E-Proceeding ICSEM 2013

e-proceeding
INTERNATIONAL CONFERENCE ON STRUCTURAL

ENGINEERING AND MECHANICS (ICSEM 2013)
December 20-22, 2013
Co-Editors: Pradip Sarkar, Robin Davis P. and Srinivas Sriramula
ISBN 978-93-80813-26-4
DEPARTMENT OF CIVIL ENGINEERING

NATIONAL INSTITUTE OF TECHNOLOGY ROURKELA
ODISHA - 769 008, INDIA
TABLE OF CONTENTS

Paper Authors Title Page

No. No.
R. P. Khandelwal; Calculation of inter-laminar shear stresses in laminated

002 Anupam Chakrabarti and composite shallow shell using least square of error (LSE) 1
Pradeep Bhargava method
Ajay Kumar; Anupam
Finite element analysis of free vibration of laminated
003 Chakrabarti and Pradeep 12
composite spherical shells with cut-outs
Bhargava
004 S. N. Patel Nonlinear response of laminated composite stiffened plates 19
Revived technique for compressive strength of CFRP

Manav Mittal and
006 confined concrete cylinders using Gaussian process 36
Pijush Samui
regression
M.N.A. Gulshan Taj and

008 Bending analysis of functionally graded sandwich plates 41
Anupam Chakrabarti
V. Chitra and Effect of lay-up sequences and imperfections on the

009 48
R.S. Priyadarsini dynamic buckling of CFRP cylindrical shells
Evaluation of shear strength of RC members using different

010 Suprabhat. Jakati 58
methods
Shivaji T Bidgar and The study of tension stiffening and crack band width in
012 65
Partha Bhattacharya reinforced concrete beam under flexure
Soumi Bhattacharyya and Effect of mass ratio on the performance of a TMD with
015 71
Aparna (Dey) Ghosh non-optimal damping
Chhabirani Tudu and Study of torsional behaviour of rectangular RC beam

016 79
Asha Patel wrapped with GFRP
Jyotirmoya Dutta
Control of wind induced vibration in transmission line
017 Majumdar and 88
towers using tuned liquid column damper
Aparna (Dey) Ghosh
Swetapadma Panda;
Stress analysis around spiral casing of Francis Turbine of a
018 Joygopal Jena and 96
Hydel power house by finite element method
Bidyadhar Basa
i

No. No.
Benazir F. Ahmed and Influence of location of staircase on seismic behaviour of

019 105
Kaustubh Dasgupta RC flat slab building
Arijit Acharjya and Influence of staircase and elevator core on twisting

020 113
Kaustubh Dasgupta behaviour of RC wall-frame buildings
P. Bhattacharya and Random vibration analysis of RCC structure using pseudo-

021 121
S. Karar excitation method
A. Roy and Aparna (Dey) Tuned liquid damper system for seismic vibration control of
022 129
Ghosh elevated water tanks
Manish K. Singh; Sailesh
Adhikari; Sumit Study on stability of natural draft hyperbolic RC cooling
023 137
Mazumder and tower shell under wind induced flow field
Amiya K. Samanta
R. Velmurugan and An experimental and numerical study of natural-fiber
025 145
Jafar Sadique P composites subjected to different environmental conditions
N Murli Krishna and Evaluation of performance point using energy based

026 154
Md Masihuddin Siddiqui pushover analysis
Arghya Sengupta and

028 Seismic behaviour of R/C frames with biaxial interaction 164
Rana Roy
Pawan Thakur and Seismic behaviour of plan-asymmetric structures under

029 174
Rana Roy spectrally matched records
Syed H. Basha and Influence of masonry properties on lateral load response of

031 185
Hemant B. Kaushik reinforced concrete frames
Somdatta Goswami;
Adaptive response surface method in structural response
032 Subrata Chakraborty and 194
approximation under uncertainty
Shymal Ghosh
Avaya K. Baliarsingh; P.
Prediction of fatigue crack propagation life in single edge
035 K. Ray; B. B. Verma and 203
notched beam using exponential model
Deepak K. Agarwalla
Deepak K. Agarwalla;
Fuzzy logic based fault diagnosis of structures using
036 Dayal R. Parhi and 210
vibration responses
Amiya K. Dash
ii

No. No.
Komathi M. and Strengthening of columns for shear in reinforced concrete

037 222
Amlan K.Sengupta buildings
Shakti P. Jena and

038 Dynamic analysis of cantilever beam with moving mass 230
Dayal R. Parhi
Saswati De and Behaviour of two-span continuous steel box RC deck

039 238
K. Bhattacharya composite bridges
Prediction of hysteretic energy demand in steel moment

Shantaram Parab; Jatin
042 resisting frames using multivariate adaptive regression 250
Alreja and Pijush Samui
splines
H.D. Chalak; Anupam
Chakrabarti; Mohd.
043 Buckling analysis of laminated soft core sandwich plate 258
Ashraf Iqbal and Abdul
Hamid Sheikh
A. Subbulakshmi and Nonlinear analysis of square concrete filled steel tubes
044 266
P. Jayabalan under constant axial load and monotonic lateral loading
Bijan Kumar Roy and Reliability based design of TMD system considering system
045 275
Subrata Chakraborty parameter uncertainty in seismic vibration control
Performance evaluation of concrete against rebar corrosion

046 Bulu Pradhan 284
in composite chloride-sulphate exposure conditions
Potentiodynamic polarization study on bare steel in

Fouzia Shaheen and
048 concrete powder solution extracts contaminated with 291
Bulu Pradhan
chloride and sulphate ions
Arun Mukherjee; Sreyashi Non-linear vibration and dynamic response of epoxy-based
049 Das (nee Pal) and A. Guha laminated composite plate with square cut-out at elevated 300
Niyogi temperature
Seismic vulnerability assessment of Indian standard code
M. V. N. Sivakumar and
053 designed midrise and high-rise RC framed structures using 308
B. Vineel Kumar
fragility analysis
Snehal Kaushik and Seismic behaviour of slab-structural wall junction in rc

054 316
Kaustubh Dasgupta building
Sanjay Goswami; Partha

Investigating effectiveness of dynamic strain response for
055 Bhattacharya and Subham 324
damage detection in composite beams
Rath
iii

No. No.
A. Vanuvamalai and
056 Finite element analysis of road tunnel 332
K.P. Jaya
Tushar Kanti Dey ,

Optimization of a fiber reinforced polymer web core skew
059 Anupam Chakrabarti and 339
bridge
Umesh Kumar Sharma
G. V. Rama Rao; M.
Cost effective structural systems for high rise buildings
060 Pavan Kumar and 347
subjected to seismic load
Sk. Rasool
Soumya Bhattacharjya; Probabilistic assessment of safety, economy and robustness
061 Urmi Saha; Asish Modak of steel structure including parameter uncertainty in is: 800- 356
and Baidyanath Sarker 2007 format
A computer aided fem based numerical solution for
A. K. Nayak; R. A. Shenoi
062 transient response of laminated composite plates with cut- 363
and J. I. R. Blake
outs
C. Mohanlal and Optimal design of a retractable roof structure using real

063 372
N. Murali Krishna coded genetic algorithm
The change in stress distribution pattern within the earthen

064 Nityananda Nandi 381
dam due to the effect of seepage
Bhaskar Ghosh and

Fundamental period of vertically irregular frames – a
065 Subhashish Roy 394
modification over the codal stipulation
Chowdhury
Maganti Janardhana; Jogi
Structural behaviour of an RC building on sloping ground
066 Naidu P. and Hymavathy 403
under earthquake load
K.
067 V. R. Kar and S. K. Panda Thermal stability analysis of functionally graded panels 416
M. Mishra and
068 Properties of rubberised fly ash concrete 422
K. C. Panda
G Muthukumar and Influence of opening on the dynamic structural response of

073 430
Manoj Kumar rectangular slender shear wall for different damping ratios
S S S Sastry;
Studies on the channel fitting in aircraft using classical and
074 Anasuyeswar K; Ayush 447
finite element methods
Mathur and Shailaja B
P.V. Katariya and Modal analysis of laminated composite spherical shell

076 454
S.K. Panda panels using finite element method
iv

No. No.
Hipparagi A. K.;
Enhancement of shear ductility for HSC beams with
077 Injaganeri S.S. and 463
minimum shear reinforcement
Mahadevgouda H.
Spectral strain energy based approach for system

083 J S Ali and S Gupta 471
identification in structures
A. Syed Mohamed and Towards an eco-friendly concrete with waste glass and rice
085 478
C. O. Arun husk ash
K. V. S. N. Murthy and A study on analysis of turbo generator foundation resting on

086 486
B. Dean Kumar pile foundations for various earthquake zones in india
Rajiv Verma and

087 Transient analysis of an elliptical journal bearing 493
Puneet Mathur
A.S. Balu and Response surface based analysis of structures with fuzzy
088 503
B.N. Rao variables
Shiv Shanker
Seismic design factors for steel moment frames with
090 Ravichandran and 510
masonry in-fills: Part 1
Richard E. Klingner
Shiv Shanker
Seismic design factors for steel moment frames with
091 Ravichandran and 521
masonry in-fills: Part 2
Richard E. Klingner
K. Sharma; V. Bhasin;
Simulation of bi-metallic interfacial crack using EFGM and
092 I. V. Singh; B. K. Mishra 532
XFEM
and R. K. Singh
Evaluation of stress concentration for a simply supported
Moon Banerjee; N. K. Jain
093 laminated composite plate with a centre circular hole under 540
and S. Sanyal
uniform transverse loading
Revathi P.; Ramesh, R. A. Effect of treatment methods on the strength characteristics

096 548
and Lavanya, K. of recycled aggregate concrete
Shashank Pandey and Finite element analysis of sandwich plates with functionally
097 556
S. Pradyumna graded material core using a layerwise theory
Irshad A Khan and Diagnosis of multiple crack of cantilever composite beam

099 564
Dayal R Parhi by vibration analysis and hybrid AI technique
Rehan A. Khan and Performance based design of rcc building under earthquake
109 571
T. Naqvi forces
v

No. No.
Effect of stiffened cut-out on vibration and parametric

111 A.K.L. Srivastava 580
excitation
H. S. Panda; S. K. Sahu Modal analysis of delaminated woven fibre composite

112 585
and P. K. Parhi plates in moist environment
Abhishek Kumar;
Seismic analysis of reinforced concrete ventilation stack
117 V. S. Phanikanth and 593
using simplified modal analysis technique
K. Srinivas
G Srikar; B Gopi and Effect of temperature on compressive behaviour of concrete

118 604
S Suriya Prakash reinforced with structural polypropylene fibres
C Sreenivasulu; R Mehar
Behaviour of masonry assemblages made of soft brick
119 Babu; S Suriya Prakash 614
under compression
and K.V.L. Subramaniam
K.V.V. Sumanth and Effect of tension stiffening on behaviour of concrete

120 624
S. Suriya Prakash columns under pure torsion
Optimal earthquake resistant design of a fixed beam using

121 M. R. Das and S. Samal 633
a simple optimization tool
Souvik Chakraborty and Uncertainty propagation using hybrid HDMR for stochastic
122 642
Rajib Chowdhury field problems
P. Dinakar and Design and development of high strength self compacting

123 657
Manu S. Nadesan concrete using metakaolin
Monalisa Priyadarshini;
Robin Davis;
124 Seismic reliability assessment of R/C stepped frames 663
Pradip Sarkar and
Haran Pragalath D C
P.R. Ravi Teja and Major indian earthquakes and earthquake risk assessment –
126 671
Sasmita Sahoo a case study
P. R. Maiti; Bhawesh
Analysis of plate resting on Pasternak and Winkler
128 Madhukar and 685
foundation due to moving load
Satyam Mandloi
Pardeep Kumar and

129 Determination of dynamic modulus of elasticity of concrete 695
R. K. Sharma
vi

No. No.
S. C. Choudhury;
Construction of high level bridge over river Mahanadi at
A. Simadri Dora;
130 6th km of Sambalpur-Sonepur road to Chadheipank-Binca 704
M. K. Tripathy and
road - a case study
P. M. Tripathy
Performance evaluation of tuned liquid damper on response
131 D. Jena and K.C. Biswal 707
of high rise structure under harmonic load
Ganesh R.; Haran
Pragalath D.C.; Robin
132 Seismic fragility analysis of axially loaded single pile 713
Davis; Pradip Sarkar and
S. P. Singh
Priyadarshi Das and Optimal design of a tuned mass damper using a simple
133 720
ManasRanjan Das optimization tool
Sourabh S. Deshpande and Analysis of thick beams using first order shear deformation
134 728
R. L. Wankhade theory
Jitendra Kumar Meher and Free vibration of multiple-stepped Bernoulli-Euler beam by

135 737
Manoranjan Barik the spectral element method
Istiyak Khan;
Attenuation of seismic response of structures using passive
136 M. K. Agrawal and 745
devices
S. B. Chafle
N. Trivedi and R. K. Experimental and analytical estimation of concrete creep

137 753
Singh for large prototype structure
N. Trivedi and R. K. Numerical simulation for VTT impact test under iris
138 760
Singh program
A. Ravi Kiran; M. K.
Synthetic ground motion generation using a semi analytical
139 Agrawal; G. R. Reddy; R. 766
model
K. Singh and K. K. Vaze
A. Ravi Kiran; P. N.
Dubey; M. K. Agrawal; G. Experimental and analytical studies of ratcheting in
140 774
R. Reddy; R. K. Singh and pressurized piping system under seismic load
K. K. Vaze
M. Bandyopadhyay and Numerical analysis of semi-rigid jointed steel frame using
141 782
A. K. Banik rotational springs
Kotla Shiva Kumar and Passive control of structures subjected to earthquake

142 791
Diptesh Das excitation using particle swarm optimization
vii

No. No.
Jagdish Malav; Swapnil

Takle; Kali P. Sethy; Properties of high strength concrete containing ultrafine
143 799
Dinakar. P and slag
U. C. Sahoo
A. K. Banik and Progressive collapse study of semi-rigid jointed frame
144 807
M. Bandyopadhyay structures: a state-of-the art review
M. K. Agrawal; A. Ravi
Kiran; G. R. Reddy; R.K.
Re- evaluation of industrial equipments for earthquake and
145 Singh; K. K. Vaze; 827
wind loads
D. K. Sakhrodia; B.
Biswas and A. Bhowmick
R. K. Sharma and Utilization of dredged material from reservoir of hydro
146 834
Pardeep Kumar project as fine aggregate in concrete
D. V. Prasada Rao and Effect of silica fume on strength of

147 843
G. V. Sai Sireesha Partially used recycled coarse aggregate concrete
Seismic response of vertically irregular structures: issues to

148 Pratika Preeti 850
be addressed
Tarapada Mandal and

149 Slope stability analysis by static and dynamic method 858
Sanjay Sengupta
Karthik Subhash A.; U. K. Damage detection in simply supported beam - an ANN

150 865
Mishra and Ganesh R. approach
Kirtikanta Sahoo; Pradip

151 Behaviour of recycled aggregate concrete 871
Sarkar and Robin Davis P.

viii

International Conference on Structural Engineering and Mechanics
December 20-22, 2013, Rourkela, India
Paper No.: 002
CALCULATION OF INTER-LAMINAR SHEAR STRESSES IN LAMINATED

COMPOSITE SHALLOW SHELL USING LEAST SQUARE OF ERROR (LSE)
METHOD
1 2 3
R.P. Khandelwal , A. Chakrabarti and P. Bhargava
1
Research Scholar, Dept. of Civil Engineering, Indian Institute of Technology, Roorkee, India
2
Associate Professor, Dept. of Civil Engineering, Indian Institute of Technology, Roorkee, India
3
Professor, Dept. of Civil Engineering, Indian Institute of Technology, Roorkee, India
Email: ravi.iitdelhi@gmail.com, anupam1965@yahoo.co.uk
ABSTRACT :
Evaluation of inter-laminar shear stresses in laminated composite shells using 2D finite element (FE) models is
difficult and it involves cumbersome post processing techniques. In this paper a simple and efficient method has
been proposed for accurate calculation of inter-laminar shear stresses in laminated composite shells by
employing a displacement based C0 FE model derived from higher order shear deformation theory (HSDT) and
a Least Square Error (LSE) method. In order to include the effect of three curvature terms in the strain
components of composite shells, Sander’s approximations are followed. In this model the first derivatives of
transverse displacement have been treated as independent variables to overcome the problem of C1 continuity in
the FE implementation associated with the above shell theory (HSDT). The LSE method is applied at the
post-processing stage, after in-plane stresses are calculated by using the above FE model based on HSDT. Thus
the proposed method is quite simple compared to the usual method of integrating the 3D equilibrium equations
for calculation of transverse shear stresses in laminated composite shells. Accuracy of the proposed method is
demonstrated in the numerical examples through comparison of the present results with those obtained from
different models based on higher order shear deformation theory (HSDT) and 3D exact and analytical solutions.
KEYWORDS: Shell, Composite, Higher order Shear deformation theory, Finite Element, LSE.
1. INTRODUCTION
Laminated composite shell structures are widely used in several engineering applications; especially in civil,
mechanical, aerospace, automobile, marine, nuclear and others because of their excellent properties especially,
high strength to weight and strength to stiffness ratios. However, composites are weak in shear due to their low
shear modulus compared to extensional rigidity. Thus the effect of shear deformation is quite significant which
may lead to failure. Moreover, the accurate evaluation of transverse shear stresses by using 2D models available
in the literature is quite cumbersome. Therefore, there is a genuine requirement for the development of a shell
model for the accurate prediction of transverse shear stresses within the framework of 2D analysis.
In order to properly design laminated composite shell structures, various shell theories and related numerical
procedures have been developed. Many of the classical theories developed for thin elastic shells are based on the
Love- Kirchhoff’s assumptions. In these theories, the normal to the mid-plane before deformation remains
normal and straight after deformation i.e. the effect of transverse shear deformation is neglected. The application
of these theories in analyzing laminated composite shells where shear deformation is very significant, may lead
to errors in the calculation of deflections, stresses and natural frequencies.
In order to include the effect of transverse shear deformations, the First order shear deformation theory (FSDT)
was developed (Reissner, 1945 and Mindlin, 1951) considering the uniform transverse shear strain throughout
the thickness of the plate. In this theory shear correction factors are required for the analysis to account for the
parabolic distribution of the transverse shear stresses.
For a better representation of the transverse shear deformations, higher order shear deformation theories (HSDT)
are proposed by many researchers (Hsu and Wang, 1970; Dong and Tso, 1972; Yang, 1973; Whitney and Sun,
1973, 1974; Bhimaraddi, 1984; Reddy and Liu, 1985; Mallikarjuna and Kant, 1992; Fraternali and Reddy, 1993;
Bhimaraddi, 1993; Huang, 1994 and others).
Reddy and Liu (1985) presented a simple higher-order shear deformation theory (HSDT) by modifying Sander’s
theory, for the analysis of laminated shells. It contains the same dependent unknowns as in the first-order shear
deformation theory (FSDT). In this theory, the displacements of the middle surface are expanded as cubic
functions of the thickness coordinate and the transverse displacement is assumed to be constant through the
thickness. This displacement field leads to the parabolic distribution of the transverse shear stresses (and zero
transverse normal strain) and no shear correction factors are used. Subsequently, Mallikarjuna and Kant (1992)
have also developed a refined shell theory for the analysis of fibre-reinforced composite and sandwich shells.
This theory is based on a higher-order displacement model and three-dimensional Hooke’s law which gives a
realistic cross-sectional deformation. Fraternali and Reddy (1993) proposed a penalty finite element model
based on layer-wise theory (Reddy, 1987) for analyzing the laminated composite shells which can be used to
calculate both the displacements and stresses accurately. This theory also predicts the accurate
through-thickness distribution of inter-laminar shear stresses which otherwise would have been very difficult to
obtain by using a usual 2D displacement based finite element model.
Bhimaraddi (1993) presented 3D elasticity solution for static response of laminated doubly curved shallow
shells on rectangular plan-form by assuming the ratio of shell thickness to its middle surface radius as negligible
as compared to unity. These 3D elasticity solutions are then utilized to compare the accuracy of various 2D shell
theories and their applicability for the range of shell thickness to length ratio. It has been observed that the
various 2-D shell theories are not suitable for the shells with thickness to length ratio greater than or equal to
1/10.
Huang (1994) presented a third order shell theory based on Reddy’s parabolic shear strain field. In this theory,
he has again introduced the shear correction factor to incorporate the continuity of inter-laminar transverse shear
stresses. Shu and Sun (1994) has also presented an improved higher-order theory for laminated composite plates
which satisfy the shear stress continuity condition across each layer interface and includes the effects of material
and ply-up patterns on the displacement field. Liew and Lim (1996) modified the simple higher order theory of
Reddy and Liu (1985) by incorporating the Lame parameter (1+z/Rx) and (1+z/Ry) for transverse strains.
Xiao-ping (1996) presented a simple higher-order shear deformation theory for laminated composite shallow
shells using Love’s first-order geometric approximation and Donnell’s simplification. This theory contains the
same number of dependent unknowns as in the FSDT. This theory also accounts the parabolic distribution of the
transverse shear stresses through the thickness of the shell and continuity of transverse shear stresses across each
layer interfaces. This theory predicts more accurate responses than first-order theory and some higher-order
theories without using the shear correction factor. However, these theories (Reddy and Liu, 1985; Xiao-ping,
1996) demand the C1 continuity of transverse displacements during their finite element implementations. Cho et
al. (1996) also proposed an efficient higher order shell theory for symmetric laminated composites. In this
theory, the in-plane displacement fields are obtained by superposing a global cubically varying displacement
field on a zig-zag varying displacement field. This theory satisfies the transverse shear stress continuity
condition at layer interfaces and zero transverse shear stress condition at the free surface. In addition to that,
they also calculated the accurate through-thickness distribution of transverse stresses for cylindrical shell in the
post processing stage after calculating the in-plane displacements based on the first order shear deformation
shell theory. Kant and Khare (1997) proposed a flat facet quadrilateral composite shell element based on
higher-order theory for the analysis of thin-to-thick laminated plates and shells.
It is also been observed that among all the higher order theories discussed above, only the theory proposed by
Yang (1973) account for twist curvature (1/Rxy), which is essential while analyzing the hyper and conoidal shells.
Pradyumna and Bandyopadhyay (2008) also studied the behavior of laminated composite shells by extending
2
the higher-order shear deformation theory developed by Kant and Khare (1997) considering three radii of
curvature. However, this theory contains some nodal unknowns which do not have any physical significance, so
incorporation of appropriate boundary conditions becomes a problem.
Recently, Kumar et al. (2012) studied the vibration characteristics of laminated composite skew hyper shells
using higher order theory. In this theory, the displacement fields are assumed in the same form as given by
Reddy (1984). In this theory, the transverse shear stresses are assumed to vary in parabolic manner across the
shell thickness. The effect of all three radii of curvature is also included in this formulation.
Conventionally, the transverse stress recovery from the 3D equilibrium equations is done by some cumbersome
integration of these equations through the thickness of laminate. The transverse stresses can also be calculated
by Reissner Mixed Variational Theorem (RMVT) by satisfaction of the inter-laminar equilibrium (Reissner,
1984, 1986).
Considering all these aspects in view, an attempt has been made in this study to develop an improved FE plate
model based on HSDT and LSE method to accurately predict the deflections and stresses of laminated
composite shallow shells for different loadings, boundary and geometric conditions. In the present study, the C0
finite element model proposed by Kumar et al. (2012) based on simple higher order shear deformation theory
(HSDT) is modified to develop a model in suitable combination of HSDT and LSE method. In the present paper
the global response (i.e. displacement) is calculated first, using this C0 FE model based on HSDT and then an
accurate prediction of transverse stresses is done by using an efficient post processing technique (LSE method).
Outstanding performance of the method (LSE) in combination with HSDT is presented through many numerical
results. The accuracy of the proposed finite element model is established by comparing the present results with
3D elasticity and other solutions.
2. MATHEMATICAL FORMULATION
In Figure 1, a laminated composite shell made of a finite number of uniformly thick orthotropic layers is shown
where the layers are oriented arbitrarily with respect to the shell coordinates (x, y, z). The mid-plane (i.e. the plane
x-y at z = 0) of the shell is taken as the reference plane. The three-dimensional shell problem has been reduced in
to two-dimensional one, by expressing the in-plane displacement components u(x, y, z) and v(x, y, z) at any point
in the shell, in terms of in-plane displacement variables and thickness coordinates by using the Taylor’s series.
The transverse displacement, w(x, y) is assumed to be constant throughout the shell thickness. Since the transverse
shear stresses actually vary parabolically throughout the shell thickness, the in-plane displacement field should
have the cubic functions of the thickness coordinate.
Figure 1 Laminated composite doubly curved shallow shell geometry and mapped shell panel
In view of the above, the displacement fields are assumed to be of the same form as given by Reddy (1984) which
3
satisfy the above criteria and may be expressed as below:
u ( x, y, z ) = u0 ( x, y ) + zθ x ( x, y ) + z 2ξ x ( x, y ) + z 3ζ x ( x, y ) (1)
v( x, y, z ) = v0 ( x, y ) + zθ y ( x, y ) + z 2ξ y ( x, y ) + z 3ζ y ( x, y )
(2)
w(x, y) = w0(x, y) (3)
where u, v and w are the total displacements of a point (x, y, z) in the shell along x, y and z directions,
respectively. The parameters u0, v0, and w0 are the displacements of a point at the mid-plane; and θ x and θ y are
the rotations of the normal with respect to the mid plane about the y and x axes, respectively. ξx , ζx , ξy and ζy are the
higher-order unknown terms which can be determined by using the zero transverse shear stress ( τ xz z =± h / 2 = 0
and τ yz = 0 ) conditions at the top and bottom surface of the shell.
z =± h / 2
Using the above mentioned conditions, the displacement field given in Eqs. (1)-(3) may be written as
4 z 2 4 z 3 ∂w 4z2 4z3 *
u = u0 + zθ x (1 − ) − ( ) = u + zθ (1 − )− ψx (4)
3h 2 3h 2 ∂x
0 x
3h 2 3h 2
4 z 2 4 z 3 ∂w 4z 2 4z3 *
v = v0 + zθ y (1 − 2 ) − 2 ( ) = v0 + zθ y (1 − 2 ) − 2 ψ y (5)
3h 3h ∂y 3h 3h
The linear strain-displacement relations according to Sander’s approximation are given as:
∂u w ∂v w ∂v ∂u 2 w
εx = + , εy = + , γ xy = + + , (6)
∂x Rx ∂y Ry ∂x ∂y Rxy
∂u ∂w u v ∂v ∂w v u
γ xz = + − A1 − A1 and γ yz = + − A1 − A1 (7)
∂z ∂x Rx Rxy ∂z ∂y Ry Rxy
where Rx and Ry are the radii of curvature of the shell along x and y directions respectively, and Rxy is the twist
radius of curvature.
After substituting the expressions from Eqs. (4), (5) and (3) into the Eqs. (6) and (7), one may get,
4z 2 4z3 *
ε x = ε x 0 + zK x (1 − )− Kx (8)
3h 2 3h 2
4z2 4z3 *
ε y = ε y 0 + zK y (1 − )− Ky (9)
3h 2 3h 2
4z 2 4z3
γ xy = γ xy 0 + zK xy (1 − 2 ) − 2 K xy* (10)
3h 3h
2 3
4z 4z 4z3
γ xz = φx + zK xz (1 − 2 ) − 2 K xz* − 2 K xz** (11)
3h 3h 3h
2 3
4z 4z 4z3
γ yz = φ y + zK yz (1 − 2 ) − 2 K *yz − 2 K ** yz
(12)
3h 3h 3h
⎧⎪ ∂u w ∂v w ∂u ∂v 2 w0 ⎫⎪
where {ε x 0 , ε y 0 , γ xy 0 } = ⎨ 0 + 0 , 0 + 0 , 0 + 0 + ⎬ (13)
⎪⎩ ∂x Rx ∂y Ry ∂y ∂x Rxy ⎭⎪
4
⎧ u ⎪⎫
{φ ,φ } = ⎪⎨ ∂∂wx + θ
x y x − A1
u0 v
− A1 0 ,
∂w
Rxy ∂y
v
+ θ y − A1 0 − A1 0 ⎬ (14)
⎩⎪ Rx Ry Rxy ⎭⎪
⎧⎪ ∂θ x ∂θ y ∂θ x ∂θ y ∂ψ x* ∂ψ *y ∂ψ x* ∂ψ *y ⎫⎪
{ x y xy x y xy } ⎨ ∂x , ∂y , ∂y + ∂x , ∂x , ∂y , ∂y + ∂x ⎬
K , K , K , K *
, K *
, K *
= (15)
⎪⎩ ⎭⎪
⎧⎪ θx θy θy θx ψ x* ψ *y ψ *y ψ x* ⎫⎪
{K xz , K yz , K *
xz , K *
yz } = ⎨ − A1 − A1 , − A1 − A1 , − A1 − A1 , − A1 − A1 ⎬ (16)
⎪⎩ Rx Rxy Ry Rxy Rx Rxy Ry Rxy ⎪⎭
and {K **
xz yz } = {θ x + ψ x , θ y + ψ y }
, K ** * *
(17)
A1 is a tracer by which the mathematical formulation can be converted to appropriate shear deformation theory.
For example, by taking A1 as zero, the formulation can be reduced to the shear deformable Love’s first
approximation while in the present formulation based on HSDT the value of A1 is taken as 1.0.
The stress–strain relationship of a lamina (say, k-th lamina) having any fiber orientation with respect to structural
axes system (x-y-z) may be expressed as
⎧σ x ⎫ ⎡ Q11 Q12 Q16 0 0 ⎤ ⎧ε x ⎫
⎪ ⎪ ⎢ ⎥ ⎪ ⎪
⎪σ y ⎪ ⎢Q12 Q 22 Q 26 0 0 ⎥ ⎪ε y ⎪
⎪ ⎪ ⎢ ⎥ ⎪ ⎪
⎨τ xy ⎬ = ⎢Q16 Q 26 Q 66 0 0 ⎥ ⎨γ xy ⎬ (18)
⎪ ⎪ ⎢ ⎥ ⎪ ⎪
⎪τ xz ⎪ ⎢ 0 0 0 Q 55 Q 45 ⎥ ⎪γ xz ⎪
⎪τ yz ⎪ ⎢ 0 0 Q 45 Q 44 ⎥⎦ k ⎪⎩γ yz ⎪⎭
⎩ ⎭ ⎣ 0
or {σ } = ⎡⎣Q ⎤⎦ {ε }k
(19)
The rigidity matrix ⎡⎣Q ⎤⎦ can be evaluated by material properties and fiber orientation following usual
k
techniques for laminated composites.
By using the Eqs. (8)-(17) and (3), the strain field vector may be written as
{ε } = [ H ]{ε }
0 (20)
where the [ H 0 ] matrix consists of terms containing z and some term related to material properties having the
{}
dimension 5 ×17 , ε is the strain field vector and {ε } is the modified strain vector at the reference plane, that
is at the mid plane,
{ε } = {ε x0 , ε y 0 , γ xy 0 , φx , φ y , K x , K x* , K y , K *y , K xy , K xy* , K xz , K xz* , K yz , K *yz , K xz** , K **yz }
T
(21)
The nodal unknowns for the present FE model may be defined with respect to the global coordinate system as
shown in Fig. 1 as
{δ } = {u 0 v0 w0 θ x θ y ψ x* ψ *y } (22)
The actual displacement fields require C1 continuity of the transverse displacement for the finite element
implementation. In order to avoid the usual difficulties associated with C1 continuity requirement, the derivatives
of w with respect to x and y are expressed as independent variables as follows:
∂w ∂w
= ψ x* and = ψ *y (23)
∂x ∂y
Now the potential energy (V) of the laminated shell under the action of transverse load of intensity q may be
written as
V = ∫ {ε } [ D ]{ε } dxdy + Pλ − ∫ w q dxdy
T
(24)
5
n
where, [ D ] = ∑[H ] ⎡Q ⎤ [ H 0 ] dz
T
0 ⎣ ⎦k (25)
k =1
The penalty term may be expressed using the Eq. (23) as below:
T
⎧ ∂w ⎫ ⎧ ∂w ⎫ ⎧ ∂w ⎫ ⎧ ∂w ⎫
T
Pλ = γ ∫∫ (⎨ −ψ x* ⎬ ⎨ −ψ x* ⎬ + ⎨ −ψ *y ⎬ ⎨ −ψ *y ⎬)dxdy (26)
⎩ ∂x ⎭ ⎩ ∂x ⎭ ⎩ ∂y ⎭ ⎩ ∂y ⎭
where γ is the penalty parameter.
In the present study, a nine node quadrilateral C0 continuous isoparametric element has been used. The
generalized displacement field variable {δ} and the element geometry at any point (x, y) within the element are
expressed in terms of nodal variables as follows.
9 9 9
δ = ∑ N iδ i , x = ∑ N i xi and y = ∑ N i yi (27)
i =1 i =1 i =1
where Ni is the standard shape functions used for the associated nodes.
The strain vector in equation (20) may be expressed in terms of {δ} containing nodal unknowns,
{ε } = [ B]{δ } (28)
where [B] is the strain displacement matrix having the dimension, 17 × 63 .
By using the principle of minimization of total potential energy of the system, one may get the equilibrium
equation as below
[ K ]{δ } = {P} (29)
where [K] is the element stiffness matrix and {P} the nodal load vector.
The above linear simultaneous global equations are solved for the problem of the laminated shell after
incorporation of appropriate boundary conditions.
The in-plane stresses are calculated with the help of constitutive relationship by using the condition of stress
continuity as in Eq. (18). It is very difficult to accurately evaluate the transverse shear stresses using the
constitutive relationship for these types of 2D FE model. Therefore, a simple post processing technique based on
the least square error (LSE) method is used to calculate the inter-laminar shear stresses (Khandelwal et al., 2012).
The LSE method proposed is based on minimization of the sum of the square of error by satisfying the 3D
equilibrium equations. The derivatives of in-plane normal and shear stresses with respect to x and y and derivative
of inter-laminar shear stresses (assumed to have a quadratic variation across the thickness) with respect to z are
substituted in the 3D equilibrium equations to obtain the error function of sum of squares. This function is then
minimized with respect to the inter-laminar shear stress which results in a set of linear algebraic equations. These
equations are then solved to obtain the inter-laminar shear stresses. The LSE method used in the present
formulation is similar to the one given in Ref. (Khandelwal et al. , 2012).
3. NUMERICAL RESULTS AND DISCUSSIONS
3.1. Convergence study
The problem of a simply supported cross ply (0/90/90) spherical shell, subjected to uniformly distributed
loading of magnitude q with a/b = 1 and Rx = Ry= R is studied for different radius to span ratio (R/a) and
thickness ratio (h/a). The spherical shell is analyzed by taking different mesh divisions and the normalized
deflections, w = ( wE2 h3 / qa 4 )103 obtained at the midpoint of the shell are presented in Table 1 along with the
results given by Reddy and Liu (1985) based on HSDT and by Reddy (1984) based on exact analytical solution.
The results show that the present model is quite capable of predicting the deflections with sufficient accuracy. It
6
may be observed in Table 1, that a mesh size of 12x12 shows good displacement convergence and as such it is
being used for all subsequent analysis. It is found that the present results are in good agreement with HSDT
results of Reddy and Liu (1985) as well as the closed form solution of Reddy (1984).
Although the individual layers possess different orientations but they have equal thickness and material property
(E1=25E2, E2 = E3, G12 = G13 = 0.5E2, G23 = 0.2E2, ν12 = 0.25 and ν13 = 0.01). This is also applicable to all the
subsequent problems unless mentioned otherwise.
Table 1 Non-dimensional central deflections ( w ) of a simply supported cross-ply laminated spherical shell under
UDL
R/a Reference Theory h/a = 0.01 h/a = 0.1
5 Present (2x2)a HSDT 1.5879 10.8092
Present (4x4) HSDT 1.5062 10.4379
Present (8x8) HSDT 1.5116 10.4232
Present (12x12) HSDT 1.5119 10.4228
Present (16x16) HSDT 1.5119 10.4227
Reddy and Liu (1985) Exact analyticalb 1.5092 10.3320
Reddy (1984) Exact analyticalc 1.5118 9.7937
10 Present (12x12) HSDT 3.6468 10.7824
Reddy and Liu (1985) HSDT 3.6438 10.7743
Reddy (1984) Exact analytical c 3.6445 10.1100
20 Present (12x12) HSDT 5.5527 10.8761
50 Present (12x12) HSDT 6.4901 10.9026
100 Present (12x12) HSDT 6.6498 10.9064
a
Entries inside the parenthesis indicate mesh division
b
Results based on HSDT
c
Results based on Sander’s shell theory
3.2. Simply supported cross-ply (0/90) laminated spherical shell subjected to UDL
In this example, the problems of a simply supported cross ply (0/90) laminated spherical shell (a/b = 1 and Rx =
Ry= R), of equal individual layer thickness have been solved taking radius to span ratio (R/a) as 10 while the
thickness ratio is considered as h/a = 0.1.
The transverse shear stresses are calculated by using the constitutive relationship as well as by the LSE method. A
multiplication factor, m= h/q0a is used to convert the transverse shear stresses in non-dimensional forms, where a
is the least lateral dimension of the shell projection and q0 is the amplitude of the applied load at the shell
mid-surface. The abbreviations ‘LSE’ and ‘Const’ preceded by an underscore in the legends of the figures indicate
quantities evaluated using LSE method and constitutive relations, respectively.
7
0.6 0.6
HSDT_Equil
0.5 0.5 HSDT_Const
0.4 0.4 Fraternali and Reddy (1993)
Normalized Thickness (z/h)

0.3 HSDT_LSE 0.3 h/a = 0.1

0.2 HSDT_Const 0.2 R/a =1
Fraternali and Reddy
0.1 0.1
(1993)
0.0 0.0
0.0 0.2 0.4 0.6 0.8
-0.1
h/a = 0.1 , R/a =1 -0.10.0 0.2 0.4 0.6
-0.2 -0.2
-0.3 -0.3
-0.4 -0.4
-0.5 -0.5
-0.6 (a) Normalized Transverse Shear Stress τxz (b) Normalized Transverse Shear Stress τyz
-0.6
Figure 2 Thickness variation of transverse shear stresses τxz and τyz of a simply supported cross-ply (0/90)
laminated spherical shell under UDL (h/a = 0.1, R/a = 10)
Thickness variation of non-dimensional transverse shear stress, τxz and τyz obtained at the centre of the left edge (x
= 0 and y = a/2) and centre of the bottom edge (x = a/2 and y = 0) of the shell respectively, are presented in Figure
2 along with the results reported by Fraternali and Reddy (1993) based on layer-wise theory.
It is observed in Figure 2 that for 0/90 ply orientations, the stresses calculated using the proposed combined model
based on HSDT and LSE method gives results which are very close to those obtained by Fraternali and Reddy
(1993). It may also be observed in the same figure (Figure 2) that the values of transverse shear stresses calculated
by constitutive relationships do not match well with those obtained by Fraternali and Reddy (1993).
3.3. Simply supported cross-ply (0/90/0) laminated spherical shell subjected to UDL
In this example, the problems of a simply supported laminated three-layer (0/90/0) cross ply spherical shell of
equal individual layer thickness have been solved taking different radius to span ratio (R/a) for thickness ratio
(h/a) = 0.1 and aspect ratio (a/b) = 1.
Through-the thickness variation of non-dimensional transverse shear stress, τyz, obtained at the centre of the
bottom edge (x = a/2 and y = a/2) of the mapped shell are shown in Figure 3 for two different radius to span ratio
(R/a) as 5 and 10.
0.6 0.6
HSDT_Equil HSDT_Equil
0.5 HSDT_Const 0.5 HSDT_Const
0.4 h/a = 0.1 0.4 h/a = 0.1
0.3 R/a =5 0.3 R/a =10

0.2 0.2
0.1 0.1
0.0 0.0
-0.10.0 0.1 0.2 0.3 0.4 -0.10.0 0.1 0.2 0.3 0.4 0.5
-0.2 -0.2
-0.3 -0.3
-0.4 -0.4
-0.5 -0.5
(a) Normalized Transverse Shear Stress τyz (b) Normalized Transverse Shear Stress τyz
-0.6 -0.6
Figure 3 Thickness variation of transverse shear stress τyz of a simply supported cross-ply (0/90/0) laminated
spherical shell under UDL (h/a = 0.1, R/a = 5, 10)
8
It may also be observed in the same figure (Figure 3) that the values of transverse shear stresses calculated by
constitutive relationships do not match well with those obtained by the LSE method.
3.3. Simply supported cross-ply (0/90/90/0) laminated spherical shell subjected to UDL
In this example, the problems of a four-layer (0/90/90/0) cross ply simply supported laminated spherical shell, of
equal individual layer thickness have been studied to see the effect of (R/a) ratio. The thickness ratio (h/a) is taken
as 0.1. The thickness distribution of non-dimensional transverse shear stress, τyz obtained at the centre of the
bottom edge (x = 0 and y = a/2) for different (R/a) = 5, 10, 20, 50 and 100. In order to observe the effect of
curvature, the problem of a simply supported laminated composite square plate (h/a = 0.1) subjected to uniformly
distributed load is considered here. The normalized value of transverse shear stresses obtained by using the
constitutive relationship and the LSE method are presented in the Figure 4 along with the exact 3D elasticity
solution given by Pagano (1970) for laminated plates.
It may be observed from Figure 4(a) that the transverse shear stresses calculated by the constitutive relationship
does not match with the 3D elasticity solution for plates for higher value of R/a (= 100). It is also clear from the
figure that for all (R/a) values, the curves seems to be symmetric about the mid-plane (i.e., z = 0) which means that
the effect of curvature is not captured in the calculation of transverse shear stresses based on constitutive
relationship.
It is clear from the Figure 4(b) that the transverse shear stresses obtained by the LSE method shows the effect of
(R/a) ratio. For lower (R/a) ratio, the curves seems to be un-symmetric about the mid-plane (i.e., z = 0) which is
the proper trend of the curve. Also for higher value of R/a ratio (= 100), the values of transverse shear stresses
obtained by LSE method, match quite well with the 3D elasticity solution (Pagano, 1970).
0.6 HSDT_Consti (R/a = 5) 0.6 HSDT_LSE (R/a = 5)
HSDT_Consti (R/a = 10) HSDT_LSE (R/a = 10)
0.4 HSDT_Consti (R/a = 20) 0.4
HSDT_LSE (R/a = 20)

h/a = 0.1 HSDT_LSE (R/a = 50)
0.2 0.2
h/a = 0.1
0.0 0.0
-0.2 -0.2
HSDT_Consti (R/a = 50)
-0.4 HSDT_LSE (R/a = 100)
HSDT_Consti (R/a = 100) -0.4
Plate Pagano (1970)
Plate Pagano (1970)
-0.6 -0.6
0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5 0.6
(a) Normalized Transverse Shear Stress τyz
(b) Normalized Transverse Shear Stress τyz
Figure 4 Thickness variation of non-dimensional transverse shear stress τyz of four layer (0/90/90/0) laminated
spherical shell (thickness ratio h/a = 0.1)
4. CONCLUSIONS
In this paper, a numerical model has been proposed for the first time in combination of a new C0 finite element
model based on higher order shear deformation theory (HSDT) along with a least square error (LSE) method to
accurately calculate the inter-laminar shear stresses for different problems of laminated shells. The proposed
analysis is done in two steps. The FE model based on HSDT is used first to calculate the deflections and in-plane
stresses. The LSE method is then utilized in the second step, to accurately predict the transverse shear stresses
from the results of deflections and in-plane stresses obtained by the FE analysis based on HSDT at the first stage.
Therefore, the proposed combined model may be recommended as a simple and efficient 2D method to accurately
9
calculate the deflections as well as the stresses (in-plane and transverse) for all types of laminated composites
shallow shells.
ACKNOWLEDGEMENTS
Financial support obtained for the present work from the Department of Science and Technology SERC, New
Delhi, India under the grant DST-536-CED is gratefully acknowledged.
REFERENCES
[1] Reissner, E. (1945). The effect of transverse shear deformation on the bending of elastic plates. ASME
Journal of Applied Mechanics, 12(2), 69-77.
[2] Mindlin, R.D. (1951). Influence of rotary inertia and shear in flexural motions of isotropic elastic plates.
ASME Journal of Applied Mechanics, 18(2), 31-38.
[3] Hsu, T.M., and Wang, J.T.S. (1970). A theory of laminated cylindrical shells consisting of layers of
orthotropic laminae. AIAA Journal, 8(12), 2141-2146.
[4] Dong, S.B., and Tso, F.K.W. (1972). On a laminated orthotropic shell theory including transverse shear
deformation. ASME Journal of Applied Mechanics, 39, 1091-1096.
[5] Yang, T.Y. (1973). High order rectangular shallow shell finite element. Journal of the Engineering
Mechanics Division, 99, 157-181.
[6] Whitney, J.M., and Sun, C.T. (1973). A higher order theory for extensional motion of laminated composites.
J. Sound and Vibration, 30, 85-97.
[7] Whitney, J.M., and Sun C.T. (1974). A refined theory for laminated anisotropic cylindrical shells. Journal
of Applied Mechanics, 41(2), 471-476.
[8] Bhimaraddi, A. (1984). A higher order theory for free vibration analysis of circular cylindrical shells. Int. J.
Solids Structures, 20(7), 623-630.
[9] Reddy, J.N., and Liu, C.F. (1985). A higher-order shear deformation theory of laminated elastic shells.
International Journal of Engineering Science, 23(3), 319-330.
[10] Mallikarjuna, Kant, T. (1992). A general fibre-reinforced composite shell element based on a refined shear
deformation theory. Computers & Structures, 42(3), 381-388.
[11] Fraternali, F., Reddy, J.N. (1993). A penalty model for the analysis of laminated composite shells. Int. J.
Solids Structures, 30(24), 3337-3355.
[12] Reddy, J.N. (1987). Generalization of two-dimensional theories of laminated composite plates. Commun.
Appl. Numer. Meth., 3, 173-180.
[13] Bhimaraddi, A. (1993).Three-dimensional elasticity solution for static response of orthotropic doubly
curved shallow shells on rectangular planform. Composite Structures, 24(1), 67-77.
[14] Huang, N.N. (1994). Influence of shear correction factors in the high-order shear deformation laminated
shell theory. Int. J. Solids Strut., 31, 1263-1227.
10
[15] Shu, X., Sun, L. (1994). An improved simple higher-order theory for laminated composite plates.
Computers & Structures, 50(2), 231-236.
[16] Liew, K.M., and Lim, C.W. (1996). A higher-order theory for vibration of doubly curved shallow shells.
ASME, Journal of Applied Mechanics, 63, 587-593.
[17] Xiao-ping, S. (1996). An improved simple higher-order theory for laminated composite shells. Computers
and Structures, 60(3), 343-350.
[18] Cho, M., Kim, K., and Kim, M.H. (1996). Efficient higher-order shell theory for laminated composites.
Composite Structures, 34, 197-212.
[19] Cho, M., and Kim, M.H. (1996). A postprocess method using a displacement field of higher-order shell
theory. Composite Structures, 34, 185-196.
[20] Kant, T., and Khare, R.K. (1997). A higher-order facet quadrilateral composite shell element. Int. J. for
Numerical Methods in Engng., 40, 4477-4499.
[21] Pradyumna, S., and Bandyopadhyay, J.N. (2008). Static and free vibration analyses of laminated shells
using a higher-order theory. Journal of Reinforced Plastics and Composites, 27, 167-186.
[22] Kumar, A., Bhargava, P., and Chakrabarti, A. (2012). Vibration of laminated composite skew hyper shells
using higher order theory. Thin-Walled Structures, 63, 82-90.
[23] Reddy, J.N. (1984). A simple higher-order theory for laminated composite plates. ASME, Journal of
Applied Mechanics, 51, 745-752.
[24] Reissner, E. (1984). On a certain mixed variational theory and a proposed application. Int. J. for Numerical
Methods in Engng., 20, 1366-1373.
[25] Reissner, E. (1986). On a mixed variational theorem and on shear deformable plate theory. Int. J. for
Numerical Methods in Engng., 23, 193-200.
[26] Daniel I. M., and Ishai O. (2006). Engineering Mechanics of Composite Materials. 2nd Edition Oxford
University Press.
[27] Reddy, J.N. (1984). Exact solutions of moderately thick laminated shells. ASCE, Journal of Engineering
Mechanics, 110(5), 794-809.
[28] Pagano, N.J. (1970). Exact solutions for rectangular bidirectional composite and sandwich plates. J. Comp.
Materials, 4, 20-34.
[29] Khandelwal, R.P., Chakrabarti, A., and Bhargava, P. (2012). An efficient FE model and Least Square Error
method for accurate calculation of transverse shear stresses in composites and sandwich laminates.
Composites: Part B, 43, 1695–1704.
[30] Soldatos, K.P., Shu, X. (1999). On the stress analysis of cross-ply laminated plates and shallow shell panels.
Composite Structures, 46, 333-344.
11
Paper No.: 003
FINITE ELEMENT ANALYSIS OF FREE VIBRATION OF LAMINATED

COMPOSITE SPHERICAL SHELLS WITH CUTOUTS
Ajay Kumar1, Anupam Chakrabarti2 and Pradeep Bhargava3

1
Assistant Professor, Dept. of Civil Engineering, National Institute of Technology Patna, Patna, India
2
Associate Professor, Dept. of Civil Engineering, IIT Roorkee, Roorkee, India
3
Professor, Dept. of Civil Engineering, IIT Roorkee, Roorkee, India
Email: sajaydce@gmail.com, anupam1965@yahoo.co.uk,bhpdpfce@iitr.ernet.in
ABSTRACT:
Free vibration of laminated composite spherical shells with cutout is studied by using a 2D finite element (FE)
model based on higher order shear deformation theory (HSDT). The proposed model incorporates all the three
radii of curvatures including the cross curvature and satisfies zero transverse shear stress conditions at shell top
and bottom. The formulation includes Sanders’ approximation for doubly curved shells considering the effects
of rotary inertia and transverse shear. A realistic parabolic distribution of transverse shear strains through the
shell thickness is assumed and the use of shear correction factor is avoided. Since composites are weak in shear
hence modelling shear deformation appropriately is important which is done in HSDT more accurately than in
FSDT. The purpose of this study is to find effect of cutout size on free vibration of laminated composite
spherical shells. For this a C0 FE model is developed using a nine noded curved isoparametric element with
seven nodal unknowns at each node. C1 continuity requirement associated with HSDT is circumvented
effectively. The validation with available published results based on first order shear deformation theory (FSDT)
show good agreement. Free vibration of laminated composite spherical shells with cutouts has been done for
first five modes by varying boundary conditions and curvature of composite spherical shells. Many new results
are presented which are useful for future researchers.
KEYWORDS: composite, spherical shell, cutout, free vibration, finite element method, higher order shear
deformation theory
1. INTRODUCTION
The free vibration of composite as well as isotropic plate with cutout was studied by different researchers from
time to time. Reddy [2] investigated large amplitude flexural vibration of composite plate with cutout. Later in
1989, Malhotra et al. [3] studied free vibration of composite plate with cutout for different boundary conditions.
One of the early reports on free vibration of curved panels with cutout was due to Sivasubramonian et al. [4].
They analyzed the effect of cutouts on the natural frequencies of plates with some classical boundary conditions.
The plate had a curvature in one direction and was straight in the other. The effect of fiber orientation and size
of cutout on natural frequency on orthotropic square plates with square cutout was studied using Rayleigh-Ritz
method. Later Sivakumar et al. [5], Rossi [6], Huang and Sakiyama [7], and Hota and Padhi [8] studied free
vibration of plate with various cutout geometries. Chakravorty et al. [1] analyzed the effect of concentric cutout
on different shell options. Again in 1999, Sivasubramonian et al. [9] studied the effect of curvature and cutouts
on square panels with different boundary conditions. The size of the cutout (symmetrically located) as well as
curvature of the panels is varied. Hota and Chakravorty [10] published useful information about free vibration
of stiffened conoidal shell roofs with cutout. Later Nanda and Bandyopadhyay [11] studied the effect of
different parametric variation on free vibration of cylindrical shell with cutout using first order shear
deformation theory (FSDT).
A scrutiny of the literature on vibration of composite spherical shells with cutouts indicates that there is
no C0 FE results based on HSDT of composite spherical shells with rectangular cutouts. In the present paper the
free vibration of composite spherical shells with rectangular cutouts based on HSDT, is studied considering
different curvatures, thicknesses and boundary conditions.
2. GOVERNING EQUATIONS
A laminated shell made of a finite number of uniformly thick orthotropic layers oriented arbitrarily with respect
to the shell co-ordinates (x, y, z) is taken as shown in Figure 1. The reference plane (i.e., mid plane) is defined at
z = 0.
Figure 1 Inplane displacement across thickness of shell
The displacement fields are assumed in the form given by Reddy and Liu [17] and may be expressed as,
u ( x , y , z ) = u 0 ( x , y ) + zθ x ( x , y ) + z 2ξ x ( x , y ) + z 3ζ x ( x , y )
v ( x, y , z ) = v0 ( x, y ) + zθ y ( x, y ) + z 2ξ y ( x, y ) + z 3ζ y ( x, y )
w( x, y ) = w0 ( x, y ) (1)
where u, v and w are the displacements of a general point (x, y, z) in an element of the laminate along x, y and z
directions, respectively. The parameters u0, v0, and w0 denote the displacements of a point (x ,y) on the mid plane
of shell, θ x and θ y are the rotations of normals with respect to the mid plane about the y and x axes. The
functions ξx , ζx , ξy and ζy can be found using the conditions of zero transverse shear strains at the top and bottom
surfaces of the shell.
The linear strain-displacement relations according to Sanders’ approximation are shown in Eq. (2) below:
∂u w ∂v w ∂v ∂u 2w
εx = + A1 ε y = + A1 γ xy = + + A1
∂x Rx ∂y Ry ∂x ∂y Rxy (2)
∂u ∂w u v ∂v ∂w v u
γ xz = + − A1 − A1 γ yz = + − A1 − A1
∂z ∂x Rx Rxy ∂z ∂ y Ry Rxy
3. FINITE ELEMENT FORMULATION
A nine-noded curved isoparametric C0 element shown in Figure 2, with seven unknowns per node is used in the
present finite element model. The FE code is developed to take the position and size of cutout. The free
vibration analysis is done to determine the natural frequencies from the following,
([ K ] − ω2n[ M ]) = [0] (3)
13
Figure 2 Nine noded curved isoparametric element with typical node numbering and gauss points
4. RESULTS AND DISCUSSION
4.1. Comparison and Convergence Studies
All analyses are carried out corresponding to the convergence mesh size: 20 x 20 (full shell). Comparison
studies are first made to establish the correctness of the present FE model based on HSDT.
Table 4.1 Frequency parameter of a simply supported, cross ply (0/90), square (a x a) laminate having a
central rectangular cut-out. ( a /b =1, h/a = 0.01, E1/E2 = 25, G12 = G13 = 0.5E2, G23 = 0.2E2 and ν12 = 0.25 )
First four frequency parameters ( )

Cutout size References
1 2 3 4
Present (20 × 20) 9.065 25.438 25.491 38.056

0.2a × 0.2a
Sheikh et. al [18] 9.281 26.231 26.239 38.308
4.2. New results
For all the investigations, unless otherwise mentioned, the geometric properties are taken as: a/b = 1 and the
material properties are: E1/E2 = 25, G12 = G13 = 0.5E2, G23 = 0.2E2 and ν12 = 0.25. Non-dimensional frequency
parameter, = ωa2/h .
4.2.1 Effect of curvature on free vibration frequency of composite spherical shell with cutout
The fundamental frequencies of 0/90/0/90 composite spherical shells with cutout are presented in Tables 4.2 and
4.3 for various curvatures under simply supported boundary condition with h/ a= 0.01 and 0.1 respectively.
From Table 4.2, it is observed that in 0/90/0/90 composite spherical shells with cutout frequency
parameter decreases as the curvature increases. The frequency parameter increases with increase in
mode number. Cutout frequency parameter depends on the size of cutout.
14
Table 4.2 Frequency parameter of a simply-supported, cross ply (0/90/0/90), spherical shell having a central
rectangular cutout, h/a=0.01
First five mode numbers

Cutout size R/a
1 2 3 4 5
20 14.585 37.978 38.131 55.562 81.051
0.2a × 0.2a
100 12.817 37.147 37.251 55.263 80.603
20 14.283 29.116 29.680 50.871 62.981
0.4a × 0.4a
100 12.000 27.975 28.470 50.650 62.925
20 16.000 24.707 25.726 43.571 46.087
0.6a × 0.6a
100 13.240 23.221 24.130 43.599 45.687
20 13.876 30.632 36.647 53.822 76.953
0.4a × 0.2a
100 12.004 29.643 35.714 53.586 76.662
20 12.782 15.440 37.903 42.074 71.105
0.8a × 0.4a
100 11.754 14.362 37.293 41.845 70.532
20 14.264 34.759 37.158 54.933 79.161
0.6a × 0.2a
100 12.408 33.859 36.242 54.670 78.795
Table 4.3 Frequency parameter of a simply-supported, cross ply (0/90/0/90), spherical shell having a central

Cutout size R/a
1 2 3 4 5
20 10.965 22.991 23.010 23.379 23.536
0.2a × 0.2a
100 10.948 22.992 23.009 23.382 23.514
20 10.443 17.979 18.400 25.054 25.143
0.4a × 0.4a
100 10.419 17.980 18.370 25.055 25.143
20 11.439 15.648 16.352 22.344 25.259
0.6a × 0.6a
100 11.409 15.639 16.327 22.435 25.159
20 10.333 17.763 23.391 23.434 23.910
0.4a × 0.2a
100 10.313 17.768 23.365 23.434 23.913
20 10.036 10.693 21.643 21.779 25.329
0.8a × 0.4a
100 10.016 10.690 10.691 21.676 21.791
20 10.657 20.568 23.247 23.296 23.465
0.6a × 0.2a
100 10.638 20.573 23.221 23.296 23.468
From Table 4.3, it is observed that in 0/90/0/90 composite spherical shells with cutout frequency parameter
15
decreases as the curvature increases, but this decrease is lesser compared to h/a = 0.01. The frequency
parameter increases with increase in mode number. Cutout frequency parameter depends on the size of
cutout.
From Tables 4.2-4.3, for same cutout frequency parameter is more for h/a = 0.01compared to h/a = 0.1
4.2.2 Effect of boundary condition on free vibration frequency of composite spherical shell with cutout
The fundamental frequencies of 0/90/0/90 composite spherical shells with cutout are presented in Tables 4.4 and
4.5 for various curvatures under clamped boundary condition with h/a = 0.01 and 0.1 respectively.
Table 4.4 Frequency parameter of a clamped, cross ply (0/90/0/90), spherical shell having a central

Cutout size R/a
1 2 3 4 5
20 36.044 59.100 59.589 83.403 108.985

0.2a × 0.2a
100 30.878 57.243 57.605 83.190 108.419
20 44.889 53.430 55.035 77.801 82.008

0.4a × 0.4a
100 41.737 51.528 53.019 77.676 81.352
20 83.925 84.855 89.991 95.407 100.507

0.6a × 0.6a
100 82.550 83.706 88.777 94.488 100.293
20 36.538 46.853 60.248 79.527 102.177

0.4a × 0.2a
100 32.507 44.622 58.628 79.399 101.556
20 45.444 45.448 72.422 72.443 123.183

0.8a × 0.4a
100 43.206 43.210 72.306 72.319 122.979
20 36.325 53.132 59.048 81.989 105.330

0.6a × 0.2a
100 31.716 51.121 57.158 81.830 104.829
decreases as the curvature increases. The frequency parameter increases with increase in mode number.
Cutout frequency parameter depends on the size of cutout.
decreases as the curvature increases, but this decrease is lesser compared to h/a = 0.01. The frequency
parameter increases with increase in mode number. Cutout frequency parameter depends on the size of
cutout.
From Tables 4.4-4.5, for same cutout frequency parameter is more for h/a = 0.01compared to h/ a= 0.1
From Tables (4.2-4.5), it is observed that frequency parameter for the same cutout size is more in clamped
case than that in simply supported case.
16
Table 4.4 Frequency parameter of a clamped, cross ply (0/90/0/90), spherical shell having a central

Cutout size R/a
1 2 3 4 5
20 19.794 28.599 28.800 39.766 45.184

0.2a × 0.2a
100 19.697 28.570 28.751 39.767 45.172
20 25.247 28.510 29.095 37.176 38.572

0.4a × 0.4a
100 25.186 28.478 29.051 37.162 38.574
20 39.353 39.949 41.379 43.335 46.196

0.6a × 0.6a
100 39.321 39.924 41.350 43.314 46.194
20 20.810 24.306 30.394 38.310 42.525

0.4a × 0.2a
100 20.738 24.268 30.353 38.307 42.505
20 26.071 26.074 36.573 36.587 51.405

0.8a × 0.4a
100 26.028 26.031 36.564 36.578 51.390
20 20.379 26.376 29.281 39.255 43.492

0.6a × 0.2a
100 20.295 26.343 29.235 39.254 43.479
5. CONCLUSIONS
In the present study, the free vibration analysis based on higher order theory of laminated composite spherical
shells with cutout has been studied using a C0 finite element model by incorporating all three radii of curvatures
developed by the authors. Cutout frequency parameter of composite cylinrical shell decreases as the curvature
increases. The frequency parameter increases with increase in mode number. Frequency parameter for the
same cutout size is more in clamped case than that in simply supported case. Many new results are generated
which should serve as benchmark for future research. As such the present model may be recommended for wide
applications to solve many engineering problems related to vibration of composite shells with cutout.
REFERENCES
[1] Chakravorty, D.; Sinha, P. K. and Bandyopadhyay, J. N. (1998). Applications of FEM on free and forced
vibration of laminated shells. Journal of Engineering Mechanics, 124(1), 1–8.
[2] Reddy, J. N. (1982). Large amplitude flexural vibration of layered composite plates with cutouts. Journal of
Sound and Vibration, 83(1), 1–10.
[3] Malhotra, S.K.; Ganesan, N. and Veluswami, M.A. (1989). Vibration of composite plates with cut-outs.
Journal of Aeronautical Society of India, 41, 61–64.
[4] Sivasubramonian, B.; Kulkarni, A.; Venkateswara Rao, M.G. and Krishnan, A. (1997). Free vibration of
curved panels with cutouts. Journal of Sound and Vibration, 200(2), 227–234.
17
[5] Sivakumar, K.; Iyengar, N.G.R. and Deb, K. (1999). Free vibration of laminated composite plates with
cutout. Journal of Sound and Vibration, 221(3), 443–465.
[6] Rossi, R.E. (1999). Transverse vibrations of thin, orthotropic rectangular plates with rectangular cutouts
with fixed boundaries. . Journal of Sound and Vibration, 221(4), 733–736.
[7] Huang, M. and Sakiyama, T. (1999). Free vibration analysis of rectangular plates with variously-shaped
holes. Journal of Sound and Vibration, 226(4), 769–786.
[8] Hota, S.S. and Padhi, P. (2007). Vibration of plates with arbitrary shapes of cutouts. . Journal of Sound
and Vibration, 302(4-5), 1030–1036.
[9] Sivasubramonian, B.; Rao, G.V. and Krishnan, A. (1999). Free vibration of longitudinally stiffened curved
panels with cutout. Journal of Sound and Vibration, 226(1), 41–55.
[10] Hota, S.S. and Chakravorty, D. (2007). Free vibration of stiffened conoidal shell roofs with cutouts.
Journal of Vibration and Control, 13(3), 221–240.
[11] Nanda, N. and Bandyopadhyay, J.N. (2007). Nonlinear free vibration analysis of laminated composite
cylindrical shells with cutouts. Journal of Reinforced Plastics and Composites, 26(14), 1413–1427.
[12] Qatu, M.S. (2004). Vibration of Laminated Shells and Plates, Elsevier, London, UK.
[13] Vasiliev, V.V.; Jones, R. M. and Man, L. I. (1993). Mechanics of Composite Structures, Taylor & Francis,
Boca Raton, Fla, USA.
[14] Dey, A.; Bandyopadhyay, J.N. and Sinha, P.K. (1992). Finite element analysis of laminated composite
paraboloid of revolution shells. Computers and Structures, 44(3), 675–682.
[15] Cook, R.D.; Malkus, D.S. and Plesha, M.E. (1989). Concepts and Applications of Finite Element Analysis,
John Wiley & Sons, New York, NY, USA.
[16] Sahoo, S. and Chakravorty, D. (2005). Finite element vibration characteristics of composite hypar shallow
shells with various edge supports. Journal of Vibration and Control, 11(10), 1291–1309.
[17] Reddy, J.N. and Liu, C.F. (1985). A higher-order shear deformation theory of laminated elastic shells.
International Journal of Engineering and Science, 23,319-330.
[18] Sheikh, A.H.; Haldar, S.; Sengupta, D. (2004). Free flexural vibration of composite plates in different
situations using a high precision triangular element. Journal of Vibration and Control, 10, 371–386.
18
Paper No.: 004
NONLINEAR RESPONSE OF LAMINATED COMPOSITE STIFFENED

PLATES
S.N.Patel
Dept. of Civil Engineering, BITS Pilani, Pilani Campus, Pilani-333031, (Raj), India
shuvendu@pilani.bits-pilani.ac.in
ABSTRACT
This paper deals with the geometric nonlinear bending and stress response of laminated composite stiffened
plates subjected to uniform transverse loading. The eight-noded degenerated shell element and three-noded
degenerated curved beam element with five degrees of freedom per node is adopted in the present analysis to
model the plate and stiffeners respectively. The Green-Lagrange strain displacement relationship is adopted to
formulate the matrices. The total Lagrangian approach is taken in the formulation. The results by this method
are compared with the available results along with the results obtained from finite element software ABAQUS
and the conclusions are made. The nonlinear response of deflection and stress are reported.
KEYWORDS: Degenerated shell element, Degenerated curved beam element, Nonlinear analysis, Green-
Lagrange nonlinearity, Stiffened plate and Laminated composite.
1. INTRODUCTION
With the increased application of fiber reinforced composites in various fields, research on their behaviour for
different structural form has also increased. Most commonly used structural forms are plates, used in aircraft,
ship, automotive and many other industries. The performance, i.e. strength/stiffness to weight ratio of the plates
is enhanced by adopting suitable stiffened forms. The literature dealing with the flexural behaviour of laminated
composite stiffened plates is few and mostly restricted to small deflection theory. While analyzing or designing
a laminated composite stiffened plate with small deflection theory(neglecting the nonlinear strain-deformation
relationship) one seldom uses the full capacity of the material. However, one can see by considering the
nonlinear strain-deformation relationship, the structural component is capable of taking more load before failure
than that is found out by not considering the nonlinearity. So an appropriate design can be made by considering
the geometric nonlinearity to use the material economically. At higher loads the transverse deflection of plate is
large compared to its thickness. The bending and stretching coupling comes into play and the load deflection
behaviour becomes non-linear. At large deflection level, membrane stresses are produced which give additional
stiffness to the structure. So a large deflection analysis provides accurate responses. For composite plates,
coupling effects arising from ply orientation makes the matter more complicated.
The large deflection analysis of un-stiffened composite plates has been extensively studied in the recent years
(Chia 1988). The large deflection analysis of isotropic stiffened plates are reported by some
authors(Mukhopadhyay and Satsangi (1984), Liao and Reddy (1990), Koko and Olson (1991), Rao et al.
(1993), Sheikh and Mukhopadhyay (2000) and Ojeda et al.(2007))
The available literatures for nonlinear bending analysis of laminated composite stiffened plates are few. Turvey
(1983) has analyzed the uniformly loaded ring-stiffened composite stiffened plates for large deflection. In his
modeling, the effect of stiffener in the plate has been incorporated by considering statically equivalent local
body forces acting on the plate. The large deflection equations have been solved by finite difference scheme.
Liao and Reddy(1990) have investigated the large deflection behaviour of composite stiffened plates and shells
by the finite element method. They have used degenerated three dimensional shell elements and associated
curved beam elements which have been derived from the degenerated elements by imposing appropriate
kinematic constraints. The incremental equations of motion developed using the principle of virtual

displacement of a continuum and the total Lagrangian concept have been solved by the Newton-Raphson
iteration procedure. The experimental investigation on the flexural behaviour of composite stiffened plate in the
nonlinear range has been carried out by Hyer et al. (1990). They have used the STAGS (Almroth and Brogan
1978) computer code to compare the experimental results with the analytical ones. STAGS models the stiffened
plate as shell branches. Chattopadhyay et al. (1995) have performed the large deflections analysis of laminated
composite stiffened plates using an eight noded isoparametric element. The element formulation is based on
Reissner–Mindlin’s hypothesis with a total Lagrangian description of motion. The nonlinear equilibrium
equations are solved by the Newton-Raphson iteration procedure. Recently, Ojeda et al. (2007) have carried out
the large deflection finite element analysis of isotropic and composite plates with arbitrary orientated stiffeners.
In the present analysis, the nonlinear bending analysis of laminated stiffened plate subjected to uniform
transverse loading is carried out. The nonlinear response of deflection and stress are reported. The eight-noded
degenerated shell element and three-noded degenerated curved beam element with five degrees of freedom per
node is adopted in the present analysis to model the plate and stiffeners respectively. The Green-Lagrange strain
displacement relationship is adopted and the total Lagrangian approach is taken in the formulation. The
resulting nonlinear equations are solved by the Newton-Raphson iteration technique along with the incremental
method. The results by this method are compared with the available results along with the results obtained from
finite element software ABAQUS and the conclusions are made.
2. FORMULATION
The plate skin and the stiffeners are modeled discretely. The plate is modeled with Ahmad et al.’s (1970)
degenerated shell element with Green-Lagrange strain displacement relationship. The element contains five
degree of freedom per node, θz is neglected. Shear correction factor of 5/6 is adopted in the stress-strain
relationship for transverse shear stresses. A three noded degenerated curved beam element with five degree of
freedom per node is taken to model the stiffeners. As the in-plane rigidity of the plate is very high the sixth
degree of freedom (θz) in the beam element is also neglected, as the bending of stiffener in the in-plain direction
of the plate skin is insignificant. To consider the torsional rigidity of the beam adequately, a torsional correction
factor is introduced in the formulation of the stiffener. The formulation of stiffener element is also based on the
formulation of Ahmad et al.’s (1970) degenerated shell element. The Green-Lagrange strain displacement
relationship is adopted in this case also.
2.1 Shell Element
The formulation of the shell element is based on the basic concept of Ahmad et al. (1970), where the three-
dimensional solid element used to model the shell is degenerated with the help of certain extractions obtained
from the consideration that the dimension across the shell thickness is sufficiently small compared to other
dimensions. The detail derivation of this element for isotropic case is available in the literature (Ahmad et al.
1970; Zienkiewicz 1977; Rao 1999).
The element has a quadrilateral shape having eight nodes as shown in Fig.1a where the external top and bottom
surfaces of the element are curved with linear variation across the shell thickness. Figure 1b shows the global
Cartesian and local co-ordinate system at any node i. The geometry of the element can be nicely represented by
the natural coordinate system (ξ, η and ζ) where the curvilinear coordinates (ξ-η) are in the shell mid-surface
while ζ is linear coordinate in the thickness direction. According to the isoparametric formulation, these
coordinates (ξ, η and ζ) will vary from -1 to +1 on the respective faces of the element. The relationship (Eqn.1)
between the global Cartesian coordinates (x, y and z) at any point of the shell element with the curvilinear
coordinates holds good. This is the geometry of an element, which is described by the coordinates of a set of
points taken at the top and bottom surfaces, where the line joining a pair of points (itop and ibottom) is along the
thickness direction i.e., normal to the mid-surface at the ith node point. The line joining the top and bottom
points are the normal vector (V3i) at the nodal point i.
20

∑ , ∑ , (1)
Where Ni are the quadratic serendipity shape functions in (ξ-η) plane of the two-dimensional element.
Eqn. (1) may be rewritten in terms of mid-surface nodal coordinates with the help of unit nodal vectors (v3i)
along the thickness direction as,
∑ , ∑ , (2)
Where, l3i, m3i and n3i are direction cosines of the nodal vector (V3i), i.e. components of unit nodal vectors (v3i),
v3i is the unit vector along (V3i) direction, hi is the thickness at node i.
4
8
ς 7
η
1
, z'
ξ
3
5
6
Fig.1a Eight-noded quadrilateral degenerated shell element in curvilinear co-ordinates.
z, w
y, v
θ
, ‐ x, y'
,z'
θ
, y, x'
x, u
Fig.1b Global Cartesian co-ordinate (x, y and z) and local co-ordinate system at any node i.
Two orthogonal tangential vectors V2i and V1i are formed at the node i which are normal to V3i vector. The two
tangential vectors V2i and V1i not necessarily follow ξ and η directions. The unit vectors along V2i and V1i
directions are v2i and v1i. The local co-ordinates x', y' and z' are directed along V1, V2 and V3 directions
respectively. The directions cosines of x',y' and z' and V1, V2 and V3 are same as the components of unit vectors
v1, v2 and v3. The displacement u,v and w are along the global coordinates x,y and z directions. Similarly the
local displacement components u',v' and w' are along the local coordinates x',y' and z' directions. The rotations
21

of the mid surface normal θx and θy are taken about the local coordinates y' and x' or v2 and v1 directions
respectively.
The displacement field(Eqn.3) of a point within the element can be defined with the help of three mid surface
nodal translational displacement(ui, vi and wi) along the global Cartesian co-ordinates directions and two
rotational components θxi and θyi about the local coordinates y' and x' directions.
∑ , ∑ , (3)
Where, l1i,m1i and n1i are direction cosines of the nodal vector(V1i), i.e components of v1i, l2i,m2i and n2i are
direction cosines of the nodal vector(V2i), i.e components of v2i, and {d} is nodal displacement vector,
{d} =[ u1 v1 w1 θx1 θy1 u2 v2 ---- ----- θx8 θy8 ]T (4)
The strain displacement relationship with Green-Lagrange strain of the element in local co-ordinate system(x'-
y'-z') can be expressed as,
(5)
(6)
After performing number of operations using equation (2) and (3) we can write,
(7)
(8)
′ ′
Where, and are strain-displacement matrices with respect to linear and nonlinear strain components
respectively in local co-ordinate system(x'-y'-z'). The normal strain along direction is neglected.
Knowing, the stress-strain relationship of the laminated composite material in each layer in its material axis
system (1-2-3), the stress-strain relationship in the local co-ordinate systems(x'-y'-z') can be found out by simple
transformation. Here material axis 3 is directed along z' direction. The material axes 1-2 lie in x'-y' plane but it
can be oriented at some angle θ. After transformation the stress-strain relationship in the local co-ordinate
systems can be written as,
(9)
After finding ′ , ′ and matrices the secant stiffness matrix can be expressed as,
(10)
This secant stiffness matrix is not symmetric. To efficiently use the storage scheme which is used in linear
analysis, this non symmetric scant stiffness matrix can be made symmetric (Wood and Schrefler 1978) as,
(11)
Where, matrix is stress-strain matrix in local co-ordinate system, and and are stress matrix in local
co-ordinate system for linear and nonlinear parts of the strain respectively.
22

The tangent stiffness matrix, which is used in the nonlinear solution of the equilibrium equation by Newton-
Raphson method, can be written as,
(12)
2.2. Stiffener Element

The derivation of the stiffener element (Fig.2) is based on the basic concept used to derive the shell element. In
this case the stiffener element modeled with three dimensional solid element is degenerated with the help of
certain extractions obtained from the consideration that the dimension across stiffener depth as well as breadth
is small compared to that along the length.
η
ζ
ξ
bs
ds 3
2
e 1 Reference axis (Shell
element nodal line)
Fig.2 Degenerated curved beam element
The stiffener element follows an edge of a shell element where the parameters of three nodes lying on that shell
element edge are used to express the geometry and deformation of the stiffener utilizing compatibility between
shell and stiffeners. It helps to eliminate the involvement of additional degrees of freedom for the modeling of
stiffeners. The stiffener element having any arbitrary curved geometry is mapped into a regular domain in ξ-η-ζ
co-ordinate system where all these coordinates vary from –1 to +1. Again ξ is taken along the stiffener axis
while η and ζ are taken along the width and depth directions respectively. It has been found that the vectors ,
and are quite useful for the representation of geometry and deformation of the stiffener element. For the
stiffener element a similar set of vectors , and are used and these may be obtained from those of the
shell element ( , and ) as
, and (13)
where ( - ) is oriented at an angle of with respect to ( - ) and follows the stiffener axis (Fig.3).
θs
Fig.3 Orientation of stiffener local axis with respect to shell local axis (Plan View)
The direction of the vectors , and are the direction of the local co-ordinate system x',y' and z'
directions respectively for the stiffeners, which are used in the stiffener element's stain-deformation and stress-
23

strain relationships. The component of these vector are direction cosines of x',y' and z' lines with respect to the
global (x-y-z) co-ordinate. , and are direction cosine of x' axis for stiffener, those are same as the
component of unit vector , , and are direction cosine of y' axis for stiffener, those are same as the
component of unit vector and , and are direction cosine of z' axis for stiffener, those are same as
the component of unit vector .
With these vectors, the co-ordinates of any point within the stiffener may be expressed in terms of co-ordinates
(xi, yi, zi) of those three nodes of the corresponding shell element edge as,
∑ ∑ ∑ (14)
Where bs is stiffener width, ds is its depth and e is the eccentricity (distance of the stiffener axis from the shell
mid surface, taken positive when the stiffener is attached to the top of the plate and taken negative when the
stiffener is attached to the bottom), are the expressions of the quadratic shape functions along ξ-direction.
Now considering the deformation of the stiffener element, the present formulation differs from the usual one
(Bathe 1996, Ferguson and Clark 1979 and Liao and Reddy 1990) where six degrees of freedom are generally
taken to represent the biaxial bending apart from torsion and axial deformation. In the present study the bending
of the stiffener in the tangential plane of the shell is not considered. This has helped to eliminate the
involvement the sixth degrees of freedom θz like that of shell element. Moreover the usual formulation (Bathe
1996, Ferguson and Clark 1979 and Liao and Reddy 1990) overestimates the torsional rigidity and it cannot be
corrected simply with some correction factor since it got mixed with other terms. Moreover in these cases the
warping displacement in the stiffener element is neglected. However, the present formulation facilitates to treat
it nicely where a torsion correction factor is introduced for parallel as well as perpendicular stacking
schemes(Fig.4a and Fig.4b). Actually this is the primary object for the reformulation of the stiffener element.
Based on this the displacement components at any point within the stiffener may be expressed as
∑ ∑ (15)
Where, and are rotations about local y' and x' axes (i.e. and ) of the stiffener node respectively.
and can be written as,
and (16)
Using Eqn.(16) in Eqn.(15), the displacement components at any point within the stiffener can be expressed
with the help of unit vectors and rotations at shell mid surface node as,
∑ ∑ (17)
Where, is the displacement vector of stiffener element, which is same as the displacement vector of the
three nodes of the shell element coming in line with the stiffener element,
The matrix [Tvi] is used to make the component of translational displacement along at shell mid-plane zero,
since the bending of the stiffener in the tangential plane of the shell is not considered. Its effect should be
24

insignificant since bending deformation in this mode will be very small due to high in-plane rigidity of the shell
skin. Moreover the flexural rigidity of stiffener in this mode is usually found to be small.
The matrix [Tvi] used in the above equation may be expressed with the help of , as,
0 0 0 (18)
The strain displacement relationship with Green-Lagrange strain of the stiffener element in local co-ordinate
system(x'-y'-z') of the stiffener can be expressed as,
(19)
(20)
(21)
The strains and are neglected in the stiffener element.
Similar to the shell element, the stress and strain components at any point within the stiffener element are taken
in a local axis system ( x ′ - y ′ - z ′ ) corresponding to , and . The relationship between them may be
expressed as,
0 0
0 0 or (22)
0 0
Where, βs is the shear correction factor taken as 5/6. The torsion correction factor βt and other rigidity
parameters in the rigidity matrix in eqn (22) are presented for two different types of stacking arrangements
of the stiffener as shown in Fig. 4a(parallel stacking) and Fig.4b(perpendicular stacking). For both the
arrangements, the fibers are oriented with respect to x' axis. The material axis system is 1-2-3. Axes 1-2 is
oriented with x'-r by angle θ, while direction 3 is along the direction of s . In both the arrangements, the stress-
strain relationship of a lamina in its axis system ( x ′ - r - s) as shown in Fig. 4a and Fig.4b, after transforming
from their material axis system(1-2-3) can be written as,
0 0
0 0
0 0 (23)
0 0 0
0 0 0
In ( x ′ - r - s) axes system for Ply arrangement-I, r corresponds to y' and s corresponds to z', while in Ply
arrangement-II, s corresponds to y' and r corresponds to z', of the local axis system(x'-y'-z'). The orientation of x'
of (x'-r-s) axis system is same as the orientation of x' of (x'-y'-z') in both cases of arrangement. So the rigidity
parameters ′ in Eqn.(22) is found out separately for both arrangements of ply with respect to ( x ′ - r - s) axes
25

system. The local axes are required to be found out in each node and integration point to evaluate the strain-
displacement matrices. In a particular layer ′ remains constant. It will change from layer to layer. Even if
two layers have same ply orientation, the torsion correction factor βt changes from layer to layer.
1
3
2
x'
Fibers oriented w.r.t x'

i
s x'
z'
y'
nth layer
1st layer
Fig.4a Ply arrangement-I in stiffener (Parallel Stacking)
1
2
x'
3
x'
z'
r
y'
Fibers
oriented
w.r.t x'
axis
nth layer
1st layer
Fig.4b Ply arrangement-II in stiffener (Perpendicular Stacking)
26

The rigidity parameters of equation (22) may be obtained by modifying the equation (23), utilizing the
conditions ( 0) and ( 0).
For ply arrangement-I, the rigidity parameters will be,
(24)
(25)
(26)
∑
∑
(27)
For ply arrangement-II, the rigidity parameters will be,
(28)
(29)
(30)
∑
(31)
∑
Where nls is the number of layers of the stiffener rib and k is the factor to get torsion constant of an isotropic
beam having rectangular section, which is a function of ds/bs ratio of the rectangular section (Timoshenko and
Goodier 1951).
After finding , and matrices, the secant and tangent stiffness matrices of the stiffener element
can be obtained by following the same procedure as done in the shell element.
The secant and tangent stiffness matrices of all elements of the plate and stiffeners are calculated and assembled
properly to form the global secant and tangent stiffness matrices of the structure. The load vector is calculated.
The nonlinear equations are solved by the Newton-Raphson iteration technique along with the incremental
method using Cholesky decomposition method using the skyline storage scheme. The tolerance is defined with
respect to the residual load.
3. RESULTS
The convergence and validation of the present formulation is tested first, considering the systems solved by
previous researchers and all results are reported.
3.1 Two bay rectangular isotropic stiffened plate
An isotropic stiffened plate (1000mm×500mm×3mm) with a central stiffener (depth=18mm, breadth=10mm)

along the shorter direction, attached at the bottom of the plate, clamped on all sides(Fig.5) is considered for the
convergence study and validation purpose of the present formulation. The Young’s modulus of the plate is
71700 N/mm2 and Poisson’s ratio is 0.33. The central deflections of the plate with different mesh size for two
loading intensity are presented in Table.1. It is seen from the table that the results are converging quickly with
the increase in the mesh size. The 8×8 mesh size of the full plate is sufficient to get the converged result.
27

500mm
A
18mm
10mm
1000mm
z y
x
Fig.5 Isotropic stiffened plate
Table.1. Central deflection of the stiffened plate
Mesh size Central deflection in mm

(full plate) Load intensity(0.1 N/mm2) Load intensity(1.0 N/mm2)
2×2 4.766 23.480
4×4 3.418 16.251
6×6 3.284 15.426
8×8 3.268 15.272
10×10 3.258 15.307
The result of this problem with 8×8 mesh of full plate is compared with the results of Sheikh and
Mukhopadhyay (2000) by spline finite strip method, Ojeda et al. (2007) by finite element method, Koko and
Oslan (1991) by super element method as well as by semi-analytic finite strip method and ANSYS solved by
Ojeda et al.(2007) in Fig.6, to further validate the formulation. Sheikh and Mukhopadhyay (2000) and Ojeda et
al. (2007) have used the Green-Lagrange strain-displacement relationship with von Karman’s assumption.
However, in the present formulation, full Green-Lagrange strain-displacement relationship is used. It is seen
from Fig.6 that the results obtained in the present formulation are matching well with others’ results. Moreover,
the stress( ), including both bending and stretching at the centre of the plate at point A(Fig.5) at the top fiber is
calculated for different mesh size for full plate and for quarter plate taking the symmetry condition. The load is
applied on the top surface of the plate(-ve z-direction). The result of the stresses for the loading intensity of
1N/mm2 for different mesh sizes are shown in Table.2. The stresses are calculated properly at the gauss points.
But the point A is not a gauss point. So the calculated stress at this point may not be very accurate. In the
present analysis the stress at this point is calculated directly, not by extrapolating from gauss points. We can
take the result of stress with 9×5 mesh size of the quarter plate as the converged result and further the results are
compared with the results of Sheikh and Mukhopadhyay (2000) by spline finite strip method, Ojeda et
al.(2007), Rao et al.(1993) and Ray and Satsangi (1998) by finite element method in Fig.7. It is observed that
the results are matching well.
28

16
14
Central deflection in mm
12
10
Present
8 Ojeda et. al.(2007)
6 Koko and Oslan(1991)
4 Sheikh and Mukhopadhyay(2000)
ANSYS(Ojeda et.al.2007)
2
0
0 0.2 0.4 0.6 0.8 1
Load intensity in N/mm2
Fig.6. Central deflection of isotropic stiffened plate
Table.2 Stress ( ) at point A at the top extreme fiber of the plate
Full Plate Quarter Plate
Mesh Size in N/mm2 Mesh Size in N/mm2
4×4 225.901 2×2 222.714
6×6 331.095 4×4 419.195
8×8 422.664 6×3 524.192
10×10 485.120 6×6 534.484
12×12 538.626 8×3 605.792
14×14 585.035 8×4 618.361
16×16 623.484 8×5 618.104
8×8 618.271
9×5 650.540
9×9 649.388
29

700
600
500
Stress (N/mm2)
400
Present
300 Ray and Satsangi (1998)
200 Sheikh and Mukhopadhyay (2000)
Ojeda et al.(2007)
100
Rao et al.(1993)
0
0 0.2 0.4 0.6 0.8 1 1.2
Load Intensity (N/mm2)
Fig.7. Stress at point A at the top extreme fiber of the isotropic stiffened plate
3.2 Cross-stiffened laminated composite stiffened plate
A simply supported cross stiffened composite square plate (2438mm × 2438mm × 6.35mm) as shown in Fig.8 is
taken to further validate the formulation. This plate has been analyzed for cross-ply(0/90) and angle-ply(-45/45)
lamination schemes. The stiffeners are attached on both sides of the plate on both x and y directions as shown in
the Fig.8. The lamination scheme in the plates and stiffeners are same. The stiffener depth, ds=6.35mm and
breadth, bs=20mm. The material properties of the plate as well as stiffeners are E1=25E2, E2=7031N/mm2,
G12=G13=0.5E2, G23=0.2E2 and ν12=0.25. Two types of simply supported boundary conditions, BC-1 and BC-2
are considered.
BC-1,
Side-1 and Side-2, u=w=φx (φx is the rotation about y-axis)

Side-3 and Side-4, v=w=φy (φy is the rotation about x-axis)
BC-2,
Side-1 and Side-2, v=w=φx (φx is the rotation about y-axis)

Side-3 and Side-4, u=w=φy (φy is the rotation about x-axis)
In this case also it is seen that, the results are converging quickly with the increase in the mesh size. The 8×8
mesh size of the full plate is sufficient to get the converged result. So in this case, 8×8 mesh of the full plate is
adopted. The central deflections of the plate with BC-1 are presented in Fig.9a and Fig.9b for angle-ply and
cross-ply lamination scheme respectively along with the results obtained from ABAQUS finite element
software. The stiffeners are modeled with shell element in ABAQUS and the required shells are tied up with tie
constraint (surface-surface). The element used in the ABAQUS model is S8R5.
30

layer-2 (90)
Side-2 layer-1 (0)
layer-2 (90)
Side-3 layer-1 (0)

Side-4 layer-2 (90)
layer-1 (0)
[Cross-section with cross-ply]
Side-1 z layer-2 (45)

y layer-1 (-45)
layer-2 (45)
layer-1 (-45)
x layer-2 (45)
layer-1 (-45)
[Cross-section with angle-ply]
Fig.8. crossed-stiffened plate with cross-section at stiffener with two layers
8
7
6
Central Deflection (mm)
5
4
3
Present
2
ABAQUS,Static,General,Nonlinear
1 ABAQUS, Static, Riks, Nonlinear
0
0.0E+00 4.0E-05 8.0E-05 1.2E-04 1.6E-04
2
Load intensity (N/mm )
Fig.9a. Central deflection of the crossed-stiffened plate with (-45/45) lamination scheme (BC-1)
31

12
10
Central Deflection (mm)

8
4 Present
ABAQUS,Static,General,Nonlinear
2
ABAQUS, Static, Riks, Nonlinear
0
0.0E+00 4.0E-05 8.0E-05 1.2E-04
Fig.9b. Central deflection of the crossed-stiffened plate with (0/90) lamination scheme (BC-1)
Results for this example with BC-2 have been previously reported by Chattopadhyay et al. (1995), Liao and
Reddy (1990) and Ojeda et al.(2007). The result of this problem with BC-2 for angle-ply lamination and cross-
ply lamination scheme is compared with the finite element results of Chattopadhyay et al., Liao and Reddy and
Ojeda et al. along with the results obtained from ABAQUS software, in Fig.10a and Fig.10b respectively. The
results are matching well.
6
Central deflection in mm
3 Liao and Reddy (1989)

Chattopadhyay et al. (1995)
2 Ojeda et al.(2007)
Present
1 ABAQUS,Static,General, Nonlinear
ABAQUS, Static,Riks,Nonlinear
0
0.0E+00 2.0E-05 4.0E-05 6.0E-05 8.0E-05 1.0E-04 1.2E-04
Fig.10a. Central deflection of the crossed-stiffened plate with (-45/45) lamination scheme (BC-2)
32

Central deflection in mm 5
3 Liao and Reddy (1989)

Chattopadhyay et al. (1995)
Ojeda et al.(2007)
2 Present
ABAQUS,Static,General, Nonlinear
1 ABAQUS, Static,Riks,Nonlinear
0
0.0E+00 2.0E-05 4.0E-05 6.0E-05 8.0E-05 1.0E-04 1.2E-04
Fig.10b. Central deflection of the crossed-stiffened plate with (0/90) lamination scheme (BC-2)
Again the stress ( ), including both bending and stretching effect, at the centre of the plate, at the bottom fiber
of the skin (not stiffener) is calculated. The load is applied on the top surface of the plate (-ve z-direction). The
variation of this stress with the load intensity is presented in Fig.11. For this, we have taken the (0/90)
lamination scheme for both plate skin and stiffeners, with BC-1 boundary condition. Whole plate is modeled
with 8×8 mesh.
8
7
6
Stress (N/mm2)
5
4
3
2
Variation of Stress
1
0
0.0E+00 4.0E-05 8.0E-05 1.2E-04
Load intensity (N/mm2)
Fig.11. Variation of stress ( ),in the centre of the stiffened plate at the bottom fiber of the skin
33

4. CONCLUSIONS
The conclusions of the present analysis can be summarized as,
1. The nonlinear deflection and stress response of the laminated stiffened plate is presented.
2. The results of the present formulation are matching well with the available results.
3. In the cross-stiffened composite plate the boundary conditions BC-1 and BC-2 has significant effect on
the deflection of the plate in the (0/90) lamination scheme, where as the effect is insignificant in case of
(-45/45) lamination scheme.
4. The present formulation can be extended to other types of laminated composite stiffened plates and
shell
REFERENCES
[1] Chia, C.Y. (1988), "Geometrically nonlinear behaviour of composite plates - a review." Appl. Mech. Rev.,
Vol. 41, 439-50.
[2] Mukhopadhyay, M. and Satsangi, S.K. (1984), "Isoparametric stiffened plate bending element for the
analysis of ships’ structures." Trans. RINA, Vol. 126. 144–151.
[3] Liao, C. L. and Reddy, J. N. (1990), "Analysis of anisotropic stiffened composite laminates using a
continuum-based shell element." Comput. Struct., Vol. 34, 805-8I5.
[4] Koko, T.S. and Olson, M.D. (1991), "Non-linear analysis of stiffened plates using super elements." Int. J.
Numer. Methods Eng., Vol. 31, 319–343.
[5] Rao, D.V., Sheikh, A.H. and Mukhopadhyay, M. (1993), "A finite element large displacement analysis of
stiffened plates." Comput. Struct., Vol. 47, 987–993.
[6] Sheikh, A.H. and Mukhopadhyay, M. (2000), "Geometric nonlinear analysis of stiffened plates by the
spline finite strip method." Comput. Struct., Vol. 76, 765-785.
[7] Ojeda, R., Prusty, B.G., Lawrence, N. and Thomas, G. (2007), "A new approach for the large deflection
finite element analysis of isotropic and composite plates with arbitrary orientated stiffeners." Finite Elem.
Anal. Des., Vol. 43, 989-1002.
[8] Turvey, G.J.(1983), "Axisymmetric elastic large deflection behaviour of stiffened composite plates",
Composite Struct., Vol. 2, 72-88.
[9] Hyer, M. W., Loap, D. C. and Starnes, J. H. (1990), "Stiffener/skin interactions in pressure-loaded
composite panels." AIAA J., Vol. 28, 532-537.
[10] Almroth, B. O. and Brogan, F. A. (1978), "The STAGS Computer Code. NASA CR-2950."
[11] Chattopadhyay, B., Sinha, P.K. and Mukhopadhyay, M. (1995), "Geometrically nonlinear analysis of
composite stiffened plates using finite elements." Composite. Struct., Vol. 31, 107-118.
[12] Ahmad,. S., Irons, B.M. and Zienkiewicz, O.C. (1970), “Analysis of thick and thin shell structures by
curved finite elements.” Int. J. Numer. Methods Eng., Vol. 3, 419-451.
[13] Zienkiewicz, O.C. (1977). “The Finite Element Method.” Tata Mc-Graw Hill Publishing Company
Limited, New Delhi.
34

[14] Rao, J.S. (1999). “Dynamics of Plates.” Narosa Publishing House, New Delhi.
[15] Wood, R. D. and Schrefler, B. (1978), “Geometrically nonlinear analysis-a correlation of finite element
methods.” Int. J. Numer. Methods Eng., Vol. 12(4), 635–642.
[16] Bathe, K.J. (1996), Finite Element Procedure, Prentice-Hall of India Private Limited, New Delhi.
[17] Ferguson, G.H. and Clark, R.D. (1979), “A variable thickness curved beam and shell stiffener with sheat
deformation”, Int. J. Num. Met. Eng., 14, 581-592.
[18] Timoshenko, S.P. and Goodier, J.M. (1951), Theory of elasticity, Tokyo, McGraw-hill, Kogakusha.
[19] Ray, C. and Satsangi, S.K. (1998), “Geometric Nonlinear Static Analysis of Laminated Hat Stiffened
Plates", J. of Reinf. Plast. and Composite, Vol 17(14),1264-1280.
35

Paper No.: 006

REVIVED TECHNIQUE FOR COMPRESSIVE STRENGTH OF CFRP

CONFINED CONCRETE CYLINDERS USING GAUSSIAN PROCESS
REGRESSION
Manav Mittal1 and Pijush Samui2

1
Undergraduate Student, School of Mechanical and Building Sciences, VIT University, Vellore-632014,
Tamil Nadu, India,
2
Professor, (Director-in Charge), Centre for Disaster Mitigation and Management, VIT University,
Vellore-632014, Tamil Nadu, India,
Email: manav.mittal2010@vit.ac.in, pijush.phd@gmail.com
ABSTRACT:
The paper discusses the potential and accuracy of Gaussian Process Regression (GPR) for the prediction of
compressive strength of carbon fiber reinforced fiber (CFRP).Concrete lateral expansion capacity decreases in
presence of sufficient lateral forces. There by increasing its compressive strength under the uniaxial load.
Generally carbon fiber is used for reinforcement in columns for increasing lateral forces and has proved to be
very beneficial. Hence it is very substantial matter to calculate the compressive strength of carbon fiber
reinforced fiber. Basically, GPR is a Bayesian non-parametric regression model. Gaussian process functions
make it possible to conduct exact Bayesian inference by simple matrix manipulations. Process includes
developing a Gaussian process regression for calculating compressive strength followed by its training with 70
percent of the data, the rest 30 percent of the data is used for testing of the program. It was proved that Gaussian
process regression is more agreeable and promising method in terms of accuracy for estimating the compressive
strength of CFRP than previously developed linear genetic programming.
KEYWORDS: CFRP confinement, Linear genetic programming, Formulation, Gaussian process regression,
Concrete compressive strength
1 INTRODUCTION

When sufficient lateral forces are acting on concrete, makes it impossible for it to expand laterally when
uniaxial load is applied; due to this the compressive strength of the concrete increases. Also the ability of
concrete in this case is called confinement. Hence generally this technique is applied in case of columns and has
proved to be very beneficial. There are various techniques which are commonly employed for this confinement,
namely transverse reinforcement in different types like rectangular ties, circular hoops, spirals and many more.
But the most common method is the use of fiber reinforced polymers, which are used for confinement of
concrete columns. If we start to count the advantages of FRPS (fiber reinforced polymers) over the steel we will
have a long list which include uniform confining action to the entire cross section, ease of application, and can
be quickly installed, constant physical conditions of the strengthened elements and also have more resistance to
corrosion than steel. But as day is followed by night, the pros of a thing also come with cons and in this case the
cost. All the advantages of the FRPS increase the cost of construction which is generally 5-6 times compared
with that of steel. The retrofit of reinforced concrete columns with FRP composite jackets has become
increasingly common in regions of high seismicity. The compressive stress-strain behavior of FRP confined
concrete cylinders is essentially nonlinear. The initial portion of the stress-strain response typically follows that
of the unconfined concrete.
After achieving the unconfined concrete strength, the response of the FRP-confined concrete softens, this
softening can occur with either a localized descending branch that may stabilize as the dilation of the concrete
core progresses or it may exhibit a bilinear behavior until the FRP composite jacket fail. The below diagram
shows the typical response of FRP confined cylinder, namely:
1. Normalized axial stress vs. Axial strain

2. Normalized axial stress vs. lateral strain
3. Normalized axial stress vs. Volumetric strain
The unconfined compressive strength is taken as base to normalize stress. The mechanics of confinement is
depends on two factors, the tendency of concrete to dilate and the radial stiffness of the concrete. Out of various
types of fiber reinforced polymers used, carbon fiber reinforced plastic is most widely used owing to it various
advantages over other types of FRPS. The advantages of CRPF as observed by many researchers are anti-
corrosive, easy cutting and construction, as well as high strength to weight ratio and high elastic modulus. These
features have increased the usage of CRPF in strengthening of reinforced concrete structures.
Extensive research has taken place in this field which is the most successful technology in civil engineering
world. Generally these researches are related to effect of CFRP confinement on the strength and deformation
capacity of concrete columns. Due to this, a large number of empirical and theoretical model have come out.
Hence to have more accuracy as demanded by many situations. We propose finding compressive strength of
CFRP confined cylinders using GPR.
Fig. 1. Typical response of FRP- Concrete
2. GAUSSIAN PROCESS REGRESSION
The GPR model is a non-parametric black-box model which searches for the relationship among measured
data. It differs from other black-box classification approaches since it does not try to approximate the modeled
system by fitting constraints of particular basis functions. GPR model uses kernel functions similar to Support
Vector Machines (SVM) and Relevance Vector Machines (RVM) [7, 8]. It gives normally distributed output,
37

articulated in terms of mean and variance. The former value represents the most likely output and the latter can
be inferred as the measure of its confidence. The latter value is dependent on the amount and quality of
available identification data. Let us consider the following dataset:
{(xk, yk)}NK=1 x R y R (1)
Fig. 2. The illustrative figure of CFRP confining concrete cylinder
This study uses D, t, f’f , f’co and PU as input variable and fcc as output variable. So, x = [D, t, f’f , f’co , PU]
and y = [fcc]. The above dataset have been drawn from the following noise process.
yi = f (xi )+ ε , ε ~ N( 0, σ 2) (2)
For a given input x*, GPR defines a gaussian predictive distribution over the output y* with mean
μ = K(x*,x)(K(x, x )+ σ 2 I)−1y (3)
Where K(x,x) is kernel function and I is identity matrix and variance is represented by following equation:
Σ = K( x *, x *) − σ 2 I – K( x * , x)( K (x, x )+ σ 2 I)−1 K (x, x* ) (4) A suitable covariance function and its
parameter are required to develop the GPR model. For a fixed value of gaussian noise, GPR is trained by
maximizing marginal likelihood.
Fig. 3 Geometrical parameters of CFRP concrete cylinder
38

3 GPR FOR PREDICTIO
ON OF COM
MPRESSIVE
E STRENGTH
H OF CFRP
As
A stated aboove, various methods havve been propoosed for the pre-
p diction of compressiive strength oof CFRP.
This
T study em u of Gaussian GPR usees 71 out of 101 cases for CFRP in ordder to train thhe model
mploys the use
while
w rest forr testing the developed model.
m The daata are normallized betweenn 0 and 1 beffore being used in the
model
m as follo
owing:
Dnorm = (D – Dmin)/(Dmax – Dmin) (5)
tnorm = ((t – tmin)/(tmaax – tmin) (6)
f’f(norm)= ( f’f – f’f(min)))/(f’f(max) – f’f(min)) (7)
f’co(norm) = ( f’
f co – f’co(minn))/(f’co(max)) – f’co(min)) (8)
min))/(PU(maxx) – PU(min))
PU(norm)= ( PU – PU(m (9)
Where,
W max and min subsscript shows the maximum
m and minimuum values off input and ouutput data whhile norm
subscript dennotes normalizzed value.
Normalized
N v
values of D, t, f’f , f’co annd PU of trainning dataset are
a given as input in GPR
R model. The value of
the on coefficientt ‘R’ obtainedd from GPR model is 0.9261 Fig.1 shows performaance of testinng dataset
t co-relatio
in R This value is achievedd at a radial basis functioon width ‘σ’’ = 0.04 andd noise ‘ε’ = 3. The
i terms of R.
developed
d model is then tested
t using teesting datasett (see table 1).
Fig 4: Perfoormance of GPR

G Model inn prediction off compressivee strength of CFRP.
C
4 RESULTS AND DISCU

USSIONS
Actual
A fcc values
v for teesting datasett are compared with alrready existingg models inncluding the recently
developed
d G
GPR model (ssee Table 1).. The comparrison has beeen shown in terms of errrors (predicteed values
subtract meaasured values)). The mo deld with thee s ma l l e s t RSME
R can bbe consideredd as the besst model.
From
F Fig. 2, we can infer that the GPRR model prediicts compresssive strength wwith much leess error as compared
c
to
t previouslyy existing moodels. The study
s shows that the GPRG has stroong potentiall to predict accurate
compressive
c strength of CFRP.
C
39
Table 1. Coomparison of errors
e estimatted by GPR and
a other moddels for testingg dataset
5 CONCLU
USIONS
This
T study haas investigateed the use off GPR for preediction of co ompressive strength
s of CF
FRP. The value of R
obtained
o from
m GPR is 0.99261 which is i almost sam me as neural networks. It was found that t result off GPR is
comparable
c t the neural network. Buut GPR is muuch easier to
to o use than neeural networkk as it workss on less
variables.
v Heence, GPR is proved
p to be much better option
o than neural networkk. Also, GPR is applicablee in other
areas
a as well.. It gives goodd prediction values
v will lesss error perceentage.
REFERENC
R CES
[1] Vintzileoou E, Panagiootidou E (200 or predicting the mechaniccal propertiess of FRP

08) An empirrical model fo
confined
c conccrete. Constru
uct Build Matter 22:841–8554.
[2] Xiao Y, Wu H (20000) Compressive behavior of concrete confined

c by carbon fiber composite jaackets. J
Mater
M Civil Eng
E 12(2):139
9–146.
[3] Cevik A,
A Guzelbey IH (2008) NeuralN networrk modeling of strength enhancementt for CFRP confined
concrete
c cylinnders.Build Environ
E 43:7551–763
[4] Norwegiaan Council fo

or Building Standardizatio
S on. Concrete structures deesign rules NS
N 3473. Stocckholm:
Norwegiann Concrete Association;
A 19
992.
[5] Spoelstraa MR, Montii G (1999) FRP-confined

F d concrete model.
m mpos Const 3(3):143–150 strength
J Com
concrete-fuzz
c zy logic. Cem
m Concr Res 2005;
2 35:15311–8.
[7] Matthys S, Taerwe L, L Audenaert K (1999) Teests on axiallly loaded conncrete colum mns confined by fiber
reinforced
r poolymer sheet wrapping. In n: Proceedinggs of the 4thh internationaal symposium
m on fiber reeinforced
polymer
p reinfforcement forr reinforced concrete structtures, pp 217–
–228.
[8] Baykasogglu A, Dereli T,

T Tanıs S (2004) Predictioon of cement strength usinng soft compu
uting techniquues. Cem
Concr
C Res 344(4):2083–209 90
J Priestley MJN, Park R (1988) Theeoretical stresss–strain moddel for confinned concrete. J Struct
[9] Mander JB,
E 114(8):1804–1849
Eng
[10] Lam L, Teng JG (20001) Strength models for ccircular concrrete columns confined by FRP composites. In:
Burgoyne
B CJJ (ed) Proceedings of thhe internationnal conferencce on fibre rreinforced pllastics for reeinforced
concrete
c strucctures. Londoon, pp 835–8444
40
Paper No.: 008
BENDING ANALYSIS OF FUNCTIONALLY GRADED SANDWICH PLATES

1 2
M.N.A. Gulshan Taj and Anupam Chakrabarti
1
Research Scholar, Dept. of Structural Engineering, Indian Institute of Technology, Roorkee, India
2
Associate professor, Dept. of Structural Engineering, Indian Institute of Technology, Roorkee, India
Email: gulshantaj19@yahoo.co.in, anupam1965@yahoo.co.uk
ABSTRACT :
This paper presents a bending analysis of functionally graded sandwich plates using higher order shear
deformation theory. For this purpose, an efficient C0 finite element model is utilized with seven nodal unknowns
for each node of the Lagrangian element. The realistic parabolic variation of transverse shear variation is
incorporated in the kinematics field and thus eliminates the need of shear correction factor. Rule of mixture is
used to ascertain the material properties of the plate along the thickness direction. Thereafter, it is assumed that
the material properties are graded in the thickness direction to presume the continuous and smooth variation by
following a simple power law distribution. Convergence and validation study is performed to test the efficacy
and accuracy of the element. Numerical examples are presented to show the influence of various parameters
such as thickness ratio and volume fraction index on the deflection parameter of the plate. Furthermore,
important conclusions are highlighted concerning the bending of functionally graded sandwich plates which
may serve as benchmark data for designers and scientists involved in the field.
KEYWORDS: Sandwich plates, Functionally graded material, Finite element method, higher order shear
deformation theory, bending analysis.
1. INTRODUCTION
A conventional composite material suffer from the phenomena known as thermal stress concentration and shear
jump at the interfaces and hence demands the new class of materials that can alleviate the related issues,
especially under ultra high temperatures. To address the issue advanced composite materials coined as
functionally graded materials are discovered and introduced to the engineering society in Japan in 1984
(Koizumi, 1993). Typical FGM consists of mixture of two isotropic materials in a predetermined fashion to
achieve the desirable properties. In the process, the superior material properties of each constituent are derived
to obtain the FGM structure. In general ceramic and metal are best known for their application in an FGM
structure. Further, an FGM structure ensures the smooth and continuous variation of material properties in the
preferred direction, usually along the thickness direction. FGM sandwich plates are often found their application
in aerospace, nuclear and other thermal related areas by virtue of their superior thermal properties. Due to the
aforementioned reasons it is vital to analyze the bending response of FGM sandwich plates under mechanical
loading.
Many analytical models were developed based on 3D elasticity solution and utilized to study the effect of FG
core in the sandwich construction having orthotropic face sheets (Anderson, 2002a, 2002b; Anderson, 2003). It
is observed that except for the case of thin face sheets and large span to thickness ratios, the introduction of FG
core will not play any significant role in minimizing the interfacial transverse shear stresses. In addition to
above work, Kashtalyan and Menshykova (2009) developed a 3D elasticity solution to study the effect of FG
core in sandwich panels. For the core portion, exponential variation of Young’s modulus is considered for the
ceramic and metal components. Discontinuity noticed in in-plane normal and shear stresses across layer
interfaces can be considerably reduced by the introduction of FGM core in sandwich arrangement. Further the
author and his associates (Woodward and Kashtalyan, 2011) extended their work to study the influence of
different loading patterns on FGM plate response. Venkataraman and Shankar (2003) presented the elasticity
solution to predict the variation of shear stresses by considering the FGM beam with functionally graded core.
Das et al., (2006) developed a triangular element for the thermo-elastic analysis of functionally graded core in
sandwich type construction. A single FGM soft core with relatively stiff orthotropic face sheets is modeled for
the investigation. Natarajan and Manickam (2012) derived the kinematics field including zig-zag term to
address the slope discontinuities exist at the layer interfaces. In the study, mechanical and thermal loading with
sinusoidal variation is considered to generate the results. A special form of displacement field that incorporate
the classical plate theory (CPT), first order shear deformation theory (FSDT), third order shear deformation
theory (TSDT) and sinusoidal shear deformation theory (SSDT) by appropriate substitution of variables in the
displacement field is proposed by Zenkour (2005) for the analysis of sandwich plate having homogenous core.
A two variable refined plate theory is developed for the analysis of simply supported FGM sandwich plates by
Abdelaziz et al., (2011). It was observed that a proposed refined plate theory can handle the deflection and
stresses as identical as third order shear deformation theory. Neves et al., (2012) performed the static analysis of
functionally graded sandwich plates based on kinematics field that incorporates the hyperbolic sine terms in the
in-plane displacement fields. In addition, the discontinuity of slopes at the layer interface has been taken care by
means of the term called Murakami zig-zag function (Murakami, 1986) in the in-plane response. Recently,
Carrera’s unified formulation has been extended to study the static analysis of functionally graded sandwich
plates based on higher order shear deformation theory (Neves et al., 2013). Tounsi et al., (2013) proposed the
refined trigonometric shear deformation theory that incorporate the bending and shear components in the
transverse displacement field.
From the close observation of the past literature related to FGM sandwich plates, it is manifested by the authors,
that there exists scarcity of results for FGM sandwich plates based on higher order shear deformation theory.
Keeping this in mind, in the present investigation an attempt has been made to present the bending analysis of
FGM sandwich plates based on third order shear deformation theory in conjunction with finite element method.
Formulation is incorporated using nine node Lagrangian element free from shear locking and contains seven
nodal unknowns at each node. Parametric study is performed by considering various geometric properties and
important key conclusions regarding are analysis of FGM sandwich skew plate are derived and submitted.
2. MODELING AND FINITE ELEMENT FORMULATION
2.1. Functionally Graded Sandwich Plate (FGSP)
A solid rectangular plate of length a, width b, thickness h made of functionally graded material is considered in
the present study. Rectangular Cartesian co-ordinate system (x, y and z) is used to define the boundary of the
plate and the mid-plane of the three layer sandwich plate is represented by the thickness co-ordinate z=0. The
top and bottom edges of the plate are defined by z=+h/2 and z=-h/2, respectively. In general, core/face sheet
may be designed to utilize the FGM concept. For numerical analysis, thickness of the layers are represented
from bottom of the plate (z=-h/2). Exponential law (Jin and Batra, 1996), power law (Suresh and Mortensen,
1998) and sigmoid function (Chung and Chi, 2001) are the most widely used methods by the researchers to
estimate effective material properties of the sandwich plate. In the present study, power law (Suresh and
Mortensen, 1998) is used to describe the material property at any height z of the plate, and it is given by
P( z ) = ( Pt − Pb )Vc + Pb (1)
Here, P(z) denotes the material properties, viz. elastic modulus, shear modulus, Poisson’s ratio, density, thermal
expansion, and thermal conductivity of the material at any height z. Pt and Pb represent the material properties at
top (z=h/2) and bottom (z=-h/2) portion of each layer of sandwich plate. In the present Eq.(1) is used to derive
the Young's modulus of the FGM plate. The volume fraction of ceramic Vc may assume different form
depending upon the type of material distribution exists at each layer of the plate.
42
2.1.1 Sandwich plate with FGM skin– Type I
In this case, the plate is modeled in such a way that the core portion is occupied by the homogenous (ceramic)
material, wherein the top and bottom of the plate are represented by metal distribution. Since, we vary the
thickness of the each face in the sandwich plate; the plate may exhibit the symmetric or anti-symmetric
properties with respect to the mid-plane (reference plane). In order to study the effect of thickness variation
among three layers on deflection parameter, a simple notation is used in all the numerical examples considered.
For example, a notation of 1-2-1 (i.e., bottom face-core-top face thickness) is used to represent that the top and
bottom face sheets have same thickness, whereas the core is modeled with twice the bottom/top thickness of the
plate. In such a case the volume fraction Vc present in Eq.(1) assumes the following form of distribution.
n
⎛ z − h0 ⎞
Vc = ⎜
(1)
⎟ ,for layer1
⎝ h1 − h0 ⎠
Vc (2) = 1.0,for layer 2and 3 (2)
n
⎛ z − h3 ⎞
Vc (3) = ⎜ ⎟ ,for layer 3
⎝ h2 − h3 ⎠
Depending on the layer number, the term Vc, appear in Eq. (1) to estimate the effective material property is
replaced by the appropriate expression from Eq. (2). The exponent ‘n’ appears in the expression of volume
fraction Vc is the positive definite parameter which may be termed as volume fraction index, power law
exponent or non-negative number in the literature.
2.1.2 Sandwich plate with FGM core – Type II
In the second case, the bottom and top faces of the plates are represented by metal and ceramic, respectively. By
this option, a sandwich plate with isotropic face sheets and core having smooth and continuous variation of
material properties (FGM distribution) could be modeled. The volume fraction Vc of each layer is defined by the
following expression.
Vc(1) = 0, for layer1

k
⎛1 z ⎞
V c
(2)
= ⎜ + c ⎟ ,for layer 2and 3 (3)
⎝ 2 hc ⎠
Vc(3) = 1.0, for layer 3
2.2. Finite Element Formulation
Based on Reddy’s higher order shear deformation theory, variation of the in plane displacements u, v and
transverse displacement w over the thickness of the plate may be expressed as
4z3
u ( x, y, z ) = u0 + zθ x − γx
3h 2
4 z3
v( x, y , z ) = v0 + zθ y − 2 γ y (4)
3h
w( x, y ) = w0
where u0, v0 and w0 are the displacements in the x, y and z directions of any point in the reference plane; θx, θy
are the bending rotations in y and x directions, respectively; γx and γy are the assumed shear rotations in x, y
directions respectively. The brief explanation for the element description and derivation of governing equation is
43
reported elsewhere (Taj and Chakrabarti, 2013) and not discussed here for the sake of brevity. The final
governing equation for bending analysis based on minimum potential energy principle can be written as
[ K ]{d} ={ p} (5)
3. NUMERICAL RESULTS AND DISCUSSION
A computer program in MATLAB(R2013b) environment is developed based on the higher order shear
deformation theory in conjunction with the finite element formulation. A nine node Lagrangian element with
seven nodal unknowns is adopted to mesh the proposed FGM plate element. The formulation leads to the use of
C0 element which is widely used in most practical applications as compared to C1 elements. To validate the
present formulation Aluminium/Zirconia plate is considered with thickness parameter h=0.1m. The developed
program is capable to handle different types of boundary conditions and combinations of different boundary
conditions. For the sake of conciseness, only the simply supported boundary condition is adopted in the present
investigation. Two cases of FGM sandwich structures are considered: an FGM plate with homogeneous core
having functionally graded layer as skin and functionally graded core with homogenous materials as skin layers.
Based on the thickness of each layer different schemes are used to perform the numerical examples. For
example thickness scheme 1-1-1 (bottom skin layer-core-top skin layer) represents the FGM plate comprises of
three layers of equal thickness. It is to be noted that thickness is considered from the bottom of the plate for all
the thickness schemes assumed.
w
The non-dimensional deflection parameter w = for type I and type II FGM plate based on two types of core
h
materials is furnished under Table 1 and Table 2, respectively. Five types of thickness schemes are chosen to
incorporate the numerical data. Core thickness is increased from left to right wherein the value of power law
exponent (n) is increased from top to bottom in the Table. The value of power law is chosen such that it
represents the case of both homogeneous and FGM plate properties. Increase in power law index from
homogenous case to graded type increases the deflection of the plate at the center. This is due to the fact that as
the n value rises it represents the FGM plate with increased metal content having low stiffness modulus
compared to homogeneous plate. Among the different thickness schemes considered for the study the scheme
corresponds to 2-1-2 ensures maximum defection compared with other thickness schemes for both type I and
type II plates. For type I plates, 2-2-1 thickness scheme ensures minimum deflection while 2-1-1 type of plates
are turned to be better choice when the plate is modeled with FGM core.
Table 1. Deflection of sandwich plate with type I arrangement (simply supported, h=0.1, a/b=1)
n 1-1-1 2-1-2 2-1-1 2-2-1 1-2-1
Ceramic 0.2016 0.2016 0.2016 0.2016 0.2016
1 0.2573 0.2641 0.2043 0.2055 0.2495
2 0.2593 0.2652 0.2063 0.2083 0.2531
5 0.2632 0.2671 0.2098 0.2138 0.2605
10 0.2669 0.2688 0.2128 0.2191 0.2678
50 0.2718 0.2709 0.2168 0.2264 0.2778
Metal 0.2718 0.2709 0.2168 0.2264 0.2779
44
Table 2. Deflection of sandwich plate with type I arrangement (simply supported, h=0.1, a/b=1)
n 1-1-1 2-1-2 2-1-1 2-2-1 1-2-1
Ceramic 0.1895 0.1895 0.1895 0.1895 0.1895
1 0.2824 0.2947 0.2885 0.2751 0.2679
2 0.3224 0.3391 0.2955 0.2857 0.3006
5 0.3606 0.3777 0.3287 0.3173 0.3344
10 0.3746 0.3892 0.3392 0.3297 0.3492
50 0.3869 0.3979 0.3475 0.3402 0.3640
Metal 0.3870 0.3980 0.3477 0.3405 0.3641
Further the influence of thickness ratio (a/h) is studied and shown in Table 1 (a) and (b) for the case of plate
with FGM skins and FGM core, respectively. For both the cases, value of ‘a’ is kept as constant and thickness
is varied from 0.1 to 0.02. As the thickness elevates from 0.1 to 0.02, a sharp rise in displacement component is
manifested and this is tendency is common for all the types of sandwich schemes. Since no core portion exists
for 1-0-1 type of sandwich plate, it records maximum deflection compared to other types. Significant deviation
from other types of schemes is observed for 2-1-1 sandwich plate having FGM as core portion, and the
deviation is increasing as the thickness ratio increases.
Figure 1 Effect of thickness ratio (a/h) on transverse displacement for type I and type II FGM sandwich plates
Table 3. Non-dimensional stresses for sandwich plate with FGM face sheets
Sandwich type σ xx σ yy τyz τxz
1-0-1 5.1685 2.7841 0.2877 0.2877
2-1-2 5.6370 3.0364 0.2674 0.2672
1-1-1 6.0090 3.2361 0.2582 0.2582
2-2-1 6.8141 3.6705 0.2389 0.2381
1-2-1 6.8204 3.6738 0.2491 0.2408
45
Non-dimensional axial and shear stresses are furnished in Table 3 for FGM skin sandwich plate. Isotropic plate
(i.e., 1-0-1) ensures minimum axial and shear stresses compared with other sandwich plates.
4. CONCLUSIONS
Finite element formulation for FGM sandwich plate is presented based on the third order shear deformation
theory. The theory does not require any shear correction term since the realistic variation of shear stresses are
incorporated. Different kinds of sandwich plates are considered to show the bending response of plate under
transverse mechanical load. A plate with thickness scheme 2-1-2 ensures maximum deflection, and it is
irrespective of plate model (type I and type II).When thickness ratio increases the displacement also increases
and response of different sandwich plates shows less deviation. Stresses for FGM sandwich plates are higher
compared to isotropic plate. Increase in volume fraction index increases the metal content with low stiffness and
thus reports maximum displacement for all the sandwich plates.
REFERENCES
[1] Koizumi, M. (1993). The concept of FGM. Ceramic Transaction, 34(1), 3-10.
[2] Anderson, T.A. (2002a). A 3-D elasticity solution for a sandwich composite with a functionally graded core
subjected to transverse loading. Proceedings of the 10th US–Japan Conference on Composite Materials,
Palo Alto, 82–90.
[3] Anderson, T.A. (2002b). Three-dimensional elasticity analysis of sandwich composites with functionally
graded core. Proceedings of the 43rd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and
Materials Conference, Denver, 22–25.
[4] Anderson, T.A. (2003). 3-D elasticity solution for a sandwich composite with functionally graded core
subjected to transverse loading by a rigid sphere. Composite Structures, 60, 265–274.
[5] Kashtalyan, M.; Menshykova, M. (2009). Three-dimensional elasticity solution for sandwich panels with a
functionally graded core. Composite Structures, 87, 36–43.
[6] Woodward and Kashtalyan. (2011). 3D elasticity analysis of sandwich panels with graded core under
distributed and concentrated loadings. International Journal of Mechanical Sciences, 53, 872–885.
[7] Satchi Venkataraman,; Bhavani, V. and Sankar. (2003). Elasticity Solution for Stresses in a Sandwich
Beam with Functionally Graded Core. AIAA Journal, 41 (12), 2501-2512.
[8] Das, M.; Barut, A.; Madenci, E.and Ambur, D.R. (2006). A triangular plate element for thermo-elastic
analysis of sandwich panels with a functionally graded core. International Journal of Numerical Methods
in Engineering, 68, 940–966.
[9] Natarajan, S. and Ganapathi Manickam. (2012). Bending and vibration of functionally graded material
sandwich plates using an accurate theory. Finite Element in Analysis and Design, 57, 32–42.
[10] Zenkour, A.M. (2005). A comprehensive analysis of functionally graded sandwich plates: Part 1-Deflection
and stresses. International Journal of Solids and Structures, 42, 5224–5242.
[11] Hadj Henni ABDELAZIZ; Hassen Ait ATMANE; Ismail MECHAB; Lakhdar BOUMIA; Abdelouahed
TOUNSI; Adda Bedia El ABBAS. (2011). Static Analysis of Functionally Graded Sandwich Plates Using
an Efficient and Simple Refined Theory. Chinese Journal of Aeronautics, 24, 434-448.
46
[12] Neves, A.M.A.; Ferreira, A.J.M.; Carrera, E.; Cinefra, M.; Jorge, R.M.N. and Soares, C.M.M. (2012a).
Static analysis of functionally graded sandwich plates according to a hyperbolic theory considering
Zig-Zag and warping effects. Advances in Engineering Software, 52, 30–43.
[13] Murakami, H. (1986). Laminated composite plate theory with improved in-plane responses. Journal of
Applied Mechanics, 53, 661-666.
[14] Neves, A.M.A..; Ferreira, A.J.M.; Carrera, E.; Cinefra, M.; Roque, C.M.C.; Jorge, R.M.N. and Soares,
C.M.M. (2013). Static, free vibration and buckling analysis of isotropic and sandwich functionally graded
plates using a quasi-3D higher-order shear deformation theory and a meshless technique, Composites: Part
B, 44, 657–674.
[15] Abdelouahed Tounsi; Mohammed Sid Ahmed Houari; Samir Benyoucef; El Abbas Adda Bedia. (2013). A
refined trigonometric shear deformation theory for thermoelastic bending of functionally graded sandwich
plates. Aerospace Science and Technology, 24(1), 209-220.
[16] Jin, Z.H. and Batra, R.C. (1996). Stress intensity relaxation at the top of an edge crack in a functionally
graded material subjected to a thermal shock.. Journal of Thermal Stresses, 19, 317-339.
[17] Suresh, S. and Mortensen A. (1998). Fundamentals of functionally graded materials. First. Ed. IOM
Communications, London.
[18] Chi, S.H. and Chung, Y.L. (2006). Mechanical behavior of functionally graded material plates under
transverse load-Part II: Numerical results. International Journal of Solids and Structures, 43, 3675–3691.
[19] Reddy, J.N. (1987). A generalization of two dimensional theories of laminated composite plates,
Communications in Applied Numerical Methods, 3, 173-180.
[20] Gulshan Taj M.N.A. and Anupam chakrabarti, (2013). Static and dynamic analysis of functionally graded
skew plates. ASCE Journal of Engineering Mechanics, 139(7), 848-857.
47
Paper No.: 009
EFFECT OF LAY-UP SEQUENCES AND IMPERFECTIONS ON THE

DYNAMIC BUCKLING OF CFRP CYLINDRICAL SHELLS
1 2
V. Chitra and R.S. Priyadarsini
1
Postgraduate Student, Dept. of Civil Engineering, College of Engineering, Trivandrum, Kerala, India.
2
Assistant Professor, Dept. of Civil Engineering, College of Engineering, Trivandrum, Kerala, India.
Email: itzchitra.v@gmail.com, priyadarsini_rs@yahoo.com
ABSTRACT :
Composite cylindrical shells especially made of laminated fibre composites are extensively used
in aerospace applications due to their weight sensitive property. Design of composite shells in aerospace
structures need to agree with the loading conditions, both static and dynamic. Under working conditions, axial
compression is predominant in such shell structures. These axial loads lead to instability condition like buckling
which causes severe damage to cylindrical shells under dynamic conditions. Impact loads fall in class of
dynamic loads that act on structure for time duration less than the natural period of the structure. Landing
impact of an aircraft is categorized as a condition where instability condition arises in very short duration. Also,
shell structures are prone to imperfections during erection and/or during working conditions that reduces the
load carrying capacity of the shells significantly. This paper presents the results of the numerical study
conducted to understand the behaviour of laminated composite circular cylindrical shells with and without
imperfections under axial impact. In shells with imperfections, the dynamic buckling loads are smaller than that
of perfect shells. Lesser the imperfection amplitude, greater will be the influence on the rate of reduction the
dynamic buckling loads under axial impact.
KEYWORDS: Axial compression, cylindrical shells, dynamic buckling, impact loads, imperfections,
laminated composites.
1. INTRODUCTION
The use of advanced fibre composite shells in aerospace industry is increasing nowadays mainly due to their
exceptionally high strength to weight ratio, high stiffness to weight ratio etc. when compared to conventional
metallic structures. Among different types of loading, axial compression is predominant in such shell structures.
Under axial compressive loads, these light-weight thin walled shell structures suffers large drop in load carrying
capacities under instabilities like buckling. During their lifetime, there can be situations wherein the short
duration, large amplitude loads governs the design of such thin shell structures. Impact loads belong to the class
of short duration loads possessing large magnitude and occurring within small duration. The sudden application
of thrust from rocket ignition engine is an example for dynamic axial impact on cylindrical shells. Being very
thin, these shell structures are highly imperfection sensitive. They are susceptible to imperfections during
erection and/or during working conditions. Often, there are discrepancies reported between the theoretical and
experimentally determined load carrying capacities of thin cylindrical shells in different literature works. The
reason behind the discrepancy is the presence of geometric imperfections in the shell structure [1,2,3]. The
panels and cylindrical shells made of composites are restricted for use in their post-buckling zone and also the
predicted buckling loads of shell structures are reduced by empirical ‘knock down factors’ that accounts for
imperfection sensitivity. Hence, it is important to study the effect of the imperfections on the load carrying
capacities of shells under static and dynamic loads [2].
The data regarding the actual imperfection associated with the composite cylindrical shells may be difficult to
obtain. In such conditions, the use of “artificial” imperfections in the shell models may be followed for the
analytical studies. Artificial imperfections in the shell models may be assumed as per Koiter’s theory.
According to this theory, the imperfections on the shape of the static buckling mode may be imposed in the
shells. The magnitude of such imperfections may be assumed as normalised with respect to the thickness of the
shell [2]. This kind of imperfections yields the lowest buckling loads and hence are referred to as “worst”
imperfections [4].
The buckling of imperfection-sensitive structures such as cylindrical shells under dynamic loads has been
studied intensively for a few decades where majority of the descriptive works have been reported on the
isotropic shells. Hutchinson and Budiansky [5] carried out dynamic buckling study on long isotropic cylindrical
shells using closed form solution technique. Their work assumed that the shells possessed imperfections in the
form of Eigen modes. The results of the work gave an idea regarding the chances that the dynamic loads may be
lesser than the static loads under short load durations. Kirkpatrick and Holmes [6] analysed the effect of
imperfections on the dynamic buckling of isotropic cylindrical shells assuming modal imperfections and
real-time imperfections in the shell geometry. The shells were subjected to external blast loads and the work was
evaluated both experimentally and analytically. This work showed that the shells simulate the actual dynamic
behaviour when the realistic imperfections are imposed; and the approximate imperfections in the form of single
mode may be assumed in absence of realistic imperfection data for shells. Yaffe and Abramovich [7] studied the
dynamic buckling responses of isotropic stiffened cylindrical shells having Eigen mode imperfections. The
shells were loaded with suddenly applied axial loads having half-sine pulse shape. The work reported that when
the frequency of loading equals the natural frequency of the shells, the dynamic buckling loads were
experiencing a slight decrease. Catellani et al. [8] performed analytical studies on the dynamic buckling
behaviour of simply supported isotropic cylindrical shells having asymmetric and axisymmetric modal
imperfections under sinusoidal axial loads. They reported that the dynamic critical loads were less affected by
the presence of imperfections in the isotropic shells when compared to the static critical loads.
Relatively very few works have been reported to have investigated the imperfection sensitivity of the composite
shells for dynamic buckling conditions. Schokker et al. [9] conducted numerical studies on the dynamic
instability of the unstiffened and stiffened composite cylindrical shells under hydrostatic pressure. Shells made
of isotropic, anisotropic and orthotropic materials were tried and their responses under long duration pulse loads
were investigated. This work identified the occurrence of unbounded responses when the shells attain limiting
conditions of dynamic buckling under lateral pulse loads. Tanov and Tabiei [10] conducted studies on the
behaviour of graphite-epoxy cylindrical shells under suddenly applied lateral pressure using finite element
method. They analysed the behaviour of shells having different lengths and thicknesses, considering small
imperfections in the shells based on the Koiter’s theory. The dynamic critical loads were computed for different
shells using both the equations of motion and the phase-plane approaches. However, these works made use of
long duration pulses for the loading the composite shells. Huyan and Simitses [11] investigated the stability of
boron/ epoxy laminated cylindrical shells having imperfections subjected to dynamic axial loads and bending
moments. The duration of the loading used in the work were both the infinite and finite ones and the effect of
the imperfections were followed based on the Koiter’s theory. The problem was solved numerically using the
finite element approach and the equations of motion approach were followed for calculating the dynamic limit
loads. Eslami et al. [12] performed dynamic buckling studies to investigate the buckling behaviour of
boron/epoxy laminated cylindrical shells loaded with axial compression with the help of general layerwise
theory. Axisymmetric imperfections were assumed for the shell models and different impulse load profiles were
used for loading the shell. These studies on the composite cylindrical shells demonstrated that the dynamic
buckling loads are strongly dependent on the load duration under suddenly applied impulsive loads. Both [11,12]
of these studies accounted for the buckling behaviour of laminated shells having simply supported boundary
conditions which does not occur in the composite shells used in practice.
Buckling studies on the composite cylindrical shells having fixed end conditions subject to axial impulse are
less reported. Bisagni and Zimmermann [13] used the finite element analysis to study the buckling behaviour of
the fibre composite cylindrical shells under impulsive axial loading. This work imposed the imperfections on
the shell models following Koiter’s theory and the static and dynamic buckling behaviour was studied in the
work for composite shells by varying the imperfection magnitudes. Later, Bisagni [14] carried out similar work
on the dynamic buckling behaviour of the fibre composite shells under impulsive axial compression where the
49
imperfection assumed in the model were real time imperfections measured during the quasi-static experimental
study of the shells. The common finding of these works indicates that the dynamic buckling loads, like the static
ones, decreases under the influence of the imperfections in them. Even though, the works reported this
conclusion for the simply supported and clamped boundary conditions in the shells, the laminated composite
shells considered in their study constitute only four numbers of plies. For practical use, the fibre composite
shells will have more number of plies to minimise the coupling effects.
Therefore, the present paper aims to study the buckling behaviour of composite cylindrical shells with and
without imperfections subjected to axial impact. Imperfections were imposed on the cylindrical shells as per
Koiter’s theory. The buckling analysis was carried out for the composite cylindrical shells using the finite
element package, ABAQUS.
2. NUMERICAL MODELLING OF COMPOSITE SHELLS
Laminated composite cylindrical shells made of carbon fibre reinforced plastic (CFRP) having eight plies are
used for the study. The material properties of the laminate are shown in Table 1 and each ply is 0.125 mm thick.
The length of the shell was taken as 400 mm which had a mean diameter of 300 mm with a nominal radius to
thickness ratio, 150 [3]. In this work, the first static buckling mode of the composite cylindrical shells extracted
from the linear static buckling analysis is applied as the imperfection shape. The imperfection amplitude is
expressed in the form of a non-dimensionalised parameter, a/t, where ‘t’ is the shell thickness. In this study, the
imperfection magnitudes (a/t) ranging from 0.05, 0.10, 0.25 and 0.50 were considered.
Table 1. Mechanical properties of the CFRP laminae
Elastic Elastic Shear

Poisson’s
Modulus Modulus Modulus Density
Ratio
E11 E22 G12 (kg/m3)
ν12
(N/mm2) (N/mm2) (N/mm2)
134780 9250 4800 0.286 1700
Three lay-up sequences were considered for the composite cylindrical shells such as [0°/90°/0°/90°]s,
[0°/60°/-60°/0°]s and [75°/-75°/75°/-75°]s, where 0° corresponds to the axial direction of the shell and the
stacking sequence is taken from outside to inside. The lay-up sequences [0°/90°/0°/90°]s, [0°/60°/-60°/0°]s were
chosen because they were reported in different literatures. The lay-up [75°/-75°/75°/-75°]s was reported to show
good static buckling behaviour in terms of the limit load values and axial deformation[15,16]. However, it was
found that this lay-up showed critical dynamic behaviour under ramp loading conditions. Hence the lay-up
[75°/-75°/75°/-75°]s was accounted for this work to study their dynamic buckling behaviour under axial impact.
All the laminates used in the study were symmetric lay-ups in order to avoid coupling properties in the
laminated composite shells.
The complete shell geometry was modelled for the composite cylindrical shells since the number and the shape
of the buckling waves occurring in the longitudinal and transverse directions are not known in advance. The
composite shell models were discretised using four node shell elements with reduced integration designated as
S4R, having six degrees of freedom at each node. These elements are suitable for both the static and dynamic
analyses in ABAQUS. The mesh convergence study for the shell models was performed with the aid of the
linear static (Eigen value) buckling analysis. A model containing a total of 13680 elements (180 elements in the
circumferential direction, 76 elements in the axial direction) was chosen for the further analysis considering the
computational time and accuracy of the results. The geometry of the cylindrical shell with loading is shown in
Figure 1(a).The nodes at the supported end of the shell are fully fixed against translation and rotation and all
movements except the displacement in the axial direction are fixed at the loaded end. The load is applied as
uniformly distributed shell edge load in the form of step pulse as shown in Figure 1(b).
50
(a) Cylindrical shell with axial loading (b) Step pulse used for loading
Figure 1 Model geometry and loading diagram
3. SOLUTION METHODOLOGY
The buckling analysis of composite cylindrical shells was carried out using the finite element software package,
ABAQUS. The buckling analysis of the shell subjected to the static axial compression is performed by linear
(static) analysis and nonlinear static method using ABAQUS/ Standard. A frequency (free vibration) analysis is
also performed in order to determine the natural frequency of the shell and hence the natural period. The
dynamic buckling analysis of the shell under impulsive loading is then carried out by the explicit integration
scheme used in ABAQUS/Explicit. For impact loading, the load of constant magnitude is suddenly applied for
finite duration. Different values of load duration, greater than and less than the natural period of the shell are
considered for the current study. The dynamic buckling loads under the axial impact are calculated using the
equations of motion approach, the Budiansky-Roth criterion. This technique is easily applicable to the
computational methods such as the finite element method. The dynamic buckling loads thus obtained are then
compared with the static buckling loads.
4. STATIC BUCKLING OF COMPOSITE CYLINDRICAL SHELLS
Two different analyses namely linear static and nonlinear static analysis are carried out in this work. Linear static
buckling analysis gives the theoretical buckling strength of shells without considering the effect of imperfections.
The linear buckling strength will be greater than the actual buckling strength since it does not take into account,
the nonlinear behaviour of shells. Hence, a nonlinear static buckling analysis using modified Riks procedure is
also performed to find the exact load carrying capacity of the shell. Nonlinear static buckling analyses are
conducted on shells with and without imperfections. The value of the buckling loads from the linear and nonlinear
static buckling analyses for the shells- [0°/90°/0°/90°]s, [0°/60°/-60°/0°]s and [75°/-75°/75°/-75°]s are reported in
Table 2. The buckling modes of the three different composite cylindrical shells obtained from the linear and
nonlinear static analyses are shown in Figure 2.
Table 2. Results of static buckling analyses
Linear static Nonlinear static load

Lay-up Load (kN)
(kN) a/t = 0 a/t = 0.05 a/t = 0.10 a/t = 0.25 a/t = 0.50
[0°/90°/0°/90°]s 97.24 93.23 83.82 77.35 64.08 55.40
[0°/60°/-60°/0°]s 151.50 142.44 140.22 131.02 101.93 82.82
[75°/-75°/75°/-75°]s 107.80 104.93 99.85 94.70 82.70 68.69
51
Figure 2 Static buckling modes of composite cylindrical shells
The largest static buckling load was found for composite shell having lay-up [0°/60°/-60°/0°]s followed by the
composite shells having lay-ups [75°/-75°/75°/-75°]s and [0°/90°/0°/90°]s. The linear static buckling mode of the
shells with lay-ups [0°/60°/-60°/0°]s and [75°/-75°/75°/-75°]s were spiral shaped while that of the lay-up
[0°/90°/0°/90°]s was diamond shaped. The linear static buckling mode of the shells with lay-ups [0°/90°/0°/90°]s
and [75°/-75°/75°/-75°]s were axi-symmetric ring shaped while that of the lay-up [0°/60°/-60°/0°]s was similar to
diamond shaped with buckled waves directed inwards.
5. DYNAMIC BUCKLING ANALYSIS OF COMPOSITE CYLINDRICAL SHELLS UNDER AXIAL

IMPACT
Dynamic buckling due to impact load of constant magnitude applied for finite duration is investigated in the study.
The Budiansky- Roth criterion is followed in the study where the equations of motion are solved for various
values of the load parameter and thus the systems responses are obtained. The load value at which there exists a
sudden change in the responses is taken as the critical load for that particular duration. In the present study, the
axial impact in the form of step pulse is transferred as suddenly applied uniformly distributed shell edge load
having constant magnitude with finite duration. From the frequency extraction (free vibration) analysis, it was
found that the natural frequency of the composite cylindrical shells is 443 Hz and hence the natural period of the
system is around 2 ms. Based on this, five time durations of pulse loading namely T = 1ms, 2.5 ms, 5 ms, 10 ms
and 15ms were considered. Such a selection of load duration enables to understand the behaviour of shell under
impulsive axial loading for different durations greater than and less than the natural period of shells.
Figure 3 (a) presents the axial displacement of a point on the loaded end of the laminated cylinder [0°/90°/0°/90°]s
without imperfections subjected to suddenly applied axial compression for duration of 15ms for different load
magnitudes. In this case, the curve corresponding to 75 kN show regular response and the cylinder vibrates about
its equilibrium position. From 78 to 80 kN, there is a sudden change between the two responses. This indicates the
dynamic buckling condition and the load average equal to 79 kN is regarded as the dynamic buckling load of the
composite shell with lay-up [0°/90°/0°/90°]s for 15 ms load duration. Figure 3(b) and 3(c) shows a similar plot of
axial displacement versus time for composite shell with layup [0°/60°/-60°/0°]s and [75°/-75°/75°/-75°]s at T =
15ms respectively. Using this criterion to estimate the dynamic buckling loads, dynamic analyses are performed
for different time durations on cylinders with all three layups with and without imperfections and the results are
reported in Table 3.
52
Figure 3 Variation of axial displacement for different load magnitudes corresponding to a load duration of
T = 15ms
Table 3. Dynamic buckling loads for three lay-ups for different time durations
Imperfection
amplitude 0 0.05 0.10 0.25 0.50
Lay-up (a/t)
T (ms) Dynamic buckling loads (kN)
1 205 95 84 72 67
2.5 111 77 74 67 63
[0°/90°/0°/90°]s 5 88 76 73 66 63
10 79 76 73 66 63
15 79 76 73 66 63
1 207 137 133 133 133
2.5 123 109 109 106 99
[0°/60°/-60°/0°]s 5 109 99 97 97 97
10 97 97 97 96 93
15 87 94 97 92 93
1 101 81 79 78 77
2.5 62 62 61 61 58
[75°/-75°/75°/-75°]s 5 58 54 53 49 49
10 51 49 49 48 49
15 51 49 49 48 49
53
6. RESULTS AND DISCUSSIONS
Dynamic buckling of composite shells due to axial impact was studied for shells for three different lay-ups,
[0°/90°/0°/90°]s, [0°/60°/-60°/0°]s and [75°/-75°/75°/-75°]s and for different imperfection amplitudes. Using the
static buckling analysis, the static buckling loads of composite cylindrical shells with and without imperfections
were also found and compared with the dynamic buckling loads for these composite cylindrical shells. Figure 4
shows the variation of the dynamic buckling loads with load durations for the composite shell having lay-up
[0°/60°/-60°/0°]s for different imperfection amplitudes. Generally, the dynamic buckling loads of composite
cylindrical shells were found to decrease with increasing load duration. The same trend was observed for
composite shells for all the lay-ups for various imperfection amplitudes. The reasons behind this trend may be the
increase in the shell stiffness for shorter durations of load. Also, it is believed that the stress wave transmission
between impacted and loaded ends leads to reduction in shell stiffness for longer load durations [11, 12, 14].
(a) a/t = 0 (b) a/t = 0.05
(c) a/t = 0.10 (d) a/t = 0.25
(e) a/t = 0.5
Figure 4 Dynamic buckling loads for composite cylindrical shells- [0°/60°/-60°/0°]s having different imperfection
magnitudes
54
Collectively, the results of the dynamic buckling loads show that the rate of reduction in dynamic buckling
loads is higher for the composite shells having smaller imperfection amplitudes. Also, such a reduction in
dynamic buckling load is observed for load durations less than the natural period of the composite shells. For
longer durations of impact, the ratio between dynamic to static buckling load remains almost the same. Hence it
may be seen that the effect of the imperfections on the dynamic buckling loads under axial impact may be well
assessed using geometric imperfections of smaller amplitude rather than the larger ones.
The dynamic to static buckling load ratios of different composite cylindrical shells for various imperfection
amplitudes are shown in Table 4. In the case of all the lay-ups and for all imperfection amplitudes, the ratio of
dynamic to static buckling load were found to decrease with increasing load duration. For shells with lay-ups,
[0°/90°/0°/90°]s and [0°/60°/-60°/0°]s when the imperfection amplitudes were a/t = 0, 0.05 and 0.10 and for load
duration less than the natural period of the shell, the ratio of dynamic to static buckling loads were large. As the
impact duration increases, the ratio of dynamic to static buckling load decreases. Also for all the three lay-ups,
When the imperfection amplitude was about a/t = 0.25 and 0.5, the ratio of dynamic to static buckling loads were
almost the same, which was close to unity for all load durations. But for the shell having lay-up
[75°/-75°/75°/-75°]s, the ratio of dynamic to static buckling load was less than unity for all values of imperfection
and load duration.
Table 4. Ratio of dynamic to static buckling loads for composite cylindrical shells
Imperfection
amplitude 0 0.05 0.10 0.25 0.50
Lay-up (a/t)
T (ms) Ratio of dynamic to static buckling loads
1 2.20 1.13 1.09 1.12 1.21
2.5 1.19 0.92 0.96 1.05 1.14
[0°/90°/0°/90°]s 5 0.94 0.91 0.94 1.03 1.14
10 0.85 0.91 0.94 1.03 1.14
15 0.85 0.91 0.94 1.03 1.14
1 1.45 0.98 1.02 1.27 1.61
2.5 0.86 0.78 0.83 1.04 1.20
[0°/60°/-60°/0°]s 5 0.77 0.71 0.74 0.95 1.17
10 0.67 0.68 0.74 0.94 1.12
15 0.61 0.67 0.74 0.90 1.12
1 0.96 0.81 0.83 0.94 1.12
2.5 0.59 0.62 0.64 0.74 0.84
[75°/-75°/75°/-75°]s 5 0.55 0.54 0.56 0.59 0.77
10 0.49 0.49 0.52 0.58 0.77
15 0.49 0.49 0.52 0.58 0.77
The composite shell with lay-up [0°/90°/0°/90°]s showed smaller static buckling loads and higher dynamic
buckling loads which were nearly twice the static loads in comparison with other lay-ups. Whereas, the shell
with lay-up [75°/-75°/75°/-75°]s showed satisfactorily good static buckling load but very low dynamic buckling
loads in comparison with other lay-ups. Also, the presence of imperfections affect the shell with lay-up
[75°/-75°/75°/-75°]s greatly than the other lay-ups, with the lay-up [0°/90°/0°/90°]s having less sensitivity. A
55
satisfactory performance was observed for the shell with lay-up [0°/60°/-60°/0°]s in terms of static and dynamic
buckling capacities (Refer Table 3).Therefore, the lay-up sequence used for the composite cylindrical shells
(perfect and imperfect) play an important role in their behaviour under axial impact. For a more general
statement, the study must be extended by considering more number of lay-ups for the composite cylindrical
shells.
7. CONCLUSION
The buckling behaviour under axial impact were investigated for composite cylindrical shells having three
lay-ups viz.[0°/90°/0°/90°]s, [0°/60°/-60°/0°]s and [75°/-75°/75°/-75°]s. The effect of the imperfections in the
composite shells under axial impact was also studied using the finite element method. In nutshell, composite
cylindrical shells experience dynamic buckling loads larger than static buckling loads, if their duration is very
short as compared to the natural period of the system and less than the static buckling load for longer durations.
This means that taking dynamic buckling loads as the design load for dynamic problems might be unsafe under
longer duration pulse loading. The ratio between the dynamic buckling load and the static buckling load is of
practical importance as it gives a direct indication of the load carrying capacity of the shell structures exposed to
rapidly applied loads. The behaviour of the composite shells were found to vary with different lay-up
sequences.Also, when the imperfection amplitude in the composite cylindrical shell increases, the dynamic
buckling load decreases for a particular duration of load. But, the rate of decrease of load is higher for smaller
magnitudes of imperfection. The present work shows that the dynamic buckling loads of composite cylindrical
shells under axial impact conditions are affected by aspects such as the lay-up sequence and the amplitude of
imperfections. Hence the careful selection and control of such aspects may help in achieving satisfactory
efficiency of the composite cylindrical shells under axial impact.
REFERENCES
[1] Huhne C.; Zimmermann R.; Rolfes R., and Geier B. (2002). Sensitivities to geometrical and loading
imperfections on buckling of composite cylindrical shells, Proceedings of the European Conference on
Spacecraft Structures, Materials and Mechanical Testing, France.
[2] Simitses G.J. (1986). Buckling and postbuckling of imperfect cylindrical shells: a review, Applied Mechanics
Review, 39(10), 1517-1524.
[3] Priyadarsini R.S.; Kalyanaraman V., and Srinivasan S.M. (2012). Numerical and experimental study of
buckling of advanced fiber composite cylinders under axial compression, International Journal of Structural
Stability and Dynamics, 12(4), 1-25.
[4] Bisagni C. (2000). Numerical analysis and experimental correlation of composite shell buckling and
post-buckling, Composites: Part B, 31, 655-667.
[5] Hutchinson J.W. and Budiansky B. (1966). Dynamic buckling estimates, AIAA Journal, 4(3), 525-530.
[6] Kirkpatrick S.W. and Holmes B.S. (1989). Effect of initial imperfections on dynamic buckling of shells,
Journal of Engineering Mechanics, 115(5), 1075-1093.
[7] Yaffe R. and Abramovich H. (2003). Dynamic buckling of cylindrical stringer stiffened shells, Computers
and Structures, 81, 1031-1039.
[8] Catellani G.; Pellicano F.; Dall’Asta D. and Amabili M. (2004), Parametric instability of a circular
cylindrical shell with geometric imperfections, Computers and Structures, 82, 2635-2645.
56
[9] Schokker A.; Sridharan S., and Kasagi A. (1996). Dynamic buckling of composite shells, Computers and
Structures, 59(1), 43-53.
[10] Tanov R., and Tabiei A.(1998). Static and dynamic buckling of laminated composite shells, Proceedings of
the fifth LS – DYNA Users Conference, South field.
[11] Huyan X. and Simitses G.J.(1997). Dynamic buckling of imperfect cylindrical shells under axial
compression and bending moment, AIAA Journal, 35(8), 1404-1412.
[12] Eslami M.R.; Shariyat M. and Shakeri M. (1998). Layerwise theory for dynamic buckling and postbuckling
of laminated composite cylindrical shells, AIAA Journal, 36(10), 1874-1882.
[13] Bisagni C. and Zimmermann R. (1998). Buckling of axially compressed fiber composite cylindrical shells
due to impulsive loading, Proceedings of European Conference on Spacecraft Structures, Germany.
[14] Bisagni C. (2005). Dynamic buckling of fiber composite shells under impulsive axial compression, Thin
Walled Structures, 43, 499-514.
[15] Altenbach H. and Becker W. (2003). Modern Trends in Composite Laminate Mechanics, International
Center for Mechanical Sciences, Springer, USA.
[16] Steinmüller P.; Breitsch A.; Degenhardt R. and Rohwer K. (2008). Investigations of axially loaded
unstiffened CFRP cylindrical shells subject to single perturbation loads, Proceeding of 2nd International
Conference on Buckling and Postbuckling Behaviour of Composite Laminated Shell Structures, Germany.
57
Paper No.: 010
EVALUATION OF SHEAR STRENGTH OF RC MEMBERS USING

DIFFERENT METHODS
Suprabhat. Jakati
Engineer, Atkins Bangalore

Email: jakatisuprabhat@gmail.com
ABSTRACT:
The accurate estimation of shear strength is very important to ensure the ductile failure of the RC flexural
members due to sudden and brittle nature shear failure. There are different theories explaining the shear
behaviour of the RC members and the truss analogy concept is the most widely accepted in which the concrete
acts as compression strut and the stirrups take the tension. Indian standard (IS 456:2000) adopts the fixed strut
model in which the angle of compression strut with the horizontal is fixed. Eurocode-2 (EN 1992-1-1-2004)
adopts the variable strut model in which the inclination the compression strut with the horizontal varies between
22° and 45° (cot-1 2.5 to cot-1 1). This paper presents the comparison in the calculation of shear strength of
sections between Indian Code and Euro code
KEYWORDS: Shear Strength, Ductile failure, fixed strut model, variable strut model
1. INTRODUCTION
Figure 1 represents the distribution of principal stress along the span of the homogenous beam. The direction of
principal compressive stresses takes the form of an arch, while the tensile stresses have the curve of the catenary
or suspended chain. Towards mid-span, where the shear stresses are low and the bending stresses are
predominant the direction of stresses tend to be parallel to the axis of the beam. Near the supports where
shearing forces are greater, the principal stresses become inclined, greater the shear force greater is the
inclination angle of the stresses. The inclined tensile stresses near to the supports cause the diagonal cracking of
the concrete. To arrest this cracking appropriate reinforcement should be provided either in form of stirrups or
inclined bars.
The actual behaviour of the reinforced concrete in shear is complex and is difficult to analyse theoretically.
Many simplified procedures are adopted using the experimental results. IS 456:2000 has adopted the fixed strut
model where are the newly introduced Euro code (BS EN 1992-1-1-2004) adopts the variable strut model.
Figure 1: Distribution of principal stress along the length of homogenous beam

2. SHEAR STRENGTH OF SECTIONS
2.1. Concrete sections that do not require design shear reinforcement.
The total shear is transmitted between planes in a reinforced concrete beam by combination of (i) Shear
resistance of uncracked portion of concrete ( Vcz) (ii) Vertical component of the aggregate locking (Vay) (iii)
Dowel force in the tension reinforcement (Vd) (iv) Shear resistance of the shear reinforcement ( Vs). In
sections that do not require shear reinforcement (lightly loaded beams and slabs) the component Vs is
totally absent.
V= Vcz + Vay + Vd + Vs (1)
The expressions to calculate the shear capacity of beam without shear reinforcement given in the Codes are
empirical and are compared in the Table 1.
Table 1- Expression to calculate the shear capacity of beam without shear reinforcement
BS EN 1992-1-1-2004 IS 456:2000
1/3
VRdc = [ CRdc k (100ρ1fck) ] bwd kN τc= 0.85√(0.8fck ) (√(1+5β)-1)/(6β) N/mm2
vRdc = [ CRdc k (100ρ1fck)1/3 ] N/mm2 β= (0.8fck)/(6.89pt) ≤ 1
VRdc min = 0.035 k3/2fck1/2
Where,
VRdc = Shear capacity of the Beam without shear
reinforcement. Where,
CRd,c = Coefficient whose value is 0.12
fck = Concrete cube compressive strength.
k = Size effect coefficient = 1+√( 200/d) ≤ 2
pt= Percentage longitudinal reinforcement
fck = Concrete cylinder compressive strength.
100ρ1= Percentage longitudinal reinforcement
The graphical representation of the above formulae of Table 1 for different beam depths is shown in Figure 2.
From these graphs it can be seen that the depth of the cross section has an inverse relation with the strength of
the concrete section know as the size effect (Kani, 1967), however IS 456:2000 does not reflect the size effect
on the concrete capacity of the section. Also the difference in values of the shear stresses varies between the two
codes from 5 % for a depth of 230 mm to around 30 % at a depth of 1200 mm, the Indian code predicting higher
values.
Also ,If the percentage of longitudinal reinforcement is zero (plain cement concrete), still the concrete has some
strength which the Euro code gives as VRdc min , but IS 456:2000 fails to express the minimum shear strength of
plain cement concrete.
59
(a) Eurocode
(a) Indian Standard

Figure 2: Variation of shear stress with the depth for different percentage steel
2.2. Concrete sections that require design shear reinforcement.
Tie and strut model is a generally accepted model for evaluating the effects of shear reinforcement and its
calculations. In this model, the compression and the tension are carried by the concrete in compression zone and
the tension steel. The concrete strut is assumed to carry the compressive force and the steel reinforcement the
tensile force.
There are two ways in which this truss model is treated, the first by the variation of the strut angle θ and the
second is the distribution of shear force i.e.wheather the truss should carry all the shear force or shear force is
60
carried by both the truss and concrete. IS 456:2000 (BS 8110 as well as ACI 318) adopts the first method in
which the strut angle θ is set to 45° and the total shear is the sum of shear carried by the concrete and the shear
reinforcement. BS EN 1992-1-1 adopts the second method, in which all the shear is assumed to be carried by the
shear reinforcement and the strut angle takes any value between 22° and 45°.
2.2.1 Maximum force in the diagonal strut member
Concrete is strong in compression, with its compression capacity depending upon the grade of concrete, but
there lies an upper limit on the compression capacity. The shear applied must lie within the permissible
compressive stress to avoid the crushing failure of the concrete strut. The expression for computing the
maximum permissible compression capacity of the section in shear is tabulated in Table 2.
Table 2- Expression to calculate the Maximum permissible shear capacity of a concrete section.
BS EN 1992-1-1-2004 IS 456:2000
VRd,max = bwzαcwν1fcd /(cot θ+tan θ) kN τcmax = 0.62√fck N/mm2

vRd,max = αcwν1fcd /(cot θ+tan θ) N/mm2
where, where,
VRd,max = max shear capacity of the section τcmax = the max shear capacity of the section
bw =width of the section fck = the design compressive strength of
z =the lever arm ( 0.9d) concrete
ν1 =the efficiency factor= 0.6(1-fck/250)
fcd =the design compressive strength of
concrete
5.0 4.0
4.0 3.0
VRd,max
3.0
τcmax
2.0 M20
2.0 M20 M25
M25 1.0
1.0 M30
M30
0.0 0.0
20 25 30 35 40 45 20 25 30 35 40 45
Strut angle θ Strut angle θ
Figure 2: Variation of Maximum shear stress capacity with the strut angle for different grades of concrete.
From the above graphs it can be seen that the maximum shear capacity of the concrete strut varies directly
proportional with the strut angle (Fig 3(a)) and there is no unique value, the designer needs to calculate VRd,max
for the chosen value of strut angle unlike where it is constant value for a given concrete grade given by IS
456:2000 (Fig 3(b)).
61
2.2.1 Calculation of vertical shear reinforcement
If the nominal shear stress exceeds the shear strength of concrete but less that the maximum capacity of the strut,
shear reinforcement in the form of stirrups needs to be provided. One major difference lies in distribution of the
shear between concrete and stirrups between BS EN 1992-1-1: 2004 and IS 456:2000. According to the Euro
code all the shear is resisted by the stirrups provided with no direct contribution from the shear capacity of
concrete itself, whereas according to IS 456:2000 the shear to be resisted by the stirrups is the total shear minus
the shear capacity of the concrete section. Table 3 provides a procedure comparison between the 2 codes.
BS EN 1992-1-1-2004 IS 456:2000
1. Calculation of the total shear to be resisted 1. Calculation of the total shear to be resisted
(VEd). (VEd).
2. Calculation of the maximum shear capacity of 2. Calculation of nominal shear stress ( τv=
the section VRd, max for strut angle of 22 ° and Vu/bd)
45°. 3. Calculation of shear capacity of concrete
3. Calculation of strut angle θ using the section depending on the percentage of
expression longitudinal steel provided (τc)
4. Calculation of τc max and if τv> τcmax section
θ = 0.5 sin-1 ( VEd/ VRd,max (45)) to be redesigned.
5. The vertical shear reinforcement is calculated
The value of θ must lies between 22° and 45°. as
If the value is less than 22° then θ= 22° and if
the value is greater than 45° then the stress As/s = ( Vu- Vc)/ (0.87fy d )
exceeds the value of VRd,max and the section
needs to be redesigned.
4. The vertical shear reinforcement is calculated 6. The minimum shear reinforcement is

as calculated as
Asw/s = VEd/ (0.78fyk d cot θ) Asv/bsv = 0.4/0.87fy
5. The minimum shear reinforcement is

calculated as 7. The maximum stirrup spacing is 0.75d ≤ 300mm
Asw min/s = 0.08 fck0.5bw / fyk
6. The maximum stirrup spacing is 0.75d

(1+cotα) ≤ 300mm, α is the inclination of the
shear reinforcement to the longitudinal axis of
the beam.
Another major difference between BS EN 1992-1-1: 2004 and IS 456:2000 is the need of additional longitudinal
reinforcement due to shear. An additional longitudinal force is developed in the tension steel due to the
horizontal component of the force in concrete strut. According to BS EN 1992-1-1 the additional longitudinal
force is given by
ΔFtd = 0.5 VEd cot θ (2)
And the area of reinforcement can be calculated as

ΔAst = 0.5 VEd cot θ/ (0.87fyk) (3)
62
The variation of ΔFtd/VEd with strut angle is shown in figure 3.
1.5
ΔFtd/VEd
1.0
0.5
0.0
20 25 30 35 40 45 50
Strut angle θ
Figure 3: Variation of ΔFtd/VEd with the strut angle.

3. PUNCHING SHEAR OR TWO WAY SHEAR
Punching shear failure is a localised shear failure around a concentrated load on a slab, commonly occurring in
flat slabs, pad foundation and raft foundation. The critical surface for checking the punching shear is a perimeter
known as basic control perimeter. The selection of basic control perimeter is of prime importance as the
perimeter closer to the loaded area has lesser perimeter length and hence more stressed and vice versa. The
shape of the basic control perimeter is also a parameter that is of importance. Figure 4 shows the shape of the
perimeter used by BS EN 1992-1-1:2044 and IS 456:2000. The form proposed by Euro code is much realistic
but the used of rectangular perimeter by IS code is much of practical approach as the shear reinforcement if
need must be provided within the perimeter and providing reinforcement within rectangular grid is easier.

BS EN 1992-1-1-2004 IS 456:2000
Figure 4: Basic control perimeter shapes adopted by BS EN 1992-1-1:2004 and IS 456:2000.
Having chosen the basic form of the perimeter, its location form the loaded area is of importance. Since the
treatment of shear is empirical there is no specific distance for location of the control perimeter. The same
strength could be obtained by using lower shear strength with longer perimeter or with higher shear strength and
shorter perimeter and hence with the shear strength considered within the particular code BS EN 1992-1-1
adopts the basic control perimeter at a distance 2d and IS 456:2000 adopts at a distance d/2.
63
4. CONCLUSIONS
1. The concrete shear strength from IS 456 (τc) is significantly higher that from EC-2 (VRdc). Since the
EC-2 equations are based on latest theoretical developments and experiments, the higher shear strength
predicted by IS 456 may leads to unsafe design for shear. Hence the provisions for concrete shear
strength from IS 456 (τc) have to be rechecked based on latest experimental results to ensure its
accuracy.
2. The size effect of the concrete beams on the concrete shear strength is a well known problem. The
procedure for concrete shear strength, EC-2 accounts for size effect where as IS-456 ignores it.
3. Generally the EC-2 estimates the maximum shear capacity of the compression strut higher than IS-456.
The degree of over estimation increases as grade of concrete increases.
4. The requirement of the additional longitudinal reinforcements due to shear is major difference between
the provisions in EC-2 and IS 456.
ACKNOWLEDGEMENT
I sincerely thank Dr. Anil Kumar M.V (Design Engineer- Atkins, Bangalore) for his constant support and
motivation, without which this paper would not be possible.
REFERENCES
[1] BS EN 1992-1-1:2004- Eurocode 2 -Design of Concrete structures.
[2] IS 456: 2000- Plain and reinforced concrete- Code of Practice
[3] Pillai, U. and Menon, D (2003). Reinforced Concrete Design, Tata McGraw Hill, New Delhi
[4] R.S. Narayanan and A.Beeby (2005). Designers Guide to EN 1992-1-1 and EN 1991-1-2, Thomas Telford
[5] Bill Mosley, John Bungey and Ray Hussel (2007). Reinforced concrete design to Eurocode 2, Palgrave
Macmillan
64
Paper No.: 012
THE STUDY OF TENSION STIFFENING AND CRACK BAND WIDTH IN

REINFORCED CONCRETE BEAM UNDER FLEXURE
Shivaji T Bidgar 1* and Partha Bhattacharya 2

1
Research Scholar QIP (Poly), Department of Civil Engineering, Jadavpur University Kolkata, 700032
(Head of Applied Mechanics Department, Government Polytechnic Osmanabad 413513)
2
Associate Professor Department of Civil Engineering Jadavpur University Kolkata 700 032
Email: stbidgar@gmail.com p_bhatta@yahoo.com
ABSTRACT:
An effect of tension stiffening is important in the analysis of monotonic behavior of reinforced concrete (RC)
beams under flexure. The tensile strength of concrete is limited and hence when reinforced concrete beam is
subjected to flexure, the cracks are observed in tensile zone at certain intervals. The concrete between the
adjacent cracks is intact and carries some tensile stresses along with the reinforcement. The study of these
phenomena is carried out in “Nonlinear Finite Element Analysis of RC Beams” under monotonic loading. The
total strain based analysis is carried out with modified compression field theory and secant stiffness formulation.
The concrete before cracking or crushing is assumed as linear, isotropic and elastic material and after cracking
and crushing it is assumed as orthotropic material with stresses and strains in two principal directions as coaxial.
KEYWORDS: Reinforced concrete, nonlinear, tension stiffening, isotropic, orthotropic, crack band width.
1 INTRODUCTION
The study of failure of reinforced concrete structural members is the topic of research since long time. The
reinforced concrete is composite material consisting of concrete and reinforcing steel bars. The concrete itself is
a heterogeneous material having mixture of cement, fine aggregate, coarse aggregate, water and admixtures.
The numerical analysis of reinforced concrete is thought-provoking issue due to the intricacies in modeling
concrete, reinforcement and bond between concrete and reinforcing bar. The behavior of concrete in tension and
compression is altogether different and should be modeled separately. The response of reinforced concrete
member under monotonic increasing load is different at different loading stages. For early loading stage the
reinforced concrete material behaves linearly when concrete and steel both are in elastic range, as the load goes
on increasing the concrete in tension goes into nonlinear range due to limited strength in tension. The concrete
in compression is linear up to one third of fc and there after it is nonlinear. On the other hand steel is linear up to
yielding and there after it is strain hardening. In between the failure of bond between concrete and
reinforcement took place in the bond zone due to the slip. It is difficult to capture these entire phenomenons in
the numerical analysis of reinforced concrete members to obtain the response which is similar to the
experimental results. The aim of this paper is to present the issues involved in the failure of reinforced concrete
beam in tension and its importance in achieving over all load deformation response. A strain based nonlinear
finite element analysis procedure is adopted for this analysis.
Literature review reveals that, the beginning of finite element analysis of reinforced concrete members was
due to Ngo & Scordelis (1967). Subsequently many researchers developed different models for linear and
nonlinear analysis of structural members using various approaches. Vecchio & Collins (1986) and Vecchio
(1989) developed the modified compression field theory (MCFT) for nonlinear finite element analysis of
reinforced concrete members like slab panels, beams, shear walls. The linear finite element analysis procedure
with some modification is used for nonlinear analysis. The stress based approach and smeared rotating crack
model with secant stiffness formulation to avoid negativity of stiffness in tension softening and compression
softening region was the basis of their analysis. The plane stress finite element analysis using MCFT with strain
based approach was suggested by Wie, Wu, Liew & Zhishen (2006) for predicting the behavior of reinforced
concrete member in flexure. The effect of the current strength of concrete and current fracture energy of
concrete on load deformation response of RC member was found to be insignificant, Wie, Wu & Liew (2008).
The study of tension stiffening of concrete in tension zone of reinforced concrete member was investigated
by many researchers Alih and Khelil (2012), Perera and Mutsuyoshi (2011), Sokolov (2010), Ng, Lam and
Kwan (2009), Khalfallah (2008), Wu and Gilbert (2008) through experiments and numerical methods. Many of
them investigated this effect in pull out test on reinforced concrete member and RC bar subjected to axial and
uniform tension. Perera and Mutsuyoshi (2011) carried out experimental study of tension stiffening behavior of
high strength concrete tension member. Khalfallah (2008) considered local bond slip phenomenon to study the
tension stiffening of cracked flexural reinforced concrete beam. Sokolov (2010) concluded on the basis of their
study of tension stiffening model for reinforced concrete beams that tension stiffening reduces with increase in
reinforcement ratio and load deformation response is more accurate when tension stiffening model is included in
finite element formulation. The aim of this paper is to present the results of analysis of the failure of concrete
under tensile stresses and the variation of stresses and strains in the cracked and un-cracked concrete elements in
nonlinear finite element analysis of reinforced concrete beam.
2 MATERIALS AND METHODS
The nonlinear finite element analysis of reinforced concrete beam was carried out with four point monotonic
loading. The formulation of program in MATLAB 2009R with following tools is developed and the load
deformation response is obtained.
a) Modified compression field theory (MCFT) as suggested by Vecchio and Collins (1986) and
Vecchio (1989) is used with smeared rotating crack model and secant stiffness formulation.
b) The concept of equivalent principal strains and principal stresses originally suggested by Darwin
and Pecknold (1977) which was adopted by Wei et al (2006) and (2008) for their total strain based
nonlinear analysis is used for this formulation.
c) In the finite element analysis of RC beam, Concrete is modeled as plane stress, four node
isoparametric element with two degrees of freedom at each node. All reinforcing bars are modeled
as one dimensional two node bar element with two degrees of freedom at each node. The
reinforcement is placed at the edge of concrete element with their nodes coinciding.
d) The constitutive relationship for concrete in tension is assumed as bilinear and for concrete in
compression a parabolic curve as suggested by Vonk (1992) is used here. The ultimate cracking and
crushing strains are obtained from mode I and mode II fracture energy of concrete in tension and
compression, Bazant and Oh (1993). The concrete is assumed to be linear elastic and isotropic
before cracking and crushing. Once the concrete gets cracked and crushed it is assumed to be
orthotropic with equivalent principal strains and principal stresses as coaxial in the directions
perpendicular to crack and parallel to crack. The strains and stresses in the direction normal to crack
are tensile and they are taken as positive and in the direction parallel to crack are compressive and
taken as negative. The Poisson’s ratio effect is considered in the linear analysis and is taken as zero
for nonlinear analysis. The modulus of elasticity of the concrete in principal direction normal to
crack Ec1 and parallel to crack Ec2 are obtained from the equivalent strains and stresses in the
respective direction from the constitutive relationships for concrete in tension and compression.
e) The reinforcement is modeled as tri-linear curve which assumed as elastic perfectly plastic with
strain hardening up to ultimate strain. The perfect bond is assumed between concrete and
reinforcement for the present analysis.
f) The grade of concrete M30 is used for the present analysis with steel reinforcement as Fe415 HYSD
bars for longitudinal reinforcement and Fe250 MS bars for vertical 2 legged stirrups.
g) Owing to the symmetry only half of the RC beam is modeled. The details of RC beam, loading and
supports are as shown in Fig. 1(a) and finite element meshing in Fig. 1(b).
66
Figure 1 (a) Showing details of RC beam used for nonlinear finite element analysis
Figure 1 (b) RC beam finite element discretization, 12x40 mesh
h) An incremental iterative procedure is used for the present nonlinear analysis. The convergence of
moduli’s of concrete in principal direction normal to crack Ec1 and parallel to crack Ec2 is obtained
by norms of vector. The maximum 50 iterations are sufficient to get the satisfactory results of load
deformation response. The smaller load steps are considered in the region of high nonlinearity and
in the region where yielding of steel took place. The results of load deformation obtained from
numerical analysis of the reinforced concrete beam are compared with the experimental results.
3 RESULTS AND DISCUSSION
3.1. Experimental and Numerical load deformation of RC beam
The results of load and deflection at the center of beam are compared with experimental results obtained in the
laboratory. The results of numerical analysis are in well agreement with the experimental values and it is as
shown in Fig. 2.
67
Load vs Deflection at Center of Beam
35
30
25
Load in kN 20
15
Present Model Deflection
Experimental Deflection
10
0
0 1 2 3 4 5 6 7 8 9 10
Deflection in mm
Figure 2 Load verses central deflection for RC beam as shown in Fig.1 (a) (Concrete M30,
Steel Fe415 main bars, Fe250 stirrups, Poisson’s ratio=0.15)
3.2 Tension Stiffening and Crack Band Width
The detail analysis of reinforced concrete beam in tension zone is carried out to study the variation of strains
and stresses in concrete element, cracking and uncracking of elements and variation of stresses and strains in
adjacent bar elements. It is observed that for element mesh 12x30 alternate concrete element shows cracks and
for 12x40 every third element and for 12x50 every fourth element shows cracks. The elements in between
adjacent cracked elements remains un-cracked as the strains and stresses in these elements are within the linear
zone. When the strain rate in one of the element increase at very fast before final cracking the strain in the
adjacent element reduces and goes back into the linear zone. These un-cracked elements carry some tensile
stresses along with the bar elements. This phenomena observed in the reinforced concrete is known as tension
stiffening. The stress-strain variation for cracked and un-cracked element obtained from the nonlinear finite
element analysis is shown in Fig. 3(a) and Fig. 3(b).
Strain vs Stress in Element E1 Equivalent Strain vs Stress for Element 2

2 2
1.8 1.8
1.6 1.6
1.4 1.4
S tre s s in M P a
1.2
S tre s s in M P a
1.2
1 1
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Equivalent Strain Element E1 x 10
-3 Eqivalent strain Element 1 x 10
-4
(a) Cracked Concrete Element (b) Un-Cracked Concrete Element

Figure 3 Stress-Strain patterns for Concrete Elements in Tension Zone
If we assumed that the crack in the element appears at the center and if we calculate the crack band
width i.e. center to center distance between two adjacent cracks, it shows that the cracks appear at regular
interval. The convergence for crack band width is obtained for finer size mesh. The details of element size,
68
meshing, sequence of cracking and crack band width are shown in the Table No.1.
Table No. 1 Details of FE Mesh Size and Crack Band Width
Case Lx in Ly in nx ny Element Cracking sequence of Crack Band

No. m m Size in mm Element in bottom layer width in mm
1 0.75 0.15 30 12 25x12.5 E1, E3, E5… 50.00
2 0.75 0.15 40 12 18.75x 12.5 E1, E4, E7… 56.25
3 0.75 0.15 50 12 15x 12.5 E1, E5, E9… 60.00
Note – Lx - Half beam length, Ly - Depth of beam, nx – Number of element in X direction. ny – Number of
element in Y direction. ( E1, E2, E3…. are the Plane stress elements (concrete) in the bottom layer of RC beam
measured from center towards right.)
Equivalent Strain in Direction1 vs Load

20
E1
18 E2
E3
16 E4
14
12
Load in kN
10
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Strain in Elements E1, E2, E3 and E4 -3
x 10
Figure 4. The variation of equivalent strain in crack normal direction verse Load
(For elements in bottom layer from center of beam)
4 CONCLUSIONS
The load deformation response of reinforced concrete beam under monotonic load is well matching with the
experimental values up to yielding of steel reinforcement. The strain in the un-cracked elements is well within
the linear zone, which indicate that these elements are intact till the failure of beam and they carry tensile
stresses along with steel bar elements. The phenomenon of tension stiffening is well captured in the present
model for nonlinear finite element analysis of reinforced concrete beam. The crack band width obtained here are
on lower side. The crack band width obtained in the experiment shows large variation and hence nothing can be
concluded regarding crack band width.
REFERENCES
[1] Vecchio, F. J., & Collins, M.P.,( 1986), ‘The modified compression field theory for reinforced concrete
elements subjected to shear’, Journal of American Concrete Institute, 83(22), 219-231.
[2] Darwin, D. & Pecknold, D. A., (1977), ‘Nonlinear biaxial stress-strain law for concrete’, Journal of
Engineering Mechanics Division, 103 (EM2), 229-241.
69
[3] Vecchio, F.J., (1989), ‘Nonlinear Finite Element Analysis of Reinforced Concrete Membranes’, American
Concrete Institute Structural Journal, 86-S4, 26-35.
[4] Wei, H, Wu, Y. F., Liew, K.M., & Wu, Z., (2006) A 2D total strain based constitutive model for predicting
the behavior of concrete structures, International Journal of Engineering Science, 44, 1280-1303.
[5] Wei, H, Wu, Y.F., & Liew, K.M., (2008)A fracture energy based constitutive model for analysis of
reinforced concrete structures under cyclic loading, Computer Methods in Applied Mechanics and
Engineering, 197, 4745-4762.
[6] Bazant Z.P., & Oh B.H., (1983), Crack band theory for fracture of concrete, RILEM Material and
Structures, 16(93), 155-177.
[7] Vonk, R.A., (1992), Softening of concrete loaded in compression, PhD Thesis, Eindhoven University of
Technology, Einthoven, The Netherlands,.
[8] Perera S.V.T.J. and Mutsuyoshi H., (2011), Tension stiffening behavior of high strength concrete tension
members, Annual Research Journal of SLSAJ, (11), 10-18.
[9] Ng, P. L., Lam, J.Y.K. and Kwan A.K.H., (2010), Tension stiffening in concrete beams. Part I: FE analysis,
Proceedings of the Institution of Civil Engineers, Structures and Buildings, 163(1), 19-28.
[10] Khalfallah, S., (2008), Tension stiffening bond modeling of cracked flexural reinforced concrete beams,
Journal of Civil Engineering and Management, 14(2), 131-137.
[11] Sokolov, A., (2010), Tension stiffening model for reinforced concrete beams, PhD Thesis, Vilnius
Gediminas University.
[12] Wu, H. Q., and Gilbert R. I., (2008), An experimental study of tension stiffening in reinforced concrete
members under short-term and long term loads, UNCIV Report No. R-449, The University of New South
Wales, Sydney, Australia.
[13] Stramandinoli, R. S. B., & LaRovere, H. L., (2008), ‘An efficient tension stiffening model for nonlinear
analysis of reinforced concrete members’, Engineering Structures, 30(7), 2069-2080.
70
Paper No.: 015
EFFECT OF MASS RATIO ON THE PERFORMANCE OF A TMD WITH

NON-OPTIMAL DAMPING
1 2
Soumi Bhattacharyya and Aparna (Dey) Ghosh
1
Junior Research Fellow, Department of Civil Engineering, Bengal Engineering and Science University,
Shibpur, Howrah, India.
2
Associate Professor, Department of Civil Engineering, Bengal Engineering and Science University, Shibpur,
Howrah, India.
Email: soumibhttchr86@gmail.com, aparna@civil.becs.ac.in
ABSTRACT :
The Tuned Mass damper (TMD) is a very popular and well-established control device for the vibration
suppression of structures subjected to wind and earthquake forces. The basic spring-mass-dashpot model of the
TMD is often used by researchers to model the action of other passive control devices such as the sloshing
action of the Tuned Liquid Damper (TLD) in a simplistic manner. The design parameters of a TMD for
optimum performance include the tuning ratio, damping ratio and mass ratio. It is general practice to optimize
the first two parameters for a given value of mass ratio. It has been observed that increase of mass ratio leads to
an improvement of the TMD performance, but this is provided the tuning and damping ratios of the TMD are
optimized for the corresponding values of mass ratio. In this paper, an investigation has been carried out on the
effect of mass ratio on the performance of a TMD with non-optimal damping, such as in the case of the SDOF
idealization of the fundamental convective mode of the TLD. Frequency domain formulations of the
displacement transfer function of a base-excited SDOF structural system with (i) attached TMD and (ii) attached
TLD modeled as a 2-DOF system, have been presented. The input excitation has been characterized by a white
noise power spectral density function (PSDF). The performance of the damper systems has been examined on
the basis of reduction in the root mean square (rms) value of the structural displacement. Numerical results
reveal that a far lower value of optimum mass ratio exists when the damping of the TMD is non-optimal and
that its performance is also very sensitive to the mass ratio as compared to the case of optimal damping.
KEYWORDS: TMD, TLD, non-optimal damping, mass ratio, transfer function.
1. INTRODUCTION
The Tuned Mass Damper (TMD) is a well-established vibration control device used in mechanical and structural
applications. The first TMD was installed in the Centerpoint Tower in Sydney, Australia [1] to reduce the
wind-induced motions of the structure. Since then, there have been a number of successful structural
implementations of the TMD worldwide, such as in the Citicorp Center, New York, John Hancock Tower,
Boston, Funade Bridge Tower, Osaka, CN Tower, Toronto, etc. [2]. The concept of the TMD as a dynamic
vibration absorber was first developed by Frahm in 1909 [3]. Since then, considerable research has been carried
out on the TMD performance, to enable proper selection of design parameters. Den Hartog developed the basic
principles of undamped and damped dynamic vibration absorbers with an undamped structural system and the
procedure for proper selection of design parameters [4, 5]. In 1952, Bishop and Welbourn carried out the
analysis of dynamic vibration absorbers with a damped structural system [6]. Falcon et al. devised an
optimization procedure to minimize the peak response and obtain maximum effective damping in the structural
system [7]. The empirical formulas for correction factors of these optimum absorber parameters for a lightly
damped structural system were developed by Ioi and Ikeda in 1978 [8]. Further studies on absorber parameters
were carried out by Warburton and Ayorinde, Randall et al., Thompson, Fujino and Abe, Rana and Soong, Lee et
al., Bekdaş and Nigdeli [9, 10, 11, 12, 13, 14, 15]. A study by Lee [16] on the optimization of the ratio of the
TMD mass to the structural mass, i.e. the mass ratio, reported that the value of the optimum mass ratio is much
larger than those practically feasible and hence does not appear to be useful in practice.
The chief design parameters of the TMD are its tuning ratio (i.e. ratio of the damper frequency to the natural
frequency of the structure), damping ratio and mass ratio. In general practice, the values of tuning ratio and
damping ratio are optimized for a given value of mass ratio. The best TMD performance is achieved with the
highest value of mass ratio that is practically feasible, given the constraints regarding the vertical load carrying
capacity of the structure and the difficulties associated with the placement of a heavy mass on the top of the
structure, and the corresponding optimum tuning ratio and damping ratio of the mass damper. However, cases
concerning the TMD model of a dynamic vibration absorber may arise where the flexibility in the choice of the
design parameters is limited from some other considerations as well. For instance, in case of the Tuned Liquid
Damper (TLD), the simplified model of the sloshing modes of the liquid involves the basic spring-mass-dashpot
model of the TMD. As it is difficult to properly cater for the damping due to the nonlinear sloshing action of the
liquid, a linearized analysis often considers a fixed value of the damping ratio of the single-degree-of-freedom
(SDOF) system representing the fundamental sloshing mode of the liquid [17]. The SDOF system is tuned to the
structural mode to be controlled and hence may be termed a TMD. The performance of this TMD with increase
in mass ratio is not the same as that of the conventional TMD as the damping ratio remains fixed and is
non-optimal. To investigate these in detail, the present study has been carried out on the effect of mass ratio on
the performance of a TMD with non-optimal damping and has then been extended to the case of the
performance of the fundamental convective mode of a TLD which is idealized as a TMD. Frequency domain
formulations of the displacement transfer function of a base-excited SDOF structural system with attached TMD
and with an attached TLD, modeled as a 2-DOF system, have been presented. The performance of the damper
systems has been examined on the basis of the reduction in the root mean square (rms) value of the structural
displacement.
2. FORMULATION OF TRANSFER FUNCTION FOR TMD-SDOF STRUCTURAL SYSTEM
The structure is modeled as a SDOF system by considering only the predominant mode. The symbols to denote
parameters of the TMD-SDOF structural system are indicated in Table 1. The mathematical model of a
TMD-SDOF structural system subjected to a horizontal base acceleration, &z&(t ) , is shown in Fig. 1. Let u(t)
denote the horizontal displacement of ms relative to ground motion and y(t) denotes the horizontal displacement
of the TMD relative to the structure.
The normalized equation of motion of the TMD, md, is expressed as
2 1
where, 2 ⁄ and ⁄ .
The normalized equation of motion for ms yields
2 2 2
where, 2 ⁄ , ⁄ and ⁄ .
On Fourier transforming Eqs.1-2 and through appropriate substitutions, the input output relation between the
base acceleration and the displacement response of the structure equipped with the TMD is,
3
where, 4
is the transfer function relating the displacement of the structure to the input base acceleration.
72
Table 1 Symbols for parameters of
TMD-SDOF structural system

Parameters SDOF structural
TMD
system
Mass
Stiffness
Damping
coefficient
Damping
ratio
Natural
frequency
Figure 1 Mathematical model of
The transfer function 1⁄ 2 relates the displacement of a SDOF system

representing the mass to the base acceleration. Similarly, 1⁄ 2 is
the transfer function relating the displacement of a SDOF system representing the mass to the input
acceleration. Further, and are the Fourier transforms of the corresponding time dependent
variables u(t) and respectively. If the ground acceleration is characterized by a white noise power spectral
density function (PSDF) of intensity S0, then the PSDF of the displacement response of the structure is denoted
by | | [18]. The root mean square (rms) value of the displacement response of the structure,
, can be numerically evaluated by computing the square root of the area under the corresponding PSDF curve.
3. NUMERICAL STUDY ON TMD-SDOF STRUCTURAL SYSTEM
The numerical study has been conducted with an SDOF

structural system having a mass of 20000 kg ( ) and
time period 1.5 s. The structural damping has been
assumed to be 2%. The ground motion has been
characterized by a white noise PSDF of intensity 100
cm2/s3. Using the formulation in the preceding section,
the transfer function has been obtained and the percent
response reduction of the structure has been calculated.
The relation between mass ratio and response reduction
has been shown in Fig. 2. The optimum damping ratios
( ) and optimum tuning ratios (νopt) for each mass
ratio have been calculated using the empirical formula
given by Ioi and Ikeda [8] and are shown in Table 2.
From Fig. 2 it is observed that initially with the increase
in mass ratio, the percent response reduction increases
sharply, but with the further increase in mass ratio the Figure 2 Percent response reduction with
response reduction becomes almost constant and finally it mass ratio for νopt and for a TMD.
decreases when the mass ratio is very large. The optimum
mass ratio has been obtained as 43%, which is much larger than the practically feasible values [16].
Further, a study has been carried out with TMDs having some fixed values of the damping ratio ( ). For a fixed
value of , a range of mass ratios are considered and the optimum tuning ratio is evaluated in each case. The
resulting response reductions are presented in Fig. 3. From Fig. 3 it is observed that the lower values of damping
73
ratios have lower values of optimum mass ratio. Moreover, when the damping ratio is very low, the percent
response reduction decreases with the increase in mass ratio and even becomes negative when the value of
damping ratio is as low as 0.1%.
To examine this further, the displacement transfer function curves of the TMD-SDOF system for 5% mass ratio
and aforesaid damping ratios have been plotted in Fig. 4 and have been compared with that for the optimum
damping ratio (13.63%) of the TMD. From Fig. 4, the effect of tuning is manifested in the formation of two
peaks from the single peak of the structure alone case. Further it is observed that the flattening of the peaks is
maximum when the damping ratio is optimum. But for values of lower than the optimum, the two peaks of
the TMD-SDOF structural system become sharper. When is very low (0.1%) the area under the
displacement transfer function curve of TMD-SDOF structural system becomes larger than that of the SDOF
structural system alone and a negative percent response reduction is obtained. This is because while the mass of
the damper reduces the frequency of the combined system leading to increased displacement response, the very
low damping value is ineffective in reducing the response.
Figure 4 Displacement transfer function of

Figure 3 Percent response reduction with mass structure for 5% mass ratio, νopt and
ratio for νopt and different for a TMD. different for a TMD.
Table 2 Values of and νopt corresponding to different mass ratios

Mass
Ratio 1 2 3 4 5 6 7 8 9 10
(%)
6.36 8.84 10.72 12.28 13.63 14.84 15.94 16.95 17.88 18.75
(%)
νopt 0.9846 0.9745 0.9647 0.9551 0.9456 0.9364 0.9273 0.9184 0.9096 0.9011
4. FORMULATION OF TRANSFER FUNCTION FOR TLD-SDOF STRUCTURAL SYSTEM
The TLD is modeled as a 2-DOF system, comprising the fundamental convective mode and the impulsive mode.
The fundamental convective or sloshing mode is that which is tuned to the structural frequency for vibration
control while the impulsive mode, representing the portion of the liquid that moves rigidly with the container,
has a very high frequency and cannot be used for tuning to structural frequencies. The sloshing mode is assumed
to have a damping ratio of 1% [17]. This value of damping ratio would be far below the optimal value of a TMD
corresponding to the mass ratios being considered. The mathematical model of the TLD-SDOF structural
system has been shown in Fig. 5. The symbols to denote parameters of the TLD system are indicated in Table 3.
The expressions for , , and are given by Veletsos and Tang [17].
74
The normalized equation of motion of the convective mass of the tank water, , is expressed as
2 5
The normalized equation of motion of the impulsive mass of the tank water, , is expressed as
2 6
The normalized equation of motion for yields
2 2 2 7
where, 2 ⁄ , ⁄ , ⁄ and ⁄ .
Table 3 Symbols for parameters of TLD
TLD system
Parameters Convective
Impulsive mode
mode
Mass
Damping
ratio
Natural
frequency
Mass Ratio ⁄ ⁄
Figure 5 Mathematical model of

TLD-SDOF structural system
On Fourier transforming Eqs.4-6 and through appropriate substitutions, the input output relation between the
base acceleration and the displacement response of the structure equipped with the TLD system is given below
8
where, 9
is the transfer function relating the displacement of the structure to the input base acceleration. Also
1⁄ 2 and 1⁄ 2 are the transfer functions
relating the displacements of SDOF systems representing the masses and to the input base acceleration
respectively. The displacement response of the structure has been evaluated numerically by the previously
mentioned procedure.
5. NUMERICAL STUDY ON TLD-SDOF STRUCTURAL SYSTEM
The influence of mass ratio (1% to 10%) on the response reduction of the example structure under study
equipped with a TLD has been investigated with optimum tuning ratio (νopt) and 1% damping ( ) (Table 4). The
structural damping has been assumed to be 2%. The variation in the response reduction achieved by the TLD
system, in terms of the rms displacement of the structure, over a range of tuning ratio values, for different values
of mass ratios has been presented in Fig. 6. The transfer function curves for different mass ratios have been
75
presented in Fig. 7. In each case the optimum tuning ratio, evaluated numerically, has been considered. The rms
response reduction varies from 12.61% to 16.18%. It is observed that the control performance of the TLD does
not increase significantly with increase in mass ratio. Moreover, the value of percent response reduction
decreases when the mass ratio exceeds 2% (see Table 4).
Figure 6 Percent response reduction with Figure 7 Displacement transfer functions of

tuning ratio for different mass ratios structure with νopt and 1% for different
1% mass ratios
Figure 8 Percentage of fundamental sloshing

mass of water to total water mass for Figure 9 Percent response reduction with mass ratio
different mass ratios for νopt and 1% damping ratio for a TLD and a TMD.
To investigate the reason behind this phenomenon, the percentage of fundamental sloshing mass of water to the
total water mass in the TLD has been plotted against the mass ratio (see Fig.8). Here it is observed that the
percentage of water mass taking part in sloshing decreases with the increase in mass ratio. Thus the effective
mass ratio ( / is not increasing with the increase in mass ratio; and for a particular mass ratio the
effective damper mass is larger in case of TMD than that of a TLD. To examine this further, a comparison has
been made of the response reduction achieved by the TLD with increasing mass ratio considering optimum
tuning ratio (νopt) and 1% damping ( ) with the results of TMD (see Fig.9). In case of the TMD, the maximum
response reduction is 16.28% corresponding mass ratio is 1.3% and in case of the TLD the maximum response
reduction is 16.19% corresponding mass ratio is 1.7%. Though the trends of the two curves are similar, for
lower mass ratios the performance of the TMD is better, but when the mass ratio exceeds the optimal values, the
TLD provides a better performance than the TMD. The fixed value of 1% damping ratio assumed both in case
76
of the TLD sloshing mode and for the TMD is much below the optimal value for the mass ratios considered.
Greater the deviation from the optimal value, lower is the response reduction achieved by the damper. As the
mass ratio increases, lower effective damper mass in case of the TLD provides the slightly better performance of
the TLD as compared to the TMD.
Table 4 Optimum tuning ratio, response reduction (%), masses and frequencies of 2-DOF TLD for
different mass ratios
Mass Ratio 1 2 3 4 5 6 7 8 9 10
(%)
Optimum
Tuning Ratio 0.98 0.96 0.95 0.93 0.92 0.90 0.89 0.87 0.86 0.85
(νopt)
Response 15.86 16.18 15.95 15.57 15.12 14.64 14.14 13.63 13.12 12.61
reduction (%)
Mass of water 200 400 600 800 1000 1200 1400 1600 1800 2000
in TLD (kg)
Convective
Mass of water 149.65 286.91 416.43 545.17 667.73 795.56 916.30 1048.00 1170.50 1293.10
in TLD (kg)
Impulsive
Mass of water 37.699 89.861 150.240 211.280 278.880 340.780 410.240 467.950 535.580 602.980
in TLD (kg)
Frequency of
convective
4.105 4.021 3.979 3.896 3.854 3.770 3.728 3.644 3.602 3.561
mode ( )
(rad/s)
Frequency of
impulsive
2723.50 1922.70 1562.00 1369.40 1225.50 1139.20 1059.40 1014.60 964.25 922.69
mode ( )
(rad/s)
6. CONCLUSION
A frequency domain investigation has been carried out on the performance of a TMD with a fixed value of
damping ratio, which is non-optimal for the considered mass ratio. As an illustrative case, a TLD-structure
system has been taken where the fundamental sloshing mode of the TLD is modeled as a TMD with a fixed
value of damping ratio. The optimum tuning ratio has been calculated empirically in case of a TMD while in
case of a TLD it has been evaluated numerically. The displacement transfer functions of a base-excited SDOF
structural system with attached TMD and attached TLD, modeled as a 2-DOF system, have been formulated in
frequency domain. The ground motion has been characterized by a white noise PSDF of intensity 100 cm2/s3.
First, the study with the TMD shows that the lower values of damping ratios have lower values of optimum
mass ratio. Moreover, when damping ratio is as low as 0.1%, the response reduction becomes negative in case
of higher mass ratios; because while the mass of the damper reduces the frequency of the combined system
leading to increased displacement response, the very low damping value is ineffective in reducing the response.
Further, the investigation with the TLD reveals that for a fixed damping ratio the performance of the damper
deteriorates, if the ineffective mass of the damper is increased. So, it can be concluded that to achieve a better
response reduction, a lower mass ratio should be provided when the value of damping ratio is lower than the
optimal value.
77
REFERENCES
[1] Kwok, K. C. S. and MacDonald, P.A. (1987), Wind-induced Response of Sydney Tower, Proc. First National
Struct. Eng. Conf., 19-24.
[2] Soong T.T. and Dargush G.F. (1997). Passive Energy Dissipation Systems in Structural Engineering, Wiley:
New York.
[3] Frahm, H. (1909). Device for Damped Vibrations of Bodies, U.S. Patent No. 989958, Oct. 30.
[4] Den Hartog, J. P. (1956), Mechanical Vibrations, 4th Edition, McGraw-Hill, NY.
[5] Ormondroyd, J. and Den Hartog, J.P. (1928). The Theory of the Dynamic Vibration Absorber, Trans. ASME,
APM-50-7, 9-22.
[6] Bishop, R. E. D. and Welbourn, D. B. (1952). The Problem of the Dynamic Vibration Absorber, Engineering,
Lond., 174, 769.
[7] Falcon, K. C., Stone, B. J., Simcock, W. D. and Andrew, C. (1967). Optimization of Vibration Absorbers: A
Graphical Method for Use on Idealized Systems with Restricted Damping, J. Mech. Eng. Science, 9, 374-381.
[8] Ioi, T. and Ikeda, K. (1978). On the Dynamic Vibration Damped Absorber of the Vibration System, Bulletin
of Japanese Society of Mechanical Engineering, 21(151), 64-71.
[9] Warburton, G. B. and Ayorinde, E. O. (1980). Optimum Absorber Parameters for Simple Systems,
Earthquake Eng. Struct. Dyn., 8, 197-217.
[10] Randall, S. E., Halsted, D. M. and Taylor, D. L. (1981). Optimum Vibration Absorbers for Linear Damped
Systems, J. Mech. Des., ASME, 103, 908-913.
[11] Thompson, A. G. (1981). Optimum Tuning and Damping of A Dynamic Vibration Absorber Applied to a
Force Excited and Damped Primary System, J. Sound Vib., 77, 403-415.
[12] Fujino, Y. and Abe, M. (1993). Design Formulas for Tuned Mass Dampers based on a Perturbation
Technique, Earthquake Eng. Struct. Dyn., 22, 833-854.
[13] Rana, R. and Soong, T.T. (1998). Parametric Study and Simplified Design of Tuned Mass Damper,
Engineering Structures, 20-3, 193-204.
[14] Lee, C.L., Chen, Y.T., Chung, L.L., Wang, Y.P. (2006). Optimal design theories and applications of tuned
mass dampers, Engineering Structures, 28-1, 43-53.
[15] Bekdaş, G. and Nigdeli, S.M. (2013). Mass Ratio Factor for Optimum Tuned Mass Damper Strategies,
International Journal of Mechanical Sciences, 71, 68-84.
[16] Lee, J. (1990). Optimal Weight Absorber Designs for Vibration Structures Exposed to Random Excitations,
Earthquake Eng. Struct. Dyn., 19, 1209-1218.
[17] Veletsos, A.S. and Tang, Y. (1990). Soil-Structure Interaction Effects for Laterally Excited Liquid Storage
Tanks, Earthquake Engineering and Structural Dynamics, 19, 473-496.
[18] Newland D.E. (1993). An introduction to random vibrations, Spectral and Wavelet Analysis, Longman: New
York.
78
Paper No.: 016
STUDY OF TORSIONAL BEHAVIOUR OF RECTANGULAR RC BEAM

WRAPPED WITH GFRP
1 2
Tudu.Chhabirani and Patel.Asha
1
Asst. Professor, Dept. of Civil Engineering, DRIEMS, Cuttack, Odisha, India
2
Professor, Dept. of Civil Engineering, NIT Rourkela, Odisha, India
Email: rani.chhabi@gmail.com
ABSTRACT:
Fibre Reinforced Polymer (FRP) as an external reinforcement is used extensively to address the strength
requirements related to flexure and shear in structural systems. But the strengthening of members subjected to
torsion is yet to be explored. Torsion failure is an undesirable brittle form of failure. An attempt is made here to
study the behaviour and performance of reinforced concrete members strengthened with externally bonded
Glass FRP (GFRP) sheets subjected to pure torsion. In this study experimental investigation on the torsional
behaviour of rectangular RC beams strengthened using glass fibre reinforced polymer (GFRP) sheets are carried
out. Rectangular RC beams externally bonded with GFRP sheets were tested to failure using a symmetrical two
point static loading system. All total four RC beams cast were tested for this experimental study. All of them
having same reinforcement detailing were found to be weak in torsion. One beam was used as a control beam
and three beams were strengthened using different configurations of GFRP sheets. Experimental data on load,
angle of twist and failure modes of each of the beams were obtained. The effect of different amount and
configuration of GFRP on ultimate load carrying capacity and failure mode of the beams were also investigated.
The experimental results show that externally bonded GFRP can increase the torsional capacity of the beam
significantly.
KEYWORDS: Angle of twist, Flexure, GFRP, Shear, Torsion, Ultimate load
1.INTRODUCTION
During its whole life span, nearly all engineering structures ranging from residential buildings, an industrial
building to power stations and bridges faces degradation or deteriorations. The structural retrofit problem has
two options, repair/retrofit or demolition/reconstruction. Demolition or reconstruction means complete
replacement of an existing structure may not be a cost-effective solution and it is likely to become an increasing
financial burden if upgrading is a viable alternative. Therefore, repair and rehabilitation of bridges, buildings,
and other civil engineering structures is very often chosen over reconstruction for the damage caused due to
degradation, aging, lack of maintenance, and severe earthquakes and changes in the current design requirements.
This problem needs development of successful structural retrofit technologies.
Previously, the retrofitting of reinforced concrete structures, such as columns, beams and other structural
elements, was done by removing and replacing the low quality or damaged concrete or/and steel reinforcements
with new and stronger material. However, with the introduction of new advanced composite materials such as
fiber reinforced polymer (FRP) composites, concrete members can now be easily and effectively strengthened
using externally bonded FRP composites. Retrofitting of concrete structures with wrapping FRP sheets provide
a more economical and technically superior alternative to the traditional techniques in many situations because
it offers high strength, low weight, corrosion resistance, high fatigue resistance, easy and rapid installation and
minimal change in structural geometry. In addition, FRP manufacturing offers a unique opportunity for the
development of shapes and forms that would be difficult or impossible with the conventional steel materials.

Early efforts for understanding the response of plain concrete subjected to pure torsion revealed that the material
fails in tension rather than shear. Structural members curved in plan, members of a space frame, eccentrically
loaded beams, curved box girders in bridges, spandrel beams in buildings, and spiral stair-cases are typical
examples of the structural elements subjected to torsional moments and torsion cannot be neglected while
designing such members. In addition, torsion is usually associated with bending moments and shearing forces,
and the interaction among these forces is important. Thus, the behaviour of concrete elements in torsion is
primarily governed by the tensile response of the material, particularly its tensile cracking characteristics. .
Reinforced concrete (RC) beams have been found to be deficient in torsional capacity and in need of
strengthening. Similar to the flexure and shear strengthening, the FRP fabric is bonded to the tension surface of
the RC members for torsion strengthening. In the case of torsion, all sides of the member are subjected to
diagonal tension and therefore the FRP sheets should be applied to all the faces of the member cross section.
However, it is not always possible to provide external reinforcement for all the surfaces of the member cross
section. In cases of inaccessible sides of the cross section, additional means of strengthening has to be provided
to establish the adequate mechanism required to resist the torsion.
Most of the research projects investigating the use of FRP focused on enhancing the flexural and shear
behaviour, ductility, and confinement of concrete structural members. A limited number of experimental studies
were conducted to investigate torsion strengthening of RC members (Ameli and Ronagh 2007; Hii and Al-
Mahadi 2006; Rahal and Collins 1995; Behera et al. 2008). Santhakumar et al. (2007) presented the numerical
study on unretrofitted and retrofitted reinforced concrete beams subjected to combined bending and torsion. An
experimental investigation together with a numerical study on reinforced concrete beams subjected to torsion
that are strengthened with FRP wraps, were presented by Ameli et al. (2007). In the same year Chalioris
addressed an analytical method for the prediction of the entire torsional behaviour of reinforced concrete (RC)
beams strengthened with externally bonded fibre-reinforced-polymers (FRP) materials. In more recent studies,
Deifalla and Ghobarah (2010) developed an analytical model for the case of the RC beams strengthened in
torsion. Zojaji and Kabir (2011) developed a new computational procedure to predict the full torsional response
of reinforced concrete beams strengthened with Fiber Reinforced Plastics (FRPs), based on the Softened
Membrane Model for Torsion (SMMT).
The aim of present work is to study behaviour and performance of RCC rectangular beams strengthened with
externally bonded Glass Fiber Reinforced Polymer (GFRP) subjected to combined flexure and torsion. In the
present work the behaviour of rectangular reinforced concrete beams, strengthened with GFRP is observed to
know the practical feasibility of its application in the construction industry. Four numbers of rectangular
reinforced concrete beams are cast. All these beams except one beam are bonded with GFRP fabrics using
epoxy in different size and orientations. These beams are subjected to torsion by applying gradually increasing
static loading at the two cantilever moment arm of the beam to evaluate the increase in the torsional strength due
to retrofitting. The variables considered in the experimental studies include
(1) Fiber orientation (45o, 90o oriented glass fiber fabrics)
(2) GFRP configuration (continuously fully wrapped and wrapped in strips)
2. EXPERIMENTAL DETAILS
2.1. Materials Used
For the beam specimen, the design compressive strength of M20 was used. The design mix proportions and the
quantities of different materials used for the casting of four beams are provided in Table-1. 53 grade Konark
Portland Slag Cement (PPC) confirming to IS: 1489 – 1991, having a specific gravity of 2.5 was used. Locally
available river sand having a fineness modulus of 2.61, specific gravity of 2.51 was used. Crushed granite
coarse aggregates of 20mm and 10mm size, having a fineness modulus of 6.97 and specific gravity of 2.78,
were used. Fibre mat used for wrapping on reinforced concrete beams was Nitowrap EP Glass Fibre (GF). The
orientation of fibre was bidirectional with density 2.6 g/cc. By using INSTRON 1195 universal tensile test
80

machine the ultimate strength and Young’s modulus of GFRP was found out 30.89 KN and 6158 MPa
respectively. High yield strength deformed bars of Fe 415 were used as reinforcement in the concrete beam.
Table 1 Design Mix Proportions
Sand Course
Description Cement Water
(Fine Aggregate) Aggregate
Mix proportion
1 1.8 3.6 0.5
(by weight)
Quantities of materials
46 84 168 26
3
(in Kg/m ) for each beam
2.2. Details of Test Specimen
Three types of beams were casted for torsional test out of which one was control beam, the second one was
beam fully wrapped with GFRP over full a length of 0.65m and two numbers of the third type first vertical strips
of GFRP at 900 with the beam at 5cm width at 15cm spacing and the last one was beam with inclined strips of
GFRP oriented at an angle 450 to the beam at 15cm spacing. The reinforcement used in the beam and the its
detailing at section A-A is shown in the Figures1 (a, b). A steel mould satisfying the above specimen size was
used. Before casting, machine oil was smeared on the inner surfaces of moulds. Concrete was mixed using
tilting type laboratory mixing machine and was poured into the moulds in layers. Each layer of concrete was
compacted well with vibrator. The specimens were removed from mould after 24 hours of casting and then
cured using jute bags. All specimens were moist cured for 28 days.

1650

3, 12 mm dia
800 A
150

2, 16mm dia 6 mm dia @ 150 mm
300 A
6 mm dia @ 130mm
150 3, 12mm diameter

Top view of Torsional Beam

Figure 1(a) Reinforcement Detailing of the Beams
81

150
6mm dia @ 130 mm
2, 10 mm diameter
250
2, 16 mm diameter
Figure 1(b) Reinforcement Detailing of Beam at Sec-A-A
2.3. Wrapping of Glass Fibre
For torsion test, full wrapping of beam-2 was done with GFRP sheets and 50mm GFRP strips were wrapped
around the beam-3 and beam-4 with epoxy resin along four sides of the beam. Two type of GFRP orientation
900 and 450 was followed to study the torsional effect of the reinforced concrete beam wrapped with 50mm
GFRP strips. Two layers of GFRP sheets were externally glued to the beams except control beam-1.
2.4. Test Setup
The beams were tested in the loading frame of “Structural Engineering” Laboratory of National Institute of
Technology, Rourkela. Two-point loading is conveniently provided by the arrangement shown in Figure 2. The
load is transmitted through a load cell and spherical seating on to a spreader beam. The spreader beam is
installed on rollers seated on steel plates bedded on the test member with cement in order to provide a smooth
levelled surface. The test member is supported on roller bearings acting on similar spreader plates. The
specimen is placed over the two steel rollers bearing leaving 50 mm from the ends of the beam. The load is
transmitted through a load cell via the square plates kept over the flange of the beam at a distance 105mm from
the end. Loading was done by Hydraulic Jack of capacity 100 Tones.
2.5. Testing Procedure
All the beams were tested pure torsion. The central point load was applied to the beam for the fixity of the beam
and to avoid the uplifting at the time of uplifting force. The torsional force was applied by means of two
cantilever section or moment arms in the opposite sides of the beam. The hydraulic jacks were fixed at the
middle of cantilever portion to apply the twisting force. The hydraulic jacks were used to apply the twisting load
at same speed. The gradual increase in load and the deformation in the strain gauge reading are taken throughout
the test. The dial gauge reading shows the deformation. The load at which the first visible crack is developed is
recorded as cracking load. Then the load is applied till the ultimate failure of the beam. The deflections was
taken at a distance 0.3 m away from the middle of the main beam as the maximum value of deflection persists at
this point which is further used for the calculation of angle of twist.
82

250
800
1650
Figure 2 Schematic view of loading pattern on the beam (Test Setup)
3. EXPERIMENTAL RESULTS AND DISCUSSIONS
From the test result summarized in Table 2, it can be seen that GFRP wrapping relatively increases the torsional
moment carrying capacity of the beam. The full wrap of GFRP strips shows more torsional moment carrying
capacity. Figs. 3(a, b, c and d) shows the ultimate torsional moment carrying capacity of the torsion beams with
different configurations. The control specimen of the torsion test failed at a torsional moment of 35.1 kNm
whereas the Beam-2 failed completely in torsion at a load 210KN and torsional moment of 56.7KNm. It was
found that the crack were appeared making an angle 500-600 with the main beam on the side faces and top
surface. In Beam-3, at the load of 100 KN a cracking sound was heard as it completely failed in torsion at a load
216KN and torsional moment of 58.32KNm. The increase strength of beam-3 was 66.15% as compared to
control beam. Beam-4 was strengthened by wrapping with four 5cm Bidirectional GFRP strips at a distance
15cm from each other over the beam under torsion. The GFRP strips were wrapped over the beam by making an
angle 450 with the main beam. At the load of 100 KN cracking sound was heard. The Beam 3 failed completely
in torsion at a load 188KN and torsional moment 50.76KNm. The increase strength of beam was 55.38% as
compared to the control beam. After the test is done the GFRP sheets were removed and the crack pattern was
observed. It was found those cracks were appeared making an angle 400 with the main beam on the side faces
and 400 at the top surface.
Figure 3(a) Control Beam Figure 3(b) Beam-2
83

Figure 3(c) Beam-3 Figure 3(d) Beam-4
Figure 3 (a, b, c and d) Crack pattern on four experimental beams
Table 2 Summary of Test Results of Torsion Test
Beam Designation Torsional Moment Angle of Remarks

(KNm) twist(rad/m)
Beam-1(Control 21.6 0.010 First hairline crack
Beam) appeared (80KN)
35.1 Ultimate
load(130KN)
Beam-2(Full Wrap 35.1 First cracking sound
0.00829
with GFRP) heard(140KN)
Ultimate
56.7
load(210KN)
Beam-3(GFRP strip 29.7 0.00684 First hairline crack
at 900) appeared (100KN)
45.5 0.0118 Rupture and
debonding occurred
58.32 Ultimate
load(216KN)
Beam-4(GFRP strip 29.7 0.00624 First hairline crack
at 450) appeared (90KN)
48.6 0.011 Debonding Occurred
54.54 Ultimate
load(200KN)
3.1. Twisting Characteristics
The torsional moment versus twisting angle measured up to the yield stage for various beams are presented in
Figure 4. The control beam shows typical moment versus twisting angle curve defined by an initial linear
portion up to cracking after which, there is steep increase in twisting angle with increase in torsional moment.
The same was observed in the beams wrapped with GFRP fully and in 50mm strips.
84

Fig 4 Torsional moment Vs Twisting angle
3.2. Ultimate Load Carrying Capacity
The ultimate torsional moment carrying capacity of various beams is presented in the Table 2. The GFRP
wrapping increases the torsional moment carrying capacity of the reinforced concrete beams. The increase
strength of fully wrapped beam-3 was 66.15% as compared to control beam.
3.3. Modes of Failure
The beams tested in torsion failed by parallel shear crack pattern. In control beam the initial hairline cracks
appeared at load of 80 KN. Later on with increase in loading values the crack propagated further. It was
observed that those cracks were appeared making an angle 400-500 with the main beam. The cracks were
developed in a spiral pattern all over the main beam which later leads to the collapse of the beam in torsional
shear. For other three beams excluding control beam, after the test is done the GFRP sheets were removed and
the crack pattern was observed. It was found that the crack were appeared making an angle 500-700 with the
main beam on the side faces and top surface. The Fig.6 shows the failure pattern of torsional beams.
4. CONCLUSIONS
Based on the experimental investigation on the torsional behaviour of four different rectangular RC beams the
following inferences can be made,
1) The GFRP wrapping increases the torsional moment carrying capacity of reinforced concrete beams.
2) Two types of configurations were studied out of which strip with 900 shows 66.15% increment in torsional
moment compared to 450.
3) In addition the results indicate that the most effective configuration was the full-wrap of GFRP.
ACKNOWLEDGEMENT
The authors thankfully acknowledge the Department of Civil Engineering, for providing research facilities in
the Structural Engineering Laboratory of National Institute of Technology, Rourkela, India.
REFERENCES
[1] ACI Committee 440 (1996) State of Art Report on Fiber Reinforced Plastic.
85

[2] Ameli, M. and Ronagh, H.R. (2007). Behaviour of FRP strengthened reinforced concrete beams under
torsion. Journal of Composites for Construction, 11(2), 192-200.
[3] Ameli, M., and Ronagh, H. R. (2007). Analytical method for evaluating ultimate torque of FRP
strengthened reinforced concrete beams. Journal of Composites for Construction, 11, 384–390.
[4] Amir, M., Patel, K. (2002). Flexural strengthening of reinforced concrete flanged beams with composite
laminates. Journal of Composites for Construction, 6(2), 97-103
[5] Andre, P., Massicotte, Bruno, Eric, (1995). Strengthening of reinforced concrete beams with composite
materials: Theoretical study. Journal of composite Structures, 33, 63-75.
[6] Arbesman, B. (1975). Effect of stirrup cover and amount of reinforcement on shear capacity of
reinforced concrete beams. MEng thesis, Univ. of Toronto.
[7] Arduini, M., Tommaso, D. A., Nanni, A. (1997). Brittle Failure in FRP Plate and Sheet Bonded
Beams. ACI Structural Journal, 94 (4), 363-370.
[8] Belarbi, A., and Hsu, T. T. C. (1995). Constitutive laws of softened concrete in biaxial tension-
compression. ACI Structural Journal, 92, 562-573.
[9] Chalioris, C.E. (2006). Experimental study of the torsion of reinforced concrete members.
Structural Engineering & Mechanics, 23(6), 713-737.
[10] Chalioris, C.E. (2007a). Torsional strengthening of rectangular and flanged beams using carbon fibre
reinforced polymers – Experimental study. Construction & Building Materials, in press (available online
since 16 Nov. 2006).
[11] Chalioris, C.E. (2007b). Tests and analysis of reinforced concrete beams under torsion retrofitted with
FRP strips. Proceedings 13th Computational Methods and Experimental Measurements (CMEM 2007),
Prague, Czech Republic.
[12] Chen, J.F. and Teng, J.G. (2003a). Shear capacity of fiber-reinforced polymer-strengthened reinforced
concrete beams: Fiber reinforced polymer rupture. Journal of Structural Engineering, ASCE, 129(5), 615-
625.
[13] Chen, J.F. and Teng, J.G. (2003b). Shear capacity of FRP-strengthened RC beams: FRP debonding.
Construction & Building Materials, 17, 27-41.
[14] Deifalla A. and Ghobarah A.(2010). Full Torsional Behavior of RC Beams Wrapped with FRP: Analytical
Model. Journal of Composites for Construction, 14, 289-300.
[15] Gobarah,A., Ghorbel, M., and Chidiac, S. (2002). Upgrading torsional resistance of RC beams using FRP.
Journal of Composites for Construction, 6,257–263.
[16] Hii, A. K. Y., and Al-Mahadi, R. (2007). Torsional capacity of CFRP strengthened reinforced concrete
beams. Journal of Composites for Construction, 11, 71–80.
86

[17] Hii, A.K.Y. and Al-Mahaidi, R. (2006). An experimental and numerical investigation on torsional
strengthening of solid and box-section RC beams using CFRP laminates. Journal of Composite Structures,
75, 213-221.
[18] Hsu, T.T.C. and Mo, Y.L. (1985). Softening of concrete in torsional members – Theory and tests. ACI
Journal Proceedings, 82(3), 290-303.
87

Paper No.: 017
CONTROL OF WIND INDUCED VIBRATION IN TRANSMISSION LINE

TOWERS USING TUNED LIQUID COLUMN DAMPER
1 2
Jyotirmoya Dutta Majumdar and Aparna (Dey) Ghosh
1
Engineer, Transportation Department, AECOM India Pvt. Ltd. India
Email: catchjyotirmoya@gmail.com.
2
Howrah, India
Email: aparna@civil.becs.ac.in, aparrnadeyghosh@gmail.com.
ABSTRACT:
The tuned liquid column damper (TLCD) is a passive damper in which the oscillation of liquid (water) in a
U-tube with orifice(s) acts as the energy dissipative mechanism. In this study, a TLCD system has been
proposed for the response mitigation of a transmission line tower (TLT) under wind loading. A study on the
performance of the TLCD system attached to a TLT modelled as a single-degree-freedom (SDOF) system has
been carried out in frequency domain. A TLT structure of significant height that would be susceptible to
dynamic wind loading has been taken under consideration. The performance of the TLCD system has been
examined on the basis of reduction in the root mean square (rms) value of the structural displacement. The
governing differential equations for the liquid within the damper and for the structure coupled with TLCD
system have been derived. Thereby in frequency domain, the input-output relation of the TLCD-SDOF
structural system has been developed. The input excitation has been characterized by three different power
spectral density functions (PSDF) – Harris, Kaimal and Davenport. A detailed parametric optimization study
has been performed considering different values of mass ratio, tuning ratio, length ratio, head loss coefficient
and wind speed. The results indicate that substantial response reduction can be achieved by the TLCD.
KEYWORDS: Transmission line tower, TLCD, wind induced vibration, PSDF
1. INTRODUCTION
A transmission line tower (TLT) is a steel lattice tower, used to support overhead electricity conductors for
electric power transmission. The mechanical support system of the transmission line represents a significant
portion of the total cost and it plays an important role in reliable power transmission. Apart from its own dead
load, the effect of wind load on the tower is a very predominant one.
Control of the wind excitations of TLTs can prevent the frequent failure of these structures and also lead to a
more economic design. Battista et al. (2003) studied nonlinear pendulum-like dampers for the attenuation of the
TLT’s wind induced displacements. Chen et al. (2009) carried out studies on control of wind induced response
of TLTs by using magneto-rheological dampers. Amongst the various structural protection technologies
available, passive control is cost-efficient, reliable and relatively easier to implement and appears an ideal
option here.
The tuned liquid damper or TLD is one of the most effective passive dampers and has been well-implemented
for the control of wind induced vibration of structures such as the Nagasaki Airport Tower, the Yokohama
Marine Tower, Shin-Yokohama Prince Hotel, Tokyo air traffic control towers and the Sakitama bridge (Soong
and Dargush (1997)). The major advantages of liquid dampers over other types of passive dampers include low
initial cost, virtually maintenance free operation and ease of frequency tuning.
The tuned liquid column damper or TLCD is a type of liquid damper in which the liquid oscillates in a
U-shaped container to counteract the forces acting on the structure. Damping in the TLCD occurs due to head
loss experienced by the liquid column moving through orifice(s) provided in the horizontal portion of the
U-tube. Some advantages of the TLCD over other liquid dampers are higher volumetric efficiency with respect
to a given amount of liquid, consistent performance across a wide range of excitation levels as the natural
frequency of the oscillating liquid column in independent of the amplitude of the input excitation, clear
mechanism of damping that can be quantified in a definite manner. Also the TLCD frequency can be modified
by simply changing the length of liquid column in the tube. In this paper, a frequency domain study has been
carried out on the possible suppression of wind excitation of a TLT by the TLCD.
2. MODELING OF TLCD-SDOF SYSTEM
The TLCD has been considered to be composed of a U-tube like container with orifice(s), installed in the
horizontal portion of the U-tube (see Fig. 1). The cross-sectional area of the tube has been denoted by A and the
length of the horizontal portion, measured between centre to centre of the two columns of the tube, has been
termed as b. The mass of the liquid in the container has density ρ. The total length of the liquid column in the
U-tube has been denoted by l. The coefficient of head loss, ξ of the TLCD is controlled by the orifice opening
ratio. The TLCD is rigidly attached to the TLT structure which has been modelled as a SDOF system. M, K and
C respectively denote the mass, stiffness and damping of the SDOF system. The representative model of the
structure-TLCD system has been shown in Fig. 2.
Figure 1: Definition sketch for Tuned Liquid Column Damper (TLCD)
Figure 2: Model of TLCD-SDOF Structural System
89
3. FORMULATION OF TRANSFER FUNCTION
Let us consider a linear SDOF structural system subjected to a horizontal wind force F(t). A TLCD is rigidly
attached to the structure (see Fig. 2). Let (t), (t) and (t) respectively denote the horizontal displacement,
velocity and acceleration along the DOF of the structure relative to the ground. The equation of motion for M
may be written as
M&x&(t ) + Cx&(t ) + Kx(t ) = F (t ) + f (t ) (1)
The force transmitted from the TLCD to the structure is denoted by f(t) which is given by
f (t ) = −{ρAl&x&(t ) + ρAbu&&(t )} (2)
where, u(t) is the change in elevation of the liquid column.
Normalizing Eq. (1) by M and applying Fourier transformation, the following are obtained.
1
X (ω ) = H 1 (ω )[ F (ω ) + f (ω )] (3)
M
where, f (ω ) = ρ Al ω 2 X (ω ) + ρ AB ω 2U (ω ) (4)
1
and H 1 (ω ) = (5)
ωs 2
− ω + 2iωω s ξ s
2
Here, ξs and ωs denote the damping ratio and natural frequency of the structural model.
The equation of motion of the liquid column as derived by Saoka et al. (1988) is given by
1
ρAl u&&(t ) + ρAξ u& (t ) u& (t ) + 2 ρAgu (t ) = − ρAb&x&(t ) (6)
2
By adopting an equivalent linearization procedure (Iwan and Yang (1972)), the following linear equation may
be written to represent the nonlinear system in Eq. (6).
ρ Al u&&(t ) + 2 ρ AC p u& (t ) + 2 ρ Agu (t ) = − ρ Ab &x&(t ) (7)
where, Cp is the equivalent linearized damping coefficient, given by, C p = σ u& ξ 2π , σ being the standard
deviation of the liquid velocity.
Normalization of Eq. (7) with respect to the mass of liquid in the damper, i.e. ρAl, and subsequent Fourier
transformation leads to the following equation in the frequency domain:
b
U (ω ) = H 2 (ω ) ω 2 X (ω ) (8)
l
1
where, H 2 (ω ) = (9)
C
ω l 2 − ω 2 + 2iω p
l
The natural frequency of the LCD is ω l = 2 g l .
90
The substitution of Eq. (8) in Eq. (4) leads to
ρAb 2ω 4
f (ω ) = ρAl ω X (ω ) +
2
H 2 (ω ) X (ω ) (10)
l
Again on substituting the above expression of f(ω) in Eq.(3), the input-output relation between the structural
displacement and the wind force in frequency domain may be written as
X (ω ) = H (ω ) F (ω ) (12)
H 1 (ω )
where, H (ω ) = (13)
ρ Ab 2 ω 4
M − ρ Al ω 2 H 1 (ω ) − H 1 (ω ) H 2 (ω )
l
Here H(ω) is the transfer function relating the input force to the output displacement. If μ = ρAl/M be the mass
ratio and α = b/l be the length ratio, the transfer function H(ω) may be expressed as
H 1 (ω )
H (ω ) =
(
M 1 − μω H 1 (ω ) − μα 2ω 4 H 1 (ω ) H 2 (ω )
2
) (14)
If the wind force is characterized by a Power Spectral Density Function (PSDF) of intensity SF, then the PSDF
of the displacement response of the structure, denoted by Sx(ω), is expressed by (Newland (1993))
S x (ω ) = H (ω ) S F (ω )
2
(15)
The rms value of the displacement response of the structure, σx, can be numerically evaluated by computing the
square root of the area under the corresponding PSDF curve as given in Eq. (15).
4. NUMERICAL STUDY
Resonant dynamic response does not appear to be a major problem with transmission line system. Although the
suspended lines themselves usually have natural frequencies less than 1 Hz, the resonant response is largely
damped out because of the very large aerodynamic damping. The natural frequency of TLTs up to 50 m in
height are normally greater than 1 Hz and hence resonant response is also negligible. Thus, except for extremely
tall supporting towers and long line spans, the peak response of TLT can be safely computed, neglecting the
resonant dynamic response (Holmes (2007)). Thus, in order to take dynamic response into consideration, a very
high supporting tower has to be selected.
The TLT considered in this study is based on the tall electricity pylons existing in the Jiangyin, Jiangsu Province,
China (www.powerengineeringint.com Vol: 13, Issue: 6). The power line section running over the river has a
span width of 2,303 m. It is carried by two 346.5 m tall suspension electricity pylons, each weighing 4,192 tons.
They are identical lattice towers of square cross section with an area of 68 m × 68 m at the base and 8 m × 8 m
at the top. Based on the specifications of this tower, an example TLT is considered with the following data:
• Mass of the tower: 4200 ton.
• Height of the tower: 350 m.
• Cross-sectional area: Square cross-section with 68 m x 68 m dimension at base and 8 m x 8 m
dimension at top.
• Assumed Solidity ratio: 0.15
• Frontal Area of the structure: 1767 m2
• Height at which the TLCD is installed: 300 m.
91
To estimate the fundamental time period of this TLT, an empirical formula is used (Balendra et al. (1995))
TP = h 2 (750(wb + wt )) (16)
where, Tp is the time period in sec, h is the height of tower, wb and wt are respectively bottom and top width of
the tower. The fundamental time period is thereby evaluated as 2.15 sec (i.e. fundamental frequency is 2.981
rad/s or 0.474 Hz). As it is a steel lattice tower, the damping ratio here is taken as 2%.
Regarding the TLCD, it not considered to be installed at the top of the tower. This is because unlike buildings or
water tanks TLTs do not have enough space available at the top for installation. So it is assumed that the TLCD
is attached to the tower at a height where the cross-arms are located, which are at a height of 300 m in the
present example case. The wind force is calculated corresponding to this height. Three different PSDFs have
been used to characterize the wind force, namely those given by Harris (Balendra et al. (1995)), Davenport
(Blevins (1990)) and Kaimal (Murtagh et al. (2005)).
The sensitivity of the structural response reduction by the TLCD to parameters such as head loss coefficient ξ,
mass ratio µ, tuning ratio ν (ratio of frequency of a TLCD to that of the structure) and length ratio α, is analyzed
by a parametric study and an optimal design is sought. The study has been carried out for different values of
mean wind velocity. From practical considerations, it is not feasible to provide a very high mass ratio, since that
would increase the gravity load on the structure significantly. Thus a few discrete values of mass ratio µ, which
are feasible from design point of view, are considered. First, to study the effect of the headloss coefficient ξ , ν is
fixed at unity and the mean wind velocity U10, is taken as 20 m/s. The resulting percent displacement reductions
for several values of α, using the Harris spectrum as input, are plotted in Figs. 3a and 3b for µ = 0.5% and 2%
respectively. From these figures it is observed that when ξ is small, the response reduction increases sharply
with increase in ξ. It attains a peak value at ξ = ξopt and then reduces gradually. Thus, for low values of ξ the
response reduction is highly sensitive to ξ and so the design value of ξ must never be below ξopt. It is also noted
that ξopt increases with increase in µ but is practically independent of α. Increase in α only has the effect of
increasing the response reduction. The results have also been evaluated for a higher value of the mean wind
velocity (graph not shown here) and the observations remain unchanged.
a. µ = 0.5%, ν=1, and U10=20 m/s b. µ = 2%, ν=1, and U10=20 m/s
Figure 3: Variation of rms displacement reduction with ξ and α
Next, the sensitivity of the response reduction to tuning ratio ν is investigated. The ξopt values as obtained from
Fig. 3 are taken and the percent reduction in rms displacement for several α values and for ν ranging from 0.5 to
1.5 are evaluated and presented in Figs. 4a and 4b for two different values of the mean wind velocity U10.
92
a. µ = 0.5%, ξ = ξopt and U10=20 m/s b. µ = 0.5%, ξ = ξopt and U10=25 m/s
Figure 4: Variation of rms displacement reduction with ν and α
The figures show that the response reduction is highly sensitive to ν. It drops sharply if ν shifts from optimal
value. So care has to be taken while tuning the TLCD. Further, the figures indicate that the νopt value is
independent of α and U10. The variation in response reduction with ν has also been examined for different mass
ratios (figures not shown here). It is found that νopt is very close to unity and reduces slightly with increase in µ.
In general ν should not be allowed to vary outside a range of 0.95 – 1.05 for better performance. The foregoing
study also confirms that larger µ and α values result in better control performance in general. The restriction on
providing higher µ has already been mentioned. The value of α is restricted by the maximum displacement of
the liquid column |umax| and cannot exceed the value of (l-b)/2. If the value of α is raised above this threshold
value, it destroys the basic characteristics of the TLCD.
To obtain a quantitative estimate of the optimal parameters and corresponding response reductions for different
PSDF of wind force, similarly parametric studies have been performed using Davenport and Kaimal spectrums
and the results are presented in Table 1. The length ratio is kept fixed at 0.9. Three different values of mass ratio,
namely 0.5%, 1% and 1.5%, have been considered. Also, two different values of the mean wind velocity, U10 ,
equal to 20 m/s and 25 m/s, have been taken. From Table 1 it can be observed that for the Davenport spectrum,
which has a higher power intensity, the response reductions are low. For the Kaimal spectrum on the other hand,
which has a lower intensity, the response reductions achieved by the TLCD are the highest. Response reductions
increase with increase in mass ratio. However, there is no significant effect of the mean wind velocity on the
damper performance.
The displacement transfer function curves for the structure alone and with TLCD having optimal parameters as
indicated in Table 1 are plotted in Fig. 5. The wind load input is characterized by the Harris PSDF and U10=20
m/s is considered. It is observed that the single peak located at the frequency of the SDOF structural system has
been replaced by two peaks of reduced magnitude in the presence of the TLCD. This reflects the effect of tuning
and the equality of the peaks reflects the effect of optimal damping.
93
Table 1: Optimum parameters of TLCD under different PSDF of wind load
Mass Ratio µ=0.5% µ=1% µ=2%
Type of input spectrum Harris Davenport Kaimal Harris Davenport Kaimal Harris Davenport Kaimal
ξopt 0.27 0.13 0.15 0.57 0.28 0.3 1.24 0.59 0.7
U10=20 m/s
νopt 0.99 0.99 0.99 0.99 0.99 0.99 0.98 0.98 0.98
reduction in
rms
15.69 13.9 20.22 20.49 18.08 26.55 25.0 22.34 33.22
displacement
(%)
ξopt 0.16 0.08 0.1 0.35 0.16 0.22 0.74 0.35 0.45
U10=25 m/s
νopt 0.99 0.99 0.99 0.99 0.99 0.99 0.98 0.98 0.98
reduction in
rms
16.16 14.17 20.24 21.12 18.45 26.6 26.22 24.30 33.25
displacement
(%)
Figure 5: Displacement Transfer Functions of structure alone and with TLCD for optimized condition
5. CONCLUSIONS
Through this work, the applicability of TLCD as a wind induced vibration control device for a very high TLT,
for which resonant dynamic response is a major problem, has been investigated. A real life very high TLT has
been identified and modeled as a SDOF system. A formulation of the displacement transfer function of a
wind-excited viscously damped SDOF system equipped with TLCD has been developed in frequency domain
The wind force has been characterized by three standard PSDF, namely those of Harris, Kaimal and Davenport.
The rms displacement reductions due to the TLCD have been evaluated under various parametric variations.
The key points observed during the study have been summarized below.
94
• Substantial rms displacement response reduction of the order of 14%-33% can be achieved by the
TLCD in the present example case study.
• The TLCD performance is found to be highly sensitive to the tuning ratio, ν. Response reduction drops
sharply with detuning of the TLCD with the structural frequency. It may be suggested that ν should not
be allowed to vary outside a range of 0.95 – 1.05 for better control performance.
• Response reduction increases with increase in mass ratio. But from practical considerations, it is not
feasible to provide a very high mass ratio since that would increase the gravity load in the structure
significantly.
• Response mitigation by the TLCD also appears to increase with increase in length ratio of the TLCD.
However there is a threshold value above which if the value is raised it destroys the basic characteristics
of the TLCD.
Overall the TLCD is a promising wind induced vibration control device for high TLTs.
REFERENCES
[1] Balendra, T., Wang, C. M. and Cheong, H. F. (1995), Effectiveness of tuned liquid column dampers for
vibration control of towers, Engineering Structures, 17 (9), 668-675.
[2] Battista, R. C., Rodrigues, R. S. and Pfeil, M. S. (2003), Dynamic behavior and stability of transmission
line towers under wind forces, Journal of Wind Engineering and Industrial Aerodynamics, 91(8),
1051-1067.
[3] Blevins, Robert D. (1990), Flow-Induced Vibration, Van Nostrand Reinhold, New York.
[4] Chen, B., Zheng, J. and Qu, W. (2009), Control of wind-induced response of transmission tower line
system by using magnetorheological dampers, Int. J. of Structural Stability and Dynamics, 09 (4).
[5] Holmes, John D. (2007), Wind loading on structures, Taylor and Francis, USA.
[6] Iwan, W.D. and Yang, I. M. (1972), Application of statistical linearization techniques to nonlinear
multidegree-of-freedom systems, Journal of Applied Mechanics, 39, 545-550.
[7] Murtagh, P. J., Basu, B. and Broderick, B. M. (2005), Along-wind response of a wind turbine tower with
blade coupling subjected to rotationally sampled wind loading, Engineering Structures, 27, 1209–1219
[8] Newland, D. E. (1993). An introduction to random vibrations, spectral and wavelet analysis. Longman,
London, U.K.
[9] Soong, T. T. and Dargush, G. F. (1997), Passive energy dissipation systems in structural engineering, John
Wiley and Sons, New York.
[10] Saoka, Y., Sakai, F., Takaeda S. and Tamaki, T. (1988), On the suppression of vibrations by tuned liquid
column dampers, Annual Meeting of JSCE, Japan Society of Civil Engineers, Tokyo, Japan.
95
Paper No.: 018
STRESS ANALYSIS ARROUND SPIRAL CASING OF FRANCIS TURBINE OF

A HYDEL POWER HOUSE BY FINITE ELEMENT METHOD
1 2
Swetapadma Panda , Dr. Joygopal Jena and Bidyadhar Basa 3
1
Asst.Professor, Dept. of Civil Engineering, Institute of Technical Education and Research, Odisha, India
2
Professor, Dept. of Civil Engineering, Institute of Technical Education and Research, Odisha, India
3
Assoc. Professor, Dept. of Civil Engineering, Institute of Technical Education and Research, Odisha, India
Email: swetapadma@soauniversity.ac.in, joygopaljena@soauniversity.ac.in,
bidyadharbasa@soauniversity.ac.in
ABSTRACT
A Hydel Powerhouse structure can be broadly separated into three parts namely substructure, intermediate
structure and super structure. Substructure is a massive concrete structure embedding many hydro-mechanical
and electrical equipments such as spiral case, speed ring, control system for governing the flow of water, draft
tube and foundation for generator and rotor assembly. The present study is limited to the concrete around the
spiral casing. The spiral case which is connected to penstock at the end has a gradually reduced cross-sectional
area in order to distribute the water evenly. The geometrical shape of the spiral case is a complex one and it is
difficult to compute the stresses in the surrounding concrete which is essentially a three dimensional one. The
knowledge regarding the actual structural behavior of the spiral case is inadequate and prevalent design
practices are only approximate. It is therefore useful for the designer to know the true behavior of the
surrounding concrete in order to work out an economic and safe design.
For the three dimensional analysis of such complex shaped structure two tools are tools available; Finite
Element Method and Three-dimensional Photo Elasticity Method. P. S. Nigam (2) has analyzed similar structure
by using Photo elasticity method. In this study attempt has been made to compute the stresses in concrete
around spiral casing due to various loadings using Finite Element Method. The ANSYS CAE (Computer-Aided
Engineering) software was used in conjunction with 3D CAD (Computer Aided Design) solid geometry to
simulate the behavior of mechanical bodies under thermal / structural loading conditions. ANSYS automated
FEA (Finite Element Analysis) technologies from ANSYS, Inc. were used to generate the results. The
substructure model is assumed as a solid cube enclosed on all three sides except at penstock opening. The
analysis has been done for an equivalent uniformly distributed load (U.D.L) considering all total loads such as;
loads transferred from the super structure, self load of concrete, loads generator and stator including impact and
short circuit loads etc., acting on an annular area over the generator floor. Data considered by Nigam has
been considered for the present study and the results are compared with that of obtained by Nigam.
KEY WORDS: Concrete around Spiral Casing, 3D Stress analysis, FEA, Photo Elasticity Method
1. INTRODUCTION
Hydraulic turbines are the machines which use the Potential energy of water and convert into Mechanical energy.
The mechanical energy developed by the turbine is used in running an electric generator for generation of
electricity. One of the essential requirements of the hydro-electric power generation is the availability of
continuous source of water with a large amount of hydraulic energy. Such source of water may be made from a
reservoir formed by constructing dam across a river or by diverting a continuous flow through a water conductor
system. In order to keep the velocity of water constant throughout its path around the runner, the cross sectional
area of casing is gradually decreased. The water from penstock enters the scroll casing (spiral casing) which
completely surrounds the runner. The purpose of providing casing is to provide an even distribution of water
around the circumference of turbine runner, thus maintaining an approximating constant velocity for the water.
It provides a uniform distribution of water flow to the runner. The steel case is used for heads over 30 m,
designed for the maximum static head plus the pressure ring caused by water hammer. Large capacity and
high head cases are embedded in concrete for additional support and decreased vibrations. A vertical Francis
Turbine is generally placed with a vertical shaft located at an opening encircled by a spiral casing embedded in
concrete. The Fig.1 shows general arrangement of spiral casing in cross sectional plan and elevation.
Figure 1: Sectional Arrangement of Francis Turbine

Source: P.S. Nigam
1.1 Literature Review
P. S. Nigam (1985)2 has done three dimensional analysis of concrete around spiral case by photo
elastic method. Very few works have been done till now evolving exact mathematical analysis for
solving, 3-D structures, involving loads, geometrical discontinuities, theoretically this analysis could
be done using various analytical methods including FEM. E.C.Kalkani, (1995)3 worked on expected
displacement and stresses in encasing concrete of a Francis Turbine scroll case by using finite element
stress analysis. In his study, he has examine the minimum principal stresses in various cases which
were found to be compressive and has higher concentrations close to be stay ring supports. Wang
Hai-Jun et al (2007)7 has worked on stress analysis method for periphery concrete of steel spiral case
keeping internal water pressure. They have worked on the basis of supporting system analysis of steel
spiral case keeping internal water pressure during construction. Hegao Wu et al (2011)9 have worked
on structural analysis of the embedded spiral case in the “three gorge hydropower station”. Their
research performs a three dimensional non-linear finite element static analysis of the concrete structure
based on a damaged plasticity model to study the reliability of spiral case and the safety of the turbine
units under different loading coordinates. Wenliang Ma and Wuxing Chen, (2011)10 have studied on
joint load bearing of steel spiral case on surrounding concrete. His paper adopts FEM to carry out
97
simulating analysis for spiral case structure of Shuikou hydropower station.
1.2 Loads And Forces On Spiral Case
The scroll casing taken here is surrounded by plain cement concrete. The Fig. 2 shows the different locations of
load on the plan of scroll casing of the Francis Turbine.
C = Super structure column load.
W = Water load in scroll casing.
S = Generator startor load
R = Generator rotor load.
P = Load through speed ring.
Figure 2: Internal Plan of Spiral Case Showing The Relative Position Of Loads
2. METHODS OF ANALYSIS
Very few works are available in literature related to present study. It has already been stated that for such
complicated structure has no standard method of design practice is available and few approximate methods are
adopted. Methods employed for 3-D analysis are Photo elasticity technique and FEM.
2.1 Stress Analysis under Photo Elasticity Technique
Method employed for finding stress in 3-D model photo elasticity technique is stress freezing method. It is
based on the diphase behavior of polymer materials when heated. The polymeric materials are composed of long
chain hydrocarbon molecules some of which are well bonded into a 3-D network of primary bonds. However a
large number of molecules are less solidly bonded together into shorter secondary chains. At room temperature
both primary and secondary molecular bonds act to resist deformation due to applied load. However as the
temperature of the polymer is increased the secondary bonds breakdown and the primary bonds in effect carry
the entire applied load. Since the secondary bonds constitute a very large portion of the polymer, the deflections
which the primary bonds undergo are quite large yet elastic in character. If the temperature of the polymer is
lowered to room temperature while the load is maintained on the model, the secondary bonds will be reformed
between the highly elongated primary bonds and serve to lock them into their extended positions. When the load
is removed the primary bonds relax to a very modest degree but the main portion of their deformation is not
recovered. The elastic deformation of the primary bonds is permanently locked into the model by the reformed
secondary bonds. As these deformations are locked in on a molecular scale, the accompanying birefringence is
maintained in any small section cut from the original model. The cutting or slicing process may relieve the
molecular layer on each face of a slice cut from a model but this relieved layer is so thin relative to the thickness
98
of the slice that the effect is not observed.
2.2 Stress Analysis under Finite Element Analysis
The finite element method is a numerical procedure for analyzing structures, which permit the calculations of
stresses and deflections. The most distinctive feature of finite element method that separates it from others is the
division of a given domain into a set of sub-domains, called elements. The finite element procedure produces
many simultaneous algebraic equations which are generated and solved on a digital computer. Results are rarely
exact. However, errors are minimized by processing more equations, and results accurate enough for
engineering purposes are obtainable at reasonable cost. Using such elements, the structural idealization is
obtained merely by dividing the original continuum into segments, all the material properties of the original
system being retained in the individual elements. This is known as discretization. Instead of solving the problem
for entire body in one operation, the solutions are formulated at each constituent unit and combined to obtain the
solution for the original structure. A 4 noded tetrahedral element is shown in Fig 3. for a typical finite element ‘e’
defined by nodes i ,j, k, l. As it is a 3-dimensional element, so it will have three components of displacement, i.e.,
u, v, w in x, y, and z directions , respectively.
Figure 3: A 4 Nodded Tetrahedral Element

3. SOFTWARE USED
The ANSYS CAE (Computer-Aided Engineering) software was used in conjunction with 3D CAD (Computer
Aided Design) solid geometry to simulate the behavior of mechanical bodies under thermal / structural loading
conditions. ANSYS automated FEA (Finite Element Analysis) technologies from ANSYS, Inc. was used to
generate the results. The scenario presented here represents one complete engineering simulation. The definition
of a simulation includes known factors about a design such as material properties per body, contact behavior
between bodies and types and magnitude of loading conditions. The results of a simulation provide insight into
how the bodies may perform and how the design might be improved. Multiple scenarios allow comparison of
results given different loading conditions, materials or geometric configurations.
3.1 Processes in ‘ANSYS’
ANSYS program has many finite element analysis capabilities, ranging from a simpler linear, static analysis to a
complex, non-linear, transient dynamic analysis. The steps in ANSYS workbench 10.0 are as follows:
(i) Building the model
(ii) Define an element type
(iii) Define material properties
(iv) Meshing the model into elements & nodes
(v) Apply structural loads & structural support
(vi) Obtain solution
(vii) View the results in ANSYS Report.
99
3.2 ANSYS inputs
(i) Material : Concrete 2

(ii) Unit : SI Metric system
(iii) Convergence criteria : Not defined
(iv) Alert criteria : Not defined
4. THE MODEL
A 3-D model of the scroll casing has been prepared having the dimension according to Nigam( 4) .It is subjected
to a uniformly distributed load through a circular ring over the spiral case of the Francis Turbine to determine
the 3D stresses.
4.1. Assumptions in the Model
The material is within its elastic limit. The material is isotropic. The structural model is enclosed on all four
sides except the penstock opening. Shape of spiral case is as per U.S.B.R. Design. The diameter of the scroll
case is gradually reduced.
4.2 Properties of Encasing Concrete
Young’s Modulus of Elasticity of concrete 3.0 x 1010 Pa.Poisson’s Ratio of concrete 0.18. Compressive ultimate
strength of concrete 4.1 x 107 Pa.Density of concrete 2300.00 kg/m3 Tensile ultimate strength of concrete 5.0 x
106 pa.Thermal expansion of concrete is 1.4 x 10-5/c.Specific heat 780.0 J/Kg.0C.Thermal conductivity 0.72
W/m0C.Tensile yield strength and compressive yield strength 0.0 pa.
4.3 Description of the Model
The bounding box for the model measures (0.11 x 0.11 x 0.08) m oriented along the global X,Y & Z axes.
Total mass = 2.1 kg.Total volume = 9.15 x 10-4- m3. Reducing diameter of the penstock are 0.0259m,
0.0249m, 0.0206m, 0.0165m, 0.00975m respectively. The diameter of the stay vane 38.2 mm.
Figure 4: Structural Model and Its Sectional View with Inner Passage
100
5. LOADING CONDITIONS
The model was analyzed under one loading condition that is the generator load. So a uniformly distributed load
is considered to be acting through a circular ring simulating generator load. A 4 noded tetrahedral element
defined by nodes i, j, k, l is taken for the purpose of meshing the model. As it is 3-dimensional element, so it
will give three components of stress in X, Y, Z direction. The table below shows the details of generator (60MW)
load.
Table 1: Details of Generator load
1. Vertical load including hydraulic thrust at lower bracket level 510x6=3060 kN

2. Stationary part of generator 175 x 6 = 1050 kN
3. Wt. of stator pedestral 65 x 6 = 390 kN
4. Wt. of barrel 1450 kN
5. Wt. of sidewall of housing 840 kN
6. Reaction of floor 700 kN
Total 6900 kN
Beam area = 40.2 m2

Intensity of U.D.L./m2 = 6900/40.2 = 17.5 kN/ m2
So a U.D.L. of 0.18 Pa is applied through a circular ring over the model as shown in figure.
Figure 5: Meshed Model and Loaded Model with The U.D.
6. ANALYSIS OF RESULTS
The results obtained by the FEM model are compared with those obtained by Photo-Elastic Method. The shape
of the spiral case adopted is without column and the dimensions are as per U.S.B.R. criterion. The graphs
plotted below shows the variations of normal stress in X, Y, Z directions both in PEM by Nigam and FEA by
ANSYS Work Bench.
101
Table 2: Stress Values Obtained by FEM
Minimum Maximum Alert

Name Scope Orientation Minimum Maximum
Occurs On Occurs On Criteria
"Normal Stress" "Model" X Axis -0.05 Pa 0.04 Pa Solid Solid None
"Normal Stress 2" "Model" Y Axis -0.17 Pa 0.02 Pa Solid Solid None
"Normal Stress 3" "Model" Z Axis -0.05 Pa 0.03 Pa Solid Solid None
"Total
"Model" Global 0.0 m 2.35×10-14 m Solid Solid None
Deformation"
The results are presented in the form of stress contours at different time interval by ANIMATION.
0.6
0.5
0.4
0.3 PEM
0.2 FEM
0.1
0
1 2 3 4 5 6
Figure 6: Normal Stress in X-Direction Slice No. Vs SCF Curves By PEM & FEM
0.18
0.16
0.14
0.12
0.1
PEM
0.08
FEM
0.06
0.04
0.02
0
1 2 3 4 5 6 7
Figure 7: Normal Stress In Y-Direction Slice No. Vs SCF Curves By PEM & FEM
102
0.6
0.5
0.4
0.3 PEM
FEM
0.2
0.1
0
1 2 3 4 5 6
Figure 8: Normal Stress In Z-Direction Slice No. Vs SCF By PEM &FEM
7. CONCLUSION
The stress patterns obtained by FEM analysis for an uniformly distributed load through a circular ring of spiral
case are found generally comparable to the stress patterns obtained by Nigam(4 ). From Photo-Elastic
experimental method FEM is a more convenient and less time consuming tool for the analysis of complex
structure like spiral case. The 3-Dimensional stress patterns obtained by FEM analysis for the given loading
condition reveal following:
IN X-DIRECTION: The results of analyses by two methods, it is seen that both shows a closely similar stress
patterns in the concrete. In both of analyses the tension zone is developing or increasing in X-direction due to
the given loading condition.
IN Y -DIRECTION: From the e results of analyses by two methods, it is observed that stress patterns in the
vertical directions of the concrete differ from each other. However similarity occurs in the development of
compressive stress due to the given loading condition. The reason of variation of stress for a particular direction
i.e., in Y-direction, could not be ascertained. Further studies could be done for the same.
IN Z-DIRECTION: From the results of analyses by two methods, it is seen that both shows that the stress
patterns is in a reducing manner towards the central portions of the concrete block. So the analysis of normal
stress in Z-directions by both Photo-Elasticity and FEM yields similarly.
As Photo-Elastic Method is an experimental method, it is not possible to analyze the model under the combined
effect of loading and to see the total deformation. But FEM analysis, use of ANSYS Software enables us to
analyze the total deformation of the model under 3-Dimensional stress.
103
Figure 9: Total Deformation
REFERENCES
[1] George W & Arthur H.N. (1984), “Design of Concrete Structures” International Students Edition, Mc
Graw Hill Inc. New York.
[2] Nigam P.S (1985), Handbook of Hydro Electric Engineering, Nem Chand and Brothers, Roorkee.
[3] E.C. Kalkani (Nov. 1993), “Expected Displacements and stresses in the encasing concrete of a Francis
Turbine. Scroll casing. Civil Dept. National Technical University, Patission – 42, Athens 10682, Greece.
[4] A wickstrom (1997), Structural Analysis of Francis Turbine Runners using ADINA”, Kvaerner Energy,
N-0135, OSLO, Norway.
[5] O.P Gupta (1999), “Finite and Boundary Element Methods in Engineering” Oxford Publication, New
Delhi.
[6] P.N. Modi & Seth (2002) “Hydraulics and Fluid mechanizes and Hydraulic Machine, Standard Book
House, New Delhi.
[7] Wang Hai-Jun et al (2007), “Research on Stress Analysis Method for Periphery Concrete of Steel Spiral
Case Keeping Internal Water Pressure” China Rural Water and Hydropower
[8] G. Rmamurthy (2010) “Applied Finite Element Analysis, J.K. International Publishing House, New
Delhi.
[9] Hegao Wu et al (2011),”Structural analysis of the embedded spiral case in the three gorges hydropower
station”, Practice periodical on structural Design and Construction, ASCE
[10] Wenliang Ma and Wuxing Chen, (2011), “Study on joint load bearing of steel spiral case and
surrounding concrete”, applied mechanics and materials, Vol.101-102, pp.451-454.
[11] IS: 456-2000, “Code Of Practice For Plain And Reinforced Concrete”. Bureau of Indian Standards, New
Delhi.
[12] IS: 4247 (Part-III) – 1978, “Code of Practice for Structural Design of Surface Hydel Power Stations: Part
III Substructure”, Bureau of Indian Standards, New Delhi.
104
Paper No.: 019
INFLUENCE OF LOCATION OF STAIRCASE ON SEISMIC BEHAVIOUR OF

RC FLAT SLAB BUILDING
1 2
Benazir F. Ahmed and Kaustubh Dasgupta
1
Ph.D. Student, Dept. of Civil Engineering, Indian Institute of Technology Guwahati, Guwahati, India
2
Assistant Professor, Dept. of Civil Engineering, Indian Institute of Technology Guwahati, Guwahati, India
ABSTRACT :
This tem Flat plate system, composed of two-way slabs without beams possesses major advantages over the
conventional moment resisting frame such as architectural flexibility, lower building heights, easier formwork
and shorter construction time. A major concern with such a system is its susceptibility to punching shear failure,
generated by the combination of gravity loading and seismically induced unbalanced moment. Further the
location and configuration of staircase affect the strength and stiffness of the building by inducing static
eccentricity which leads to possible torsional effects in the building under the action of earthquake loading.
In the present study, a symmetric plan RC flat plate building is analyzed using the computer program SAP2000
using IS: 1893 (Part 1):2002 prescribed design load combinations. The columns are modeled using two-noded
frame elements and floor slab is modeled using four-noded shell element. For the staircase, the landing slabs
(mid-landing and floor landing) and the waist slabs are modeled using shell elements. The building is assumed
to be founded on rocky stratum, so all the degrees of freedom are restrained at the bottom nodes of the columns.
From linear static analysis results, it is observed that the unbalanced moments at the connections increase
significantly when subjected to earthquake load combinations. The maximum magnitude is found in either
1.5DL+1.5EL or in 1.2DL+1.2LL +1.2EL load combination. Also unbalanced moments go on increasing as we
move to an upper storey. The punching shear stress around the column vicinity increases significantly under
seismic action since unbalanced moments get transferred as eccentric shear stress at the punching shear critical
section. The corner columns, edge columns and the columns connecting the staircase are found critical. The
same trend is observed with varying positions of staircase though there is variation of magnitude of moments
across the building models.
KEYWORDS: Flat slab, Flat plate, Slab-Column connection, Unbalanced moment, Punching shear,
Column strip moment
1. INTRODUCTION
A Reinforced Concrete (RC) flat slab or flat plate building consists of slab and columns without beams. Under
the combined effect of gravity loading and lateral earthquake forces, unbalance moment is generated at the
slab-column junction. This may lead to the punching shear mode of failure and can further lead to progressive
collapse of these types of structures (Robertson and Johnson, 2006). As the loads are directly transferred to the
supporting columns in such a building, there is a tendency of punching shear failure at the slab-column
connection. Hence, drop panels (i.e., a projection below the slab around the column) and column capital (i.e.,
the flared portion of the column at its head as it meets the floor slab or drop panel) are provided to achieve
adequate strength in punching shear. If no drop panel or column head is present, the system is referred to as flat
plate rather than flat slab (Wight and MacGregor, 2009).
Various codes of practice (IS:456-2000, 2000; ACI:318-08, 2008) prescribe different methods of analysis of flat
plate buildings, namely (a) direct design method, (b) equivalent frame method, (c) finite element method etc. In
direct design method, the slab is divided into a certain number of panels and the bending moments are computed
in each panel along each direction. In equivalent frame method, an equivalent plane frame is considered for
analysis of the building along a particular direction by taking the stiffness contribution of slabs. The structures
having irregular types of plans with which the equivalent frame method has limitations in analysis can be
analyzed without any difficulties by the finite element method (Wight and MacGregor, 2009).
The incorporation of staircase in a building plan causes a concentration of lateral stiffness at a particular
location; this leads to a shift of center of stiffness at any particular floor level. Consequently, torsional effects in
the vertical members may lead to undesired failure mechanisms during strong earthquake shaking. The
possibility of failure also depends on the location of staircase and configuration of the building. In the past
studies, the effect of location of staircase on the seismic response of a flat plate building has not been
investigated.
2. UNBALANCED MOMENT
Unbalanced moment is the bending moment occuring at the slab-column junction due to unequal spans or
unequal loading on either side of the junction (Alexander and Simmonds, 1987) (Figure 1(a)). Moreover when a
flat slab structure resists lateral forces, there is transfer of moment at every connection (Hawkins et al., 1989)
(Figure 1(b)). Under seismic action, when the columns undergo drift, an extra unbalanced moment is generated
at the slab-column connection due to the shift of the vertical load transmitted through the columns. The
unbalance moment is transferred in two ways, namely (a) as flexure over the column width (c) plus three times
the slab thickness (h) and (b) as eccentric shear at a critical section located at a distance of d/2 from the column
face, where, d is the effective depth of the slab. Thus unbalanced moment transfer through flexure is Mf =γf Munb
and unbalanced moment transfer through eccentric shear is Mv = (1- γf) Munb = γv Munb, where, Munb is the
unbalanced moment to be transferred at the connection, γf the fraction of the unbalanced moment to be
transferred by direct flexure [=1/{1+2/3×(b1/b2)0.5}] with b1 and b2 the total width of the critical section
measured perpendicular to the axis about which the moment acts and total width parallel to the axis respectively
and γv the fraction of the moment transferred by eccentric shear (Kang and Wallace, 2005).
3. PUNCHING SHEAR
Punching shear is a type of failure of RC slabs subjected to high localized forces. In flat slab structures, this
occurs at column support points. Flat plates exhibit higher stress at the column connections and are likely to fail
due punching shear rather than flexural failure. It is to be noted that the punching shear failure may occur due to
concentration of shear forces and the combination of unbalanced bending and twisting moments.
Such a concentration of shear force and moments leads to asymmetrical stress distribution around the column
plate connections (Figure 2). In the vicinity of a column support, a two-way concrete plate must carry transverse
shear resulting from gravity load and an additional unbalanced moment. For design, ACI 318-08 limits the
calculated maximum shear stress on a critical section located at a distance half the slab effective depth from the
face of the support. Both the vertical shear and a fraction of the unbalanced moment are assumed to contribute
to shear stresses on the critical section which varies linearly over the critical section. The magnitude of the shear
stress vu at any point on the critical section (Varghese, 2005; Choi, 2005; Elgabry and Ghali, 1996) is given by
vu = Vu/bod + γvx Munbx· ey/Jcx + γvy Munby· ex/Jcy (1)
where, Vu is the direct shear due to gravity load on the critical section, b0 is the width of the section, d is the
effective depth of the section, γvx and γvy are the fractions of unbalanced moment transferred by eccentric shear
about the principal x and y direction respectively, Munbx and Munby are the unbalanced moments along x and y
directions respectively, ex and ey are the distance between a point on the critical section and its centroid in the
principal x and y direction respectively and Jcx and Jcy are the polar moments of inertia of the section along x and
y directions respectively.
106
(a)
(b)
Figure 1 Development of unbalanced moment at the slab-column connection due to (a) gravity
loading (Alexander and Simmonds, 1987) and (b) seismic loading (Hawkins et al., 1989)
Figure 2 Resultant shear stress plot due to vertical load and moment transfer (Hawkins et al., 1989)
4. MODELING AND ANALYSIS OF FLAT PLATE STRUCTURE
The finite element program chosen for the linear static analysis study was SAP2000 V 15.0.1 (CSI, 2011) and
the entire building was modeled. Every floor slab was divided into column strip, middle strip and panel. Column
strip is a design strip with a width on each side of the column centerline equal to 0.25l2 or 0.25l1, whichever is
less where l1 is the span length or center to center distance between the supports, in the direction of the moments
to be determined and l2 is the span length or center to center distance between supports in the direction
perpendicular to l1. The middle strip is a design strip bounded by two column strips. A panel is the portion of the
slab enclosed by the column centerlines at the four corners (Ahmed, 2013).
107
4.1. Modeling
4.1.1 Flat slab-Column model- Geometric Idealization
The chosen prototype is that of a hypothetical residential building with flat slabs of 200mm thickness and the
columns of 400mm × 400mm square section in each floor with a regular arrangement in the upper storeys. The
floor plan and elevation dimensions are shown in Figure 3. To investigate the influence of the location of
staircase on the building response, the locations were varied to create four different models (Figures 4 and 5).
3.04m
1.52m
3.04m
3.04m
3.20m 3.04m
3.04m
3.04m
3.20m
3.04m
3.04m
1.52m 1.50m
1.60m 3.20m 2.74m 3.20m 1.60m 1.60m 3.20m 2.74m 3.20m 1.60m
(a) (b)
Figure 3. Five-story flat –slab building - (a) elevation (b) plan
(a) (b)
Figure 4. Locations of staircase in the plans of (a) Model 1 and (b) Model 2
108
(a) (b)
Figure 5. Locations of staircase in the plans of (a) Model 3 and (b) Model 4
4.1.2 Finite Element Modeling
For the buildings, the floor slabs and columns are modeled using four noded thin shell elements and two noded
frame elements respectively in SAP2000 program. As the buildings are assumed to be founded on rocky stratum,
all the translational and rotational degrees of freedom at the bottom level nodes are restrained. Thus, soil-
structure interaction is not considered in the analysis. The brick masonry infill wall is modeled by an equivalent
pin-jointed strut made of the same material having width as the thickness of the infill and depth as one-fourth
the diagonal length of the infill. The brick masonry wall is assumed to have negligible resistance against tensile
cracking and hence the wall is modeled to carry only axial compressive force. The ends of the strut element are
released against generation of shear force and bending moment (Kaushik et al., 2009). The staircase component
comprises of two landings one at each of the floor levels, a mid-landing at mid-level of the storey and two waist
slabs. The two floor landings and the mid-landing are directly supported on the columns. The width of the stair
flight is 1.6m. In each flight, there are ten steps with the riser being 150mm and the tread being 300mm. The last
rise of the stair is considered as 200mm so as to match the total height of the rises with the storey height.
For incorporating the brick masonry infill load along the four edges (covering the peripheral columns) on the
flat plate, strips of 115mm are created on either side of the column point and continued throughout the periphery
(Ahmed, 2013). The infill load then as a uniformly distributed load is placed on the created peripheral strip of
total thickness 230mm. The thin shell elements (flat plate slab) are also meshed for obtaining unbalanced
moments, based on the calculated Column Strips on either side of the column points.
4.1.3. Loads
Apart from the self weight, imposed loads of intensities 4.5 kPa and 1.5kPa are considered on floor slabs and
roof respectively as per the Indian Standard IS:875 (IS:875 (Part 2), 1987). In additional to the above vertical
loading, lateral earthquake forces are also considered along the two orthogonal directions as per IS 1893
(IS:1893 (Part 1), 2002) using equivalent static method. All the four models are analyzed and results obtained
for four code-prescribed load combinations, namely (a) 1.5(DL +LL), (b) 1.5(DL+EL), (c) 1.2(DL+LL+EL) and
(d) 0.9DL +1.5EL, where DL, LL and EL refer to dead load, live load and earthquake load respectively.
109
4.2 Analysis and Results
In the building models, the corner slab-column junctions and the slab-column junctions supporting the staircase
are investigated to identify the critical regions for the transfer of maximum unbalanced moment and the
maximum punching shear stress. The column strips moments (moments along the width and length of the
building) are evaluated at each column point. For estimating unbalanced moments, the moments on either side
of the centerline of a particular column are observed and the resultant moment is reported as the unbalanced
moment (Ahmed, 2013).
In the mentioned regions of the building models, it is observed that the unbalanced moments significantly
increase under earthquake loading as compared to those due to gravity loading only. The highest magnitude is
found to be under either combination (b) or (c) as reported in Section 3.1.3. At interior column-slab junctions
around the staircase, however the maximum unbalanced moment is obtained under combination (d), though the
magnitude is low. Also the magnitudes of unbalanced moments increase with the elevation of the floor slab.
This is possibly due to the larger magnitude of lateral force applied at the upper floor levels which lead to larger
unbalanced moments locally. For all the models, the corner slab-column junctions show larger unbalanced
moments as compared to the staircase column positions. In models 3 and 4 with the orientation of the staircase
along the length of the building, the unbalanced moments are quite low under earthquake loading along the
same direction; this is due to the additional lateral stiffness of the model mobilised from the bracing action of
the staircase.
The unbalanced moment gets transferred as eccentric shear stress at the critical section for punching shear; this
adds up with the punching shear due to gravity loading and leads to increase of total punching shear stress under
seismic loading. Using Equation (1), the punching shear stress is obtained at the slab-column junctions at all the
floor levels for earthquake loadings both along length and width of the building. The total punching shear stress
also depends on the critical section. At the corner column position, full peripheral dimension of the critical
section is available (the distance being considered as column dimension + effective depth/2 on all four sides of
the column face) for the transfer of load. However at the slab-column junctions around the staircase, the full
dimensions of the critical section are not available. This leads to lesser polar moment of inertia of the section as
compared to the corner junctions. Thus, the resulting punching shear stress is quite high at these regions
although the unbalanced moments are less.
Among all the models, punching shear stress is observed to be the maximum in Model 4 for seismic loading
along the length of the building. Figure 6 shows variation of punching shear stress against the floor slab
elevation (ground to fifth floor) for corner slab-column junction and slab-column junction around the staircase
in Model 1. Thus, the maximum value is obtained for EQ-X case which represents the seismic load along the
length of the building. In case of staircase columns, even gravity loading (GL case) mobilised more punching
shear stress as compared to the earthquake load along the width of the building (EQ-Y case). Among all the
models, the maximum punching shear stress was obtained 1.85 MPa (Ahmed, 2013) and this exceeded the
code-prescribed punching shear strength for M25 concrete (IS:456, 2000). Thus the flat plate is vulnerable to
possible punching shear failure.
4. CONCLUSIONS
The following salient conclusions are drawn from the analysis results:
1. Unbalanced moments show a significant increase in magnitude when subjected to seismic loading. The
magnitude of the moment increases with the height of the floor.
2. The corner slab-column connections generate higher magnitude of unbalanced moment as compared to the
moment generated for the slab-column connections around the staircase. This leads to high punching shear
stress at the corner slab-column connections.
3. Although the slab-column connections around the staircase carry lower values of unbalanced moments, high
punching shear stresses are generated due to lower polar moment of inertia at the critical section of such
connections.
110
EQ-X
Punching Shear Stress (MPa)

Punching Shear Stress (MPa)
EQ-Y
GL
Floor Level Floor Level

(a) (b)
Figure 6. Punching shear stress distribution for a (a) Corner column (b) Edge column supporting the
staircase in Model 1 (Ahmed, 2013).
4. Due to larger lateral stiffness of the building along its length, as compared to the stiffness along width, the
unbalanced moments evaluated along length are of lower values as compared to those along width. The same
comparison holds good for punching shear stress also.
5. For the orientation of staircase along the length of a building, the unbalanced moments and punching shear
stresses (along the direction of staircase) are observed to be less due to the bracing action of the staircase.
6. The flat plate is observed to be vulnerable against possible punching shear failure as the punching shear stress
demand at many connections is found to exceed the slab punching shear strength.
ACKNOWLEDGEMENT
The support and resources provided by Department of Civil Engineering, Indian Institute of Technology
Guwahati and Ministry of Human Resources and Development, are gratefully acknowledged by the authors.
REFERENCES
[1] Robertson, I., and Johnson, G. (2006). Cyclic Lateral Loading of Nonductile Slab-Column Connections.
ACI Structural Journal, 103 (3), 356-364.
[2] Wight, J. K., and MacGregor, J. G. (2009). Reinforced Concrete, Mechanics and Design. Fifth Edition,
Pearson Education, Inc.
[3] IS:456-2000 (2000). Indian Standard Plain and Reinforced Concrete – Code of Practice (Fourth Revision).
Bureau of Indian Standards, New Delhi, India.
[4] ACI:318-2008 (2008). Building Code Requirement for Structural Concrete (ACI 318-08) and Commentary.
American Concrete Institute, Farmington Hills, Michigan, USA.
[5] Alexander, S. B. and Simmonds, S. H. (1987). Ultimate Strength of Slab-Column Connections. ACI
Structural Journal, 84 (3), 255-261.
[6] Hawkins, N. M.; Bao, A. and Yamazaki, J. (1989). Moment Transfer from Concrete Slabs to Columns. ACI
Structural Journal, 86(6), 705-716.
[7] Kang, T. H. K. and Wallace, J. W. (2005). Dynamic Responses of Flat Plate Systems with Shear
Reinforcement. ACI Structural Journal, 102 (5), 763-773.
111
[8] Varghese, P. C. (2005). Advanced Reinforced Concrete Design. Second Edition, PHI Learning Private
Limited, New Delhi.
[9] Choi, J.-W. (2005). A Method for Effective Beam Widths of Slabs in Flat Plate Structures under Gravity
and Lateral Loads. Structural Engineering and Mechanics, 21(4), 451-468.
[10] Elgabry, A. A. and Ghali, A. (1996). Moment Transfer by Shear in Slab-Column Connections. ACI
Structural Journal, 93(2), 187-196.
[11] CSI, (2011), SAP2000 Analysis Reference SAP Manual, Computers and Structures Inc., California, USA.
[12] Ahmed, B. F. (2013). Investigation of Seismic Behaviour of RC Flat Plate Building with Staircase by Static
Analysis. Master of Technology Project Report, Indian Institute of Technology Guwahati, Guwahati, India.
[13] IS:875 (Part 2)-1987 (1987). Indian Standard Code of Practice for Design Loads (Other than Earthquake)
for Buildings and Structures Part 2 Imposed Loads (Second Revision). Bureau of Indian Standards, New
Delhi, India.
[14] IS: 1893 (Part 1)- 2002, (2002), Indian Standards Criteria for Earthquake Resistant Design of structures,
Bureau of Indian Standards, New Delhi.
[15] Kaushik, H. B.; Rai, D. C. and Jain, S. K. (2009). Effectiveness of Some Strengthening Options for
Masonry-Infilled RC Frames with Open First Storey. Journal of Structural Engineering ASCE, 135(8), 925-937.
112
Paper No.: 020
INFLUENCE OF STAIRCASE AND ELEVATOR CORE ON TWISTING

BEHAVIOUR OF RC WALL-FRAME BUILDINGS
1 2
Arijit Acharjya and Kaustubh Dasgupta
1
Assistant Professor, Dept. of Civil Engineering, Birbhum Institute of Engineering and Technology, Siuri, India
2
Email: acharjya.arijit@gmail.com, kd@iitg.ac.in
ABSTRACT:
Staircase and elevator core are important components of typical multi-storeyed Reinforced Concrete (RC)
moment-resistant frame building. The location and configuration of staircase and elevator core affect the
strength and stiffness of the building, and these affect its seismic behaviour during earthquake shaking. The
change in location of staircase and elevator core leads to possible torsional effects in the building during
earthquake shaking.
In the present study, a symmetric plan RC frame building is analysed using the computer program SAP2000
under the design load combinations as per Indian Earthquake Code IS:1893 (Part 1):2002. The beams and
columns are modelled using two-nodded frame elements and floor slab is modelled using four-nodded shell
element. The elevator core wall and the slab components of the staircase are modelled using shell elements. The
locations of staircase and elevator core wall are changed across the same building plan. The variation of
maximum forces and moments in the frame members for the different building models was obtained from the
linear static analyses. Accordingly, the locations of the critical frame members were also obtained. The centre of
stiffness of the building, considering the staircase, was determined by providing equivalent shear walls along the
staircase bays; the dimensions and location of the shear walls were obtained through consistent deformation
method.
Due to the presence of staircase and elevator core, the change in the torsional behaviour of the building models
was computed. Subsequently, the use of shear wall in resisting the twisting effects of the building was illustrated.
The moments in the staircase beam, due to dead and live loads only, were compared with the moments obtained
during the twisting action under earthquake forces. The lateral bending of staircase beam, due to earthquake
loads, is also determined and this needs to be considered for its design.
KEYWORDS: Staircase, Elevator core, RC wall, Plan asymmetry, Center of stiffness
1. INTRODUCTION
In a multistoried RC frame building, the staircase is generally cast monolithically with the structural frame
system. By virtue of its configuration, it behaves similar to a stiff truss attracting large lateral forces. When a
building is subjected to seismic forces in the direction of the flight of stairs, the stairway behaves analogous to a
diagonal strut. This aspect is generally not considered in design of staircase components, e.g., the waist slab of
the staircase is generally considered as a simply-supported beam (Pillai and Menon, 2009). As the staircase is a
stiff element and the present of staircase make building plane unsymmetrical, twisting behavior gets induced in
the building. In a multistoried building, the RC structural wall or shear wall can be positioned in a way such that
it can contribute to an efficient lateral force resisting system as well as minimise the twisting effects (Acharjya,
2013).
During earthquake shaking, the lateral loads on buildings with staircase and elevator core are resisted by the
combined action of moment resistant frames, staircase and RC wall of elevator core. Due to large lateral
stiffness, the staircase components attract large forces. Various degrees of damages have been observed in the
different components (steps, landing beams etc.) of staircases during past earthquakes, particularly (a) 2008
Wenchuan earthquake (Li and Mosalam, 2012) and (b) 2011 Sikkim earthquake (Kaushik and Dasgupta, 2011).
Also, due to mobilisation of large forces and moments in the frame members around the staircase, different
modes of ductile and brittle failure (e.g., short column effect due to the presence of mid-landing) have also been
observed in those members.
In buildings with RC walls around elevator core, separate shear walls may not be required to resist the large
lateral shear force during strong earthquake shaking. Although there have been past studies on seismic
behaviour of multistoried buildings with elevator core walls (Munir and Warnitchai, 2012; Adebar et al., 2007),
the influence of varying positions of staircase and elevator core together have not been carried out. In the
present study, the locations of staircase and elevator core are changed in the same plan of the building and
nonlinear static analysis is carried out to obtain a comparative behaviour of the building models. The level of
asymmetry caused by the presence of both staircase and elevator core is also investigated.
2. MODELING AND BUILDING DETAILS
A hypothetical five-storied building, with symmetric plan, is considered with plan dimensions of 15.52 m × 9.14
m and a height of 16.7 m (Figures 1a and 1b). The floor-to-floor height is taken as 3.04 m and the columns are
extended 1.5 m below the ground level. All the columns and floor beams have the cross sections as 300 mm ×
300 mm and 225 mm × 300 mm respectively. The thickness of RC wall and floor slab are assumed to be 200
mm and 125 mm respectively. The grades of concrete and reinforcement are M25 and Fe415 respectively. The
building is assumed to be located in Guwahati city under Seismic Zone V as per the Indian earthquake code IS:
1893 [IS:1893 (Part 1), 2002]. The building model, without staircase and elevator core, is named as Model-A
A B C D E F 3.04m
4
3.20m 3.04m
3
2.74m 3.04m
2
3.20m 3.04m
1
3.04m 3.20m 3.04m 3.20m 3.04m
3.04m
1.5m
3.20m 2.74m 3.20m
Figure 1: Views of Model-A: (a) typical floor plan and (b) typical elevation along width of the building
The staircase is positioned at four different locations across the same building plan (Model-A), and they are
named as Model-B, Model-C, Model-D and Model-E (Acharjya, 2013). In each of the models, an elevator core
is positioned adjacent to the staircase, and the models having both staircase and elevator core are named as
114
Model-1, Model-2, Model-3 and Model-4 respectively (Figures 2a, 2b, 2c and 2d). Further, two RC walls are
also provided between gridlines C and D at the outer edge of the building and along gridlines 1 and 2 along the
length of the building. Again all models are modified without changing the positions of staircase and elevator
core; new RC walls are introduced with length and thickness such that the overall twisting behaviour of the
building gets reduced. After adding the new RC wall in the 8 models, as mentioned earlier, they are named as
Model-BM, Model-CM, Model-DM, Model-EM, Model-1M, Model-2M, Model-3M and Model-4M
respectively.
A B C D E F A B C D E F
4 4
3.20m
3 3
2.74m
2 2
3.20m
1 1
3.04m 3.20m 3.04m 3.20m 3.04m 3.04m 3.20m 3.04m 3.20m 3.04m
(a) (b)
A B C D E F A B C D E F
4 4
3.20m
3 3
2.74m
2 2
3.20m
1 1
3.04m 3.20m 3.04m 3.20m 3.04m 3.04m 3.20m 3.04m 3.20m 3.04m
(c) (d)
Figure 2: Typical floor plans of (a) Model-1, (b) Model-2, (c) Model-3 and (d) Model-4.
2.1. Modeling
Using the computer program SAP2000 (CSI, 2011), the beams and columns are modeled as frame elements and
the slabs as shell elements. Rocky stratum is assumed to be present at the bottom of the building. Thus, all the
six degrees of freedom (DOFs), namely (a) three translational DOFs (u1, u2, u3) and (b) three rotational DOFs
(r1, r2, r3) are restrained at the bottom nodes of base columns.
115
For the staircase, the waist slab, floor landing slab and mid landing slab are modeled as shell elements. To
determine the center of stiffness of the building with staircase, the staircase is replaced by RC wall along both
the directions to match the lateral stiffness. Although, difference in the nodal values of lateral displacement is
observed during substitution of staircase with equivalent RC walls, the lateral stiffness equivalence could be
established based on similarity in the overall response of the models.
Modeling of elevator core: The RC walls of the elevator core are modeled as shell elements. The column frame
elements (i.e., boundary elements of core walls) and the beam frame elements are also discretized along with the
shell elements so that proper connectivity and load transfer mechanism are mobilised at the connection nodes.
All the shell element nodes at the bottom of the wall are restrained against all the six degrees of freedom.
Modeling of RC wall: The torsion-resistant RC walls are modeled using equivalent frame elements (Agarwal
and Shrikhande, 2006). It is connected with adjacent frame with stiff beam. Floor landing and slab are
discretized at the junction of the shear wall and slab. All the six degrees of freedom are restrained at the bottom
nodes of base shear wall.
Load distribution on model: Apart from the self-weight, the building is also subjected to the finishing load and
live load. Live load in floor, Roof Floor, stair slab, waist slab, elevator car are 1.5, 0.75, 3, 3, 5.44 respectively
and dead load of stair slab, waist Slab, elevator car due to floor finish 0.6,1.5, 7 respectively. They are taken
from as per IS 875-1987 [7]. In additional to the above vertical loading, the system is analyzed against
earthquake loading along the two orthogonal directions (Global X and Global Y) as per IS 1893 (Part 1): 2002
[1] using Equivalent Static Method for different load combination.
3. ANALYSIS AND RESULTS
The influence of the combined action of staircase and elevator core is reflected in the comparative plot of nodal
displacements along X and Y directions (Figure 3) and floor rotation about the vertical axis is shown in Figure 4.
From Figure 3, it is observed that due to the presence of staircase and elevator core, the displacement gets
reduced due to increase of the building stiffness. Also, the presence of elevator core in building model changes
the direction of floor rotation about vertical axis (Figure 4). So by proper positioning of elevator core or by
addition of extra RC wall, the torsional behaviour can be possibly reduced.
Location of core wall

To obtain the centre of stiffness of a particular floor, the model is discretized into bays along grid lines 1 to 4
and A to F as explained earlier for Model-A (Figure 1). Using the lateral stiffness of individual frames, the
distances of centre of stiffness of the building from the X and Y axes are obtained as,
6
∑ SPF × D
n =1
n
CS-X = (1)
TS − X
7
∑ SPF × D
n =1
n
CS-Y = (2)
TS − Y
where, CS-X and CS-Y are the distances of centre of stiffness from Y and X axes respectively, SPF is the
stiffness of the portal frame, Dn the distance of the nth frame from the main axis and TS-X and TS-Y the total
stiffness of the building along X and Y directions respectively.
To obtain the lateral translational stiffness of the elevator core wall, the nodal rotations about the horizontal axes
are restrained at appropriate locations. For minimising the torsional effect, the revised locations of centre of
mass and centre of stiffness are obtained as,
DLC × CMY + EDCw × VCs CwY

Distance of centre of mass from Y axis (m) = (3)
DLC + EDCw
116
Maximum Displacement (m)
(a)
Maximum Displacement (m)
(b)
Figure 3: Comparison of maximum lateral displacements for earthquake loading along (a) X direction and (b) Y
directions respectively
Floor rotation about vertical
axis (rad)
Figure 4: Comparison of rotation of floor about vertical axis for earthquake loads along X and Y directions.
DLC × CMX + EDCw × VCs CwX

Distance of centre of mass from X axis (m) = (4)
DLC + EDCw
SYC × CsYC + C × (VCs − 1.52) + d × (VCs + 1.52)
Distance of centre of Stiffness from Y axis (m) = (5)
SYC + c + d
SXC × Cs XC + C × (VCs − 1.37 ) + d × (VCs + 1.37 )
Distance of centre of Stiffness from X axis (m) = (6)
SYC + c + d
117
where, DLC is the dead load of Model-C, CMX and CMY the distances of centre of mass from X and Y axes
respectively, EDCw effective dead load of core wall, SXC and SYC the stiffness of core wall along X and Y
directions respectively for Model-C, CSXC and CSYC the distances of centre of stiffness from X and Y axes
respectively, VCsCwY and VCsCwX the distances of variable centre of stiffness of core wall along Y and X axes
respectively and VCs the variable stiffness of core wall.
Although the perimeter of the shifted core wall gets reduced, mostly the horizontal nodal displacements are
observed to be low (Table 1). Also, the nodal rotation about the vertical axis gets significantly reduced in all the
cases. As the elevator is always located adjacent to the staircase, an additional rotation of the floor is always
observed. So to counter the rotational effect, a better alternative may be the provision of additional RC structural
walls.
Table 1: Improvement in behaviour due to change in location of core wall
Output Improvement in %
Joint
Case U1 U2 R3
95 EQ-X -12.5 54.8 59.3
126 EQ-X -12.7 64.4 60.7
211 EQ-X -43.4 55.9 60.2
252 EQ-X -43.5 64.6 62.3
For every model, the RC wall is provided in such way so that the nodal rotation about the vertical axis is
minimised as well as the symmetry is maintained. This is achieved locating the centers of mass and stiffness as
close as possible. Other locations of shear wall used in various models are along gridlines A, F, 1 and 4 between
grid lines 2-3 and C-D respectively. The required stiffness of shear wall is obtained as,
7 2
∑
0
SF × Dn + ∑ CS × D − CM × (TSF − TCS )
1
X= (7)
CM − L
where, X is the required stiffness, SF the stiffness of the portal frame, n the number of portal frame; Dn=
distance of nth frame from the axis, CS the core stiffness, CM the distance of centre of mass, TSF the total
stiffness of the frame L the length of shear wall from the axis and TCS the total core stiffness.
The dimensions of the RC wall are changed by trial-and-error method so that the lateral stiffness remains
approximately the same after addition of the wall. Suitable modeling aspects are taken care during the
mentioned procedure. The lengths of RC wall for various models are shown in Table 2. As exact matching of
displacement and rotation are not possible, the small differences are tabulated in Table 3. With the provision of
the additional wall, the displacement and rotation are reduced in all the cases.
Table 2: Length of RC wall for different models
Model Name Direction Original R3 resisting Total shear wall
Model (m) shear wall length (m)
Model-1 Along X 6.08 2.00
5.50
Along Y 0 3.50
7.58
Along Y 0 4.50
Model-3 Along X 6.08 0
2.50
Along Y 0 2.50
9.00
Along Y 0 4.00
118
Table 3: Difference in Displacement and Rotation for Different Models
Model-1 with RC wall Model-1 without RC wall
Joint Load
U1 U2 R3 U1 U2 R3
Location case
(m) (m) (rad) (m) (m) (rad)
EQ-X 0.0059 0.0000 0.0001 0.0073 0.0010 0.0002

Joint 1-F
EQ-Y -0.0006 0.0040 0.0000 -0.0028 0.0032 -0.0005
EQ-X 0.0038 0.0002 0.0000 0.0038 0.0002 0.0000

Joint 1-F
EQ-Y 0.0000 0.0018 0.0000 0.0000 0.0018 0.0000
EQ-X 0.0042 0.0000 0.0000 0.0042 0.0000 0.0000

Joint 1-F
EQ-Y 0.0003 0.0040 0.0000 0.0003 0.0040 0.0000
4. CONCLUSIONS
The following salient conclusions are drawn from the present study:
(a) Presence of staircase reduces the peak shear force and bending moments in the frame members of the
building. Mostly the maximum shear force and bending moment occur in the members adjacent to the
elevator core wall or staircase.
(b) The twisting behaviour of the building may be reduced by properly locating the elevator core wall and the
staircase in the building plan.
(c) Provision of RC wall of adequate length and at appropriate location in the plan of the building leads to
reduction in overall twisting behaviour of the building. It also improves lateral shear capacity considerably.
5. ACKNOWLEDGEMENT
REFERENCES
Pillai, U., and Menon, D. (2009). Reinforced Concrete Design. 3rd Edition, Tata McGraw-Hill.
Acharjya, A. (2013). Investigation of Seismic Behavior of RC Wall-Frame Building with Staircase and Elevator
Core by Static Analysis. Master of Technology Project Report, Indian Institute of Technology Guwahati,
Guwahati, India.
Li, B. and Mosalam, K. M. (2012). Seismic Performance of Reinforced Concrete Stairways during the 2008
Wenchuan Earthquake. Journal of Performance of Constructed Facilities, ASCE, published online in August
2012.
Kaushik, H. B. and Dasgupta, K. (2012). Observations on Performance of Structures during 18 September 2011
Sikkim (India) Earthquake. Proceedings of 15th World Conference on Earthquake Engineering, Lisbon, Portugal,
Paper No. 5281.
119
Munir, A. and Warnitchai, P. (2012). The Cause of Unproportionately Large Higher Mode Contributions in the
Inelastic Seismic Responses of High-Rise Core-Wall Buildings. Earthquake Engineering and Structural
Dynamics, 41(15), 2195-2214.
Adebar, P., Ibrahim, A. M. M. and Bryson, M. (2007). Test of High-Rise Core Wall: Effective Stiffness for
Seismic Analysis. ACI Structural Journal, 104(5), 549-559.
IS: 1893 (Part 1)- 2002, (2002), Indian Standards Criteria for Earthquake Resistant Design of structures, Bureau
of Indian Standards, New Delhi.
CSI, (2011), SAP2000 Analysis Reference SAP Manual, Computers and Structures Inc., California, USA.
Agarwal, P., and Shrikhande, M. (2006). Earthquake Resistant Design of Structures. Prentice Hall of India.
IS:875 (Part 2)-1987 (1987). Indian Standard Code of Practice for Design Loads (Other than Earthquake) for
Buildings and Structures Part 2 Imposed Loads (Second Revision). Bureau of Indian Standards, New Delhi,
India.
120
Paper No.: 021

RANDOM VIBRATION ANALYSIS OF RCC STRUCTURE USING PSEUDO-

EXCITATION METHOD
P. Bhattacharya1 and S. Karar2

1
Associate Professor, Dept. of Civil Engineering, Jadavpur University, Kolkata, Kolkata-700032
2
M. E. Scholar, Dept. of Civil Engineering, Jadavpur University, Kolkata, Kolkata-700032
ABSTRACT:
An efficient random vibration algorithm, the pseudo-excitation method (PEM), is developed for the random
vibration analysis of seismic responses of structures. A finite element model of RCC beam and beam column
sub-assemblage is developed in MATLAB environment. Plane stress elements are used to model the concrete
and one dimensional bar elements are used to simulate the reinforcement in both the structure. The beam is
subjected to a random load having constant PSD. A static and dynamic analysis of the beam is done and
compared with the analytical solution. The PSD of the deflection of the beam under pseudo excitation is
compared with the same under classical random loading.
In case of the beam column assembly the stress analysis is done under static load applied on the beams. The
stresses developed are compared with the analytical solutions. Graphical representation of failure pattern in the
linear range is observed using Von-Mises failure criteria. To simulate the actual behaviour of a building, masses
of the upper stories and lateral stiffness of the columns of 1st storey are lumped at the appropriate locations of
the beam column sub-assemblage. Support acceleration having PSD based on the Kanai-Tazimi model is
applied at the base. The first natural frequency of this model is compared with a lumped mass model of same
storey high. The stresses in different region of the joint are determined under the support acceleration.
KEYWORDS: PEM, PSD, random, plane stress Von-Mises, Kanai-Tazimi.
1. INTRODUCTION
A special feature of the earthquake excitation of the structures, compared with most other form of dynamic
excitation, is that it is applied in the form of the support motions rather than by point or uniformly distributed
load. Ground motion induced by earthquake is random in nature. Due to complex nature of the formation of the
seismic waves and their travel path before reaching the structure of interest, a stochastic approach may be the
most suitable for generating earthquake excitation model. During the past decades, as large sets of strong motion
recordings become available, efforts are made to reduce the uncertainties associated with the modelling.
Although the ground motion is a typical non-stationary process when the stationary process is used to establish
the ground motion models, it is usually believed that this only reflects its strong motion stage.
In the process of determining response of structure under earthquake, the power spectral density (PSD)
function is a useful tool to express the response of the structure in terms of the excitation function. If the power
spectral density of the displacement is known then the power spectral density of the response of interest like
strain, stress, pressure etc can be computed using transfer function between the displacement and the response of
interest. The response of structures under the action of ground acceleration can be estimated in two ways:
(1) The structure is assumed to be lumped mass model. In case of lumped mass model the inertial coupling
terms are all zero.
(2) The structural mass is presented as a consistent mass matrix as in a distributed system. Here the coupling
terms are present between the degrees of freedom.
In case of the consistent mass matrix where mass coupling is present between the degrees of freedom
the mass matrix is no longer diagonal. Hence, the participation factor formulation contains that portion of the
mass matrix which denotes the coupling between the support degrees of freedom and free degrees of freedom.

When a structure is assumed to be distributed and supported at the base, the structure is supported at more than
one point. The structure in this condition is subjected to excitation that may be different at different support
points. This type of excitation is called multi-support excitation. So the distributed system with base excitation
problem is considered to be multi-support excitation.
2. THEORY
2.1 Pseudo-Excitation method (PEM)
2.1.1 Structural responses due to stationary random excitation
The PEM for stationary excitation problem can be explained by the Figure 1. A linear system is subjected to a
zero-mean-valued stationary random excitation whose power spectral density (PSD) Sxx (ω) has been specified.
If y and z are two arbitrary stationary harmonic responses due to the pseudo harmonic excitation
(1)
Then the response y and z can be evaluated as:
(2)
(3)
Also it can be shown that
. | | (4)
. (5)
Here, * represents complex conjugate. If the pseudo responses are two arbitrary harmonic response vectors,
{y(t)} and {z(t)}, it can be similarly verified that the corresponding PSD matrices would be
, (6)
Now, consider a structure subjected to a single random excitation. Its equation of motion is
(7)
Figure 1: Pseudo excitation and responses for a system

In which x(t) is a stationary random process of which the PSD Sxx(ω) is known. {p} is a given constant vector.
In order to solve this equation, let, , then equation (7) becomes the following equation:
(8)
Its stationary solution is
(9)
Using its first q modes for mode-superposition, then
∑ , (10)
Where,
2 (11)
122

In which ωj, {φj}, are the jth natural angular frequency, mode and damping ratio respectively. The PSD matrix
of {y} will then be,
(12)
Substituting equation (10) into equation (12) and expanding it gives
∑ ∑ (13)
This is well known complete quadratic combination (CQC) formula. It involves the cross-correlation terms
between all participant modes. If the cross terms are neglected, that leads to the following approximate square
root of the sum of squares (SRSS) formula,
∑ | | (14)
2
The main efforts for equation (13) are q vector multiplication operations, while equation (14) contains only q
such vector operations.
Also for an arbitrary response of interest ξ, when its pseudo-harmonic response is computed by using the
PEM, the corresponding PSD can be obtained from the following equation
` (15)
2.2 Theory of Multi-Support Excitation
0
(16)
Where M, C, K: mass, damping, and stiffness matrices corresponding to the free dof’s u; , and : mass,
damping and stiffness matrices corresponding to the support dof uR that are excited by the random loading. ,
, : mass, damping and stiffness matrices that couple the free and restrained dof; R: the reaction vector.
It should be noted that support dof’s that are not excited by any random excitation need not be included
in equation (16). For support excitation dynamic problems, the free displacement ‘u’ are decomposed into
pseudo-static and dynamic component:
(17)
The pseudo-static displacements can be obtained from equation (16) after excluding the inertia and damping
terms and replacing u by us,
(18)
Where (19)
th
The elements of the i column of the ‘A’ matrix are the pseudo-static displacements due to a unit displacement
of the support dof’s excited by the ith base PSD input. It will be shown next that the ‘A’ matrix is required to
compute the participation factors. To this end, substituting equation (17) and (18) into (16) to arrive at:
(20)
The damping forces on the right hand side are small compared to the inertia forces and can be neglected:
(21)
The undamped free vibration mode shapes , 0 of the structures restrained at the support point can be used
to decouple the equations of the motion (21) to yield:
2 Γ (22)
Where the participation factors Γj are given by,
Γ (23)
and ξj are the modal damping ratio. The generalized coordinated yj are related to the mode shapes φj in the
following manner:
, 1,2 , … … , (24)
123

Where n is the number of modes to be included in the random vibration analysis. For mass normalised mode
shapes 1. The participation factors are then given by,
Γ (25)
In order to express the participation factors entirely in terms of eigensolutions, the eigenproblem is written from
equation of motion (16) in the following manner:
0
(26)
0 0
Once again, Rj is the reaction vector for the jth mode of vibration at the support points that are excited by
random loading. The eigen problem can be divided into two sets of equations, one corresponding to the free
dof’s and the other corresponding to the support dof’s excited by random loads:
0 (27)
(28)
Taking transpose of equation (27) and (28):
(29)
(30)
Post multiplying both side of equation (29) by us and using the relation in equation (18) to obtain:
(31)
From equation (30) and (31) it is found that
(32)
Performing matrix augmentation of equation (32) can be rewritten as follows:
(33)
0
Similar to the dynamic components displacement dof’s equation (24) the quasi-static part can also be expressed
in terms of mode shapes and generalized coordinates:
(34)
If this quasi-static generalized coordinates zj is defined as:
(35)
Then by virtue of equation (32), zj can be written as:
(36)
Substituting these generalized coordinates into equation (34), us can be obtained as:
(37)
Comparing this with the expression for the quasi-static solution (equation (18)), derived earlier, the quasi-static
matrix A can be expressed in terms of mode shapes, natural frequencies and modal reactions:
(38)
Using equation (38) the expression for participation factor from equation (25) can be redefined as
Γ
(39)
Because the mass matrix is normalized such that 1, the following simplified expression is obtained
for the computation of participation factor as:
Γ (40)
This method of computation of participation factor eliminates the need for additional quasi-static or any
other solution besides modal solutions, thereby saving significant computation cost.
124

3.
3 RESULT AND DISCU USSION
A simmply supportted beam is model
m using plane stress element. Thee variation of displacemeent along
length
l directiion is measurred and comppared with thee analytical reesult as tabulaated in Table 1. The free vibration
v
analysis
a of th
he model givees the natural frequencies of
o the beam. The analyticaal and FEM model m frequenncies are
given
g in Tablle 2.
Table 1: Com
mparison of Displacement
D of the midpooint of the beaam between annalytical and finite elemennt model
Analyytical Result FE Modell with 20X4 FE Modell with 100X4
(m) Elemeents (m) Elemeents (m)
0.00118
0 0.00115 0.000116
Table 2: Analytical
A andd Finite elemeent results on Natural
N frequuencies of thee beam
Natural
1st 2nd
n
3rd
Frrequency (Hz))
FEM 48.76 1766.99 357.30
Analytical 46.34 1788.50 356.60
3.1
3 Responsee characteristtics evaluation with variouus power specctral densitiess of excitation
n
A pseudo ranndom load of uniform PSD D is applied att quarter poin
nt of the beamm. The load is assumed to beb a zero
mean
m white noise
n pseudo random.
r Respponses are obbtained for diffferent forcingg PSD functioon to have the idea of
relationship
r b
between the excitation
e and
d response. TThe change in n the responsse due to the change in thhe power
content
c in thee excitation iss observed.
3.1.1
3 Comparrison of respo
onse with Diff
fferent PSD off Excitation
Table 3: Caase study and comparison of
o the responsse of the nodee which is exccited with a pseudo
p random
m force
Resultts (At first funndamental freequency)
Case Input data Dissplacement Displaccement
(abssolute value) PSD
PSD = 100 m2/s3;
Frequuency range = 0-5000 radiaan/sec;
Case
Frequuency intervall =5 radiaan/sec; 1.8885x10-5 m. 3.554x100-10 m2s.
1
Damp ping ratio ξ=00.001;
Numbber of natural frequency co onsidered = 5
The value
v nput PSD is reeduced to 1 m2/s3 with all the
of the in t
Case
other parameter rem maining samee, i.e., the forrcing PSD is 1.8885x10-6 m. 3.554x100-12 m2s.
2
reducced by 1/100.
Figure 2: Dissplacement-frrequency plott (Case1) Figure 3:

3 Displacemeent PSD-frequ
uency plot (C
Case1)
The
T plot of th he displacemeent and the plot of the PSD
D of displaceement at the eexcitation point over the frrequency
range
r for casse 1 are show o the Plot of the displacem
wn in Fig 2 annd Fig 3 respeectively. Also ment of the exxcitation
point
p and thee Plot of the PSD of displlacement of the
t excitation n point over tthe frequencyy range for caase 2 are
shown in Fig 4 and Fig 5 respectively.
r
125

Figure 4: Diisplacement-ffrequency ploot (Case 2) Figure 5: Displacem

ment PSD-freq
quency plot (C
Case 2)
3.1.2
3 Discusssion
From
m the Figure 2 and 4 it is observed
o thatt the displaceement responsse is proportiional to the exxcitation
power
p spectraal density and
d it changes with
w the squarre root of thee value of pow
wer spectral density
d of the force.
On thhe other handd, if the power spectral deensity of the displacementt is of interesst, the powerr spectral
density
d of thee displacemennt is directly proportion to the power spectral denssity of force [as
[ shown in Figure 3
and
a 5]. With the relationsh hip between th t response established now the compaarison of
he random exxcitation and the
actual
a random m to pseudo random proccess is of inteerest. These relationships
r show good confirmation
c with the
theoretical
t relationship bettween responnse and excitaation power sp
pectral densityy.
3.2
3 Random vibration ana
alysis
In thiis section, ranndom vibratioon analysis of

o the beam iss performed. The geometrry, material prroperties
and
a the bounndary conditioons of the beam are kept similar
s to thaat used in static and free vibration
v anallysis. To
perform
p the response
r analyysis two meth
hods are adoppted.
Pseuddo Excitationn Method (PE EM) proposedd by Lin et all. [5] is first implemented
i for random vibration
v
analysis
a of thhe structures. It converts the random excitation off given poweer spectral deensity into eqquivalent
harmonic
h force. Details off the method are explainedd earlier in chhapter three. A pseudo ranndom load off uniform
power
p spectrral density is applied at qu
uarter point of
o the beam. The load is assumed to be b a zero meaan white
noise
n randomm force (whitee noise is a ranndom processs that has connstant power sspectral densiity over the frrequency
range).
r The response analy ysis under pseeudo excitatioon is perform
med in the freqquency domaiin. The displaacements
in
i the frequenncy domain are a calculatedd and then corrresponding power
p spectraal densities off the displacem
ment are
evaluated.
e
On the
t other hannd, classical random
r vibrattion analysis is performedd in the time domain. Thee random
load
l that is giiven at the sam
me location of
o the beam iss also assumed to be zero mean
m white noise force. Neewmark-
Beta
B method d is used to perform the response annalysis in thee time domaiin. Fast Fourrier transform m of the
displacement
d t in time dom main is carriedd out to get tthe displacem
ment in frequeency domain. With the hellp of the
displacement
d t found in freequency domain, power sppectral densitty of the dispplacement in frequency doomain is
calculated.
c
3.2.1
3 Input daata for classiccal random looading and psseudo random
m loading
The
T following g data is taken into consideration for the both random
m and pseudoo random exciitation:
Power
P spectraal density =10000 m2/s3
Frequency
F raange = 0-50000 radian/sec or 800 Hz
Frequency
F inntervalΔω = 5 radian/sec;
Total
T time =11 sec
Time
T intervall = 1/20times the value of the largest naatural frequenncy taken in coonsideration
Maximum
M naatural frequen
ncy taken = 5
The random load ding, developped using sppectral repressentation method, is baseed on the following
fo
[13]
equation
e
126

√2 2 Δ cos
Where Δ j=1, 2 ... N with upper bound (circular) frequency ωΝ = NΔω generally known also as the cut-
off frequency beyond which the power spectral density function S (ω) is considered negligible. Also φj (j =1, 2
... N) are independent random variable distributed over the range (0, 2π ) . Here N is taken to be 1001 and ωΝ is
taken as 5000 radian/sec. The force in the frequency domain is shown in Figure 6.
Figure 6: Classical random forces in frequency domain

3.2.2: Output
The displacement PSD of the excitation point in classical random vibration analysis and the displacement PSD
of the same point in pseudo excitation method are compared as shown in Figure 9. It is observed that response
computed in both the methods compares well.
Figure 9: Comparison of displacement PSD:

(i) Pseudo Excitation (red line) (ii) Random excitation (blue line)
3.2.3 Discussion
The response of the beam is determined in term of displacement and power spectral density of the same
using both the methods. It has been observed from the previous results and plots that response using pseudo
excitation method is close to the actual random response in frequency domain. Also the computation time taken
by the pseudo random technique is much lower than the actual random response computation. Therefore, it can
be concluded that the pseudo random excitation can be used without compromising with the accuracy.
4.CONCLUSION
This paper deals with the relationship between the power spectral densities of excitation and response. It is
shown that the response PSD is proportional to the excitation PSD with the help of different case studies
presented. The calculation of displacement is crucial so that the strain and stress at the different regions of the
structure. It has been proved that the displacement is directly proportional to the PSD of the excitation. On the
other hand, the response due to the force generated using pseudo excitation method gives good conformation
with the classical random force. This will simplify the procedure of random vibration analysis like seismic
127

analysis of complex structures and reduce the computational cost. The seismic analysis of a distributed system
like beam column joint subjected to ground acceleration model is carried out, but due to lack of space the details
could not be presented in the present article.
REFERENCES
[1] Kanai, K. (1957) “Semi empirical Formula for the Seismic Characteristics of the Ground”, University of
Tokyo, Bull., Earthquake Research Institute, Vol. 35, pp. 309-325.
[2] Tajimi, H. (1960) “A Statistical Method of Determining the Maximum Response of a Building Structure
during an Earthquake”, Proc. 2nd World Conference on Earthquake Engineering, Tokyo and Kyoto, Vol.
II, pp. 781-798.
[3] Kiremidijian, S., Winterstein, S. R. (2000) “Modelling of Earthquake Ground Motion in the frequency
domain”, Department of Civil and Environmental Engineering, Stanford university, Report no 134.
[4] Lin, J., Zhao, Y., Zhang, Y. (2001) “Accurate and highly efficient method for structural stationary/ non-
stationary random responses”, Computer Methods and Mechanics Engineering, Vol. 191 pp 103-111.
[5] Li, Q.S., Zhang, Y. H., Wu, J. R., Lin, J.H. (2004) “Seismic random vibration analysis of tall buildings”,
Science Direct, Engineering Structures, Vol. 26 pp 620-629.
[6] Xu, Y., Xu, W., Mahmoud, G. M., Lei, Y. (2005) “Beam–beam interaction models under narrow-band
random excitation”, Physica A, Science Direct, Elsevier Vol. 346, pp 372–386.
[7] Paez, T. L. (2006) “The history of random vibrations through 1958”, Mechanical Systems and Signal
Processing, Elsevier, Vol. 20, pp 1783- 1818.
[8] Li, J., An, Z. (2008) “Random function model research on Strong ground motion”, The 14th World
Conference on Earthquake Engineering, October 12-17, Beijing, China.
[9] Lin, J. H., Zhang, Y.H., Zhao, Y. (2011) “Pseudo Excitation Method and some recent developments”,
Procedia Engineering, Elsevier, Vol. 14, pp 2453-2458.
[10] Cacciola, P. (2012) “Simulation of variable random process through the spectral-representation-based
approach”, 5th International Conference on Reliable Engineering Computing, (REC).
[11] Zavoni, E. H., Vanmaecke, E. H. (1994) “Seismic random vibration analysis of multi-support excitation
problem”, Journal of Engineering Mechanics, Vol. 120, no. 5, ASCE.
[12] Zavoni, E. H., Salgado, E. O., Barrnco, F. (1996) “Random vibration analysis of non-classically damped
muti-support structural systems”, Eleventh World Conference on The Earthquake Engineering, paper no.
1166. Elsevier.
[13] Alkhaleefi, A. M., Ali, A. (2002) “An efficient multi point support motion random vibration analysis
technique”, Computers and Structures, Elsevier, Vol. 20, pp 1689-1697.
[14] Lin, J.H., Zhang, Y.H., Li, Q.S., Williams, F.W. (2004) “Seismic spatial effects for long-span bridges,
using the pseudo excitation method”, Engineering Structures, Elsevier, Vol. 26, pp 1207–1216.
128

Paper No.: 022
TUNED LIQUID DAMPER SYSTEM FOR SEISMIC VIBRATION CONTROL

OF ELEVATED WATER TANKS
A. Roy 1 and A. (Dey) Ghosh2

1
Junior Research Fellow, Department of Civil Engineering, Bengal Engineering and Science University, Shibpur,
Howrah, India
2
Howrah, India
Email: anuja.civil@gmail.com, aparnadeyghosh@gmail.com
ABSTRACT:
Elevated water tanks are critical facilities, the continued functioning of which, during and after an earthquake, is
of immense importance. Being top-heavy structures, they are vulnerable to seismic forces and their seismic
protection is thereby extremely significant. Amongst the various passive control devices used for the mitigation of
structural vibrations, the Tuned Liquid Damper (TLD) is simple and convenient to install and yet a very effective
system. In this paper, the seismic vibration control of elevated water tanks by a TLD system is investigated. Here,
reinforced concrete tank with shaft type support is considered and it is modeled as a single-degree-of-freedom
(SDOF) system. For the TLD, a detailed modeling based on the established shallow water wave theory is used.
Simulation studies on a realistic example of an elevated water tank are carried out with several recorded
accelerograms of different earthquakes as the base input. The performance of the TLD system is examined on the
basis of the reduction in the peak value of the structural displacement. Results indicate that the TLD system is very
effective for the seismic response mitigation of the considered elevated water tank. Further, a comparison is also
carried out on the results obtained by using a more simplistic SDOF system model for the fundamental sloshing
mode of the TLD.
KEYWORDS: Elevated Water Tank, Tuned Liquid Damper, Seismic Vibration Control, Shallow Water,
Sloshing
1. INTRODUCTION
Elevated water tanks are strategically important structures from the view point of structural engineering. These
are critical facilities that must remain serviceable even in the regions of highest shaking intensity following a
design level earthquake. Despite the horizontal flexibility of such water tanks, they are quite vulnerable to seismic
forces due to the concentrated large masses over atop tall towers. The damage and failure of these structures due to
earthquakes is drawing enormous attention thesedays. Rai [1] in his study on the performance of elevated tanks
in the Bhuj 2001 earthquake noted the low redundancy and ductility of the elevated water tank structure
especially for tanks with circular shaft-type stagings to be the reason behind poor performance. Moreover
conventional ways of strengthening elevated water tanks against seismic loading make the design stiffer, thus
increasing the likelihood of failure. So, to guarantee the serviceability of these tank structures, besides having
strict design criteria and high-quality construction practices it is judicious to provide them with seismic protection
devices.
Studies on the passive control of the seismic vibration of elevated liquid storage tanks by base isolation were
carried out by several researchers and their results validate its efficacy. Shenton and Hampton [2] made a
comparison of the seismic response of an isolated elevated water tank structure to those of the corresponding fixed
base tank with the conclusion that seismic isolation is efficient for the practical range of tank capacities. Shrimali
and Jangid [3, 4] studied the application of linear elastomeric bearings, resilient-friction base isolator and friction
pendulum system for the seismic response mitigation of elevated liquid storage tanks. Shrimali [5, 6] also applied
sliding system and friction pendulum system in earthquake response mitigation of elevated cylindrical liquid
storage tanks under bi-directional excitation. These studies yielded promising results in the area of seismic
isolation of elevated liquid storage tanks. However base isolation is a pricy affair and not even simple to
implement also. Hence in recent times, the usage of passive control strategies draws immense concern. Potty and
Nambissan [7] considered viscous and friction dampers as retrofit measures for elevated steel water tanks. For
controlling earthquake induced vibration of reinforced concrete elevated water tanks there is also instance of the
usage of the Tuned Mass Damper (TMD) [8].
The popular passive control device, the TLD, is studied for the seismic vibration control of elevated water tanks
in this paper. The TLD has several advantages over other vibration absorbers like the TMD, such as, low
installation, operational and maintenance costs, simple installation in case of existing structure, easy adjustment
of natural frequency, function not restricted to unidirectional vibration, no need for low friction bearing surface,
effectiveness even against small amplitude vibrations etc.
Frahm, Modi and Welt [9] were among the early researchers on the TLD to reduce structural response under wind
or earthquake forces. Fujino et al. [10] revealed that the fundamental sloshing frequency of TLD must be tuned to
the natural frequency of the structure for obtaining a higher damping value of the TLD under small-amplitude
structural vibration. Koh et al. [11] concluded that the liquid sloshing motion is extremely sensitive to the natural
frequency of the damper as well as the amplitude and frequency content of the excitation spectrum. Later on Kim
et al. [12] experimentally validated this observation. Yu et al. [13] proposed an equivalent TMD model of the
TLD and illustrated its satisfactory performance over a wide range of excitation amplitude values. The study by
Banerji et al. [14] reported the need for greater mass ratio of the TLD to be more efficient for larger structural
damping. The performance of unidirectional and bidirectional TLD under random vibrations was studied by Tait
et al. [15]. Tait and Deng [16] compared the performance of TLD with different tank geometries. Currently,
numerous wide-reaching applications of the TLD are there as a passive vibration mitigation device in civil
engineering structures [9].
In the present work, a time domain formulation for the elevated water tank structure, modelled as a SDOF
system, with TLD is developed. The water mass inside the tank container is assumed to move rigidly with the
tank. A detailed nonlinear model simulating sloshing motion of TLD suggested by Sun et al. [17] is involved in
this formulation. A numerical study is carried out on an example shaft supported reinforced concrete elevated
water tank. The simulation study is performed using various recorded accelerograms of past earthquakes. In
addition, a comparison is also performed on the results obtained by using a simpler SDOF system model for the
fundamental sloshing mode of the TLD.
2. MODELLING OF TLD-SDOF STRUCTURAL SYSTEM
Fig.2.1 shows a rigid rectangular tank having a length 2a, width b and height H. It is partially filled with liquid
which has a free upper surface. The undisturbed or mean liquid depth is h. The tank is subjected to a horizontal
excitation at its base, xs(t). The liquid is incompressible and irrotational. The liquid pressure is constant at the
liquid free surface. The liquid particle motion develops only in the vertical x-z plane. The effect of internal friction
between the fluid particles is appreciable only in the boundary layer near the solid surfaces. Regarding sloshing of
liquid, the effect of wave breaking is also considered. The model of a TLD-SDOF system subjected to a horizontal
base acceleration &z&(t ) is shown in Fig.2.2. The mass, stiffness and damping of the SDOF system are ms, ks and cs
respectively. ξs and ωs denote the damping ratio and the natural frequency of the same respectively.
The governing equations describing the liquid motion are
∂u (η ) ∂u (η ) ∂η ∂ 2η ∂η
+ (1 − TH2 )u (η ) + C 2fr g + ghσφ 2 = −Cda λu (η ) − &&
xs
∂t ∂x ∂x ∂x ∂x (2.1)
130
∂η ∂[φ u (η )]
+ hσ =0
∂t ∂x (2.2)
where u(η) = u(x,η,t) = the horizontal velocity of the liquid particle at the free surface,
TH = tanh[k( h + η )], k = wave number, g = the acceleration due to gravity
z
η
o x
hb h
a a
Fig 2.1 Definition sketch for liquid motion in rectangular Tuned Liquid Damper (TLD)
ks
x(t)
cs
..
z (t )
Fig.2.2 Model of TLD- SDOF structural system
tanh(kh) tanh[k (h + η )]
σ= , φ= ,
kh tanh(kh)
&x&s = The horizontal acceleration of the TLD base
1 1   2h  
where, λ = (ωlυ ) 1 +   + S  = the damping parameter
(η + h) 2   b  
S = surface contamination factor, ωl = the characteristic frequency of liquid motion.
The fundamental linear sloshing frequency of the water in the TLD tank is given by Sun et al. [17]
131
πg 
ωl =  tanh(π∆ ) (2.3)
 2a 
h
where ∆ = = undisturbed water depth/the tank length
2a
 h 2ω l 
C fr = 1.05, C da = 0.57  ( x s ) max 
 aυ 
The horizontal base shear force of the TLD is the difference between the hydrostatic pressure on left and right
ρ gb 
walls of the tank: F = ( η n + h ) − (η o + h ) 
2 2
2  
whereη n and η o are the free surface elevations at the right and left walls of the tank respectively.
Let x ( t ) denote the horizontal displacement of the SDOF system relative to ground motion.
Consideration of the dynamic equilibrium of the SDOF-TLD system gives
.. .. .
ms {x (t ) + z (t )} + cs x (t ) + k s x(t ) = F (2.4)
3. NUMERICAL STUDY
For numerical investigation an example elevated water tank structure is considered. To examine the performance
of TLD for the passive vibration control of elevated water reservoir, the following realistic reinforced concrete
tank structure with shaft type staging is taken. The relevant data that are assumed to define the structure are listed
below:
Height of shaft support = 30 m; Mean diameter of shaft = 3 m; Thickness of shaft wall = 0.125 m; Inner height of
the tank container = 4 m; Mean diameter of the tank container = 10 m; Thickness of the container wall = 0.2 m;
Thickness of the container bottom = 0.2 m; Thickness of the top cover of the container = 0.2 m; Height of water in
the tank container = 3.8 m
Based on these assumed data and considering the elevated water tank structure as a cantilever, the mass and
stiffness of the structure are calculated for the full condition of the tank. The fundamental natural frequency of
the structure is found to be 2.52 rad/sec for full condition (time period = 2.49 sec). The value of the structural
damping is assumed to be 1%. The TLD is tuned, with a tuning ratio of unity, to the frequency corresponding to
the full condition of the tank and mass ratio of 1% is considered.
The fourth-order Runge-Kutta method is being employed for time history analysis. The example structure is
subjected to various recorded accelerograms of different earthquakes. The time histories of the structural
displacement for the structure alone and with the TLD system for full condition of the tank are presented in
Fig.3.1-Fig.3.4. The performance criterion of the TLD is the reduction in the peak value of the displacement of
the elevated water tank system. The response reductions achieved in different cases are tabulated below in Table
3.1:
132
Table.3.1 Response reduction achieved by the TLD for different inputs
Sl. Percentage Reduction in Peak

Input
No. Structural Displacement
1 Elcentro 27
2 Parkfield 25.9
3 Bhuj(Longitudinal) 39
4 Bhuj(Transverse) 51.1
0.4 0.2
without TLD without TLD
0.3
Structural Displacement (m)

0.15
with TLD with TLD

0.2 0.1
0.1 0.05
0 0
-0.1 -0.05
-0.2 -0.1
-0.3 -0.15
-0.4 -0.2
0 10 20 30 40 50 60 0 5 10 15 20 25 30 35 40 45
Time (s) Time (s)
Fig.3.1 Displacement Time History of Fig.3.2 Displacement Time History of

SDOF Structural system to Elcentro SDOF Structural system to Parkfield
earthquake earthquake
0.15 0.15
without TLD without TLD
with TLD with TLD
0.1 0.1
0.05 0.05
0 0
-0.05 -0.05
-0.1 -0.1
-0.15 -0.15
-0.2 -0.2
0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140
Time (s) Time (s)
Fig.3.3 Displacement Time History of SDOF Fig.3.4 Displacement Time History of SDOF
Structural system to Bhuj earthquake Structural system to Bhuj (Transverse
(Longitudinal component) component) earthquake
Now the displacement time histories of the SDOF structure to Elcentro and Parkfield earthquake are
compared with those for a linearized model of TLD in Fig.3.5 and Fig.3.6. The comparison of response reduction
achieved by TLD for simple and rigorous modeling is given in Table.3.2.
133
Table.3.2 Comparison of Percent Response Reduction achieved by the TLD with different models
Percent Response Reduction in peak displacement
Input Earthquake
Linear Model Nonlinear Model
Elcentro 4.4 27
Parkfield 4.5 25.9
40.00
without TLD
30.00 Linear model of TLD
20.00 Nonlinear model of TLD
10.00
0.00
-10.00
-20.00
-30.00
-40.00
20.00
without TLD
Linear model of TLD
15.00
Nonlinear model of TLD
10.00
5.00
0.00
-5.00
-10.00
-15.00
-20.00
4. CONCLUSIONS
An investigation into the applicability of TLD as passive vibration control device for elevated water tanks was
undertaken. For this purpose, a formulation of the equations of motion of a base-excited, viscously damped SDOF
system equipped with a TLD system was developed. The formulation considers the continuum model for
134
simulating liquid sloshing within the TLD container, which is a highly nonlinear phenomenon. This nonlinearity
of damping is taken into consideration in the performance evaluation of the damper. The results in time domain
study considering several recorded accelerograms show that substantial reduction in peak value of structural
response is achieved by the TLD in the tuned condition. For all the cases wave breaking is taking place. Further,
the performance of the nonlinear model of TLD is compared with that of the SDOF model of the fundamental
sloshing mode of the TLD. The values of response reductions in case of the nonlinear model are much more than
those of the linear case, thereby indicating that there is a large underestimation of the TLD performance when the
simplified SDOF system model of the sloshing mode is considered. This is chiefly due to the huge dissipation of
energy that takes place due to sloshing and/or wave breaking phenomenon in the liquid of the TLD. This cannot be
captured by the linearized TLD model. Overall, the TLD system presents an effective and easy to implement
control scheme for the seismic safety of elevated water tanks.
REFERENCES
[1] Rai, D.C. (2002). Review of Code Design Forces for Shaft Supports of Elevated Water Tanks. Proceedings
of 12th Symposium on Earthquake Engineering, IIT Roorkee, Dec. 16-18, 1407-1418.
[2] Shenton III, H.W., and Hampton F.P. (1999). Seismic Response of Isolated Elevated Water Tanks. Journal of
Structural Engineering, ASCE, 125(9), 965–976.
[3] Shrimali, M.K., and Jangid, R.S. (2003). Earthquake Response of Isolated Elevated Liquid Storage Steel
Tanks. J Constr Steel Res, 59, 1267–1288.
[4] Shrimali, M.K., and Jangid, R.S. (2006). Seismic Performance of Elevated Liquid Tanks Isolated by Sliding
Systems. Proceedings of first European conference on earthquake engineering and seismology, Geneva,
Switzerland, 3–8 Sept, paper no. 1784.
[5] Shrimali, M.K. (2007). Seismic Response of Elevated Liquid Storage Steel Tanks under Bi-direction
Excitation. Steel Structures, 7, 239-251.
[6] Shrimali, M.K. (2008). Earthquake Response of Elevated Liquid Storage Tanks Isolated by FPS under
Bi-direction Excitation. Advanced Vibration Engineering, 7(4), 389–405.
[7] Potty, N. S., and Nambissan, S. (2008). Seismic Retrofit of Elevated Steel Water Tanks. ICCBT 2008 –
C-0999-108.
[8] Jaiswal, O. R. (2004). Simple Tuned Mass Damper to Control Seismic Response of Elevated Tanks. 13th
World Conference on Earthquake Engineering, Vancouver, B. C. Canada, August 1-6, Paper No.2923.
[9] Soong, T.T., and Dargush G.F. (1997). Passive Energy Dissipation Systems in Structural Engineering, Wiley,
UK.
[10] Fujino, Y., Pacheco, B.M., Chaiseri, P, and Sun L.M. (1988). Parametric Studies on Tuned Liquid Damper
(TLD) using Circular Containers by Free-Oscillation Experiments. Structural Engineering/Earthquake
Engineering, JSCE, 5(2), 381–391.
[11] Koh, C.G., Mahatma, S., and Wang C.M. (1994). Theoretical and Experimental Studies on Rectangular
Liquid Dampers under Arbitrary Excitations. Earthquake Engineering and Structural Dynamics, 23, 17–31.
[12] Kim, Y.M., You, K.P., Cho, J.E. and Hone, D.P. (2006). The Vibration Performance Experiment of
Tuned Liquid Damper and Tuned Liquid Column Damper. Journal of mechanical science and technology,
20(6), 795-805.
[13] Yu, J.K., Wakahara, T. and Reed, D.A. (1999). A Non-Linear Numerical Model of the Tuned Liquid
Damper. Earthquake Engineering and Structural Dynamics, 28,671-686.
[14] Banerji, P., Murudi, M., Shah, A. H. and Popplewell, N. (2000). Tuned Liquid Dampers for Controlling
Earthquake Response of Structures, Earthquake Engineering and Structural Dynamics, 29, 587-602.
135
[15] Tait, M. J., Isyumov, N., and El Damatty, A. A. (2008). Performance of Tuned Liquid Dampers. Journal of
Engineering Mechanics, 134, 417-427.
[16] Tait, M. Z., and Deng, X. (2010). The Performance of Structure-Tuned Liquid Damper Systems with
Different Tank Geometries. Structural Control and Health Monitoring, 17, 254–277.
[17] Sun, L.M., Fujino, Y., Pacheco, B.M. and Chaiseri, P. (1992). Modelling of Tuned Liquid Damper, Journal
of Wind Engineering and Industrial Aerodynamics, 41-44, 1883-1894.
[18] Veletsos, A. S. and Tang. Y. (1990). Soil-Structure Interaction Effects for Laterally Excited Liquid Storage
Tanks. Earthquake Engineering and Structural Dynamics, 19, 473-496.
136
Paper No.: 023
STUDY ON STABILITY OF NATURAL DRAFT HYPERBOLIC RC COOLING

TOWER SHELL UNDER WIND INDUCED FLOW FIELD.
1 1 1 2
Manish K. Singh , Sailesh Adhikari , Sumit Mazumder , Amiya K. Samanta
1
Undergraduate Student, Department of Civil Engineering, NIT Durgapur, WB, India,
2
Associate Professor, Department of Civil Engineering, NIT Durgapur, WB, India
email-manish08aec@gmail.com, aksnitd@gmail.com
ABSTRACT :
The natural draft cooling tower is a very important and essential component in the thermal, nuclear power
stations and industrial power plant. Hyperbolic shape of the cooling tower is preferred due to its superiority in
stability and higher efficiency. Also It is easier to build and stronger among various types of surfaces. Except
seismic loading, wind constitutes the main loading for the natural draft cooling towers. The wall of the cooling
tower is subjected to non-axisymmetric intensities of wind pressure due to its massive elevation. Simulation of
the model has been done using the platform of the Finite Element Software ABAQUS and the wind induced
flow field as described in relevant Indian code of practice has been expressed in terms of Fourier series in the
analytical field within the advanced options of the software. Mainly the investigation has attempted to throw
some light on the cross-sectional distortion or the ovalisation of the circular shell due to such distribution of
wind load so as to take care of the brazier effect. It has been observed that as such the deformation pattern of the
wall does not follow the elastic bending theory. At the same time one modified model has been considered by
choosing cylindrical shell beyond the throat of the said cooling tower for the purpose of comparison. The results
obtained for two models have been presented along with the critical /design values so as to assess the
potentiality, reliability of such methods in predicting behavior of such structural system.
KEYWORDS: RC Shell Wall, Natural Draft Hyperbolic Cooling Tower, Simulation of Wind Flow Field,
ABAQUS, Ovalisation.
1. INTRODUCTION
Cooling towers are heat removal devices used to transfer process waste heat to the atmosphere. Cooling towers
may either use the evaporation of water to remove process heat and cool the working fluid to near the wet-bulb
air temperature or solely on air to cool the working fluid to near the dry-bulb air temperature. Common
applications include cooling the circulating water used in oil refineries, chemical plants, power stations and
industrial cooling. The towers vary in size from small units to very large hyperboloid structures that can be up to
200 meters tall and 140 meters in diameter. Hyperbolic shape of the cooling tower is preferred due to its
superiority in stability and higher efficiency. At the bottom, the widening of the tower provides higher surface
area for water to cool and the narrowing effect helps accelerate the laminar flow. Also It is easier to build and
stronger among various types of surfaces. Due to their complexities in geometry, the analysis of such type of
structures has attracted attention of many researchers. In the absence of earthquake loading, wind constitutes the
main loading for the natural draft cooling towers. There is no guideline in the Indian code of practice to design
hyperbolic cooling tower against wind load which is important, except distribution of such loads along the
periphery of cylindrical stack like structures in empty condition.
There are a very few investigations related to this aspect, in particular to the natural draft hyperbolic cooling
tower. Samanta et. al.(2010) did a study to consider such wind flow field on cylindrical structure such as single
silo and have shown the venerability of such configuration under wind load with a special emphasis on
ovalisation instability. In the investigation of Hyuk Chun Noh (2005), the behavior of a reinforced concrete
cooling tower shell in hyperbolic configuration is presented including the geometrical nonlinearity as well as
material nonlinearities in the concrete considering layered concrete shell. Only the crack pattern was
investigated to address more clearly the failure mechanism of the cooling tower under consideration. The failure
of the cooling tower is addressed through the yielding of the meridional reinforcement and the yielding of
reinforcement propagates in the circumferential direction until the ultimate load is reached. In all other works,
specific attention was not made in regard to the ovalisation instability so far.
In this presentation, the stability and deformation effects on the shell of reinforced concrete cooling tower due to
wind induced flow field as per relevant Indian Standard code of Practice is presented and the results obtained
has been discussed. For the sake of study, the geometrical configuration and mechanical properties of a
hyperbolic natural draft cooling tower which is located in Andal, West Bengal is considered. The wall of the
cooling tower is subjected to non-axis symmetric intensities of wind pressure due to its massive elevation.
Simulation of the model has been done using the platform of the Finite Element Software ABAQUS and the
wind induced flow field has been expressed in terms of Fourier series in the analytical field within the advanced
options of the software. Mainly the investigation has attempted to throw some light on the cross-sectional
distortion or the ovalisation of the circular shell due to such distribution of wind load so as to take care of the
brazier effect.
2. MODELS AND METHODOLOGY
Two types of reinforced concrete cooling tower models have been considered for the analysis of the behavior of
the shell wall under static wind loads.
(A) Model 1 (B) Model 2

Figure 1 Profile of Hyperbolic cooling tower
Two types of reinforced concrete cooling tower models have been considered for the analysis of the behavior of
the shell wall under static wind loads. For Model-1, the total height of the tower from the ground level is 125.91
meters. Top level of racker column lies at the height of 9.4 meters from the ground level. For the sake of
simplicity, the modeling and analysis has been performed only for the portion of the tower above the racker
column as shown in Figure 1. From the level 9.4 m to 14.4 m level the shell has the straight profile and
thickness is uniformly varying from 800mm to 350mm. From 14.4 m to 125.91 m the shell has the hyperbolic
138
profile. From height 14.4m to 28.4 m the thickness varies from 350mm to 225mm. From 28.4m to 119.91 m the
shell have the uniform thickness of 225 mm and from 119.91 m to 125.91m the thickness increases from
225mm to 600mm uniformly. Throat level lies at the height of 94.43m from the ground level. Diameter at Base,
Throat and Top are 96.30m, 49.89m and 59.60m respectively. For Model-2, the Geometry is similar to the
previous except for the portion above the throat i.e. for the top 40 meters of the tower, geometry is cylindrical
with 225mm thickness. The diameter of the cylindrical portion is 49.89m.
The deformation of the structure under the influence of self weight and wind load is analyzed with the help of
ABAQUS, which utilizes the finite element method of analysis considering its mechanical properties as
homogeneous, isotropic and elastic. The hyperbolic shell wall is modelled using 8-noded standard linear
hexahedral elements (C3D8R) with the element sizes as 200mm so as to avoid any mess distortion. A total of
6136 elements for model 1 and 9086 elements for model 2 have been generated. An encastre support condition
has been stimulated at the bottom most level of the shell wall. The entire geometry for both the models were
instanced and meshed followed by application of boundary condition and load.
3. WIND PRESSURE ASSESSMENT
The wind pressure, which varies along the height, (being constructed at Andal, Durgapur) has been calculated as
per IS 875 – 1987 (Part III) with Basic wind speed (Vb) = 47 m/s. Risk coefficient (k1) and Topography factor
(k3) are taken as 1.07 and 1.00 respectively, whereas Terrain height factor (k2) varies with the height of the
cooling tower. Hence the design wind speed, (Vz) = k1.k2.k3.Vb. A new importance factor k4=1.3 is also taken into
consideration to calculate modified Vz=65.377 k2, assuming it to be a very important structure in cyclonic zone
and design wind pressure (p) = 0.60 (Vz)2. This variation wind pressure along the height is calculated in Table-1
and also has been shown in Figure-1.
Table 1 Wind Pressure Distribution along the height

Height upto (m) Up to 10m 15 20 30 50 100 150
k2 0.82 0.85 0.91 0.96 1.02 1.10 1.15
2
PZ (N/m ) 172.42 1941.20 2123.44 2363.29 2667.73 3103.06 3391.22
3.0
1.5 9.4-15m
15-20m
θ
Wind Pressure (kN/m )
20-30m
2
0.0 30-40m
40-50m
50-60m
-1.5 60-70m
70-80m
80-90m
-3.0
90-100m
100-110m
-4.5 110-120m
120-126m
-6.0
0 20 40 60 80 100 120 140 160 180

0
Circumferential Angle ( )
(A) Wind Pressure at different Levels /Heights (B) Wind distribution as per IS : 875(III)
Figure 2 Wind Flow Field on Circular Cross Section
At any level of the cylindrical structure, the wind pressure as calculated above is again not uniform on the
surface area as the external coefficient value changes with the angle subtending at centre of cross section as
shown in the adjacent sketch (Figure-2). The respective external coefficients(Cs) are calculated for height to
139
diameter ratio(h/d) by interpolation from the relevant table of IS code by interpolation. Now these coefficients
are being multiplied by wind pressure as calculated in Table-1 to get actual wind load(Pz) on the cylindrical
surface of the cooling tower.
In view of applying the same load pattern on the structure in the ABAQUS platform, the ‘Cs’ has been
expressed in terms of Fourier Series in a close form solution. Here the Fourier cosine series upto 8th harmonics
representing the coefficient in (0,2π) has been used as ; Cs= 0.5*a0 + a1Cos (Φ) + a2Cos (2Φ) + a3Cos (3Φ) +
a4Cos (4Φ) + a5Cos (5Φ) + a6Cos (6Φ) + a7Cos (7Φ) + a8Cos (8Φ). Where, ai are Fourier coefficients and Φ is
the angle subtending at the centre of cross section. The pressure coefficient has been calculated with the help of
Fourier function at 150 intervals and plotted in Figure 2 and compared with the coefficients obtained from IS
875. It has been noted that the coefficient obtained from calculation is in close agreement with that obtained
from the code. The wind loads calculated thus has been applied on shell wall of the cooling tower in the models
prepared in ABAQUS platform using advanced user facilities.
The deformed configuration and the stress contours of the cooling tower models analyzed by ABAQUS are
shown in Figure 3. Scale factor have been adopted to magnify the deformation of the shell wall. The
comparative study between the two different types of cooling tower .i.e. hyperbolic and
hyperbolic-cum-cylindrical has been conducted and the results are compared to see how difference in profile
parameter affects the response of the structure towards the applied load. The vertical deformation, Von mises’
stress and total deflection of the shell at various level .i.e. at the top level and the critical section in the
circumferential direction is shown and comparison is made at that level to check that which type is more
adversely affected due to the static wind loading. At the same time the ovalisation of the structure is analyzed
and a view has been made to have a say that which profile is more adversely affected due to the ovalisation
effect occurring due to the static wind loading.
Model-1
Model-2
Wind Flow Field Deformed Configuration Von Mises’ Stress Contour

Figure 3 Load, Deformation and Stress Contour
140
It has been observed that as such the deformation pattern of the wall does not follow the elastic bending theory.
At the same time one modified model (#2) has been considered by choosing cylindrical shell beyond the throat
of the said cooling tower for the purpose of comparison. The results obtained for two models have been listed
along with the critical /design values so as to assess the potentiality, reliability of such methods in predicting
behaviour of such structural system.
(A) Von Mises’ stress at the critical section
(B) Total deflection at the top level

Figure 4 Cross-sectional Deformation
It is observed that the stresses in the hyperbolic profile are more than the latter profile at the critical section. The
Von mises’ stress in the first profile in windward direction is 6.33 MPa and at critical section it is 11.4 MPa but
in the second profile the same stress is 5.31MPa in the windward direction and 9.23MPa at critical section. But
the second profile deform more than the first. At the top level due to less rigidity of second profile arising due to
the absence of ring beam is causing grater ovalisation whereas due to higher rigidity at top the first profile is
very much effective in resisting the deformation which can cause the failure at top. Figure 4 shows distortion of
the vertical section in the plane of wind direction and perpendicular to the plane of wind direction respectively.
It also shows that the level of maximum shortening of diameter of cooling tower is near about at the throat level.
Figure 5 shows the variation of the hoop stress, vertical stress and Von mises’ stress along the height at
windward side, leeward side and 780 to windward side, for both the models side by side. It is observed that the
variation of the hoop stress values along the height is to some extent irregular in case of model 2 and the same is
regular and smooth for model 1 except at the bottom and top junctions.
In the windward direction, at bottom level, the hoops stress, vertical stress and von mises’ stress at the bottom of
the 1st model are 1.79MPa, 5.81 MPa and 6.14 MPa respectively while the foresaid stresses in the second model
is 1.73 MPa, 5.42 MPa and 5.72 MPa. The maximum values of hoops stress, vertical stress and von mises’ stress
is at same height(about 25m above ground level for both the model and the value of stresses for the 1st model
are 6.06 MPa 14.62 MPa 14.85 MPa and for the second model 5.99MPa 13.15 MPa and 13.29 MPa. In the
Leeward direction, at the bottom level, the hoops stress, vertical stress and von mises’ stress at the bottom of the
1st model in are 1.84MPa, 6.52MPa and 6.92MPa respectively while the foresaid stresses in the second model is
141
1.80 MPa, 6.12 MPa and 6.50MPa. The maximum values of hoops stress, vertical stress and von mises’ stress is
at same height(about 25m above ground level for both the model and the value of stresses for the 1st model are
5.25 MPa 16.50 MPa and 17.04MPa and for the second model 5.23 MPa, 15.27 MPa and 15.70 MPa.
120 120
100 100
: Inside Nodes
: Outside Nodes
: Inside Nodes
Level (m)
80
Level (m)
WINDWARD SIDE 80
: Outside Nodes
WINDWARD SIDE
: Inside Nodes
60 : Outside Nodes (a) 60 : Inside Nodes (b)
LEEWARD SIDE
: Outside Nodes
LEEWARD SIDE
40 : Inside Nodes 40
: Outside Nodes
0
At 78 to WINDWARD SIDE : Inside Nodes
: Outside Nodes
0
20 20 At 78 to WINDWARD SIDE
0
0
-7.5 -6.0 -4.5 -3.0 -1.5 0.0 1.5 3.0
-7.5 -6.0 -4.5 -3.0 -1.5 0.0 1.5 3.0
Hoop Stress for Model-1 (MPa) Hoop Stress for Model-2 (MPa)
120 120
100 100
: Inside Nodes
: Outside Nodes
Level (m)
Level (m)
80 WINDWARD SIDE 80
: Inside Nodes
: Outside Nodes
: Inside Nodes WINDWARD SIDE
60 : Outside Nodes (a) 60 (b)
LEEWARD SIDE : Inside Nodes
: Outside Nodes
LEEWARD SIDE
40 : Inside Nodes 40
: Outside Nodes
0 : Inside Nodes
At 78 to WINDWARD SIDE : Outside Nodes
0
20 20 At 78 to WINDWARD SIDE
0 0
-21 -18 -15 -12 -9 -6 -3 0 -21 -18 -15 -12 -9 -6 -3 0
Vertical Stress for Model-1 (MPa) Vertical Stress for Model-2 (MPa)
120 120 : Inside Nodes

: Inside Nodes : Outside Nodes
0
: Outside Nodes At 78 to WINDWARD SIDE
WINDWARD SIDE
100 100
: Inside Nodes
: Outside Nodes
LEEWARD SIDE
Level (m)
Level (m)
80
80
: Inside Nodes 60
60
: Outside Nodes
0
At 78 to WINDWARD SIDE
40 : Inside Nodes
40 : Outside Nodes
: Inside Nodes WINDWARD SIDE
: Outside Nodes
LEEWARD SIDE 20
20
0
0 5 10 15 20 0 2 4 6 8 10 12 14 16 18 20
Von Mises Stress (MPa) Von Mises'Stress (MPa)
Figure 5 Stress Distribution along Height
142
At Critical section, at bottom, the hoops stress, vertical stress and von mises’ stress at the bottom of the 1st
model in are 2.0 MPa, 7.69 MPa and 8.23 MPa respectively while the foresaid stresses in the second model is
1.98 MPa, 7.45 MPa and 7.96 MPa. At the top, the values for the above stresses are 2.34 MPa, 0.27 MPa and
2.53MPa for the 1st model and 0.51MPa, 0.24MPa and 0.67MPa for the 2nd model. The maximum values of
hoops stress, vertical stress and von mises’ stress is at same height(about 25m above ground level for both the
model and the value of stresses for the 1st model are 4.95MPa, 20.05MPa and 21.13MPa and for the 2nd
model 4.92MPa, 19.21MPa 20.19MPa. These values have also been listed in Table 2 for the sake of easy
understanding and clarity. The maximum values at the three levels in each side for both the models have been
tabulated below.
Table 2 Critical Stresses along the height

Model 1 Model 2
Remarks
σvon σvert σhoop σvon σvert σhoop
6.14 5.81 1.79 5.72 5.42 1.73 Bottom Level
Windward
Side
0.37 0.25 1.50 0.33 0.24 0.37 Top Level

14.85 14.62 6.10 13.32 13.15 5.90 Maximum
6.92 6.52 1.84 6.50 6.12 1.78 Bottom Level
Leeward
Side
2.13 0.26 1.91 0.37 0.25 0.18 Top Level

17.04 16.50 5.25 15.70 15.27 5.23 Maximum
8.23 7.69 2.00 7.96 7.45 1.80 Bottom Level
Section
Critical
2.53 0.27 2.34 0.67 0.24 0.51 Top Level

21.13 20.05 4.95 20.19 19.21 4.92 Maximum
Table 2 shows the values of critical stresses required for the design of the shell wall. The stress values in the
leeward direction are not critical except the vertical stress and hence only the same has been included in the
table. Mainly the values of hoop stress and vertical stress are of significant concern to the professional
/practicing engineers for the sake of adequate detailing and precautionary measures required to be taken. It also
hints that there is a band along the height of the wall which draws attention from Figure 4. It has been observed
that in general, the height of the band varies from 50-60% of the wall height which is mostly affected by such
type of load. At the same time, the junction is also critical for both the type of the stresses, which has almost a
height of 0.5-1.0m which also needs to be strengthened while designed against such forces.
5. CONCLUSION
The investigation shows the variation as well as nature of hoop, vertical & Von Mises stress along the tower
height respectively in windward side, leeward side and critical section. From the close inspection of all the
graphs it is observed that the hoop stress values are critical in windward side and same occur at about one-third
height of the wall from the base level. Von Mises stresses are critical at the junction in leeward side for the same
height of the wall in windward side. Hoop stresses are produced due to ovalisation of wall cross section. Von
Mises stresses are critical at the junction in leeward side as well as mid height of the wall in windward side. The
hoops stress variation along the entire height is almost same for both types of profile, but at top level it become
almost constant for the second profile while it is varying in the first one. The stress and deformation variation in
the leeward direction become significant in the hyperbolic-cum-cylindrical profile than hyperbolic one. But the
radial deformation become more pronounced in the latter profile which create a significant difference in
selecting the type of profile for industrial use. At the end both the profile is almost equally efficient in resisting
the stress but the second profile is less effective when it comes to resist the deformation, which is very much
significant for tall structure in case of wind loading.
143
REFERENCES
[1] ABAQUS User’s manual. Version 6.10. SIMULIA India-303, V.C.C. Plot No.1, LSC, Near Gulabi
Bagh- Sindoran Kalan- New Delhi-110052
[2] Amiya K. Samanta, Prasanta Patra and P. Ray (2010). Wind Induced Deformation and Bending Effect
on Ground Elevated RC Cylindrical Empty Silo Wall. Procs. of International Conference on Innovative
World of Structural Engineering (ICIWSE-2010),September 17- 19, 20I0, Aurangabad, Maharashtra,
India, 633-639.
[3] Dieter Busch, Reinhard Harte, Wilfried B. Kratzig, Ulrich Montag, (2002). New natural draft cooling
tower of 200 m of height. Engineering Structures 24(12), 1509–1521.
[4] Hyuk Chun Noh, (2006). Nonlinear behavior and ultimate load bearing capacity of reinforced concrete
natural draught cooling tower shell. Engineering Structures, 28(3), 399–410.
[5] IS 11504-1985, Criteria for structural design of RCC Natural Draught Cooling Towers, Bureau of
Indian Standards, New Delhi , India.
[6] IS 875 (Part-3)-1983,Indian Standard Code of Practice for Design loads for Buildings and Structures,
Bureau of Indian Standards, New Delhi, India.
[7] J. Noorzaei, Ali Naghshineh, M.R. Abdul Kadir, W.A. Thanoon, M.S. Jaafar(2006). Nonlinear
interactive analysis of cooling tower–foundation–soil interaction under unsymmetrical wind load.
Thin-Walled Structures, 44(9), 997–1005.
[8] N. Krishna Raju, (1988), Advanced Reinforced Concrete Design, CBS Publishers & Distributers, New
Delhi.
[9] P.A. MacDonald, J.D. Holmes, K.C.S. Kwok(1990). Wind loads on circular storage bins, silos and
tanks. II- Effect of grouping.. ,Journal of Wind Engineering and Industrial Aerodynamics, 34(1), 77-95.
[10] Ramesh Babu, G., Selvi Rajan, S., Harikrishna, P., Lakshmanan, N. and Arunachalam, S. (2013).
Experimental determination of wind-induced response on a model of natural draught cooling tower,.
Experimental Techniques, 37, 35–46.
144
Paper No.: 025
AN EXPERIMENTAL AND NUMERICAL STUDY OF NATURAL-FIBER

COMPOSITES SUBJECTED TO DIFFERENT ENVIRONMENTAL
CONDITIONS
1 2
R. Velmurugan and Jafar Sadique P
1
Professor, Department of Aerospace Engineering, CompTech, IIT Madras, Chennai, 600036
2
Graduate Student, Department of Aerospace Engineering, CompTech, IIT Madras, Chennai, 600036
Email:jafarsadiquempm@gmail.com, ramanv@iitm.ac.in
ABSTRACT:
In the last decades the bio composites have been widely used in the construction, automobile and aerospace
industries. The use of such composites in structural application as a low cost material has created a new horizon
for investigation. This work focuses on the numerical and experimental analyses of polymeric composites
fabricated with polyester resin and coir fibers. Not only the interface transition zone but also the heterogeneity of
natural fibers affects the mechanical behavior of these composites. The surface modification of coir fibers is
made with alkali treatment. The changes in the mechanical properties of the composites with different alkali
composition are studied. The major materials used in this experiment are coir yarn and general purpose polyester.
Composite fabrication; with treated and untreated coir yarn fabricated by the use of metallic rod and light
stitching is carried out at 2 or 3 places (in order to keep the fiber aligned during fabrication) in compression mold.
The composites are then cut into specified according to ASTM standard for each different test. The entire test
(static tensile, flexural, impact and water absorption test) is conducted according to ASTM standard. The results
are compared and effect of alkali treatment on fiber with the help of mechanical properties is carried out. In
addition, scanning electron micrographs obtained from coir fibers were used to study the effect of surface
modification. Finite element analysis is carried out based on the elastic properties obtained via micromechanical
models and experimental tests. The software Ansys® 13.0 is used to simulate the bio composites under tensile
loadings. Micromechanical models such as “rule of mixture” and “Halpin-Tsai” were used to estimate the
effective properties of the bio composites. The experimental result compared with the numerical results and the
error of these result were calculated to evaluate the interface condition of the constitutive phases and to verify the
efficiency of the FEA.
KEYWORDS: Natural Fibers, Composite, Surface Modification, Micromechanics, Mechanical Properties
1. INTRODUCTION
A structural composite is a material system consisting of two or more phases on a macroscopic scale, whose
mechanical performance and properties are designed to be superior to those of the consistent material acting
independently. Normally, composite material consists of two total different phases. The reinforce phase which is
usually discontinuous, stiffer, and stronger, while the matrix phase is typically the less stiff and weaker and
continuous. Composite materials appeared in the sixties of the last century. Composite spreads so widely during
short period of time due to their high mechanical properties and their low weight. They have been used for
variety applications although they were born originally in the aerospace industry where light weight plays an
important role in selecting appropriate materials. All parts made of composites are designed to sustain long life,
this advantage has become disadvantage due to the fact that composite are not easy to dispose off after their
proposed life. At the moment there are two ways to dispose composites after use. They are either buried in
landfills, or burnt up. However both methods are expensive and cause pollution to the environment.
For these reasons, recent research work focus on deriving composite materials that can degrade in soil after they
are put in landfills, or composite that can be recycled. Natural fiber plays an important role in developing
biodegradable composites to resolve the current ecological and environmental problems. They are cheap,
renewable, completely or partially recyclable, and biodegradable. The natural fibre containing composites are
more environmentally friendly, and are used in transportation such as automobiles, railway coaches, aerospace
etc., military applications, building and construction industries (ceiling, paneling, and partition boards),
packaging, consumer products etc. Natural fibre include those made from plants, animal and mineral sources.
The background of the research that has been carried out is the awareness that the demand of coir and coir
products is slowly decreasing and that other profitable markets have to be found for this commodity. The best
way to bring the existing coir industry to a higher level is the development of new coir products with higher
added value. One possible technology that could fulfill this goal is the use of coir fibre in composite components.
It is clear that composite technologies allow the producer to add much more value to his product than the
processing and trading of raw material.
Demand for natural fibre based composite originated from automotive industry. Research focused on locally
grown fibers, for instance European automotive industry mainly uses flax and hemp, whereas the Brazilian
automotive industry mostly applies locally grown sisal. Using the automotive industry as stepping stone towards
other markets, each natural fibre based composites has found other applications, depending on the specific
properties of the fibers. Among the available natural fibers, coir is produced in large quantities in south Asian
countries, such as India, Indonesia, Philippines, etc.; Coir based composites, depending on its specific
characteristics, could also find a position within the wide scale of applications and products in and outside the
automotive industry [4]. Until now little research on coir composites has been carried out, so far only some
material characteristics have been studied while research on processing in combination with product
development has not been undertaken.
2. MATERIALS AND METHODS
2.1. Experimental
The raw materials used in this work are

• Coconut coir fibers
• Polyester resin and Hardener
Coir is a lignocellulose natural fiber. It is a seed-hair fiber obtained from the outer shell, or husk, of the coconut.
The individual fiber cells are narrow and hollow, with thick walls made of cellulose. Coir fibers were obtained
from local sources was used for the preparation of composites. Two types of biocomposites were investigated in
this work, one fabricated with yarn type fibers and another with short fibers. The yarn coir fibers reinforced one
is used to study the fiber architecture effect on mechanical properties and to validate the numerical analyses and
evaluating the interface condition of the constitutive phases. And the other, the short coir fibers reinforced
composites was used to study the effect of water absorption. This present work focusses on the use of general
purpose polyester. This resin is mixed with 1.5-2 wt. % cobalt naphtenate (accelerator) and 1.5-2 wt. % methyl
ethyl ketone peroxide (catalyst) is used as matrix material.
2.1.1 Surface modification of fiber
Three different types of composites have been fabricated with three different fibers treatment such as untreated,
treated with 5wt% NaOH solution and treated with 10wt% NaOH solution. Each composite consisting of 25%
of fiber and 75% of polyester resin. The designations of these composites are given in Table 2.1. These fibers
were soaked in an aqueous solution of NaOH (5wt % and 10wt %) for 72 hours. After each interval, the fibers
were removed from the solution, washed with fresh water several times and finally rinsed with distilled water to
remove any excess NaOH sticking to the fibre surface and finally dried in vacuum oven at 50° C for 24 hours.
146
Table 2.1: Designation of composites
Composites Composition
C1 Polyester (75wt %) +Untreated Coir fiber (25wt %)
C2 Polyester (75wt %) +Alkali Treated Coir fiber with 5wt% NaOH solution (25wt %)
C3 Polyester (75wt %) +Alkali Treated Coir fiber with 10wt% NaOH solution (25wt %)
2.1.2 Composite fabrication and Testing
Compression moldings were used to make specimens. The procedure while making specimens are following,
first, the mold is polished and then mold releasing agent is applied on the surface and coir yarn aligned
manually by the use of metallic rod and light stitching is carried out at 2 or 3 places in order to avoid the
presence of residual stress after cure. The thickness of the laminate was maintained by the use of a metallic
frame. The resin mixed with accelerator and catalyst and then poured on to the aligned yarn fibers placed in the
mold. When the yarn fibers were completely wet by the resin, a slight pressure is applied with a steel roller to
remove the entrapped air and the mold is closed, pressed and cured at room temperature for 24 hours. After the
curing process, test samples were cut to the required sizes prescribed in the ASTM standards. Mechanical Tests:
After fabrication the test specimens were subjected to various mechanical tests as per ASTM standards. The
standards followed were ASTM, ASTM-D 3039/D 3039M, ASTM-D 790-03 and ASTM-D 256-05 for tensile
tests, flexural tests and impact tests, respectively. To obtain a statistically significant result for each condition,
five specimens were tested to evaluate the mechanical properties. The tensile tests were conducted at a speed of
2 mm/min at room temperature. Three-point bending (flexural) tests were carried out on the specimen at room
temperature. The specimen is placed on to two supports having a 40 mm span length between the supports. The
speed of the jaws was set to 2 mm/min. Izod impact tests were conducted for all specimens at room temperature.
2.1.3 Water Absorption Test
The effect of water absorption on coconut coir fibers reinforced composites were investigated in accordance
with ASTM D5229/D5229M. This is a gravimetric test method that monitors the change over time to the
average moisture content of a material specimen by measuring the total mass change of a coupon that is exposed
on two sides to a specified environment. By this test we will get weather the material satisfy the Fickian
moisture diffusion property and there by the moisture diffusivity constant and the moisture equilibrium content.
In this test, the percent moisture mass gain versus time is monitored for a specimen that is maintained in a
steady-state environment at a known temperature and moisture exposure level until the material reaches
effective moisture equilibrium.
The moisture absorption specimens are cut in to according to ASTM standard, the specimens cut (50×25×3mm)
from plates precautions shall be taken to avoid notches, undercuts, rough or uneven surfaces or delamination
due to inappropriate machining methods. Aluminum foil is bonded to the edges such that moisture absorption
through the edges is essentially eliminated. The pre-conditioning of the specimen is done with vacuum oven at
50° C and then they were allowed to cool till they reached room temperature. Label the specimens so as to be
distinct from each other in a manner that will both be unaffected by the test and not influence the test, the
specimens are labeled using marker. Record the oven-dry specimen mass as the baseline mass, Wb, for moisture
absorption. The specimens are then placed in the conditioning chamber, which has previously reached the
specified steady-state environment. At the end of each time interval, the specimens are removed from the
conditioning chamber and wiped with an absorbent lint-free towel free. Then specimens kept for some time to
come to laboratory temperature, immediately weighed the specimens to the required precision. Recorded each
measurement as Wi, along with the corresponding total elapsed time and the time interval since the previous
weighing, and return the specimen to the conditioning chamber. Then calculated and recorded the percent mass
change at each time interval by Equation 2.1.
M, % = × 100 (2.1)
147
Where:
W = current specimen mass, g, and W = oven-dry specimen mass, g.
Water absorption tests were conducted by immersing the composite specimens in both distilled water (in
aluminum dry) at room temperature and at elevated temperature 80° C (conditioning chamber). The test at room
temperature was conducted only for 169 hours to get the initial behavior and for the comparison purpose
between the three composites in this investigation. From the experimental studies conducted by Girisha.C et al.
[1], we know that it will take time to reach moisture equilibrium. The elevated temperature test was conducted
until the water content reached saturation. The moisture absorption was calculated by the weight difference. The
percentage weight gain of the samples is measured at different time intervals and the moisture content versus
square root of time is plotted.
2.1.4 Finite Element Analysis (FEA)
Micro mechanical models and experimental tests were performed to obtain the mechanical properties individual
composite phases and the biocomposites in order to evaluate the finite element model and the interface
conditions. Two biocomposites were investigated in this section, one fabricated with untreated and another with
treated with 5wt % NaOH coir yarn fibers. The matrix phase and others are same as in experimental
investigation, fabrication also done with compression molding method. Tensile tests were carried out to
characterize the fibers and the matrix phase according to ASTM standards. The elastic properties were used to
perform the micro mechanical analyses of the composites reinforced with 30% of fibre volume fraction. Micro
mechanical models such as “Rule of mixture” and “Halpin -Tsai” were used to estimate the effective properties
of the biocomposites. The Rule of mixture considers a perfect interface condition and the Halpin -Tsai considers
a semi-empirical model. Table 2.2 shows not only the modulus of elasticity (E) of the fibers and the matrix, but
also the results obtained via micro mechanical analyses.
Table 2.2: Composites Properties

Material Method E(GPa)
Coir fiber Experimental 4.3 (4-6)
Polyester resin Experimental .6282
Experimental 1.98 to 2.4
Rule of mixture .280
Coir Composite
Halpin-Tsai .772
The software Ansys® 13 is used to simulate the biocomposites under tensile loadings. The numerical model was
fixed in one of the ends and the load was uniformly distributed in the cross section of the other end with eight
node elements, and the load increasing its intensity from 0 up to 3000 KPa, with increments of 500 KPa. The
model dimension is set as in experiment, 120mm×25mm×3.5mm. The loading levels were chosen to assure a
linear elastic analysis during the simulation. The pressure increments were applied in the numerical specimen,
∆
and the displacements were collected. Based on these results the strain (∈) was calculated by equation; ∈=
(where ∆L, is the length variation (mm) and L is the original length (mm) of the specimen). The numerical
stress/strain curves were plotted and compared with the experimental tensile test in order to evaluate the
interface condition.
3. RESULT AND DISCUSSION
3.1 Mechanical Characteristics of Composites
The characterization of the composites reveals that the surface modification of fiber is having significant effect
on the mechanical properties of composites. The properties of the composites with different surface
modification of fiber under this investigation are presented in Table 3.1.
148
Table 3.1: Mechanical Properties of the composites
Tensile
Flexural Impact Energy Water Diffusivity
Composites Strength
Strength (MPa) (mJ/ mm2) (mm2/s)
(MPa)
C1 16.220 54.8042 31.58330 0.105815
C2 17.346 60.9633 40.47466 0.250170
C3 15.949 49.7160 28.97081 0.074975
3.1.1 Effect of NaOH concentrations on Tensile Strength
The test results for tensile strengths of all the three composites are presented in Figure 1. It is seen that the
tensile strength of the composite increases with 5% NaOH treatment but it decreases with 10% NaOH treatment
compared to untreated composites. It is well known that for fiber reinforced composites, the interfacial zone
plays a leading role in transferring the load between fiber and matrix which consequently affects the mechanical
properties. As we know that tensile failure depends mainly on the fiber and matrix adhesion, the increased value
of tensile strength in case of alkali treated composite may be due to the improved wetting of coir with polyester
but in case of 10% NaOH treated composites the decrease may be due to cell wall thickening which leads to
poor adhesion with polyester resin.
Figure 1: Effect of NaOH treatments on tensile strength.
3.1.2 Effect of NaOH concentrations on Flexural Strength
Figure 2 shows the comparison of flexural strengths of the composites obtained experimentally from the bend
tests. It is interesting to note that flexural strength also following the same manner as the tensile strength, the
increased value of flexural strength in case of alkali treated composite may be due to the increase in effective
surface area available for contact with the matrix.
Figure 2: Effect of NaOH treatment on tensile strength.
149
3.1.3 Effect of NaOH concentrations on Impact Strength
The impact energy values of different composites recorded during the impact tests are given in Table 3.1. It
shows that the resistance to impact loading of coconut coir fiber reinforced polyester composites improves with
increase in NaOH concentration as shown in Figure 3. High strain rates or impact loads may be expected in
many engineering applications of composite materials. The suitability of a composite for such applications
should therefore be determined not only by usual design parameters, but by its impact or energy absorbing
properties.
Figure 3: Effect of alkali treatment on impact strength.
We know that the unidirectional orientation of continuous fibers in the composite produces the highest strength
and modulus in the fiber direction .While comparing the fiber architecture, tensile strengths and impact
strengths are approximately same and magnitudes of flexural strengths are quite different compared with
mechanical properties mentioned by J. Rout et al. [2] in section3.1. There can be two reasons for this in the
strength properties of these composites compared. One possibility is that the fibre dimensions of the various
individual cells are said to be dependent on the type of species, location and maturity of the plant. The flexibility
and rupture of the fibre is affected by the length to diameter ratio of the fibre and this also determines the
product that can be made from it. And the second possibility is the structure of the individual fiber cells are
narrow and hollow, with thick walls made of cellulose but to be useful, a yarn has to be basically of
homogeneous thickness, strength and suppleness, determined accurately according to the specific use of the
yarn and its specifications. This can result, during the incorporation of fibers into the matrix, air or other
volatiles may be trapped in the material. The trapped air or volatiles exist in the laminate as micro voids, which
may significantly affect some of its mechanical properties. This makes it difficult for their frequent use as
construction material.
3.2 The effect of a chemical treatment on the structure and morphology of coir fibers
Figure 4: Scanning electron micrographs of fibers (a-untreated, b-treated (5wt% NaOH), c-treated (10wt%
NaOH))
150
Scanning electron microscopy (SEM) was used to analyze the effect of the treatment on the structure and
morphology of the fibers. SEM photographs of the surface of the coconut coir fibers are shown in figures. From
figure 4.4 it can be seen that the untreated fibers surface contains randomly distributed globular protrusions.
From the analysis of the SEM photographs, the clear difference found between the surface morphology of the
untreated and treated fibers. The untreated fibers have an outer surface layer, which is completely removed by
the treatment. This layer is full of randomly distributed organic material with globular protrusions. Upon the
removal of this outer layer, a rougher, but more ordered structure is revealed. White dots are produced on the
surface of coir by NaOH treatment as a result of the removal of these globular protrusions which would increase
the mechanical interlocking between the fibre and the polyester resin.
3.3 Water Absorption Test Results
3.3.1 Water absorption at room temperature
Figure 5: Initial pattern of the water absorption curve at room temperature.
The water absorption behavior of the coir composites was observed by Girisha.C et al. [1], and they found out
the water absorption pattern of coir composites at room temperature following Fickian behavior. Figure 5 shows
the initial pattern of the water absorption curve at room temperature. From Figure it can be seen that the water
absorption for all specimens is linear in the beginning and this shows that the process is very similar to a Fickian
diffusion process. The moisture absorption increased with increasing alkali concentrations, the increasing water
absorption is may be due to the increase in interfacial area between the fiber and the matrix.
3.3.2 Water absorption at elevated temperature
Figure 6: Water absorption curve at elevated temperature.
The variation of moisture content at 800C is shown in the figure 6. For C1, C2 and C3 specimens the percentage
of moisture absorption is 4.4%, 5.5% and 10.8% respectively. From figure it is evident that the water absorption
properties at higher temperature did not follow Fick’s law, the effect of alkali concentration in surface treatment
on water absorption can be clearly seen. The rate at which the equilibrium is reached much more rapid for the
151
800C specimens than the samples immersed at room temperature and higher temperatures accelerate the
moisture uptake behavior.
3.4 Finite Element Analyses
Table 3.2 shows the numerical strain values and the error, corresponding to the percentage difference between
the experimental and numerical results obtained for each stress level. Figure 7 shows the stress-strain curves
obtained via finite element analyses for the numerical models (Rule of mixture and Halpin-Tsai) and
experimental test. The curve identified as “Experimental modulus” represents the numerical simulation data
based on the elastic properties via mechanical testing.
Table 3.2: Experimental and numerical stress-strain values and error percentage
Experimental
Experimental Rule of Mixture Halpin-Tsai
Modulus
Stress
(KPa) Strain Strain Error % Strain Error % Strain Error %
0 0 0 0 0 0 0 0
500 0.0031 0.001796 42.06 0.000651 78.98 0.002585 16.60
1000 0.005763 0.003592 37.66 0.001303 77.39 0.005171 10.28
1500 0.008933 0.005389 39.67 0.001954 78.12 0.007756 13.17
2000 0.012683 0.007185 43.35 0.002606 79.45 0.010341 18.46
2500 0.017473 48.60 0.003257 81.35 0.012927 26.02

0.008981
3000 0.024733 0.010777 56.42 0.003909 84.19 0.015512 37.28
Figure 7: Stress-strain behaviors of coir fiber composites by finite element analysis.
The large percent error indicates that the model was not able to predict the mechanical behavior of the
composites, even when the experimental elastic property was used. The simulation using the experimental
modulus of elasticity achieved better results compared to the micro mechanical analysis model but the error is
higher when the applied force is higher, this may be due to the miss alignment of the sample at the last moments
of the test. The FEA based on the elastic properties from the Rule of Mixture exhibited better results compared
152
to Halpin Tsai model. It is evident that the coir composites shows reverse manner while comparing with finite
element analysis of sisal fiber composites by Le andro José da Silva et al.[3]. In this way, it is possible to
conclude that the coir fibers provide a poor adhesion with the matrix phase even with surface treatment of the
fiber, showing a big difference between the numerical (rule of mixture and Halpin Tsai) and the experimental
results.
REFERENCES
[1] Girisha.C, Sanjeevamurthy, and Gunti Ranga Srinivas, Sisal/Coconut Coir Natural Fibers – Epoxy
Composites: Water Absorption and Mechanical Properties, International Journal of Engineering and Innovative
Technology (IJEIT) Volume 2, Issue 3, September 2012.
[2] Rout, J, M. Misra, S.S. Tripathy, S.K. Nayak, A.K. Mohanty,The influence of fibre treatment on the
performance of coir-polyester composites, Composites Science and Technology 61 (2001) 1303–1310.
[3] Le andro José da Silva, Túlio Hallak Panzera, André Luis Christoforo, Luís Migue l Pe re ira Durão and
Francisco Antonio Rocco Lahr, “Numerical and Experimental Analyses of Biocomposites Reinforced with
Natural Fibers”, International Journal of M aterials Engineering 2012, 2(4): 43-49.
[4] Anna Kicińska-Jakubowska, Edyta Bogacz & MałgorzataZimniewska, “Review of Natural Fibers. Part
I—Vegetable Fibers”, Journal of Natural Fibers, 9:150–167, 2012.
[5] Geethamma, VG, G. Kalaprasad, Gabriel Groeninckx and Sabu Thomas, “Dynamic mechanical behavior of
short coir fiber reinforced natural rubber composites”, Composites: Part A 36 (2005) 1499–1506.
[6] Herrera-Franco, PJ., & A. Valadez-Gonza ´lez, “A study of the mechanical properties of short natural-fiber
reinforced composites”, Composites: Part B 36 (2005) 597–608.
[7] Kasama Jarukumjorn, Nitinat Suppakarn, “Effect of glass fiber hybridization on properties of sisal fiber–
polypropylene composites”, Composites: Part B 40 (2009) 623–627.
[8] Mallick, PK, (2012), “Fiber-Reinforced Composites- materials, manufacturing, and design”, CRC
Press.
[9] Puglia, D, J. Biagiotti and J. M. Kenny, “Application of Natural Reinforcements in Composite Materials for
Automotive Industry”, Journal of Natural Fibers, 1:3, 23-65
[10] Morsyleide F. Rosaa, Bor-sen Chiou, Eliton S. Medeiros, Delilah F. Wood, Tina G. Williams, Luiz H.C.
Mattoso, William J. Orts and Syed H. Imam, “Effect of fiber treatments on tensile and thermal properties of
starch/ethylene vinyl alcohol copolymers/coir bio composites”, Bio resource Technology 100 (2009) 5196–
5202.
153
Paper No.: 026
EVALUATION OF PERFORMANCE POINT USING

ENERGY BASED PUSHOVER ANALYSIS
Prof N Murli Krishna1 and Md Masihuddin Siddiqui2
1
H.O.D, C.E.D; University College of Engineering; Osmania University; Hyderabad
2
Asst. Professor, C.E.D; Muffakham Jah College of Engineering & Technology; Hyderabad
Email: nmuralikrishna1956@gmail.com, masihuddin@mjcollege.ac.in.
ABSTRACT
Nonlinear static procedures (NSPs) are now widely used in engineering practice to predict seismic demands in
building structures. Seismic demands using NSPs can be computed directly from a site-specific hazard
spectrum. One of the well known NSP is Static Pushover Analysis which is explained in detail in many
international codes, viz., ATC40, FEMA273, FEMA356, Eurocode-8 etc. Pushover analysis leads to the
formation of the capacity curve of the structure, also known as Pushover curve. A Pushover curve is a plot of
base shear versus roof displacement. The use of roof displacement as the overall capacity index of the structure
in Pushover analysis, though convenient and practical for the engineer, lacks a consistent theoretical basis. In
addition, the area under the Pushover curve represents work of external forces that lacks natural meaning. This
paper evaluates the performance point of a regular building using Energy Approach of Static Pushover Analysis
method. It proposes the replacement of roof displacement that is used in classic “base shear – roof
displacement” pushover curve with an equivalent energy-based displacement, calculated as a function of the
work of external lateral forces acting on the structure. The Energy based methodology is applied on
characteristic multi-storey concrete building which is also analyzed using Conventional Pushover Analysis.
Thus the skeleton force-displacement (capacity) curve of the structure is converted into energy capacity plot
which is superimposed over the corresponding demand plot for the given hazard level to determine the
Performance point.
1. INTRODUCTION :
The purpose of the pushover analysis is to evaluate the expected performance of a structural system by
estimating its strength and deformation demands in design earthquakes by means of a static inelastic analysis,
and comparing these demands to available capacities at the performance levels of interest. The evaluation is
based on an assessment of performance parameters, like global drift, inter-storey drift, inelastic element
deformations (either absolute or normalized with respect to a yield value), deformations between elements and
connection forces. The inelastic static pushover analysis can be viewed as a method for predicting seismic force
and deformation demands, which accounts in an approximate manner for the redistribution of internal forces
occurring when .the structure is subjected to inertia forces that no longer can be resisted within the elastic range
of structural behavior. In pushover analysis procedures, the behavior of the structure is characterized by a
capacity curve. In nearly all cases, the capacity curve is a plot of the base shear force versus the displacement of
the roof. The seismic demands are determined throughout the structure based on the peak roof displacement
estimated in each of the modal pushover analyses. When nonlinear behavior develops in the pushover analysis,
the displacements of the floors and roof will increase disproportionately with increasing load, in general. The
arbitrary choice to plot the base shear as a function of the roof displacement introduces an arbitrariness to the
inelastic portion of the capacity curve. For systems with sharply defined yield points, disproportionate increases
in displacements over the height of the building, primarily affects the post-yield stiffness of the capacity curve.
Where yielding is more gradual, disproportionate increases in the roof displacement may, in addition, affect the
effective yield strength that is determined for the structure, when methods such as those described in ATC-40[4]
[I997] are used. Roof displacements may increase at a decreasing rate or may even reverse, leading to capacity
curves that display unusual behaviour, a literal interpretation of the capacity curves obtained in these cases
would indicate that the structure does not always absorb energy in a pushover analysis, but instead, may be a
source of energy for some inelastic regimes. Such an interpretation implies a violation of the first law of
thermodynamics, and points out the degree to which the use of the roof displacement can be misleading. There
is no doubt that external work is consumed by the deformations of plastic hinges (and any changes in
recoverable strain energy) that take place in a monotonic pushover analysis. The notion that the structure may be
a source of energy is a consequence of the arbitrary choice to use the roof displacement as the index (abscissa)
of the capacity curve.
2. ENERGY BASED APPROACH FOR THE PUSHOVER ANALYSIS:

A more general characterization of the resistance developed by a multi-degree-of-freedom (MDOF) structure
under increasing lateral loads in a pushover analysis by explicit consideration of the work done throughout the
entire structure gives better results of the response of the structure. This retains the utility of both first mode and
multiple mode methods of analysis while improving their theoretical basis. Rather than relying on the
displacements at an arbitrary location, absorbed energy (or equivalently, external work) is used to characterize
the resistance of the structure to lateral forces. The resulting capacity curves appear to be generally compatible
with those obtained from traditional first mode pushover analyses, and thereby provide a stronger theoretical
underpinning for the use of conventional first mode pushover analyses. At its core, the capacity curve of a
structure represents the development of resistance to lateral forces as a function of increasing lateral
displacements. The capacity curve has great value in characterizing the degree of nonlinearity that may develop
in a first or predominant "mode", recognizing, of course, that the onset of nonlinearity causes changes in modal
properties and invalidates modal superposition. Rather than relying on the roof displacement the use of energy
absorbed by the structure in each modal pushover analysis to determine the corresponding capacity curve of the
equivalent SDOF system, recognizing that the behavior of the MDOF system and its analogous SDOF system
can be appreciated from both conventional and energy-based perspectives. The energy-based formulation
developed below avoids the arbitrary selection of a single floor (or roof) location as the parameter for
representing the capacity curve, and may be used with single or multimode analysis procedures.
Fig.1 Extension of De,n definition to elastic and inelastic domain.
3. ENERY-BASED PUSHOVER ANALYSIS PROCEDURE

In the energy-based pushover approach, the capacity curve associated with each modal pushover is determined
based on the work done in the analysis. The work is computed incrementally typically for each step in the
pushover analysis.
155
Analysis Procedure: Step-by-step implementation of the Energy-Based POA procedure is presented.
1. Compute natural frequencies, ωn, and modes, φn, for linear-elastic vibration of the structure. The
conventional modal analysis has to be carried out for calculating the mass stiffness matrices. Later the
determinant of the |k – m * ωn2|= 0, for a trivial solution, where n is the number of modes. Sequentially the
modal frequency vectors are established by one of the values in the vector as 1 and solving for the other values
of the vector.
[K – M * ωi2] * Фin = 0
2. Define the force distribution: Sn = m.Фn. The force distribution over the structure is according to the mode
shape of the structure and will be as defined by the mode shapes for higher modes also.
3. Apply the force distribution characterized in step 2 incrementally, and record the base-shears and associated
story-displacements. The structure should be pushed just beyond the expected targeted roof displacement, in the
selected mode.
4. Employing the energy-based pushover approach presented by Hernandez-Montes et al. [2], the capacity curve
associated with each modal pushover analysis is determined based on the work done/energy absorbed in
displacing the structure. The work is computed incrementally typically for each step in the pushover analysis.
The increment in the energy-based displacement of the nth mode ESDOF system, ∆De, n, is obtained as:
------- Expression gives the response in the elastic domain of the structure.
∆De,n = ∆En x Vb,n --------expression gives the response for an incremental step of the POA.
where En = increment of work done by the lateral forces acting through the displacement increment associated
with one step of the nth mode pushover analysis and Vb,n = base-shear at that step of the pushover analysis,
which is equal to the sum of the lateral forces at that step. The incremental displacements, ∆De, n, are
accumulated (summed) to obtain the displacement De,n, of the ESDOF system at any given step in the modal
pushover analysis.
5. Idealize the pushover curve as a bilinear curve using the ATC-40[4] procedure.
6. Analyze the equivalent SDOF system of each mode through the POA method. Subsequently, obtain the
plastic moments of the members, for the given displacement De, n.
7. Energy shapes (ψ1, ψ2, ψ3…) as shown in Fig. 2b are obtained by normalizing the cumulative rotation on each
floor by the second floor’s rotation. The first-mode energy Ea1 is then distributed to each floor level and the base
of the frame with the chosen energy shape ψk. From these values of energy distribution vectors, the energy
distribution for each and every storey level is calculated selecting the appropriate energy shape vector as
suggested by C.C Chou and C.M Uang [1].
8. This procedure is repeated for each mode separately and finally, the desired responses are extracted from
pushover database values at the target displacement. However, in this process, each energy-based displacement
obtained from each record is a target displacement, separately for Energy-Based POA. The desired responses
are extracted from pushover database values at every target displacement. This method is more accurate,
although huge amounts of calculation are needed.
156
Fig.2 (a) Frame subjected to lateral forces fim. (b) Pushover curve of a single floor i.
9. Determine the Demand curve for each mode as describes earlier and plot the energy based capacity spectrum
and demand curve in a same graph for the respective modes, and the intersection of this demand curve with the
energy based capacity spectrum defines the performance point.
4. CASE STUDY
The building considered is having plan 4x3 bays of lengths 5m, 3m, 3m and 5m in X-direction and 4m, 3m and
4m in Y-direction of reinforced concrete ordinary moment resisting space frame of 7 storey configuration. Here
stiffness of the infill is neglected in order to account the nonlinear behavior of seismic demands.
The preliminary building data required for analysis assumed are presented in table 1 and the elevation and plan
view of the model is shown in Fig.3
Table 1 Assumed Preliminary data required for the Analysis of the frame
Sl.no Variable Data

1 Type of structure Moment Resisting Frame
2 Number of Stories 7
3 Bottom storey height 2.5m
4 Floor height 3m
5 Live Load 5.0 kN/m
6 Dead load 10.0 kN/m
7 Materials Concrete (M30) and Reinforced with HYSD bars (Fe415)
8 Size of Columns C1-500X500 mm C2-600X400 mm
9 Size of Beams 300x400 mm
10 Depth of slab 120mm thick

11 Specific weight of RCC 25 kN/m3
12 Zone II
13 Importance Factor 1
14 Response Reduction Factor 5
15 Type of soil Medium
157
Fig. 3 Elevation and Plan view of the structure in ETABS
4.1 Analysis Using Response Spectrum Method
The Structure is initially analysed & designed for Gravity loads and Response Spectrum. The Linear dynamic
analysis of the building models is performed using ETABS 9.7.4[5]. The lateral loads generated by ETABS
correspond to the seismic zone II and 5% damped response spectrum given in IS 1893-2002[3]. The fundamental
natural period values are calculated by ETABS, by solving the eigenvalue problem of the model. Thus, the total
earthquake load generated and its distribution along the height corresponds to the mass and stiffness distribution
as modeled by ETABS. The response spectrum for the structure is defined in the SAP 2000 V.15.0 [6] as Spec 1
(Spectrum) and the response spectrum analysis data is defined in the load cases defined. Fig. 4 shows the details
of the Response history analysis case. The response spectrum analysis is run and the members are designed for
the analysis results obtained, by IS-456:2000 design specifications.
158
Fig. 4 Response Spectrum Load Case Data
4.2 Pushover Analysis for Different Modes
The Structure is then subjected to pushover forces, however before the actual analysis the Non-linear hinges
needed to be defined to the model. After assignment of the hinges, the pushover cases are defined under the
conjugate monitored displacement. The Dead loads case is assigned to non-linear static case, and the pushover
case data is given as shown in the Fig 5. The top displacement of the structure is set to 4% of the total height of
the structure at one of the top storey nodes. The degree of freedom (U1) for the structure is in the direction of
the application of the pushover loads i.e., along the x-axis.
Figure 5 Pushover Analysis Load Cases Data

The following table contains the results of the push over analysis and energy based pushover analysis carried
out for the structure as discussed earlier. The results are of different pushover cases and each pushover case is
for specific mode and direction i.e., PUSH-1Y represents pushover in mode 1 and in Y-direction. The values
tabulated are of the critical modes of the structure, which have been considered in the analysis.
Table 2 Results of the POA and EBPOA of the structure.
POA EBPOA
Case
Max Base Shear Max Disp. Max Base Shear Max Displ.
(kN) (m) (kN) (m)
PUSH-1Y 1635.9364 0.2831 1635.9364 0.215
PUSH-2X 1860.1984 0.2945 1860.1984 0.223
PUSH-3Y 3828.0293 -0.0357 3828.0293 0.030
The results for EBPOA are obtained from the equations proposed by Hernandez Montes [2], while the results for
POA have been obtained directly from the ETABS output file. The use of the roof displacement in the
conventional POA approach leads to an apparent stiffening in the post – yield response, while the energy-based
159
approach shows monotonic softening with increasing displacement. Because each plastic hinge has a bilinear
moment -rotation relationship the apparent stiffening after initial yielding is not physically reasonable, but
instead must be viewed as a consequence of the arbitrary choice to index the capacity curve by displacement
that increases disproportionately in the nonlinear regime.
5.1 Capacity Curves
The pushover curves are obtained using the procedure given by Prof. Hernandez Montes [2].The pushover
analysis for different modes is performed and the capacity curves are plotted from the analysis results obtained
for POA and Energy-Based Pushover analysis. Fig.6a, 6b and 6c shows the comparison of pushover curves for
the first, second & third modes. It is seen that the energy based pushover curve nearly coincides with the
conventional pushover capacity curve for the first two modes. In conventional pushover analysis the pushover
curve is getting reversed which was the major drawback of this method, however in energy based method, the
pushover curve is maintaining its shape and there is no mode reversal.
Fig. 6a Push-over curve for mode 1 Fig. 6b Push-over curve for mode 2
Fig. 6c Push-over curve for mode 3
To develop the performance point, the pushover curve is converted into Capacity spectrum curve using bilinear
capacity spectrum. Fig. 7a shows capacity spectrum curve and Fig. 7b shows the bilinear curve of the first mode
pushover case.
160
Fig. 7a Capacity Curve for mode 1 Fig.7b Bilinear capacity spectrum for mode 1
5.2 Demand Curves
Response spectra is plotted for zone factor-2B and effective peak acceleration (EPA) of the ground is 0.24*g as
per ATC-40 as shown in Fig. 8a. It is then converted into Acceleration Displacement Response Spectra (ADRS)
format i.e. spectral acceleration v/s spectral displacement as shown in Fig. 8b. And then a series of family of
response spectra is plotted corresponding to different effective damping values (βeff) starting from 5% to
maximum of 29% for type B structure. Demand curve is generated.
Fig. 8a Response Spectrum Fig. 8b Response Spectrum

(Standard Format) (ADRS Format)
5.3 Performance Point
The energy capacity and demand curves at yield of the frame are overlapped and the performance point which is
the intersection of capacity curve and demand curve calculated. Fig 9a, 9b & 9c shows the performance point of
the structure for the first three modes calculated using Energy Based Pushover Analysis.
161
Fig. 9a Demand curve & Capacity curve for Mode 1 Fig. 9b Demand curve & Capacity curve for Mode 2
Fig. 9c Demand curve and Capacity curve for Mode 3
6. CONCLUSIONS
The general NSPs show the application of the force profile, and the response of the structure is calculated
accordingly, mostly without consideration of the energy or work done in the system. The adaptive procedure
should be based on the equations of motion and if this is the case, adaptive procedures depend entirely upon the
input forces. Hence the response is associated with the energy stored in the structure.
The capacity curve determined in a pushover analysis reveals the rate at which a structure develops resistance to
lateral forces of a given pattern; this resistance has been viewed conventionally from the point of view of roof
displacement or alternatively from the perspective of the energy absorbed in the lateral load analysis. For
yielding systems, disproportionate increases in displacements affect the slope of the capacity curve. The
response of the SDOF systems with negative post-yield stiffness is known to be sensitive to the post-yield
stiffness, and therefore, estimates made using analogous SDOF systems may be sensitive to the displacement
index used for the capacity curve. For higher mode pushover analyses, the roof displacements get reversed,
leading to the reversal of modes.
162
The energy absorbed by the MDOF structure in the pushover analysis has been used to derive an energy-based
displacement that characterizes the work done by the equivalent SDOF system. Thus, in contrast to the
conventional view of pushover analysis, pushover analysis has been equivalently viewed in terms of the work
done (or absorbed energy). The data associated with roof displacements has been used to determine the energy-
based displacement, De,n. The capacity curve of the equivalent SDOF system was then obtained using
conventional transformations of the base shear together with the energy-based displacements, as illustrated
schematically. The energy-based capacity curve was defined to match the capacity curves obtained using
conventional approaches for the first and higher mode analyses in the elastic and inelastic domains. The energy-
based capacity curves are able to correct the anomalies observed for the higher mode pushover curves. It is
concluded that the energy-based formulation provides a stronger theoretical basis for establishing the capacity
curves of the first and higher mode equivalent SDOF systems.
REFERENCES
[1] Chung-Che Chou and Chia-Ming Uang (2001), “A procedure for evaluating seismic energy demand of
framed structures”, † Department of Structural Engineering; University of California; San Diego; La Jolla;
CA 92093; U.S.A.
[2] Hernandez-Montes, E., Kwon. O.S. and Aschheim, M. (2004), “An energy-based formulation for first- and
multiplemode nonlinear static (pushover) analyses", J. Earthquake Eng., 8(1), pp. 69-88.
[3] IS-1893:2002, Criteria for Earthquake resistant Design of structures.
[4] ATC-40 Document by Applied Technical Council.
[5] ETABS 9.7.4, Structural Analysis Program, Computers and Structures, Inc., Berkeley, CA.
[6] SAP2000 V.15.0, Structural Analysis Program, Computers and Structures, Inc., Berkeley, CA.
163
Paper No.: 028
SEISMIC BEHAVIOUR OF R/C FRAMES WITH BIAXIAL INTERACTION

Arghya Sengupta 1 and Rana Roy2
1
Graduate Student, Department of Applied Mechanics, Bengal Engineering and Science University, Shibpur,
Howrah 711 103, INDIA.
2
Associate Professor, Department of Applied Mechanics, Bengal Engineering and Science University, Shibpur,
Howrah 711 103, INDIA [Corresponding author]
Email: sgarghya@gmail.com, rroybec@yahoo.com.
ABSTRACT
Current seismic design practice for ductile moment-resisting frames aims to ensure ‘strong-column, weak-
beam’, i.e., the strength of a column under combined bending and axial load at yield must be greater than the
yield capacities of the beams framing into it. In the codified approach, structures are analyzed along a principal
direction of the system under seismic force calculated using a reduced smooth elastic spectrum and also the
capacities are estimated based on the uni-axial member behavior. However, during seismic shaking, columns
are usually subjected to bi-directional ground motion while the beams undergo uni-directional motion only. The
present investigation, in this backdrop, attempts to systematically examine the change in response under
selected near-field records. A simple one-story frame supported by four reinforced concrete (R/C) columns is
considered. Columns are modeled using fiber model to adequately capture bi-axial interaction. Trends in
response observed in terms of variation of drift and hysteretic energy dissipation characteristics is summarized
and interpreted. The study may prove useful for practical design of R/C moment-resisting frames.
KEYWORDS: Bi-axial, fiber, reinforced concrete, seismic, fiber model, drift, hysteretic energy.
1. INTRODUCTION
In the framework of performance-based seismic design, precise assessment of structural demand is imperative
rather than a conservative estimation with simplifying assumptions. In the current seismic design practice,
structures are analyzed along a principal direction of the system under seismic force estimated using a reduced
smooth elastic spectrum. Further, the capacities are estimated based on the uni-axial member behavior.
However, during seismic shaking, lateral load-resisting elements like columns, are usually subjected to bi-
directional ground motion that may considerably increase inelastic demand (Oliva and Clough, 1987; Sakai and
Sheikh, 1989). Existing design approaches (e.g., ASCE/SEI 7-05) do not explicitly account for the bi-axial
interaction except for systems with a specific form of irregularity (Type 5) where ‘100% of the forces for one
direction plus 30% of the forces for perpendicular direction’ is suggested. According to the “30%-rule,” the total
response is computed as 100% of the maximum response for excitation in one direction plus 30% of the
maximum responses for excitation in the other horizontal direction; the combination leading to the largest
response is considered for design (Clough and Penzien, 1993). Responses under uni-directional components
separately applied to the extent of 100% and 30% along the principal directions of the structures are often
combined through SRSS. It is apparent that these approaches are untenable at least for systems excited well into
the post-elastic range.
In this backdrop, behavior of lateral load-resisting members under cyclic loading is reviewed. Upon the
accepted premise of capacity design philosophy to dissipate energy primarily through inelastic bending of
beams, majority of the experimental works and modeling scheme aims at uni-axial flexure without axial force.
In fact, experimental results on members with medium to high shear span under uni-axial cyclic loading with no
axial force is abundant (e.g., Burns and Siess, 1962; Bertero et al., 1969; Brown and Jirsa, 1971; Hansen and
Conner, 1972; Wight and Sozen, 1975). Experimental studies are also conducted on R/C members under cyclic
flexure and constant axial force (e.g., Atalay and Penzion, 1975; Priestley and Park, 1987; Otani and Cheung,
1981).
On the other hand, research progress on the seismic behavior of structures accounting bi-axial bending is rather
limited. Comprehension on the impact of multi-axial loading, developed from the limited experimental and
analytical studies, are summarized in the literature (CEB, 1996). Experimental studies on square (Okada et al.,
1976; Otani et al., 1980, Saaticioglu, 1984; Takizawa and Aoyama, 1976; Takiguchi and Kokusho, 1977) and
circular (Karlson et al., 1973; Kobayashi et al., 1984) columns, with constant axial load or at its exclusion, offer
limited information on stiffness, deformability and energy dissipation of columns under multi-axial loading.
These studies reveal that the effect of bi-axial interaction may cause early yielding of the member and
significantly add to the degradation of stiffness and strength of R/C members. Results of experimental shaking
table test under bi-axial loading on frames with rectangular column (Oliva and Clough, 1987) clearly recognize,
unlike uni-axial loading, yielding in columns and not in beams contrary to the design intents.
Thus, the seismic demand and capacity of columns estimated at the exclusion of bi-axial interaction may not be
pragmatic. Further, behavior of axially loaded R/C members subjected to cyclic bi-axial bending may increase
the degradation of strength and stiffness and reduce the capacity of the columns. On the other hand, three
dimensional character of the response of the frame structures during actual seismic event does not cause similar
detrimental effect in beams. Thus, the bi-axiality of the column moments and three-dimensional nature of the
overall response contribute together against the current design aim (‘strong column-weak beam’). This
motivates the authors to explore the implications of biaxial interaction in R/C frames relative to the uni-axial
counterpart. The trend in results summarized and interpreted may prove useful to develop appropriate seismic
design strategy.
2. REVIEW OF THE EXISTING ANALYTIC MODEL
Different analytical models have been conceptualized to date for the nonlinear analysis of R/C frame structures
under cyclic loading. Typically, hysteretic response of R/C member is a complex combination of a number of
phenomena, viz., flexural rotation, shear deformation including shear sliding and the bond-slip.
Two basic approaches are available to model the behaviour of structural members under cyclic loading. In the
former, inelasticity is averaged at the ends, while in the other; damage is monitored at various sections along the
length. These are referred to as lumped and distributed plasticity models respectively. Comprehensive review on
this subject is available elsewhere (Zeris 1986; Takizawa and Aoyama 1976; Aktan et al. 1973). In Lumped
nonlinearity models, using simplified interaction plasticity surfaces and flow rules (Nigam, 1967, Nigam and
Housner, 1969, Takizawa and Aoyama 1976), effect of bi-axial loading on column has been investigated. Such
models generally predict overall response reasonably. Local response cannot, however, be adequately
represented. These models are also extended to incorporate the effect of axial load (Fukuzawa et al. 1988; Otani
et al. 1979; Chen. 1982). Apart from such Yield surface models, Lai et al. (Lai et al., 1984) developed a Spring
model involving five springs that is observed to demonstrate good prediction of R/C columns under bi-axial
loading and constant axial load. Studies to estimate the effective properties and locations of such springs
(Ghusan and Saiidi, 1989; Lai, 1987; Jiang and Saiidi, 1990) have also been conducted.
With the advent of increasing computational power, Distributed inelasticity fiber models are often preferred.
This model has the unique capability to represent the cross-section behavior through employing uni-axial stress-
strain relationship of the discrete fibers. This facilitates straightforward representation of bi-axial loading and
interaction between flexural strength in orthogonal directions. In a fibre model, the global response curve
represents the integral of the local state of each section at the level of each Gauss point of each fibre. The global
force-displacement curve is not pre-defined but is the result of the integral of all the section forces. Any
variation of the axial force and/or application of biaxial bending moments are automatically taken into account
in the equilibrium equations. The present paper is a systematic attempt to investigate the influence of bi-axial
interaction on simple R/C frames under seismic events employing such advanced fiber based modeling.
165
3. IDEALIZATION OF THE SYSTEM
Simplified building model, idealized as rigid diaphragm (rigid in plane and flexure), is considered (Figure 1).
The structure is symmetric in configuration with a square deck of side 4.0m supported by four columns each of
height 3.0 m. Mass is assumed to be lumped at the centroid of the deck. Mass is adjusted to set the building
periods as 0.2 sec., 1.0 sec and 3.0 sec. respectively representative of short, medium and long period systems.
3.1 Modeling of frame members
Structure is modeled using standard finite element software (SeismoStruct: Version 6). Time-domain analysis
excludes the effect of geometric non-linearity and includes material inelasticity only. Structural members are
modeled using distributed plasticity fiber elements, which use member cross-section properties and uni-axial
constitutive behavior to explicitly define element hysteretic behavior. Classical Forced-based formulations,
details of which are available in the literature (Alemnder and White, 2005; Hjelmstad and Taciroglu, 2005), are
adopted. The concrete constitutive behavior is modeled using non-linear constant confinement concrete model
(Mander et al., 1988) while Menegotto-Pinto steel model (Menegotto and Pinto, 1973) with 5% strain hardening
is used to model constitutive behavior of reinforcing steel. In this distributed plasticity fiber model, the
structural member is modeled as inelastic element, the source of such inelasticity being defined at sectional
level. The global inelasticity of the frame is then obtained by integration of the contribution provided by each
controlling section. Member cross-section is discretized into two hundred fibers and four Gauss sections
(controlling section) per element are considered to ensure convergence. The Gauss-Lobatto quadrature rule is
used to numerically integrate the forced-based elements. This numerical scheme is deemed adequate following
the excellent investigation on the numerical issues involved in distributed inelasticity modeling of R/C frame
elements for seismic analysis purpose (Calabrese et al., 2010).
4. GROUND MOTIONS
Seismic response is known to be sensitive to the ground motion characteristics such as frequency content,
pattern of pulses, duration of shaking, fault-rupture mechanism etc. In this context, to comprehend the basic
trend in response, three ground motion histories are selected (available at PEER ground motion database:
http://peer.berkeley.edu). The moment magnitude for the ground motions ranges between 5.8 and 6.9. Closest
site-to-fault-rupture distance ranges between 4.0 km and 8.5 km and hence the selected records may be treated as
near-field records. Average shear wave velocity in the top 30 meters of the site (Vs30),being around 280 m/s,
represents a site class D (FEMA 356: 2000). The acceleration-time histories along with the relevant response
spectra for each component of ground motions are furnished in Figure 2.
5. METHOD OF ANALYSIS
Nonlinear constant confinement concrete model (Mander et al., 1988), is adopted considering no tensile strength
while grade of concrete, strain at peak stress and confinement factor are chosen respectively as 30 MPa, 0.0028
and 1.5. Menegotto-Pinto steel model (Menegotto and Pinto, 1973) is used considering grade of steel as 415
MPa with a strain hardening parameter of 0.005. Relevant other parameters in such model viz., transition curve
initial shape parameter, transition curve shape calibrating coefficient (A1), transition curve shape calibrating
coefficient (A2), isotropic hardening calibrating coefficient (A3), isotropic hardening calibrating coefficient
(A4) and fracture/buckling strain are respectively taken as -20, -18.5, -0.15, 0, -1 and -0.1 following the
guideline of the software (SeismoStruct: Version 6). Axial load ratio, P/(fckAg) where P, fck and Ag are
respectively axial load, compressive strength of concrete and gross sectional area are considered to vary as 10%,
20%, 25% and 29%.
Capacity curve of the structural system so modeled is established by conducting a conventional static pushover
analysis where lateral load is applied employing automatic response control algorithm (Izzuddin 1991) through
166
standard software (SeismoStruct: Version 6). Yield capacity of the structure is estimated from capacity curve of
the structure so established. Effect of axial force is not considered in this process. Subsequently, computing
elastic base shear of the system under a specific ground motion appropriate scale factor is determined to
simulate a response reduction factor (R) taken herein as 2.
Under specified ground acceleration histories, standard equations of motion (refer to Clough and Penzion, 1993;
Chopra, 2007) of the system are solved in the time domain using Hilber-Hughes-Taylor integration scheme.
Hilber-Hughes-Taylor parameters α, γ and β are chosen respectively as -0.1, 0.6 and 0.3025. Time step of
integration is taken to be sufficiently small to ensure convergence based on the computed displacement and
rotation with a precision limit of 0.1mm and 0.0001 rad. 2% of stiffness proportional damping based on initial
stiffness is considered to constitute damping matrix.
With the intent of assessing influence of bi-directional effect, two sets of analysis are conducted. In the first set,
each component of ground motion is applied separately along X (or Y) direction. Drifts of the column under
such uni-directional excitation are obtained from these two separate analyses as ∆U1 and ∆U2 respectively with
and without accounting axial load. Drift so computed at different percentage of axial load is normalized by the
corresponding drift estimated without axial load. Maximum normalized drift for the components are averaged
and mean maximum normalized drift is presented as a function of the corresponding axial load ratio (Figure
3(a)).
In the second, both the components of a ground motion are simultaneously applied and drift along x and y-
directions, viz., ∆XBi and ∆YBi computed from such single analysis are geometrically combined as [∆XBi2 +
∆YBi2]0.5. Drift under bi-directional seismic attack is evaluated with and without axial force and mean maximum
normalized drift under bi-directional shaking is computed. Variation of such response parameter with change of
axial force is furnished in Figure 3(b). It is observed that the drift tends to consistently reduce in presence of the
axial load up to a certain limit for systems with all periods. This may be a consequence of the increase of section
capacity related to the transition from tension control to balanced state. To explicitly identify the impact of
biaxial interaction, drift under bi-directional excitation normalized to that due to uni-directional excitation is
presented in Figure 3(c). Results suggest that the impact of bi-directional excitation may increase drift even to
the extent of 40% for stiff systems. For medium to long period systems, such amplification may range between
5-20%.
Albeit drift is recognized to be a good descriptor of structural performance, quantifying hysteretic energy
dissipation may offer better insight into the impact of bi-axial interaction under seismic loading, in particular.
To this end, base shear-top displacement hysteresis diagrams are presented in Figure 4 for the three selected
ground motions. Hysteresis loops under uni-directional and bi-directional shaking, furnished separately for X
and Y directions, are also superimposed for comparison. Component identifier of the ground motions applied in
two orthogonal directions of the building (viz., X and Y) are also specified therein. It deserves mention that the
hysteretic response including interaction effect, although presented separately in X and Y directions, occurs
together in case of bi-directional excitation. However, responses evaluated at the exclusion of bi-axial
interaction under two components of ground motions are uncoupled. Thus the areas of the hysteresis loops in X
and Y directions under bi-directional excitation are added and compared to the individual maximum between
the areas of the hysteresis loops in X and Y directions under uni-directional excitation (Figure 5). Such
normalized hysteretic energy also tends to demonstrate good agreement with that observed in terms of drift. The
influence of axial load, for medium to long period systems, may increase hysteretic energy dissipation to the
extent of 10 to 25%. (refer to Figure 5(a) and 5(b)). However, demand appears to decrease in presence of axial
load for short period system. Similar trend in behavior is observed under both uni-directional and bi-directional
excitation. Hysteretic energy dissipation under bi-directional shaking relative to uni-directional counterpart as
presented in Figure 5(c) may range from 40 – 140%.
167
It may be interesting to evaluate the combined impact of bi-axial interaction with axial load relative to uni-axial
shaking without axial load. Response under bi-axial shaking with axial load is normalized to that under uni-axial
shaking without axial load (Figure 6(a)). This shows that the drift demand may be estimated through uni-
directional shaking (assuming no axial load) with adequate margin unless load is very low. However, hysteretic
energy dissipation may be grossly underestimated from such uni-directional analysis (except for very stiff
system) (Figure 6(b)).
7. SUMMARY AND CONCLUSIONS
Existing practice of estimating inelastic seismic demand of lateral load-resisting structural members usually
employs overly simplistic combination of the responses under uni-directionally applied ground motion
components. This study on the effect of the possible influences of bi-directional interaction using a
representative structural model such as columns offers the following preliminary observations:
(a) Bi-axial interaction seems to amplify the drift of the structures particularly for stiff to medium period
systems. Such magnification is generally subdued in the long period. However, the associated influence on
the hysteretic energy dissipation appears to be potentially magnified in the entire period range.
(b) It is observed that the inclusion of axial load effect may reduce the drift demand under both uni-directional
and bi-directional shaking relative to companions without axial load. However, the overall damage may be
more in presence of axial load as is apparent from the trend in variation of hysteretic energy dissipation.
Further, the impact of bi-directional interaction as compared to the unidirectional counterpart in presence of
the similar axial force in both situations may considerably increase the response.
(c) Analysis of the combined influence of bi-axial interaction and axial load relative to uni-axial shaking
without axial load suggests that the drift demand may be conservatively estimated through uni-directional
analysis (assuming no axial load). This may, however, grossly underestimate the overall damage as
revealed from hysteretic energy dissipation (except for very stiff system).
The preliminary observations of the present investigation, therefore, recognize the inadequacy of the current
design approach and prepare the framework to explore the interaction effect. Further investigations on
multistory structures with suites of ground motions are currently underway by the authors.
8. DIAGRAMS
(a) (b) (c)
Figure 1: Idealization of R/C Frame (a) Geometrical details; (b) Details of section X-X;
(c) Typical reinforced concrete element.
168
Spectral acceleration
Acceleration (m/s2)
Acceleration (m/s2)
NGA_no_547_B-ZAK360 NGA_no_547_B-ZAK270
NGA_no_547_B-ZAK360
NGA_no_547_B-ZAK270
(m/s2)
Time (sec) Time (sec) Period (sec)
(a) Event: Chelfant Valley-01 dt. 20.07.1986, Station: Zack Brothers Ranch
(M = 5.8, r = 6.4 km, Fault type = SS, Vs30 = 271 m/s).
Acceleration (m/s2)
Acceleration (m/s2)
NGA_no_180_H-E05230 NGA_no_180_H-E05140 NGA_no_180_H-E05230
NGA_no_180_H-E05140
(m/s2)
(b) Event: Imperial Valley dt. 15.10.1979, Station: EL-Centro-Array#5
(M = 6.5, r = 4.0 km, Fault type = SS, Vs30 = 206 m/s).
Acceleration (m/s2)
Acceleration (m/s2)
NGA_no_802_STG090 NGA_no_802_STG000
NGA_no_802_STG090
NGA_no_802_STG000
(m/s2)
(c) Event: Loma Prieta dt. 18.10.1989, Station: Saratoga-Aloha Ave

(M = 6.9, r = 8.5 km, Fault type = RV-OBL, Vs30 = 371 m/s).
Figure 2: Ground motion histories and associated spectra.

normalized drift
Mean maximum
∆ 1.0 sec
3.0 sec
0.2 sec
normalized drift
Mean maximum
∆ ∆
normalized drift
Mean maximum
∆
∆ ∆
Axial load ratio (%) Axial load ratio (%) Axial load ratio (%)
(a) (b) (c)
Figure 3: Change of drift due to bi-directional interaction incorporating axial load

Uniaxial
Biaxial
T = 0.2 sec T = 1.0 sec T = 3.0 sec

Base shear (kN)
Base shear (kN)
Base shear (kN)
N=0 N = 20 N = 29
NGA_no_547_B-ZAK360 NGA_no_180_H-E05230 NGA_no_802_STG090
Roof displacement (m) Roof displacement (m) Roof displacement (m)
Figure 4: Typical hysteresis behavior of Column under unidirectional and bi-directional

169shaking.
Mean normalized
Mean normalized
Mean normalized
hysteretic energy
hysteretic energy
hysteretic energy
1.0 sec
3.0 sec
0.2 sec
Axial load ratio(%) Axial load ratio(%) Axial load ratio(%)
∆
(a) (b) (c)
Figure 5: Change of hysteretic energy due to bi-directional interaction incorporating axial
load
1.0 sec
3.0 sec
0.2 sec
Mean normalized
hysteretic energy
normalized drift
Mean maximum
∆
∆
Axial load ratio (%) Axial load ratio (%)

(a) (b)
Figure 6: Change of drift and hysteretic energy due to bi-directional

interaction incorporating axial load
1.0 sec
3.0 sec
T = 0.2 sec T = 1.0 sec T = 3.0 sec
0.2 sec
%
Axial load ratio (%) Axial load ratio (%) Axial load ratio (%)
(a) (b) (c)
Figure 7: Percentage change in ratio of eccentricity to lateral dimension of the
deck incorporating axial load
REFERENCES
[1] Aktan, A.E., Pecknold, D.A.W. and Sozen, M.A. (1973). Effect of Two-dimensional Earthquake Motion
on a Reinforced Concrete Column, Structural Research Series No. 399, University of Illinois, Urbana.
[2] Alemdar, B.N. and White, D.W. (2005). Displacement, Flexibility and Mixed Beam- Column Finite
Element Formulations for Distributed Plasticity Analysis, Journal of Structural Engineering, ASCE,
131(12), 1811-1819.
[3] Assa, B. and Nishiyama, M. (1998). Prediction of Load-Displacement Curve of High-Strength Concrete
Columns under Simulated Seismic Loading, ACI Structural Journal, 95(5), 547-557.
170
[4] Atalay, M. and Penzien, J. (1975). Behaviour of Critical Regions of Reinforced Concrete Components as
Influenced by Moment, Shear and Axial Force, Earthquake Engineering Research Center, University of
California, Berkeley, CA, Report no. UCB/EERC 75/19.
[5] Bertero, V.V.; Bresler, B. and Liao, H. (1969). Stiffness Degradation of Reinforced Concrete Members to
Cyclic Flexural Moments, Earthquake Engineering Research Center, University of California, Berkeley,
CA, Report no. UCB/EERC 69/12.
[6] Brown, R.H. and Jirsa, J.O. (1971). Reinforced Concrete Beams under Load Reversals, ACI Journal,
68(5), 380-390.
[7] Burns, N.H. and Siess, C.P. (1962). Load Deformation Characteristics of Beam - Column Connections in
Reinforced Concrete, Structural Research Series No. 234, Civil Engineering Studies, University of
Illinois at Urbana - Champaign.
[8] Calabrese, A.; Almeida, J.P. and Pinho, R. (2010). Numerical Issues in Distributed Inelasticity Modeling
of R/C Frame Elements for Seismic Analysis, Journal of Earthquake Engineering, 14(S1), 38-68.
[9] Chen, P. (1982). Generalized Plastic Hinge Concepts for 3D Beam – Column Elements, Earthquake
Engineering Research Center, University of California, Berkeley, CA, Report no. UCB/EERC 82/11.
[10] Chung, Y.S.; Meyer, C. and Shinozuka, M. (1998). Modeling of Concrete Damage, ACI Structural
Journal, 86(3), 259-271.
[11] Clough, R.W. and Johnson, S. B. (1966). Effect of Stiffness Degradation on Earthquake Ductility
Requirements, Proceedings 2nd Japan Earthquake Engineering Symposium, 227-232.
[12] Clough, R.W. and Penzien, J. (1975). Dynamics of Structures, McGraw-Hill, New York.
[13] Chopra, A.K. (2007). Dynamics of Structures – Theory and Applications to Earthquake Engineering,
Pearson Education, New Jersey.
[14] Correia, A.A. and Virtuoso F.B.E. (2006). Nonlinear Analysis of Space Frames, Proceedings of the 3rd
European Conference on Computational Mechanics: Solids, Structures and Coupled Problems in
Engineering, Mota Soares et al. (Eds.), Lisbon, Portugal.
[15] Dutta, S.C. (1995). Hysteresis Models for Nonlinear Behaviour of RC Members, Asian Journal of
Structural Engineering, 1(1), 37-70.
[16] Dowell, R.K.; Seible, F. and Wilson, E.L. (1998). Pivot Hysteresis Model for Reinforced Concrete
Members, ACI Structural Journal, 95(5), 607-617.
[17] CEB 231: The International Federation for Structural Concrete. (1996). R/C Frames under Earthquake
Loading, Thomas Telford Limited, London.
[18] Fragiadakis, M.; Pinho R. and Antoniou S. (2008). Modelling inelastic buckling of reinforcing bars under
earthquake loading," in Progress in Computational Dynamics and Earthquake Engineering, Eds. M.
Papadrakakis, D.C.; Charmpis, N.D.; Lagaros, Y.and Tsompanakis, A.A., Balkema Publishers – Taylor &
Francis, The Netherlands.
[19] Filippou, F.C.; Popov, E.P. and Bertero V.V. (1983). Effects of bond deterioration on hysteretic
behaviour of reinforced concrete joints, Earthquake Engineering Research Center, University of
California, Berkeley, CA, Report no. UCB/EERC 83/19.
171
[20] Fukuzawa, E.; Isozaki, Y. and Fuzisaki, K. (1988). Elastic – Plastic Earthquake Response Analysis of
R/C Frame in Consideration Fluctuation of Axial Forces on Columns, Proceedings 9th World Conference
on Earthquake Engineering, 8, Japan, 549-554.
[21] Izzuddin, B.A. (1991). Nonlinear Dynamic Analysis of Framed Structures, PhD Thesis, Imperial College,
University of London, London, UK.
[22] Jiang, Y. and Saiidi, M.S. (1990). 4-Spring Element for Cyclic Response of R/C Columns, Journal of
Structural Engineering, ASCE, 116(4), 1018-1029.
[23] Karlson, B.; Aoyama, H. and Sozen, M. (1973). Spirally Reinforced Concrete Columns Subjected to
Loading Reversals Simulating Earthquake Effects, Proceedings 5th World Conference on Earthquake
Engineering, Rome, Italy, 803-806.
[24] Kobayashi, K.; Kokusho, S.; Takiguchi, K. and Yang-Boo, C. (1984). Study on the Restoring Force
Characteristics of R/C Column to Bidirectional Deflection History, Proceedings 8th World Conference on
Earthquake Engineering, San Francisco, California, 537-544.
[25] Kwak, H.G.; Kim, S.P. and Kim, J.E. (2004). Nonlinear Dynamic Analysis of R/C Frames using Cyclic
Moment – Curvature Relation, Structural Engineering and Mechanics, 17(3-4), 357-378.
[26] Lai, S.S.; Will, G.T. and Otani, S. (1984). Model for Inelastic Biaxial Bending of Concrete Members,
Journal of Structural Division, ASCE, 110(11), 2563-2584.
[27] Mander, J.B.; Priestley, M.J.N. and Park, R. (1988). Theoretical Stress – Strain Model for Confined
Concrete, Journal of Structural Engineering, ASCE, 114(8), 1804-1823.
[28] Mpampatsikos, V.; Nascimbene, R. and Petrini L. (2008). A critical review of the R/C. frame existing
building assessment procedure according to Eurocode 8 and Italian Seismic Code, Journal of Earthquake
Engineering, 12(SP1), 52-58.
[29] Nigam, N. (1967). Inelastic Interactions in the Dynamic Response of Structures, EERL 67-64, Earthquake
Engineering Research Laboratory, California Institute of Technology, Pasadena, California.
[30] Okada, T.; Seki, M. and Asai, S. (1976). Response of Reinforced Concrete Columns to Bidirectional
Horizontal Force and Constant Axial Force, Bulletin of Earthquake Research Center, (10), University of
Tokyo, Tokyo, Japan, 30-36.
[31] Oliva, M. and Clough, R. (1987). Biaxial Seismic Response of R/C Frames, Journal of Structural
Engineering, ASCE, 113(6), 1264–1281.
[32] Otani, S. and Cheung, V.W.T. (1981). Behaviour of Reinforced Concrete Columns under Biaxial Lateral
Load Reversals, (II) Test without Axial Load, Publication 81-02, Department of Civil Engineering,
University of Toronto.
[33] Otani, S.; Cheung, V. and Lai, S. (1980). Reinforced Concrete Columns Subjected to Biaxial Lateral
Load Reversals, Proceedings 7th World Conference on Earthquake Engineering, 7, Istanbul, Turkey, 525-
532.
[34] Otani, S.; Cheung, V. and Lai, S. (1980). Behavior and Analytical Models of Reinforced Concrete under
Biaxial Earthquake Loads, Proceedings 3rd Canadian Conference on Earthquake Engineering, 2,
Montreal, Canada.
172
[35] Priestley, M.J.N. and Park, R. (1987). Strength and Ductility of Concrete Bridge Columns under Seismic
Loading, ACI Structural Journal, 84(1), 61-76.
[36] Roufaiel, M.S.L. and Meyer, C. (1987). Analytical Modeling of Hysteretic Behavior of Reinforced
Concrete Frame, Journal of Structural Engineering, ASCE, 113(3), 429-444.
[37] Roy, R. (2009). Inelastic Seismic Behaviour of Multistoried Structures Highlighting the Effects of
Asymmetry, Soil-flexibility, Degradation and Blast Loading, PhD Thesis, Bengal Engineering and
Science University, Shibpur, India.
[38] Saiidi, M.S.;. Ghusn, G. and Jiang, Y. (1989). Five-Spring Element for Biaxially Bent R/C Columns,
Journal of Structural Engineering, ASCE, 115(2), 398-416.
[39] Sakai, K. and Sheikh, S.A. (1989). What Do We Know About Confinement in Reinforced Concrete
Columns? (A critical review of previous work and code provisions), ACI Structural Journal, 86(2), 192-
207.
[40] Saatcioglu, M. (1984). Reinforced Concrete Columns Subjected to Uniaxial and Biaxial Load Reversals,
Proceedings 8th World Conference on Earthquake Engineering, 6, San Francisco, California, 585-592.
[41] Taciroglu, E. and Hjelmstad, K.D. (2005). The Variational Basis of Nonlinear Flexibility Methods for
Structural Analysis of Frames, Journal of Engineering Mechanics, ASCE, 131(11), 1157-1169.
[42] Takeda, T.; Sozen, M.A. and Nielsen, N.N. (1970). Reinforced Concrete Response to Simulated
Earthquake, Journal of the Structural Division, ASCE, 96(12), 2557-2573.
[43] Takizawa, H. and Aoyama, H. (1976). Biaxial Effects in Modelling Earthquake Response of R/C
Structures, International Journal of Earthquake Engineering and Structural Dynamics, 4(6), 523-552.
[44] Takiguchi, K. and Kokusho, S. (1977). Hysteric Behaviour of Reinforced Concrete members Subjected to
Biaxial Bending Moments, Proceedings 6th World Conference on Earthquake Engineering, 3, New Delhi,
India, 3262.
[45] Watson, S. and Park, R. (1994). Simulated Seismic Load Tests on Reinforced Concrete Column, Journal
of Structural Engineering, ASCE, 120(6), 1825-1849.
[46] Wight, J.K. and Sozen, M.A. (1975). Strength Decay of R/C Columns under Shear Reversals, Journal of
the Structural Division, ASCE, 101(ST5), 1053-1065.
[47] Zeris, C.(1986). Three Dimensional Non-linear Response of Reinforced Concrete Building, Thesis
presented to the University of California, Berkeley, Calif., in partial fulfillment of the requirement for the
degree of Doctor of Philosophy.
173
Paper No.: 029
SEISMIC BEHAVIOUR OF PLAN-ASYMMETRIC STRUCTURES UNDER

SPECTRALLY MATCHED RECORDS
Pawan Thakur 1 and Rana Roy2
1
Graduate Student, Department of Applied Mechanics, Bengal Engineering and Science University, Shibpur,
Howrah 711 103, INDIA.
2
Associate Professor, Department of Applied Mechanics, Bengal Engineering and Science University, Shibpur,
Howrah 711 103, INDIA [Corresponding author]
Email: pawan2355@gmail.com, rroybec@yahoo.com.
ABSTRACT
In the context of performance-based seismic design (PBSD), ground motions are often scaled to certain
convenient target spectra derived from probabilistic seismic hazard analysis (PSHA). While Uniform Hazard
Spectrum (UHS) is more widely used, Conditional Mean Spectrum (CMS) is proposed as a more desirable target
for scaling of real accelerograms. In this backdrop, a set of far-fault (FF) ground motions are spectrally scaled,
using wavelets, to both UHS and CMS of the same region. Seismic demand of plan-asymmetric structures under
bi-directional ground motions is computed under both scaled and seed records in the elastic range. Spectral
matching of both the horizontal components of real records to a single hazard spectrum is observed to be
adequate to predict the amplification in response due to asymmetry. Further, such scaling expectedly reduces the
variability in predicted magnification from one ground motion to other.
KEYWORDS: Asymmetry, seismic, scaling, UHS, CMS
1. INTRODUCTION
Currently there is a major shift in the basic framework of performance-based seismic design (PBSD) from
deterministic (FEMA 356: 2000; ASCE/SEI 7-05: 2006) to probabilistic approach (ATC 2009). In probabilistic
seismic hazard analysis (PSHA), using ground motion attenuation relationship, Uniform Hazard Spectrum (UHS) or
Conditional Mean Spectrum (CMS) are derived (Baker and Cornell 2006; Baker 2011). Thereafter, scaling of ground
motions to such hazard spectra is performed to estimate structural response at an intended hazard level. Response
spectral matching over a period range of interest reduces the variance in the resulting structural response and hence
a smaller number of analyses are required to obtain the same confidence in median structural response quantities
(Hancock et al. 2008). However, since scaling of ground motion may affect its basic characteristics, adequacy of
spectrally matched records to estimate structural response is still contentious.
It may be reckoned that the real structures are often irregular in plan due to architectural and functional reasons. In
view of the intense seismic vulnerability of such horizontally irregular systems, numerous investigations (e.g., Kan
and Chopra 1977a; Kan and Chopra 1977b; Tso and Dempsey 1980; Tsicinias and Hutchinson 1981; Humar and
Kumar 1998) have been carried out using parametrically defined equivalent single storey model. These studies
generally employ modal analysis technique or response history analysis under real or artificial records. Outcomes of
such studies primarily constitute the basis of the equivalent static design procedure prescribed in current codes
(International Association of Earthquake Engineering, IAEE 2000).
Thus, in the backdrop of increasing application of spectrally matched records in seismic design, present study
attempts to evaluate the adequacy of ground motions scaled to selected hazard spectra (UHS and CMS). Response
under seed records is also computed to realize the implications of scaling of records on the response of asymmetric
system. Simplified single storey models with different representative uncoupled periods of vibration are considered
in the elastic range. Far-fault (FF) records, collected from sites with soil class D, are considered in the analysis.
2. REFERENCE SPECTRA AND GROUND MOTIONS
Uniform Hazard Spectrum (UHS) for 2% probability of exceedence (PE) collected from the website of United States
Geological Survey (USGS) is presented in Figure 1. This spectrum corresponds to Los Angelas (Latitude: 34.0890 N;
Longitude 118.4350 W) typically representative of site class D (stiff soils with shear wave velocity between 182
m/sec to 366 m/sec) (FEMA 356, 2000). Conditional Mean Spectrum (CMS) of the same region is also obtained
from USGS for different conditioning periods (ranging from short to long period). Characteristic conditioning
periods are selected as 0.2 sec., 1.0 sec. and 3.0 sec. for the CMS. It is evident from Figure 1 that the ordinates of
UHS are greater than CMS at all periods other than the period at which CMS is conditioned.
Ground motion characteristics such as frequency content, pattern of pulses, duration of shaking, fault-rupture
mechanism etc. are known to potentially influence the seismic response of structures. Hence, a number of ground
motion records are selected from PEER strong motion database. Ground motions are so chosen that the tectonic
environments of the records are similar to those of the reference spectra and, therefore, records from different
countries with comparable tectonic environments are also adopted. Selection of the records is also made based on the
similarity of their characteristics in terms of the magnitude, closest site-to-fault rupture distance and soil
classification. Table 1 includes 11 pairs of seed ground motions used in the present study. The moment magnitudes
for the ground motions range between 6.2 and 7.3 and closest site-to-fault-rupture distance varies between 17.6 km
and 43.2 km. These may thus be representative of far-fault (FF) records. Shear wave velocity and fault type are also
presented in the table. Average shear wave velocity in the top 30 meters (Vs30) represents a soil class D.
Ground motions are scaled separately to UHS and CMS considering a period range of 0.2T to 1.5T (as recommended
in ASCE/SEI 7-10) where T is the fundamental period of the structure selected as 0.2 sec., 1.0 sec. and 3.0 sec.
Evidently, in case of scaling with CMS as reference target, conditioning period T* is identical with T. Scaling of
ground motion adjusts a real accelerogram such that it tends to have a good match with a given spectrum. This
scaling, using wavelets, is performed using standard software Seismo-Match (Version 1.3.0). Further details of the
methodology are available elsewhere (Abrahamson 1992; Hancock et al. 2006).
Response of the asymmetric structure is computed under bi-directional seismic excitation (both seed and scaled
records). For this purpose, it is desirable that the components of ground motions should be scaled pair-wise (Grant
2011) selecting appropriate ‘major-axis’ and ‘minor-axis’ spectra derived from orientation-dependent ground
motion measure (Hong and Goda 2007). However, such ‘major’ and ‘minor’ axis spectra may often be identical
(Grant 2011) as assumed herein.
3. IDEALIZATION OF THE SYSTEM AND METHODOLOGY
Single storey rigid diaphragm model with three degrees of freedom, two translations in two mutually orthogonal
directions and one in-plane rotation, is used in the present study. Typical structural layout of such asymmetric plan
system along with the symmetric companion is shown in Figure 2. In the parametric study, mass of the system is
adjusted to achieve certain uncoupled lateral periods (T), while different torsional periods (Tθ) are set by changing
the distribution of mass as regulated by radius of gyration. Eccentricity of the asymmetric system is expressed in the
terms of err/D and erθ (Figure 2(ii)) following a previous study (Roy and Chakrovorty 2013).
175
Under specified ground acceleration histories, standard equations of motion (Clough and Penzion 1993; Chopra
2007) of the system so modeled are solved in the time domain using Newmark’s β-γ scheme which considers
constant average acceleration over each incremental time step. While Newmark’s parameters γ and β are chosen
respectively as 0.5 and 0.25, iterations are performed in each incremental time step using modified Newton-
Raphson technique. The time step of integration is taken as less than T/1000 second to ensure convergence. 2% of
critical damping in each mode of vibration is considered to constitute damping matrix. Further details of the
idealization and computational scheme are available elsewhere (Roy and Chakrovorty 2013).
During bi-directional seismic shaking in so-called bi-directionally eccentric structures, eccentricities along two
principal directions result in two torsional moments. The effect of torsion seems to be amplified if the moments
generated due to eccentricities in each direction are additive in nature, while the mutually cancelling nature of such
moments tend to lower the impact of torsion. Such addition or cancellation of two torsional moments depends on
the relative sense of eccentricities. Further, interaction between such pair of torsional moments (additive/cancelling)
on a bi-directionally eccentric system may be additive or cancelling depending upon the ground motion
characteristics (in phase or out of phase). Thus maximum displacement response considering both senses of
eccentricities are computed for load-resisting elements oriented in both the principal directions and compared for
flexible and stiff sides (The lateral load-resisting edge element with lesser stiffness is designated as the flexible
element and the opposite edge element with greater stiffness is referred to as the stiff element) separately.
Systems with varying period (T), viz., 0.2 sec., 1.0 sec and 3.0 sec., representative of short, medium and long
periods, are analyzed under seed records. Responses under such ground motions scaled to UHS and CMS are also
computed. Maximum in-plane deformation of the edge elements due to the coupled effect of translation and torsion
in an asymmetric system and those of the reference symmetric model are separately computed under each
earthquake history. Such deformation for load-resisting elements oriented in both the principal directions is
compared and the greatest ones are regarded as maximum element displacement. In asymmetric system, responses
of flexible and stiff sides in both the principal directions are separately compared and the higher response is
considered as maximum element displacement for the respective side. Maximum element displacement of
asymmetric system normalized by that of the symmetric system is referred to as maximum normalized element
displacement and is then computed under each record to identify the impact of asymmetry exclusively. In view of
the strong dependence of seismic torsional behaviour on fundamental torsional-to-lateral period ratio (τ) of the
corresponding uncoupled system, response quantities are estimated for different τ (viz., 0.5, 1.0 and 1.5)
representative of torsionally stiff to flexible systems. Parameters quantifying asymmetry, viz., err/D and erθ are
considered to vary in the range of 0.02 to 0.22 (with interval of 0.04) and 00 to 900 (with interval of 150)
respectively.
Figure 3a and Figure 3b describe the variation of maximum normalized element displacement respectively for both
flexible and stiff sides as a function of err/D and erθ separately. Figures show that the amplification in response
increases with increase of err/D, while such variation is relatively stable with erθ. This observation is somewhat
intuitive. Increase of err/D directly increases earthquake induced torque and hence the related displacement. On the
other hand, change of erθ is associated with the increase of the eccentricity along one principal axis and a
proportionate decrease along the other. Thus the net torque generated due to lateral excitation tends to be nearly
invariant and hence the amplification. On further scrutiny, magnification in response seems to increase in long
period systems and is relatively subdued in the short period range. This trend follows consistently in both
torsionally flexible and stiff systems.
Figure 3a and Figure 3b also include the normalized response under seed records alongside the same under records
scaled to both UHS and CMS. In general, figures suggest that the mean amplification in response may be estimated
176
using spectrally matched records with reasonable accuracy. It may be appreciated that this maximum element
displacement corresponds to the response at a particular instant of the ground motion history comprising pulses of
widely varying frequencies. Thus sporadic discrepancies in response may not be unlikely even after spectral
matching.
In this context, to examine the advantage of ground motion scaling, dispersion in the amplification predicted by
different records are computed. To this end, coefficients of variation (COV) of the predicted maximum normalized
element displacement are superimposed in Figure 3a and Figure 3b respectively for flexible and stiff sides. These
demonstrate that COV reduces consistently for spectrally matched records. COV values under seed and scaled
records are further compiled and presented in the form of a bar chart (Figure 4). This clearly subscribes to the
reduction in the possible variability in the predicted seismic response. Moreover, there also exists some latitude to
infer that both UHS and CMS exhibit a comparable performance to reduce scatter in response.
It may be recalled that the response of asymmetric system is sensitive to the bi-directional components of a ground
motion. This has led to the proposition of a pair-wise scaling of the components of an earthquake record to estimate
the response of the asymmetric system in particular (Hong and Goda 2007; Grant 2011). Present study, in contrast,
scales ground motion components independently to the same spectrum ‘as is more common in practice’ (Grant
2011). However, magnification in displacement demand seems to closely match with the normalized demand due to
unscaled records. Thus, the relevance of pair-wise scaling, albeit physically attractive, deserves to be further
examined in the context of evaluation of torsion-induced demands in plan-asymmetric system.
Response of asymmetric structures (under seed and scaled records) normalized by that of reference symmetric
systems is presented to identify the impact of asymmetry alone. Variation of maximum normalized element
displacement for both flexible and stiff sides is studied. Observations of the present study lead to the following
broad conclusions.
1. Use of spectrally matched ground motions may lead to reasonable prediction of the amplification of seismic
demand in plan-asymmetric system. Such magnification of seismic displacement demand due to torsion is
marginally affected by the choice of reference target spectrum (viz., UHS or CMS) for scaling of motion.
2. Coefficient of variation (COV), indicating the dispersion of the results with respect to mean, considerably
reduces due to the use of such spectrally matched records.
Combining the preceding two conclusions, it seems that the choice of UHS or CMS as reference target spectra to
scale the records may not introduce significant difference in the prediction of torsion-induced additional seismic
demand of plan-asymmetric systems in the elastic range. However, scaling process reduces the scatter in the
results and, therefore, the required number of dynamic analysis can be significantly reduced.
3. Pair-wise scaling of ground motions employing rigorous ‘major-axis’ and ‘minor-axis’ spectra may not be
essential for estimating the torsional vulnerability of plan-asymmetric system. This issue deserves in-depth
investigation.
Thus, the study confirms the utility of scaling of ground motions to evaluate the response of plan-asymmetric system.
Inelastic displacement ratio defined as the maximum lateral inelastic displacement relative to maximum elastic
counterpart has been shown to be nearly independent of soil sites with average shear wave velocities higher than 180
m/s (Miranda 1999). Investigation (Chopra and Chintanapakdee 2004) with SDOF system underground motions
collected from different site class, such as B, C and D, also lead to the similar conclusions. These together may be
regarded as an indicative of the relative insensitivity of the soil class to the torsion-induced displacement parameter
177
studied herein. However, further investigation needs be conducted to be assertive on the applicability of the current
observations in other soil classes (other than D). Such investigation is currently underway by the authors.
6. TABLES
Table 1: Far-fault (FF) ground motions
Sl. Components
Vs30 Fault r
No Event Station M
Component identifier (m/s) type (km)
.
Cape NGA_no_826_EUR090
Eureka-Myrtle
1 Mendocino 339 RV 7.0 42.0
and West NGA_no_826_EUR000
04/25/92
Cape NGA_no_827_FOR090
Fortuna—
2 Mendocino 457 RV 7.0 20.0
Fortuna Blvd NGA_no_827_FOR000
04/25/92
Landers Desert Hot NGA_no_850_DSP090
3 345 SS 7.3 21.8
06/28/92 Springs NGA_no_850_DSP000
Landers Palm Springs NGA_no_884_PSA090
4 207 SS 7.3 36.2
06/28/92 Airport NGA_no_884_PSA000
Landers Yermo Fire NGA_no_900_YER360
5 354 SS 7.3 23.6
06/28/92 Station NGA_no_900_YER270
Loma Prieta APEEL 2— NGA_no_732_A02133 RV-
6 133 6.9 43.2
10/18/89 Redwood City NGA_no_732_A02043 OBL
Northridge-01 NGA_no_1019_L01090
7 Lake Hughes #1 425 RV 6.7 35.8
01/17/94 NGA_no_1019_L01000
San Fernando NGA_no_70_L01111
8 Lake Hughes #1 425 RV 6.6 27.4
02/09/71 NGA_no_70_L01021
San Fernando Palmdale Fire NGA_no_78_PDL210
9 453 RV 6.6 29.0
02/09/71 Station NGA_no_78_PDL120
San Fernando Whittier NGA_no_93_WND233
10 299 RV 6.6 39.5
02/09/71 Narrows Dam NGA_no_93_WND143
NGA_no_718_A-
Superstition
Wildlife Liquef. IVW360
11 Hills-01 207 SS 6.2 17.6
Array NGA_no_718_A-
11/24/87
IVW090
M - Moment magnitude, r - closest site-to-fault-rupture distance, Vs30 – Shear wave
velocity, SS – Strike slip, RV – Reverse, RV-OBL – Reverse oblique, N – Normal, N-
OBL – Normal oblique
178
7. DIAGRAMS
2.0
1.5
1.0
0.5
0.0
0.0 1.0 2.0 3.0 4.0 5.0
UHS CMS (T = 0.2Sec.) CMS (T = 1.0Sec.) CMS (T = 3.0Sec.)
Figure 1: Uniform hazard spectrum (UHS) and Conditional mean spectrum (CMS) adopted. (site class D)
(a) Reference symmetric system (b) Typical asymmetric model
Figure 2: Unified representation of equivalent single story system (Roy and Chakrovorty 2013).
Seed Scaled w.r.t. UHS Scaled w.r.t. CMS

Maximum normalized
1.5 1.5 1.5

element displacement
Flexible side Flexible side Flexible side
1.0 1.0 1.0
0.5 0.5 0.5
0.0 0.0 0.0

0 15 30 45 60 75 90 0 15 30 45 60 75 90 0 15 30 45 60 75 90
erθ (degree) erθ (degree) erθ (degree)
Maximum normalized
1.5 1.5
1.5
Stiff side
Stiff side Stiff side
1.0 1.0 1.0
0.5 0.5 0.5
0.0 0.0 0.0

0 15 30 45 60 75 90 0 15 30 45 60 75 90 0 15 30 45 60 75 90
erθ (degree) erθ (degree)
erθ (degree)
3a( i) err/D = 0.02, τ = 1.0
179
2.0 2.0 2.0
Maximum normalized

1.5 1.5 1.5
1.0 1.0 1.0
0.5 0.5 0.5
0.0 0.0 0.0

0 15 30 45 60 75 90 0 15 30 45 60 75 90 0 15 30 45 60 75 90
2.0 2.0 2.0
Maximum normalized
Stiff side Stiff side Stiff side

1.5 1.5 1.5
1.0 1.0 1.0
0.5 0.5 0.5
0.0 0.0 0.0

0 15 30 45 60 75 90 0 15 30 45 60 75 90 0 15 30 45 60 75 90
erθ (degree) erθ (degree)
erθ (degree)
3a(ii) err/D = 0.1, τ = 0.5
2.0 2.0 2.0

T=3.0 Sec.
Maximum normalized
1.5 1.5 1.5

1.0 1.0 1.0
0.5 0.5 0.5
T=1.0 Sec.
0.0 0.0 0.0
0 15 30 45 60 75 90 0 15 30 45 60 75 90 0 15 30 45 60 75 90
2.0 2.0 2.0
Maximum normalized
T=0.2 Sec.
1.5 1.5 1.5
1.0 1.0 1.0
0.5 0.5 0.5 Mean
0.0 0.0 0.0 COV

0 15 30 45 60 75 90 0 15 30 45 60 75 90 0 15 30 45 60 75 90
3a(iii) err/D = 0.22, τ = 1.5
Figure 3a: Variation of maximum normalized displacement response with change of erθ for FF records.
180
Seed Scaled w.r.t. UHS Scaled w.r.t. CMS
2.0 2.0 2.0
Maximum normalized

1.5 1.5 1.5
1.0 1.0 1.0
0.5 0.5 0.5
0.0 0.0 0.0

0 0.08 0.16 0.24 0 0.08 0.16 0.24 0 0.08 0.16 0.24
err/D err/D err/D
2.0 2.0 2.0

Maximum normalized

1.5 1.5 1.5
1.0 1.0 1.0
0.5 0.5 0.5
0.0 0.0 0.0

0 0.08 0.16 0.24 0 0.08 0.16 0.24 0 0.08 0.16 0.24
err/D err/D err/D
3b(i) erθ = 0o, τ = 1.0
2.0 2.0 2.0
Maximum normalized

1.5 1.5 1.5
1.0 1.0 1.0
0.5 0.5 0.5
0.0 0.0 0.0

0 0.08 0.16 0.24 0 0.08 0.16 0.24 0 0.08 0.16 0.24
err/D err/D err/D
Maximum normalized
2.0 2.0 2.0

1.5 1.5 1.5
1.0 1.0 1.0
0.5 0.5 0.5
0.0 0.0 0.0

0 0.08 0.16 0.24 0 0.08 0.16 0.24 0 0.08 0.16 0.24
err/D err/D err/D
3b(ii) erθ = 30o, τ = 0.5
181
2.0 2.0 2.0
Maximum normalized

1.5 1.5 1.5
T=3.0 Sec.
1.0 1.0 1.0
0.5 0.5 0.5
T=1.0 Sec.
0.0 0.0 0.0
0 0.08 0.16 0.24 0 0.08 0.16 0.24 0 0.08 0.16 0.24
err/D err/D err/D
2.0 2.0 2.0
Maximum normalized

1.5 1.5 1.5
T=0.2 Sec.
1.0 1.0 1.0
0.5 0.5 0.5
Mean
0.0 0.0 0.0
COV
0 0.08 0.16 0.24 0 0.08 0.16 0.24 0 0.08 0.16 0.24
err/D err/D err/D
3b ( iii) erθ = 75o, τ = 1.5
Figure 3b: Variation of maximum normalized displacement response with change of err/D for FF records.
300
SEED UHS CMS
250
200
Data points
150
100
50
0
0.00--0.05 0.05--0.10 0.10--0.15 0.15--0.20 0.20--0.25 0.25--0.30 0.30--0.35 0.35--0.40
COV range
Figure 4: Coefficient of variation of maximum normalized element displacement response of

plan asymmetric system.
REFERENCES
[1] Abrahamson, N.A. (1992). Non-stationary spectral matching, Seismol. Res. Lett., 63(1), 30.
[2] American Society of Civil Engineers (ASCE) (2006). Minimum design loads for buildings and other
structures, ASCE/SEI 7-05, Reston, Virginia.
[3] Applied technology Council (ATC) (2009). Guidelines for seismic performance assessment of buildings, ATC-
58 50% Draft, ATC, Redwood City, CA.
[4] Baker J.W. and Cornell C. A. (2006). Spectral shape, epsilon and record selection, Earthquake Engineering
Structural Dynamics, 35(9), 1077-1095.
182
[5] Baker, J.W. (2011). Conditional mean spectrum: Tool for ground motion selection, Journal of Structural
Engineering, ASCE, 137(3), 322-331.
[6] Chopra, A. K. (2007). Dynamics of structures: Theory and applications to earthquake engineering, Pearson
Education, Inc.
[7] Chopra, A.K. and Chintanpakdee, C. (2004). Inelastic deformation ratios for design and evaluation of
structures: single-degree-of-freedom bilinear systems, Journal of Structural Division, ASCE, 130(9), 1309-
1319.
[8] Clough, R.W. and Penzien, J. (1993). Dynamics of structures, McGraw-Hill Inc., International Edition.
[9] FEMA 356 (2000). Prestandard and commentary for the seismic rehabilitation of buildings, Report FEMA
356, Federal Emergency Management Agency, Washington DC, U.S.A.
[10] Grant, D.N. (2011). Response spectral matching of two horizontal ground-motion components, Journal of
Structural Engineering, 127(3), 289-297.
[11] Hancock, J.; Bommer, J.J., Stafford, P.J. (2008). Numbers of scaled and matched accelerograms required for
inelastic dynamic analyses, Earthquake Engineering Structural Dynamics, 37(14), 1585-1607.
[12] Hancock, J.; Watson-Lamprey, J.; Abrahamson, N.A.; Bommer, J.J.; Markatis, A.; McCoy, E.and Mendis, R.
(2006). An improved method of matching response spectra of recorded earthquake ground motion using
wavelets, Journal of Earthquake Engineering, 10(S1), 67–89.
[13] Haselton, C. B.; Baker, J. W.; Liel, A. B.and Deierlein, G. G. (2011). Accounting for ground-motion spectral
shape characteristics in structural collapse assessment through an adjustment for epsilon, Journal of Structural
Engineering, 137(3), 332-344.
[14] Hejal, R. and Chopra, A. K. (1989). Earthquake analysis of a class of torsionally-coupled buildings.
Earthquake Engineering and Structural Dynamics, 18(3), 305–323
[15] Hong, H.P. and Goda, K. (2007). Orientation-dependent ground-motion measure for seismic-hazard
assessment, Bull. Seismol. Soc. Am., 95(5), 1525-1538.
[16] Humar, J.L. and Kumar, P. (1998). Torsional motion of buildings during earthquakes. I. Elastic response,
Canadian Journal of Civil Engineering, 25(5), 898-916.
[17] International Association of Earthquake Engineering (IAEE) (2000), Regulation for Seismic Design-A World
List, Tokyo.
[18] Kan, C.L. and Chopra, A.K. (1977a). Effects of torsional coupling on earthquake forces in buildings, Journal
of Structural Divisions, ASCE, 103(ST4), 805-819.
[19] Kan, C.L. and Chopra, A.K. (1977b). Elastic earthquake analysis of a class of torsionally coupled buildings,
Journal of Structural Divisions, ASCE, 103(ST4), 821-838.
[20] Miranda, E. (2000). Inelastic displacement ratios for structures on firm sites, Journal of Structural Division,
ASCE, 126(10), 1150-1159.
183
[21] Pacific Earthquake Engineering research Centre (PEER) strong motion database - URL:
https://peer.berkeley.edu.
[22] Reiter, L. (1990). Earthquake Hazard Analysis: Issues and Insights, Columbia University Press, New York.
[23] Roy, R. and Chakrovorty, S. (2013). Seismic demand of plan-asymmetric structures: A re-visit, Journal of
Earthquake Engineering and Engineering Vibration, 12(1), 99-117.
[24] SeismoSoft. SeismoMatch 1.3.0 (2005) — A computer program for an application capable of adjusting
earthquake records, through wavelet addition. Online available from URL /http://www.seismosoft.com.
[25] Thenhaus, P.C. and Campbell, K. W. (2003). Seismic Hazard Analysis”, Earthquake Engineering Handbook,
W.-F. Chen and C. Schawthorn, eds., CRC Press, VIII1-50, 2003.
[26] Tsicinias, T.G. and Hutchinson, G.L. (1981). Evaluation of code requirements for the earthquake –resistant
design of torsionally coupled buildings, Proc. of Institution of Civil Engineers, 71(Part 2), 821-843.
[27] Tso, W.K. and Dempsey, K.M. (1980). Seismic torsional provisions for dynamic eccentricity, Earthquake
Engineering and Structural Dynamics, 8(3), 275-289.
[28] United States Geological Survey (USGS) Geologic Hazard Science Centre, ground motion calculator, Online
Available from URL: http://earthquake.usgs.gov/research/hazmaps/
184
Paper No.: 031
INFLUENCE OF MASONRY PROPERTIES ON

LATERAL LOAD RESPONSE OF REINFORCED CONCRETE FRAMES
Syed H. Basha1 and Hemant B. Kaushik2
1
Research Scholar, Dept. of Civil Engineering, Indian Institute of Technology Guwahati, India
2
Assistant Professor, Dept. of Civil Engineering, Indian Institute of Technology Guwahati, India
Email: s.basha@iitg.ac.in, hemantbk@iitg.ac.in
ABSTRACT:
Burnt clay brick is one of the most commonly used materials in construction of load bearing structures and as
infill walls in reinforced concrete frames. Recently, fly ash bricks are also used in building construction, both in
load bearing constructions and in reinforced concrete frames as infill walls. Though the influence of burnt clay
brick masonry on lateral load response of reinforced concrete frames has been well understood, influence of
material properties of fly ash bricks on the lateral load response of reinforced concrete frames has not been
studied in detail. In the current study, physical properties of the fly ash bricks, such as, water absorption, IRA
(initial rate of absorption), and compressive strength were determined in accordance with relevant standards.
Further, nonlinear stress strain characteristics of fly ash brick masonry were also determined by a detailed
experimental study on masonry prisms. Three mortar grades (cement:sand - 1:0:3, 1:0:4, and 1:0:6),
corresponding to strong, intermediate, and weak mortar used on construction sites, were used in constructing
masonry prims. Masonry prisms were tested under uniaxial monotonic compressive loading using a servo
controlled hydraulic actuator. These material properties of fly ash bricks were used in nonlinear analysis of
masonry infilled reinforced concrete frames, and the response of the frames were compared with those obtained
using the burnt clay bricks as masonry infills. Material properties of burnt clay brick masonry was obtained
from the previous studies. The study will be useful in developing guidelines for designing the reinforced
concrete frames taking into consideration the effects of masonry infill walls.
KEYWORDS: fly ash brick, burnt clay brick, non-linear analysis, lateral load response.
1. INTRODUCTION
The stress strain characteristics of masonry play a vital role in evaluating the capacity of infilled reinforced
concrete (RC) frames. From Indian context a detailed study on stress strain characteristics of burnt clay brick
masonry was carried out by the Kaushik et al. (2007), Sarangapani et al. (2005) etc., which provided a strong
understanding for non-linear behavior of masonry. Few studies Malaviya et al. (1999), Christy et al. (2013),
were conducted to evaluate the strength characteristics of fly ash bricks. But the non-linear stress strain
relationships of fly ash brick masonry has not determined in accordance with the relevant standards. In the
current study, an attempt has been made to evaluate the stress strain characteristics of fly ash brick masonry and
compared with the previous studies. Later, the stress strain properties of fly ash bricks were employed to
evaluate the nonlinear response of RC frames by modelling the infills as equivalent diagonal struts. Pushover
curve determined from the analytical study was compared with the experimental envelope curve of infilled
ductile frame determined in the previous study (Basha 2011). Finally a parametric study on the lateral load and
lateral stiffness on infilled RC frame was carried out with different infills.
2. PHYSICAL PROPERTIES OF FLY ASH BRICK MASONRY AND CONSTITUENTS
The physical properties of fly ash brick was carried out by evaluating the water absorption, initial rate of
absorption, compressive strength and finally stress strain characteristics of fly ash brick masonry prisms. Total
water absorption was determined in accordance with IS 3495 (IS1992b) and ASTM C67-12 (ASTM 2012a)
standards. Water absorption by percent of mass is given as the mass of water absorbed by dry brick to the mass
of the oven dry brick unit. It was observed that the water absorption of fly ash bricks ranges from 17.5 to 19 %
with an average of 18.3%. Kaushik et al. (2007) from their study on burnt clay bricks available in northern India
reported that total water absorption ranges from 11 to 13%. Initial rate of absorption of fly ash bricks was
determined by immersing the brick material in 3 mm deep water for 1 minute. The test was conducted in
accordance with ASTM C67-12 (ASTM 2012a). The average initial rate of absorption was about 5.1 kg/m2/min
whereas it varied from 0.97 to 2.42 kg/m2/min for different types burnt clay brick samples as reported by
Kaushik et al. (2007). Compressive strength of fly ash bricks is determined in accordance with ASTM C67-12
(ASTM 2012a) and IS 3495 (IS1992a). Compressive strength of bricks found to vary from 4.3 to 6.9 MPa with
an average strength of 5.7 MPa. The mean compressive strength of burnt clay bricks varied from 16.1 MPa to
28.9 MPa as reported by Kaushik et al. (2007). From the results, it can be inferred that the compressive strength
of fly ash bricks was very low when compared to the compressive strength of burnt clay bricks.
The stress strain characteristics of fly ash brick masonry was determined by testing five brick prisms under
monotonic compressive loading using servo controlled hydraulic actuator. The test was conducted in accordance
with ASTM C1314-12 (ASTM 2012b) and IS 1905 (IS 1987). Masonry prisms were constructed using three
different grades of mortar (cement:lime:sand by volume), i.e., 1:0:3 (strong), 1:0:4 (intermediate) and 1:0:6
(weak). As the size of the fly ash brick is 230×110×75 mm, and the thickness of mortar joint varies from 10-12
mm, the average height of the masonry prism was 420 mm. The specimens were instrumented with contact
epsilon extensometer and non-contact laser extensometer to measure the strain during compressive loading. The
rate of loading was in accordance with ASTM C1314-12 (ASTM 2012b) and IS 1905 (IS 1987) standards. The
stress strain characteristics of masonry prisms with different grades of mortar is shown in Figure 1 along with
the stress strain characteristics of burnt clay brick prisms from previous study (Kaushik et al. 2007). Clay brick
prisms were constructed in the previous study by Kaushik et al. (2007) using three different grades of mortar
(cement:lime:sand by volume), i.e., 1:0:3 (strong), 1:0.5:4.5 (intermediate), and 1:0:6 (weak). The average
maximum compressive strength of fly ash brick prisms found to be 4.6 MPa, 3.9 MPa and 3.1 MPa, across three
different grades of mortar, whereas it was 7.5 MPa, 6.6 MPa, and 4.1 MPa, for strong, intermediate and weak
burnt clay brick prims, respectively. From the curves, it can be observed that the load carrying capacity of burnt
clay bricks was very high when compared to the fly ash brick prisms. But the dropdown in strength of fly ash
brick prisms is very gradual in comparison with burnt clay prisms. The soft behavior observed by the fly ash
bricks might be the reason for the fly ash brick prisms to observe gentle degradation in strength.
8
Fly ash Brick 1:0:3
Compressive Stress (MPa)
7 Fly ash Brick 1:0:4

Fly ash Brick 1:0:6
6
Clay Brick 1:0:3
5 Clay Brick 1:0.5:4.5
Clay Brick 1:0:6
4
3
2
1
0
0 0.002 0.004 0.006 0.008 0.01 0.012
Strain (mm/mm)
Figure 1. Comparison of nonlinear stress- strain curves of masonry prisms.
186
3. IMPORTANCE OF MASONRY INFILLED RC FRAMES
The presence of masonry in RC frames not only imparts higher stiffness but also increases the lateral load
carrying capacity when subjected to earthquake loading. To evaluate the lateral load carrying capacity of
masonry infilled RC frames, an experimental study was carried out previously (Basha 2011), on half scale
models using fly ash brick masonry. The reinforcement details of the RC frame considered in the study is shown
in Figure 2. Two types of frames comprising of ductile bare frame and ductile frame infilled with full scale fly
ash bricks designed in accordance with relevant Indian standards were tested under displacement controlled
slow cyclic tests. The lateral load carrying capacity and lateral stiffness of infilled frame was about 2 times and
8 times that of the ductile bare frame, respectively. To evaluate the effect of infills on RC frames analytically,
non-linear static analysis was carried out by modelling the infills as equivalent diagonal struts. In the current
study, the main objective was to evaluate the global response of infilled RC frame in terms of lateral strength
and stiffness which can be sufficiently estimated by modelling the infills as equivalent diagonal struts.
Figure 2. Reinforcement details of Ductile RC frame.
4. NONLINEAR STATIC PUSHOVER ANALYSIS
Nonlinear static pushover analysis was carried to evaluate the capacity of the structure by incorporating the
nonlinearity and pushing the structure until the target displacement or collapse mechanism is developed (ATC
1996). Pushover analysis can be performed individually as force controlled or displacement controlled. In force
controlled, forces are increased monotonically until either the force reaches a target value or the structure attains
collapse, whereas in case of displacement controlled the displacements are increased monotonically until either
the displacement reaches a predefined control point or the structure has attained a collapse mechanism. As the
system displaces, the structure undergo nonlinear inelastic behavior in flexure, shear or axial compression. The
nonlinear behavior is modeled by assigning appropriate load deformation properties at predefined potential
plastic hinge locations. The nonlinearity in frame elements was modeled by assigning moment, and shear
hinges, whereas axial hinges were assigned to model the nonlinearity in masonry elements. In the current study,
nonlinear analysis was carried out on bare frame with the experimentally determined material properties and
compared with the load-deformation curve obtained from the slow cyclic test of ductile bare frame. Later, the
187
behavior of ductile frame infilled with full scale bricks were compared with the analytical model with infills
modeled as equivalent diagonal struts.
Finally, a parametric study was undertaken to evaluate the lateral strength and stiffness of the RC frame infilled
with different types of infills. In the current study, a total of six different types of infills were considered with
three different grades of mortars. The infills include fly ash bricks with three mortar grades (cement: lime:
sand), i.e., strong (1:0:3), intermediate (1:0:4) and weak (1:0:6), and brunt clay bricks with three different grades
of mortar, i.e., strong (1:0:3), intermediate (1:0.5:4.5) and weak (1:0:6). The stress strain characteristics of the
mentioned infills were discussed and presented in the previous section (Figure 1). The reason behind using
different type infills with different compressive strengths is to assess the capacity of the infilled frames and will
be useful in proposing design guidelines regarding the selection of the infills. Six different infills represent two
different brick types of which clay brick possessed the highest compressive strength compared to fly ash brick
and three different grades of mortar which are normally used on construction sites.
4.1. Modeling of RC Members
In the current study, RC members i.e., columns and beams are modeled as 2-noded frame elements with
nonlinearity lumped at specified plastic hinge locations. The dimensions of beams and columns are 115 mm ×
175 mm, respectively, and are shown in Figure 2. Shell element was used to model T-beam effect in the
analytical study. Section designer module of SAP2000 (CSI 2013) was employed to evaluate flexure hinge
properties, i.e., P-M interaction and moment-curvature properties for column and beam and is shown in Figure
3. Shear hinges were modeled as brittle and hinge losses load carrying capacity as the maximum force was
reached. The derived plastic properties were lumped at specified hinge locations (Figure 4). The plastic hinge
length and plastic hinge location was determined based on the crack pattern observed during the experimental
study. It was observed that hinges in columns were formed at a distance 90 mm from the top face of the bottom
beam and bottom face of the top beam, respectively, in case of bare frame. The plastic hinge length was about
0.5 times the depth of the column.
Figure 3. Plastic hinge properties assigned to RC members (Kaushik et al. 2009).

1500 mm
lp/2
lp/2
ws
1500 mm
188
Figure 4. Details of single strut model and plastic hinge locations.
4.2. Modeling of Masonry Infills
Accurate mathematical modeling of infills has been a distant task till date due to a number of interacting
parameters involved and many possible failure modes needed to be evaluated with high degree of uncertainty
(Asteris et al. 2011). Various modeling techniques ranging from macro modeling to micro modeling were
proposed i.e., single strut (Holmes 1961, Stafford smith 1967), multi strut (El-Dakhakhni et al. 2003;
Chrysostomou et al. 2002) to finite element modeling (Ghosh and Amde 2002). In order to evaluate the overall
global response of the structure in terms of lateral strength and stiffness, masonry may be idealized as equivalent
diagonal compressive strut. In the current study, masonry is modeled as equivalent diagonal strut. The width of
diagonal strut can be evaluated using a number of expressions given by different researchers (Holmes 1961,
Stafford smith 1967, Paulay and Priestley 1992), but it was found from the experimental study that the width
was ranging from one sixth to one third of the diagonal length of the masonry infill. The nonlinearity in
masonry was modeled using stress strain curves developed from prism tests and is shown in Figure 1. It was
observed that the plasticity was concentrated in half to three-fourth of diagonal length of strut. Therefore, the
length of the plastic hinge was assumed to be equal to three-fourth of the diagonal length of the strut. The struts
were modeled as compression elements and weak in tension and only axial hinges were assumed to develop in
masonry. The results of nonlinear analysis and their comparison with the experimental study are discussed in the
following sections separately.
5. RESULTS OF BARE FRAME
Pushover curves are plotted as base shear against the monitored roof displacement and the result of pushover for
bare frame is shown in Figure 5. From the curves, it can be observed that the infill stiffness and lateral load
carrying capacity of both the analytical and the experimental results of bare frame were matching. The
maximum lateral load recorded during the experiment was 44 kN and is found in good relation with analytically
determined capacity of ductile bare frame. The failure of ductile bare frame was largely flexural as the moment
carrying capacity of columns was reached the collapse state. Initially, the plastic hinges were formed in the
bottom portions of left and right columns, later as the monitored displacement increased flexural hinges formed
near the top beam-column junctions. Therefore, it can be ascertained that the capacity and initial stiffness of the
bare frame can be evaluated analytically by modeling the frame elements as 2-noded elements and modeling the
nonlinear deformational properties by lumped plasticity approach.
50
45
40
Base Shear (kN)
35
30
25 Experiment
20 Analytical
15
10
5
0
0 10 20 30 40 50 60 70 80
Displacement (mm)
189
Figure 5. Comparison of pushover curves for ductile bare frame.
6. RESULTS OF INFILLED FRAME
As the capacity of bare frame was ideally matched with analytically derived pushover curve, the same model
was used to evaluate the capacity of infilled frame. In case of masonry infilled ductile frame, infill was modeled
as diagonal strut possessing a compressive strength of 1.5 MPa and modulus of elasticity of 825 MPa. The stress
strain characteristic of masonry was modeled following the guidelines proposed by Kaushik et al. (2007), and is
shown in Figure 6. The result of pushover analysis along with the envelop curve of experimentally determined
ductile frame infilled with full scale fly ash bricks is shown in Figure 7. From the plot, it can be observed that
the initial stiffness and capacity of both the curves is matching. Initially the lateral force was resisted by the
masonry infills due to their large initial stiffness which attracts large forces. As the displacement steps
increased, failure in infill was observed; later on the behavior of the infill frame followed the profile of the bare
frame at larger displacement levels. The main problem with modeling of infills as equivalent diagonal struts was
its inability to capture the local failure of the frame elements i.e., shear failure. In the experimental study, it was
observed that due to the presence of infills, columns were failing in shear which is an unwanted situation,
whereas failure mechanism was flexural from the analytical study. Therefore, this phenomenon has to be
addressed in detail by modeling the infills, so that the actual collapse mechanism prevails in the frame system.
1.6
1.4
Compressive Stress (MPa)
1.2
0.8
0.6
0.4
0.2
0
0 0.0005 0.001 0.0015 0.002 0.0025 0.003 0.0035 0.004
Strain
Figure 6. Compressive stress strain curve for infill (1.5 MPa).
80
64
Base Shear (kN)
48
32
Experiment
Analytical
16
0
0 20 40 60 80
Displacement (mm)
190
Figure 7. Comparison of pushover curves of masonry infilled RC frame.
7. RESULTS OF DUCTILE FRAME INFILLED WITH DIFFERENT INFILLS
The result of pushover curve with different infills is shown in Figure 8. The lateral strength and stiffness of infill
frame increases as the strength of the brick as well the strength of the mortar grade increases. The same has been
reported by different researchers in the past studies (CEB 1996). It was reported that use of better quality mortar
leads to better bond characteristics between mortar and brick units and therefore higher cracking load. Once the
capacity of the infill is reached, the frame behaves as that of the bare frame and the same can be observed in the
Figure 8. The increase in strength for frames infilled with fly ash bricks is about 1.3 and 1.2 times for strong
(1:0:3) and intermediate (1:0:4) mortar grade infill in comparison with weak (1:0:6) mortar grade infill. It is
about 1.5 and 1.3 times for strong and intermediate mortar prisms when compared to weak mortar. The increase
in strength for frames infilled with clay bricks is about 1.5 and 1.4 times for in strong (1:0:3) and intermediate
(1:0.5:4.5) mortar grade infill in comparison with weak (1:0:6) mortar grade infill. Therefore, the increase in
strength for fly ash infilled frames and fly ash brick prisms is not very high when compared to the increase in
strength in case of frames infilled with clay bricks and clay brick prisms. The more pronouncing behavior of the
frame infilled with strong masonry is the transfer of failure mechanism from infill to column elements. This can
be attributed to frame failure mode which is seen in the form of plastic hinges in the columns or the beam-
column connections (Asteris et al. 2011). Therefore, care has to be taken regarding the technique to be
employed in modeling the infills. It can be concluded that infill can be modeled as equivalent diagonal strut if
global response of the system is of primary importance whereas, three strut model or finite element modeling
can be an alternative if global as well failure mechanisms needed to be evaluated (Kaushik et al. 2009).
250
200 Flyash brick 1:0:3

Flyash brick 1:0:4
Base Shear (kN)
Flyash brick 1:0:6

150 Clay brick 1:0:3
Clay brick 1:0.5:4.5
clay brick 1:0:6
100 Experimental
Flyash brick f'm =1.5
50
0
0 20 40 60 80
Displacement (mm)
Figure 8. Comparison of pushover curves of ductile frame infilled with different masonry.
8. CONCLUSIONS
The fly ash bricks used in the current study were soft compared to burnt clay bricks from the previous studies.
From the analytical study, it can be concluded that the global response i.e., lateral strength and stiffness of the
infilled frames can be evaluated using equivalent diagonal struts. In the current study, width of the diagonal strut
was about one sixth of the diagonal length of fly ash brick masonry wall. The lateral strength and stiffness of the
infilled frames increases as the strength of the masonry increase. In turn, the strength of the masonry increases
with increase in strength of both mortar and brick units. From the study it is observed that the average increase
in strength for frames infilled with fly ash bricks and clay bricks is about 3 and 4 times that of the bare frame,
respectively. The average increase in lateral stiffness is about 6 times and 9 times for both fly ash brick and clay
brick infilled frames when compared to the stiffness of the bare frame. From the design consideration, it can be
191
proposed that the strong masonry leads to the failure of columns which has to be taken into consideration. A
detail study needs to be carried out in order to evaluate the strength of masonry required to achieve higher
lateral strength and stiffness as well to restrict failure of columns. After the masonry fails the frames behave as
bare frames. The main draw back in modeling the masonry as equivalent diagonal struts is their inability to
predict the local failures in frames.
REFERENCES
[1] Applied Technology Council (ATC). (1996). Seismic evaluation and retrofit of concrete buildings,
ATC-40, California Seismic Safety Commission, California, V 1.
[2] Asteris, P.G., Antoniou, S.T., Sophianopoulos, D.S., and Chrysostomou, C.Z. (2011). Mathematical
macro modeling of infilled frames: State of the art. J. Struct. Eng., 137(12), 1508–1517.
[3] American Society for Testing and Materials (ASTM). (2012a). Standard test methods for sampling
and testing brick and structural clay tile, Masonry test methods and specifications for the building
industry, ASTM C67-12, 4th Ed., Philadelphia.
[4] American Society for Testing and Materials (ASTM). (2012b). Standard test method for compressive
strength of masonry prisms, Masonry test methods and specifications for the building industry, ASTM
C1314-12, 4th Ed., Philadelphia.
[5] Basha, S. H. (2011). Evaluation of ductility provisions in seismic codes for masonry infilled RC frame
buildings, M.Tech thesis, Dept. of Civil Engineering, Indian Institute of Technology Guwahati, India.
[6] Comite Euro-International Du Beton (CEB) (1996). RC frames under earthquake loading: State of the
Art Report, Thomas Telford, London.
[7] Christy, C. F., Tensing, D., and Shanthi, R. M. (2013). Experimental study on compressive strength
and elastic modulus of clay and fly ash brick masonry. J. Civ. Eng. and Const. Tech., 4(4), 134–141.
[8] CSI (2013). Structural Analysis Program (SAP2000) ̶ Advanced, Static and Dynamic Finite Element
Analysis of Structures. Computers and Structures, Inc., Berkeley, USA.
[9] Chrysostomou, C. Z., Gergely, P., and Abel, J. F. (2002). A six-strut model for nonlinear dynamic
analysis of steel infilled frames. Int. J. Struct. Stab. Dyn., 2(3), 335–353.
[10] El-Dakhakhni, W. W., Elgaaly, M., and Hamid, A. A. (2003). Three –Strut Model for Concrete
Masonry –Infilled Steel Frames. J. Struct. Eng., 129(2), 177–185.
[11] Ghosh, A. K., and Amde, A. M. (2002). Finite element analysis of infilled frames. J. Struct. Eng.,
128(7), 881–889.
[12] Holmes, M. (1961). Steel frames with brickwork and concrete infilling. ICE Proc., 19(4), 473–478.
[13] Indian Standards (IS). (1987). Indian standard code of practice for structural use of unreinforced
masonry, IS1905, 3rd Rev., Bureau of Indian Standards, New Delhi, India.
[14] Indian Standards (IS). (1992a). Indian standard methods of test of burnt clay building bricks—Part 1:
Determination of compressive strength, IS 3495, 3rd Rev., Bureau of Indian Standards, New Delhi,
India.
192
[15] Indian Standards (IS). (1992b). Indian standard methods of test of burnt clay building bricks—Part 2:
Determination of water absorption, IS 3495, 3rd Rev., Bureau of Indian Standards, New Delhi, India.
[16] Kaushik, H. B., Rai, D. C., and Jain, S. K. (2007). Stress-strain Characteristics of clay brick masonry
under uniaxial compression. J. Mater. Civ. Eng., 19(9), 728–739.
[17] Kaushik, H. B., Rai, D. C., and Jain, S. K. (2009). Effectiveness of some strengthening options for
masonry-infilled RC frames with open first story. J. Mater. Civ. Eng., 135(8), 925–937.
[18] Malaviya, S. K., Chatterjee, B., and Singh, K. K. (1999). Fly ash –an emerging alternative building
material. Fly ash utilization for valued added products, National Metallurgical Laboratory,
Jamshedpur, 59-67.
[19] Paulay, T., and Priestley, M. J. N. (1992). Seismic design of reinforced concrete and masonry
buildings, Wiley-Interscience, New York.
[20] Sarangapani, G., Venkatarama Reddy, B. V., and Jagadish, K. S. (2005). Brick-mortar bond and
masonry compressive strength. J. Mater. Civ. Eng., 17(2), 229–237.
[21] Stafford-Smith, B. (1967). Methods for predicting the lateral stiffness and strength of multi-storey
infilled frames. Build. Sci., Vol. 2, Pergamon Press, Oxford, U.K., 247–257.
193
Paper No.: 032
ADAPTIVE RESPONSE SURFACE METHOD IN STRUCTURAL RESPONSE

APPROXIMATION UNDER UNCERTAINTY
1 2 3
Somdatta Goswami Subrata Chakraborty Shymal Ghosh
1
PhD Student, Dept.t of Civil Engineering, Bengal Engineering and Science University, Shibpur, Howrah, India.
Email: somdatta89@gmail.com
2
Professor, Department of Civil Engineering, Bengal Engineering and Science University, Shibpur, Howrah,
India. Email: schak@civil.becs.ac.in
3
PhD Student, Dept.t of Civil Engineering, Bengal Engineering and Science University, Shibpur, Howrah, India.
Email: gh13shyamal@gmail.com
ABSTRACT:
Repeated evaluation of structural response is required in reliability analysis, optimization, and model updating
due to iterative nature of computational algorithms. If the structure of interest is represented by a complex finite
element (FE) model, large number of computations involved can rule out many approaches due to the expense
of carrying out many runs involving considerable computational time. Furthermore, the structural FE models are
often constructed by using commercial FE software and during each iteration one needs to run the FE software
which limits the popular applications of FE method in practice. Hence, alternative techniques for efficient
computation of response of complex structures by overcoming aforementioned drawbacks while retaining the
accuracy is of paramount importance to structural engineering community. Response Surface Method (RSM)
has emerged as an effective solution to such problems. For large complex structures characterized by large
number of input variables, in order to make an efficient response surface requiring less number of FE executions
is very much desirable. In this regard, the Moving Least Square Method (MLSM), a local approximation
approach fits the response curve more accurately than Least Square Method (LSM), basically a global
approximation method. The focus of the present work is on the application of adaptive RSM for efficient
computation of structural response. For this, RSM is proposed using redundant design (i.e. SD with 2 or 3 levels)
in the framework of MLSM.
KEYWORDS: Complex Structure, Response Surface Method, Moving Least Square Method
1. INTRODUCTION
The reliability analysis, optimization, and model updating of structures typically needs to evaluate structural
response repeatedly due to iterative nature of most of the computational algorithms. Nowadays, the mechanical
response of structures is obtained with much improved numerical modelling for better accuracy. Among techniques
that can perform such modelling, the finite element (FE) method is probably the most used due to its ability to deal
with complex geometries and highly nonlinear constitutive behaviours. Thus, when the structure of interest is
represented by a complex FE model, large number of computations involved can rule out many algorithm for
reliability analysis, optimization, model updating etc. due to the expense of carrying out many runs involving
considerable computational time. Furthermore, the structural FE models are often constructed by using various
commercial FE software like ANSYS, ABAQUS, NASTRAN etc. and during each iteration one needs to run the
FE software which limits the popular applications of FE method for the above mentioned applications in
practice. Hence, alternative techniques for efficient computation of response of complex structures by
overcoming such drawbacks while retaining the accuracy are of paramount importance to structural engineering
community. Response Surface Method (RSM) has emerged as an effective solution to such problems.
The RSM is one of the most sparking developments in structural reliability analysis. It is highly suitable for the
case where the performance function has no known closed form expression and need to be evaluated by
numerical methods such as FE method. The use of RSM, originally proposed by Box and Wilson [1], is still
subjected to further researches [2-4]. The developments that has been taken place in the field of RSM can be
studied under three subheads: (i) Studies dealing with experimental design in physical space or in the standard
normal space for a better control of the distance between numerical experiments [5-10], (ii) Studies dealing with
the forms of RS functions i.e. use of polynomial RS like simple linear RS [5], a quadratic RS without mixed
terms [4,6], use of neural network [8] etc. and (iii) studies on various methods of experimental design in RSM
[4-7]. The application of RSM in structural reliability was first proposed by Faravelli [11] and the subsequent
works [4-7] set the tone for application of RSM for reliability analysis of large and complex structural system.
For constructing the RS various design of experiment (DOE) are used e.g. Saturated Design (SD), Factorial
Design (FD), Central Composite Design (CCD) etc.
It is well known that the accuracy of evaluating a performance function and its gradient largely depends on the
capability of a meta-model to capture the nonlinearity and local variation of behaviours. Though, the CCD
approximates very well compared to other designs, it requires enormous input structural response information
compared to other DOE approaches. Thus, in order to make an efficient RS requiring less number of FE
executions is very much desirable for large complex structures characterized by large number of input variables.
In this regard, the Moving Least Square Method (MLSM), a local approximation approach fits the response
curve more accurately than Least Square Method (LSM), basically a global approximation method [12-14]. The
focus of the present work is on the application of adaptive RSM for efficient computation of structural response.
For this, RSM is proposed using redundant design (i.e. SD with 2 or 3 levels) in the framework of MLSM.
Numerical study is performed to study the efficiency and accuracy of the proposed algorithm compared to the
CCD based RSM and direct simulation results.
2. RESPONSE SURFACE METHOD
The RSM is a set of mathematical and statistical techniques designed to gain a better understanding about the
overall response by DOE experiments and subsequent analysis of experimental data. The method primarily
uncovers analytically complicated or an unknown relationship between several inputs and desired output
through empirical models (non-mechanistic) in which the response function is replaced by a simple function
(often polynomial) that is fitted to data at a set of carefully selected points (referred as DOE), normally obtained
from experimental investigation or numerical simulation. In a sense, RSM is a system identification procedure,
in which a transfer function relating the input parameters (loading and system conditions) to the output
parameters (response in terms of displacements, stress, etc.) is obtained in a suitable way. To create a response
surface that will serve as a surrogate for the FE simulation model, the basic process consists of calculating
predicted values of the response features at various sample points in the parameter space by performing an
experiment at each of those points.
2.1 Design of Experiments
The DOE to select different sample points in the variable space is very important in the RSM application as it
will directly influence the efficiency and the accuracy of the response surface model fitting procedure. There are
various sample point selection methods, developed suitably for different types of RSM applications. The
popularly used methods in structural reliability analysis are the Saturated Design (SD), the Central Composite
Design (CCD) and the Factorial Design (FD). In SD the points chosen are the mean values of the response
variable xi (as centre point) and at axial points, xi = ±ℎ , where ℎ is a positive integer. The points chosen
for a type-3 two variables RSM are shown in figure 1a, hi are considered to be 1.0 in figure 1. The corner, axial
and centre points are selected in the CCD for fitting a second order response surface model. As it requires a
relatively large number of sample points, the CCD method is generally chosen only in the later stage of RSM
application when the number of important variables is reduced reasonably. The CCD consists of a complete 2k
factorial design (factorial design is detailed in next paragraph), centre point and two axial points on the axis of
each input variable at a distance of γ from the design centre, where γ = 2 in order to make the design cyclic.
195
For example, the sample points for a Type 3 two variables model are shown in figure 1.b. In between the SD and
the CCD methods, there is another popular design method known as the FD method (Figure 1.c), which selects
the corner points only. The full FD method generates q sample values for each co-ordinate, thus producing a
total of sample points for k variables. The 2k factorial points (i.e. q=2) for a sample of two variable problem
are presented in figure 3. It can be noted that even for moderate values of q and k this may become prohibitively
expensive (e.g. for q=3 and k=5, number of minimum sample points required is 243). However, for lower values
of q and k the method is quite effective (e.g. for q=2 and k=2, number of minimum sample points required is 4).
Figure 1.a The SD for two Figure 1.b The CCD for two variables Figure 1.c The FD for two
variables RSM RSM variables RSM problem
2.2. Conventional LSM based Response Surface Method
If there are n response values yi corresponding to n numbers of observed data, xij (denotes i-th observation of the
input variable xj in a DOE), the relationship between the response and the input variables can be expressed by
Eq. 1.
y = Xβ + ε y (1)
In the above multiple non-linear regression model X, y, β and εy are the design matrix containing the input
data from the DOE, the response vector, the unknown co-efficient vector and the error vector, respectively.
Typically, the quadratic polynomial form used in the RSM is as following:
k k k
y = β0 + ∑ βi xi + ∑ ∑ βij xi x j (2)
i =1 i =1 j =1
The LSM of estimation technique is usually applied to obtain the unknown polynomial coefficient by
minimizing the error norm defined as in Eq. 3.
2
n  k k k 
)  = ( y - Xβ ) ( y - Xβ )
T
L = ∑  yi − β0 − ∑ βi xi − ∑ ∑ βij xi x j (3)
i =1 i =1 i =1 j =1 
And the least squares estimate of β is obtained as,
{XT y}
−1
β =  XT X  (4)
 
Once the polynomial coefficients β are obtained from the above equation, the response y can be readily
evaluated for any set of input parameters. To fit an accurate model within reasonable time, it is required that the
initial input data (X and y) are selected judiciously. The main disadvantages of applying LSM to determine the
196
coefficients of approximating polynomials to obtain a suitable response surface are limitations in the shapes that
linear models can assume over long ranges, possibly poor extrapolation properties, and sensitivity to outliers. As
the explanatory variables become extreme, the outputs of the linear model will also always more extreme. This
means that linear models may not be effective for extrapolating the results of a process for which data cannot be
collected in the region of interest. Of course extrapolation is potentially dangerous regardless of the model type.
Finally, while the method of least squares often gives optimal estimates of the unknown parameters, it is very
sensitive to the presence of unusual data points in the data used to fit a model. One or two outliers can sometimes
seriously skew the results of a least squares analysis. This makes model validation, especially with respect to
outliers, critical to obtaining sound answers to the questions motivating the construction of the model. Moreover,
LSM is global approximation of the data. One of the limitations of using least squares method in analysis is that
outliers, which are significantly bad observations, can skew the results because they have more impact. This
impact is because the square of a number grows large faster than the number. It is better to reject the outliers
using some other method prior to using least squares on the remaining data. Of course, this must be
substantiated because rejecting data otherwise is bad practice.
2.3 MLSM Based Proposed Response Surface Method
The fundamental concepts of the LSM and MLSM based RSM construction is explained considering figure 3.
The DOE points (or the scattered input data points) are denoted by the dots. The LSM yields a global
approximation, which is represented by the dotted curve. But, such predicted responses fail to capture the actual
trend of the responses within a local domain which is pointed out by the large circles. On the other hand, owing
to in-built weight factors, the MLSM produces, basically a local approximation approach and can capture the
zone-wise variations of the actual responses. Considering the scattered data as a whole, only one approximation
curve can be obtained by the LSM based prediction. On the contrary, with the MLSM based approach, there
exist separate approximation functions for each calculation point of interest. Therefore, the coefficients of the
response surface model are not constants but are varying with respect to the position of the desired point of
interest. This locally weighted approximation can be obtained from the consideration of the effective data near
the desired location and the data is weighted according to the distance from this location. The concepts of the
LSM and the MLSM based RSM approximations.
Figure 2 The concept of LSM and MLSM based Figure 3 The space truss: (a) 3D view, (b) side view and
RSM approximations (c) top view
In essence, the MLSM based RSM can be considered as a weighted least squares method that has varying
weight functions with respect to the position of approximation. Re-calling equation (4) the modified error norm
Ly(x) can be defined as the sum of the weighted errors,
197
Q
Min. L y ( x ) = ∑ Wiε i2 = òT W(x )ò = (Y - xβ)T W(x)(Y - xβ)
i −1
(5)
−1
which yields β( x ) = xT W(x)x  xT W( x )y
 
In the above equation, W(x) is the diagonal matrix of the weight function and it depends on the location of the
associated approximation point of interest (x); Q is the total number of sample points. W(x) can be obtained by
utilizing the weighting function such as constant, linear, quadratic, higher order polynomials, exponential
functions etc. [23] as described below,
 Constant 1
 
 Linear 1 − d / RI
if (d / RI < 0), Quadratic 1 − (d / RI )2
w( x − xI ) = w(d ) =  Polynomial
(6)
 1 − 6(d / RI )2 + 8(d / RI )3 − 3(d / RI )4

 Exponential exp(−d / RI )

if (d / RI > 0) w(d ) = 0
Where, d is the distance of the point where the approximate response is required to the origin of the
approximating domain and R I is the radius of the sphere of influence, chosen as twice the distance between the
centre point and the extreme most experimental point. The value of RI is so chosen in order to secure sufficient
number of neighbouring experimental points so as to avoid singularity. More details about the calculation of RI
can be found elsewhere [12, 14-15]. It can be noted from equation (8) that the weight associated with a
particular sampling point xi decays as the point x moves away from xi. The weighting function has its maximum
value of 1.0 at 0.0 normalized distance and 0.0 (minimum value) outside of 1.0 normalized distance, i.e. w (0.0)
= 1.0, w (d/RI>1.0) = 0.0. The function decreases smoothly from 1.0 to 0.0. The exponential form of the weight
function is used in the present numerical study. Eventually, the weight matrix W(x) can be constructed by using
the weighting function in diagonal terms as bellow:
 w( x − x1) 0 ... 0 
 0 w( x − x 2) ... 0 
W ( x) =   (7)
 ... ... ... ... 
 
 0 0 ... w( x − xn ) 
Other form of weight functions are also adopted in the previous study [26]. By minimizing the modified error
norm Ly(x), the coefficients β can be obtained from Eq. 7. It can be noted that the coefficient β is a function of
the location or position x. Thus, the procedure to calculate ‘β’ is a local approximation, and ‘moving’ processes
performs a global approximation over the entire design domain. It is expected that not only the number of
iteration to obtain the converged RS will be reduced but also the accuracy of computed reliability using RS
obtained by the proposed MLSM based approach will be also improve.
3. NUMERICAL STUDY
The example considered is a thirty nine bar truss structure as shown in Figure 3. The height of the structure is 16
m. Its basis is an equilateral triangle of side 6.93 m. The model consists of 15 nodes and 39 elements grouped
into 4 design variables. A vertical load of 2kN is applied at all nodes of the truss. The horizontal load F is acting
at the top three nodes of the truss in the x-direction which is considered to be random. The statistical properties
of various random variables involve in this problem are depicted in Table 1.
198
Table 1 39-Bar Truss: Characteristics of the Random Variables
Probability Density
Random Variables Mean Value COV
Function
E(kN/m2) Young’s Modulus Normal 2.1*108 10%
F(kN) Horizontal Loading Normal 8000 20%
D(mm) Diameter Uniformly Distributed 140 5%
T(mm) Thickness Uniformly Distributed 4 5%
To demonstrate the effectiveness of the MLSM based RSM in approximating the response of structure, the
variation of mean horizontal displacement at node ‘A’ with number of simulation as obtained by the SD based
RSM and CCD based RSM and direct MCS are shown in figure 4 a and b, respectively. Both the LSM and
MLSM based RSM as mentioned in the previous sections are used to approximate the response for comparative
study. The SD method consists of nine data points and the CCD needs twenty five data points for four random
variables considered in the present study. The response evaluations at those points are performed by FE analysis
of space truss using two-nodded truss element having three degrees of freedom at each node. For comparative
study, the responses are also obtained by the direct MCS technique. For this, random samples of the four
random variables are generated following the associated known probabilistic information as furnished in table 1.
Also the same input (as generated for direct MCS) are entered in all the RS models to obtain the approximate
response from the associated RS models. From figure 4 it is evident that the SD-LSM cannot predict the FE
results with sufficient accuracy. Though, the SD-MLSM approximates reasonably; there is scope of further
improvement of the prediction as notable variation with direct MCS results observed. As expected it is evident
from figure 4b that the CCD-MLSM gives the best approximations of responses, but more numbers of FE
response evaluations are required in the DOE.
7.28 7.28
SD_LSM CCD_LSM
SD_MLSM CCD_MLSM
Mean Displacement (mm)
Direct MCS 7.27 Direct MCS

7.26
7.26
7.24
7.25
7.22
7.24
7.20 7.23
2000 4000 6000 8000 10000 2000 4000 6000 8000 10000 12000
Number of Simulation Number of Simulation
Figure 4 Comparison of the response predicted by Figure 5 Comparison of the response predicted by CCD
SD (LSM and MLSM) (LSM and MLSM)
Now to further improve the SD-MLSM results, the redundant design (RD) are used. Figure 6 shows the
variation of the mean displacement of node A, along the x-axis with number of simulation for DOE considering
level 1, level 1+2 and level 1+2+3 data points considering both LSM and direct MCS. Figure 7 shows the
variation of mean displacement of node A, along the x-axis with the number of simulation for level 1, level 2
and level 3 DOE data for SD-MLSM and direct MCS. From figure.7 it can be noted that if 3 levels of data are
used to construct the RS curve, it better approximates, rather almost close to direct MCS results.
199
7.30
SD1_LSM SD2_LSM SD1_MLSM SD2_MLSM
SD3_LSM Direct MCS SD3_MLSM Direct MCS

7.28 7.28
7.26
7.24 7.24
7.22
7.20 7.20
2000 4000 6000 8000 10000 2000 4000 6000 8000 10000
Number of Simulation Number of Simulation
Figure 6 Comparison of the response obtained by Figure 7 Comparison of the response obtained by
different levels of SD-LSM different levels of SD-MLSM
Furthermore, to study the effectiveness of the MLSM based RSM, various statistical metrics i.e. the Root Mean
Square Error (RMSE), the co-efficient of determination (R ) and the average prediction error (εm) usually used
to check the validity of the meta-models are computed for both the LSM and MLSM based RSM [27-28]. These
are briefly discussed in the following. The RMSE is defined as:
∑ ( yˆ
i =1
i − yi ) 2 / p , (8)
In the above, yî is the predicted response obtained by the meta-model and yi is the actual response obtained
by the direct MCS for ith sample point. The total number of sample (p) is taken as one lakh for the present
numerical computation. The co-efficient of determination (R2) and the average prediction error ( m) is defined as:
p p
| yî − yi |
∑ ( yî − yi )2 ∑ (100 yi
)
And ε m =
i =1
R2 = i =1
p
(9)
p
∑(y − y )
i =1
i i
2
For a good metamodel, these coefficients should be as close as possible to the value of 1.0 and RMSE should be
minimum. The trends of the numerical results as shown in table 2 clearly indicate the superiority the MLSM
based RSM.
Table 2 The performance of LSM and MLSM based RSM
RMSE R2 εm
LSM 82.50% 0.654 9.3%
Level 1
MLSM 26.41% 0.934 2.71%
LSM 26.28% 0.954 2.67%
SD Level 1+2
MLSM 26.18% 0.956 2.65%
LSM 26.25% 0.957 2.66%
Level 1+2+3
MLSM 26.15% 0.964 2.64%
LSM 12.1% 0.984 1.96%
CCD
MLSM 11.2% 0.986 1.38%
200
4. CONCLUSIONS
An efficient MLSM based RSM method is investigated for approximating responses of mechanical model;
otherwise typically which need to be obtained by FE method. The approach is intended for response evaluation
of large complex system. Thus, simplified DOE i.e. SD is adopted for efficient computation. The CCD-MLSM
approach gives the best approximation followed by the CCD-LSM, SD-MLSM, SD-LSM respectively in order
of accuracy when a single level is considered in the DOE. The results of SD-MLSM can be improved drastically
if the number of level in the DOE is increased. Though, the CCD gives very accurate results yet applicability of
SD seems to be relevant considering the fact that CCD requires 2k+2k+1 design points whereas SD requires
2k+1 points. Thus, the number of FE runs required for large complex structures (e.g. for 21 variables one needs
more than twenty lakhs FE runs to generate response surface by CCD) makes CCD impractical. It has been
found that the MLSM based RSM captures the trend of the responses, obtained by the direct MCS based results,
much more efficiently and more accurately. The statistical error metrics, i.e. RMSE, R2 and ε m are also found
to be better. By studying the various norms in table 2 one can easily realise that though the SD-MLSM provides
better values compared to SD-LSM, particularly when number of levels are increased. Though, the SD-MLSM
approximates reasonably; the difference with CCD is still notable and there is scope of further improvement of
the prediction. This can be possibly attempted considering specific application of responses. However, this
needs more study.
REFERENCES
[1] Box, G.E. P. and Wilson, K. B. (1954). The exploration and exploitation of response surfaces: some general
considerations and examples’, Biometrics 10, 16–60.
[2] Das, P. K. and Zheng Y. (2000). Improved response surface method and its application to stiffened plate
reliability analysis. Engineering Structures; 22(5): 544–551.
[3] Roussouly, N.; Petitjean, F.; Salaun M. (2013). A new adaptive response surface method for reliability
analysis.Probabilistic Engineering Mechanics; 32:103-115.
[4] Khuri A. I. and Mukhopadhyay, S. (2010). Response surface methodology, WIREs Computational Statistics;
2; 128–149.
[5] Bucher, C. G. and Bourgund, U. (1990). A fast and efficient response surface approach for structural
reliability problems. Structural Safety; 7: 57-66.
[6] Kim, S. H. and Na, S. W. (1997). Response surface method using vector projected sampling points.
Structural Safety; 19:3-19.
[7] Liu, Y. W. and Moses, F. (1994). A sequential response surface method and its application in the reliability
analysis of aircraft structural systems. Structural Safety; 16:39-46.
[8] Rajashekhar, M. R. and Ellingwood, B. R. (1993). A new look at the response surface approach for
reliability analysis, Structural Safety; 12:205-220.
[9] Lemaire, M.; Mohamed, A.; Flore`s, M. O. (1997). The use of finite element codes for the reliability of
structural systems. In: 7th IFIP WG 7.5 Working Conf on Reliability and Optimisation of Structural Systems,
Boulder, CO 2-4 April, 1996. Elsevier (Pergamon), 223-30.
[10] Enevolsen, I.; Faber, M. H.; Sorensen, J. D. (1994). Adaptive response surface technique in reliability
estimation. In: Schueller, Shinozuka, Yao, eds. Structural safety and reliability:1257–64.
[11] Lemaire, M. (1998). Finite element and reliability: combined methods by response surface. In:
Frantziskonis GN. Ed., Probabilities and materials tests, models and applications for the 21st century,
NATO ASI series 3 high technology; 46. Kluwer Academic, 317-31.
[12] Faravelli, L. (1989). Response surface approach for reliability analyses. Journal of Engineering Mechanics
ASCE; 115(2); 2763-2781.
201
[13] Kim, C.; Wang, S.; Choi, K. K. (2005). Efficient response surface modeling by using moving least-squares
method and sensitivity. AIAA Journal; 43(1): 2404-2411.
[14] Bhattacharjya, S. and Chakraborty, S. (2011). Robust optimization of Structures subjected to stochastic
earthquake excitation under limited information on system parameters uncertainty, Engineering
Optimization; 43(12):1311-1330.
[15] Taflanidis, A. A.; Cheung S.H. (2012). Stochastic sampling using moving least squares response surface
approximations. Probabilistic Engineering Mechanics 2012; 28: 216-224.
202
Paper No.: 035
PREDICTION OF FATIGUE CRACK PROPAGATION LIFE IN SINGLE EDGE

NOTCHED BEAM USING EXPONENTIAL MODEL
Avaya K. Baliarsingh1, P.K. Ray2, B.B. Verma3, Deepak K. Agarwalla4

1,4
Research Scholar, Dept. of Mechanical Engineering, National Institute of Technology Rourkela, India
2
Professor, Dept. of Mechanical Engineering, National Institute of Technology Rourkela, India
3
Professor, Dept. of Metallurgy, National Institute of Technology Rourkela, India
Email: deepak_cet08@yahoo.co.in, dips50k@gmail.com
ABSTRACT :
Metal beams are extensively used in structures, automobile sectors and machine components. Some of their
applications include connecting rod of IC engine, shafts, axles, and gears, structures members of bridges and
also components of machines. Most of them experience fluctuating or cyclic load condition in their service life.
The occurrences of high amplitude cycles are also vary to these common components. The monitoring and
modeling of fatigue crack growth are more significant for the stability and safety of machines and structures.
The primary objective of the present investigation is to develop a fatigue crack propagation model for single
edge notched (SEN) cracked beams behavior. The metal beams under investigation are of (EN8) medium carbon
steel. The four point bend test has been conducted on pre-cracked specimen (straight notch at the center) using
Instron 8800 servo-hydraulic machine. All the tests have been conducted in air and at room temperature. In the
present investigation an attempt has been made to develop a fatigue life prediction methodology by using an
Exponential Model for single edge notched (SEN) cracked beams. The predicted results have been compared
with experimental crack growth data. It has been observed that the results obtained from the experimental model
are in good agreement with experimental data.
KEYWORDS: Fatigue crack propagation, crack profile, constant amplitude loading, life prediction,
exponential model.
1. INTRODUCTION
Failure due to repeated loading, that is fatigue, has accounts for at least half of this mechanical failure. No exact
data is available, but many books and articles have suggested that between 50 to 90 per cent of all mechanical
failures are due to fatigue, most of this is unexpected failures [1]. In many situations a beam experiences
fluctuating loading conditions. This may initiate and propagate a crack. The monitoring and modeling of fatigue
crack growth is more significant for the stability and safety of machines components, bridges, aircraft and
structures. In this project (EN8) medium carbon steel beam is used. In fatigue fracture the stress is generally
below the yield stress. In general ductile material deforms before fracture and gives warning before failure of a
component but in case of fatigue failure the ductile materials fails suddenly. This becomes more significant
when failure is related to automobile sectors or machinery parts in which heavy loads or continuous work being
done. In a dynamic world, however, failure occurs at stresses much below the material‘s ultimate strength or
yield strength. This phenomenon, failing at relatively low stresses, came as quite a surprise to most engineers in
the early years of metal component design and manufacturing. The other frustrating aspect is that the material
exhibited no sign of its tiredness or fatigue and could fail without much warning. This could be more dangerous
if proper selection of design criteria is not selected and validation of those criteria with experiment is not done.
There are many areas where the fatigue criteria should be in mind before designing the component.
An extensive literature survey has been conducted to review the earlier research and to identify the
research gap. Stephens et al. [2] reported that fatigue crack growth curve for constant amplitude loading
consisting of the crack growth rate (da/dN) versus the stress intensity factor range (∆K) in the log–log scale
typically includes three regions. Region-I is the near threshold region and indicate the threshold (∆Kth) value
which there is no observable crack growth. Microstructure, mean stress, frequency, and environment mainly
control Region I crack growth. Region II corresponds to stable macroscopic crack growth that is typically
controlled by the environment. In Region III the fatigue crack growth rates are very high as they approach to
instability. In Region III crack growth is often ignored in practice due to the insignificant fatigue life remaining
upon entering the region. Structural engineers have been utilizing numerical tools/ software packages of Finite
Element Method or Boundary Element Method to assess their designs for strength including crack problems.
BEM has emerged as a powerful alternative to Finite Element Method (FEM) for cases where better accuracy is
required due to situations such as stress concentration (as in the case of a crack), or an infinite domain problem.
Since BEM only requires discretization of surfaces (in case of 3D problems) and discretization of lines (in case
of 2D problems), it allows modeling of the problem becoming simpler. Aliabadi [3] reported various
applications of BEM to solve solid mechanics problems. Boundary element formulations for modeling the
nonlinear behavior of concrete were reported by Aliabadi and Saleh [4]. Fatigue crack growth is required for the
assessment of residual fatigue life, or for a damage tolerance analysis to aid structural design. In this paper
fatigue crack growth of corner crack in wide flange I-beam under constant amplitude bending loading are
presented. A quarter-elliptical corner crack in a prismatic solid is analyzed as benchmarking model for the
available analytical solution [5] prior to making further modeling of the cracks. Fatigue is defined as the process
of progressive localized permanent structural change occurring in a material subjected to conditions that
produce fluctuating stresses at some point or points and that may culminate in cracks or complete fracture after
certain number of fluctuations [6]. The stress value in case of fatigue failure is less than ultimate tensile stress
and may be below yield stress limit of the material. Cyclic stresses are characterized by maximum, minimum
and mean stress, the range of stress, the stress amplitude, and the stress ratio. Depending on the complexity of
the geometry and the loading, one or more properties of the stress state need to be considered, such as stress
amplitude, mean stress, biaxiality, in-phase or out-of-phase shear stress, and load sequence[7]. Frost and Pook [8]
proposed that the crack growth occurs under cyclic loading not as a consequence of any progressive structural
damage but merely because unloading reshapes the crack tip at each cycle and this sequence is responsible for
the formation of striations [9]. Therefore the increment of crack growth in each cycle can be related to the
changing crack tip geometry during its opening and closing and a simple model based on. Lal and Weiss [10]
proposed a static fracture model, that the crack was advanced by a distance (during each cycle) over which the
maximum normal stress exceeded the critical fracture stress of the metal. Their model was moderately
successful in predicting fatigue crack growth except it did not account for the crack tip blunting phenomena and
their model used material constants that had no physical significance. However Zheng [11] improved the La1
and Weiss model by modifying the static fracture portion and obtained material constants that are defined by the
tensile properties of the metal. Briefly, Zheng found that upon loading the crack opens elastically and the crack
tip becomes blunt, which occurs at a stress intensity threshold denoted as Kth. The exponential model has been
developed by Mohanty et. al. [12-15] for prediction of fatigue crack growth in SENT specimen for constant
amplitude loading as well as variable amplitude loading. However, this model was not developed for beams
specimens. In the present investigation an attempt has been made to use the exponential model for fatigue crack
propagation life in cracked beams subjected to constant amplitude loading.
Reviewing the above works by various researchers, it has been found that various numerical and
experimental techniques have been adopted by various researchers to predict the fatigue crack propagation in
beams of various materials. In this work a new technique i.e. based on exponential model has been used to
predict fatigue crack propagation in single edge notched cracked beam. The results from the proposed method
has been validated by the experimental results using (EN8) medium carbon steel under constant amplitude
loading condition with different stress ratios.
2. EXPONENTIAL MODEL
Fatigue crack propagation, a natural physical process of material damage, is characterized by rate of increase of
crack length (a) with number of cycles (N). It requires a discrete set of crack length vs. Number of cycle data
generated experimentally. Unlike monotonic tests, fatigue test data are usually scattered. Therefore curve fitting
of experimental a-N data was done which are usually scattered. The exponential model is based on the
exponential nature of fatigue crack propagation with number of loading cycles. The exponent (known as specific
204
growth rate) of the proposed exponential model has been correlated with various physical variables like crack
driving parameters, crack resisting parameter, and material properties in non-dimensional forms. Finally the
validation of the model has been done with experimental data in order to compare its accuracy in predicting
fatigue life of cracked beams.
Use of exponential model was first suggested by Thomas Robert Malthus (1766-1834) for the prediction of
growth of human population/bacteria. He realized that any species could potentially increase in numbers
according to an exponential series. The differential equation describing an exponential growth is,
dP (1)
= rP
dt
Where P is population, t is time and the quantity r in the above equation is the Malthusian parameter, also
known as specific growth rate. The solution of the above differential equation is
P (t ) = Po ert (2)
This equation is called law of growth.
In our present research equation (2) is modified and form of exponential equation of the proposed exponential
model is as follows:
(N − Ni )
mij
a j = ai e j
(3)
The exponent, i.e. specific growth rate (mij) has been calculated by taking logarithm of the above equation as
follows:
 aj 
ln  
ai
mij =   (4)
( N j − Ni )
In conventional differential equation model of Paris-Erdogan, there has been a physical inconsistency when the
constants of the crack growth rate equation are randomized as per dimensional analysis point of view [16]. In
case of the proposed exponential model, this type of inconsistency does not arise as the specific growth rate ‘mij’
is a dimensionless parameter. The specific growth rate mij is given by:
 ∆k   K Max   σ ys 2 a 
A B C
m=     2  (1 − R ) D (5)
 kC   kC   kC 
Where m is correlated with two crack driving forces ∆K and Kmax as well as material parameters Kc, R, σYS and is
represented by equation (5).
2.1 Formulation of the Model
The modified exponential equation is given as:
(N − Ni )
mij
a j = ai e j
(6)
 aj 
ln  
ai
mij =   (7)
( N j − Ni )
Here Nj and Ni represent number of cycles in ith step and jth step respectively, and aj and ai are the crack lengths in
ith step and jth step respectively. mij is specific growth rate in the interval (j-i).
The specific growth rate mij has been calculated for each step from experimental result of fatigue test (a-N data)
according to equation (7).
205
The different m is fitted by 3rd degree polynomial. The predicted m values have been calculated for different
specimens by using formula:
 ∆k   K Max   σ ys 2 a 
A B C
m=   k   k 2  (1 − R )
D (8)
 C   C   C 
k
where A, B, C, and D are curve fitting constants whose average value for four specimens have been presented in
the table 1.
From the bending moment equation; σ/y=M/I=E/R
σ max and σ min have been calculated according to the loads and specimen dimensions for four points bend test.
Stress range (∆ σ) = σmax -σmin
The stress intensity factor Kmax has been calculated by following equation,
Stress intensity factor kmax = σ max πaF ( g ) (9)
The stress intensity factor range ∆K has been calculated by following equation,
Stress intensity factor range ∆k =∆ σ πaF ( g ) (10)
Where F (g) is the shape factor related to the geometry of the component
6 2 tan ( π a / 2W )
f (a / W ) =  0.923 + 0.199 {1 − sin (π a / 2W )}4  (11)
cos (π a / 2W )  
Where F (g) = a/W
2.2 Validation of the Model

The predicted number of cycles or fatigue life has been given by:
N Pj =
ln
()
aj
ai
+N i
(12)
mij
The predicted values of specific growth rate (mij) of the tested specimen have been calculated by putting the
average values of the curve fitting constants (for specimen no. 1, 2, 3, 4,) in equation (5.3.3). For validation of
proposed exponential model; fatigue life is calculated (for specimen no. 5 and 6) by using the equation 12.
Table 1 Value of coefficients for exponential model
Material A B C D
EN8 7574.09 -8566.67 499.31 4.84
Fig. 1 ( a-N) curve of EN8 medium carbon steel beam (exponential

206
3. EXPERIMENTAL MODEL
The fatigue crack propagation tests have been conducted on (EN-8) medium carbon steel beams. All the tests
have been conducted in a servo-hydraulic dynamic testing machine (Instron-8800) using through cross-sectional
of cracked beam specimen under load control mode. A four point bend fixture has been fabricated for
conducting fatigue crack propagation tests. Before conducting the test, COD (crack opening displacement)
gauge has been calibrated for single edge straight notched cracked beams specimens. All the tests have been
conducted in air and at room temperature.
Table 2 Specimen and notch dimension of beam
Specimen Dimension (mm)

Length of beam(L) 510
Width of beam(B) 25
Thickness of beam (t) 25
Notch length of beam(l) 3
Upper span length(S/2) 225
Lower span length(S) 450
The test has been conducted on Instron 8800 as shown in

fig. 2. It consists of a servo hydraulic loading system,
support for the specimen and various instruments for
measurement of data. A servo hydraulic controlled
actuator of ±250 KN capacity and ±50 mm displacement
has been used for loading. The support system consists of
two pedestals with two rollers, at a span of 450 mm and a
pair of inner loading rollers with a span of 225 mm, which provides four-point bending. The test specimen has
been gripped between rollers. Four points bending test has been implemented in which the mid-section of the
specimen where the notch is located has been subjected to pure bending. The monitoring of crack has been done
using a COD gauge with data acquisition system, the data storage system has been integrated with machine
which stores data at every point of testing. Beam test arrangement has been loading the beam under four point
bending up to large scale plastic deformation with periodic significant unloading so as to create a beach mark on
the crack surface. After the test, fractured surface has been extracted by Power saw and it has been cut in such a
way that the crack surface is not damaged. Then the extracted crack length has been examined in scanning
electron microscope (SEM) at various loading stages. The crack surface has been again examined in travelling
microscope. The loading has been given in the form of sinusoidal wave. The tests have been conducted at room
temperature and laboratory air environment under constant amplitude sinusoidal loading at stress ratio of 0.1 &
0.3 and frequency 3 Hz. The load range applied during the fatigue crack initiation and growth test have been of
the order of 8 KN and 11KN.

The basic aim of present work has been to develop a fatigue crack propagation model for SEN beam without going
through numerical integration process. The specific growth rate (m) has been an important parameter of our model.
The value of m has been correlated with two crack driving forces (∆K and Kmax), and with material parameters
fracture toughness (KC), yield strength (σYS), stress ratio(R), and Young’s modulus (E) by curve fitting. The
experimental a-N data of four specimens have been used for formulation of model, and its validation has been
checked by experiments conducted on different specimens. The average value of curve fitting constants for beam
has been found out by using the constants fatigue life of a beam specimen. The predicted result by using
207
exponential model has been compared with the experimental result. The a-N curve obtained from proposed
exponential model and that obtained from experimental data have been compared with other curves. It has been
found that the predicted results have been in good agreement with the experimental results.
5. CONCLUSIONS
The conclusions drawn from the current research have been depicted below.
The calibration curve of COD gauge has been found to be straight line, which shows linear relationship between
COD gauge and crack depth of straight notched beam. The crack front profile is approximately thumbnail shape
(N )
m ij
in nature as the crack depth increases. Exponential model of the form a j = aie j − Ni has been used for
other specimen geometries and can also be used to determine the fatigue life in beams without going through
numerical integration. The proposed exponential model can be effectively used to predict fatigue crack
propagation under constant amplitude loading condition with different stress ratios.
REFERENCES
[1] Suresh, S. (1991). Fatigue of Materials, Cambridge University Press, Cambridge.
[2] R.I. Stephens, A. Fatemi, R.R. Stephens and H.O. Fuchs, Metal Fatigue in Engineering, New York: Wiley
Interscience, 2001
[3] M.H. Aliabadi, the Boundary Element Method, Chichester: John Wiley, 2002.
[4] M.H. Aliabadi and A.L. Saleh, Fracture Mechanics Analysis of Cracking in Plain and Reinforced Concrete
Using the Boundary Element Method, Engineering Fracture Mechanics, 2002, 69, 267-280.
[5] A. Neves, S.M. Niku, J.M.W. Baynham, and R.A. Adey, Automatic 3D Crack Growth Using BEASY. In:
Proceedings of 19th Boundary Element Method Conference, Computational Mechanics Publications,
Southampton, 1997, pp. 819–827.
[6] ASM handbook, fatigue and fracture, Vol.19, 1996, pp. 507-543.
[7] Mechanical Metallurgy by George E. Dieter (TMH); 1961.
[8] Frost, N.E., L. P. Pook and K. Denton, 1971. “A Fracture Mechanics Analysis of Fatigue Crack Growth
Data for Various Materials”, Engineering Fracture Mechanics, vol. 3(2), pp. 109-126.
[9] Chell, G. G. Developments in Fracture Mechanics - I, Applied Science Publishers Ltd., London, 1979.
[10] La1, D. N. and V. Weiss, “A Notch Analysis of Fracture Approach to Fatigue Crack Propagation”,
Metallurgical Transactions, vol. 9A, pp. 413-425, 1978
[11] Zheng, Xiulin and A. Manfred, “Fatigue Crack Propagation in Steels”. Engineering Fracture Mechanics,
vol. 18(3), pp. 965-973, 1983
[12] J. R. Mohanty, B. B. Verma, and P. K. Ray, “Prediction of fatigue crack growth and residual life using an
exponential model: Part I (constant amplitude loading),” International. Journal of Fatigue. Elsevier, vol. 31,
pp. 418-424, 2009
[13] J. R. Mohanty, B. B. Verma, and P. K. Ray, “Prediction of fatigue crack growth and residual life using an
exponential model: Part II (mode-I overload induced retardation),” International Journal of Fatigue.,
Elsevier, vol. 31, pp. 425-432, 2009.
208
[14] J. R. Mohanty, B. B. Verma, and P. K. Ray, “Prediction of fatigue life with interspersed mode-I and mixed
mode (I & II) overloads by an exponential model: Extensions and Improvements,” Engineering Fracture
Mechanics, Elsevier, vol. 76, pp. 454-468, 2009.
[15] J. R. Mohanty, B. B. Verma, and P. K. Ray, “Prediction of fatigue life with interspersed mode-I and mixed
mode (I & II) overloads by an exponential model: Extensions and Improvements,” Engineering Fracture
Mechanics, Elsevier, vol. 76, pp. 454-468, 2009.
[16] J. R. Mohanty, B. B. Verma, and P. K. Ray, “Evaluation of Overload-induced Fatigue Crack Growth
Retardation Parameters using an Exponential Model,” Engineering Fracture Mechanics, Elsevier, vol. 75,
pp. 3941-3951, 2008.
209
Paper No.: 036
FUZZY LOGIC BASED FAULT DIAGNOSIS OF STRUCTURES USING

VIBRATION RESPONSES
Deepak K. Agarwalla1, Dayal R. Parhi2, Amiya K. Dash

1
Research Scholar, Dept. of Mechanical Engineering, National Institute of Technology Rourkela, India
2
Professor, Dept. of Mechanical Engineering, National Institute of Technology Rourkela, India
3
Professor, Dept. of Mechanical Engineering, I.T.E.R., Bhubaneswar, India
Email: deepak_cet08@yahoo.co.in, dayalparhi@yahoo.com,amiyadash@yahoo.com
ABSTRACT :
Most of the engineering applications in recent times can’t be imagined without the services of structures,
whether it is mechanical, civil, aerospace etc. So maintaining the structural integrity is the top most important
agenda for any engineer, scientist or researcher. Generally, structures are designed keeping in mind it’s long time
applicability. But during its performance period, it undergoes various loading conditions which pull down the
load carrying capacity of the structural beam member which may be the prime source of generation of faults i.e.
damage or crack. Therefore it is evident for a researcher to identify the damage intensity and position, which
enable the engineer to take proper measure to ensure longer life span to the structural member. In this research
work, proper diagnosis has been accomplished by subjecting the faulty structure to vibration and consequent
comparison of vibration responses such as relative natural frequencies and relative mode shape differences with
respect to the intact structural member. The alteration in vibration responses has been used to formulate the
fuzzy logic based model for diagnosis of damage or fault. In this paper, structural members of Al and steel have
been considered for the experimentation and the subsequent results have been validated by finite element
analysis using commercially available analysis software. The change in vibration responses from the
experimental method has been utilized as the input for fuzzy logic based model, which in turn diagnose the
damage effectively.
KEYWORDS: Fault, vibration response, relative natural frequency, relative mode shape difference, finite
element analysis, fuzzy logic
1. INTRODUCTION
Most of the structural failures encountered are caused by material fatigue and presence of damages in structures.
Therefore, damages of any form are to be diagnosed as earliest as possible to maintain the integrity of the
structures. In spite of the existence so many traditional methods, but presence of any damage can’t be ensured
without diagnosing the entire structure. In the current section, a fuzzy logic based technique has been proposed
for structural damage identification. The proposed methodology utilizes the modal characteristics of the beam
structure using Artificial Intelligence (AI) techniques and anticipates the position and severities of the damage
present in the system. Fuzzy system has the capability to imitate the human behavior by following the different
reasoning phases in order to make the computer program behave like humans. Fuzzy logic can specify mapping
rules in terms of words rather than numbers. Another basic concept in Fuzzy logic is the fuzzy if–then rule
which is mostly used in development of fuzzy rule based systems. Fuzzy logic is a convenient way to map an
input space to an output space and is one of the tools used to model a multi-input, multi-output system. Hence
the fuzzy approach can be effectively employed to develop a damage diagnostic tool using the vibration
responses of structures.
Nahvi et al. [1] have used natural frequencies and mode shapes of the beam structure as input parameters for
analytical and finite element method to identify the crack in cantilever beam. Salam et al. [2] have analyzed the
lateral vibration of an Euler-Bernoulli beam accompanied with a single edge open crack to compare the mode
shapes of damaged and undamaged beam by a interpreted formula for the stress correction factor in terms of the
damage characteristics. Murigendrappa et al. [3] have diagnosed damaged pipes of aluminium & mild steel with
water as fluid medium at different pressure utilizing the variations in natural frequencies. . Owolabi et al. [4]
have investigated the position and severity of crack for Al fixed-fixed and simply supported beams by fixing the
first three natural frequencies and mode shapes. Gounaris et al. [5] have established a functional relationship
between the crack parameters and modal responses assuming development of an open crack and the results are
approved by the Eigen mode comparison of damaged and undamaged beam. Wang et al. [6] have investigated
the crack position and severity of a composite cantilever having a surface crack by establishing a dependency
between frequencies of the vibrating cantilever beam and material properties. Boutros et al. [7] have identified
the transient and gradual abnormalities using fuzzy logic approach with the help of four condition monitoring
indicators. They have compared the fuzzy based technique in two different applications with satisfactory
validation. Miguel et al. [8] have developed a decision making segment based on fuzzy logic the damage
diagnosis applications. The input and output parameters of an isolation system are successfully utilized in
laboratory equipments to maintain the uncertainty level with in the acceptable range. Parhi [9] has developed a
fuzzy inference based guiding mechanism for multiple robots working in obstacles hindered environment. They
have been designed to navigate in an environment without hitting any obstacles along with other robots. Fox [10]
has analyzed the role of fuzzy logic in medical diagnosis and reveals the presence of various concerns regarding
the information-processing techniques in the development of medical computing. Dash & Parhi [11] have used
the fuzzy logic based techniques to detect the cracks in a cantilever beam of uniform cross section. They have
utilized the dynamic characteristics such as change in natural frequencies and mode shapes as input to the fuzzy
model to predict the crack position and severity, which is subsequently validated by finite element and
experimental methods. Angelov et al. [12] have used the developed fuzzy system for image classification in on
line mode utilizing the approaches cited for improving the realization of on line fuzzy classifier. Zimmermann
[13] has obtained the comparatively effective solution for linear vector maximum problem through fuzzy linear
programming approach for solving linear vector maximum problem. Sugumaran et al. [14] have proposed a
decision tree of a fuzzy classifier for selecting best few FEMture that will differentiate the fault condition of the
bearing from given trained samples. The vibration signal from a piezoelectric transducer is captured for different
types of fault condition of bearing and is used to build the fuzzy rules. The results obtained from the fuzzy
classifier are encouraging when compared with that of the experimental analysis. Kim et al. [15] have come up
with a computer based damage diagnosis system for concrete structures using Fuzzy set theory. They have
applied the enhanced technique to diagnose the damage using the damage attributes as building blocks to model
the fuzzy inference system and the results obtained are very encouraging when compared with the standard ones.
Parhi [16] has developed a fuzzy inference based navigational control system for multiple robots working in a
clumsy environment. They have been designed to navigate in an environment without hitting any obstacles
along with other robots.
From the above literature survey, it has been found that researchers have successfully diagnosed the fault
in the structure using different methodologies. Moreover, fuzzy logic based artificial intelligence techniques
have also been used by some researchers to identify the fault in the structures. In the current work, fuzzy logic
based models have been developed for Al and steel cantilever beams using the Gaussian elements for predicting
the possibilities of faults in the structural members. The vibration responses such as first three relative natural
frequencies and relative mode shape differences have been obtained from the experimental method with
subsequent validation using finite element method and utilized as the input to the fuzzy logic based model to
diagnose the fault in the Al and steel structural members.
2. Finite Element Method
In the current section, Finite element method (FEM) has been performed on the cantilever beam with different
damage characteristics to evaluate the modal parameters, which are subsequently used for damage diagnosis.
The pictorial view of the cantilever beam with damage severity a1 subjected to axial and bending load has been
presented in fig. (1). The relationship between the displacement and the forces can be expressed as;
uj – ui Uj
Covl (1)
θj – θi = Øj
211
R11 -R12
Where overall flexibility matrix Covl can be expressed as, Covl =
-R21 R22
ui (Ui) Lc uj (Uj)
a1
θj (Øj)
θi (Øi)
Le
Fig. 1 Damaged cantilever beam element subjected to axial and bending forces.
The forces acting on the beam element for finite element analysis are shown in Fig. 1.
Where,
R11: Deflection in direction 1 due to load in direction 1
R12= R21: Deflection in direction 1 due to load in direction 2
R22: Deflection in direction 2 due to load in direction 2.
Under this system, the flexibility matrix Cintact of the intact beam element can be expressed as;
uj – ui Uj Le/EA 0
= Cintact and Cintact = 0 Le/EI (2)
Where, θj – θi Øj
The displacement vector in equation (2) is for the intact beam.

The total flexibility matrix Ctot of the damaged beam element can now be obtained by,
Le/EA+ R11 -R12 (3)
Ctot = Cintact + Covl =
-R21 Le/EI+ R22
Through the equilibrium conditions, the stiffness matrix Kc of a damaged beam element can be obtained as [17]
Kc=DCtot-1DT (4)
Where D is the transformation matrix and expressed as;

-1 0
0 -1
D= 1 0
0 1
In the current analysis, the computational analysis software has been used to determine the vibration responses
i.e. the natural frequencies and mode shapes of the damaged and undamaged Al and steel cantilever beam. The
FEM tool follows the methodology for solution of the stiffness matrix Kc as mentioned in equation (4) and
derive the modal response parameters of the intact and damaged beam members. The results from the FEM for
the first three modes of the damaged beam with different damage configurations are compared with that of the
experimental analysis and are presented in Table 1.
3. Experimental Method
The experimental analyses have been carried out to measure the natural frequencies and mode shapes of the
damaged and undamaged Al and steel cantilever beam with different damage characteristics. The schematic
diagram of the experimental set up has been shown in Fig. 2. Experiments have been performed on the damaged
beam structures with different damage locations and damage severities to validate the results obtained from
212
finite element analysis and the proposed Fuzzy logic based model. The modal parameters in terms of natural
frequencies and mode shapes for the first three modes of vibration have been recorded by suitably positioning
the accelerometer at the damaged sites across the length of the Al and steel structural beam members. The
required instrumentation for measuring the vibration parameters of the cantilever beam consists of nine numbers
of elements and they are (a)Vibration exciter, (b)Delta Tron accelerometer, (c)Al and steel cantilever beam
platforms, (d)Vibration Analyzer, (e)Vibration Monitor, (f)Function Generator, (g)Power amplifier, (h)Power
supply, (i) Distribution box.
Fig. 2 Schematic diagram of experimental setup

Out of several observations (relative natural frequencies and relative mode shape differences) obtained from the
experimental analysis for measuring the modal response for different damage configurations of the Al and steel
cantilever beam, nine numbers of data set have been presented in table 1 and table 2 corresponding to nine
different damage configurations. The relative natural frequencies & relative mode shape differences presented in
the table no. 1 and table no. 2 from the several experimental observations are for the three relative damage
positions (0.25, 0.50 and 0.75) and three relative damage severities (0.375, 0.500 and 0.625). For the above said
damage configurations the first three relative natural frequencies and relative mode shape differences derived
from the FEM and experimental analysis have been presented and compared in the table no. 1 and table no. 2.
4. Fuzzy Logic Based Method
The understanding of the input data and the output data for a real application is often vague due to the intricate
dependencies of the input and output variables of the working domain. Fuzzy logic based methodology posses
the approximation FEMtures by the help of fuzzy membership functions and fuzzy IF-THEN rules. In the
process of development of a fuzzy model, the domain knowledge helps in selecting the appropriate membership
functions and development of fuzzy rules. This membership functions are designed by using the suitable fuzzy
linguistic terms and fuzzy rule base. The fuzzy rule base or the conditional statements are used for fuzzification
of the input parameters and defuzzification of the output parameters. The fuzzy model takes the input
parameters from the application at a certain state of condition and using the rules it will provide a controlled
action as desired by the system. In the current research Gaussian fuzzy models have been used for diagnosis of
damage in Al and steel cantilever beam elements.
The inputs to the fuzzy model for damage identification in the current analysis comprises,
Relative first natural frequency = “FNF”; Relative second natural frequency = “SNF”;
Relative third natural frequency = “TNF”; Relative first mode shape difference = “FMD”;
Relative second mode shape difference = “SMD”; Relative third mode shape difference = “TMD”
The linguistic term used for the outputs are as follows;
Relative damage position = “RDP”
Relative damage severity = “RDS”
213
Fig. 3 Fuzzy controller for current analysis
4.1. Analysis of the fuzzy controller used for damage identification
Inputs Outputs
FNF
Fuzzy Model
SNF RDS
TNF
FMD
RDP
SMD
TMD
Fig. 4 Gaussian fuzzy model
Based on the fuzzy subsets, the fuzzy control rules are defined in a general form as follows:
If (FNF is FNFi and SNF is SNFj and TNF is TNFk and FMD is FM D l and SMD is SMD m and TMD is TMD n )
then RDP is RDPijklmn and RDS is RDS ijklmn (5)
Where i=1 to 9, j=1 to 9, k = 1 to 9, l= 1 to 9, m= 1 to 9, n= 1 to 9.
As “FNF”, “SNF”, “TNF”, “FMD”, “SMD”, “TMD” have ten membership functions each. From equation (4.1),
two set of rules can be written as,
If (FNF is FNFi and SNF is SNFj and TNF is TNFk and FMD is FMD l and SMD is SMD m and TMD is TMD n )
then RDS is RDSijklmn
(6)
If (FNF is FNFi and SNF is SNFj and TNF is TNFk and FMD is FMD l and SMD is SMD m and TMD is TMD n )
then RDP is RDPijklmn
According to the usual fuzzy logic control method [11, 16], a factor Wijklmn is defined for the rules as follows:
Wijklmn = µ fnf i (freq i ) Λ µ snf j (freq j ) Λ µ tnf k (freq k ) Λ µ fmd l (moddif l ) Λ µ smd m (moddif m ) Λ µ tmd n ( moddif n )
Where freqi , freqj and freqk are the first, second and third relative natural frequencies of the cantilever beam with
damage respectively ; moddifl, moddifm and moddifn are the average first, second and third relative mode shape
differences of the cantilever beam with damage respectively. By applying the composition rule of inference [11,
16], the membership values of the relative damage position and relative damage severity, (position)RDP and
(severity)RDS can be computed as;
µ RDP (position) = Wijklmn Λ µ RDP (position)
ijklmn ijklmn (7)
µ RDS ( severity ) = Wijklmn Λ µ RDS ( severity )
ijklmn ijklmn
The overall conclusion by combining the outputs of all the fuzzy rules can be written as follows:
214
µ RDP (position) = µ RDP (position) ∨ ... ∨ µ RDP (position) ∨ ...∨ µ RDP (position)
111111 ijklmn 10 10 10 10 10 10 (8)
µ RDS(severity) = µ RDS (severity) ∨ ... ∨ µ RDS (severity) ∨ ....∨ µ RDS (severity)
111111 ijklmn 10 10 10 10 10 10
The crisp values of relative damage position and relative damage severity are evaluated using the centre of
gravity method [16] as:
(position ⋅ µRDP (position) ⋅ d(position)
Re lativedamageposition = RDP = ∫
∫ µRDP(position) ⋅ d(position) (9)
(severity) ⋅ (severity) ⋅ d(severity)
Re lativedamageseverity = RDS = ∫
µ RDS
∫ µRDS(severity) ⋅ d(severity)
Fig. 5(a) Membership functions for relative natural Fig. 5(b) Membership functions for relative natural
frequency for second mode of vibration.
frequency for first mode of vibration.
Fig. 5(c) Membership functions for relative natural Fig. 5(d) Membership functions for relative mode shape
frequency for third mode of vibration. difference for first mode of vibration.
Fig. 5(e) Membership functions for relative mode shape Fig. 5(f) Membership functions for relative mode shape
difference for second mode of vibration. difference for first mode of vibration.
Fig. 5 (g) Membership functions for relative damage severity.
Fig. 5 (h) Membership functions for relative damage position.

215
Fig. 5 (a) - (h) Membership functions for Gaussian fuzzy model.
Table 1 Description of fuzzy linguistic terms.
Membership Functions Linguistic

Definition of the Linguistic terms
Name Terms
Low ranges of relative natural frequency for first mode of
L1F1,L1F2,L1F3 FNF 1 to 3
vibration
Medium ranges of relative natural frequency for first mode of
M1F1,M1F2, M1F3 FNF 4 to 6
vibration
Higher ranges of relative natural frequency for first mode of
H1F1,H1F2,H1F3 FNF 7 to 9
vibration
Low ranges of relative natural frequency for second mode of
L2F1,L2F2,L2F3 SNF 1 to 3
vibration
Medium ranges of relative natural frequency for second mode
M2F1,M2F2,M2F3 SNF 4 to 6
of vibration
Higher ranges of relative natural frequencies for second mode
H2F1,H2F2,H2F3 SNF 7 to 9
of vibration
Low ranges of relative natural frequencies for third mode of
L3F1,L3F2,L3F3 TNF 1 to 3
vibration
Medium ranges of relative natural frequencies for third mode of
M3F1,M3F2,M3F3 TNF 4 to 6
vibration
Higher ranges of relative natural frequencies for third mode of
H3F1,H3F2,H3F3 TNF 7 to 9
vibration
S1M1,S1M2,S1M3 FMD 1 to 3 Small ranges of first relative mode shape difference
M1M1,M1M2,M1M3 FMD 4 to 6 medium ranges of first relative mode shape difference

H1M1,H1M2,H1M3 FMD 7 to 10 Higher ranges of first relative mode shape difference
S2M1,S2M2,S2M3 SMD 1 to 3 Small ranges of second relative mode shape difference
M2M1,M2M2,M2M3 SMD 4 to 6 medium ranges of second relative mode shape difference
H2M1,H2M2,H2M3 SMD 7 to10 Higher ranges of second relative mode shape difference
S3M1,S3M2,S3M3 TMD 1 to 3 Small ranges of third relative mode shape difference
M3M1,M3M2,M3M3 TMD 4 to 6 medium ranges of third relative mode shape difference
H3M1,H3M2,H3M3 TMD 7 to 10 Higher ranges of third relative mode shape difference
SP1,SP2,SP3,SP4 RDP 1 to 4 Small ranges of relative damage position
MP1,MP2,MP3,MP4 RDP 5 to 8 Medium ranges of relative damage position

HP1,HP2, HP3,HP4 RDP 9 to 12 Higher ranges of relative damage position
LS1,LS2,LS3,LS4 RDS 1 to 4 Lower ranges of relative damage severity
MS1,MS2,MS3,MS4 RDS5 to 8 Medium ranges of relative damage severity
US1,US2,US3,US4 RDS 9 to 12 Upper ranges of relative damage severity
216
Table 2 Examples of twenty fuzzy rules to be implemented in fuzzy model.
Sl.
Examples of some rules used in the fuzzy model
No.
If FNF is H1F1, SNF is M2F2, TNF is M3F1, FMD is H1M2, SMD is H2M4, TMD is H3M3,
1
then RDS is LS4 and RDP is SP3.
If FNF is L1F4, SNF is L2F4, TNF is L3F4, FMD is H1M1, SMD is H2M1, TMD is H3M2,
2
If FNF is L1F3, SNF is L2F4, TNF is L3F4, FMD is M1M2, SMD is H2M2, TMD is H3M3,
3
If FNF is H1F2, SNF is H2F1, TNF is H3F1, FMD is H1M3, SMD is H2M4, TMD is H3M4,
4
then RDS is MS2 and RDP is SP3.
If FNF is M1F1, SNF is L2F2, TNF is L3F3, FMD is H1M1, SMD is H2M1, TMD is H3M2,
5
If FNF is L1F1, SNF is L2F2, TNF is L3F3, FMD is H1M3, SMD is M2M1, TMD is H3M4,
6
then RDS is MS2 and RDP is MP1.
If FNF is L1F4, SNF is L2F4, TNF is L3F4, FMD is M1M2, SMD is H2M1, TMD is H3M1,
7
If FNF is H1F1, SNF is M2F2, TNF is M3F1, FMD is H1M2, SMD is H2M2, TMD is H3M2,
8
If FNF is L1F1, SNF is L2F4, TNF is L3F4, FMD is M1M1, SMD is M2M1, TMD is M3M2,
9
then RDS is LS1 and RDP is MP2.
If FNF is M1F1, SNF is L2F2, TNF is L3F1, FMD is M1M2, SMD is M2M2, TMD is H3M1,
10
If FNF is M1F1, SNF is M2F1, TNF is M3F1, FMD is H1M3, SMD is H2M3, TMD is H3M4,
11
If FNF is M1F1, SNF is L2F1, TNF is L3F1, FMD is H1M3, SMD is H2M2, TMD is H3M3,
12
If FNF is M1F2, SNF is M2F1, TNF is M3F1, FMD is M1M1, SMD is H2M1, TMD is H3M2,
13
then RDS is MS2 and RDP isMP2.
If FNF is H1F2, SNF is H2F1, TNF is H3F1, FMD is H1M4, SMD is H2M1, TMD is H3M1,
14
If FNF is M1F1, SNF is L2F1, TNF is L3F2, FMD is S1M1, SMD is S2M2, TMD is H3M1,
15
If FNF is L1F4, SNF is L2F4, TNF is L3F4, FMD is H1M2, SMD is S2M1, TMD is H3M2,
16
If FNF is M1F1, SNF is L2F3, TNF is L3F1, FMD is S1M2, SMD is M2M1, TMD is S3M1,
17
If FNF is L1F1, SNF is L2F1, TNF is L3F1, FMD is H1M2, SMD is H2M2, TMD is H3M2,
18
If FNF is H1F2, SNF is H2F1, TNF is H3F1, FMD is S1M2, SMD is H2M3, TMD is H3M1,
19
If FNF is L1F3, SNF is L2F4, TNF is L3F4, FMD is S1M3, SMD is S2M2, TMD is S3M3, then
20
RDS is LS3 and RDP is SP3.
217
Inputs
Rule no 5 of Table 2 is effectuated Rule no 15 of Table 2 is effectuated
Output
0.1703 Relative damage severity
0.3614 Relative damage position

Fig. 6 Resultant values of relative damage severity and relative damage position from Guassian
fuzzy model when Rules 5 and 15 of Table 2 are effectuated.
The Gaussian fuzzy model has been used to identify the relative damage severity and relative damage
position of Al as shown in fig. 6. The outputs have been generated by the execution of rules 5 and 15 of table
no. 2. The same procedures have been adopted to determine the relative damage severity and relative damage
position of steel. The outputs from the FEM, experimental and Fuzzy logic model for Al and steel have been
presented and compared in table no. 3 and table no. 4 respectively.
218
Table 3 Comparison of damage characteristics of Al cantilever beam obtained from Fuzzy logic, FEM
and Experimental method
Fuzzy
FEM Experimental
FNF SNF TNF FMD SNF TNF Gaussian
RDS RDP RDS RDP RDS RDP
0.997 0.998 0.999 0.007 0.003 0.37 0.37
0.0021 0.50 0.49 0.375 0.50
9 6 2 5 7 4 5
0.995 0.988 0.998 0.001 0.003 0.48
0.0043 0499 0.49 0.48 0.500 0.50
6 5 6 1 6 9
0.991 0.970 0.998 0.014 0.009 0.62 0.62
0.0083 0.49 0.48 0.625 0.50
0 2 3 0 7 4 4
0.992 0.996 0.994 0.030 0.016 0.37 0.37
0.0016 0.26 0.27 0.375 0.25
4 4 4 1 1 6 6
0.983 0.991 0.988 0.009 0.037 0.50 0.49
0.0039 0.26 0.27 0.500 0.25
7 8 0 4 3 1 6
0.966 0.984 0.975 0.014 0.029 0.62 0.62
0.0089 0.25 0.26 0.625 0.25
1 1 2 7 3 6 4
0.999 0.999 0.990 0.068 0.043 0.37 0.37
0.0169 0.76 0.74 0.375 0.75
9 9 8 2 5 4 1
0.999 0.999 0.980 0.042 0.071 0.49 0.49
0.0359 0.76 0.72 0.500 0.75
8 6 5 9 1 9 7
0.999 0.993 0.960 0.046 0.031 0.62 0.62
0.0724 0.74 0.73 0.625 0.75
5 1 4 5 3 6 2
Table 4 Comparison of damage characteristics of Steel cantilever beam obtained from Fuzzy logic, FEM and
Experimental method
Fuzzy Experimenta
FEM
FNF SNF TNF FMD SNF TNF Gaussian l
RDS RDP RDS RDP RDS RDP
0.997 0.996 0.999 0.006 0.004 0.37 0.37
0.0021 0.374 0.49 0.49 0.50
9 9 6 9 4 4 5
0.995 0.986 0.999 0.001 0.004 0.49 0.50
0.0043 0.501 0.49 0.49 0.50
5 6 2 9 5 9 0
0.990 0.967 0.998 0.013 0.010 0.62 0.62
0.0081 0.624 0.51 0.48 0.50
8 6 8 8 5 4 5
0.992 0.996 0.994 0.028 0.017 0.37 0.37
0.0018 0.376 0.25 0.27 0.25
1 3 1 8 9 3 5
0.983 0.991 0.987 0.012 0.034 0.49 0.50
0.0041 0.499 0.24 0.24 0.25
3 8 5 9 1 8 0
0.965 0.984 0.974 0.012 0.027 0.62 0.62
0.0094 0.626 0.24 0.26 0.25
3 2 4 2 4 3 5
0.999 0.999 0.990 0.060 0.046 0.37 0.37
0.0174 0.374 0.74 0.76 0.75
9 9 5 7 1 7 5
0.999 0.998 0.979 0.047 0.069 0.50 0.50
0.0099 0.499 0.74 0.73 0.75
8 9 8 7 3 1 0
0.999 0.991 0.959 0.041 0.037 0.62 0.62
0.0743 0.624 0.76 0.74 0.75
4 3 3 9 5 3 5
219
This section of the paper list down the discussion in regard to the results obtained from the FEM, experimental
analysis and Fuzzy logic model. The FE model describes the required procedures from equation no. 1 to 4 to
compute the vibration parameters (relative natural frequency and relative mode shape difference) from the
damaged model subjected to axial and bending load as shown in fig. (1). The experimental setup for vibration
analysis has been shown in fig. (2). The experimental section describes the complete instrumentation used to
derive the relative first three natural frequencies and relative first three mode shape differences of the damaged
composite cantilever beam with different damage scenarios. The results from both FEM and experimental
analysis for Al and steel have been compared and shown in table no. 1 and table no. 2 and a close agreement has
been found between the results. The Fuzzy logic based controller with six inputs i.e. first three relative natural
frequencies and relative mode shape differences and two outputs i.e. relative damage severity and relative
damage position have been shown in fig. (3). Fig. 4 shows the Gaussian membership function which has been
used in the current research. Fig. 5 illustrates the Gaussian membership functions as inputs and outputs with
suitable range definition. The resultant values of relative damage severity and relative damage position have
been shown in fig. 6. The table (1) presents the description of fuzzy linguistic terms and table (2) presents the
twenty examples of fuzzy rules which have been implemented in the proposed model. The table (3) and table (4)
presents a comparison of results derived from the FEM, experimental and the Fuzzy logic based model used for
damage diagnosis in the Al and steel cantilever beam models. From the analysis of results from table (3 and
table (4), it is evident that the fuzzy logic based model gives results in close proximity to actual as compared to
the finite element analysis.
6. CONCLUSIONS
The conclusions drawn by analyzing the results from the above mentioned sections are depicted below.
The deviations in the mode shapes are quite significant at the vicinity of the damaged site present on the Al and
steel cantilever beam which acts as the basis not only to identify the presence of damage but also, the position
and severity of the damage. The finite element analysis has been carried out for computing the vibration
parameters of the damaged and intact Al and steel cantilever beams at the first three modes of vibration. The
major deviations in the mode shapes have been observed at the third mode of vibration. The experimental results
are found to be in close agreement with the finite element results. The developed Fuzzy logic based model with
Gaussian membership function has predicted the damage characteristics present in the Al and steel cantilever
beams effectively and the results are found to be close to the experimental results. Hence, the Fuzzy logic based
model can be used as a robust tool for diagnosing the structural integrity of engineering structures.
REFERENCES
[1] H. Nahvi and M. Jabbari, “Crack detection in beams using experimental modal data and finite element
mode”, International Journal of Mechanical Sciences, 2005, Vol. 47(10), pp. 1477-1497.
[2] A. Salam, Y. Alsabbagh, O.M. Abuzeid and Mohammad H. Dado, “Simplified stress correction factor to
study the dynamic behavior of a cracked beam”, Applied Mathematical Modelling, 2007, Vol. 33, pp.
127–139.
[3] S.M. Murigendrappa, S.K. Maiti and H.R. Srirangarajan, “Experimental and theoretical study on crack
detection in pipes filled with fluid”, Journal of Sound and Vibration, 2004, Vol. 270, pp.1013–1032.
[4] G. M. Owolabi, A. S. J. Swamidas and R. Seshadri, “Crack detection in beams using changes in frequencies
and amplitudes of frequency response functions”, Journal of Sound and Vibration, 2003, Vol. 265, pp.1-22.
[5] G.D. Gounaris and C.A. PapadoPoulos, “Analytical and experimental crack identification of beam structures
in air or in fluid”, Computers and Structures, 1997, Vol. 65 (5), pp.633-639.
[6] K.Wang, J.D.Inman and R.F.Charles, “Modeling and analysis of a cracked composite cantilever beam
vibrating in coupled bending and torsion”, Journal of sound and vibration, 2004, Vol. 284 (1-2), pp.23-49.
220
[7] T. Boutros and M. Liang, “Mechanical fault detection using fuzzy index fusion”, International Journal of
Machine Tools & Manufacture, 2007, Vol. 47, pp. 1702–1714.
[8] L. J. de Miguel and L. F. Blazquez, “Fuzzy logic-based decision-making for fault diagnosis in a DC motor”,
Engineering Applications of Artificial Intelligence, 2005 Vol. 18, pp.423–450.
[9] D. R. Parhi, “Navigation of mobile robot using a fuzzy logic model”, Journal of Intelligent and Robotic
Systems, Theory and Applications, 2005, Vol. 42, pp. 253-273.
[10] J. Fox, “Some observations on fuzzy diagnosis and medical computing”, International Journal of
Bio-Medical Computing, 1977, Vol. 8, pp.269–275.
[11] A. K. Dash, D.R. Parhi, “Development of an inverse methodology for crack diagnosis using AI technique”,
International Journal of Computational Materials Science and Surface Engineering (IJCMSSE), 2011, Vol.
4(2), pp.143-167.
[12] P. Angelov, E. Lughofer, X. Zhou, “Evolving fuzzy classifiers using different model architectures”, Fuzzy
Sets and Systems, 2008, Vol. 159, pp.3160–3182.
[13] H.J. Zimmermann, “Fuzzy programming and linear programming with several objective functions”, Fuzzy
Sets and Systems, 1978, Vol. 1, pp.45–55.
[14] V. Sugumaran and K.I. Ramachandran, “Fault diagnosis of roller bearing using fuzzy classifier and
histogram FEMtures with focus on automatic rule learning”, Expert Systems with Applications, 2011, Vol.
38(5), pp.4901-4907.
[15] Y. M. Kim, C.K. Kim and G. H. Hong, “Fuzzy set based crack diagnosis system for reinforced concrete
structures”, Computers and Structures, 2007, Vol. 85, pp.1828–1844.
[16] D. R. Parhi, “Navigation of mobile robot using a fuzzy logic model”, Journal of Intelligent and Robotic
Systems, Theory and Applications, 2005, Vol. 42, pp. 253-273.
[17] M. H. Dado, “A Comprehensive Crack Identification Algorithm for Beams Under Different End
Conditions”, Applied Acoustics, 1997, Vol. 51, pp. 381-398.
221
Paper No.: 037
STRENGTHENING OF COLUMNS FOR SHEAR IN REINFORCED

CONCRETE BUILDINGS
Komathi M.1 and Amlan K.Sengupta 2

1
Ph.D. Student and 2Associate Professor
Dept. of Civil Engineering, Indian Institute of Technology Madras, Chennai, India
Email: komathimurugan@gmail.com, amlan@iitm.ac.in
ABSTRACT:
The columns in the open ground storeys of reinforced concrete multi-storeyed buildings performed poorly
during the Bhuj earthquake in 2001. Such columns were predominantly subjected to brittle shear failure,
causing collapse of the storeys in several cases. To strengthen such existing buildings, in addition to global
retrofit measures, the deficient columns need to be strengthened by local retrofitting technique, such as concrete
jacketing.
The present research investigates the improvement of the capacity and behaviour of columns retrofitted by
concrete jacketing, specifically for shear. The paper presents a review of the literature on modelling of
strength and behaviour of columns under shear, strengthening of columns by concrete jacketing, and aspects of
the interface behaviour between old and new concrete for jacketing. Results of a numerical analysis of a
typical model mid-rise building are presented to study the effect of shear deficiencies of columns in the open
ground storey. Two conditions of the ties of the columns were investigated: adequately designed and
inadequately designed for shear under gravity loads. From the pushover analyses, it is demonstrated that with
deficient ties, brittle shear hinges form in the columns which lead to a sudden failure of the building. An
experimental programme is under progress to study the performance of shear-critical columns, before and after
retrofitting with concrete jacketing.
KEYWORDS:
Building, Column, Concrete jacket, Reinforced concrete, Retrofit, Shear
1. INTRODUCTION
The earthquakes in the recent decades have exposed the vulnerability of multi-storeyed reinforced concrete
buildings in India (Jabalpur, 1997; Bhuj, 2001; Andaman islands, 2004; Sikkim, 2011). Among the different
types of failures observed during the earthquakes, the failure due to damage of columns is more significant.
There can be various types of deficiencies of the columns. Here, the observations are limited only to the shear
deficiencies. Columns with reduced height-to-depth ratio in an open ground storey for parking of cars,
columns adjacent to partial height infill walls or window openings, shorter columns in the ground storey with
sloping foundation profile, columns adjacent to split levels, are examples of shear-critical columns. In absence
of adequate amount, spacing and anchorage of the ties, shear failure of the columns is brittle and occurs
throughout the depth of the section (Figure 1).
The vulnerability of existing deficient buildings can be reduced by seismic retrofitting. The retrofit techniques
aim to improve the capacity and performance of a structure, as well as its components. They can be broadly
classified as local or global strategies. One of the local methods to improve the flexural and shear capacities of
a column is jacketing. The jacketing can be done using additional reinforcing bars and concrete, ferrocement,
steel sections and plates, or fibre-reinforced polymer composites. The present study investigates the execution
and effect of concrete jacketing, and the modelling of a jacketed column in a numerical analysis of a building.
The paper provides a brief review of several aspects of shear behaviour of columns, including strengthening by
concrete jacketing, and an analysis of a typical building showing the consequence of shear failure of columns.
Figure 1 Shear failure of a column in an open ground storey during the Bhuj earthquake
(Murty et al., 2002)
2. STRENGTHENING OF COLUMNS BY CONCRETE JACKETING
Concrete jacketing of an existing column involves preparation of the concrete surface, fabrication and anchoring
of additional longitudinal bars and transverse ties, and casting a layer of concrete. The interface between the old
and new concrete should be effective in transferring the shear stresses. The additional longitudinal bars should
be anchored to the foundation at the base and should continue through the floor slabs above, in order to enhance
the flexural capacity of the column. The ties are to be provided preferably with 135o hooks, with adequate
extensions at their ends. The new concrete should be workable enough to be cast in a narrow space, or added as
shotcrete.
The advantages and drawbacks in adopting this technique are listed in the following table (Handbook on Seismic
Retrofit of Buildings).
Table 1 Advantages and drawbacks of concrete jacketing

Advantages Drawbacks
• Enhancement of flexural and shear capacities, • Increase in column size and decrease in carpet
and increase in ductility of a column area of a floor
• Distributed increase in lateral stiffness of a • Involves drilling of existing concrete for
storey with jacketing of several columns placement of longitudinal bars through slabs, and
• Improvement in durability of the existing ties at the joints
columns, avoidance of corrosion and fire • Disruption to the users of the building
protection needs
• Tend to be cost-effective
3. LITERATURE REVIEW
To model the strength and behaviour of a jacketed column, it is necessary to study the modelling of reinforced
concrete columns. The review also covers the studies on jacketed columns, and the interface between existing
concrete and the jacket layer.
223
3.1. Shear Strength of Columns
The shear strength of a column under increasing lateral load was expressed as a sum of concrete contribution, tie
contribution based on a truss mechanism, and a contribution of the axial load based on a strut mechanism
(Priestley et al., 1994). The concrete contribution depends on the flexural deformation. In the model
proposed by Sezen and Moehle (2004), the contribution of concrete includes the effect of axial load, and both
the concrete and tie contributions are reduced with flexural ductility demand.
3.2. Shear Behaviour of Columns
The lateral stiffness of a column after cracking can be studied based on a truss analogy (Park and Paulay, 1975).
The Modified Compression Field Theory (MCFT) was developed to predict the post-cracking non-linear shear
deformation of wall type members, based on two dimensional equilibrium and compatibility conditions for the
concrete struts and rebar ties, and respective constitutive relationships (Vecchio and Collins, 1986). The
MCFT has been used to predict the deformation of a column due to flexure and shear, under an increasing lateral
load and in presence of axial load (Mostafaei and Vecchio, 2008). The analysis is based on the average strains
in the strut between the end (section of maximum moment) and the mid-height (inflection point) of the column.
A model relating the shear capacity to the lateral drift at shear failure was developed by Elwood and Moehle
(2005). A piece-wise linear lateral load versus shear deformation model, connecting the points of cracking,
maximum load and onset of degradation was developed by Sezen (2008). Fibre-based modelling of the
column section under cyclic lateral load has been extended to include shear deformation (Ceresa et al., 2009;
Mullapudi and Ayoub, 2013).
3.3. Performance of Jacketed Columns
Short column specimens deficient in shear and strengthened by jacketing, were tested under double-curvature
loading by Bett et al. (1988). The jacketed specimens were found to exhibit increase in lateral stiffness and
strength. Column-joint sub-assemblage specimens strengthened for flexure and tested under cyclic loading,
behaved ductile with increased strength and reduced degradation of strength (Rodriguez and Park, 1994).
Stoppenhagen et al. (1995) investigated the effect of jacketing damaged columns in a column-spandrel-beam
frame. The shear failure of columns could be avoided after jacketing. Effect of concrete jacketing using
three procedures such as welding of the ends of a tie, inclusion of L-shaped dowel bars in the existing column,
and bent-down bars connecting the longitudinal reinforcement of the existing column and those of the jacket,
were investigated by Vandoros and Dritsos (2008). When the tie ends were welded, then there was no effect of
the additional connections between the existing concrete and the jacket concrete, on the initial stiffness and
capacity of a specimen. However, the specimen with bent-down bars showed more energy dissipation.
3.4. Behaviour of Concrete-Jacket Interface
The interface between concrete in the original column and the jacket concrete should be such that both of them
act monolithically. The shear transfer between the concrete surfaces is an important issue since it can lead to
delamination of the jacket. The options of improving shear transfer are roughening of the surface of the
original column, use of bonding agent, and use of dowel bars or bent-down bars.
From cyclic tests of push-off specimens, it was observed that the amount and embedment depth of interface
reinforcement improved the shear capacity of the interface even at large levels of slip (Bass et al., 1989).
Investigation of jacketed column specimens with interfaces prepared by surface roughening, applying bonding
agent, or steel connectors, showed that satisfactory monolithic behaviour can be achieved even without these
methods (Julio et al., 2005). The observation was of course restricted to columns with high bending
moment-to-shear ratio. Ratio of key parameters such as yield moment, yield drift etc. of a jacketed column to
224
that of a monolithic column, were noted to study the effect of different types of interface preparation (Bousias et
al., 2007). These were not found to be significantly affected by the diverse methods of interface preparation.
4. NUMERICAL ANALYSIS
4.1. Building Data
A typical model mid-rise building with an open ground storey was analysed to study the effect of shear failure
of columns. Some basic information of the building is provided in Table 2. Figure 2 shows the plan and
elevation of the building.
Table 2 Building data

Plan dimension 14m × 21m
Building layout Number of storeys G+6
Type of foundation Raft
Beam size (at all levels) 200mm × 300mm
Column size (in all storeys) 230mm × 350mm
Sizes of members Slab thickness (at all levels) 125mm
Thickness of brick infill External: 230mm

walls Internal: 120mm
Grade of concrete M20
Materials
Grade of steel Fe415
Seismic zone V
Importance factor I 1
Seismic parameters as
per IS 1893: 2002 3.0 for ordinary moment
Response reduction factor R
resisting frame
Type of soil Medium (Type II)
A reduced height of the open ground storey was considered for car parking, to limit the height-to-depth ratio of
the columns. The sizes of slabs, beams and columns were selected to satisfy conventional gravity load design
criteria. The building was modelled and analysed using SAP2000. The beams and columns were modelled
using frame elements. Considering the raft foundation, the base of a ground storey column was assumed to be
fixed. A slab was not modelled, but the dead and live loads were distributed on the supporting beams as
triangular and trapezoidal loads. Considering a slab to act as a rigid horizontal diaphragm under seismic
loading, a diaphragm constraint was assigned at each floor level. An infill panel was modelled as a single
equivalent diagonal compression strut. The dead and live loads were considered as per conventional gravity
load design. The seismic loads were calculated based on IS 1893 (Part 1): 2002.
225
a) Plan at foundation level b) Side elevation
Figure 2 Plan and elevation of building
4.2. Linear Analysis for Gravity and Lateral Loads
Linear analyses were performed for the gravity and lateral loads to obtain the member forces for various loads
cases and their combinations. Figure 3 shows the high the moment and shear forces in the ground storey
columns, under lateral loads. The reinforcement in the frame members was designed based on the forces due
to gravity loads only.
a) Shear force diagram b) Bending moment diagram

Figure 3 Typical member force diagrams from lateral load analysis
226
4.3. Modelling for Static Non-Linear Analysis
Pushover analyses were done on the building model to study the sequence of failure of the members. The
non-linear behaviour of the beams, columns and equivalent struts were modelled by assigning point plastic
hinges at suitable locations. The moment hinge properties for the columns and beams were calculated based
on the values of yield and ultimate moments, and the hinges were assigned at the ends of the members. The
shear hinge properties were calculated based on the truss analogy, and the hinges were assigned at mid-height
locations of the columns, and at the ends of the beams. The axial hinge properties for the equivalent struts
were calculated based on a parabolic-plastic load-deformation relationship, and the hinges were assigned at the
middle of the struts.
The shear hinges for the ground storey columns were calculated for two conditions of the ties. First, the ties
were considered to be adequately designed at least for gravity loads, with adequate anchorage to achieve the
design yield stress (referred to as strong in shear). Second, the ties were considered to be inadequately
designed even for gravity loads, with deficient anchorage (referred to as weak in shear). The hinge properties for
one typical ground storey column are shown in Figure 4. For each of the two orthogonal horizontal directions,
a gravity pushover analysis was followed by a displacement controlled lateral pushover analysis, with a
parabolic distribution of the lateral loads over the height of the building.
(a) Flexural hinge b) Shear hinges

Figure 4 Hinge properties for a typical ground storey column under service axial load
4.4. Results
a) Formation of shear hinges b) Formation of flexural hinges

Figure 5 Failure of ground storey columns for push along the direction of frames
227
The primary results obtained from the pushover analyses were the pushover curves and the status of the hinges.
It was found that if the ties in the ground storey columns are inadequate even for gravity loads, it leads to the
formation of brittle shear hinges (Figure 5a) with imminent collapse of the structure (Figure 6). If the ties are
adequately designed for gravity loads, then it leads to the formation of flexural hinges (Figure 5b), with
increased lateral load resistance and deformability (Figure 6). Of course, the sway mechanism in the ground
storey still persists.
Figure 6 Base shear versus roof displacement curves showing the influence of adequacy of
shear capacity of columns
4.5. Proposed Experimental Study
An experimental programme is under progress to study the performance of shear-critical columns, before and
after retrofitting with concrete jacketing. Reference and strengthened column specimens in column-foundation
sub-assemblages will be tested under monotonic and cyclic lateral loads. Several practical aspects are
considered in the selection and preparation of the specimens: application of the lateral load with suitable
boundary conditions, application of vertical load, mode of expected failure of the specimens, preparation of the
surface of a specimen to be retrofitted, anchorage of the additional longitudinal bars at the foundation of a
specimen, anchorage of the additional ties, casting of the jacket concrete, etc.
5. CLOSURE
Various aspects of strengthening of columns for shear by concrete jacketing are presented in the paper. The
literature review covers modelling of strength and behaviour of columns, previous investigations on concrete
jacketing of columns, and interface between old and new concrete. The numerical analysis demonstrates the
effect of shear failure in the columns of the open ground storey of a typical mid-rise building. The results from
the pushover analysis show that when the columns fail in shear, the failure of the building is sudden, as
compared to that caused by the columns failing in flexure. The ongoing experimental study aims to investigate
the effect of concrete jacketing on the behaviour of such shear-critical columns.
REFERENCES
[1] Bass, R.A.; Carrasquillo, R.L. and Jirsa, J.O. (1989). Shear Transfer Across New and Existing Concrete
Interfaces. ACI Structural Journal, 86 (4), 383-393.
228
[2] Bett, B.J.; Klingner, R.E. and Jirsa, J.O. (1988). Lateral Load Response of Strengthened and Repaired
Reinforced Concrete Columns. ACI Structural Journal, 85 (5), 499-508.
[3] Bousias, S.N.; Biskinis, D.; Fardis, M.N. and Spathis, A.L. (2007). Strength, Stiffness and Cyclic Deformation
Capacity of Concrete Jacketed Members. ACI Structural Journal, 104 (5), 521-531.
[4] Ceresa, P.; Petrini, L.; Pinho, R. And Sousa, R. (2009). A Fibre Flexure-Shear Model for Seismic Analysis of
RC-Framed Structures. Earthquake Engineering and Structural Dynamics, 38 (5), 565-586.
[5] Elwood, K.J. and Moehle, J.P. (2005). Drift Capacity of Reinforced Concrete columns with Light Transverse
Reinforcement. Earthquake Spectra, 21 (1), 71-89.
[6] Handbook on Seismic Retrofit of Buildings (2008). Editors: Chakrabarti, A.; Menon, D. and Sengupta, A.K.
Narosa Publishing House.
[7] IS 1893 (Part 1): 2002. Indian Standard Criteria for Earthquake Resistant Design of Structures - Part 1:
General Provisions and Buildings, Bureau of Indian Standards, New Delhi.
[8] Julio, E.N.B.S.; Branco, F.A.B. and Silva, V.D. (2005). Reinforced Concrete Jacketing - Interface Influence on
Monotonic Loading Response. ACI Structural Journal, 102 (2), 252-257.
[9] Mostafaei, H. and Vecchio, F.J. (2008). Uniaxial Shear-Flexure Model for Reinforced Concrete Elements.
Journal of Structural Engineering, 134 (9), 1538-1547.
[10] Mullapudi, T.R.S. and Ayoub, A. (2013). Analysis of Reinforced Concrete Columns Subjected to Combined
Axial, Flexure, Shear and Torsion Loads. Journal of Structural Engineering, 139 (4), 561-573.
[11] Murty, C.V.R.; Goel, R.K. and Goyal, A. (2002).Reinforced Concrete Structures. Earthquake Spectra, 18
(A), 149-185.
[12] Park, R. and Paulay, T. (1975). Reinforced Concrete Structures, John Wiley & Sons Inc., New York.
[13] Priestley, M.J.N.; Verma, R. and Xiao, Y. (1994). Seismic Shear Strength of Reinforced Concrete Columns.
Journal of Structural Engineering, 120 (8), 2310-2329.
[14] Rodriguez, M. and Park, R. (1994). Seismic Load tests on Reinforced Concrete Columns Strengthened by
Jacketing. ACI Structural Journal, 91 (2), 150-159.
[15] Sezen, H. (2008). Shear deformation Model for Reinforced Concrete Columns. Structural Engineering and
Mechanics, 28 (1), 39-52.
[16] Sezen, H. and Moehle, J.P. (2004). Shear Strength Model for Lightly Reinforced Concrete Columns. Journal
of Structural Engineering, 130 (11), 1692-1703.
[17] Stoppenhagen, D.R., Jirsa, J.O. and Wyllie, L.A. (1995). Seismic Repair and Strengthening of a Severely
Damaged Concrete Frame. ACI Structural Journal, 92(2), 177-187.
[18] Vandoros, K.G. and Dritsos, S.E. (2008). Concrete Jacket Construction Detail Effectiveness when
Strengthening RC Columns. Construction and Building Materials, 22 (3), 264-276.
[19] Vecchio, F.J. and Collins, M.P. (1986). The Modified Compression-Field Theory for Reinforced Concrete
Elements Subjected to Shear. ACI Structural Journal, 83 (2), 219-231.
229
Paper No.: 038
DYNAMIC ANALYSIS OF CANTILEVER BEAM WITH MOVING MASS.

1 2
Shakti.P.Jena and Dayal.R. Parhi
Dept of Mech Engg, NIT, Rourkela, Odisha, India
ABSTRACT
The dynamic actions of the structures under the influence of moving loads is a significant subjects of great
engineering applications such as in the field of transportation and in the intend of machining process In the
present study; the behavior of a structure containing a mass travelling on a cantilever beam is considered to
describe the response of the beam. The mass is brought to move by an applied external force. The consequence
of magnitude and velocity of the moving mass on the dynamic actions of the consequent structure has been
explored theoretically and computationally with various boundary states. By the use of continuum mechanics,
the differential equation have been developed for the structures which are consequently solved by fourth order
Runge-Kutta method to observe the dynamic deflection of the structures.
Keywords: cantilever beam, moving mass, dynamic deflection, Runge-Kutta Method.
1. INTRODUCTION
In the era of transport, structures are subjected to moving masses or loads. As comparison to other types of
dynamic loading, the moving load differs in its location. Due to this, the moving load dilemma becomes an
interesting subject in the study of structural dynamics. The dynamic analysis of structures under moving loads
has been the subject of numerous researches over the years, with an increased interest in the last decade, mainly
due to development of new high-speed vehicles and common drive to upgrade the existing infrastructure.
Jeffcott [1] has first described a technique to find the inexact solution for the problem of vibration under the act
of moving and fluctuating loads. Stanisic et al. [2] developed a hypothesis to calculate the response of a plate to
a multi-mass moving structure by Fourier transformation technique. The analysis shows that for the same
natural frequency of the plate, the resonance is reached prior by considering the moving multi-masses system
than by moving multi-force system. Stanisic and Hardin [3] have examined to find out the response of a simple
supported beam subjected to random no of moving masses by Fourier technique in the presence of external load.
Stansic and Euler [4] have developed an article for analyzing the dynamic nature of the structures with moving
masses by operational calculus method. Akin and Mofid [5] have developed an analytical-numerical method to
compute the dynamic nature of beams subjected to moving mass with arbitrary boundary states by and
examined the result with FEM. They have used the Continuous System Modeling Programming (CSMP) to
solve the system of ordinary differential equations. Olsson [6] has investigated the dynamic reaction of a simply
supported beam with a continuous traversing load at regular speed. Parhi and Behera [7] have examined the
dynamic deflection of a cracked cantilever subjected to moving mass by Runge-Kutta technique taking it
account the local stiffness matrix.
Michaltsos et al. [8] have examined the dynamic response of a simply supported regular beam along with
constant traversing mass & speed by considering the consequence of the mass. Kwon et al. [9] have studied the
characteristics of vibration control of bridges along with traversing loads by utilizing a tuned mass damper
(TMD).They have established a TMD to decrease the reaction of the bridges produced by the traversing loads.
Mahmoud [10] has investigated a system to calculate the stress intensity factor (SIF) for distinct & dual edge
cracks in structures with traversing mass. Mahmoud [11] has analyzed the consequence of cracks on the
dynamic character of a simple undamped Euler-Bernoulli’s beam. Yoon and Son [12, 13] explored the
consequence of cracks on the dynamic character of the cracked pipe carrying fluid with traversing mass and tip
mass. Lin and Chang [14] have investigated a methodical technique to determine the response of a cracked
cantilever beam along with a intense traversing load by using transfer matrix method. Yoon et al. [15] have
examined the influence of the location and deepness of every crack on the natural frequency of a simply
supported double crack beam. Yang et al. [16] have examined the free and forced vibration of FGM-Euler
Bernoulli’s beam having open edging cracks with axial load and a intense traversing mass by utilizing revolving
spring & modal expansion technique. Lin [17] has investigated the dynamic reaction of a cracked
simply-supported structure with a traversing load. Al-Said [18] has investigated a crack detection algorithm to
detect the crack size & position in a stepped cantilever beam with intense slowly traversing loads. Pala and Reis
[19,20] have analyzed the dynamic study of simply supported structure with crack along with traversing load by
considering the consequence of centripetal & Coriolis’s forces. Siringoringo and Fuzino [21] have examined the
bridge fundamental frequency by utilizing the reaction of a passage vehicle ultimately. Han et al. [22] have
analyzed the dynamic character of a geared rotor structure with a slant crack shaft and examined the
consequences of slant crack on rotating characteristics, parametric unsteadiness for the structures under
unbalanced forces. Azam et.al [23] have studied the dynamic response of a Timoshenko beam for the case of
moving sprung mass system and compared that with moving mass and moving load systems.
In the current study, the dynamic reaction of an untracked cantilever beam carrying a moving has been
analyzed .The consequences of mass and speed on the reaction of the beam have been investigated analytically
and numerically by 4th order Runge-Kutta technique.
2.THEORETICAL ANALYSIS
v
β z
Dynamic analysis of the structure with moving mass has been analyzed below in details. Let us consider an
uniform cantilever beam of length ‘L’ , beam mass ‘m’ per unit length subjected to a traversing mass ‘M’ with
velocity ′ ′. Basing upon the postulations of constant inertia, considering the damping effect, rotary and
shearing effects The principal equation of motion of the beam with moving mass is given by-
( , ) ( , )
+ = (x,t) (1)
Where P(x,t)= Applied external force = ( ) ( − ) + ( ) (2)
=Dirac delta function, ( ) = General loading (zero in this case)
( ) =Force due to the moving mass

= − ( + ) (!, ); ! = (3)
0, ∀ ≠ $
( − $ ) = % )*+ *) ℎ+- +- +../0*1 +2/* 34) = $ 5
= +∞
So now 6$ 3( ) ( − $ )1 = 3( $ ), 0 < $ < ∞.
∞
So here, the above equation can written as

8
6$ 3( ) ( − $ )1 = 3( $ ) , 0 < $ < 2 (4)
8
And 6$ 3( ) ( − $ )1 = 0, $ < 0, $ >2 (5)
231
( , ) ( , )
Now from eqn (1) + =[ − ; +
< (!, )].This eqn can be reformed as
( , ) ( , )
+ =[ − ; + < (!, ) ( − !) (6)
Where ! = =The distance travelled by the mass from the fixed end .
Considering the solution of the eqn (6) in series form ;
=( , ) = ∑@ ?AB =? ( )C? ( ) (7)
Dℎ*)* =? ( ) = E *- +2/*1 Function of the beam (without-M, with the same boundary states at the
- F node. To get =? ( ) ; the eqn will be
=? GH ( ) − I?GH ( ) = 0 (8)
K
Jℎ*)* I?GH = L
MN
The general solution of equation (8) can be written as :
=? ( ) = [a cosh(I? ) + T .E-ℎ(I? )U + [V V4.(I? ) + 1 .E-(I? )U (9 )
Where a, b,c & d are constant coefficients depending upon the boundary condition.
Tn(t) is the function of time to be calculated.
The R.H.S of the equation (6) can be rewritten as:
X − ; + < (!, )Y ( − !) = ∑∞
?AB =? ( )Z? ( ) (10)
n n n
Putting the value of eq (7) in eq (10) ;Now eq (10) becomes
X − ; + < ∑∞
?AB =? (!)C? ( )Y ( − !) = ∑?AB =? ( )Z? ( )
∞
(11)
Multiplying both sides of the eq (11) by =[ ( ) and integrating over the entire beam length, the eq (11) now
n n
becomes:
^ ^ ∞
] ]
∞
\[ − ( + )² _ =? (!)C? ( )U ( − !)=[ ( ) = \ _ =? ( )Z? ( )=[ ( )1

] ]!
$ ?AB $ ?AB
^ ^
] ]
∞
\ =[ ( ) ( − !)1 − _ \ =[ ( ) à + b =? (!)C? ( )c ( − !)1

] ]!
$ ?AB $
∞ ^
= _ \ =[ ( )=? ( )Z? ( )1 (12)

?AB $
So by the condition of Dirac-Delta function from eqn (4) and (5)
^
6$ =[ ( ) ( − !)1 = ( )=[ (!) (13)
^
∑∞
?AB 6$ =[ ( ) f; + < =? (!)C? ( )g ( − !)=[ ( ) =
^
∑∞
?AB( + )² =? (!)C? ( ) 6$ =[ ( ) ( − !)1
= ∑∞
?AB( + )²=? (!)C? ( )=[ (!)U (14)
^ 8
The right hand side of eqn (12) )=? ( )Z? ( )1 = ∑∞ ?AB 6$ =[ (
?AB Z? (
∑∞ ) =? ( )1 ) 6$ =h (
Now from the orthogonality principle and orthogonal properties, the above expression can be rewritten as :
^ ^ ^
ZB ( ) \ =[ ( )=B ( )1 + Z ( ) \ =[ ( )= ( )1 … … . . +Z[ ( ) \ =[ ( )=? ( )1 + ⋯

$ $ k
^
= Z[ ( ) 6$ =[ ( )=? ( )1 [Cℎ* 24J*) *)0. +-E.ℎ*.U
Z[ ( )m[ = m[ Z[ ( ) (15)
^ 0 , - ≠ n
Its because 6$ =[ ( )=? ( )1 = fm - = ng
[,
Again from equation (12) i.e.
232
^ ^
] ] ^
∞ ∞
\ =[ ( ) ( − !)1 − _ \ =[ ( )( + )² =? (!)C? ( ) ( − !)1 = _ \ =[ ( )=? ( )Z? ( )1 .

] ]! $
$ ?AB $ ?AB
Applying eqn(13) , (14) and (15) in the above expression –
^ ^
] ]
∞
\ =[ ( ) ( − !)1 − _( + )² =? (!)C? ( ) \ =[ ( ) ( − !)1

] ]!
$ ?AB $
∞ ^
= Z? ( ) _ \ =[ ( )=? ( )1
?AB $
( )=[ (!) − ∑∞
?AB( + )² =? (!)C? ( )=[ (!) = Z[ ( )m[ (16)
o
Now, Z[ ( ) = [ − ∑∞
?AB( + )² =? (!)C? ( )U =[ (!) (17)
pq
By accounting the values from equation (6) and (10) , we can again write
( , ) ( , )
+ = ∑∞ ?AB =? ( )Z? ( ) (18)
Now combining equation (17) and (18)
( , ) ( , ) o
+ = ∑∞
?AB =? ( ) p t − ∑uAB( +
∞
)²=u (!)Cu ( )v =? (!) (19)
L
Jℎ*)* w = -4 43 . + *. , But =( , ) = ∑∞
?AB =? ( )C? ( ) [E. * 3)40 *w-(7)U
Under this condition equation (19) can be written as :
] x y∑∞
?AB =? ( )C? ( )z ] y∑∞
?AB =? ( )C? ( )z
+
] x ]
] ]
∞ ∞
= _ =? ( ) [ − _( + )²= (!)Cu ( )U=? (!) (20)

m? ] ]! u
?AB uAB
So now equation (20) can be written as :
∞ ∞
_ = x ? ( )C? ( ) + _ =? ( )C?, ( )
?AB ?AB
] ]
∞ ∞
= _ =? ( ) [ − _y( + )²= (!)Cu ( )zU=? (!) (21)

m? ] ]! u
?AB uAB
But from equation (8) =? x ( ) − I? x =? ( ) = 0.
So now putting the values of = x ? ( ) E- eqn (21); we get
∞ ∞
_ I? x =? ( )C? ( ) + _ =? ( )C?, ( ) =
?AB ?AB
o
?AB =? (
∑∞ ) X
pL
− ∑∞
uAB ; + < =u (!)Cu ( )Y =? (!) (22)
Now equation (22) can be modified as
o }~
∑∞
?AB =? ( ){ I? x C? ( ) + C?, ( )| − p t − ∑∞
uAB( + < (!)Cu ( )=? (!) (23)
L
So equation (23) must satisfy for any values of
o
I? x C? ( ) + C?, ( ) − p [ − ∑∞ uABy( + )²=u (!)Cu ( )zU=? (!) (24)
L
3.NUMERICAL ANALYSIS
Beam Size= (1.2×0.05×0.006) mm ; Beam type= Mild steel .

M=Mass of the moving object =1kg and 2kg
V=Velocity of the moving object= 5.5 m/s , 22.5 m/s and 41.5m/s.
233
M=1kg and V=5.5m/s
3.5
3
Displacement in cm 2.5
2
1.5 x=l
1 x=vt
0.5
0
0 0.05 0.1 0.15 0.2 0.25
Time‘t’ in sec
(Fig-1)
M=2kg and V=5.5 m/s

7
Displacement in cm
6
5
4
3
2 x=l
1 x=vt
0
0 0.05 0.1 0.15 0.2 0.25
Time ‘t’ in sec

(Fig-2)
M=1kg and V=22.5m/s

Displacement in cm
0.3
0.2
x=l
0.1
x=vt
0
0 0.01 0.02 0.03 0.04 0.05 0.06
-0.1
Time ‘t’ in sec

(Fig-3)
234
M=2kg and V=22.5 m/s
0.5 Displacement in cm
0.4
0.3
0.2 x=l
0.1 x=vt
0
-0.1 0 0.01 0.02 0.03 0.04 0.05 0.06
Time ‘t’ in sec

(Fig-4)
M=1kg and V=41.5m/s
0.05
0.04
Displacement in cm
0.03 x=l
0.02 x=vt
0.01
0
-0.01 0 0.005 0.01 0.015 0.02 0.025 0.03
-0.02
-0.03
-0.04
-0.05
Time‘t’ in sec
(Fig-5)
M=2kg and V=41.5 m/s
0.1
x=l
Displacement in cm
0.05 x=vt
0
0 0.005 0.01 0.015 0.02 0.025 0.03
-0.05
-0.1
(Fig-6)
235
4.RESULTS AND DISCUSSION
1.When the velocity of the moving mass increases ,the deflection at the end point of the beam decreases; It’s
because at higher velocity of the moving mass ,lower modes are not excited, which mainly contribute larger
dynamic displacement of the beam.
2. From Fig-2, 3 and Fig-5,6; The dynamic deflection at the free end decreases sharply with time and again
increases at higher mass velocity. Its because of higher modes of vibration are more prominent than that of the
1st mode. Time ‘t’ in sec (Fig-6)
3. In case of x= vt ; the dynamic deflection mainly depends upon the weight of the moving mass; But in case of
x=l; the dynamic deflection depends upon the mode of vibration of the beam.
5.CONCLUSION
The deflection of the beam structure mainly depends upon the velocity of the moving mass. As the moving mass
velocity increases, the dynamic deflection of the beam decreases; which is a peculiar phenomenon. However as
the magnitude of the moving increases, the deflection increases.
REFERENCES
[1]. Jeffcott.H ; On the vibration of beams under the action of moving loads, Philosophical magazine series7.
1929, Vol.8 (48),pp.66-97.
[2]. M. M. Stanisic, Baltimore, Md., J. C. Itardin and Y. C. Lou, Lafayette, Ind. On the response of the plate to a
multi-masses moving system, Acta Mechanica, 1968,vol.5,pp.37-53.
[3].M.M Stanisic and Hardin.J.C,On the response of beams to an atbitry number of moving masses, J.Franklin
Institute,1969,Vol.187(2),pp.115-123.
[4]. M.M.Stanisic,On the dynamic behavior of the structures carrying moving masses,Ingenieur-Archiv of
Applied Mechanics,1985,Vol.55(3),pp.176-185.
[5]. John E.Akin and Massood Mofid,Numerical solution for the response of beams with moving mass, Journal of
Structural Engg,1889, Vol.115(1),pp.120-132.
[6]. M.Olsson, On The Fundamental Moving Load Problem, Journal of sound and vibration, 1991, Vol. 145(2),
pp.299-307
[7]. D.R Parhi and A.K.Behera , Dynamic deflection of a cracked beam with moving masses, Journal of
mechanical engineering science,1997,Vol.211,pp.77-87.
[8]. G.Michaltos,D.Sophianopolos and A.N.Kounadis ,The effect of a moving mass and other parameters on the
dynamic response of a simply supported beam, Journal of Sound and Vibration,1996,Vol.1919(3),pp.357-362.
[9]. Ho-Chul Kwon ,Man-Cheol Kim and In-Won Lee, Vibration control of bridges under a moving load, Journal
of computer and structures,1998,Vol.66(4),pp.473-380.
[10]. Mohamed.A.Mahmoud, Stress intensity factors for single and double edge cracks in a simple beam subject
to a moving load, International Journal of Fracture, 2001, 111, pp.151-161.
236
[11]. Mohamed.A.Mahmoud, Effect of cracks on the dynamic response of a simple beam subject to a moving
load, Journal of Rail and Rapid,2001,vol(3),pp-207-215.
[12] Han-Ik Yoon and In-Soo Son, Influence of tip mass on dynamic behavior of cracked cantilever pipe
conveying fluid with moving mass, Journal of Mechanical Science and Technology, 2005,
vol(19),No-9,pp-1731-1741.
[13]. Han-Ik Yoon and In-Soo Son, Dynamic behavior of cracked simply supported pipe conveying fluid with
moving mass, 2005, Journal of Sound and Vibration, vol (292).pp.941-953.
[14]. Hai-Ping Lin and Shun-Chang Chang, Forced responses of cracked cantilever beams subjected to a
concentrated moving load, 2006, International Journal of Mechanical Sciences 48 vol(48), pp.1456-1463.
[15]. Han-Ik Yoon, In-Soo Sona, Sung-Jin Ahn, Free Vibration Analysis of Euler-Bernoulli Beam with Double
Cracks, 2006, Journal of Mechanical Science and Technology ,vol(21),pp.476.485.
[16]. J. Yang, Y. Chen, Y. Xiang, X.L. Jia, Free and forced vibration of cracked inhomogeneous beams under an
axial force and a moving load, 2007, Journal of Sound and Vibration 312 (2008) pp.166–181.
[17]. Hai-Ping Lin, Vibration Analysis of a Cracked Beam Subjected to a Travelling Vehicle, 2007, 14th
International Conference on Sound and Vibration, vol (140).
[18]. Samer Masoud AL-Said, Crack Detection in Stepped Beam Carrying Slowly Moving Mass, 2007, Journal
Vibration and Control, vol(14).pp.1903-1920.
[19]. Murat Reis and Yasar Pala, Vibration of a cracked cantilever beam under moving mass load, 2009,
Journal of Civil Engineering and Management, vol-18(1), pp.106-113.
[20]. Murat Reis and Yasar Pala, Dynamic Response of a Cracked Beam under a Moving Mass Load, 2012,
Journal of Engineering Mechanics, ASCE, vol (558).
[21]. Siringoringo and Fuzino, Estimating Bridge Fundamental Frequency from Vibration Response of
Instrumented Passing Vehicle: Analytical and Experimental Study, 2011, Journal of Advances in Structural
Engineering, vol-15(30, pp.443-459.
[22]. Qinkai Han, Jingshan Zhao and Fulei Chu, Dynamic analysis of a geared rotor system
Considering a slant crack on the shaft, 2012, Journal of sound and vibration,vol-331,pp.5803-5823.
[23]. S.Eftekhar Azama, M.Mofidb, R.Afghani Khoraskanic, Dynamic response of Timoshenko beam under
moving mass, 2013, Journal of Scientia Iranica, vol-20(1), pp.50-56.
[24]. L.Fryba, Vibration of solids and structures (1999), Third Edition, Thomas Telford Ltd , Prague.
[25]. W.T.Thomson, Theory of Vibration with Application (2002) Third Edition.CBS Publishers , New Delhi.
237
Paper No.: 039
BEHAVIOUR OF TWO-SPAN CONTINUOUS STEEL BOX RC DECK

COMPOSITE BRIDGES
Saswati Dey1, K. Bhattacharya2
1
Junior Lecturer, Department of Architecture, Women’s Polytechnic, Chandernagar, W.B.
2
Professor, Department of Civil Engineering, NIT Durgapur, W.B
ABSTRACT
Behavior of a two-span continuous open web Steel-Box-Girder Bridge with Reinforced Concrete (RC) Deck is
analyzed. The bridge is a part of an existing bridge which is located at North West England, having total length
of 172.5m with five spans. The behavior of the junction between the two spans having lengths = 46m and
39.25m, subjected to formula loading (BD37/88) as UDL with or without earthquake loading is studied. As it is
expected the flexural stresses on flanges show large shear-lag effect that even leads to material non-linearity in
them. Similarly the deck is subjected to flexural tension over support, and thus becomes cracked. The
diaphragms are subjected to huge reaction; particularly the intermediate one so, non-linearity may occur in them
either by buckling or yielding. Non-linear analysis is carried out by modeling RC as tension-stiffening cum
compression-crushing material; the steel as elasto-plastic material and the reinforcing bars (Rebar) as elastic
membrane. The stresses are evaluated using ABAQUS modeling of the bridge and compared them with
classical method of analysis. Bearing reactions are also obtained.
To what extend these parameters are influenced by Earthquake Excitation is also studied under simulated
spectrum-compatible acceleration time histories.
KEYWORDS: Steel-Box-RC Deck Bridge, Non-linear analysis, Diaphragm, Bearing reactions, Shear lag
effect, Earthquake analysis.
1. INTRODUCTION
The methods of analysis and design of box girders have been surveyed several times to date, for example the
prismatic folded plate theory, (Goldberg and Leve, 1957) considered the box girder to be made up of an
assemblage of folded plates where the independent membrane stresses are merged with the classical plate
theory for analyzing bending and twisting of the component plates. The analysis is limited to straight, simply
support prismatic box girders composed of isotropic plates with no interior diaphragms. The theory was later
marginally improved by (Scordelis, 1966) by taking the effect of interior diaphragms.
The thin-walled elastic beam theory (V1asov, 1961) was intended for analyzing single cell girder which was
later modified (by Abdel et.al. 1968; Wright, Robinson et.al. 1967) by introducing the theory of longitudinally
or transversely stiffened plate elements and with rigid or deformable diaphragms.
Chen and Yen (1980) have combined the basic features of steel-concrete composite box girders on the basis of
classical elastic theory. The steel plates used for the configuration of the box may have been stiffened or un-
stiffened. This work has detailed the application of simple flexural theory to torsion with warping and
finally to dis-torsion effects on box having interior diaphragms. The reinforced concrete (RC) deck had been
modeled by using theory of matrix. The theory was validated with some experimental results. However, the
limitation of that work is, it studied only the simple supported bridge and so the effect of cracking of the RC
deck, which happens to be obvious in case of continuous box-girder bridge, remains unaddressed.
Topkaya, Yura et.al (2005) have carried out experimental as well as field investigation on the composite action
of box-girder bridge by giving emphasis on the load transfer mechanism of shear studs between RC deck and
top flanges of the girder, wherefrom it was proven that composite action develops at very early concrete ages,
for example by twenty-eight days.
The diaphragms over bearings are one of the vulnerable parts in steel box with reinforced concrete (RC) deck
bridges, which are used to fail either by buckling or yielding. Therefore, to enhance buckling resisting capacity,
bearing stiffeners are welded to them. Horizontal stiffeners are also provided in certain cases, particularly near
the man-holes provided in diaphragms for maintenance purposes.
The loads of the super structures e.g. bridges are transferred to the peers through bearings. The bearings are
placed concentrically beneath the diaphragms which are in-turn attached to the peers through fasteners. The
reactions in fasteners are primarily vertical however also accompanied by transverse and longitudinal stresses
that depend upon the geometry of the bridge e.g. i) curvature and skew angles of the bridge, ii) continuity of the
bridge over several supports iii) loads etc.
The analysis and design of diaphragm, bearing stiffeners, bearings with fasteners etc. carried out by classical
methods are available in literature. Detail analysis using FEM is very rare. In-fact the authors couldn’t find any.
Therefore, analysis is carried out for the junction of interior two spans of an existing bridge located at North
West England, having total length of 172.5m with five spans, (D.C. Iles, 1984). Out of five spans, attention is
given to the junction of the largest span (46m) and the neighboring one of 39.25m, ignoring the presence of
other spans to discard the relieving effects would have been appeared from them. The total width of the RC
deck is 12.2m having 7.3m carriageway, 1.85m wide footpath, 0.5m space for fencing and 0.2m for service line.
The deck is symmetric about its center line.
FEM analysis under dead and live load at Ultimate Limit State (BD 37/88) using ABAQUS modeling is carried
out and compared with that are presented in literature (D.C. Iles, 1984). The aim of the study is to assess,
1. The shear lags effect on the flexural stress distribution across the width of the bridge at the intermediate
support.
2. To study the equivalent Von-Mises stress on the Diaphragm, Bearing Stiffeners etc.
3. Effect of cracking of RC deck on the stresses stated above.
4. Non-linear response of the bridge when the material of box i.e. steel is considered as Elasto-Plastic.
5. Transverse reactions developed in fasteners, and comparing the results between static and earthquake
analysis.
The paper is organized by describing bridge modeling followed by Static analysis. Later earthquake analysis is
described in brief.
2. COMPONENTS OF BRIDGE
The model is consisting of two open-web steel boxes, RC deck, internal stiffeners in a box, diaphragms, bearing
stiffeners and the connecting I beams that act as the global bracings of the bridge. The different elements are
described below,
2.1 Diaphragms
The diaphragms are made of 20mm steel plate consisting of four vertical stiffeners and two horizontal stiffeners
which are placed at both sides of it. The stiffeners are of 25mm thick steel plate. A manhole is provided into
the diaphragm for maintenance purpose, (Figure1). Four noded shell elements are used in diaphragms.
239
Figure1 Diaphragm with Bearing Stiffeners
2.2 RC Deck
RC deck of 12.2m wide and 200mm thick is reinforced with top and bottom layers of 25 mm reinforcing bars
having spacing @ 150mm c/c in both directions. The two meshes are placed having 50mm effective cover each
from top and bottom faces of the deck. The deck is modeled with 20 noded elements while the reinforcement
mesh is modeled as membrane having 6.53mm equivalent thickness. These two types of elements are embedded
element wise. The meshing is shown in Figure 2.
R C d e c k w i t h e m b a d d e d to p a n d b o t to m r e in f or c e m e n t
Figure 2 RC Deck with Finite Element Meshing
2.3 Steel Boxes and Stiffeners
Two trapezoidal steel boxes were provided under the deck, made up of 12.5mm thick plates for webs, 47.5mm
plates for bottom flange, 24 mm for top flange (Figure 3). Each box is consisting of a combination of 4 types of
stiffeners, i.e. two transverse web stiffeners and one transverse bottom flange stiffener and one top tie stiffener
in addition to one diagonal stiffener. All stiffeners are made up of 12.5mm thick plates. These combinations of
stiffeners are placed at the intermediate position of the boxes.
Five I-beams are used as bracings between two boxes for load distribution purpose as well as for keeping the
trapezoidal boxes at their own position tightly. These I-beams are provided at supports and selected
intermediate positions. The bridge under consideration is shown in (Figure 4).
240
Figure 3 Two trapezoidal boxes connected with cross beams
Figure 4 The Complete Composite Box-Girder Bridge
3. MATERIAL MODELLING
The analysis is carried out by considering the RC deck (M40) as Elasto-Plastic material under compression and
a Tension Stiffening material in tension, invoking Smeared Crack Modelling due to the presence of rebars (Park
and Pauley,1975; ABAQUS ). The steel (S355 J2) is also modeled as elastic-plastic having yield strength =
355 MPa. However, the rebars are considered as elastic. These stress-strain diagrams concrete and steel are
given in Figures 5 (a-c).
241
MODELING OF RC, TENSION STIFFENING
sigma/sigma_c
1.2
0.8
Failure stress
Sig ma/sig ma_c

0.6
0.4
St rain soft ening
0.2
0
0.00E+00 2.00E-04 4.00E-04 6.00E-04 8.00E-04 1.00E-03 1.20E-03
-0.2
Epsilon-epsilo n-c
(a) Tension Stiffening Modeling of Concrete
MODELING OF CONCRETE, UNDER COMPRESION

compressive
stress
5.00E+07
4.50E+07
4.00E+07
Compressive stress
3.50E+07
3.00E+07
2.50E+07
2.00E+07
1.50E+07
1.00E+07
5.00E+06
0.00E+00
0.00E+00 1.00E-02 2.00E-02 3.00E-02 4.00E-02
-5.00E+06
Plastic strain
(b) Compressive Stress -Strain Curve of Concrete (c) Stress-Strain Curve of S355 J2 Steel
(Stress in MPa) (Stress in N/m2)
Figure 5 Stress-Strain Diagrams of Concrete and Steel
4. STATIC ANALYSIS
The bridge is analyzed with the loads as prescribed by BD-37/88, a British Code appropriate for the bridge
stated here. The code stipulates following loads as a) Self Weight, b) Super Imposed Dead Load (SIDL)
comprising of screed concrete over the carriage way, and appropriate dead load over footpath, permanent
formworks etc. The Live Load comprises of a) HA loading which is a formula based UDL and/or b) Abnormal
242
Vehicle lading (HB loading). As suggested in the code the lanes may have a) HA loading in one lane with HB
in another, or b) both the lanes may have HA loadings. The paper highlights the results corresponding to later
one i.e. type-b combination.
4.1 Design Load
4.1.1 H.A. Load
Type HA loading consists of a uniformly distributed load and a knife edge load combined. It is a formula
loading. As per BD 37/88 the load lengths in excess of 50m but less than 1600m the UDL shall be derived from
the equation:
W= 36(1/L) 0.11 1
and, the KEL per notional lane shall be taken as 120 kN. The Design Load is comprised of other loads over the
self-weight as non-structural mass as, ‘Super Imposed Dead Load ‘(SIDL) due to screed concrete over the
carriage way = 4 kN/m2, equivalent patch load = 6kN/m2 over the Foot Path which are inclusive of permanent
form work.
4.1.2 H.B. Loading
Abnormal Vehicle lading is defined by 30 to 45 units of axel loads of a four axel vehicle, when one unit = 10
kN. The spacing of inner axels varies from 6 to 11m. The analysis is done by considering 30 units as axel
loading which is recommended for country roads and used by the designer (D.C. Iles, 1984).
4.2. Serviceability Limit State Analysis
The static analysis in Serviceability Limit State (SLS) is carried out with both un-cracked and cracked
modelings of RC deck. It is needless to say that Shear Lag effect is noticed prominently over the continuous
support where shear is at its peak. The stresses obtained in bottom flange, RC deck, rebars over continuous
support are presented in Figures (6-8), where stresses are plotted in N/m2.
The flexural stress distribution in the bottom flange (1.2 m wide) over the continuous support of a box girder is
shown in Figure (6), from which the effect of shear lag appears significant. The stresses obtained from both
cracked and un-cracked modeling of deck are shown in the figure from where it can be concluded that the stress
distribution remains practically unaffected by the modeling of the concrete.
In contrast to the above, the stress distribution in RC deck from its edge to the centerline of the deck, as
expected, is highly influenced by the modeling of concrete Figure (7). For example, while the stress in deck
appears more than 5 MPa by un-cracked modeling, has come down to about 2 MPa due to tension stiffening
effects. Nevertheless, the shape of the stress profile in cracked and un-cracked modeling is varying markedly
which is due to the quantum of cracking developed at different locations in the deck. The zone, which is
supposed to be having maximum flexural stress in elastic analysis, is suffering by maximum cracking thus,
carrying minimum stresses in cracked modeling.
The stress distribution in rebars is shown in Figure (8). The stresses in the bars are increased in cracked
modeling with respect to the un-cracked modeling of concrete. The stress released by the RC deck on cracking,
is received by the rebar due to presence of bond stress and thus the stresses of top bar for non-linear cases are
greater than that of linear cases. The linear and non-linear curves are similar in pattern indicating similar shear
lag effect. The stresses are in peak at the webs and are decreasing as the distance from the web increases.
243
4.2.1 Stresses in Diaphragm
The stresses on diaphragm both in vertical and horizontal sections are given in Figures (9-10). Both the
diagrams show a) in-plane compressive stresses (S11) and b) equivalent Mises Stress. The Figure (9) shows that
at any vertical section the compressive stresses decreases from bottom to top of the diaphragm, while the stress
remains almost (Figure 10) constant at any horizontal section of the diaphragm. The same is also true for Von-
Mises stress. It is also seen from these figures that the concrete modeling i.e. cracked/un-cracked doesn’t have
any appreciable effects on the magnitude of the stresses.
Figure 6. Compressive Stress in bottom flange Figure 7. Stress in RC deck (from edge to center of Deck)
Figure 8. Stress in top bar (from edge to center of Deck) Figure 9. Diaphragm stresses at Vertical Section
244
Figure 10. Stresses in Diaphragm at Horizontal Section
4.3 Ultimate Limit State Analysis
As it was seen above, that due to redistribution of internal stress between concrete and rebars (because of the
presence of bonding) the stresses in various components of a Box-Girder remained unaffected, the analysis is
further carried out under Ultimate Limit State (ULS) by considering un-cracked modeling of concrete.
The various load factors and material factors are (BD-37/88) as follows.
A. Loads to be taken in each combination with appropriate γfL

Dead Load: Steel = 1.05, Concrete = 1.15
Super Imposed Load: Deck and Footpath = 1.75
Live Load: HA alone = 1.5
HA + HB = 1.3
B. The Partial Safety Factor over Yield Stress γ3 = 1.1

C. The additional partial safety factor γm = 1.0
Using these partial factors linear analysis is carried out which shows that there are certain elements particularly
at the bottom part of the diaphragm and also in bottom flange where the resultant stresses (Von-Mises) appear
exceeding the yield stress (i.e. 355 MPa), thus indicating the need for considering non-linear stress-strain curves
for steel. The steel, grade S355 J2 was used by the designer (D.C. Iles, 1984) the stress-strain diagram of which
was shown earlier (Figure 5-c). However, there was no indication of non-linearity in the design. For example,
the stresses in some of the elements of diaphragm and bottom flange which exceeded the yield stress are given
in Tables (1 – 2) along with their designed values which highlights the order of variation. Once the non-linear
analysis is done, it brings down the stresses near to the yield stress.
245
Table 1 Stresses in Diaphragm at its Bottom Part Table 2 Stresses in Bottom Flange over Support
NON
ELEMENT Design LINEAR
NON LINEAR
Design LINEAR S Value ANALYSIS
ELEMENTS LINEAR ANALYSI
Value ANALYSIS NUMBER (N/mm2) ( N/mm2)
NUMBER ANALYSIS S(N/mm2)
(N/mm2) 2
(N/mm )
(N/mm2)
19131 450.811 355.240
13796 373.101 355.066
7864 402.597 355.146
288.2
13795 333.3 361.78 355.099
7849 366.701 355.084
13794 365.36 355.028
7863 404.362 355.170

13793 364.996 355.009
But as this non-linearity occurs in very limited
zones, its effect is not at all significant in overall response of the bridge. For example, the reaction forces at the
mid-support and the vertical deflection at the midpoint of 46m span of a box girder are given in the Table 3
which shows marginal variation. However, would the cracking of the RC is considered, the mid-span deflection
become much larger which could not be evaluated because of excessive cracking of concrete leading towards
matrix singularity at element level.
Table 3 Reaction at Mid-Bearing and mid-Span Deflection (46m)
Rfy Vy (in mm)
Linear Analysis 16770.69kN -182.766
Non-Linear
Analysis 16791.72kN -184.427
4.4 Bearings and Fasteners under Diaphragm
Box girders are torsionally stiff and strong- it is usually adequate to provide only one bearing under each box at
intermediate support. The bearing is placed concentrically with the diaphragm and for a simple-supported
straight bridge; it is generally designed for vertical loads neglecting transverse loadings. However, the same is
not true for a continuous bridge. Whether it is a straight or a curved bridge, depending on the number of spans
and loadings, the horizontal reactions (transverse and longitudinal) in the bearings are developed.
Fasteners are used to anchor the bearings with the peers. For designing the fasteners, meticulous evaluation of
all three directional forces is necessary. The net resultant horizontal forces either in longitudinal or transverse
on a bearing are not very large however, from a fastener’s point of view it is significant. To explain it better, a
schematic diagram of nodal points where boundary conditions are applied is given Figure (11).
246
Web
Bottom Flange
2214 562 563 Diaphragm
39.25 m span Bearing Stiffeners 46 m span Z
2216 564 565 X
Web
Figure 11 Nodal Points where Boundary Conditions are applied
The reaction forces at the nodes where they are significant are given in Table (4). The net resultant forces either
in longitudinal or transverse on a particular bearing are not very large, but from fasteners point of view they are
significant.
Table 4 Reaction Peak Forces at Middle-Bearing
Node Rfx(kN) Rfy(kN) Rfz(kN)

Number
562 -706.15 2337.87 -
564 686.97 2154.67 -
565 - - 2459.77
2216 - - -2279.31
Such reactions under earthquake excitations sometime become even more critical. Time-history analysis is
carried out with simulated earthquake data corresponding to Design Basis Earthquake (DBE) of IS: 1893
(2002). However for load combinations, due to the non-availability of revised code for bridges, we have
followed the earlier version of IS 1893, (1984) which has stated that, during earthquake load calculation i) No
live load be considered in longitudinal direction, ii) Only 25% of live load be considered in transverse and
vertical directions, iii) The earthquake load will act with 50% of the live load over deck. The zone factor is
considered as 0.36 which is the highest factor in India, the Importance factor = 1.75, and the Response
reduction factor = 4 (Part -1).
Two different loads combinations are considered, namely,
1.5 (DL + SIDL) + 1.5 (EL) (2)
And,
1.2 (DL + SIDL) + 1.2 (LL) + 1.2 (EL) (3)
247
And the combination as in Eq. (2) appears critical. All three components of earthquake are considered in
analysis and the dynamic reactions in fasteners are evaluated. Comparing them with the static reactions it
appears that the transverse reaction during earthquake is about 56% more than that from static analysis. The
results for selective nodes are given in table (5)
Table 5 Peak Reactions at the Junction of Two Spans (Static and Dynamic)
Node Rfx(kN) Rfy(kN) Rfz(kN)

Number
Static Dynamic Static Dynamic Static Dynamic
562 -706.15 -1100 2337.87 1870.0 -
565 - - - - 2459.77 2200.00
CONCLUSION
From this exercise it is concluded that,
• Modelling of RC deck only affects the stress distribution between the concrete and the reinforcement
bars. The load that is released by the concrete due to cracking gets redistributed in the rebars keeping
the overall load carrying capacity of the RC deck unchanged.
• Because of above internal redistribution of stresses in RC deck due to cracking, the stresses in steel
boxe girders remain unaffected, and so for the detail study of joint encompassing failure due to
buckling or yielding, non-linear modeling of concrete is not required.
• The stresses evaluated by FEM are larger than those obtained from classical methods. The non-linearity
which is encountered by FEM remained unattended by the designer.
• The sum total of (net) horizontal reactions on a bearing appears insignificant. But from a fasteners point
of view, which is location specific, the reaction force varies and meticulous estimation is required for
their design.
• The earthquakes influence the reactions on fasteners to a great extent, for example, there is about 56%
increase in transverse reaction during earthquake presented herein.
REFERENCE
[1] ABAQUS Documentation and Application
[2] Abdel-Samed, S. R. Wright, R. N. and Robinson A. R. (1968) Analysis of box girders with diaphragms,
Journal of the Structural Division, ASCE, Proc. Paper 6153 94(10).
[3] Bakht, B. and Jaeger, L.G. (1985), Bridge Analysis Simplified, ISBN 0-07-100145-X, McGraw-Hill Book
Company.
[4] BD-37/88, Design manual for roads and bridges, THE HIGHWAYS AGENCY, SCOTTISH
EXECUTIVE DEVELOPMENT DEPARTMENT, UK
[5] Chen. Y.S. and Yen. B.T,(1980). Analysis of Composite Box girder, Fritz Engineering Laboratory,
Report No 380.12(80), Lehigh University.
248
[6] Topkaya, C. Yura, J.A. Williamson, E.B. and Frank, K.H. (2005), Composite Action during Construction
of Steel Trapezoidal Box Girder Bridges, Research Project 0-1898-2,
[7] Transportation research boureau of engineering research, the University of Texas. Austin, Texas
Department of Transportation, and U.S. Department of Transportation, Federal Highway Administration.
[8] Iles D.C. (1984), Design Guide for Composite Box Girder Bridges, The Steel Construction Institute, ISBN
1 85942 0060.
[9] Goldberg, J. R. and Leve, H. L.(1957), Theory of Prismatic folded plate structures, IABSE, 17
[10] IS 1893(Part-1), (2002), Criteria for earthquake resistant design of structures, General Provisions and
Buildings, BIS, ND 110002.
[11] Park, R. and Paulay, (1975), T. Reinforced Concrete Structures, John Wiley and Sons.
[12] Scordelis, A. C., Analysis of Simple Supported box girder bridges, Report No. SESM-66~17, Department
of Civil Engineering, University of California, Berkeley, CA., October 1966.
[13] Merrison, J. (1973), Inquiry into the basis of design and methods of erection of steel box girder bridges,
Interim Design and Workmanship Rules, Department of Environment, Scottish Development Department,
Welsh Office, February.
[14] V1asov, V. Z.(1961), Thin walled elastic beams, Report, The Israel Program for Scientific
Translation, Jerusalem, U. S. Department of Commerce, Washington, D. C.
[15] Wright, R. N., Abdel-Samed, S. R. and Robinson, A. R. (1967),‘Analysis and design of closed section
girder bridges with diaphragms, American Iron and Steel Institute, Project 110.
249
Paper No.: 042
PREDICTION OF HYSTERETIC ENERGY DEMAND IN STEEL

MOMENT RESISTING FRAMES USING MULTIVARIATE ADAPTIVE
REGRESSION SPLINES
Shantaram Parab1, Jatin Alreja2, and Pijush Samui3
1
Undergraduate Student, School of Mechanical and Building Sciences (SMBS), VIT University
Vellore - 632014, Tamil Nadu
2
Undergraduate Student, School of Mechanical and Building Sciences (SMBS), VIT University
3
Professor & Director, Centre for Disaster Mitigation and Management (CDMM), VIT University,
Email- shantaram_186@hotmail.com, jateen_alreja2005@yahoo.com, pijush.phd@gmail.com
ABSTRACT:
Hysteretic energy demand estimation is the first significant step in energy-based seismic design of structures.
The hysteretic energy dissipated by systems undergoing dynamic loading is often thought to represent a useful
measure of their performance when subjected to earthquake excitation. This paper presents a new model for the
Prediction of hysteretic energy demand in steel moment Resisting frames using Multivariate Adaptive
Regression Spline (MARS). MARS is a non-parametric regression technique and can be seen as an addition
to linear models that automatically models non-linearity and interactions between variables. This model is used
to establish a relation between the Hysteretic energy demand and several effective parameters such as
earthquake intensity, number of stories, soil type, period, strength index, and the energy imparted to the
structure. A total of 27 data sets (input-output pairs) are used, 23 of which are used to train the model and 4 are
used to test this model. The data is normalized between 0 and 1 prior to being used in the model. The datasets
used in this study are derived from experimental results. The performance and validity of the model is further
tested on different steel moment resisting structures. The proposed model provides a very high correlation
coefficient (R =0.9864), and a low root mean squared error (RMSE=828.22) value. The developed MARS has
been compared with the Linear Genetic Programming (LGP). Accurate results portray the strong potential of
MARS, as a feasible and reliable tool to predict the hysteretic energy demand.
KEYWORDS: MARS, Hysteretic Energy, Mathematical Modelling
1. INTRODUCTION:
The hysteretic energy, absorbed by a structural system during a seismic event that is strong enough to induce a
certain amount of nonlinearity to the system, has been recognized by several researchers as a potentially useful
seismic performance indicator (e.g., Park et al. 1987; Bojorquez et al. 2011).The energy dissipated by hysteresis
in structures under reversible load patterns is one of the most widely used parameters to evaluate cumulative
effect of load history (cumulative damage). It is used to identify different behaviors between elements whose
failure is dominated by flexion or shear, but also many others phenomena. However, it is necessary to evaluate
this parameter to have a complete record of load-displacement history of each member of the studied structure.
Hysteretic Energy is a parameter to measure effects of duration caused to a building after an earthquake strikes.
It also takes into account the cyclic elastic behavior of the structure. Out of the different energy components the
hysteretic energy is more important since it is directly in relation with the cumulative ductility. The hysteretic
energy and its distribution throughout the structure are dependent on both the structural systems and the ground
motion. The structural failure happens when the earthquake-induced hysteretic energy demand for a structure is
larger than the hysteretic energy dissipation capacity of the structure. It is considered as the key parameter to
construct earthquake resistant buildings or to fix already existing structures and modify them into earthquake
resistant structures. Hence it is important to calculate the hysteretic energy for buildings exceptionally for low
rise buildings.
The hysteretic energy demand takes into account the effects of the duration of the earthquake and the
cyclic-plastic deformation behavior of the structure. A monotonic demand parameter, such as peak inelastic drift
or displacement, cannot represent this cumulative cyclic damage. A design approach based on hysteretic energy
demand, thus, has the potential to account for the damage potential explicitly. Hence, the modelling of
Hysteretic Energy demand is necessary.
There are many methods available for the modelling of Hysteretic energy demand of steel moment resisting
structures, but they face some major disadvantages too. One of the methods is:
Genetic Algorithms: Certain optimization problems (they are called variant problems) cannot be solved by
means of genetic algorithms. There is no absolute assurance that a genetic algorithm will find a global optimum.
Like other artificial intelligence techniques, the genetic algorithm cannot assure constant optimization response
times.
Over the past few years, a different modelling method has become very popular and used by many researchers
for a wide range of engineering applications. MARS (Multivariate Adaptive Regression Splines) has a strong
potential for predicting the output (Hysteretic Energy) with high correlation and precision to the experimental
value. Multivariate adaptive regression splines (MARS) is a form of regression analysis introduced by Jerome
Friedman in 1991. MARS models are more flexible than linear regression models. It can handle both continuous
and categorical data and it is simple to understand and interpret.
The aim of this study is to predict the hysteretic energy demand of steel moment resisting structures. Totally 27
input-output data-sets were collected, trained and tested by means of MARS model. This model has been
developed using MATLAB software with the input parameters: Earthquake intensity (I), Number of stories (Ns),
Soil type (Z), Fundamental period (T), Strength index (σ) and Hysteretic energy to energy imparted to the
structure (EI) ratio (EH/EI) and the output parameter Hysteretic Energy(EH/m). The obtained results have been
compared by experimental ones to evaluate MARS model for predicting the Hysteretic Energy Demand.
2. MULTIVARIATE ADAPTIVE REGRESSION SPLINES (MARS):
Multivariate adaptive regression spline is an implementation of techniques introduced by Friedman (1991),

which is used for predicting the values of a continuous dependent variable using a set of predictor or
independent variables. This method makes no assumption about the functional relationship between the
dependent and independent variables, instead MARS derives this relationship from a set of coefficients and
basis functions that are directed from the regression data. MARS model is implemented in 2 steps where in the
first step we build the model and add basis functions to increase the complexity until maximum complexity is
achieved. Then in the second step we do a backward calculation to eliminate the least significant basis function
from the model. Given below is an algorithm to explain MARS modelling.
The MARS algorithm builds models from two sided truncated functions of the predictors (x) of the form:
− , >
− = }
, ℎ
(1)
These serve as basis functions for linear or nonlinear expansion that approximates some true underlying
function f(x).
The MARS Splines model for a dependent (outcome) variable y, and M terms, can be summarized in the
following equation:
251
= = + ∑ ! " !, (2)
Where the summation is over the M terms in the model, and bo and bm are parameters of the model (along with
the knots t for each basis function, which are also estimated from the data). Function H is defined as:
#$% & ' $,% ( = ∏+

$ *$% (3)
Where , -,. is the predictor in the kth of the mth product. For order of interactions K=1, the model is additive
and for K=2 the model pairwise interactive.
During forward stepwise, a number of basis functions are added to the model according to a pre-determined
maximum which should be considerably larger (twice as much at least) than the optimal (best least-squares fit).
After implementing the forward stepwise selection of basis functions, a backward procedure is applied in which
the model is pruned by removing those basis functions that are associated with the smallest increase in the (least
squares) goodness-of-fit. A least squares error function (inverse of goodness-of-fit) is computed. The so-called
Generalized Cross Validation error is a measure of the goodness of fit that takes into account not only the
residual error but also the model complexity as well. It is given by
∑6
37 23 4 3
5
GCV = 9 5
(4)
8 :
6
With
; = 1 + => (5)
Where N is the number of cases in the data set, d is the effective degrees of freedom, which is equal to the
number of independent basis functions. The quantity c is the penalty for adding a basis function. Experiments
have shown that the best value for C can be found somewhere in the range 2 < d < 3.
2.1. Development of MARS model:
Out of the 27 data sets which are available, 23 datasets are used to train the MARS model and 4 datasets are
used to test the accuracy of the model. Table 1given below shows the data set used for developing the MARS
model. The data was normalized between 0 and 1 before being used in the model as following:
? ? C@
?@AB = ? (6)
D ? C@
252
Table 1- The Database used for model development
S. No I Ns Z T (s) σ EH/EI EH/m (cm/s)2

1 1 3 1 1.0109 0.22 0.84 4160
2 1 3 2 1.0109 0.22 0.8 4011

3 1 3 3 1.0109 0.22 0.87 5751
4 1 9 2 2.2862 0.11 0.7 415

5 1 9 3 2.2862 0.11 0.81 835
6 1 20 1 3.7863 0.058 0.78 1503
7 1 20 2 3.7863 0.058 0.68 1081

8 1 20 3 3.7863 0.058 0.8 1849
9 2 3 1 1.0109 0.22 0.88 10948
10 2 3 2 1.0109 0.22 0.86 11219
11 2 3 3 1.0109 0.22 0.9 13953
12 2 20 1 3.7863 0.058 0.83 4075

13 2 20 2 3.7863 0.058 0.75 3131
14 2 20 3 3.7863 0.058 0.83 4496

15 3 3 1 1.0109 0.22 0.9 20140
16 3 3 2 1.0109 0.22 0.88 21383
17 3 3 3 1.0109 0.22 0.91 24613
18 3 9 2 2.2862 0.11 0.82 2484
19 3 9 3 2.2862 0.11 0.88 4387

20 3 20 1 3.7863 0.058 0.85 7656
21 3 20 2 3.7863 0.058 0.79 6202

22 3 20 3 3.7863 0.058 0.85 8173
23 1 9 1 2.2862 0.11 0.85 1760
24 2 9 1 2.2862 0.11 0.89 4809

25 2 9 2 2.2862 0.11 0.78 1241
26 3 9 1 2.2862 0.11 0.9 9042

27 2 9 3 2.2862 0.11 0.86 2361
253
The assessment of the model is done on the basis of coefficient of regression value R which is calculated using
the formula
∑@ G G
C7 FDC FD FHC FH
E= (7)
I∑@ G @ G
C7 FDC FD I∑C7 FHC FH
Where Eai and Epi are the actual and predicted values, respectively, E a and p
are mean of actual and
E
predicted E values. For an effective and good model the R value should be close to one. Also while comparing
the models the values of R is compared and the model with R value closer to one and higher than the other is
considered better and used.
3. RESULTS AND DISCUSSIONS:
MARS model has been developed to predict hysteretic energy demand of steel moment resisting frames which
is one of the prime parameter in designing of steel structures. The complete model is divided into two phrases
that is forward and backward pass which is also used by recursive partitioning trees. This model has been
developed using MATLAB software for training and prediction of the hysteretic energy. The model has been
trained using 23 data sets and 4 data sets are used to validate the model. This model gives a more precise value
of hysteretic energy demand as compared to the already existing models. The model will be highly useful in
designing the steel moment resisting frames.
3.1. Graphs:
1.2
Normalised Predicted Hysteretic Energy
Correlation Coefficient (R) = 0.9864

1
0.8
0.6
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1 1.2
Normalised Experimental Hysteretic Energy
Fig. 1- Correlation of the measured and predicted Hysteretic Energy values for Training Set
254
0.25
Correlation Coefficient (R) = 0.9890

Normalised Predicted Hysteretic Energy
0.2
0.15
0.1
0.05
0
0 0.05 0.1 0.15 0.2
Normalised Experimental Hysteretic Energy
Fig. 2- The correlation of the measured and predicted Hysteretic Energy values for Testing Set
The performance of the MARS model is shown in Fig. 1 and Fig. 2. The best value of R (Coefficient of
Correlation) is 0.9890 for the training set and the value of R for the testing set is 0.9864. The RMSE value
calculated using the predicted values by the MARS model is 1000.13 for the training set and 828.22 for the
testing set. The values of Correlation coefficient and Root Mean Squared Error show that the proposed MARS
model is suitable and can predict Hysteretic Energy values very close to the experimental values.
The Hysteretic Energy Equation formulated according to MARS is given below:
J = −0.0335 + ∑QORS ;O PO (8)
Where, y = Eh/m, Ci = Coefficient, βi(x) = Basis Function .
Table 2 – Details of Basis Function
Βi(x) Wi Function
TU
Β1(x) 0.9945 Β1(x) = max(0, TV
– 0.73913)
Β2(x) 0.4756 Β2(x) = max(0, T -0.45950)

Β3(x) 0.7679 Β3(x) = max(0, 0.45950 –T)
Β4(x) 1.6201 Β4(x) = Β3(x) * max(0, I -0.5)
Β5(x) -0.7776 Β5(x) = Β3(x) * max(0, 0.5 – I)
Β6(x) -0.6819 Β6(x) = Β2(x) * max(0, 0.5 – I)
Fig 3 and Fig 4, shown below shows the comparison between GSA and MARS models for modelling
Hysteretic Energy in terms of Correlation Coefficient (R) and Root Mean Squared Error (RMSE).
255
0.9870
0.9864
Correlation Coefficient (R)

0.9865
0.9860
0.9855
0.985
0.9850
0.9845
0.9840
GSA MARS
Fig 3 – Comparison between MARS and GSA models in terms of Correlation Coefficient (R)
1600
Root Man Squared Error (RMSE)
1400 1346.1
1200
1000
828.22
800
600
400
200
0
GSA MARS
Fig 4 – Comparison between MARS and GSA models in terms of Root Mean Squared Error (RMSE)
These results prove that the MARS model is more accurate and reliable for the prediction of the Hysteretic
Energy Demand in Steel Moment resisting Structures.
4. CONCLUSION:
This study presents MARS based approach for the prediction of Hysteretic Energy Demand. The major
conclusions are given below:
1. MARS is an effective approach for the evaluation of the Hysteretic Energy Demands of Steel Moment
Resisting Frames.
2. Comparison of Results of MARS in terms of R showed that MARS models are capable to predict effective
results for Hysteretic Energy values.
3. MARS model is much better than the Genetic-based Simulated Annealing (GSA) model based on
root-mean-square-error (RMSE) performance criteria.
4. Results show that the proposed MARS model could achieve the best performance in terms of prediction
accuracy.
256
REFERENCES:
[1] Pakize Taylan, Gerhard-Wilhelm Weber and Fatma Yerlikaya Ozkurt (2010). A new approach to
multivariate adaptive regression splines by using Tikhonov regularization and continuous optimization.
Springer,18,377-395
[2] Amir Hossein Gandovi, Amir Hossein Alavi and Abazar Asghari, Hadi Niroomand, Ali Matin Nazar
(2013). An innovative apporoach for modeling of hysteretic energy demand in steel moment resisting
frames. Springer Verlag London-2013.
[3] Manish Rathore, Amarnath R. Chowdhary and Siddhartha Ghosh. Approximate method for estimating
hysteretic energy demand on Plan- Asymmetric building. Journal of Earthquake Engineering, 15:1,
99-123.
[4] Tholen Prasanth, Siddhartha Ghosh and Kevin R Collins (2008). Estimation of hysteretic energy demand
using concept of modal pushover analysis. John Wiley and Sons Ltd., 37,975-990.
[5] A.k kazantzi and D Vamvatsikos (2012).A study on the correlation between dissipated hysteretic energy
and seismic performance.15th World Conference on Earthquake Engineering.
[6] Amarnath R. Chowdhary and Siddhartha Ghosh (2010).Inclusion of p-∆ effect in the estimation of
hysteretic energy demand based on modal pushover analysis. ISET Journal of Earthquake Technology, 47,
75-86.
[7] Javier De Andres, Pedro Lorca, Francisco Javier De Cos Juez and Fernando Sanchez Lasheras
(2011).Bankruptcy forecasting: A Hybrid approach using fuzzy c- means clustering and multivariate
adaptive regression splines (MARS).Expert Systems with Application, 38, 1866-1875.
[8] Dengju Yao,Jing Yang and Xiaojuan Zhan(2011).Predicting breast cancer Survivability using random
forest and multivariate adaptive regression splines(MARS). Electronic and Mechanical Engineering and
Information Technology (EMEIT) International Conference on 2011, 5, 2204-2207.
[9] Claudio Oyarzo Vera (2006). Evaluation of energy dissipated by hysteresis in R/C shear walls using
parameters of displacement. First European conference on Earthquake Engineering and Seismology.
[10] Jalali-Heravi M, Asadollahi-Baboli M and Mani-Varnosfaderani A (2009). Shuffling multivariate adaptive

regression splines and adaptive neuro-fuzzy inference system as tools for QSAR study of SARS inhibitors.
Journal of Pharmaceutical and Biomedical Analysis, 50,853-860.
257
Paper No.: 043
BUCKLING ANALYSIS OF LAMINATED SOFT CORE SANDWICH PLATE
H.D. Chalak1, Anupam Chakrabarti1, Mohd. Ashraf Iqbal1 and Abdul Hamid Sheikh2
1
Department of Civil Engineering, Indian Institute of Technology Roorkee, Roorkee, INDIA
2
School of Civil, Environment and Mining Engineering, University of Adelaide North Terrace, Australia
Email: chalakhd@yahoo.co.in, anupam1965@yahoo.co.uk, iqbal_ashraf@rediffmail.com,
ahsheikh@civeng.adelaide.edu.au
ABSTRACT :
The buckling behavior of laminated soft core sandwich plates with general lamination configuration is studied
using new C0 element developed based on higher order zigzag theory (HOZT). In this theory, the in-plane
displacement variations are considered to be cubic for both the face sheets and the core while the transverse
displacement is assumed to be quadratic within the core and constant in the faces beyond the core. The plate
theory ensures a shear stress-free condition at the top and bottom surfaces of the plate. Thus, the plate theory has
all of the features required for an accurate modelling of laminated sandwich plates. As very few elements based
on this plate theory (HOZT) exist and they possess certain disadvantages, an attempt has been made to develop
this new element. A nine node C0 quadratic plate finite element is implemented to model the HOZT for the
present analysis. The proposed model satisfies the condition of transverse shear stress continuity at the layer
interfaces and the zero transverse shear stress condition at the top and bottom of the plate. The nodal field
variables are chosen in such a manner to overcome the problem of continuity requirement of the derivatives of
transverse displacements i.e. no need to impose any penalty stiffness in the formulation. The accuracy of the
present C0 finite element in predicting the critical loads of laminated plates is shown by considering different
features of the laminated plates.
KEYWORDS: Sandwich plate, Soft core, Zigzag theory, Finite element, Stress continuity.
1. INTRODUCTION
The need for lighter, stronger and stiffer structure has motivated the use of composite laminates and sandwich
structures. Sandwich construction is a special type of laminated structure with low strength core and high
strength face sheets in the form of composite laminates. Thus the effect of shear deformation is quite significant
and needs special attention. Secondly, the use of high strength materials in such construction leads to slender
sections, thus making buckling as a primary mode of failure of the member subjected to axial compressive
forces. These buckling modes depend upon the material properties and relative stiffness of the face sheets and
core. For these kinds of structures, exact three dimensional (3D) solutions are required to predict its response
accurately. Such 3D solutions are available only for simple boundary conditions and geometries [1-2]. 3D finite
element (FE) solution is computationally expensive and often intractable. Hence, the development of an
appropriate mathematical two dimensional (2D) model to represent the behavior of fiber reinforced composite
laminates has drawn a considerable amount of attention. These plate theories (2D) can be broadly divided into
two categories based on their assumed displacement fields: 1) Single layer theory and 2) Layer-wise theory.
The structural behavior of sandwich laminate cannot be accurately predicted by first order shear deformation
theory (FSDT) as it assumes uniform transverse shear strain over the entire plate thickness. To compensate for
the actual parabolic variation of the shear stress, it (FSDT) requires a shear correction factor [3-4]. The further
improvement comes in the form of higher order shear deformation theory (HSDT), where the higher order
variation of in-plane displacement through the thickness is considered to represent the actual warping of the
plate cross-section due to which it (HSDT) becomes free from shear correction factor [5]. It [HSDT] gives a
continuous variation of shear strain across the thickness, which gives discontinuity in the shear stress
distribution at the layer interfaces due to different values of shear rigidity at the adjacent layers. But the actual
phenomenon is just the opposite i.e., the shear strain is discontinuous and the shear stress is continuous at the
layer interfaces [6]. In order to consider above aspect, the refined plate theories developed starting with
layer-wise plate theories [7]. In layer-wise plate theories the unknown displacement components are taken at all
the layer interfaces, which give a zigzag pattern of through thickness variation for the in-plane displacement to
represent the desired shear strain discontinuity at the layer interfaces. As the number of unknowns increases
directly with the increase in the number of layers due to which these (layer-wise) theories required huge
computational effort. The problem of layer wise theories has been overcome by considering the unknowns at all
the interfaces are expressed in terms of those at the reference plane. This is achieved by satisfying the condition
of shear stress continuity at the layer interfaces.These theories are known as zigzag theories (ZZT) in general. In
some improved version of these theories [8], the condition of zero transverse shear stresses at the plate/beam top
and bottom was also satisfied. The unknown transverse displacement fields across the depth in addition to that
in the reference plane are essential for accurate prediction of the variation of transverse deflection. This can be
done by using sub-laminate plate theories but the number of unknowns will increase with the increase in the
number of sub-laminates. As the introduction of additional unknowns in the transverse displacement fields
invites additional C1 continuity requirements in its finite element implementation by using the zigzag theory, its
application is not encouraged in a practical analysis.
Based on higher order zigzag theory, Pandit et al. [9] proposed a model for the buckling analysis of sandwich
plates with soft compressible core. The core is modeled as a 3D elastic continuum. By defining the separate
field variable for the derivatives of transverse displacement, the authors [9] overcame the problem of C1
continuity. Due to this some constrains have been imposed which are enforced variationally through penalty
approach. However, choosing suitable value for the penalty stiffness multiplier is a well known problem in the
finite element method. Recently Chakrabarti et al.[10] has overcame the shortcomings of the FE model
presented by the authors [9] and proposed a C0 one dimensional FE model based on higher order zigzag theory
to predict the static response of the soft core sandwich beam structures. The displacement fields are chosen in an
effective manner so as no need to impose any penalty stiffness.
Keeping all this in mind, buckling problems of laminated soft core sandwich plates having different geometry
and boundary conditions are solved using a recently developed C0 FE model based on higher order zigzag
theory. The in-plane displacement fields are assumed as a combination of a linear zigzag function with different
slopes at each layer and a cubically varying function over the entire thickness. The out of plane displacement is
considered to be quadratic within the core and constant in the face sheets. The model satisfies the transverse
shear stress continuity conditions at the layer interfaces and the conditions of zero transverse shear stress at the
top and bottom of the plate. The isoparametric quadratic plate element has nine nodes with eleven field variables
at each node. The displacement fields are chosen in such a manner that there is no need to impose any penalty
stiffness in the formulation. The element may also be matched quite conveniently with other C0 elements.
2. MATHEMATICAL FORMULATIONS
To ensure a piecewise parabolic variation of transverse shear strains across the thickness with discontinuity at
the layer interface as expected in a layered plate, the in-plane displacement fields [10] are chosen as follows:
nu −1 nl −1
U = u0 + zθ x + ∑( z − ) H ( z − )
i =1
ziu ziu α xu
i
+ ∑ ( z − z ) H ( − z + z )α
j =1
l
j
l
j
i
xl + β x z 2 +ηx z3
(1)
nu −1 nl −1
V = v0 + zθ y + ∑ ( z − z ) H ( z − z )α + ∑ ( z − z ) H ( − z + z )α
i =1
u
i
u
i
i
yu
j =1
l
j
l
j
i
yl + β y z 2 + η y z3
(2)
259
where, u0 and v0 denotes the in-plane displacements of any point on the mid surface, θx and θy are the rotations
of the normal to the middle plane about y-axis and x-axis respectively, nu and nl are number of upper and lower
layers respectively, βx, βy, ηx and ηy are the higher order unknown, α xu
i
, α iyu , α xlj and α ylj are the slopes
of i-th/ j-th layer corresponding to upper and lower layers respectively and H ( z − ziu ) and H (− z + z lj ) are
the unit step functions.
The transverse displacement is assumed to vary quadratically through the core thickness and constant over the
face sheets [10] and it may be expressed as,
W = l1wu + l2 w0 + l3 wl for core

= wu for upper face layers
= wl for lower face layers (3)
where wu, w0 and wl are the values of the transverse displacement at the top layer, middle layer and bottom layer
of the core, respectively, and l1, l2 and l3 are Lagrangian interpolation functions in the thickness co-ordinate.
The stress–strain relationship of an orthotropic layer/ lamina (say k-th layer) having any fiber orientation with
respect to structural axes system (x-y-z) may be expressed as
σ xx   Q11 Q12 Q13 Q14 0 0  ε xx 

σ    ε 
 yy  Q 21 Q 22 Q 23 Q 24 0 0   yy 
σ zz  Q  ε zz 
   31  
{σ } = Q  {ε }
Q32 Q33 Q34 0 0 
 =   or (4)
 τ xy  Q 41 0  γ xy 
K
Q 42 Q 43 Q 44 0

τ   γ 
 xz   0 0 0 0 Q55 Q56   xz 

 τ yz   0 0 0 0 Q 65 Q 66  K γ yz 

where {σ } , {ε } and QK  are the stress vector, the strain vector and the transformed rigidity matrix of
k-th lamina, respectively.
Utilizing the conditions of zero transverse shear stress at the top and bottom surfaces of the plate and imposing
the conditions of the transverse shear stress continuity at the interfaces between the layers along with the
conditions, u = uu at the top and u = ul at the top of the plate, βx, ηx, βy , ηy, α xu
i
, α iyu , α xlj , α ylj ,
(∂wu / ∂x) , (∂wl / ∂x) (∂wu / ∂y ) and (∂wl / ∂y ) may be expressed in terms of the displacements u0,v0, θx,, θy,
uu, ul, vu and vl as
{B} = [ A]{α } (5)

where,
{B} = {β x η x β y η y α 1xu α xu2 …α xunu −1 α 1xl α xl2 …α xlnl −1 α 1yu α yu2 …α yunu −1 α 1yl α yl2 …α ylnl −1
( ∂wu {
∂x )( ∂wu ∂y )( ∂wl ∂x )( ∂wl ∂y )} , {α } = u0 v0 θ x θ y uu vu ul vl }
T T
and the elements of [A] are dependent on material properties. It is to be noted that last two entries of the vector
{B} helps to define the derivatives of transverse displacement at the top and bottom faces of the plate in terms
of the displacements u0, v0, θx, θy, uu, vu, ul and vl to overcome the problem of C1 continuity as mentioned
before.
260
Using the above equations, the in-plane displacement fields as given in Eq.(1-2) may be expressed as
U = b1u0 + b2 v0 + b3θ x + b4θ y + b5uu + b6vu + b7ul + b8vl (6)

V = c1u0 + c2 v0 + c3θ x + c4θ y + c5uu + c6vu + c7ul + c8vl (7)
where, the coefficients bi’s and ci’s are function of thickness coordinates, unit step functions and material
properties.
The generalized displacement vector { δ } for the present plate model can now be written with the help of Eq.
(3), (6) and (7) as
{δ } = {u0 v0 w0 θ x θ y uu vu wu ul vl wl }
T
Using strain-displacement relation and Eq. (1)-(5), the strain field may be expressed in terms of unknowns (for
the structural deformation) as
{ε } = {ε } + {ε }
L NL
where, the linear strain part is
{ε } =  ∂∂Ux ∂V ∂W ∂U ∂V ∂U ∂W ∂V ∂W 
+
∂y ∂z ∂x ∂y ∂z
+
∂x ∂z
+
∂x 
or {ε } = [ H ]{ε } (8)
and non-linear strain part can be written as

1 2(∂W ∂x)2 + 1 2(∂U ∂x) 2 + 1 2(∂V ∂x) 2 
 
{ε } = 1 2(∂W ∂y ) 2 + 1 2(∂U ∂y ) 2 + 1 2(∂V ∂y )2  or {ε } = 1 2 [ AG ]{θ }
T
(9)
NL
 2
NL
1 2(∂W ∂x) + 1 2(∂U ∂x) + 1 2(∂V ∂x) 

2 2
where,
{θ } = [∂W ∂x ∂W ∂y ∂U ∂x ∂U ∂y ∂V ∂x ∂V ∂y ] or {θ } = [ H G ]{ε } = [ H G ][ B]{δ }
and
{ε } = [u0 v0 w0 θ x θ y uu vu wu ul vl wl (∂u0 ∂x) (∂u0 ∂y ) (∂v0 ∂x) (∂v0 ∂y ) (∂w0 ∂x) (∂w0 ∂y ) (∂θ x ∂x)
(∂θ x ∂y ) (∂θ y ∂x) (∂θ y ∂y ) (∂uu ∂x) (∂uu ∂y ) (∂vu ∂x) (∂vu ∂y ) (∂wu ∂x) (∂wu ∂y )
(∂ul ∂x) (∂ul ∂y ) (∂vl ∂x) (∂vl ∂y ) (∂wl ∂x) (∂wl ∂y )]
and the elements of [H], [AG] and [HG] are functions of z and unit step functions.
With the quantities found in the above equations, the total potential energy of the system under the action of
transverse load may be expressed as
Π e = U s –U ext (10)
where Us is the strain energy and U ext is the energy due to the externally applied in-plane load.
Using Eq. (3) and (6), the strain energy (Us) is given by
∑ ∫∫∫{ε } ∫∫{ε }
1 1
Us = L
T
[Q k ]{ε }L dxdydz = L
T
[ D]{ε }L dxdy (11)
2 2
k =1
261
where,
n
[ D] = ∑ ∫ [H ] [Q ][H ]dz
k =1
T
k (12)
and the energy due to application of external in-plane load can be calculated as
n
∑ ∫∫∫{ε } ∫∫{ε } [G]{ε }dxdy

1 1
U ext . = NL
T
[ S i ]{ε }NL dxdydz = T (13)
2 2
k =1
where,
n
[G ] = ∑ ∫ [H ] [S ][H ]dz
k =1
T i
and the stress matrix [Si] may be expressed in terms of in-plane stress components of the i-th layer as
σ xx τ xy 0 0 0 0 
 
 τ xy σ yy 0 0 0 0 
 0 0 σ xx τ xy 0 0 
S  = 
i 
   0 0 τ xy σ yy 0 0 
 
 0 0 0 0 σ xx τ xy 
 
 0 0 0 0 τ xy σ yy 
In the present problem, a nine-node quadratic element with eleven degrees of field variables (u0, v0, w0, θx, θy, uu,
vu, wu, ul , vl and wl) per node is employed.
Using finite element method the generalized displacement vector {δ} at any point may be expressed as
{δ } = ∑Ni {δ }i (14)
i =1
{δ } = {u0 v0 w0 θ x θ y uu vu wu ul vl wl }
T
where, as defined earlier, δi is the displacement vector corresponding
to node i, Ni is the shape function associated with the node i and n is the number of nodes per element, which is
nine in the present study.
With the help of Eq. (14), the strain vector {ε} that appeared in equation (8) may be expressed in terms of
freedom as
{ε } = [ B]{δ } (15)
Where [B] is the strain-displacement matrix in the Cartesian coordinate system.
The elemental potential energy as given in Eq. (10) may be rewritten with the help of Eq. (11)-(15) as
∬ ∬{δ } [ B] [G ][ B]{δ } dxdy

1 1
Пe = {δ }T [ B ]T [ D ][ B ]{δ } dxdy − T T
2 2
1 1
= {δ } [ K e ]{δ } − λ {δ } [ K G ]{δ }
T T
(16)
2 2
262
where,
[ K e ] = ∫ [ B ]T [ D ][ B ] dx (17)
[ KG ] = ∫ [ B ]T [G ][ B ] dx (18)
where, [Ke] and [KG] are stiffness matrix and geometrical stiffness matrix.
The equilibrium equation can be obtained by minimizing Пe as given in Eq. (16) with respect to {δ} as
[ K e ]{δ } = λ [ KG ]{δ } (19)

where λ is a buckling load factor.
The skyline technique has been used to store the global stiffness matrix in a single array and simultaneous
iteration technique is used for solving the buckling equation.
3. NUMERICAL RESULTS
In this section, the examples on the buckling analysis of laminated sandwich plate having a soft core is analyzed
by using the proposed FE model based on higher order zigzag theory. The accuracy and applicability of the
proposed FE model based are demonstrated by solving a number of problems having different features. The
results obtained are compared with the published results and finally new results are also reported.
The non-dimensional quantities used, to show different results are as follows
λl 2
Buckling load, λ=
h 2 ETf
where z and ETf are the location across the depth and transverse modulus of elasticity of face layer
respectively.
3.1. Laminated Composite Plate (0/90/0) with Different Boundary Conditions
In this example a simply supported three layer composite plate is considered for the analysis. The layers are of
equal thickness. The material properties used in this problem are E1/E2 = 40, G12 = G13 = 0.6E2, G23 = 0.5 E2, ν12
=ν13 =ν23 = 0.25. In Table 1, the results for the non-dimensional critical biaxial buckling loads are presented
mainly to study the rate of convergence and also for validation for the plate problems with different boundary
conditions. The thickness ratio (a/h) is considered as 10. The full plate is analyzed taking mesh size (4×4), (8×8),
(12×12), (16×16) and (20×20). It may be observed in Table 1 that the non dimensional buckling loads are
converged at mesh size (16×16). A mesh size (16×16) is taken for all subsequent buckling analysis to get
sufficiently accurate results. For the comparison of the present results, the results published based on FSDT
(Ferreira et al. 2011a; Ferreira et al. 2011c) and HOZT ( Rodrigues et al. 2011) are considered. A good
agreement between the results may be observed from Table 1.
Table 1: Comparison of critical bi-axial buckling load for laminated

composite plates with different boundary conditions
Ref. SSSS CCCC
*
Present (8×8) 10.3832 13.7065
Present (12×12) 10.1487 13.4670
Present (16×16) 10.1485 13.4650
Present (20×20) 10.1485 13.4650
Rodrigues et al. [11] 10.1907 13.6629
Ferreira et al. [3] 10.1969 13.2919
Ferreira et al. [4] 10.2088 13.6581
*mesh size indicated in parenthesis
263
3.2. Multi-Layered Laminated Sandwich Plate (0/90/0/C/0/90/0)
The effect of soft core is investigated by analyzing a simply supported square laminated sandwich plate with
soft core. The ratio of the thickness of core to thickness of face sheets (tc/tf) is considered as 10. The material
properties for the composite face sheets are E1 = 131 GPa, E2 = E3 = 6.9 GPa, G12 = G23 = 6.9 GPa, G13 = 6.2
GPa, ν12 =ν13 = 0.22, ν23 = 0.49 and that of isotropic core are E1 = 6.9×10-3 GPa, ν= 10-5. In Table 2, the non
dimensional critical buckling load is reported for different thickness ratio (a/h) with different boundary
conditions mainly; SSSS, SCSC, CCCC and CFCF. The values show expected variations.
Table 2: Variation of critical bi-axial buckling load for different thickness ratio of multi-layered sandwich plate
(0/90/0/C/0/90/0) with various boundary conditions
Thickness Boundary Conditions

Ratio (a/h) CFCF SSSS SCSC CCCC
2 0.00661 0.00662 0.00663 0.00664
4 0.01123 0.01237 0.01248 0.01257
10 0.02982 0.03368 0.03468 0.03489
20 0.1004 0.1353 0.1453 0.1463
30 0.2271 0.2833 0.3036 0.3075
40 0.3469 0.4627 0.4996 0.5492
50 0.4786 0.6616 0.7368 0.8029
60 0.628 0.8661 0.9979 1.1054
70 0.791 1.0663 1.2725 1.4385
80 0.9651 1.2559 1.552 1.793
90 1.1475 1.4311 1.8297 2.1609
100 1.3352 1.5906 2.1005 2.5354
4. CONCLUSION
An improved C0 plate finite element (FE) model has been developed in this paper for the buckling analysis of
laminated sandwich plate with soft core. The higher order zigzag shear deformation theory is used to model the
in-plane displacement fields and a quadratic displacement field is used to define the transverse displacement in
the proposed model. The continuity requirement of the derivates of transverse displacement is circumvented by
effectively choosing the nodal field variables. So as there is no need to use penalty functions in the formulation
as used by many previous researchers. A nine node C0 quadratic plate finite element is successfully
implemented in the present model. Few numerical examples are solved for different problems of laminated
composite and sandwich plates. The results obtained by using the present FE model are successfully compared
with published results. The present FE model may, therefore, be recommended for the accurate analysis of
laminated sandwich plates with/without soft cores.
REFERENCES
[1] Noor A.K.. (1975). Stability of multilayered composite plates, Journal of Fiber Science and Technology, 8,
81-99.
[2] Noor, A.K.., Peters, J.M. and Burton, W.S. (1994). Three-dimensional solutions for initially stressed
structural sandwiches. ASCE J. Eng. Mech., 120(2), 284-303.
[3] Ferreira, A.J.M., Castro, L.M., Roque, C.M.C., Reddy, J.N. and Bertoluzza S. (2011). Buckling analysis of
laminated plates by wavelets. Computers and Structures, 89(7-8), 626-630.
264
[4] Ferreira, A.J.M., Roque, C.M.C., Neves, A.M.A. Jorge, R.M.N., Soares, C.M.M. and Liew, K.M. (2011).
Buckling and vibration analysis of isotropic and laminated plates by radial basis function, Composites: Part
B, 42, 592-606.
[5] Lo, S.H., Zhen, W., Sze, K.Y. and Wanji, C. (2011). An Improved In-Plane Displacement Model for the
Stability Analysis of Laminated Composites with General Lamination Configuration. Composite and
Structures, 93, 1584-1594.
[6] Sheikh, A.H. and Chakrabarti, A. (2003). A new plate bending element based on higher order shear
deformation theory for the analysis of composite plates. Finite Element in Analysis and Design, 39,
883-903.
[7] Srinivas, S. (1973). A refined analysis of composite laminates. Journal of Sound and Vibration, 30,
495-507.
[8] Kapuria, S., Dumir, P. C. and Jain, N.K. (2004). Assessment of zigzag theory for static loading, buckling,
free and force response of composite and sandwich beams, Composite Structures, 64, 317-327.
[9] Pandit, M.K., Sheikh A.H. and Singh, B.N. (2008). Buckling of laminated sandwich plates with soft core
based on an improved higher order zigzag theory. Thin-Walled Structures, 46, 1183-1191.
[10] Chakrabarti, A., Chalak, H.D., Iqbal, A. and Sheikh, A.H. (2011). A new fe model based on higher order
zigzag theory for the analysis of laminated sandwich beam with soft core. Composite Structures, 93,
271-279.
[11] Rodrigues, J.D., Roque, C.M.C., Ferreira, A.J.M., Carrera E. and Cinefra, M. (2011). Radial basis
functions-finite differences collocation and a unified formulation for bending, vibration and buckling
analysis of laminated plates, according to murakami’s zigzag theory. Composite Structures, 93(7),
1613-1620.
265
Paper No.: 044
NONLINEAR ANALYSIS OF SQUARE CONCRETE FILLED STEEL TUBES

UNDER CONSTANT AXIAL LOAD AND MONOTONIC LATERAL LOADING
A. Subbulakshmi 1 and P. Jayabalan 2
1
M.Tech Student, Dept. of Civil Engineering, National Institute of Technology, Tiruchirappalli
2
Professor, Dept. of Civil Engineering, National Institute of Technology, Tiruchirappalli
Email: subbulakshmigce@gmail.com, pjeya@nitt.edu
ABSTRACT:
In the present paper, cantilever square concrete filled steel tubular (CFST) beam-columns are considered to
study the effects of various parameters on the ultimate lateral load. Parameters are the width of steel tube,
thickness of steel tube, length to width ratio, yield strength of steel, compressive strength of concrete and axial
load level. Finite element model is created for the cantilever square CFST members under constant axial load
and monotonically increasing lateral load at the free end by using the finite element package ABAQUS. Taguchi
method is used to reduce the number of experiments for design of experiments approach. Nonlinear regression
equation is proposed from the design of experiments for finding the ultimate lateral load of cantilever square
CFST beam-columns.
KEYWORDS: CFST, Beam-column, Monotonic lateral load, Finite element analysis, Ultimate lateral load
1. INTRODUCTION
Concrete filled steel tubular (CFST) members combine the advantages of both the structural steel and concrete
materials properties. Steel members have good ductility and tensile strength while the concrete members have
high compressive strength and stiffness. Steel tube confines the concrete core which produces high compressive
strength and concrete core prevents inward local buckling of the steel tube. So CFST members have been
widely used due to their excellent seismic resistant properties such as high strength, stiffness, ductility and
energy absorption capacities.
Tomi et al. [1] studied the beam-columns for pin-pin end conditions. Their results showed that low B/t ratio
specimens had strain hardening behavior in moment curvature relationship. Hanbin et al. [2] studied the strength
deformation behavior of CFST stub columns. They concluded that CFST columns had high strength and
ductility compared to hollow steel columns. Schneider [3] studied the behavior of short CFST columns under
concentric axial compression. Their results indicated that circular columns offered more ductility than square
and rectangular columns. Brian Uy [5] conducted the experimental study on the strength behavior of short CFST
columns under combined axial load and bending moment to consider the effect of concrete strength and
slenderness. Shanmugam et al [6] presented the review of various research works carried out on the composite
columns. Elchalakani et al [7] performed the experimental study on the flexural behavior of CFST members
subjected to pure bending. They found that void filling prevented the local buckling. Varma et al [8] studied the
flexural force deformation behavior of high strength CFST beam-columns for pin-pin conditions. They
concluded that curvature ductility of high strength CFST beam-columns reduced with increase in axial load
level and width to thickness ratio.
Sakino et al [9] studied the axial load behavior of CFST columns and also they proposed the stress strain model
for concrete and steel. Ellobody et al [10] developed the nonlinear finite element model for axially loaded
square and rectangular CFST columns. Dalin Liu [11] conducted the experimental and analytical study on the
behavior of high strength rectangular CFST columns under eccentric loading. Zhong et al [12] conducted the
experimental study on the behavior of concrete filled stiffened thin walled steel columns. Local buckling of steel
tube was delayed due to stiffeners in the column. Manojkumar [13] presented the strength deformation behavior
of CFST members under flexure. Brian Uy [14] performed the test on the short and slender concrete filled
stainless steel columns under the axial compression and combined axial compression and bending. Present study
focuses on the formulation of equations for finding the ultimate lateral load of cantilever square CFST beam-
columns under constant axial load and monotonic lateral load.
3. FINITE ELEMENT MODELING
ABAQUS package is used to develop the finite element model for cantilever square CFST beam-columns under
constant axial load and monotonic lateral load. S4R and C3D8R elements are used to model the steel tube and
concrete core. Elastic Plastic model and concrete damaged plasticity models are used to describe the material
behavior of steel and concrete. Convergence study is done to get the accurate results with proper mesh size.
Interaction between the steel tube and concrete core is simulated by using the surface based interaction. Hard
contact model and Coulomb friction models are used in the normal direction and tangential directions between
the surfaces. Loading, boundary conditions and stress distributions of cantilever square CFST beam-columns are
shown in Fig 1 and Fig 2.
Fig 1: Loading and Boundary Conditions of Cantilever Square CFST Beam-Columns
Fig 2: Stress Distribution of Cantilever Square CFST Beam-Columns
267
4. EMPIRICAL MODELING
4.1 Design of Experiments
Design of experiments (DOE) is the powerful statistical technique to study the effects of multiple
variables simultaneously. Taguchi method is used to reduce the number of experiments and also to find the
proper combination of the parameters. Taguchi uses an orthogonal arrays to combine the parameters affecting
the response and the levels at which they should be varied. L27 orthogonal array is chosen here for finding the
proper combinations of the parameters since it consists six parameters each at three levels. B/t ratio is
considered as 20 to 100 for combinations. Parameter combinations are shown in Table 1. Finite element model
is created for each combination and results are obtained from the finite element analysis using ABAQUS
software. Ultimate lateral loads obtained from the finite element model (HuFEM) are compared with the
Eurocode-4 design code results (HuEC4) which are shown in Table 1.
Table 1: Combinations of Parameters for L27 Orthogonal Array and Comparisons of Ultimate Lateral Load
No Specimen B t B/t L L/B fy fc n HuFEM HuEC4 HuFEM /

(mm) (mm) (mm) (MPa) (MPa) (kN) (kN) HuEC4
1 S1 100 2 50 300 3 290 20 0.2 36.47 34.87 1.046
2 S2 100 2 50 300 3 290 38.7 0.4 37.82 39.4 0.96
3 S3 100 2 50 300 3 290 59.3 0.6 33.99 38.83 0.875
4 S4 100 3.5 28.6 500 5 386 20 0.2 43.69 41.16 1.061
5 S5 100 3.5 28.6 500 5 386 38.7 0.4 40.36 41.78 0.966
6 S6 100 3.5 28.6 500 5 386 59.3 0.6 30.36 34.64 0.876
7 S7 100 5 20 700 7 471 20 0.2 45.2 45.13 1.002
8 S8 100 5 20 700 7 471 38.7 0.4 37.59 39.91 0.942
9 S9 100 5 20 700 7 471 59.3 0.6 26.7 31.06 0.86
10 S10 150 2 75 750 5 471 20 0.4 46.12 47.8 0.965
11 S11 150 2 75 750 5 471 38.7 0.6 41.98 48.89 0.859
12 S12 150 2 75 750 5 471 59.3 0.2 66.85 62.95 1.062
13 S13 150 3.5 42.9 1050 7 290 20 0.4 32.03 35.31 0.907
14 S14 150 3.5 42.9 1050 7 290 38.7 0.6 29.58 34.33 0.862
15 S15 150 3.5 42.9 1050 7 290 59.3 0.2 44.09 46.13 0.956
16 S16 150 5 30 450 3 386 20 0.4 134.13 118.67 1.13
17 S17 150 5 30 450 3 386 38.7 0.6 117.35 102.44 1.146
18 S18 150 5 30 450 3 386 59.3 0.2 185.12 175.49 1.055
19 S19 200 2 100 1400 7 386 20 0.6 30.38 34.61 0.878
20 S20 200 2 100 1400 7 386 38.7 0.2 51.49 50.39 1.022
21 S21 200 2 100 1400 7 386 59.3 0.4 67.28 71.83 0.937
22 S22 200 3.5 57.1 600 3 471 20 0.6 114.43 112.15 1.02
23 S23 200 3.5 57.1 600 3 471 38.7 0.2 210.42 209.72 1.003
24 S24 200 3.5 57.1 600 3 471 59.3 0.4 215.12 232.8 0.924
25 S25 200 5 40 1000 5 290 20 0.6 67.01 61 1.099
26 S26 200 5 40 1000 5 290 38.7 0.2 114.65 110.78 1.035
27 S27 200 5 40 1000 5 290 59.3 0.4 109.64 126.09 0.87
4.2 Main Effects Plots
After completing the experiments as per Taguchi method, main effects plots are plotted for ultimate
lateral load using the MINITAB software. Main effects plot gives the direct effect of parameters on the
response. It is plotted by considering the means of responses at each level of parameters. Main effects plot of all
268
the parameters on the ultimate lateral load is shown in Fig 3. Response table for means of ultimate lateral load is
given in Table 2. Increase in width of steel tube, thickness of steel tube, yield strength of steel and compressive
strength of concrete increase the ultimate lateral load and increase in length to width ratio and axial load level
reduce the ultimate lateral load.
Table 2: Response Table for Means of Ultimate Lateral Load
Level B (mm) t (mm) L/B fy (MPa) fc (MPa) n

1 36.91 45.82 120.54 56.14 61.05 88.66
2 77.47 84.45 62.30 77.80 75.69 80.01
3 108.94 93.04 40.48 89.38 86.57 54.64
Delta 72.03 47.22 80.06 33.24 25.52 34.02
Rank 2 3 1 5 6 4
Length to width ratio has the most significant effect on the ultimate lateral load. Rank of each parameter
is given in response table based on the deltas of means of ultimate lateral load of each level of all parameters.
Delta is the difference between maximum and minimum values of ultimate lateral load for each parameter. L/B
ratio has the first rank. After that width of steel tube, thickness of steel tube, axial load level, yield strength of
steel and compressive strength of concrete have the ranks of 2, 3, 4, 5 and 6. Compressive strength of concrete
is the least influencing parameter on the ultimate lateral load.
Fig 3: Main Effects Plots for Ultimate Lateral Load
4.3 Nonlinear Regression Equation
Response surface method (RSM) is a combination of mathematical and statistical techniques used in
empirical study of relationships and optimization of response in which several parameters influence the
response. Response surface method is used to propose the nonlinear regression equation for finding the ultimate
lateral load of cantilever square CFST beam-columns based on the finite element model results of 27
269
experiments using the MINITAB software. Nonlinear regression equation for ultimate lateral load is given as
follows:
L
H = −163.436 + 1.2663 B + 62.4752 t − 65.5517 + 0.559 f + 1.1637 f + 82.0778 n − 0.0018 B
B
L
− 6.6763 t + 4.5538 − 0.0005 f − 0.0065 f − 208.9167 (1)
B
Where Hu - Ultimate Lateral Load

B - Width of Steel Tube
t - Thickness of Steel Tube
L/B - Length to Width Ratio
fy - Yield Strength of Steel
fc - Compressive Strength of Concrete
n - Axial Load Level
Ultimate lateral load for the above 27 experiments are found out by using nonlinear regression equation
(empirical model). Results are plotted between the finite element model results and nonlinear regression
equation results for ultimate lateral load which is shown in Fig 4.
250
Nonlinear Regression Hu (kN)
200
150
100
50
0
0 50 100 150 200 250
FEM Hu (kN)
Fig 4: Finite Element Model Vs Nonlinear Regression Equation Ultimate Lateral Load
4.4 Surface Plot and Contour Plot
Response surface plots indicate the effect of any two variables together on the response. Surface plots
are plotted between the variables and the response using MINITAB software. Contour plots are used to obtain
the proper combination of the two variables for the response. Surface plot and contour plots of ultimate lateral
load Vs width and thickness of steel tube are shown in Fig 5 and Fig 6.
270
Fig 5: Surface Plot of Ultimate Lateral Load Vs Width and Thickness of Steel Tube
Fig 6: Contour Plot of Ultimate Lateral Load Vs Width and Thickness of Steel Tube
4.5 Validation of Nonlinear Regression Equation
Experimental work conducted on the cantilever square concrete filled steel tube beam-columns by Matsui et al
[4] is taken for validation of nonlinear regression equation. Experimental (HuE) and nonlinear regression
equation results (HuN) are compared for ultimate lateral load which is given in Table 3. Fig 7 shows the plot
between the experimental and predicted ultimate lateral loads from the equation. Validation of nonlinear
regression equation of ultimate lateral load for various axial load levels with B/t ratio of 22 is shown in Fig 8.
271
Table 3: Validation of Nonlinear Regression Equation for Ultimate Lateral Load
No Specimen B (mm) t (mm) L/B fy (MPa) fc (MPa) n HuE (kN) HuN (kN) HuN / HuE
1 R22C20C 100 4.4 5 407 34.1 0.2 61.6 61.79 1
2 R22C30C 100 4.4 5 407 34.1 0.3 61.4 59.55 0.97
3 R22C40C 100 4.4 5 407 34.1 0.4 49.8 53.14 1.07
4 R22C50C 100 4.4 5 407 34.1 0.5 44.6 42.54 0.95
5 R22C60C 100 4.4 5 407 34.1 0.6 37 27.77 0.75
75
Nonlinear Regression Hu (kN)
60
45
30
15
0
0 15 30 45 60 75
Experimental Hu (kN)
Fig 7: Experimental Vs Nonlinear Regression Equation Ultimate Lateral Loads
70
Ultimate Lateral Load, Hu (kN)
60
50
40
30
20
Model
10
Matsui et al
0
0 0.2 0.4 0.6 0.8
Axial Load Level (n)
Fig 8: Validation of Nonlinear Regression Equation for Ultimate Lateral Loads with B/t Ratio = 22
272
5. CONCLUSIONS
Cantilever square concrete filled steel tubular beam-columns are modeled using finite element package
ABAQUS. Effects of various parameters such as width of steel tube, thickness of steel tube, length to width
ratio, yield strength of steel, compressive strength of concrete and axial load level on the ultimate lateral load of
cantilever square CFST beam-columns are studied. Nonlinear regression equation is proposed to find out the
ultimate lateral load of the cantilever square CFST beam-columns.
REFERENCES
[1] Masahide Tomii., and Kenji Sakino. (1979). “Experimental studies on the ultimate moment of concrete
filled steel tubular beam-columns.” Transactions of Architectural Institute of Japan, 275, 55-63.
[2] Hanbin, Ge., and Tsutomu Usami. (1992). “Strength of concrete filled thin walled steel box columns:
Experiment.” Journal of Structural Engineering, ASCE, 118 (11), 3036-3054.
[3] Stephen Schneider, P. (1998). “Axially loaded concrete filled steel tubes.” Journal of Structural
Engineering, ASCE, 124 (10), 1125-1138.
[4] Chiaki Matsui., Keigo Tsuda., Yujiro Yamaji., and Takashi Fujinaga. (1998). “Structural Performance and
Axial Load limit of Concrete Filled Steel Square Tubular Columns.” Journal of Structural and
Construction Engineering, AIJ, 504, 103-110.
[5] Brian Uy. (2000). “Strength of concrete filled steel box columns incorporating local buckling.” Journal of
Structural Engineering, ASCE, 126 (3), 341-352.
[6] Shanmugam, N.E., and Lakshmi, B. (2001). “State of art of report on steel concrete composite columns.”
Journal of Constructional Steel Research, 57, 1041-1080.
[7] Elchalakani, M., Zhao, X.L., and Grzebieta, R.H. (2001). “Concrete filled circular steel tubes subjected to
pure bending.” Journal of Constructional Steel Research, 57, 1141-1168.
[8] Amit Varma, H., James Ricles, M., Richard Sause., and Le-Wu Lu. (2002). “Experimental behavior of high
strength concrete filled steel tube beam-columns.” Journal of Structural Engineering, ASCE, 128 (3), 309-
318.
[9] Kenji Sakino., Hiroyuki Nakahara., Shosuke Morino., and Isao Nishiyama. (2004). “Behavior of centrally
loaded concrete filled steel tube short columns.” Journal of Structural Engineering, ASCE, 130(2), 180-
188.
[10] Ehab Ellobody., and Ben Young. (2006). “Nonlinear analysis of concrete filled steel SHS and RHS
columns.” Thin Walled Structures, 44, 919-930.
[11] Dalin Liu. (2006). “Behaviour of eccentrically loaded high strength rectangular concrete-filled steel tubular
columns.” Journal of Constructional Steel Research, 62, 839-846.
[12] Zhong Tao., Lin-Hai Han., and Dong-Ye Wang. (2007). “Experimental behaviour of concrete-filled
stiffened thin-walled steel tubular columns.” Thin Walled Structures, 45, 517-527.
[13] Manojkumar Chitawadagi, V., and Mattur Narasimhan, C. (2009). “Strength deformation behaviour of
circular concrete filled steel tubes subjected to pure bending.” Journal of Constructional Steel Research,
65, 1836-1845.
273
[14] Brian Uy., Zhong Tao., and Lin-Hai Han. (2011). “Behaviour of short and slender concrete-filled stainless
steel tubular columns.” Journal of Constructional Steel Research, 67, 360-378.
[15] Eurocode 4: Design of composite steel and concrete structures, Part 1.1: General rules and rules for
buildings.
[16] ABAQUS Standard User’s Manual, Version 6.11.
274
Paper No.: 045
RELIABILITY BASED DESIGN OF TMD SYSTEM CONSIDERING SYSTEM

PARAMETER UNCERTAINTY IN SEISMIC VIBRATION CONTROL
Bijan Kumar Roy1 and Subrata Chakraborty2

1
University Institute of Technology, Burdwan-713104, West Bengal, India
2Department of Civil Engineering, Bengal Engineering and Science University, Shibpur, India
E-mail: bijan.roy@rediffmail.com, schak@civil.becs.ac.in
ABSTRACT:
The optimum design of tuned mass damper (TMD) parameters is one of the most important issues with regard
to its performance in seismic vibration control. The structural displacement covariance of the protected system
under stochastic earthquake load is usually considered as the performance measure assuming deterministic
system. A major limitation of such deterministic approach is that the uncertainties in the system parameters
cannot be included in the optimization process. But, the efficiency of the system may be drastically affected if
the TMD is not tuned to the vibrating mode for which it is designed to suppress due to unavoidable presence of
uncertainty. The present study deals with reliability based design of TMD system in seismic vibration control
considering system parameter uncertainty. The maximum structural displacement is considered as the
performance index to define the failure of the system. With the aid of matrix perturbation theory using first
order Taylor series expansion, the total probability theory is used to evaluate the unconditional response of the
structure under parameter uncertainty. Subsequently, the unconditional failure probability of the structures is
obtained which is used as the objective function to obtain the optimum TMD parameters. A numerical study is
performed to elucidate the effect of parameter uncertainty in the optimization of TMD parameters and overall
safety of the structures.
KEYWORDS: Earthquake, Control, TMD, Parameter Uncertainty, Reliability
1. INTRODUCTION
The use of a TMD has been demonstrated to be very effective in protecting structures from suffering major
damage because of lateral loadings [1-2]. In the field of passive vibration control, the TMDs are amongst the
oldest control devices. The optimal design of TMD assuming deterministic system parameters are well
developed [3-5]. The most commonly used approach of damper parameter optimization assumes the load like
earthquake or wind actions as the only source of randomness. Loads are suitably modelled as a stochastic
process and the stochastic structural optimization (SSO) is performed by using the structural displacement
covariance of the protected system as the performance measure. The approach assumes that all parameters
except the load are unaffected by any source of uncertainty. A major limitation of such deterministic assumption
is that the uncertainty in the performance-related decision variables cannot be included in the optimization
process. But, the design of a TMD system based on a single nominal model of the system may fail to create a
control system ensuring satisfactory performance. The unavoidable presence of uncertainty in the system
parameters is expected to cause a variation of the optimum value of the damper parameters designed based on
deterministic assumption of system parameters [6]. The efficiency of a damper may reduce significantly if the
parameters are not properly tuned to the vibrating mode it is designed to suppress. Thus, for an efficient design
of damper system, the uncertainty associated with excitation as well as mechanical model of the structure should
be explicitly taken into account. The developments in the field of vibration control by using various passive
devices and considering system parameter uncertainty are notable [8-10].
The present study deals with the reliability based design optimization (RBDO) of TMD system to consider the
effect of system parameters uncertainty in the vibration control of primary structures subjected to stochastic
earthquake load. The probability of failure (pf ) with respect to maximum displacement of the primary structure
is considered as the performance index. With the aid of matrix perturbation theory using first order Taylor series
expansion, the total probability concept [11] is used to evaluate the unconditional response of the primary
structure considering the effect of system parameter uncertainty. For this, the conditional second order
information of response quantities are obtained in random vibration framework using the state space
formulation. Subsequently, the unconditional RMS top floor displacement of the primary structure is evaluated.
These unconditional responses are then used to obtain the unconditional pf based on first passage failure
criterion, serving as the objective function in the optimization of the TMD system. The design optimization is
also performed based on minimization of the unconditional RMS response of the top floor in order to demarcate
the importance of the RBDO procedure. A TMD attached to the primary system is considered for numerical
study to elucidate the effect of parameter uncertainty on the optimum performance of TMD system.
2. STOCHASTIC DYNAMIC RESPONSE OF TMD - STRUCTURE SYSTEM
TMD consists of an additional mass connected to a main system by a spring and a damper. The mechanical
model of a TMD structure system is represented in Figure-1which is described by a single mass ms, linked with
the base by a linear spring ks and a dashpot cs. It is excited by a base acceleration &&
yb (t ) , and it is connected with
a secondary mass mT by a spring kT and a dash pot cT. The dynamic equilibrium of the equation of the system
can be expressed as,
MY(t) & + KY(t) = r&&

&& + CY(t) yb (t ) (1)
& and Y(t)

where, Y(t) , Y(t) && , denote the relative displacement, velocity and acceleration vectors, r is the
influence coefficient vector; M C and K are respectively the mass, damping and stiffness matrices of the
combined system. The well-known Kanai-Tajimi stochastic process [12,13] which is able to characterise input
frequency content for a wide range of practical situations is adopted in the present study and && yb (t ) is
represented as,
yb (t ) = &&
&& x f (t ) + w(t )
(2)
x f (t ) + 2ξ f ω f x& f (t ) + ω f 2 x f (t ) = −w(t )
&&
In the above x f (t ) is the response of the filter representing the ground, ξ f and ω f are, respectively, the
damping ratio and the frequency of this filter, w(t) being the white noise process, representing the excitation at
the bed rock.
Figure 1 Linear model of TMD system
276
Introducing the state space vector, Z = ( yT , yS , x f , y&T , y& S , x& f )T : the stochastic response of the system can be
obtained by using the covariance matrix RZZ obtained by solving the following Lyapunov equation.
ARZZ + RZZ AT + B = 0 (3)
Where the matrix B (of size 6x6) has all null elements, except the last element on the main diagonal:
[B]6,6 = 2πS0 , where S0 is the intensity of the white noise. The system matrix A is:
 0 0 0 1 0 0 
 0 0 0 0 1 0 
0 0 0 0 0 1 (4)
A =  −ω 2 T ω 2T ω2f − 2ξ T ω T 2ξ T ω T 2ξ f ω f 

ω T − ( µω 2 T + ω 2 S ) ω2f 2ξ T ω T µ − (2ξ T ω T µ + 2ξ S ω S ) 2ξ f ω f 
2
 0 0 −ω 2 f 0 0 − 2ξ f ω f 
Various parameters in the above matrix are as follows: ωT = kT / mT , ωS = k S / mS , ξT = cT / 2 mT kT ,

ξ S = cS / 2 mS k S and, µ = mT / mS ; ωS and ωT are the natural frequency of the primary structure and TMD,
respectively; ξS and ξT are the damping ratio of concerned system and µ is the mass ratio. The state space
covariance matrix R ZZ (of size 6x6) is obtained by solution Eq. 3. The RMS displacement of TMD and the
structure can be obtained as,
σ yT = R zz (1,1) , σ ys = R zz ( 2 ,2 ) (5)
3. OPTIMIZATION OF TMD PARAMETERS
The problem of optimization of TMD system requires determining the tuning ratio ( γ ) and the damping
coefficient ξT . The design vector can be thus defined as b = (γ , ξ ) :. The conventional SSO problem under
T
random earthquake is generally formulated as the search of a suitable set of design variables, over a possible
admissible domain Ω , to minimize the desired objective function. The optimization problem so defined for
system subject to stochastic load can be transformed into a standard nonlinear programming problem and the
design variables can be obtained for a known mass ratio, damping ratio and frequency of the primary structure.
The objective function typically considered in the SSO is the mean square responses (displacement,
acceleration, stress etc.) or the failure probability or can be the total life-cycle cost of the structure. In the
present study, the RMS displacement of the primary structure is considered as the objective function. The SSO
problem so defined leads to a standard nonlinear programming problem [14],
Find b = ( γ , ξT ) to minimize: f = σ ys (6)
4. TMD PARAMETERS OPTIMIZATION UNDER UNCERTAINTY: RBDO APPROACH
The matrix A and B as described by Eq. 3 are depended on various system parameters characterizing the
primary structure, filter model and earthquake load. The response statistic evaluated under stochastic earthquake
load to solve the SSO problem as described by Eq. 6 intuitively assumes that these parameters are completely
known. However, the uncertainty in these parameters may lead to an unexpected excursion of responses
affecting the desired safety of structure. Therefore, the uncertainties considered in the system parameter in the
present study are in ωs , ξ s , ω f , ξ f and S0 which are assumed to be independent Gaussian random variables. To
277
include the effects of such parameter uncertainties, the total probability concept is one of the most important
theorems used to estimate the unconditional stochastic response of structures involving sensitivity analysis of
stochastic dynamic system. In the subsequent sections, the related formulations are briefly discussed.
The random system parameters as mentioned above are denoted by a vector u. To obtain the sensitivity of
response, the first order derivative of the basic Lyapunov equation can be obtained by differentiating Eq. 3 with
respect to the i-th parameter ui as following,
∂
A R, + R , Α T + B1 = 0, where, Β1 = Α ,ui R + RΑ ,T + (B ) (7)
ui ui ui ∂ui
The sensitivity of response (the RMS displacement as considered herein) can be obtained directly by
differentiating the corresponding expression of Eq. 5 with respect to the i-th random variable ui as following,
∂ 1 R ,ui (2, 2)
(σ y ) i.e. σ y , u = (8)
∂ui s s i 2 R (2, 2)
In which, R ,ui (2, 2) is obtained by solving Eq. 7. Now, taking the second order derivative with respect to the j-
th parameter of Eq.7, one can further obtain the following,
ΑR,u u +R,u u AT + Β2 = 0, Β2 = 2[Α,u R,u +R,u Α,T ] + [Α,u u R + RΑ,T ] (9)

i j i j i i i ui i j uiu j
The second order sensitivity of the RMS displacement can be obtained by differentiating Eq. 9 with respect to j-
th random variable uj as following:
 
σ ys ,u u = (1 / 2 R (2, 2)) R,u u (2, 2) − 0.5([R,u (2, 2)]2 / R (2, 2))  (10)
i j  i j i 
It can be noted that in order to obtain the first and second order sensitivities of RMS displacement the equations
need to be solved for each random variable involve in the problem. The second order sensitivity of the
covariance matrix can be obtained by solving Eq.9. Finally, using this result, the second order sensitivity of the
responses can be obtained from Eq. 10.The total probability concept is traditionally used to obtain the
unconditional response of the system under parameter uncertainty and suitable performance index is used in the
optimum design of TMD parameters based on such unconditional response. The approach is referred as the
RBDO. The responses of a structure subjected to stochastic earthquake load depend on the system parameters
and can be expanded in the Taylor series about the mean value of the random system parameter with the
assumptions that the random variation is small. The i-th such parameter ui can be viewed as the superposition
of the deterministic mean component ( ui ) with a zero mean deviatoric component ( ∆ui ) .The Taylor series
expansion of the RMS displacement of the primary structure about the mean values of the random system
parameters can be expressed as,
nv 1 nv nv
σ ys = σ ys ( ui ) + ∑ σ ys , u ∆ui + ∑ ∑ σ y , u u ∆ui ∆u j + .......... (11)
i =1
i 2 i =1 j =1 s i j
In the above, nv is the total number of random variables involve in the problem, σ ys ( ui ) is the mean part of the
RMS displacement of the primary structure. Assuming uncertain variables u are uncorrelated, the quadratic
approximation provides the expected value i.e. the unconditional RMS displacement as,
278
1 nv
σ ys = σ ys ( ui ) + ∑σ σ2 (12)
2 i = 1 ys , uiui ui
where, σ ui is the standard deviation of the i-th random parameter. Thus, knowing the second order information
of the response quantities using Eq. 10, the unconditional response can be obtained from Eq. 12. Finally, the
TMD parameters optimization problem as defined by Eq. 6 can be redefined for the RBDO as following,
1 nv
Find b=(γ , ξT )T to minimize: f = σ ys = σ ys ( ui ) + ∑ σ y , u u σu
2
(13)
2 i =1 s i i i
The standard gradient based techniques of optimization can be used to solve the problem. In the present study,
the standard MATLAB routine is used.
5. RELIABILITY OF TMD SYSTEM
In seismic reliability analysis, the TMD system is assumed to fail if the response crosses some threshold value
(referred as barrier). The reliability for the i-th limit state is thus determined by the first crossing of any
structural response xi to certain barrier level βi . For the TMD system considered herein, the first time bilateral
crossing of any response xi to the barrier level βi , within a time interval Td , implying failure, is expressed as
Fi = { xi βi } ; t ∈ [ 0, Td ] (14)
The conditional failure probability P ( Fi / u ) with regard to response xi based on the structural and excitation
model parameters specified by {u} can be estimated following classical Rice’s formulation as
P ( Fi ) ≅ 1 − exp  −αβi ({u}) Td  (15)

 
where α βi ({u}) is the conditional threshold-crossing rate for the i-th failure mode and Td is the duration of the
earthquake motion. With further assumption that the response is a stationary stochastic Gaussian process with
zero mean, the conditional crossing rate can be written as
σ y& s  
β2
α β i ({u} ) = exp  − i  (16)
πσ y  2σ 2 
s  ys 
where σ ys and σ y& s are the RMS of ys and y& s . In case of multiple failure events with nq number of limit state,
nq
F = ∑ Fi . This implies that the system fails if any one of the response xi exceeds its threshold βi . The mean
i =1
up-crossing rate of the system is then approximated by
nq
α = ∑ αβi ( u ) (17)
i =1
and the probability of failure P ( F ) of the controlled structural system can be approximated as
279
 Nq 
( )
P F {u} ≈ 1 − exp  − ∑ αβi ({u} ) Td  (18)
 i =1 
6. NUMERICAL STUDY
A primary structure modelled as a SDOF system with an attached TMD as shown in Fig.1, subjected to
earthquake excitation, is undertaken to elucidate the proposed RBDO of TMD system for seismic vibration
control of structures considering random system parameters. The power spectral density (PSD) of the white
noise process at bed rock, S0 is related to the standard deviation σ &&yb of the ground acceleration by:
( )
S 0 = (2ξ f σ &&y2 ) / {π 1 + 4ξ 2f ω f } . Unless mentioned otherwise, the following nominal values are assumed for
b
present numerical study: ms=5x105 kg, ks=3.5x108 N-m, ξ s = 3% , µ=3%, ωf=9π , ξf =0.4, S0=300 cm2/sec3.
The duration of the seismic ground motion is taken 20 sec. Based on these numerical values; the RMS
displacement of the unprotected system i.e. without TMD is computed to be 3.36 cm. The threshold
displacement value (β) defining the failure of the primary structure with respect to the maximum displacement
is taken as 4.5 cm.
The variations of optimum damping ratio and tuning ratio for varying mass ratio for different level of
uncertainty of the parameters are shown in figure 2 and 3 considering the unconditional failure probability as
objective function. The associated pf values are depicted in figure 4. The value of ξ s and S 0 are taken as 3%
and 300 cm2/sec3 for developing this plot. The optimum results considering probability of failure evaluated
based on conditional RMS displacement (as obtained Eq. 6) is also shown in the same plot for comparison.
Figure 2 The variation of optimum damping of Figure 3 The variation of optimum tuning ratio of TMD
TMD with varying mass ratio for different level with varying mass ratio for different level of uncertainty
of uncertainty
The results similar to figure 2 to 4 are also developed with varying damping ratio of primary structure for
different level uncertainty of the parameters and are shown in figure 5 to 7. Fig.8 shows the variations of
probability of failure with varying intensity of earthquake, quantified by the white noise intensity. For obvious
reason, the optimal failure probability is higher for higher earthquake intensity and the effect of uncertainty is
also more prominent.
From the results it is observed that the optimization results follow the general trend of deterministic case. The
effect of uncertainty i.e. visible differences of optimum results can be noted when the probability of failure is
comparatively higher. It is generally seen that the optimum tuning parameter is as such not affected much. But
there is a definite change in the optimum parameters and probability of failure considering uncertain system
parameters with respect to that of deterministic system parameter case.
280
Figure 4 The variation of pf with varying mass ratio Figure 5 The variation of optimum tuning ratio of
for different level of uncertainty TMD with varying damping ratio of structures for
different level of uncertainty
Figure 6 The Variation of optimum damping of Figure 7 The variation of pf with varying damping
TMD with varying damping ratio of structures for ratio with varying damping ratio of structures for
different level of uncertainty different level of uncertainty
The change in failure probability increases as the level of uncertainty increases. The optimum damping
parameters obtained assuming deterministic system parameters indicates the possibility of obtaining a less
probability of failure, which in reality will produce larger values. This is a potential problem in optimization
and warns the application of the damper without proper consideration of uncertainties. It may be observed from
the plots that the effect of uncertainty on safety of the primary system is more for comparatively lower damping
values. In this regard it may be pointed out here that the control devices are applied typically in reduction of
vibration level of flexible structure having smaller structural damping. Thus, the effect of uncertainty will be a
critical issue in such cases.
Figure 8 The variation of pf with varying seismic intensity
281
7. CONCLUSIONS
The reliability based optimization of TMD system of protection is studied for mitigation of seismic vibration
effect on the structures subjected to stochastic earthquake loading. The observations of optimization results are
in conformity with the well known facts in the application of TMD assuming deterministic system parameters
e.g. the probability of failure of structure is quite significantly reduced with increasing mass ratio and damping
ratio of structures as expected. However, when the system parameters uncertainties are considered, there is a
definite change in the optimum results. It is generally seen that the optimal tuning ratio of the TMD is not so
sensitive with respect system parameters uncertainty; however, there is definite change in the damping ratio of
the TMD. Though the uncertainty in the excitation parameters dominates the safety level, it is generally seen
that the uncertainties in system parameters have noticeable influence and cannot be neglected. In general, the
advantage of the TMD tends to reduce as the level of uncertainties increases as the probability of failure
increases with increasing uncertainty. It is evident that if the uncertainty in the parameters of the system is not
considered, the TMD performance is overestimated and may lead to an unsafe design. The design methodology
can be the extended to MDOF system is straightforward. However, it needs further study for interactive MDOF
complex system. The observation is made with earthquake load of specific model and needs to study further for
non-stationary earthquake model or can be extended to other random load like wind or wave etc.
REFERENCE
[1] Frahm, H. (1911) .Device for damping vibration of bodies. U.S. Patent No.958 -989.
[2] Ormondroyd, J and Den Hartog JP.(1928). The theory of the vibration absorber. Trans of American Soc
Mech Eng, 49(9), A9–A22.
[3] Warburton, G.B. and Ayorinde E.O.(1980). Optimum absorber parameters for simple systems. Earth
Engng and Struct Dyn. 8, 197–217.
[4] Thomson, A.G. (1980). Optimizing the tuned viscous dynamic vibration absorber with primary system
damping; a frequency locus method. J. of Sound and Vibration, 73, 469-472
[5] Warburton, G.B. (1981).Optimum absorber parameters for minimizing vibration response. Earth Engng
and Struct Dyn, 9, 251-262.
[6] Schuëller, G.I. and Jensen, H.A. (2008). Computational methods in optimization considering uncertainties-
an overview. Computer Method in Applied Mechanics and Engineering, 198 (1), 2–13
[7] Jensen, H.; Setareh , M. and Peek, R. (1992). TMDs for vibration control of System with uncertain
Properties. J. of Structural Engineering (ASCE), 118(2), 3285-3296.
[8] Papadimitriou, C.; Katafygiotis, L. S. and Au, S. K. (1997). Effects of structural uncertainties on TMD
design: a reliability based approach. J. Struct. Contr., 4(1), 65–88.
[9] Taflanidis, A.A.; Beck, J.L. and Angelides, D.C.(2007). Robust reliability-based design of liquid column
mass dampers under earthquake excitation using an analytical reliability approximation. Eng Struct, 29,
3525–37
[10] 10 Marano, G.C.; Sgobba, S.; Greco, R. and Mezzina, M (2008). Robust optimum design of tuned mass
dampers devices in random vibrations mitigation. J. of Sound and Vibration, 313, 472–492.
[11] Chaudhuri, A. and Chakraborty, S.(2004). Sensitivity evaluation in seismic reliability analysis of
structures. Computer Methods in Applied Mechanics and Engineering, 93, 59-68.
282
[12] Kanai, K. (1957) Semi-empirical formula for the seismic characteristics of the ground. Bulletin of
Earthquake Research Institute, University of Tokyo, 35, 309-325
[13] Tajimi , H. (1960). A statistical method of determining the maximum response of a building during
earthquake. In:Proc of 2nd World Conference on Earthquake Engineering, Tokyo, Japan.
[14] Nigam, N.C. (1972). Structural optimization in random vibration environment, AIAA Journal, 10(4): 551-
553
283
Paper No.: 046
PERFORMANCE EVALUATION OF CONCRETE AGAINST REBAR

CORROSION IN COMPOSITE CHLORIDE-SULFATE EXPOSURE
CONDITIONS
1
Bulu Pradhan
1
Assistant Professor, Dept. of Civil Engineering, Indian Institute of Technology Guwahati, India
Email: bulu@iitg.ernet.in
ABSTRACT :
The significant causes of deterioration of reinforced concrete structures are the durability problems. The
durability problems can be caused by aggressive external agents namely, chloride ions and sulfate ions. Rebar
corrosion in concrete is the most significant durability problem for reinforced concrete structures. In the present
study, the corrosion performance of steel in concrete exposed to composite chloride-sulfate environment is
evaluated. For this purpose an experimental investigation is carried out wherein different corrosion parameters
have been measured. Ordinary Portland cement, w/c ratios of 0.50 and 0.55 and steel bar (TMT steel) of
diameter 12 mm were used in the preparation of test specimens. The specimens were subjected to alternate
wetting and drying cycles with seven days of immersion of specimens in composite chloride-sulfate solutions
followed by seven days of drying. For preparation of composite solutions, different concentrations of NaCl
(1.5% and 3.5%) and MgSO4 (2.5% and 5%) were used. Different corrosion tests namely potential
measurement, relative resistivity measurement and corrosion current density by linear polarization resistance
technique was conducted at the age of 180 days. On the basis results obtained, the effect of different
concentrations of chloride and sulfate ions on corrosion parameters indicating the variation in performance of
concrete against rebar corrosion have been obtained.
KEYWORDS: Concrete; Steel; Chloride ions; Sulfate ions; Corrosion.
1. INTRODUCTION
Durability problems are encountered in reinforced concrete structures during their service life. Durability of
reinforced concrete structures is affected by surrounding environmental and exposure conditions [1]. Corrosion
of steel reinforcement in concrete is the most significant durability problem and it results in huge cost for repair
and maintenance of reinforced concrete structures. In well made and dense concrete, the steel reinforcement is
normally protected from corrosion, as concrete provides both chemical and physical protection to the steel
reinforcement [2,3]. The chemical protection is through the formation of passive layer over the steel surface,
which is formed due to high alkalinity (typically pH in excess of 13) of the concrete pore solution. The physical
protection is through the denser and relatively impermeable structure of concrete. However corrosion initiation
takes places when there is breakdown of the passive layer either due to the presence of chloride ions at rebar
level or due to reduction in pH value of concrete to a non-protective level due to carbonation [4, 5, 6]. As
compared to carbonation, chloride ions are considered to be the primary cause of steel reinforcement corrosion
in concrete.
Chlorides and sulfates present in seawater, groundwater etc. pose serious threat to the long-term durability and
serviceability of reinforced concrete structures. In such situations, in addition to chloride ions, the ingress of
sulfate ions also takes place into concrete and the mechanism of deterioration may become even more complex
because of ingress of both chloride and sulfate ions. The role of sulfate ions on the mechanism of chloride
induced rebar corrosion is not clearly known [7]. The presence of sulfate ions may significantly influence the
chloride induced corrosion of steel in concrete. Thus it is necessary to understand the behavior of concrete in
composite solutions of chloride and sulfate ions with regard to the durability of reinforced concrete structures
exposed to both marine and away from marine environment. Various corrosion monitoring techniques used for
determining the parameters related to corrosion of steel in concrete are half-cell potential measurement,
corrosion rate measurement etc. The potential measurement gives the potential value of steel reinforcement and
from the measured potential values, the probability of steel reinforcement corrosion can be obtained. The
procedure for potential measurement and the criteria for probability of steel reinforcement corrosion
corresponding to potential values are mentioned in ASTM C876 [8]. The information about actual extent of
corrosion of steel reinforcement can be obtained from corrosion rate. The different techniques used for
measurement of corrosion rate are linear polarization resistance measurement, electrochemical impedance
spectroscopy (EIS) technique, gravimetric measurement etc. In reinforced concrete structures, the detection of
corrosion and measurement of corrosion parameters can be carried out using these techniques and depending on
the extent of corrosion, the appropriate repair strategy can be adopted.
In the present study, corrosion performance evaluation of reinforced concrete subjected to composite chloride-
sulfate exposure conditions has been carried out. For this purpose, slab specimens with a centrally embedded
steel bar have been subjected to composite chloride-sulfate solutions and the corrosion parameters have been
measured to assess the corrosion performance.
2. EXPERIMENTAL WORK
In this section, details about materials, test specimen, exposure solutions and tests carried out in the present
work are presented.
2.1 Materials and Test Specimen
For preparation of concrete specimens, one type of cement i.e. ordinary Portland cement (OPC) was used as
binder. Coarse aggregates of size 20 mm MSA (maximum size of aggregate) and 12.5 mm MSA were used in
the proportion of 66% and 34% of the total mass of coarse aggregate respectively. Locally available sand was
used as fine aggregate. Tap water from laboratory was used as mixing water. Two w/c ratios used in the
preparation of specimens were 0.50 and 0.55. One type of steel such as Tempcore TMT (thermo-mechanically
treated) steel bar of diameter 12 mm was used as the steel reinforcement. Concrete mixes were designed for
similar workability with slump varying from 30-50 mm and a water content of 200 kg/m3 was fixed on the basis
of trial tests for the required slump range. The mix quantities of cement, sand and coarse aggregate were
calculated by British mix design method [9] with some modifications. The quantities of cement, sand and coarse
aggregate at w/c ratio of 0.50 are 400 kg/m3, 621.25 kg/m3 and 1153.75 kg/m3 respectively. Similarly these
quantities at w/c ratio of 0.55 are 363.64 kg/m3, 633.98 kg/m3 and 1177.39 kg/m3 respectively.
Cube specimens of size 150 mm were prepared from the concrete mixes for determination of 28 day
compressive strength. Slab specimens of size 320mm x 320mm x 52mm with a centrally embedded steel bar
were prepared from different concrete mixes for measurement of corrosion parameters. A cover of 20 mm was
provided to the steel bar at both top and bottom. The schematic diagram of the slab specimen is shown in
Figure 1. After 24 hours of preparation, cube and slab specimens were demoulded and subjected to moist curing
in a curing tank for 27 days. After completion of moist curing, the specimens were removed from the curing
tank.
2.2 Exposure Solutions
The slab specimens were exposed to composite solutions of chloride and sulfate ions. The composite solutions
of chloride and sulfate ions were prepared by adding sodium chloride (NaCl) and magnesium sulfate (MgSO4)
in required quantity of water. Sodium chloride and magnesium sulfate of analytical reagent grade were used in
285
the preparation of test solutions. Sodium chloride concentrations of 1.5% and 3.5% (by mass of water) and
magnesium sulfate concentrations of 2.5% and 5% (by mass of water) were added for preparation composite
solutions. The details about the test solutions used in the present work are presented in Table 1.
12 mm diameter centrally
embedded steel bar
320 mm
320 mm
12 mm diameter centrally embedded steel

bar with cover depth of 20 mm at both top
and bottom
52 mm
320 mm
Figure 1 Slab specimen of size 320 mm × 320 mm × 52 mm
Table 1 Details of test solutions
Sl No. Concentration of solutions Abbreviation

1 1.5% NaCl + 2.5% MgSO4 1.5%NC+2.5%MS
2 1.5% NaCl + 5% MgSO4 1.5%NC+5%MS
3 3.5% NaCl + 2.5% MgSO4 3.5%NC+2.5%MS
4 3.5% NaCl + 5% MgSO4 3.5%NC+5%MS
The slab specimens prepared from different concrete mixes were exposed to composite solutions with alternate
wetting-drying cycles. After removal from curing tank (after 27 days of moist curing), the slab specimens were
kept in laboratory exposure condition for a period of fourteen days. After that the slab specimens were exposed
to different test solutions. For this purpose replicate specimens from same concrete mix were used separately for
exposure to different test solutions. The alternate wetting and drying cycles comprised of seven days of
immersion of slab specimens in composite solutions in plastic containers followed by seven days of laboratory
drying. The alternate wetting and drying cycles was adopted to accelerate the ingress of chloride and sulfate ions
into the concrete. The exposure of slab specimens to test solutions with wetting-drying cycles was continued
and the corrosion measurement was carried out at the age of 180 days from the day of preparation of specimens.
286
2.3 Compression Testing on Cubes
After completion of moist curing, compressive strength of cubes made from OPC and w/c ratios of 0.50 and
0.55 were determined at the age of 28 days in the compression testing machine.
2.4 Measurement of Corrosion Parameters
The corrosion parameters such as half-cell potential, resistivity and corrosion current density were measured at
the age of 180 days using a corrosion monitoring instrument (Make ACM, Gill AC serial no.1542). Corrosion
current density was determined by linear polarization resistance (LPR) technique. Saturated calomel electrode
(SCE) was used as reference electrode. The LPR test with IR compensation was performed with guard ring
arrangement on the slab specimens prepared from different concrete mixes and exposed to different test
solutions. While determining the corrosion current density, the IR drop value in the cover concrete has to be
determined and compensated. This is required due to the fact that, concrete is a high resistive medium, thus IR
drop in cover concrete has to be compensated while determining the corrosion current density. Before placing
the guard ring assembly on the surface of slab specimen, conducting sponge attached to the guard ring was
wetted with soap solution so as to have proper electrical contact. For determination of corrosion current density,
the polarized surface area of the steel bar is taken as the area lying under a circle intersecting the midpoint
between the two sensor electrodes [10,11] present in guard ring assembly and only top half surface area of the
steel bar is assumed to be polarized [11,12]. In the linear polarization resistance technique, the steel bar in the
slab specimen was polarized to ± 20mV from the equilibrium potential at a scan rate of 0.1mV per second.
Using the Stern-Geary equation, the corrosion current density icorr, is obtained as follows;
i corr = B R (1)
p
Where
B = Stern-Geary constant
Rp = polarization resistance of steel
The expression for Stern-Geary constant is given by;
(β a ×β c )
B= (2)
2.3(β a +β c )
Where
βa and βc are anodic and cathodic Tafel constants respectively. The value of B is considered as 26 mV for steel
bar in active condition [7,11,13].
As already stated, the compressive strength of concrete cubes prepared from OPC and w/c ratios of 0.50 and
0.55 were determined at the age of 28 days. The 28 day compressive strength values were found to be 34.96
N/mm2 and 29.78 N/mm2 respectively at w/c ratios of 0.50 and 0.55.
The measured half-cell potential values (with respect to saturated calomel reference electrode) of steel bar in
concrete (made from OPC) exposed to composite solutions of chloride and sulfate ions of different
concentrations are shown in Figure 2. From this figure it is observed that the potential values of steel bar
exposed to all the composite solutions were more negative than – 270 mV (with respect to SCE) / -350 mV
(with respect to Cu/CuSO4 reference electrode) that correspond to greater than 90% probability of steel
reinforcement corrosion occurrence as per ASTM C876 [8] criteria. It is noted the potential values decreased
with increase in chloride ion concentration at both sulfate ion concentrations for both w/c ratios. Further the
potential values decreased with increase in sulfate ion concentration at both chloride ion concentrations at w/c
ratio of 0.50 as observed form Figure 2; however the variation is not systematic at w/c ratio of 0.55.
287
w /c = 0.50
600
w /c = 0.55
Potential value (-mV/ SCE) 500
400
300
200
100
0
1.5%NC+2.5%MS 1.5%NC+5%MS 3.5%NC+2.5%MS 3.5%NC+5%MS
Exposure solution
Figure 2 Half-cell potential of steel reinforcement in concrete

exposed to composite solutions
The relative resistivity values of concrete exposed to composite solutions of different concentrations of chloride
and sulfate ions are shown in Figure 3. The measured IR compensation value in Ohm-cm2 represents the relative
resistivity value of cover concrete.
20000 w /c = 0.50
w /c = 0.55
Relative resistivity (Ohm-cm2)
15000
10000
5000
0
Exposure solution
Figure 3 Relative resistivity value of concrete exposed to composite solutions
From Figure 3, it is observed that, the relative resistivity values increased with decrease in w/c ratio for all
exposure solutions. This is due to dense structure exhibited by concrete made with lower w/c ratio as compared
to that made with higher w/c ratio. Further it is observed that relative resistivity value decreased with increase in
sulfate ion concentration at both chloride ion concentrations at w/c ratio of 0.55; however same variation was
not observed at w/c ratio of 0.50 as observed from Figure 3. Similarly relative resistivity value decreased with
increase in chloride ion concentration at both sulfate ion concentrations for concrete made with w/c ratio of
0.50. However for w/c ratio of 0.55, there is no systematic variation in relative resistivity value with chloride ion
288
concentration at different sulfate ion concentrations in composite solutions.
The obtained corrosion current density values of steel bar embedded in concrete prepared from different mixes
and exposed to composite solutions of chloride and sulfate ions are shown in Figure 4.
w /c = 0.50
10
w /c = 0.55
Corrosion current density (µA/cm2)
0
Exposure solution
Figure 4 Corrosion current density of steel bar in concrete exposed to composite solutions
From this figure it is noted that the corrosion current density of steel bar increased with the increase in chloride
ion concentration in composite solutions at both w/c ratios. This is due to higher amount chloride ions available
at the rebar level for corrosion of steel bar. Further the corrosion current density mostly increased with increase
in sulfate ion concentration in the composite solutions except for composite solutions with 1.5% NaCl at w/c
ratio of 0.50. In addition the corrosion current density values mostly decreased with decrease in w/c ratio except
for composite solution of 3.5% NaCl + 5% MgSO4.
4. CONCLUSIONS
The present experimental investigation has been carried out determine the corrosion parameters such as half-cell
potential, relative resistivity and corrosion current density of steel in concrete exposed composite solutions of
chloride and sulfate ions and the observations regarding the variations in the corrosion parameters of steel in
concrete (made with OPC and w/c ratio of 0.50 and 0.55) exposed to four composite solutions of different
concentrations of chloride and sulfate ions have been obtained. The potential values of steel reinforcement
exposed to all the composite solutions were more negative than – 270 mV (with respect to SCE) and the
potential values decreased with increase in chloride ion concentration in the composite solutions at both sulfate
ion concentrations. Concrete made with lower w/c ratio exhibited higher value of relative resistivity as
compared to that made with higher w/c ratio. However variations in relative resistivity value with chloride ion
and sulfate ion concentrations were not consistent at both w/c ratios. The corrosion current density increased
with the increase in chloride ion concentration at different concentrations of sulphate ions in composite
solutions. Further the corrosion current density mostly increased with increase in sulfate ion concentration in the
composite solutions.
ACKNOWLEDGEMENTS
The present work is a part of the sponsored research project funded by Department of Science and Technology,
Government of India. The author wishes to express his gratitude to Department of Science and Technology,
Government of India for funding the project through Fast Track scheme for Young Scientists.
289
REFERENCES
[1] Sakr, K. (2005). Effect of cement type on the corrosion of reinforcing steel bars exposed to acidic media
using electrochemical techniques. Cement and Concrete Research, 35 (9), 1820-1826.
[2] Dehwah, H. A. F.; Maslehuddin, M. and Austin, S. A. (2003). Effect of sulfate ions and associated cation
type on the pore solution chemistry in chloride-contaminated plain and blended cements. Cement & Concrete
Composites, 25 (4-5), 513-525.
[3] Almusallam, A. A. (2001). Effect of degree of corrosion on the properties of reinforcing steel bars.
Construction and Building Materials, 15 (8), 361-368.
[4] Andrade, C.; Alonso, C. and Sarria, J. (2002). Corrosion rate evolution in concrete structures exposed to the
atmosphere. Cement & Concrete Composites, 24 (1), 55-64.
[5] Bautista, A.; Blanco, G. and Velasco, F. (2006). Corrosion behaviour of low-nickel austenitic stainless steels
reinforcements: A comparative study in simulated pore solutions. Cement and Concrete Research, 36 (10),
1922-1930.
[6] Elsener, B. (2002). Macrocell corrosion of steel in concrete - implications for corrosion monitoring. Cement
& Concrete Composites, 24 (1), 65-72.
[7] Dehwah, H. A. F.; Maslehuddin, M. and Austin, S. A. (2002). Long-term effect of sulfate ions and associated
cation type on chloride-induced reinforcement corrosion in Portland cement concretes. Cement & Concrete
Composites, 24 (1), 17-25.
[8] ASTM C 876 - 09, (2009). Standard test method for corrosion potentials of uncoated reinforcing steel in
concrete. West Conshohocken, PA.
[9] SP 23 : 1982, Handbook on Concrete Mixes (Based on Indian Standards), Bureau of Indian Standards, New
Delhi.
[10] Law, D. W.; Millard, S. G. and Bungey, J. H. (2000). Linear polarisation resistance measurements using a
potentiostatically controlled guard ring. NDT&E international, 33 (1), 15-21.
[11] Pradhan, B. and Bhattacharjee, B. (2009). Performance evaluation of rebar in chloride contaminated
concrete by corrosion rate. Construction and Building Materials, 23 (6), 2346-2356.
[12] Gu, G. P.; Beaudoin, J. J. and Ramachandran, V. S. (2001). Techniques for corrosion investigation in
reinforced concrete. Handbook of Analytical Techniques in Concrete Science and Technology, V. S.
Ramachandran, J. J. Beaudoin, eds., Noyes Publications, New Jersey, 441-504.
[13] Erdogdu, S.; Bremner, T. W. and Kondratova, I. L. (2001). Accelerated testing of plain and epoxy-coated
reinforcement in simulated seawater and chloride solutions. Cement and Concrete Research, 31 (6), 861-867.
290
Paper No.: 048
POTENTIODYNAMIC POLARIZATION STUDY ON BARE STEEL IN

CONCRETE POWDER SOLUTION EXTRACTS CONTAMINATED WITH
CHLORIDE AND SULFATE IONS
Fouzia Shaheen1 and Bulu Pradhan2
1
Research Scholar, Dept. of Civil Engineering, Indian Institute of Technology Guwahati, Assam, India
2
Assistant Professor, Dept. of Civil Engineering, Indian Institute of Technology Guwahati, Assam, India
E-mail: bulu@iitg.ernet.in
ABSTRACT:
In this paper, the findings of an experimental investigation are presented wherein the role of chloride and sulfate
ions on steel reinforcement corrosion has been studied. For this purpose, potentiodynamic polarization study has
been conducted on bare steel specimens in concrete powder solution extracts contaminated with chloride and
sulfate ions. The concrete specimens were admixed with different concentrations of sodium chloride (3% and 5%)
and sodium sulfate (3% and 6%) at the time of preparation. Portland pozzolana cement (PPC), one type of steel
(Thermex TMT) and two w/c ratios of 0.45 and 0.5 were used in the experimental investigation. Concrete powder
is obtained by crushing the concrete specimens admixed with different concentrations of chloride and sulfate ions.
From results of potentiodynamic study in concrete powder solution extracts, various zones of corrosion of steel
bar namely semi-immune zone, active zone, passive zone and pitting zone at different combinations of chloride
and sulfate ions have been identified.
KEYWORDS: Concrete powder, Steel, Chloride ions, Sulfate ions, Potentiodynamic, Corrosion
1. INTRODUCTION
Durability can be defined as the ability of concrete to withstand the damaging effects of the environment and its
service conditions until it reaches a minimum level of performance [1]. Deterioration due to reinforcement
corrosion due to chloride ions and degradation of concrete exposed to sulfate bearing environments are probably
the most important durability issues with reinforced concrete. The durability of reinforced concrete depends on
the surrounding environment and exposure conditions. Corrosion, one of the main causes of deterioration in
reinforced concrete structures, initiates due to its exposure to harmful chemicals that may be found in nature such
as in some ground water, soil, industrial effluents and sea waters. In addition some of the concrete making
ingredients are also contaminated with the aggressive ions. The most aggressive ions that effect the long term
durability of concrete structures are the chloride ions and sulfate ions [2]. Disruption of the protective layer of
ferric oxide around the steel bar takes place when the concentration of chloride ions in its vicinity exceeds a
critical threshold value [3] and it depends on various factors such as cement type, concrete mix proportions,
water/cement (w/c) ratio, C3A content of the cement, blended materials, chloride binding, steel type/ steel
surface condition, concentration of hydroxyl ions, temperature, relative humidity and sources of chloride, etc. [4,
5]. Not all the chloride ions that are in or penetrate into concrete remain free in the pore solution. Some of the
chloride ions bound with the hydration products through a chemical reaction to form calcium chloroaluminate
(Friedel’s salts) [6, 7, 8, 9], while others get absorbed to the various hydrates of cement [10, 11]. Thus, only a
portion of the chloride ions remains free. It is the free chloride ions that are mainly responsible for causing
damage to the concrete structures by corroding steel reinforcement bars. Sulfate attack is one of the major
threats to the durability of concrete. When concrete is in contact with salts containing sulfate, interactions
between solid phase and the sulfate solutions occur, leading to chemical reactions within the material and can
lead to the deterioration of the material [12]. Sulfate ions will attack some or all of the three main hydrate
compounds of hardened concrete: calcium hydroxide, calcium aluminate hydrate and calcium silicate hydrate,
depending on the type of sulfate present in the solution involved [13]. The conjoint presence of chloride ions and
sulfate ions may significantly affect reinforcement corrosion. Such a situation is representative of the marine
environments and substructures exposed to soil and ground water admixed with chloride and sulfate salts [14].
The corrosion of steel rebar in reinforced cement concrete can be represented by polarization curves. The
polarization curve is obtained by conducting potentiodynamic study which represents the relationship between the
potential and the anodic or cathodic current density. However due to the variability in the electrolytic environment
of the concrete and high resistivity of concrete medium, it is difficult and complicated to conduct the
potentiodynamic study directly on steel embedded in concrete [15]. However this difficulty can be overcome by
conducting the polarization study in simulated pore solutions. The concrete powder solution extracts can be used
for carrying out anodic polarization study. Concrete powder solution extracts nearly represents the concrete pore
solution as it is prepared from the concrete powder that is obtained by crushing the hardend concrete that
represents the mixture of cement hydrates, fine and coarse aggregates in their respective proportions. The
objective of present work is to determine different zones of corrosion namely; semi-immune zone, active zone,
passive zone, pitting zone for Thermex TMT steel reinforcement in chloride and sulfate contaminated concrete
powder solution extracts prepared from one type of cement and two w/c ratios. For this purpose potentiodynamic
linear sweep test has been carried out to obtain the anodic polarization curves. From the obtained zones of
corrosion, the state of corrosion of the steel bar in the concrete with respect to potential values can be identified
and thus the shift of potential required to bring the steel bar to a protection level can be determined.
2. EXPERIMENTAL PROGRAM
2.1 Materials and Specimen Preparation
Concrete specimens were prepared from Portland pozzolana cement (PPC) at w/c ratios of 0.45 and 0.5. The
coarse aggregates of size 20 mm maximum size of aggregate (MSA) and 10 mm MSA were used in the proportion
of 1.94:1 by mass. Coarse aggregates satisfy the grading requirements as per guidelines specified in IS: 383-1970
[16]. Sand is used as fine aggregate and it is conforming to grading zone II as per IS: 383-1970 [16]. The specific
gravity of 20 mm (MSA), 10 mm (MSA) and sand are 2.64, 2.63 and 2.61 respectively. The water content in all
the mixes was kept constant to 195 Kg/m3 for slump value of 20-50 mm.
Chloride was added to concrete as sodium chloride (NaCl) and sulfate was added as sodium sulfate (Na2SO4).
Different concentrations of NaCl and Na2SO4 were admixed at the time of preparation of concrete. The
concentrations of sodium chloride used were 3% and 5% by mass of cement and the concentrations of sodium
sulfate used were 3% and 6% by mass of cement. The proportions of cement, fine aggregate and coarse aggregate
are 1:1.418:2.635 and 1:1.661:2.953 by mass at w/c ratios of 0.45 and 0.5 respectively. Thermex TMT
(Thermo-mechanically Treated) steel bar of diameter 12 mm was used as bare steel specimen.
2.2 Preparation of Bare Steel Specimens
Steel specimens required for potentiodynamic study are cut to a length of 70mm from steel bars followed by
drilling and threading at one end and are then cleaned by using wire brush to remove any rust present on the
surface. The bare steel specimen was then coated with epoxy leaving 5mm height at the other end, which is
exposed. The bare steel specimen is shown in Figure 1.
292
5mm exposed portion
1 12mm
12mm
1 Section 1-1
70mm
3mm diameter threaded hole for connection
Epoxy coating to corrosion monitoring instrument
Figure1 Bare steel specimen
2.3 Preparation of Concrete Powder Solution Extracts
Concrete cubes of size 150 mm × 150 mm × 150 mm were prepared from different mixes. After 24 hours of
preparation, the cubes were subjected to moist curing in a curing tank for 27 days. After curing, the cubes were
kept in the laboratory condition till the period of crushing. In addition three cubes from each concrete mix were
tested for 28 day compressive strength. Concrete cubes were crushed in the compression testing machine followed
by further crushing in the abrasion testing machine at the age of 56 days from the day of preparation and then
sieved through square sieve of mesh size 150 µm. The collected powder sample was then stored in air tight plastic
containers. A photograph of concrete powder sample is shown in Figure 2.
For the preparation of concrete powder solution extracts, the powder is mixed with distilled water in 1:1
proportion by mass and then stirred for half an hour. The solution is then boiled for 15-20 minutes and allowed to
settle and cool to room temperature. After that the solution is filtered through Whatman no.1 filter paper. The
filtered solution extract is then used for potentiodynamic polarization test on the bare steel specimens [15].
Figure 2 Concrete powder sample
2.4 Compressive Strength Test
The compressive strength test was carried out on the cube specimens prepared from different concrete mixes at the
age of 28 days in the compression testing machine. The average value of three replicates was reported as the
compressive strength for a given concrete mix.
293
2.5 Potentiodynamic Linear Sweep Test
The potentiodynamic linear sweep test was performed on bare steel specimens using a corrosion monitoring
instrument (Make ACM, Gill AC serial no.1542). Saturated calomel electrode (SCE) was used as reference
electrode for measuring the potential of the steel bar specimen. For testing, the bare steel specimen was connected
to the working electrode pin of the corrosion monitoring instrument. The working electrode (WE), reference
electrode (RE) and auxiliary electrode (AE) were connected to the electrochemical cell as shown in the Figure 3.
The filtered solution extract was poured into the electrochemical cell and the potentiodynamic linear sweep test
was carried out from 0 mV to 1500 mV with offset from corrosion potential at a sweep rate of 50 mV per minute
on bare steel specimen [17].
Potentiostat
Auxiliary electrode
(AE)
Working electrode (WE)

Reference electrode (RE)
Concrete powder solution extracts

Electrochemical Cell
Figure 3 Schematic diagram of test setup for potentiodynamic sweep test
3.1 Compressive Strength
The 28 day compressive strength values of different concrete mixes are presented in Table 1.
Table 1: 28 day Compressive strength
Admixed chloride and sulfate Compressive Strength (N/mm2)

concentrations
w/c=0.45 w/c=0.5
Control mix 32.60 29.0
3%NaCl + 3%Na2SO4 38.07 35.11
3%NaCl + 6%Na2SO4 36.29 34.52
5%NaCl + 3%Na2SO4 34.80 32.15
5%NaCl + 6% Na2SO4 35.70 33.60
From the results it is observed that the compressive strength of the specimens made with a w/c ratio of 0.45
exhibited higher compressive strength as compared to those made with w/c of 0.5 irrespective of the
concentrations of chloride and sulfate ions. Further it is noted that the concrete mixes admixed with chloride and
sulfate ions showed higher compressive strength as compared to the concrete mix. This may be due to the filling
of pores with the products of hydration and products of reaction between sulfate ions (preferentially react as
294
compared to chloride ions) and hydration products there by resulting in a relatively denser microstructure as
compared to the control mix. Further it is observed that the concrete admixed with different dosages of Na2SO4
with 3%NaCl exhibited higher compressive strength as compared to the concrete admixed with different dosages
of Na2SO4 with 5%NaCl as evident from Table 1. While observing the effect of sulfate ion concentration it is
inferred that the compressive strength decreased with increase in sulfate ion concentration with 3% sodium
chloride whereas the compressive strength increased with increase in sulfate concentration with 5% sodium
chloride.
3.2 Different Zones of Corrosion of the Steel Reinforcement
Anodic polarization curves were obtained through potentiodynamic linear sweep test performed on bare steel
specimens in concrete powder solution extracts. Anodic polarization curve for Thermex TMT steel specimen in
concrete powder solution extract made from PPC at w/c ratio of 0.45 and without addition of sodium chloride and
sodium sulfate in concrete (i.e. control mix) is shown in Figure 4.
Pass/Pitt Pitting zone (Transpassivity zone without the presence

boundary of chloride and sulfate ions)
Passive zone
Act/Pass
boundary Active zone
Rest potential
Semi-immune zone
Figure 4 Anodic polarization curve for PPC, w/c = 0.45 and without addition of chloride and sulfate salts
In Figure 4 different zones of corrosion namely semi-immune zone, active zone, passive zone and pitting zone are
identified. The zone of corrosion below the rest potential is termed as semi-immune zone. In this zone the steel is
thermodynamically unable to undergo anodic reactions and thus immune to dissolution. The zone above the
semi-immune zone is termed as the active zone, wherein the increase in corrosion current density is significantly
high with small change in potential. The zone above the active zone can be termed as the passive zone. In this zone
change in current density is very less with significant increase in potential. The zone above the passive zone is
termed as pitting zone, where the anodic current density increases significantly leading to localized dissolution of
steel. In Figure 4, the active-passive zone boundary (Act/Pass) potential and Pass-Pitt zone boundary (Pass/Pitt)
potential are also shown. The anodic polarization curve at w/c ratio of 0.5 and without addition of sodium chloride
and sodium sulfate in concrete (i.e. control mix) is shown in Figure 5.
295
Figure 5 Anodic polarization curve for PPC at w/c = 0.5 and without addition of chloride and sulfate salts
From the Figure 4 and 5, it is observed that the values of rest potential, active/passive boundary potential and
passive/pitting boundary potential for control mix are -262.14 mV and -269.85 mV, -180.08 mV and -174.3 mV,
and +427.5 mV to +450.2 mV at w/c ratios of 0.45 and 0.5 respectively.
3.3 Role of NaCl and Na2SO4 Concentrations on Zones of Corrosion
The anodic polarization curves for the concrete mixes made with PPC and admixed with different concentrations
of sodium chloride and sodium sulfate are shown in Figure 6 to Figure 9 at w/c ratio of 0.45 and in Figure 10 to
Figure 13 at w/c ratio of 0.5.
Figure 6 Anodic polarization curve for concrete Figure 7 Anodic polarization curve for concrete
made with w/c = 0.45 and admixed with made with w/c = 0.45 and admixed with
3%NaCl + 3% Na2SO4 3%NaCl + 6% Na2SO4
296
3%NaCl +3% Na2SO4 3%NaCl + 6% Na2SO4
297
From these anodic polarization curves, the potential at the point of change from active to passive zone i.e.
active/passive zone boundary potential and that from passive to pitting zone i.e. passive/pitting zone boundary
potential values were obtained for each steel specimen in concrete powder solution extracts made from different
concrete mixes.
The values of rest potential, active/passive boundary potential and passive/pitting boundary potential lie in the
ranges of -338.52mV to -425.69 mV, -266.95 mV to -345.23 mV and -3.64 mV to +135.3 mV respectively for
concrete admixed with 3% NaCl irrespective of w/c ratio and admixed sodium sulfate concentration as observed
from Figure 6, Figure 7, Figure 10 and Figure 11. Similarly from Figure 8, Figure 9, Figure 12, and Figure 13, it is
observed that for concrete admixed with 5% NaCl the values of rest potential, active/passive boundary potential
and passive/pitting boundary potential varied from -357.15 mV to -405.26 mV, -270.32 mV to -330.32mV and
-27.74mV to -129.98 mV respectively irrespective of w/c ratio and admixed sodium sulfate concentration.
From Figure 6, Figure 8, Figure 10 and Figure 12, it is observed that the values of rest potential, active/passive
boundary potential and passive/pitting boundary potential for concrete admixed with 3% Na2SO4 and different
concentrations of NaCl lie in the ranges of -338.52 mV to -405.26 mV, -266.95 mV to -330.32 mV and -129.98
mV to +135 mV respectively irrespective of w/c ratio. Similarly from Figure 7, Figure 9, Figure 11, and Figure 13,
it is noted that the values of rest potential, active/passive boundary potential and passive/pitting boundary
potential for concrete admixed with 6% Na2SO4 and different concentrations of NaCl lie in the ranges of -357.15
mV to -425.69 mV, -270.32 mV to -345.23 mV and -106.5 mV to +59.93 mV respectively.
From the obtained results it is observed that the range of passive zone is significantly higher in control mix as
compared to concrete admixed with various concentrations of NaCl and Na2SO4. As the range of passive zone is
higher in control mix, the upper limit of passivity range known as the pitting potential increases and thus there is
less chance of localized corrosion.
From the anodic polarization curves obtained for Thermex TMT steel, different concentrations of admixed NaCl
and Na2SO4 and w/c ratios, it is observed that the range of passive zone decreases while that of pitting zone
increases with increase in chloride ion concentration. Thus the chance of localized corrosion increases with the
increase in chloride ion concentration. While observing the effect of sulfate ion concentration, it is inferred that
the concrete mixes admixed with 6% Na2SO4 mostly showed lower chance of localized corrosion (higher range of
passive zone) as compared to concrete mixes admixed with 3% Na2SO4 with different dosages of sodium chloride.
4. CONCLUSION
From the results of the potentiodynamic linear sweep test conducted on Thermex TMT steel in concrete powder
solution extracts prepared from concrete mixes made with PPC with different concentrations of sodium chloride
and sodium sulfate, various zones of corrosion namely, semi-immune zone, active zone, passive zone and pitting
zone have been identified in terms of potential values. The ranges of potential values corresponding to different
zones of corrosion can be used for determining the shift of potential required to bring the potential of the steel bar
in concrete contaminated with chloride and sulfate ions to the protection level i.e. in the semi-immune zone.
REFERENCES
[1] Saleh, A. (2008). Chloride induced corrosion and sulfate attack – A literature review on concrete durability.
The 3rd International conference-ACF/VCA.
[2] Wee, T. H.; Suryavanshi, A.K.; Wong, S.F. and Rahman, A. K. (2000). Sulfate resistance of concrete
containing mineral admixtures. ACI materials journal, 97 (5), 536-549.
298
[3] Pradhan, B.; Bhattacharjee, B. (2009). Half Cell Potential as an indicator of chloride-induced Rebar corrosion
initiation in RC. Journal of Materials in Civil Engineering (ASCE), 21 (10), 543–552.
[4] Alonsd, C.; Andrade, C.; Castellote, M..; Castro, P. (2000). Chloride threshold values to depassivate
reinforcing bars embedded in a standardized OPC mortar. Cement concrete Research, 30 (7), 1047–1055.
[5] Trejo, D.; Monterio, P.J. (2005). Corrosion performance of conventional (ASTM A615) and low- alloy
(ASTM A706) reinforcing bars embedded in concrete and exposed to chloride environments. Cement concrete
Research, 35 (3), 562–571.
[6] Robert, M.H. (1962). Effect of calcium chloride on the durability of the pretensioned wire in prestressed
concrete. Magazine of concrete Research, 14 (42), 143–154.
[7] Mehta, P.K. (1977). STP 629 Chloride Corrosion of Steel in Concrete (ASTM International), Effect of cement
composition on corrosion of reinforcing steel in concrete, Edited by Tonini, D.E.; Dean Jr, S.N., 72–79.
[8] Diamond, S. (1986). Chloride concentrations in concrete pore solutions resulting from calcium and sodium
chloride Admixtures. Cement concrete and Aggregates, 8 (2), 62–79.
[9] Taylor, H. F. W. (1992). Rapid determination of the chloride diffusivity in concrete by applying on electrical
field. ACI Materials Journal, 89 (1), 49–53.
[10] Ramachandran, V. S.; Seeley, R. C.; Polomark, G. M. (1984). Free and combined chloride in hydrating
cement and cement components. Materials and Structures, 17 (4), 285–289.
[11] Byfors, K.; Hansson, C. M.; Trittihart, J. (1986). Pore solution expression as method to determine the
influence of mineral additives on chloride binding. Cement and concrete Research, 16 (5), 760–770.
[12] Lothenbach, B.; Bary, B.; Le Bescop, P.; Schmidt. T.; Letterrier, N.(2010). Sulphate ingress in Portland
cement. Cement concrete Research, 40 (8), 1211–1225.
[13] Committee Euro-International Du Beton, Durable concrete structures - Design Guide. Thomas Telfrod
Services Ltd, ISBN 07277 16204.
[14] Maslehuddind, M. (1995). Role of chloride ions and strength reduction in plain and blended cements in
Sulphate environment. Construction and Building Materials, 9 (1), 25–33.
[15] Pradhan, B.; Bhattacharjee, B. (2007). Corrosion zones of rebar in chloride contaminated concrete through
potentiostatic study in concrete powder solution extracts. Corrosion Science, 49 (10), 3935–3952.
[16] IS: 383-1990 (reaffirmed 2002). Specification for coarse and fine aggregate from natural sources for
concrete. New Delhi: Bureau of Indian Standard, 1993.
[17] Kapat, Chetan.; Pradhan, B.; Bhattacharjee, B. (2006). Potentiostatic study of reinforcing steel in chloride
contaminated concrete powder solution extracts. Corrosion Science, 48 (7), 1757–1769.
299
Paper No.: 049
NON-LINEAR VIBRATION AND DYNAMIC RESPONSE OF EPOXY-BASED

LAMINATED COMPOSITE PLATE WITH SQUARE CUT-OUT AT ELEVATED
TEMPERATURE
Arun Mukherjee1, Sreyashi Das (nee Pal)2 and Dr. A. GuhaNiyogi3
1PG student, Department of Civil Engg,, Jadavpur University, Kolkata-700032

2Asst. Professor, Department of Civil Engg,, Jadavpur University, Kolkata-700032
3Professor, Department of Civil Engg,, Jadavpur University, Kolkata-700032
Email: mukherjeearun@ymail.com
ABSTRACT:
Effect of temperature on the dynamic properties of epoxy-based laminated composite plates with a square
cut-out are investigated. Eight-noded isoparametric plate elements with 6 degrees of freedom per node have
been implemented in the present computations. First order transverse shear deformation based on YNS theory is
used along with rotary inertia of the material. The analysis considers material properties of the laminae at
elevated temperature. Residual stresses due to thermal environment are taken into account. A set of new results
with various cut-out ratios and by varying the thickness of plates are presented. Results showing change in
natural frequencies of laminates at elevated temperatures for various fibre angle orientations, boundary
conditions, plate thicknesses and various cutout ratios are presented.
KEYWORDS: Free Vibration, Laminates, Thermal Load, Cutout
1. INTRODUCTION
Epoxy-based laminated composite plates with cut-outs are inevitable in naval, aeronautical, mechanical and
civil structures. A cut-out in a plate changes the behavior of the structure and stress concentration occurs near
the opening. Further, such structures are often subjected to high temperature and high dynamic load. This has
adverse effects on the strength and stiffness of structural composites. The matrix is more susceptible to the
elevated temperature than the fibre, hence, the deformation is observed to be more in the transverse direction.
The rise in thermal stress reduces the elastic modulus of the material and induces initial internal stresses, which
may affect the stability as well as the safety of the structures. Thus, knowledge of the dynamic behavior of a
structure over a range of temperatures is essential for the design of structures in a thermal environment.
In the review of literature, Whitney and Ashton [1] considered the effect of environment on the free vibration of
laminated composite plates by using Ritz method for symmetric laminates and equilibrium equation of motion
for unsymmetric laminates. Dhanaraj and Palninathan [2] showed that the natural frequency decreases linearly
with the increase in temperature. Shen and Springer [3, 4] showed that the physical change due to temperature
and moisture effects in the resin results in decreasing the tensile properties and reducing the transverse and shear
modulus. Sai Ram and P.K. Sinha [5] studied the moisture and temperature effect in free vibration of symmetric
and antisymmetric laminates accounting transverse shear deformation with simply supported and clamped
boundary condition. Lee et al. [6] have predicted the natural frequencies of composite rectangular plates with
cutouts, neglecting shear deformations and rotary inertia. Ashwini kumar and Srivastava [7] studied the free
vibration of laminated composites with delaminations around a central cutout. A FEM formulation using HSDT
is used in the analysis.
In the present investigation results are presented for symmetric and antisymmetric laminates subjected to
uniform distribution of temperature throughout its volume in absence of any external loading. Four layered
graphite/epoxy laminates with simply supported boundary conditions are considered. Lamina material properties
at elevated temperatures [5] are considered. Since the evaluation of shear correction factor from the exact theory
of elasticity is difficult in the present case, a commonly used value of 5/6 is assumed [8].
1.1. Mathematical formulation
A laminated composite plate of uniform thickness t consisting of a number of thin laminae each of which may
be arbitrarily oriented at an angle θ (Figure 1) with reference to the x-axis of the co-ordinate system is
considered. The constitutive equations for the plate, when it is
y subjected to uniform temperature is given by [5],
2
F = D ε -{FN }, (1)
1
where,
{F} = Mechanical force resultant
b
θ
x = {Nx, Ny, Nxy, Mx, My, Mxy, Qx, Qy}T,
{FN} = Thermal force resultant
= {NxN, NyN, NxyN, MxN, MyN, MxyN, QxN, QyN, 0, 0}T,
a
{ε} = {εx, εy, γ̅ xy, Κx, Κy, Κxy, φx, φy}T
A A A B B B 0 0
Figure 1: Arbitrarily oriented lamina,
A A A B B B 0 0
Axis 3 coincides with z axis
A A A B B B 0 0
B B B D D D 0 0
B B B D D D 0 0
[D] = (2)
B B B D D D 0 0
0 0 0 0 0 0 A A
0 0 0 0 0 0 A A
where, the thermal force and moment resultants are
{NNx NNy NNxy}T = ∑#$ ′ % dz (i, j = 1, 2 and 6)

!"
{MNx MNy MNxy}T = ∑#$ & ′ % dz (i, j = 1, 2 and 6) (3)

!"
where % = %' %( %'( )

= Thermal strain= [R] * , * )
(T-T0)
-./ 0 /12 0
in which [R] = , /12 0 -./ 0 8,
sin 20 −/12 20
T and To is the instantaneous and reference temperature respectively.
The stiffness coefficient of the laminate are defined as

9:
(Aij, Bij, Dij) = ∑;<$ 9:!"
′ {1, z, z2} dz, for i,j =1,2,6,
9:
and (Aij) = α∑;<$ 9:!"
′ dz, for i,j = 4,5 (4)
α is a shear correction factor, taken as 5/6, to take account for the non uniform distribution of the transverse
shear strain across the thickness of the laminate.
301
The non-linear strains can be expressed as,
εxnl = [u ,> +v ,> +w
B ,> + 2zDu,> θE,> − v,> θ>,> F + z (θ E,> +θ >,> ) /2,
εynl = [u ,E +v ,E +w
B ,E + 2zDu,E θE,E − v,E θ>,E F + z (θ >,E +θ E,E ) /2,
γxynl = [u,> u,E + v,> v,E + w
B ,> w
B ,E + zDu,E θE,> + u,> θE,E F − zDv,E θ>,> + v,> θ>,E F + z (θE,> θE,E + θ>,> θ>,E )],
γxznl = [u,> θ,E + v,> θ> + zDθE θE,> + θ> θ>,> F],
γyznl = [u,E θ,E + v,E θ> + zDθE θE,E + θ> θ>,E F]. (5)
1.2 Finite Element Formulation
Eight-noded isoparametric plate elements with 6 degrees of freedom per node have been implemented in the
present computations. The stiffness matrix, the initial stress stiffness matrix, the mass matrix and the nodal load
vectors of the element are derived by using the principle of minimum potential energy [5]. The element
displacements are expressed in terms of their nodal values by using the element shape functions and are given
by
u0 = ∑P$ J (K, L)MNO , v0 = ∑P$ J (K, L)QNO , w = ∑P$ J (K, L)R
θx = ∑P$ J (K, L)0'O , θy = ∑P$ J (K, L)0(O (6)
in which uoi, uoi, voi, wi, θxi, θyi are the displacement at a node i.
1.1.1. Element Stiffness Matrix

The linear strain matrix {ε} is obtained by expression {ε} = [B]{δe} (7)
Z/ X/
The element stiffness matrix is given by [Ke] = –Z/ YX/
S )
T S UV UW (8)
1.1.2. Element Geometric Stiffness Matrix

Z/ X/
Element stiffness matrix[5] is given by \]^
_
= YZ/ YX/
a ) b _ a UV UW (9)
1.1.3. Element Mass Matrix

Z/ X/
The elemental mass matrix[5] is defined as c^ = YZ/ YX/
J )
d J UVUW (10)
1.1.4. Element Load Vectors

The element load vector[5] due to hygrothermal forces and moment is given by
e^f = ∬ S )
h f UVUW (11)
1.1.5. Solution Process
The stiffness matrix, the geometric stiffness matrix, the mass matrix and the load vectors of the element, given by
equations (8-11), are calculated by expressing the integral in local natural co-ordinates, ξ and η of the element and
then performing Gaussian integration. Generally, a 3 point Gauss quadrature is adopted to compute the bending
stiffness of the elements, whereas a 2 point integration is applied to calculate the shear stiffness and mass matrix.
Then the element matrices are expressed accordingly to global co-ordinates from local co-ordinates.
The initial displacements (δi) are then evaluated from equilibrium equation,
[K]{δi} = {P} + {PN} (12)
i i i i i i i i
Then the initial stress resultants {Nx , Ny , Nxy , Mx , My , Mxy , Qx , Qy } are obtained from equation (1). After that,
the natural frequencies of vibration of the plate are determined from the general equation of vibration,
| \ + \σ − ω# c | = 0
302
This generalized eigenvalue problem is solved by subspace iteration method.
2. NUMERICAL RESULTS AND DISCUSSION
The finite element formulation described in the earlier section has been used to generate numerical results to study
the effects of cutout under thermal environment of a laminated composite plate. Parametric studies have been
provided for variation in cut-out ratio, stack up sequence, plate thicknesses, boundary conditions and various
temperatures.
2.1. Mesh Convergence Study
A mesh convergence study has been carried out to obtain minimum number of elements required to achieve
the non-dimensional natural frequencies as accurately as possible optimizing the computation time. For this
purpose, a square laminated composite plate having a square cutout (a/s=0.4) is taken and analyzed under
simply supported (SS) and clamped (CC) boundary conditions. The side-to-thickness ratio (=a/t) is taken as 75.
Three mesh sizes containing 144(=8x8 mesh), 200 (=10x10 mesh) and 220 (=11x11 mesh) elements are
considered for the convergence study. Details of a typical mesh having 200 elements are shown in Figure 2.
Natural frequencies obtained for different meshes under

different boundary conditions are tabulated in Table 1. It
can be observed that the fundamental frequency converges
for a mesh containing 200 elements, or more. Hence, for
subsequent studies, a mesh containing 200 (10x10 mesh)
elements is taken.
Table 1: Non-Dimensional Frequency for Different Mesh Sizes
MODE SS CC
NO. 9x9 10 x 10 11 x 11 9x9 10 x 10 11 x 11
1 14.95 14.86 14.86 14.95 14.86 14.86
2 28.06 28.02 28.02 28.06 28.02 28.02
3 28.06 28.02 28.02 28.06 28.02 28.02
Figure 2: Discretization details for a square

plate with square cutout
2.2. Validation Study
Two validation studies are made to test the accuracy of the computer program and the results are compared
with those available in the literature. The cases are as followes:
i. Validation for laminated composite plate with square cutout
ii. Validation for laminated composite plate under thermal load
2.2.1. Laminated Composite Plate With A Square Cutout

A square laminated composite plate, with all edges simply supported, is considered [7]. The lay-up sequence is
[(±45°02°)3(90°/02°/90°)2]s. Each lamina having following material properties (available from Aeronautical
Development Agency, Banglore, India): E1 = 13.00 x 1010 N/m2; E2 = 1.00 1010 N/m2; G12 = G31 = 0.5 x 1010
N/m2; G23 = 0.33 x 1010 N/m2; ν12 = 0.35; ρ = 1500kg/m3. The plate has side-to-thickness ratio(=s/t) of 75.
Nondimensionalized frequency are obtained with a square (a/s = 0.4) cutout and results are compared with those
of Ashwini Kumar and R.P. Shrivastava [7]. The results are shown and compared in Table 2. It is clear that two
results are in good agreement.
303
Table 2: First Five Non-dimensional Frequency
MODE Ashwini kumar Present % diff b d

R = R_ ∗ k m ∗ (n po )
l
et al.[7]
1 14.86 14.86 0.00
2 26.68 28.02 -5.03 Where,
3 29.91 28.02 6.31 wi =non-dimensional frequency
4 49.76 50.54 -1.58 wr = natural frequency in rad/sec
S = dimension of the side of the square
5 54.10 53.45 1.21
plate
t = thickness of the plate
ρ = density of the plate
2.2.2. Laminated Composite Plate under Thermal Load

A simply supported square plate, with side to thickness ratio 100, has been analyzed under various temperatures
and the non-dimensional frequencies are compared with those obtained by K. S. Sai Ram and P. K. Sinha [5].
The layup sequence of the laminate is taken as (0/90/90/0). Graphite-epoxy material with E1=130 GPa,
E2=9.5GPa, G12=6GPa, G13 = G12, G23 = 0.5G12, ν12 = 0.3, α1 = -0.3x10-6/°K, α2 = 28.1x10-6/°K and density =
1600kg/m3 are used. The results are shown in Table 3. Material properties are assumed to be constant for this
purpose, i.e., they do no vary with temperature. From the table it is evident that the program used, are giving
quite accurate results.
Table 3: First Four Non-dimensional Frequency under Uniform Thermal loading
MODE Sai Ram et al. Present % diff

1 8.088 8.096 -0.09
2 19.196 19.297 -0.53
3 39.324 39.489 -0.42
4 45.431 46.165 -1.62
2.3. Case Studies:
2.3.1. Study of Dynamic Behavior of Laminated Composite Plate Structure for Varying cutout size
Various sizes of square cut-outs are considered to examine the effect of cut-out size on the natural frequency of
the laminate. The laminate (1.0m x 1.0m) has its side to thickness ratio equal to 75. The layup sequence of the
laminate is (0/90/90/0). The material properties are given in section 3.2.2. The cut out ratio is varied from 0.2 to
0.8 with increment of 0.2. Two different types of boundary conditions, namely simply supported (SS) and
clamped (CC), are taken in the analysis. Results obtained are shown in Table 4. To visualize the effect of cut out
size a plot of the fundamental frequencies with various cut out sizes is presented in Fig 3.
Table 4: First Three Natural Frequency for varying cutout size
Cut out Ratio (a/s)

Mode No. Simply Supported (SS) Clamped (CC)
0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8
1 59.85 62.96 73.81 124.07 107.03 130.72 233.36 841.66
2 135.30 126.35 120.19 157.75 197.20 189.25 258.11 848.11
3 135.30 126.35 120.19 157.75 197.20 189.25 258.11 848.11
304
1000
Cutout Ratio
800
Frequency (Hz) 600 SS
400 CC
200
0
0 0.2 0.4 0.6 0.8 1
Figure 3: Graphical plot for fundamental frequency vs. varying cutout size for various boundary conditions
It is observed that for smaller cut outs, a/s =0.2 to 0.4, the fundamental frequency increases slightly, both for
simply supported and clamped edge condition. This indicates that stiffness of the plate reduces marginally when
a smaller cut out is present. With the increase in cut out size, a/s =0.4 to 0.8, the fundamental frequency
increases rapidly. A drastic increase in frequency is observed at a/s=0.8 for the plate clamped on all four edges.
Thus for moderate and large cutouts the stiffness of the plate increases rapidly with the cut out area. The results
are in conformation with the observations of Ali and Atwal [9].
For second and third modes, the frequency decreases for increasing cutout ratio up to 0.6. After that, the
frequency increases considerably. It is also observed that the second and third modes are showing similar
frequencies. This is due to the similar mode shapes occurring owing to the structure symmetry.
2.3.2. Study of Dynamic Behavior of Laminated Composite Plate Structure under Thermal Load
Here the effect of thermal load on the dynamic behaviour of laminated composite plate (1.0m x 1.0m) with a
square cutout (a/s=0.4), for different side to thickness ratio, has been analyzed. The temperature is varied from
300K to 400K. The material used in the analysis is graphite /epoxy.
Table 5: Degraded Elastic Moduli for varying Temperature

Elastic moduli Temp (K)
(GPa) 300 325 350 375 400
E1 130 130 130 130 130
E2 9.5 8.5 8.0 7.5 7.0
G12 6.0 6.0 5.5 5.0 4.75
Table 6: First Three Natural Frequency (Hz) for varying Temperature
SS CC
temp (K) temp (K)
Mode No.
300 325 350 375 400 300 325 350 375 400
1 63.0 43.0 - - - 130.7 116.8 101.2 83.4 63.3
s/t = 75 2 126.3 115.6 - - - 189.2 179.1 168.6 158.5 150.1
3 126.3 115.8 - - - 189.2 179.3 169.6 160.7 153.7
1 93.7 81.8 67.8 52.2 34.9 194.6 184.5 174.6 164.8 156.3
s/t = 50 2 186.7 178.9 170.2 161.9 155.5 279.1 271.0 262.1 253.5 246.9
3 186.7 179.0 170.5 162.6 156.5 279.1 271.1 262.3 254.1 247.9
1 224.8 219.5 211.4 203.5 197.9 459.1 451.1 441.6 432.1 425.1
s/t = 20 2 427.6 422.4 411.9 401.2 394.3 626.6 619.3 605.7 591.6 582.5
3 427.6 422.4 411.9 401.2 394.3 626.6 619.3 605.7 591.7 582.5
305
Degraded properties of the material as indicated in Table 5 have been used in the analysis. Simply supported
and clamped boundary conditions are taken in the analysis. The layup sequence is taken as (0/90/90/0). First
three natural frequencies (Hz) are shown in Table 6. The fundamental frequency is plotted in Figure 4. From the
figure, it is observed that with increase in temperature, generally the structure turns soft thus reducing its
stiffness. The decrease in natural frequency is nearly nonlinear for thinner plates (s/t=75). Instability occurs at
350K for the thinner plate at simply supported boundary condition. Clamped boundary condition imparts higher
stiffnesses to the structure than simply supported conditions. From the figure, it is also observed that, the less
stiff a structure is, chances of instability is more with increase in temperature.
500
450 s/t = 75 (SS)
400
s/t = 75 (CC)
Frequency (Hz)
350
300
250 s/t = 50 (SS)
200
150 s/t = 50 (CC)
100
50 s/t = 20 (SS)
0
280 330 380 430 s/t = 20 (CC)
Temperature (K)
Figure 4: Graphical plot for fundamental frequency vs. varying temperature for various boundary conditions and
side to thickness ratio
2.3.3. Study of Dynamic Behavior of Laminated Composite Plate Structure under Thermal Load for various
Layup Sequences
Here the effect of thermal load on the dynamic behaviour of simply supported laminated composite plate (1.0m
x 1.0m, s/t=75 ) with a square cutout (a/s=0.4), for different fibre angle, has been analyzed.
90
80 0/90/90/0
Frequency (Hz)
70
0/90/0/90
60
50 45/-45/-45/45
40 45/-45/45/-45
30
20 60/-60/-60/60
10 60/-60/60/-60
0
280 300 320 340 360 380 400 420
Temperature (K)
Figure 5: Graphical plot for fundamental frequency vs. varying temperature for various layup sequence
The temperature is varied from 300K to 400K. The material used is same as given in previous case study. The
fundamental frequencies (Hz) are plotted in Figure 5. From the Figure, it can be seen that (45/-45)2 is the stiffest,
whereas (0/90)2 have the least stiffnesses. Also, symmetric and antisymmetric cross ply , shows instability much
earlier than others. The rate of drop of natural frequency is less for (60/-60)2.
306
3. CONCLUSION
This paper is concerned with the dynamic analysis of laminated composite plate structures with a cutout under
temperature load. For the numerical simulation of the problem finite element technique has been used. Various
cutout sizes, plate thicknesses, thermal load, lay up sequence and boundary configurations have been considered
in the present study. A set of new results are presented. It has been shown that boundary condition plays an
important role in providing stiffness of a plate structure with a cutout. This is also observed that high
temperature turns the structure soft and there is reduction in stiffnesses as well as natural frequency. Less stiff
plate has higher chances of instability under higher temperature load. Also instability of a structure largely
depends on the layup sequence of the laminated composite structure.
REFERENCES
[1] Whitney, J.M. and Ashton, J.E. (1971). Effect of environment on the elastic response of layered composite
plates. American Institute of Aeronautics of Aeronautics Journal, 9, 1708-1713.
[2] Dhanraj, R. and R.Palaninathan., (1990), Free Vibration of Thermally Stressed Composite Laminates,
Journal of Sound and Vibration, Vol. 39, pp.365-378.
[3] Shen, C. H. and Springer, G. S., (1977), Effects of Moisture and Temperature on the Tensile Strength of
Composite Materials, Journal of Composite Materials, 11, 2-16.
[4] Shen, C. H. and Springer, G. S., (1977), Environmental Effects on the Elastic Moduli of Composite
Materials, Journal of Composite Materials, 11, 250-264.
[5] Sai Ram, K.S. and Sinha, P.K. (1992). Hygrothermal Effects On The Free Vibration Of Laminated
Composite Plates. Journal of Sound And Vibration, 158(1), 133-148.
[6] Lee H.P., Lim S.P. and Chow S.T. (1990), Prediction of natural frequencies of rectangular plates with
rectangular cutouts, Computers & Structures,36 (5), 861-869.
[7] Kumar, A. And Shrivastava, R.P., (2005). Free Vibration Of Square Laminates With Delamination Around
A Central Cutout Using HSDT. Composite Structures, 70, 317-333.
[8] Reddy, J.N., (1979), Free Vibration Of Anisotropic, Angle-Ply Laminated Plates Including Transverse
Shear Deformation By The Finite Element Method. Journal of Sound and Vibration, 66, 565-576.
[9] Ali, R. Atwal, S.J., (1980). Prediction Of Natural Frequencies Of Vibration Of Rectangular Plates With
Rectangular cutouts. Computers & Structures., 12, 819-823.
307
Paper No.: 053
SEISMIC VULNERABILITY ASSESSMENT OF INDIAN STANDARD CODE

DESIGNED MIDRISE AND HIGHRISE RC FRAMED STRUCTURES USING
FRAGILITY ANALYSIS
1 2
M.V.N.Sivakumar and B.Vineel Kumar
1 2
Assistant Professor, M.Tech Student
Department of Civil Engineering, Birla Institute of Technology & Science, Pilani-Hyderabad Campus,
Hyderabad, India
E-Mail: mvns@hyderabad.bits-pilani.ac.in
ABSTRACT
Accurate quantification of seismic risk requires a correct estimation of the seismic hazard and a good evaluation
of the seismic vulnerability of structures through an appropriate earthquake damage model. Existing
vulnerability assessment procedures consist of finding the expected damage level induced in each kind of
element at risk for a given hazard level. The damage data used within these methods are based on the
post-seismic surveys, expert judgment, analytical simulations or combinations of these. Hence, each data source
has associated with some advantages and disadvantages.
The vulnerability of a certain structure or a whole category of structures is determined in the form of earthquake
intensity-damage relationship. These relationships yield the probability distribution of the occurrence of damage
for a given earthquake intensity and are most frequently presented in the form of damage probability matrices
and fragility curves. Fragility curves/surfaces are useful tools for showing the probability of structural damage
due to earthquakes as a function of ground motion indices. Fragility curves are generally developed using a
single parameter to relate the level of shaking to the expected structural damage. The present study is mainly
focussed on to develop the fragility curves for midrise reinforced concrete frame buildings. Here Fragility
curves are constructed in terms of spectral displacement (Sd).
KEYWORDS: Fragility analysis, Seismic vulnerability, Framed structures, RC frame design
1. INTRODUCTION
Earthquake risk assessment is a process of estimating the probability of damage to a given structure due to
scenario earthquake. Vulnerability of a structure is an important consideration for risk assessment methodology.
The seismic vulnerability of a structure can be described as the extent of its exposure to the intensity of ground
shaking. Vulnerability assessment of a structure includes estimation of the probability of a given level of
damage to any structure for given level of earthquake [1]. The level of damage depends on performance of
structural and non-structural elements during earthquake. Vulnerability of a structure can be expressed in terms
of damage functions which represents the probability of damage level for a given earthquake.
Severe earthquakes applies complex loading to the structure. So, the structure is shaken well into the inelastic
range. To analyse the structures which were suffered from these earthquakes, we need an analysis tool which
can provide deep insight into the structural elements, which control performance during the earthquakes.
Inelastic analysis procedures only can give details about post elastic deformation of structural members. But
unfortunately Indian codes don’t have such provisions for the nonlinear analysis of structures. Pushover analysis
is one such a tool which can give required information about inelastic response of the structure [2].
Accurate quantification of seismic risk requires a correct estimation of the seismic hazard and a good evaluation
of the seismic vulnerability of structures through an appropriate earthquake damage model [3]. The seismic
vulnerability of a structure can be described as its susceptibility to damage by ground shaking of a given
intensity. The aim of a vulnerability assessment is to obtain the probability of a given level of damage to a
given building type due to a scenario earthquake [4]. Existing vulnerability assessment procedures consist of
finding the expected damage level induced in each kind of element at risk for a given hazard level [5]. Existing
vulnerability curves can be classified into four generic groups of empirical, judgmental, analytical and hybrid.
The damage data used within these methods are based on the post-seismic surveys, expert judgment, analytical
simulations and combinations of these respectively.
Though the evaluation of vulnerability of existing Reinforced Concrete (RC) Moment Resisting Framed (MRF)
buildings is not new, the application of the same techniques to non-ductile or gravity load designed buildings is
not so well developed in India. Hence, in this paper an attempt is made to assess the vulnerability of mid-rise
RCMRF structures designed as per IS code provisions. The analysis technique proposed in ATC-40 [6] is being
used for the performance evolution of RC Moment resisting framed buildings.
2. SEISMIC VULNERABILITY ASSESSMENT METHODS
Each vulnerability assessment method considers different damage scales to build vulnerability curves;
frequently used damage scales include the MSK scale [7] and the EMS98 scale [8]. In empirical vulnerability
procedures, the damage scale is used to produce post-earthquake damage statistics, whereas in analytical
procedures this is related to limit-state mechanical properties of the buildings, such as inter-storey drift capacity.
2.1 Empirical methods
Empirical curves use mainly the damage information from the past earthquakes. Observational data obtained
from post-earthquake surveys is the most realistic method of deriving fragility relationships because all the
details of the building inventory are included in addition to topography, site, source characteristics, and
soil-structure interaction effects [9]. Even though, empirical functions are derived from realistic sources of
damage. It has some drawbacks like, post-earthquake surveys do not include all the necessary information such
as the building materials and seismic design provisions and damage information of buildings.
2.2 Analytical methods
Analytical methods are based on the estimation of the damage distributions that are obtained from the response
of the structural model subjected to seismic input. The seismic input can be represented by a response spectrum
or an acceleration time-history. Fragility curves derived analytically differ based on modeling levels of the
structure such as consideration of the nonlinearities in the structures and load effects (large deformation or
P-Delta effects).
Advances in modelling capabilities coupled with a lack of sufficient earthquake ground motion data motivated
the development of fragility curves using analytical and simulation based methods. Researchers have employed
several analysis techniques at different sophistication levels, such as consideration of modelling nonlinearities,
accurate characterisation ground motion using artificial algorithms and computation of response to the dynamic
inputs, etc. to develop analytical fragility curves.
Analytical vulnerability curves adopt damage distributions simulated from the analyses of structural models
under increasing earthquake loads [10]. Significant improvements of this work from the past studies include
better representation of the structure when subjected to dynamic loads and characterisation of the potential
ground motions.Analytical approaches for the generation of vulnerability curves are becoming very attractive in
terms of the ease and efficiency by which data can be generated. A variety of analysis procedures have been
followed ranging from elastic analysis of equivalent single degree of freedom system to nonlinear analysis of
3D models. The choice made for the analysis method, structural idealization and input earthquake loading
309
strongly influence the derived fragility curve. But still these methods need to be fully exploited to the limits of
their potential.
2.3 Expert judgment
Judgmental based fragility assessment of structures based on predictions and opinions of expert panels of civil
engineers having good experience in the field of earthquake engineering. It mainly relies on judgment of some
experts who are asked to provide an estimate of the mean loss or probability of damage of a given element for
different levels of seismic loading. Probability distribution functions are fit to the expert predictions to represent
the range of damage estimates at each intensity level. Sometimes expert opinions are based on local structural
types, typical configuration, detailing. Hence if the fragility curves derivation is characterized by construction
practices that differ significantly from country to country.
Damage Probability Matrices based on expert judgement and opinion with an expert panel of 50 Earthquake
engineering experts were first introduced in ATC-13 [11] and some of the fragility curves proposed in HAZUS
[12] are also developed on judgmental basis. The main drawback of this method is that results solely rely on
individual experience of experts consulted. However, it has the advantage of not being affected by the lack of
extensive damage data in case of empirical approaches or the reliability of the structural model in case of
analytical methods.
2.4 Hybrid methods
Hybrid fragility curves are the result of a combination of methods, such as analytical and observational data, or
analytical models with expert judgment. Hybrid vulnerability curves attempt to compensate the deficiencies of
one method with some other sources like, scarcity of observational data judgmental data and modeling
deficiencies of analytical procedures by combining data from the different sources. In recent years these data are
further supplemented with experimental test results. However, due to the cost and time required for full-scale
testing and since small-scale testing is non-definitive on similitude grounds, a very limited number of
parameters can be investigated on experimental basis. Here mainly the difficulty arises with the hybrid method
is that calibration of results obtained from different methods including different sources of uncertainties.
3. FRAGILITY ANALYSIS
Fragility function of structures is the major requirement of seismic loss estimation which is widely used in the
seismic risk management. Seismic risk assessments were carried out on populations of buildings to identify the
urban areas most likely to undergo large life and economic losses during an earthquake. The results of such
studies are important in the mitigation of losses under future seismic events [13]. It has been a common trend to
express the fragility of structures to earthquakes as a relationship between the ground motion severity and
structural damage. Fragility curves and Potential Damage Matrices (PDM) is the most common format of this
relationship. Both methods describe the conditional probability of exceeding different levels of damage at given
levels of ground motion severity. As the names could suggest, fragility curves express the data in a graphical
format while PDM express it numerically.
Fragility curves represents conditional probability statements that give the likelihood that a structure will sustain
or exceed a specified level of damage for a given ground motion intensity measure. Damage functions or
fragility curves for buildings, are lognormal functions that give the probability of reaching or exceeding
different levels of damage for a given level of ground motion. Each fragility curve is characterized by a median
value of ground motion (or failure) and an associated dispersion factor (lognormal standard deviation). Ground
motion is quantified in terms of peak ground acceleration (PGA) and ground failure is quantified in terms of
permanent ground displacement (PGD). This paper describes the HAZUS [12] methodology for the estimation
of earthquake damage to buildings given knowledge of the model building type and an estimate of the level of
ground shaking (or degree of ground failure). The extent and severity of damage to structural and non-structural
310
components of a building is described by one of five damage states: None, Slight, Moderate, Extensive, and
Complete. Damage states are defined for each model building type by physical descriptions of damage to
building elements.
3.1 Description of Structural System and Classification
The structure considered for the analysis purpose is Indian Type Reinforced Concrete bare Frame Structure.
Reinforced Concrete Moment Resisting Frames (C1) from HAZUS [12] building classification was considered
to be closely matched building type to represent Indian Scenario RC Frame structures for damage assessment.
Further RC MRFs are classified based on number of stories as,
• Low-Rise (1-3 Stories),

• Mid-Rise (4-7 stories),
• High-Rise (8 and above)
3.2 Seismic Design Level
As per HAZUS [12] methodology, building stock represented by population of a given model building type
designed to either High-Code, Moderate-Code, or Low-Code seismic standards, or not seismically designed,
referred to as to a Pre-Code buildings. Older areas of construction, not conforming to modern standards, should
be considered for lower level of seismic design. For example, in areas of high seismicity, buildings of newer
construction (e.g., post-1973) are best represented by High-Code damage functions, while buildings of older
construction would be best represented by Moderate-Code damage functions, if built after about 1940, or by
Pre-Code damage functions, if built before about 1940 (i.e., before seismic codes existed). Pre-Code damage
functions are appropriate for modeling older buildings that were not designed for earthquake load.
3.3 Description of Building Damage States
Building damage varies from “none” to “complete” as a continuous function of building deformations (building
response). The Methodology predicts a structural and non-structural damage state in terms of one of four ranges
of damage or “damage states”: Slight, Moderate, Extensive, and Complete. For example, the Slight damage state
extends from the threshold of Slight damage up to the threshold of Moderate damage. General descriptions of
these damage states are provided for all model building types with reference to observable damage incurred by
structural and non-structural building components. Damage predictions resulting from this physical damage
estimation method are then expressed in terms of the probability of a building being in any of these four damage
states.
Descriptions for Slight, Moderate, Extensive, and Complete structural damage states for RC MRF model
building types are provided in table 1. For estimating casualties, the descriptions of complete damage include
the fraction of the total floor area of each model building type that is likely to collapse. Collapse fractions are
based on judgment and limited earthquake data considering the material and construction of different model
building types.
Slight Structural Damage: Flexural or shear type hairline cracks in some beams and columns near joints or
within joints.
Moderate Structural Damage: Most beams and columns exhibit hairline cracks. In ductile frames some of the
frame elements have reached yield capacity indicated by larger flexural cracks and some concrete spalling.
Non-ductile frames may exhibit larger shear cracks and spalling.
Extensive Structural Damage: Some of the frame elements have reached their ultimate capacity indicated in
ductile frames by large flexural cracks, spalled concrete and buckled main reinforcement; non-ductile frame
elements may have suffered shear failures or bond failures at reinforcement splices, or broken ties or buckled
main reinforcement in columns which may result in partial collapse.
311
Complete Structural Damage: Structure is collapsed or in imminent danger of collapse due to brittle failure of
non-ductile frame elements or loss of frame stability. Approximately 13%( low-rise), 10%(mid-rise) or 5%
(high-rise) of the total area of C1 buildings with Complete damage is expected to be collapsed.
3.4 Cumulative Damage probability matrices of particular damage state
The probability of being in or exceeding a given damage state is modelled as a cumulative lognormal distribution.
For a given median spectral displacement Sd and standard deviation β ds for a particular damage state ds , design
level the conditional probability of being in or exceeding is defined by
 1  s 
P [ ds |Sd ] =  ln  d  
 β ds  sd ,ds  
Where, sd ,ds is the median value of spectral displacement at which the building reaches the
threshold of the damage state, ds . β ds is the standard deviation of the natural logarithm of spectral
displacement of damage state, ds .
4. SEISMIC FRAGILITY ANALYSIS OF MID-RISE STRUCTURES
Fragility curves express the probability of structural damage due to earthquakes as a function of
ground motion indices. In the present study, fragility curves are constructed in terms of spectral
displacement. 5 storied and 8 storied RC framed structures are studied under seismic lateral load
pattern using nonlinear analysis tool SAP 2000 [13] to establish the performance point. After obtaining
performance point, damage state variability values are taken from HAZUS [12] to develop damage
probability functions.
A certain fragility curve shows the conditional probability that a certain damage state shall be
exceeded under an earthquake of certain intensity. Each fragility curve is defined by an Intensity
measure parameter corresponding to the damage state and the associated variability. In other words, for
analytical approaches if we use nonlinear static analysis spectral displacement value at the
performance point can be used to obtain damage probability functions. So, to estimate the variability
associated with the spectral displacement there is a compulsory need of statistical data and
unfortunately for Indian conditions such data is not available. So, median and damage state variability
parameters are taken from HAZUS [12] methodology. The set of curves are shown in figure 2 & 3 and
the damage probability matrix presented in table 1.
Table 1: Damage probability matrices

Damage Cumulative
State Probability
Slight 0.887328491
Moderate 0.634564006
Extensive 0.173227026
Complete 0.013535016
Fig.1: Analytical model of a typical 8 story frame
312
Fig.2 Fragility curves for 5-stotied building represented by the selected building
Fig.3 Fragility curves for 8-stotied building represented by the selected building
5. CONCLUSIONS
Seismic performance evaluation as per capacity spectrum method and development of fragility curves have been
generated for the considered building. These curves are site specific and the design is based on Indian Standard
codes of practices. The procedure for the development of fragility curves has been discussed and presented for
different classes of building. These buildings are designed for vertical loads and analyzed for lateral seismic
forces. HAZUS methodology has been implemented for the development of fragility functions. The seismic
response of the structures has been found to be quite sensitive to the type of the building and uncertainty
associated with the variable parameters considered in the present study. Also the seismic fragility development
accounts for uncertainties associated with the structural response and therefore, provide better confidence for the
practicing designers.
313
REFERENCES
[1] Seyedi, D. M., Gehl, P., Douglas,J., Davenne, L., Mezher, N, and Ghavamian, S., Development of seismic
fragility surfaces for reinforced concrete buildings by means of nonlinear time-history analysis, Earthquake
Engineering And Structural Dynamics, 39, 2010, pp. 91–108.
[2] Calvi G.M., Pinho R., Magenes G., Bommer J.J., Restrepo-Velez L.F., Crowley H.. Development of seismic
vulnerability assessment methodologies over the past 30 years. Journal of Earthquake Technology 43(3),
2006, 75-104.
[3] Pratibha S And M.Prasad, Seismic vulnerability of existing RC buildings in India, 13th World Conference on
Earthquake Engineering, August 1-6, 2004, Paper No. 1207.
[4] Medvedev, S. and Sponheuer, W.. MSK Scale of Seismic Intensity, Proceedings of the Fourth World
Conference on Earthquake Engineering, Santiago, 1969, Chile, Vol. 1, p.A2.
[5] Grünthal, G. (editor). Cahiers du Centre Européen de Géodynamique et de Séismologie Volume 15 –

European Macroseismic Scale 1998, European Center for Geodynamics and Seismology, 1998,
Luxembourg.
[6] ATC. Seismic Evaluation and Retrofit of Concrete Buildings, Report No. ATC-40, Applied Technology
Council, 1996, Redwood City, CA.
[7] Whitman, R.V., Reed, J.W. and Hong, S.T. Earthquake Damage Probability Matrices, Proceedings of the
Fifth World Conference on Earthquake Engineering, Rome, Italy, 1973, Vol. 2, pp. 2531-2540.
[8] Bernardini, Seismic damage to masonry buildings, 1999, BR rotterdam, Netherlands.
[9] Spence, R., Coburn, A.W. and Pomonis, A.. Correlation of Ground Motion with Building Damage: The
Definition of a New Damage-Based Seismic Intensity Scale, Proceedings of the Tenth World Conference on
Earthquake Engineering, Madrid, Spain, 1992, Vol. 1, pp. 551-556.
[10] Orsini, G.. A Model for Buildings’ Vulnerability Assessment Using the Parameterless Scale of Seismic
Intensity (PSI), Earthquake Spectra, 1999, Vol. 15, No. 3, pp. 463-483.
[11] Sabetta, F., Goretti, A. and Lucantoni, A. Empirical Fragility Curves from Damage Surveys and Estimated
Strong Ground Motion, Proceedings of the 11th European Conference on Earthquake Engineering, Paris,
1998, France, pp. 1-11.
[12] HAZUS-MH. Multi-Hazard Loss Estimation Methodology: Earthquake Model HAZUS-MH MR5
Technical Manual, 2011, Federal Emergency Management Agency, Washington DC.
[13] Rossetto, T. and Elnashai, A. Derivation of Vulnerability Functions for European-Type RC Structures
Based on Observational Data, Engineering Structures, 2003, Vol. 25, No. 10, pp. 1241-1263.
[14] Singhal, A. and Kiremidjian, A.S. Method for Probabilistic Evaluation of Seismic Structural Damage,
Journal of Structural Engineering, ASCE, 1996, Vol. 122, No. 12, pp. 1459-1467.
[15] Masi, A. Seismic Vulnerability Assessment of Gravity Load Designed R/C Frames, Bulletin of Earthquake
Engineering, 2003, Vol. 1, No. 3, pp. 371-395.
[16] Rossetto, T. and Elnashai, A. A New Analytical Procedure for the Derivation of Displacement-Based
Vulnerability Curves for Populations of RC Structures, Engineering Structures, Vol. 7, 2005, No. 3, pp.
397-409.
314
[17] Applied Technology Council. Earthquake damage evaluation data for California. ATC-13, 1985, Federal
Emergency Management Agency.
[18] Kappos, A.J., Pitilakis, K. and Stylianidis, K.C. Cost-Benefit Analysis for the Seismic Rehabilitation of
Buildings in Thessaloniki Based on a Hybrid Method of Vulnerability Assessment, Proceedings of the Fifth
International Conference on Seismic Zonation, Nice, France, 1995, Vol. 1, pp. 406-413.
[19] Mackie, K., and Stojadinovic, B. Fragility Basis for California Highway Overpass Bridge Seismic Decision
Making, PEER Report 2005/02, Pacific Earthquake Engineering Research Center . 2005, University of
California, Berkeley, CA.
[20] Karthik Narayan Ramanathan, Next Generation Seismic Fragility Curves for California Bridges
Incorporating The Evolution In Seismic Design Philosophy, Ph.D Dissertation, Georgia Institute Of
Technology, Aug 2012.
[21] FEMA. NEHRP guidelines for the seismic rehabilitation of buildings (FEMA 273). Washington (DC)
1997, Building Seismic Safety Council.
[22] Sigmudi A. Freeman, Development and use of Capacity Spectrum Method, 6th US NCEE Conference on
Earthquake Engineering,1998, Washington.
[23] Newmark N.M and W.J.Hall, Earthquake Spectra and Design, Earthquake Engineering Research Institute,
1982, Okland, California.
[24] IS 456: 2000, Indian Standard Plain and Reinforced Concrete Code of Practice (Fourth Revision)”, Bureau
of Indian Standards, New Delhi.
315
Paper No.: 054
SEISMIC BEHAVIOR OF SLAB-STRUCTURAL WALL JUNCTION IN RC

BUILDING
1 2
Snehal Kaushik and Kaustubh Dasgupta
1
Research Scholar, Dept. of Civil Engineering, Indian Institute of Technology Guwahati, Guwahati, India
2
Email: k.snehal@iitg.ac.in, kd@iitg.ac.in
ABSTRACT :
In multistoreyed Reinforced Concrete (RC) frame-wall buildings, the junctions between structural walls and
floor slabs play an important role in the transfer of lateral forces from the floor slab to the wall and finally to the
foundation below. During strong earthquake shaking, the dynamic forces may lead to high stress concentration
at the junction and subsequent localized failure. The current seismic design guidelines do not prescribe formal
analysis and design methodology for the same.
In the current study, a five-storied RC wall-frame building with symmetric plan is modeled using the computer
program SAP2000. The beams and columns are modelled using two-noded frame elements and floor slab and
structural walls are modelled using four-noded shell elements. Sufficiently fine meshing is adopted at the
junction region in order to obtain the forces and the moments. Linear static analysis is carried out under design
load combinations, as prescribed in IS: 1893 (Part 1) - 2002. The end regions of wall and slab are designed
against the critical forces and moments obtained through the load combinations.
Subsequently, lumped plasticity model is considered by assigning flexural plastic hinges at the ends of beams
and columns of all the stories of the building. Considering the material nonlinearity in structural walls, two
different models are investigated, namely (i) with elastic structural walls and (ii) with inelastic structural walls.
Inelastic behavior of walls is obtained by modeling it as a multi-layer shell element with actual stress-strain
curves of concrete and steel reinforcement. Displacement-controlled nonlinear static analysis of the models is
carried out and stress-deflection relationships at wall-slab junction are studied. Different levels of inelasticity
are developed at the end regions of the walls for analyses along the width and length of the building.
KEYWORDS: Slab-Wall Junction, Shear wall, Non-linear static analysis (Pushover).
1. INTRODUCTION
Shear walls are specially designed structural walls incorporated in buildings to resist lateral forces that are
produced in the plane of the wall due to earthquake forces. Slab-Wall connections in structure resisting lateral
forces constitute a potential weak link in the preferred load path from slabs to Walls, thereby influencing the
pattern of lateral load distribution to the vertical members of system. In the current study, linear and
displacement controlled non-linear static analysis of a five-storied RC wall-frame building with symmetric plan
is carried out using the computer program SAP2000. The stress-deflection relationships at wall slab junction are
studied for different models. Considering the material nonlinearity in structural walls, two different models are
investigated.
2. PAST RESEARCH
An economic form of construction for high-rise apartment buildings consists of spatial assemblies of shear walls
and floor slabs. In the past research, Coull and Wong [1985] had determined the distribution of shear stresses at
the slab wall junction and concluded that the junction between the floor slab and structural wall is subjected to
severe stress concentration. Coull and Chee [1983] carried out a finite element analysis of the coupled slab wall
system to obtain the induced bending moment and shear forces. A design method was proposed based on the
estimated shear forces for checking against punching shear failure in slabs. To avoid redistribution of forces
from walls to other elements not necessarily designed for lateral load resistance, Pantazopoulou and Imran
[1992] proposed that a slab-wall connection design must be done considering the stress concentration. They
observed that the design of slab-wall connection is governed by the same requirements as shear walls, although
it is a common practice to omit shear design consideration due to long connection boundary between slabs and
walls. Greeshma and Jaya [2013] have done experimental and analytical study of the floor slab – shear wall
connection by considering different reinforcement detailing when subjected to gravity and lateral cyclic loading.
The specimens detailed with 90° bent slab bars with slab shear reinforcement exhibited higher ultimate strength
compared to the other specimens. They concluded that the exterior shear wall – slab joint with slab shear
reinforcement and bent 90° at the joint can be effective in moderate to high seismic risk region.
3. LINEAR STATIC ANALYSIS
In the current study, a five-storied RC wall-frame building (Figure 1) with symmetric plan is modeled using the
computer program SAP2000 [CSI, 2011]. In case of linear static analysis, the assumption is such that the
member deformations, and also the stress-strain curves of materials, will always be linear. The material and
sectional properties of the various RC elements used in the present study are shown in Table 1.
3m
A B C D E F
4
3m
3.2 m
3m
3
3.0 m
3m
2
3.2 m
3m
1
3.0 m 3.2 m 3.0 m 3.2 m 3.0 m
(a) Plan (b) elevation

Figure 1: Model considered in the analysis.
3.1 Modeling
The building is assumed to be founded on rocky stratum and soil-structure interaction effects are not considered
in the study. Thus, the translational and rotational degrees of freedom are restrained at the bottom nodes. In
addition, the effect of unreinforced brick masonry infill walls was not considered in this analysis.
The beams and columns are modeled using two-noded frame elements and floor slab and structural walls are
modeled using four-noded shell elements. Sufficiently fine meshing is adopted at the junction region in order to
obtain the forces and the moments. Linear static analysis is carried out under design load combinations, as
prescribed in IS: 1893 (Part 1) – 2002 [IS:1893, 2002]. The end regions of wall and slab are designed against
the critical forces and moments obtained through the load combinations. The force resultants at the slab-wall
junction of the RC frame building were compared for different load combinations.
317
Table 1 Properties of RC elements used in the analysis
RC Members
Grade of concrete M25 (fck = 25 MPa)
Grade of steel reinforcement Fe415 (fy = 415 MPa)
Weight density (kN/m3) 25
Poisson’s ratio 0.3
Modulus of elasticity Ec (MPa) 25000
Column size (m) 0.30 × 0.30
Beam size (m) 0.225 x 0.30
Thickness of floor slab (m) 0.12
Thickness of shear wall (m) 0.15
3.2 Results
Linear static analysis was carried out under the action of vertical dead loads (DL), live loads (intensity taken as
3 kPa) and earthquake loads (EQ). The earthquake load is applied as horizontal forces at all nodes of each floor.
No live load was considered on the roof of the structure. The force resultants obtained for the load combination
1.5(DL+EQ) was found to be critical for the considered model. The bending moments at the wall-slab junctions
of various floor levels are obtained considering the effect of both slabs and walls. It was observed that there is
consistent increase in bending moment at wall-slab junction with the decrease in floor level.
Table 2 Maximum bending moment at slab-wall junction

Wall: 1-CD Load Case: 1.5(DL+EQ)
Floor Level Roof level Floor 4 Floor 3 Floor 2 Floor 1
Bending Moment +EQX 5.67 10.70 97.94 274.06 497.26
(kNm) -EQX -5.67 -10.70 -97.94 -274.06 -497.26
Wall: A-23 Load Case: 1.5(DL+EQ)
Bending Moment +EQY 4.60 13.08 96.69 269.4 509.04
(kNm) -EQY -4.60 -13.08 -96.69 -269.4 -509.04
4. NONLINEAR STATIC ANALYSIS
Adequate seismic performance implies that available strength and deformation capacities of a structure must be
greater than the demands imposed by a design earthquake. In nonlinear static analysis, any structure is pushed
under a pre-specified lateral force or lateral displacement [FEMA 356, 2000] with the monitoring of possible
nonlinear behaviour at various locations. Finally, the total base shear is plotted against a reference displacement
to obtain the global capacity curve for the structure. In the present study, displacement-controlled pushover
analysis of the building was carried out considering material non-linearity only.
In the present study, lumped plasticity is considered in the frame members by assigning flexural plastic hinges
at the ends of beams and columns of all the stories in the building. Regarding the behaviour of structural walls,
two different models are investigated, namely (i) with elastic walls and (ii) with inelastic walls. Inelastic
behavior of walls and slabs is obtained by modeling them as a multi-layer shell element with actual stress-strain
curves of concrete and steel reinforcement.
The length of the plastic hinge for the beams and columns was assumed as 10% of the length of the member.
Three different plastic hinges are modelled, namely (a) axial force – bending moment (P-M) hinge, (b) moment
– rotation (M-θ) hinge, and (c) shear force – shear deformation (V-∆) hinge. The P-M and M-θ hinges were
specified using the auto hinge option in SAP2000, using Tables 6-7 and 6-8 of FEMA 356. The plastic hinge
rotation capacities can be estimated by calculating moment curvature relationship using realistic material stress
strain curves as discussed in Section 3.3. After analysing the models, stress-displacement relationships at
318
wall-slab junction are studied. Different levels of inelasticity are developed at the end regions of the walls for
analyses along the directions of width and length of the building.
4.1 Models
Pushover analysis of following two models was carried out to compare the stresses at slab-wall junction.
Model 1: In this model, beams and columns are modeled using lumped plastic hinges whereas, floor slabs and
structural walls are considered elastic.
Model 2: In this model, beams and columns are modeled using lumped plastic hinges whereas, floor slabs are
elastic, while shear walls are modeled as multi–layer shell element with confined and unconfined concrete
characteristics.
4.2 Multi-layer Shell Element
In multi-layer shell element, different material properties are assigned to various layers so that the response of
the structural wall can be directly correlated with the material constitutive characteristics [Miao et al., 2006].
Two layers of vertical reinforcement are modeled with the thickness of each layer obtained by assuming
smeared reinforcement. Since the moment generated due to earthquakes will be resisted by the vertical steel
only, the thickness of only one vertical layer of steel is modeled in layers of the section and material angle of
this reinforcement is made 90⁰. Figure 2 shows a typical multi-layer shell element and the various layers defined
in the analysis.
Layer 1 Concrete layer

Layer 2
Layer 3 Smeared steel layer
Concrete layer
Layer 4
Layer 5 Concrete layer
Figure 2: Typical cross-section of a multi-layer shell element [Miao, et al., 2006].
4.3 Material Properties
In case of nonlinear static analysis, the realistic stress-strain curves for concrete have to be invoked with both
unconfined and confined characteristics. Although various models have been proposed, one of the recent models
[Karthik and Mander, 2011] has been used for the present study. The stress-strain curves used for confined
concrete, unconfined concrete and the reinforcement layers of the multi-layered element in structural wall are
shown in Figure 3.
4.4 Results
For the two models mentioned in Section 3.1, stresses are observed at the wall-slab junction section and
subsequently, the stress-displacement relationships are obtained. The displacement is monitored for a roof-level
node in the building.
319
35 600
Unconfined Concrete
30 Confined Concrete 500
25
Stress (MPa)
Stress (MPa)
400
20
300
15
200
10
5 100
0 0
0 0.005 0.01 0.015 0.02 0.025 0.03 0 0.05 0.1 0.15 0.2
Strain (mm/mm) Strain (mm/mm)
(a) (b)
Figure 3: Stress-strain curves for (a) concrete and (b) reinforcement in layered shell element
4.4.1 Model 1: Elastic Shear Wall and Slab
With elastic behaviour of the shell elements and lumped nonlinearity in the frame members, the lateral
displacement capacities of the building are 50mm and 43mm for pushover analyses in the directions of length
and width of the building respectively. The variations of the compressive and tensile stresses in concrete are
shown in Figures 4 and 5 respectively. The stresses follow a similar pattern in both the cases.
0 10 20 30 40 50 60 0 10 20 30 40 50
0 0
-2 -2
-4
-4
-6
-6
Stresses (MPa)
Stresses (MPa)
-8
-8
-10
-10
-12 Ground Floor
-12 First Floor
-14 Ground Floor
-14 Second Floor
-16 First Floor
Third Floor
-18 Second Floor -16
Roof Level
Third Floor
-20 -18
Roof Level
-22 -20
Displacement (mm) Displacement (mm)
(a) (b)
Figure 4: Variation of compressive stress in concrete layer with lateral displacement for pushover analyses in
the direction of (a) length and (b) width of the building
4.4.2 Model 2: Elastic Slab and Layered Shear Wall
With elastic shell model of slab, multi-layer shell model of structural walls and lumped nonlinearity in the frame
members, the lateral displacement capacities of the building are 43mm and 47mm for pushover analyses in the
directions of length and width of the building respectively. The low displacement capacities are primarily due to
the presence of stiff structural walls along both the directions as well as the chosen algorithm for nonlinear static
analysis.
The plastic hinge formation starts with the boundary element column ends at lower stories and then propagates
to upper stories (Figure 6). In the final stage of pushover analysis, hinges are also formed in the beams adjacent
to the walls. As the building plan is symmetric, similar hinge patters are obtained for the bays with structural
walls. Thus, the nonlinearity is observed mostly in the boundary element region and the web of the wall is still
elastic. The variation of stresses in various material layers of the wall in the junction region are shown in
320
Figures 7 to 10. As compared to other regions of slab and wall, significantly high stresses are observed in the
wall-slab junction region; this indicated vulnerability towards possible failure in this region during strong
earthquake shaking.
35 20
30
Ground Floor Ground Floor
15
25 First Floor
First Floor
Stresses (MPa)
Stresses (MPa)
Second Floor
Second Floor
20 Third Floor
Third Floor
Roof Level 10
Roof Level
15
10
5
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50
(a) (b)
Figure 5: Variation of tensile stress in concrete with lateral displacement for pushover analyses in the direction
of (a) length and (b) width of the building
For all the material layers in the walls, the observed stress-displacement curves for roof and third floor levels are
almost similar. Large variation in magnitude of stresses is observed for the lower floors. The compressive
stresses in the concrete layers (both unconfined and confined layers) do not reach the failure values, possibly
due to low lateral displacement level of the entire building. Similarly, the compressive and tensile stresses in the
reinforcement layers of the walls do not reach the yield level. Large stresses and possible failure of materials
may be observed in case of time history analysis of the building models under strong earthquake motions.
(a) (b)
Figure 6: Formation of plastic hinges in the final stage of pushover analyses in the direction of (a) length and (b)
width of the building
5. CONCLUSIONS
The salient conclusions obtained from the present study are noted as follows:
(a) For linear static analysis, maximum force resultants in the wall-slab junction region are obtained for the
load combination 1.5(DL+EQ). There is a consistent increase in bending moment with decrease in floor level.
321
0 10 20 30 40 50 0 10 20 30 40 50
0 0
-1 -1
-2 -2
Stresses (MPa)
Stresses (MPa)
-3 -3
-4 -4
Ground Floor
-5 Ground Floor -5
First Floor
First Floor
-6 -6 Second Floor
Second Floor
Third Floor
-7 Third Floor -7
Roof Level
Roof Level
-8 -8
(a) (b)
Figure 7: Variation of compressive stress in unconfined concrete layer with lateral displacement for pushover
analyses in the direction of (a) length and (b) width of the building

0 10 20 30 40 50 0 10 20 30 40 50
0 0
-0.2 -0.2
Stresses (MPa)
Stresses (MPa)
-0.4 -0.4

-0.6 First Floor -0.6 First Floor
Second Floor Second Floor
-0.8
Third Floor -0.8
Third Floor
Roof Level Roof Level
-1 -1
(a) (b)
Figure 8: Variation of compressive stress in confined concrete layer with lateral displacement for pushover
analyses in the direction of (a) length and (b) width of the building

0 10 20 30 40 50 0 10 20 30 40 50
0 0
Stresses (MPa)
Stresses (MPa)
-50 -50

First Floor First Floor
-100 -100
Second Floor Second Floor
Third Floor Third Floor
Roof Level Roof Level
-150 -150
(a) (b)
Figure 9: Variation of compressive stress in reinforcement layer with lateral displacement for pushover analyses
in the direction of (a) length and (b) width of the building
(b) Low lateral displacement levels are observed possibly due to the presence of stiff structural walls.
322
200
Ground Floor 200
First Floor Ground Floor
Second Floor First Floor
150 150 Second Floor
Third Floor
Third Floor
Stresses (MPa)
Roof Level
Roof Level
Stresses (MPa)
100
100
50
50
0
-50
0 10 20 30 40 50
0 10 20 30 40 50
Displacement (mm)
Displacement (mm)
(a) (b)
Figure 10: Variation of tensile stress in reinforcement layer with lateral displacement for pushover analyses in
the direction of (a) length and (b) width of the building
(c) For nonlinear static analysis, the stress-displacement curves are almost similar at roof level and third floor
level wall-slab junctions.
(d) The various layers of materials in layered shell model of structural wall do not show failure level stresses.
(e) For nonlinear static analysis, the wall-slab junction region shows more vulnerability towards possible
seismic failure during strong earthquake shaking than other regions of wall and slab.
6. ACKNOWLEDGEMENT
REFERENCES
[1] Coull, A., and Chee, W. Y., (1983), Design of Floor Slabs Coupling Shear Walls, Journal of Structural
Engineering ASCE, Vol. 109, No. 1, pp. 109-125.
[2] Coull, A., and Wong, Y. C., (1985), Effect of Local Elastic Wall Deformations on the Interaction between
Floor Slabs and Flanged Shear Walls. Journal of Building and Environment, 20, pp. 169-179.
[3] FEMA 356, (2000), Prestandard and Commentary for the Seismic Rehabilitation of Buildings. Federal
Emergency Management Agency, USA.
[4] Greeshma, S., and Jaya, K. P., (2013), Effect of Slab Shear Reinforcement on the Performance of Shear Wall
– Floor Slab Connection, Journal of Performance of Constructed Facilities ASCE. Vol. 27, No. 4, pp.
391-401
[5] IS: 1893 (Part-1), (2002), Indian Standards Criteria for Earthquake Resistant Design of structures, Bureau of
Indian Standards, New Delhi.
[6] Karthik, M. M., and Mander, J. B., (2011), Stress-Block Parameters for Unconfined and Confined Concrete
Based on a Unified Stress-Strain Model, Journal of Structural Division, Proceedings of the American
Society of Civil Engineers, Vol. 137, No. 2, pp. 270-273.
[7] Miao, Z.W., Lu, X.Z., Jiang, J. J., Ye, L. P., (2006), Nonlinear FE Model for RC Shear Walls Based on
Multilayer Shell Element and Microplane Constitutive Model, Computational Methods in Engineering and
Science, EPMESC X, 2006, China.
[8] Pantazopoulou, S., and Imran, I., (1992), Slab-Wall Connections under Lateral Forces, ACI Structural
Journal, Vol. 89, No. 5, pp. 515-527.
[9] CSI, (2011), SAP2000 Analysis Reference SAP Manual, Computers and Structures Inc., California, USA.
323
Paper No.: 055
INVESTIGATING EFFECTIVENESS OF DYNAMIC STRAIN RESPONSE FOR

DAMAGE DETECTION IN COMPOSITE BEAMS
Sanjay Goswami1,2, Partha Bhattacharya2 and Subham Rath2

1
Dept. of Computer Applications, Narula Institute of Technology, Kolkata, India
2
Dept. of Civil Engineering, Jadavpur University, Kolkata, India
Emails: sanjaygoswamee@gmail.com, p_bhatta@yahoo.com and subham.juciv@gmail.com
ABSTRACT:
This work particularly focuses on simulated investigation into the effectiveness of Dynamic Strain Response
(DSR), combined with Wavelet Based Signal Processing and Neural Pattern Recognition, as a mean to detect
and locate damages in beams. These beams are useful in manufacturing light weight structures such as
automobiles, aircrafts and aerospace vehicles. The methodology suggested involves a Finite Element Model of
specimen structure being excited by a transient impact vibration and the dynamic strain responses are picked up
by strain sensors at other parts of the beam. Picked signals are cleansed by low-pass digital filters to smooth out
noise and features are extracted by means of Continuous Wavelet Transform and Principal Component Analysis.
The principal components as features are then used to train a Neural Classifier which is used to classify the
signals corresponding to various damage scenarios in the structure. The damage scenarios are generated by
simulating stiffness value changes at a specific location, and moving the location across the length of the beam,
thus making Extent of Damage and Location to be the identifiable parameters of the damage. For better
performances, optimal sensor location points have been identified through repeated random simulations and
analysis. Simulation experiments have shown that the adopted approach has achieved success rates of 90.88% in
case of Damage Extent identification, while 87.98%in case of Damage Location identification. The results have
been compared with a similar simulation experiment involving displacement response (DR), in place of DSR,
and the proposed one has been found to be slightly superior to the other one.
KEYWORDS: Structural Health Monitoring, Finite Element Modeling, Vibration Signal Processing,
Wavelet Transform, Neural Networks applications.
1. INTRODUCTION
The field of Structural Health Monitoring (SHM) is an inter-disciplinary field involving Electrical Engineering,
Electronics Engineering, Computer Science, and Structural Engineering.
Several approaches have been proposed in history, ranging from tap-testing to guided-wave signal processing
based approaches. Currently prevailing vibration based techniques are specifically emphasised by modern day
researchers (Farrar et al 2001, Carden et al 2004 and Raghavan et al 2007). Qiao (2009), in his thesis, presented
some of the most relevant signal processing techniques suitable for vibration based damage detection, while
Staszewsky et al (2004) elaborated the usage of signal processing based damage detection techniques in
aerospace domain.
The sub-field of signal processing based damage detection is essentially a problem of statistical pattern
recognition (Raghavan et al 2004, Fan et al 2007), involving neural and other pattern recognition techniques.
The effectiveness of wavelet transform, particularly in case of signal processing based SHM, has been
emphasized by Sohn et al (2004) and Taha et al (2006).
In earlier works by the same authors (Goswami et al 2012a and Goswami et al 2012b), detection of single and
multiple damages were investigated using displacement responses sensed by piezoelectric displacement sensors.
The focus of the current work is to study the effectiveness of Dynamic Strain Response (DSR) data over
displacement response data, in identifying single damages, and their locations, in cantilever beam structures.
The procedure followed is explained in the following sections.
2. DEFINITION OF THE FINITE ELEMENT BEAM STRUCTURE MODEL
0.2m 0.001 m
1.5m
Fig. 1 Dimensions of the Beam Structure considered
The beam structure model considered for the present study has the following dimensions – 1.5m length, 0.2m
width and 0.001m thickness. The material properties include Elasticity modulus – 200 GPa, Poissons’ Ratio - 0.3
and Density -7800kg/cu.m. One end of the beam is fixed and the other kept free, giving it a cantilever class (Fig.1).
The free end is subjected to mechanical impact excitations, and the vibrations are picked near the fixed end to
study occurrence of any damage in the structure. The structure is discretized into 110 elements in the length
direction and 8 elements in the width direction (Fig. 2).
To localize the damages, the structure is logically divided into 7 segments along the X-direction (Fig. 2), out of
which, segments 2 to 6 are selected for introducing numerical damages and study them. Segment 1 is meant for
receiving the impact excitations, while segment 7 is meant for picking up the vibration signals from.
The structural behavior is simulated using a Finite Element Method (FEM) model on Ansys ver. 11.0 platform,
with the following model properties: Material Type - Shell 63, Section Properties - Rectangular, Element Size -
0.02m and 0.025m for alternate zones (for meshing convenience only).
∆x
982 996 Strain Measurement Point
Y
989
30 48 205 340 475 610 745 880

(0.0, 0.2, 0.0) (0.3, 0.2, 0.0) (0.5, 0.2, 0.0) (0.7, 0.2, 0.0) (0.9, 0.2, 0.0) (1.1, 0.2, 0.0) (1.3, 0.2, 0.0) (1.5, 0.2, 0.0)
Load Point 1 2 3 4 5 6 7
(Node 53)
(0.0, 0.1, 0.0)
X
(0.0, 0.0, 0.0) (0.3, 0.0, 0.0) (0.5, 0.0, 0.0) (0.7, 0.0, 0.0) (0.9, 0.0, 0.0) (1.1, 0.0, 0.0) (1.3, 0.0, 0.0) (1.5, 0.0, 0.0)
1 4 190 325 460 595 730 865
Fig.2 The F E M model of the beam with node numbers (and xyz coordinates). Produced by Ansys 11.0
3. METHOD FOR IDENTIFYING OPTIMUM MODAL STRAIN RESPONSE POSITION
After meshing the model, modal analysis was performed on it with the frequency range specified from 0-200 Hz,
and 10 mode shapes were generated. On getting 10 different mode shapes, Rotational Surface Strain values
between alternate nodes along the length and at a particular ordinate (y=0.1) were extracted for each mode. All the
325
values for a particular mode were normalized by dividing them with the maximum surface strain value for that
particular mode.
In the next step, absolute surface strain values were taken into account and sum of these absolute values from each
alternate nodes, for all the modes, were calculated. It is understandable that the nodes showing highest absolute
strain-values, in every mode, will have a higher sum compared to those which showed lower positive or negative
values (refer modal shapes in Fig. 3).
Based on this principle, the surface strain values are first converted into equivalent voltages (because of the
presence of piezoelectric sensors), corroborating with the following sets of equations;
σ
e33 lcbc
Cp =
tc (1)
Where lc, bc and tc are length, width, and thickness of the sensor respectively. Considering only the effect of strain
along 1-direction, the voltage generated by the sensor can be expressed as
d 31 YC bC ∫ ε1dx
VC =
lc
Cp
(2)
Assuming the value of ε1 to be averaged over the gauge length, we define a sensitivity parameter
Sq=d31 Yc lc bc (3)
Where Yc is the Young’s modulus of the piezoelectric material, d31 is related to the strain in the particular direction
and e33 is the permittivity. Values of all the parameters are summarized in Table 1.
Table 1
d31 Yc ∆X e33 lc bc tc Cp k
-274 71 0.04 30.1 1.5 0.2 0.001 9030 -0.01723
Putting these values into Eqs. 1, 2 and 3, we get the values of the voltage generated at different points of the
structure. The nodes with the highest voltage-value sums, for the 10 modes generated, are the most optimum
nodes that will give high strain values, irrespective of what the mode may be. In this way, five optimum nodes
showing high surface strain values are found out and the node corresponding to the highest voltage value is
selected to be the optimal node.
Various mode shapes studied and the information about the five nodes with highest strain voltage output is
summarized in the next sub-sections.
3.1. Mode shapes of the beam structure
This section illustrates the 10 mode shapes that are being studied to find the most optimal strain response points.
Mode shapes for Modes 1 through 10, along with their information, shown in Fig. 3(a) through (j), respectively.
326
(a) Mode 1 (b) Mode 2
(c) Mode 3 (d) Mode 4
(e) Mode 5 (f) Mode 6
327
(g) Mode 7 (h) Mode 8
(i) Mode 9 (j) Mode 10
Fig.3: First 10 Mode Shapes for the given beam structure.
3.2. Various strain response values and identification of the maximum strain response sites
The following table, Table 2, summarizes only max five Surface-Strain and Voltage sums, contributed by all the
modes, and the locations where these are occurring.
Table 2: Maximum Strain and Voltage Values

Max Values (Sums) Location (mm)
Surface Strain Voltages (abs) x coordinate (node)
8.738141 0.15055 1.48 (996)
7.727971 0.13315 1.46 (982)
6.717801 0.11575 1.44 (975)
5.994813 0.10329 1.22 (826)
5.880237 0.10132 0.50 (209)
5.845876 0.10072 0.94 (649)
It can be seen from the table that the positions from where the surface strain response can be maximum are
between nodes 982 and 996. Rotational strain from nodes 982 and 996 are sensed and the surface strain for the
328
node 989 (1.47, 0.1, 0.0) are computed using the following formula
−
= ∗
∆ 2
(4)
Where, strain989 is the strain at node 989, roty996 and roty982 are Rotational Degrees of Freedom about Y-axis at
nodes 996 and 982 respectively, ∆x is the distance between nodes 996 and 982, and d is the thickness of the
beam (Fig. 2).
4. SIGNAL PROCESSING BASED DAMAGE IDENTIFICATION SCHEME
Signal Noisy
Strain Acquisition
Response Signal
Principal Continuous Low Pass

Component Wavelet Filtered
Wavelet Filter
Analysis Scalogram Signal
Transform
Neural Trained
Principal Neural
Array
Features Array
Training
Fig.4 The Pattern Recognition Scheme followed
The signal processing based pattern recognition scheme followed in this work is illustrated in Fig. 4. Strain
response signals are picked up by strain sensors fitted to the structure and passed over to a low-pass digital filter
to clean out the noise in those, if any. Filtered signals are then subjected to Continuous Wavelet Transform
(CWT) (Mallat 1989, Walker 2008) to extract out time-frequency data from the signals. The time-frequency data
are stored in a matrix called Wavelet Scalogram. The Scalograms undergo Principal Component Analysis (PCA)
(Haykin 1999, Konar 2005) for dimensionality reduction and production of principal features that represent the
entire information hidden in the signals. The principal feature vectors, corresponding to all the damage cases,
are then used to train a set of Neural Networks (Haykin 1999) to identify and localize the damages.
5. NUMERICAL EXPERIMENTATION, RESULTS AND DISCUSSIONS
Severity of the damages in the beam is simulated by varying the percentage of Elasticity Modulus (E) of the
structure material from 1% to 100%, at specific zones, one percent at a time. The damaged material property is
then relocated from zone 2 to zone 6 to simulate the location of damages. The free end of the beam is then
exposed to impact excitations at the geometric location (0.0, 0.1, 0.0) (Fig.2), and the travelling waves through
the body are sensed at the designated nodes 982 and 996, and strain computed for 989 with 500 time steps.
Since the situations are numerically created, the signals generated for the damage cases are pure. To simulate
real noisy environments the signals are contaminated with 5% Gaussian White Noise (Oppenheim et al 1999),
which are then subjected to low-pass Butterworth filter (Oppenheim et al 1999) to prune-out the noise. The
cut-off frequency is taken to be 1000 Hz, while the sampling frequency is 3000 Hz. The filtered strain signals
are then subjected to Continuous Wavelet Transform (CWT), with 256 scales of Daubechies-1 mother wavelet
(Mallat 1989, Walker 2008). Principal feature vectors obtained from dimensionality reduction of Scalograms
(out of CWT) are then used to train a Neural Network system (Multi Layer Perceptron) (Haykin 1999) to
identify and locate damages in the beam.
329
Damage identification success rates from strain response approach are compared with those from displacement
response approach, and the results are summarised in Table 3, for single damage case.
Table 3: Average success rates in identifying damages – Single Damage Case

Strain Response Displacement Response
Zone 87.98% 86.22%
Severity 90.88% 89.22%
It can be seen from the results that effectiveness of damage sensitivity increases if strain response signals are
considered, compared to displacement response signals.
6. CONCLUSION
It has been found from this numerical study that Dynamic Strain Response (DSR) can be used as data for
damage investigation purposes in beam structures. Neural damage classifiers also work better in case of DSR,
compared to displacement response (DR), with DSR approach scoring 87.98% and 90.88% as success rates in
identifying Zone and Location, respectively, over DR scoring 86.22% and 89.22% for the same. The
performance in case of DSR appears to be slightly better in case of beam damage detection. However, the
effectiveness of DSR approach in other higher-dimensional and complex structures remains a subject of study
for the future.
REFERENCES
[1] Carden E. Peter and Fanning Paul (2004), Vibration Based Condition Monitoring: A Review. Structural
Health Monitoring, 3(4), pp. 355-377.
[2] Farrar Charles R., Doebling Scott W. and Nix David A. (2001), Vibration-Based Structural Damage
Identification. Philosophical Transactions of Royal Society London, 359, pp.131–149.
[3] Fan Wei and Qiao Pizhong (2011), Vibration Based Damage Identification Methods: A Review and
Comparative Study. Structural Health Monitoring, 10(1), pp. 83-111.
[4] Goswami Sanjay and Bhattacharya Partha (2012a), A Scalable Neural-Network Modular-Array
Architecture for Real-Time Multi-Parameter Damage Detection in Plate Structures Using Single Sensor
Output. International Journal of Computational Intelligence and Applications, Vol. 11(4), pp.
1250024-1250045.
[5] Goswami Sanjay and Bhattacharya Partha (2012b), Pattern Recognition for damage detection in Aerospace
Vehicle Structures. In Proceedings of the 2012 Third International Conference on Emerging Applications of
Information Technology (EAIT2012), pp. 178 – 182, Nov. 30 - Dec. 1, 2012, Kolkata, IEEE Xplore.
[6] Haykin Simon (1999), Neural Networks – A Comprehensive Foundation, 2nd ed., Pearson, New Delhi.
[7] Konar Amit (2005), Computational Intelligence – Principles, Techniques and Applications, Springer, New
Delhi.
330
[8] Mallat S. (1989), A Theory for Multi-resolution Signal Decomposition: The Wavelet Representation. IEEE
Pattern Analysis and Machine Intelligence, vol. 11(7), pp. 674-693.
[9] Oppenheim Alan V., Schafer Ronald W. and Buck John R. (1999), Discrete-Time Signal Processing, 2nd ed.
Prentice-Hall Inc., New Jersey.
[10] Qiao Long (2009), Structural Damage Detection Using Signal-Based Pattern Recognition. Ph.D.
Dissertation, Kansas State University.
[11] Raghavan Ajay and Cesnik Carlos E. S. (2007), Review of Guided-Wave Structural Health Monitoring. The
Shock and Vibration Digest, 39, pp. 91-115.
[12] Sohn H., Park G., Wait J. R., Limback N. P., and Farrar C. R. (2004), Wavelet-Based Active Sensing for
Delamination Detection in Composite Structures, Smart Materials and Structures, 13(1), pp. 153–160.
[13] Staszewski W.J. and Worden K. (2004), Signal Processing for Damage Detection. In: Staszewski W.J., Boller
C. and Tomlinson G.R. (ed.) Health Monitoring of Aerospace Structures – Smart Sensor Technologies and
Signal Processing, John Wiley & Sons, Sussex, pp. 163-206.
[14] Taha M. M. Reda, Noureldin A., Lucero J. L. and Baca T. J. (2006), Wavelet Transform for Structural Health
Monitoring: A Compendium of Uses and Features. Structural Health Monitoring, 5(3), pp.0267-0296.
[15] Walker James S. (2008), A Primer on Wavelets and their Scientific Applications, 3rd ed., Chapman &
Hall/CRC, USA.
331
Paper No.: 056
FINITE ELEMENT ANALYSIS OF ROAD TUNNEL

1 2
A.Vanuvamalai and Dr.K.P. Jaya
1
Research scholar, Div. of Structural Engineering, Dept.of Civil Engineering, Anna University,chennai, India
2
Associate Professor, Div. of Structural Engineering, Dept.of Civil Engineering, Anna University,chennai, India
Email: avanu23@gmail.com, jayakp@annauniv.edu
ABSTRACT :
Underground structures have been widely adopted and used in the construction of infrastructure in urban area in
for the recent years. . In the present study performances of tunnels were investigated analytically . The analytical
methods estimate how the rock or soil behaves after the excavation of tunnel. Bidirectional single tube tunnel
without invert tunnel are modeled with FEM software PLAXIS. New Australian Tunneling Method (NATM) is
adopted in this model. The analyses were performed on natural geometry conditions. The performance of this
transverse section depends on the strength and ductility of its nodes. This paper presents the results of analytical
investigations.
KEYWORDS: Design, Concrete Lining, NATM, Underground Structures.
1. INTRODUCTION
This paper presents the findings from the design of 10m internal diameter reinforced concrete NATM tunnel
using 2D FEM analysis.
1.1. Background and Project Description
Ladakh area of J&K State is linked by two routes Leh – Srinagar , Leh – Manali . Annualy roads remain closed
between November to March – heavy snow fall / snow avalanches. Border road organization requires the road to
be opened throughout the year.Two tunnels proposed . Z – Morh is phase 1 tunnel 2.7 km from Sonamarg
towards Srinagar near the village Sitkhari. Phase 2 connecting Sonamarg to Leh. Phase 1 investigation and
exploration taken up in the year 2010- 2011. The Eastern portal of the tunnel to be established 400m from the
Shitkari bridge of Sonamarg Town. Western portal of the tunnel to be established near km69 on Srinagar -
Sonmarg road. Twin tunnels proposed for reasons of safety , ventilation & to keep temperature under control.
Figure 1- Location of Z-Morh Tunnel

2. GEOTECHNICAL CONSIDERATIONS
2.1. Ground Conditions
The geological investigations show that the project area is located in metamorphic rocks of Quartzites , slates /
Phyllites and Dolerite & Basalt. Especially at the portal areas the metamorphic bedrock is covered by thick
accumulations of block and rock fall debris. Geological L-Section of Z-Morh Tunnel; green…Panjal Trap,
blue…Zojila formation .
Figure 2- Geological longitudinal section of Z-Morh Tunnel
2.2. Geotechnical Design Parameters
The geotechnical design parameters for analyses were derived from field and laboratory test results, and based
on experience on local and similar ground conditions. The parameters used in the static design and analyes are
presented in table 1.
Table 1 – Rock mass properties for Ground Types
γ Specific weight
Φ Friction angle
c Cohesion
E Young’s modulus (primary loading); E-module unload/reload = 3xE
ʋ Poisson’s ratio
333
Table 2 – Tunnel Lining properties
EA / EI Axial stiffness / Bending stiffness of the element per m tunnel length

W Element weight per m tunnel length
Table 3 – Tunnel Rock Bolt properties
Lspac Spacing of rock bolts in direction of tunnel length

EA Axial stiffness of rock bolts per meter tunnel
Np Maximum axial tensional force per meter tunnel
3. ANALYSIS AND DESIGN
3.1.Design Approach
Numerical analyses were performed by the finite element code PLAXIS on simplified 2-D models assuming
plane strain conditions.The geometry of the tunnel with the main primary support elements was taken from the
design (support categories A to G). For numerical modelling “homogeneous” underground situation and a
“plain” ground surface were assumed. Therefore only one tunnel half was modelled due to symmetrical
conditions. The overburden was simulated by means of a 10 m thick layer introduced in the model. The
specific weight assigned at this layer (in kN/m³) resulted from the overburden depth diminished by the
numerical model height above tunnel top (20 m). The model boundaries were determined in a distance of
approximately 3 times the tunnel diameter to the outer tunnel circumference. The numerical model was
therefore about 30 m wide and 55 m high in total including the 10 m thick overburden layer. Cover is consider
1000m.The model was fixed at the lower boundary in x- and y-direction and at the lateral boundaries in
x-direction. The upper boundary was free to move. The tunnel primary lining was additionally fixed at the top in
the symmetrical axis against rotation (= symmetrical boundary conditions for plates).The mesh generation was
done automatically by the finite element program with 15-noded triangular shaped elements. Approx. 5 m
around the tunnel excavation the mesh was refined.
334
STRESSES, PLASTIC ZONE
Figure 3 - Figure 4 -
Mean principal effective stresses direction Mean principal effective stresses (shadings)
Figure 5- Relative shear stresses (shadings)
Fig. 5 shows the plastic zone around the excavation (red coloured zone) having a depth of app. 3 m.
DEFORMATIONS
Figure 6- Deformed mesh Figure 7- Total displacement (shadings)
335
Figure 8- Total displacements of primary lining
The maximum vertical deformation occurs at the top heading with approximately 4.7 mm. The maximum
horizontal deformation at the lower part of the bench is also approximately 4 mm.
INTERNAL FORCES
Figure 9- Axial forces of primary lining Figure 10- Bending moments of primary lining
Fig. 9 & Fig. 10 show the internal forces of the final calculation step. The maximum axial forces are determined
as per analysis to approximately 1.6 MN compression and 1.1 MN tension. The value of axial tension is
resulting from numerical singularities in the connection between bench and top heading lining and is therefore
neglected in further design of the primary lining. The average axial force in the primary lining is determined to
approximately 0.25 MN compression. The bending moments are determined to approx. ±8 kN/m.
336
RADIAL BOLTING
Table one shows the axial forces in the rock bolts of the final calculation step. It can be shown that no rock bolt,
except the lower left rock bolt on the side wall, reach the yield strength.
Figure 11- Axial forces rock bolts
4. CONCLUSIONS
Modern finite element software enables very detailed and high resolution 2D analysis to be conducted for
realistic scenarios. With such tools, one no longer needs to make relatively arbitary assumptions relating to
tunnel convergence or stress relief. A static analysis can then be carried out effectively to obtain the induced
forces onto the tunnel lining due to surface force events. Some of the observations made from this analysis
are,Increase in lining thickness decreased the deflection and increased the other design parameters.
ACKNOWLEDGEMENTS
The authors would like to thank and acknowledge the co-operation and permission from the Consultancy. The
authors acknowledge the financial assistance provided by Anna University, Chennai through the Anna Cenetary
Research Fellowship research grant.
REFERENCES
[1] EN 1997-1(2004); Eurocode7 – Geotechnical Design – Part 1; general Rules,2004.
[2] Eurocode2 (EN 1992-1-1) Design of concrete structures, Part 1-1; General rules and rules for buildings.
[3] PLAXIS b.v.: PLAXIS 2D 2011, Delft, The Netherlands, 2011.
[4] Hashash,Y,Hook.J,Schmidt.B,Chiang Yao.J.(2001) ,Seismic design and analysis of underground structures,

[5] Tunneling and Underground Space Technology, Vol.16, 247-293.
[6] Dasari, G., R., Rawlings, C., G., and Bolton, M., D., 1996. Numerical modelling of a NATM tunnel
construction in London clay, Proceedings of International Symposium on Geotechnical Aspects of
Underground Construction in Soft Ground, City University, Balkema, pp. 491-49.
337
[7] Sunil Sharma, William R. Judd (1991), Underground Opening Damage From Earthquake, Engineering
Geology, 30, 263-275.
[8] Robert Rowe (1992), Tunnel Engineering In Earthquake Area, Tunnels And Tunneling, 12, 41-44.
[9] Pan Changshi (1996), The Survey Of The Study On Earthquake Resistance Of Tunnel and Underground
Structure, The World Tunnel, 5, 7-16.
[10] Zhang Yu-Er, Bai Baohong, Zhang Yaohui, Wang Zanzhi (2003), Study On Subway Tunnel’s Behavior
Due To Seismic Damage And Methods Of Analyzing Seismic Response And Earthquake-Proof Means,
22:1, 70-74.
[11] Jin-Hung Hwang, Chih-Chieh Lu (2007), Seismic Capacity Assessment Of Old Sanyi Railway Tunnels.
Tunneling And Underground Space Technology, 22:7, 433-449.
338
Paper No.: 059
OPTIMIZATION OF A FIBER REINFORCED POLYMER WEB CORE SKEW

BRIDGE
1 2 2
Tushar Kanti Dey , Anupam Chakrabarti and Umesh Kumar Sharma
1
Research Scholar, Dept. of Structural Engineering, Indian Institute of Technology, IIT Roorkee, India
2
Associate Professor, Dept. of Structural Engineering, Indian Institute of Technology, IIT Roorkee, India
Email:tkdiitr@gmail.com, tusharkantidey@hotmail.com
ABSTRACT :
This paper describes an optimum design methodology for Fiber Reinforced Polymer (FRP) web core bridge
having skew orientation. In the present study, a number of bridge deck panels have been considered with
different skew angle orientation. The FRP web core skew bridge decks are considered to be simply supported on
two opposite edges while the other two edges are free. The bridge deck is subjected to self weight and a impose
live load over a patch area as per IRC 70R loading for the tracked vehicle. Structural optimization of the FRP
skew bridge deck has been carried out considering the depth of core, thickness of top and bottom face plates as
the design parameters. Numerous trials have been considered for the optimization; obtained by varying different
design parameters. The optimization has been carried out for skew bridge deck using criteria based on the
maximum deflection and it is also checked for critical buckling load and failure load. The optimization has been
performed by using modified PYTHON script generated by Graphical User Interface (GUI) of ABAQUS CAE.
The values of the design parameters are finalised based on the trail solutions corresponding to the minimum
volume of the FRP materials used leading to least cost. Thus an efficient optimization scheme has been
developed for the FRP web core bridge deck having different skew orientation in the study.
KEYWORDS: Web core, FRP composite, skew bridge deck, finite element
1. INTRODUCTION
Fiber reinforced polymer (FRP) is gaining popularity due to its high specific strength and specific stiffness. A
rib core bridge deck/panel system having lightweight core and laminated stiff FRP face sheets provides high
mechanical performance with minimum weight. In this structure, the core is made of different rib configurations
while the laminated face sheets are prepared by stacking FRP composite layers with different orientations.
Reduction in weights would raise the live load rating of the structure and would also help in rapid replacement.
Moreover, the replacement process of these lightweight decks takes relatively lesser time which has the benefit
of having shorter duration of traffic interruption as well faster new construction as demonstrated by first
generation of all FRP bridge decks. As the concrete decks are the major contributors of the dead load in a bridge
structure, the use of lightweight sandwich decks can help to achieve a more economical design of new bridges.
The foam core sandwich construction is quite popular in many structures, but this is not preferred in bridge
decks as de-lamination at the interfaces between the core and face sheets can be produced easily by the impact
of wheels especially for bridges having larger spans.
Kellar et al. (2004) described the quasi-static and fatigue performance of hybrid bridge girders composed of
cellular FRP bridge decks and steel girders and demonstrated that the well-established design method for steel–
concrete composite girders with shear stud connections can essentially be used for the design of such FRP-steel
girders. Aref et al. (2004) presented the finite element analysis (FEA) and simple methods of analysis of an
innovative hybrid FRP-concrete bridge superstructure system. Salim et al. (2006) determined an approximate
series solution using a one-term approximation of a macroflexibility series solution including deformations for
fiber-reinforced polymer (FRP) deck-and-stringer orthotropic bridge systems for analysis of FRP composite
highway bridges. Davalos et al. (2001) developed an analytical procedure for FRP honeycomb sandwich panels
based on analytical and experimental studies. Uddin and Abro (2008) demonstrated a unique approach for the
design and manufacturing of structurally efficient low cost thermoplastic composite bridge superstructures. Qiao
et al. (2000) outlined a systematic analysis and design approach for single-span all FRP deck/stringer bridges.
The concept of light weight and heavy duty FRP honeycomb panels, with sinusoidal wave core configurations
in the plane extending vertically between face laminates was introduced by Plunkett (1997). He and Aref (2003)
developed a good and easy to handle tool to allow engineers in the design of rib core FRP bridge decks. This
tool had provisions to estimate geometries and other properties without performing detailed finite element
analysis. Keller et al. (2007) presented a new concept for a lightweight hybrid FRP bridge deck. The sandwich
construction consisted of three layers: a FRP composite sheet with T-upstands for the tensile skin, lightweight
concrete for the core and a thin layer of ultra high performance fiber reinforced concrete (UHPFRC) as a
compression skin.
Through the available literature on the optimization of FRP bridge decks it is observed that it involves a large
number of design variables and constraints. Based on the studies conducted on the optimization of all FRP
composite bridge decks, a few of them are described briefly below.
Cohn and Dinovitzer (1994) suggested that optimization could become more attractive to practicing designers if
more concrete examples of its application were available, especially for realistic structures, loading conditions,
and limit states. Salim (2000) provided an optimum design solution to minimize the cost by utilizing Excel
Solver based on a nonlinear optimization technique to analyze the impact of high cost of FRP materials on the
design of bridge system. Mantell and Hoiness (1996) proposed and validated a general methodology for
assessing minimum cost beam design for various pultruded shapes of a glass fiber reinforced composite box
beam. Park et al. (2005) performed optimization of the geometry of bridge decks and properties of FRP
materials by a three dimensional numerical technique capable of handling more complex objective functions and
constraints and validated the performance of proposed FRP deck by numerous practical tests. Qiao et al. (2008)
demonstrated a combined homogenization and multi-objective technique for the optimization of the properties
of thin-walled sinusoidal honeycomb cores. Using a commercial FE analysis engine, Kim et al. (2003)
developed an optimization procedure based upon modified genetic algorithm (GA) for the optimum deign of
GFRP bridge deck of pultruded cellular cross-section and validated the results of the code by a commercial FEA
package. Burnside et al. (1993) outlined an optimization process for FRP bridge deck with cellular and stiffened
box geometries. Qiao (1997) presented the shape optimization of FRP beams with wide-flange section in his
thesis work.
In this present paper a simple technique is proposed for the optimization of FRP rib core skew bridge deck
structures using the using finite element package ABAQUS CAE 6.8 (2008). The model of the bridge deck
system has been generated in Graphical user interface (GUI) of ABAQUS CAE 6.8 (2008). The PYTHON
(2012) script of the model has been developed in the scripting interface of the software. Various trials are found
by varying the structural design parameters (i.e., thicknesses of top and bottom face plates; thickness, depth and
number of ribs in the core). The different design points of the trails are incorporated in the PYTHON (2012)
script in the form of nested loop to obtain the deflection of various trails. The initial dimensions of the design
parameters are chosen arbitrarily considering manufacturing criteria. The different trials satisfying the design
constraints (i.e., the deflection criteria) are further considered for optimization. The optimization function has
been chosen corresponding to the minimum volume leading to the optimum cost of material. Finally, the
optimum design configurations of the skew bridge deck are analyzed to satisfy the buckling and failure load
criteria.
2. OBJECTIVE FUNCTION
In the present study design optimization is carried out only structural components volume contribution. In this
optimization study, objective function is chosen to minimize the volume of the FRP skew bridge decks. The
following function computes the volume of the FRP material used:
340
= ∗ + ∗ + ∗ ∗ − − ∗ (1)
where L, D, W, tb, tt, tw, and n are the deck length, depth, width, thickness of bottom plate, thickness of top plate,
thickness of ribs and the number of ribs respectively (Refer Figure 1)
Top face
Web
Bottom face
Figure 1 Cross-sectional view of Skew bridge deck
3. OPTIMIZATION METHODOLOGY
\ In order to optimize the various parameters (thicknesses of top face plate, bottom face plate and rib and the
number of ribs in the core) of a FRP skew rib core bridge deck it is essential to have a suitable initial
configuration to begin with. Thus in the first step of the optimization process, based upon the deflection criteria,
a configuration is adopted so that it just satisfies the governing deflection equation (Max. deflection < Span
(L)/800) as per AASHTO (1998). Once this initial value is obtained by trial and error method using the GUI of
ABAQUS CAE 6.8 (2008), necessary commands are fed into it and a model is created and run for analysis.
Upon completion of the analysis, the journal file created by the software is modified as per the set rules of a
typical PYTHON (2012) script compatible with it. The full optimization coding is done using PYHTON (2012)
scripting language with the help of GUI of ABAQUS/CAE 6.8 (2008).
A range of feasible solutions with an interval of 1mm in thicknesses of each component is incorporated in this
model script along with the introduction of nested loop. The generated nested loop allows the user to generate
various trial solutions upon execution of the model script in ABAQUS CAE 6.8 (2008). On completion of the
entire loop, trial solutions are generated by the post processing script made available to ABAQUS CAE 6.8
(2008) during the execution of the model script. The trial solutions satisfying the deflection criteria are chosen
out of the sample range and sorted in order of minimum volume used.
A second refinement optimization is carried out by choosing the range of top best trial solutions. The same
model script is reutilized for the next range of feasible solutions with various thicknesses of each component of
the FRP skew bridge deck. In the end all trial solutions of the refinement optimization satisfying the deflection
criteria are sorted according to the minimum volume used and the trial with the least volume is quoted as the
best configuration for that size of bridge deck.
It is obvious from the proposed optimization scheme that it has a wide range of application due to the
availability of multiple looping function which helps in handling multiple design parameters. The algorithm for
the modified code used in the present study is also illustrated by a typical flowchart in Figure 2.
4. MODELING
The model consist of components namely bottom face plate, unidirectional rib core, and top face plate. The
elements used for the present FE analysis are conventional shell elements having three displacements and three
341
rotational degrees of freedom. The S4R element type available in the Abaqus library has been used (S4R:
Conventional Stress/Displacement Shell, 4-node, Reduced Integration). The shape of the element is
quadrilateral having linear element geometry.
The skew bridge deck is assumed to have span along the longer dimension (i.e., 3m) and accordingly simply
supported boundary conditions are applied at the two opposite shorter edges (i.e., 7.5 m). The longer edges are
assumed as free. To manage the accuracy and computing resource, a mesh convergence study is performed. The
mesh density is increased repeatedly and the model is re-analysed until the results converge satisfactorily. In this
module, top and bottom face plate are meshed by 80 by 80 divisions and the rib core is meshed with a global
seed size of 80 mm. The interaction between the common nodes of the individual components is defined by
using the tie elements. These tie elements ensure perfect bond between the adjacent components. The analysis
is done using the general static loading option. The 70R tracked vehicle loading specification as per IRC: 6
(2000) has been adopted by including the effect of impact along with the self weight. After obtaining the
optimum configuration of the FRP skew bridge deck, it is also checked for buckling and failure criteria. The
buckling analysis is performed using the linear perturbation option. Various criteria for failure analysis are
adopted in the present study.
Figure 2 Typical flowchart for the proposed optimization code
5. DESIGN EXAMPLE
In the present study a 3 m x 7.5m bridge deck panel having laminated top face plate, bottom face plate and
unidirectional laminated rib is considered for optimization where the skew angles are varied. The depth of the
342
rib is varied according to the practical requirement. The various skew angles considered for this study are 30˚
and 45˚ in addition to the normal (0˚) configuration. The different depths considered for optimization are 150
mm, 175 mm and 200. These depths are selected to cover a wide range of available sections having different
numbers of ribs in the core. For each depth 5, 7 and 11 numbers of ribs are taken in the present study to get
different optimized configurations. The detail of the bridge panel is shown in the Fig. 1. IRC 70R track vehicle
loading (IRC: 6 (2000)) with impact effect is applied on the bridge decks in addition to the self weight. The
structural system is subjected to a typical patch load of 350 kN over an area of 4570 mm x 840 mm with an
intensity of 0.11 N/mm2. This vehicle loading is suitably arranged to get maximum deflection/maximum
bending stress or maximum shear stress. Specification regarding the permissible deflection is not available in
the Indian Standards. Therefore, the permissible deflection limit of Span (L)/800 as per FHWA's Advisory
(AASHTO (1998)) specifications is adopted in this study. The ply stacking sequences used for top and bottom
plate is 0/90/0 and rib plate is 45/-45. The material properties of FRP Composites used in the present problem
are shown in Table 1.
Table 1
Elastic properties (compressive)
Composite laminate elastic properties of FRP glass/vinlyester (QM6408)
EL 23 GPa
ET 18 GPa
GLT 9 GPa
νLT 0.25
Density 1826 kg/m3
6. RESULTS
It has been observed that the deflection criteria govern the design of FRP laminated structures in most of the
cases as they have relatively lower values of the elastic modulus. Therefore, optimization is performed for the
FRP rib core skew bridge deck based on the deflection criteria to obtain the minimum volume using iteration
method. The design parameters to be optimized are the thicknesses of top face plate, bottom face plate and ribs;
and the number of ribs. The depth of the rib is varied from 150 mm to 200 mm taking three different number
(i.e., 5, 7 and 11) of ribs and skew angle (i.e., 0˚, 30˚, 45˚) to obtain different optimized configurations for
practical applications. The best range of configurations with least volume is selected in this stage of
optimization for each configuration. Further refinement in the iteration process is performed on the range of
configurations obtained in the previous step to get the optimum design. Figure 3 represents the plot of the total
volume of the FRP skew bridge deck corresponding to different trial configurations.
1.14
1.12
1.10
Volume(X 10 mm)
8
1.08
1.06
1.04
1.02
1.00
0 20 40 60 80 100 120 140
Trail Numbers
Figure 3 Convergence of solutions
343
The optimum design values for various parameters involved for the various combinations of skew orientation of
FRP bridge deck are summarised in Table 2 for span of 3m.
Table 2 Summary of optimum design for span 3m and width of 7.5 m
Skew Depth(mm) Number of webs Optimized Configuration (mm) Volume

angle x 109(mm3)
Web thickness Top plate thickness
5 32 28.8 4.18
150 7 28 24.8 3.66
0 11 23 19.9 3.02
5 30 27.0 3.96
175 7 26 22.7 3.40
11 21 17.2 2.68
5 29 25.9 3.84
200 7 25 21.4 3.26
11 19 15.4 2.45
5 26 22.7 2.89
150 7 23 19.2 2.50
30 11 18 14.9 2.01
5 25 21.8 2.81
175 7 21 17.8 2.35
11 17 13.1 1.82
5 25 21.3 2.78
200 7 21 17.2 2.31
11 15.9 12 1.70
5 20 16.8 1.78
150 7 13 14.0 1.48
45 11 14 10.2 1.18
5 20 16.1 1.75
175 7 16 12.8 1.43
11 13 9.0 1.08
5 19.2 16.2 1.78
200 7 16 12.5 1.43
11 12 8.4 1.04
The results obtained from different analyses are presented in Table 3. Due to the identical configuration of top
and bottom plates having symmetrical ply orientation, the effect of the thickness of either of these components
is interchangeable. Therefore, the design curve in Fig. 4 are drawn considering equal and average thickness of
the top and bottom plates. The utility of these curves for design purpose is also viable.
344
32 Skew: 30°
Number of webs: 5
30 Depth of core: 150 (mm)
Thickness of webs (mm)

Depth of core: 175 (mm)
28 Depth of core: 200 (mm)
26
24
22
20
18
2.0 2.5 3.0 3.5 4.0

Span (m)
Figure 4 A typical design curve (θ=30) for web plate thickness for different Span (No. of webs: 5, Skew: 30º)
7. DISCUSSION AND CONCLUSION
A simple optimization technique for FRP rib core skew bridge deck has been proposed using a computer coding
generated in using ABAQUS CAE 6.8 (2008). The proposed method for optimization has a wide application
because it overcomes the tedious coding for various configurations of the same model. It has added advantage
due to the availability of special features of multi-looping which helps to simultaneously vary multiple design
parameters. This reduces the computational effort and produces efficient results. The sensitivity study reveals
the effect of different structural components on the design constraints.
The various conclusions that can be drawn are:
• It is observed that the design of FRP bridge decks is governed by deflection criteria due to their lower
values of elastic modulus.
• In all of the above configurations buckling is not found to be critical.
• Through the optimization study several design curves have been generated which should be useful for
the selection of the various components of a bridge deck as per requirements.
REFERENCES
[1] Keller T., Gurtler H. (2004). Quasi-static and fatigue performance of a cellular FRP bridge deck adhesively
bonded to steel girders. Composite Construction Laboratory CC Lab, Swiss Federal Institute of Technology,
BP Ecublens, Lausanne 1015, Switzerland.
[2] Aref A.J., Kitane Y., Lee G.C. (2004). Analysis of hybrid FRP-concrete multi-cell bridge superstructure.
Department of Civil, Structural, and Environmental Engineering, University at Buffalo, The State University
of New York, Buffalo, NY 14260, USA and Simpson Gumpertz & Heger Inc., Waltham, MA 02453, USA.
[3] Salim H.A., Barker M., Davalos J.F. (2006). Approximate series solution for analysis of FRP Composite
Highway Bridges. J. Compos. Constr., 10(4), 357-366.
345
[4] Davalos J. F., Qiao P., Xu X. F., Robinson J., and Barth K. E. (2001). Modeling and characterization of
fiber-reinforced plastic honeycomb sandwich panels for highway bridge applications. Compos. Struct., 52(3)
441–452.
[5] Uddin N., Abro A.M. (2008). Design and manufacturing of low cost thermoplastic composite bridge
superstructures. Engineering Structures, 30,1386–1395.
[6] Qiao P., Davalos J.F., Brown B. (2000). A systematic analysis and design approach for single-span FRP
deck/stringer bridges. Composites, Part B, 31, 593-609.
[7] Plunkett J.D. (1997). Fiber-reinforcement polymer honeycomb short span bridge for rapid installation. IDEA
Project Report.
[8] He Y., Aref A.J. (2003). An Optimization Design Procedure for fiber reinforced polymer web-core
sandwich bridge deck systems. Composite Structures, 60, 183-195.
[9] Keller T., Schaumann E., Vallee T. (2007). Flexural behavior of a hybrid FRP and lightweight concrete
sandwich bridge deck. Composite, Part A, 38, 879-889.
[10] Cohn M.Z., Dinovitzer A.S. (1994). Application of structural optimization. J. Struct. Eng., 120(2), 617-
650.
[11] Salem M. A. (2000). Optimum design of precast bridge systems prestressed with carbon fiber
reinforcement polymers. MS thesis, Concordia University, Montreal, Quebec, Canada.
[12] Mantell S.C., Hoiness B. (1996). Optimized design of pultruded composite beams. J. Reinf. Plast.
Compos., 15,758-778.
[13] Park K.T., Kim S.H., Lee Y.H., Hwang Y.K. (2005). Pilot test on a developed GFRP bridge deck.
Compos. Struct., 70, 48-59.
[14] Qiao P., Fan W., Davalos J.F., Zou G. (2008). Homogenization and optimization of sinusoidal honeycomb
cores for transverse shear stiffness. J. Sandwich Struct. Mat., 10, 385-412.
[15] Kim H.Y., Choi Y.M., Hwang Y.K., Cho H.N. (2003). Optimum design of a pultruded FRP bridge deck.
ECI
[16] Conference on Advanced Materials for Construction of Bridges, Buildings, and Other Structures III.
[17] Burnside P., Babero E.J., Davalos J.F., Rao H.G. (1993). Design optimization of an all-FRP bridge.
Proceedings of 38th Int. SAMPE Symposium.
[18] Qiao P. (1997). Analysis and design optimization of fiber-reinforced plastics (FRP) structural beam. PhD
dissertation, West Virginia University.
[19] ABAQUS CAE 6.8. (2008). Dassault Systèmes Simulia Corp.
[20] PYTHON 3.2.2. (2012). IDLE, Python Integrated Development Environment; Python Software
Foundation.
[21] AASHTO. (1998). LRFD bridge design specifications, 2nd Ed., Washington, D.C.
[22] Indian Road Congress. (2000). Standard specifications and code of practice for road bridges. Section : II
Loads and stresses., 4th revision, New Delhi.
346
Paper No.: 060
COST EFFECTIVE STRUCTURAL SYSTEMS FOR HIGH RISE BUILDINGS

SUBJECTED TO SEISMIC LOAD
G.V. Rama Rao1, M. Pavan Kumar2, Sk. Rasool3
1
Professor, Dept. of Civil Engineering, Andhra University, Visakhapatnam, Andhra Pradesh.
2
Asst. Professor, Dept. of Civil Engineering, BABA Institute of Technology, Visakhapatnam, Andhra Pradesh .
3
M.E (Strs) Student, Dept. of Civil Engineering, Andhra University, Visakhapatnam, Andhra Pradesh.
Email: pavanidea@ gmail.com
ABSTRACT:
A model for selecting cost effective moment resisting or shear wall framing system will be a valuable tool for
all decision makers. Engineers, in particular, will be able to select optimal concrete framing faster, thus
reducing design time and iterations. Furthermore, selection of economic framing system will also result in direct
cost savings for concrete buildings project. This investigation proposes a prototype cost effective model for
selecting either a moment resisting or shear wall framing system for residential buildings.
The study involves the design and cost estimation of apartment buildings representing moment resisting and
shear wall framing systems. The prototype cost effective model is calibrated for high rise buildings located in
transient earthquake regions of zone-II and zone-III. Designs are carried out according to IS 456 - 2000 using
STAAD PRO 2006 (force analysis and column design) software package. Shear walls are designed using
manual calculation according to IS: 13920-1993. The cost of the framing system is represented by the cost of
steel reinforcement alone and the framing systems are compared based on drift requirements and cost. The
preferred framing system should meet drift requirements and is lower in cost.
The results of this present study shows that the Moment Resisting Framing System is the cost effective choice
up to 5 storied apartment buildings subjected to seismic load of zone-II and Zone-III. Also the Shear Wall
Framing System is the cost effective choice for High rise buildings beyond 5 f1oors in seismic zones II & III.
KEYWORDS: MRF System, SWF System, STAAD PRO, Drift, Seismic Zone.
1. INTRODUCTION:
High rise structures have certain features. The structures are high & lead to higher vertical loads and higher
lateral loads in comparison with lower buildings. Buildings between 75 feet and 491 feet (23 m to 150 m) high
and the materials used for the structural system of high-rise buildings are reinforced concrete and steel. Vertical
loads on the high rise structures have Dead loads arise from the weight to the individual construction elements
and the finishing loads and Live loads are dependent on use depending on the number of stories, live loads can
be reduced for load transfer and the dimensioning of vertical load-bearing elements. It generally arises from
unexpected deflections, wind and earthquake loads. Unexpected Deflections arises from imprecision in the
manufacture of construction elements and larger components. Due to wind loads high-rise buildings are
susceptible to oscillation. It should not be viewed as statically equivalent loads, but must be investigated under
the aspect of sway behavior. The ability of wind loads to bring a building to sway must also be kept in mind.
Reinforced concrete frames, provided in both the principal directions, are effective in resisting both the vertical
and the horizontal loads. The brick walls are to be regarded as non load bearing filler walls only. The spacing of
frames varying from 4.0 m to 7.0m or more is closely related to the function of building. The slab thickness
should be as close to 100 mm as possible. This can be achieved by providing subsidiary beams in addition to the
frame beams. Single storey buildings of large storey heights (5.0 m or more), like electric sub-stations, etc. are
also made of framed structures, as brick walls of large height are slender and these may not be relied on to
support vertical loads.
2. METHODOLOGY:
The methodology adopted is to evaluate the cost-effective method of the moment-resisting and shear wall
framing systems for high rise building. The methodology includes analysis and design of the systems against
different vertical and lateral loads followed by cost estimation. This paper gives the detailed description of the
model considered for the framing system. For analysis and design, STAAD PRO-2006 software is used and
shear wall design was done manually. Cost estimation is done manually based on the quantities obtained from
computer analysis. Cost estimation deals with the criteria for calculating steel reinforcement quantities of
framing systems. The final step in the methodology outlines the comparison criteria which are the deflection
and steel cost. Deflection of the framing systems are checked against code requirements. The preferred
structural system must meet deflection requirements as well as lower in cost. The governing parameter is lateral
deflection.
Fig 1 Plan and elevation of 30 storey structure
3. DIMENSIONS OF MODEL STUDIED & LOADS CONSIDERED:
The plan and elevation details of the 30 storey structure chosen for comparative studies are shown in fig.1. The
analysis of any statically indeterminate structure like a frame demand prior knowledge of dimensions of
individual columns and beams of all floor levels. For this purpose the dimensions have been fixed through
348
preliminary simplified calculation of axial loads coming on columns at different floor levels and of bending
moment in beams in a typical floor solely under the action of vertical loads.
The dimension chosen for the 30 storey building are as given below
3.1 Column details:
Moment resisting & shear wall framing system

Here C1: Exterior columns, C2: Interior columns
Beam size : 500mm x 300mm , Slab thickness :120mm, Grade of concrete : M30, Grade of Steel : Fe415,Shear
wall thickness:200mm, fck :30 N/mm2, fY : 415 N/mm2
Also for the 5 storey, 15storey, 30 storied buildings the column and beams c/c distance remain same but the
section dimension varies as shown in Table 3.1.
Table 3.1Column Sizes
Range fck C1 C2
For 5 storey: Up to 5 floors 30 N/mm2 450 mm x 300mm 600mm x 300mm
For 15 Up to 15 floors 30 N/mm2 600 mm x 450mm 750 mm x 450mm

storey:
For 30 Up to 30 floors 30 N/mm2 750mm x 450mm 1200mm x 600mm

storey:
3.2 Loads Considered for the Design:
1. Dead Load: Dead load considered As per IS: 875-(part-I)

2. Live load: Live load considered from IS: 875 (Part-II)
Design live load intensity was taken as 2kN/m2
3. Seismic load: Seismic load from IS: 1893-2002
Zone factor Z=0.1 for zone-II and Z=0.16 for zone-III
Importance factor I=1
Reduction factor R= 3 for OMRF & R=5 for SMRF
3.3 Load Combinations Considered:
DL + LL, 1.5(DL +LL), 1.2(DL + LL + EL(X)), 1.2(DL + LL + EL (-X)), 1.5(DL + EL(X)), 1.5(DL + EL (-
X)), 0.9DL + 1.5EL (+X), 0.9DL + 1.5EL (–X)
4. ANALYSIS AND DESIGN:
The structure with different framing systems has been modeled using STAAD PRO-2006 software with the
above mentioned load conditions and combinations. The analysis is done for both moment resisting framing
system and shear wall framing system where as the analysis of a multi-storied frame for vertical as well as
lateral loads is a straight forward affair, incorporation of a shear wall into the system with commercially
available STAAD PRO programme was not that easy. Hence, a number of alternative methods need to be tried
out to arrive at a satisfactory method for the analysis of a frame attached to shear wall.
349
4.1. Moment-resisting framing system:
This comprises of columns and beams alone with support conditions as fixed. These columns and beams are
created using beam elements of the software. Here instead of slab panels created directly by software, the
vertical slab load calculated for per square meter was applied on the beams. For horizontal loads, the physically
present phenomenon that the floor slab at each floor level acting as very rigid horizontal beams which ensures
that the lateral deformations of all the nodes at any particular floor level are the same. This is known as
“Diaphram Action” of the horizontal slabs.
4.2. Shear wall framing systems:
This comprises of columns and beams as the framing systems but with two shear walls the end frames along
windward direction(18m span in plan) instead of columns and beams at these two ends.
4.2.1. Method using 2-noded elements for shear wall:
Here shear wall was initially created using two noded elements (i.e idealized as single column with equivalent
stiffness) and cross section was given as 3m x 0.2m and analysis was done.
5. RESULTS & DISCUSSION:
5.1. General Consideration:
Comparison of lateral-force resisting systems is done for each building category based on lateral loads, lateral
drift, material quantity and cost in terms of steel reinforcement alone. The building type is a multi storey
structure with an 18m x 40m plan area and 6m x 4.0m bay size. Lateral force considered is seismic alone for
zone-II and III. Lateral drift/deflection is checked against the requirements of clause 7.11.1 of IS 1893-2002 i.e.
Under transient seismic load the lateral sway at the top should not exceed 0.004 x hi, where hi is the storey
height of the ith floor. Cost of the framing system includes cost only of reinforcement in columns, beams, slab
and shear walls (wherever shear wall is used). The cost of foundation reinforcement is not included. Findings
are discussed below for 5, 15, 30 storey structures at zone-II and zone-III.
The steel quantity obtained here for 3D-frame and shear wall alone are combined to arrive at total weight of
steel required in case of shear wall framing system. Weight of steel obtained is shown in table5.1 below.
Table 5.1 Weight of steel obtained
No of storey’s Weight of steel obtained for (Zone-II) Weight of steel obtained for (Zone-III )
(Tons) (Tons)
Shear wall 3-D Frame Shear wall 3-D Frame
30 11.82 267.7 11.82 273.47
15 5.91 101.75 5.91 102.5
5 1.97 30.9 1.97 30.36
5.2. Comparison of steel reinforcement variation in percentage.
Comparison of steel reinforcement variation in percentage in all storey i.e 5,15 & 30 for moment resisting
framing system and shear wall framing ,taking number of storey’s along X-axis and % variation along y-axis in
Zone –II & Zone-III is as shown in the following table.
350
Table 5.2 Comparison of steel reinforcement variation in percentage in all storey i.e 5,15 & 30 for moment
resisting framing system and shear wall framing system
Load Weight of steel (Tons)

ZONE combination
5 storey 15storey 30storey
MRFS SWFS MRFS SWFS MRFS SWFS
II 1.5(DL+LL) 27.715 27.475 109.402 95.167 300.633 253.22
WORST 32.584 30.1 113.75 101.749 311.651 267.71
%Variation=steel(worst)x100
Steel(DL+LL) 117.56 109.55 104 106.91 103.66 105.72
1.5(DL+LL)
III 31.635 27.475 109.402 95.167 300.633 253.22
WORST 31.829 30.363 119.367 102.5 322.896 273.468
%Variation=steel(worst)x100 100.61 110.51 109.1 107.7 107.4 108

Steel(DL+LL)
5.3 Comparison of MRFS & SWFS system based on drift, material quantity and cost:
Moment resisting and shear wall framing are compared based on lateral drift, material quantity in terms of steel
reinforcement and cost for each lateral load category i.e. for earthquake Zone of II & III. If the two systems
meet drift requirement then cost is used to select the perfect framing system.
5.3.1. Five storey structure:
5.3.1.1. For Seismic zone-II:
Lateral Drift: The relation is shown between the lateral roof drift for the two framing systems and the drift limit
which is 12mm calculated from 0.004 x hi where hi is the storey height of ith floor .The lateral drift for the
moment resisting framing system is 3.24mm, while that of the shear wall framing system is 1.56mm. Both the
framing systems do not exceed the drift limit. Hence comparison is based on cost.
Material quantity: The moment resisting framing system has about 0.38% (0.29 tons) less steel reinforcement
Percentage and is calculated as the difference in quantities divided by quantity used in the moment-resisting
framing multiplied by 100. The moment resisting framing system is constructed from four edge columns,
twenty two external columns, and eighteen internal columns. Slab dimensions are kept constant for all storey
structures. The shear wall framing system is constructed from eighteen external columns and eighteen internal
columns .slab dimensions are same as that of MRFS.
Weight of steel required (5storey, zone-II):
351
Table 5.3.1Weight of steel required for 5 storey MRFS & SWFS system in Seismic Zone-II
Framing system Weight of the steel reinforcement

Columns & Beams slab Shear wall Total
MRFS 32.58 43 ------- 75.58
SWFS 30.9 43 1.97 75.87
Cost of shear wall framing system is 0.38% more expensive than the moment resisting framing system as can be
seen from the following table. The moment resisting framing system is the cost-effective framing system for
this case, since it is less expensive and meets drift requirements. It is therefore recommended.
Note: Cost of steel per Metric Ton is taken as RS 48,000/- as per SSR-2011 with September revision cost of
framing system (5storey, zone-II):
Table 5.3. 2 Cost of steel reinforcement for 5 storey MRFS & SWFS system in Seismic Zone-II
Cost of the steel reinforcement(Rupees)

Framing system Columns & Beams slab Shear wall Total
MRFS 15,63,840 20,64,000 -------- 36,27,840
SWFS 14,83,200 20,64,000 94,560 36,41,760
Difference Rs 13,920/-
5.3.1.2. For Seismic zone-III:
Lateral drift: The relation is shown between the lateral roof drift for the two framing systems and the deflection
limit which is 12mm calculated from 0.004 x hi where hi is the storey height of ith floor. The lateral drift for the
moment resisting framing system is 2.2mm, while that of the shear wall framing system is 1.6mm. Both the
framing systems do not exceed the drift limit. Hence comparison is based on cost.
Material quantity: The moment resisting framing system has about 0.67% (0.5 tons) less steel reinforcement
percentage which is calculated as the difference in quantities divided by quantity used in the moment-resisting
framing multiplied by 100. The moment resisting framing system is constructed from four edge columns,
twenty two external columns, and eighteen internal columns. Slab dimensions are kept constant for all storey
structures. The shear wall framing system is constructed from eighteen external columns and eighteen internal
columns. Slab dimensions are same as that of MRFS.
Weight of steel required (5storey, zone-III):
352
Table 5.3.3. Weight of steel required for 5 storey MRFS & SWFS system in Seismic Zone-III
Weight of the steel reinforcement

MRFS 31.83 43 -------- 74.83
SWFS 30.36 43 1.97 75.33
Cost of shear wall framing system is 0.67% more expensive than the moment resisting framing system as can be
seen from the following table. The moment resisting framing system is the cost-effective framing system for
this case, since it is less expensive and meets drift requirements. It is therefore recommended.
Cost of steel per Metric Ton is taken as RS 48,000/- as per SSR-2011 with September revision. Cost of framing
system (5storey, zone-III)
Table 5.3.4. Cost of steel reinforcement for 5 storey MRFS & SWFS system in Seismic Zone-III
Cost of the steel reinforcement(Rupees)

MRFS 15,27,840 20,64,000 35,91,840
SWFS 14,57,280 20,64,000 94,560 36,15,840
Difference Rs 24,000/-
5.4. Variation in Drift for the Zones II & III:
Variation = Actual Drift / Allowable Drift
MRFS SWFS Allowable MRFS SWFS

Storey Drift (mm) Drift (mm) Drift (mm) Variation in Drift Variation in
Drift
Zone - Zone- Zone II Zone- Zone II Zone- Zone II Zone- Zone II Zone-
II III III III III III
30 3.95 6.3 2.4 3.9 12 12 0.329 0.525 0.2 0.55
15 3.9 6.32 1.399 2.238 12 12 0.325 0.526 0.116 0.186
5 3.24 2.2 1.56 2.5 12 12 0.27 0.183 0.13 0.208
353
5.5. Comparison of weight of steel per unit area for the Two Systems:
Total weight of steel (Tons) Total area (sq.m) Wt/sq m (Kg/m2)

Stor
ey
Zone Zone Zone II Zone Zone Zone Zone Zone Zone Zone Zone Zone
II III III II III II III II III II III
30
567.65 578.89 535.52 541.29 21600 21600 21600 21600 26.28 26.80 24.79 25.05
15
241.75 247.36 235.66 236.41 10800 10800 10800 10800 22.38 22.90 21.82 21.88
5
75.58 74.83 75.87 75.33 3600 3600 3600 3600 20.99 20.78 21.07 20.925
5.6. Comparison of Cost per Unit Area for the two systems:
stor Total cost of steel (Rupees) Total Area(sqm) Cost/sqm

ey
Zone Zone Zone Zone Zone Zone Zone Zone Zone Zone III Zone Zone
II III II III II III II III II II III
30 2,72,47 2,77,8 2,57,0 2,59,81
,248 7,008 4,960 ,920 21600 21600 21600 21600 1261.44 1286.43 1190.04 1202.86
15 1,16,04 1,18,7 1,13,1 1,13,47
,000 3,616 1,680 ,680 10800 10800 10800 10800 1074.44 1099.40 1047.37 1050.71
5 36,27,8 35,91, 36,41, 36,15,8
40 840 76 40 3600 3600 3600 10800 1007.73 997.73 1011.6 1004.4
6. SUMMARY AND CONCLUSIONS:
The investigation presents the development of a new method for analyzing shear wall framing system and a new
streamlined model to quantitatively compare the cost-effectiveness of a lateral-load-resisting system for high
rise concrete building. Multi storey buildings with moment-resisting and shear wall framing systems are
investigated. In addition to their vertical dead and live loads, these building are designed against lateral forces
generated for seismic zones II & III . Cost-efficiency of the system is evaluated based on lateral load resistance
and cost of steel. Roof drift is limited to 0.004 x hi, where hi is the storey height of the ith floor. Cost of steel
reinforcement includes cost of column, beam, slab, shear wall reinforcement for comparing concrete framing
systems.
Based on the analytical study carried out for 3 types of structures, the following conclusions can be drawn:
1. The method of analyzing the combined system of frame plus shear wall using 2D panel size, two node
elements (Idealized as single column of equivalent stiffness) for the shear wall is less time consuming and hence
suitable for designer.
354
2. Analysis of shear walls idealized as a 4-node element is extremely time consuming & hence not suitable for
designer.
3. From the point of view of steel consumption/unit area, introduction of shear wall results in more economical
system for buildings having more than 15 storey’s of earthquake zone II & III.
4. From the point of view of drift as well as steel consumption shear wall systems are advantageous for
buildings taller than 15 stories in earthquake zones II & III.
5. From the Analysis results of MRFS & SWFS, Moment Resisting Framing system is cost effective in terms of
steel up to 5 storey’s both in Zone II & Zone III.
6. From the Analysis results of MRFS & SWFS , Shear wall Framing system is cost effective in terms of steel
from 15 storey’s and above for Zone II & III.
7. In both the systems (MRFS & SWFS), the storey drift is within the permissible limit. But when compared
with MRFS, one can effectively control the drift (i.e the drift in SWFS is only 50% of the drift in MRFS).
REFERENCES
[1] Build Science“The Structural Analysis of Tall Buildings having irregularly positioned shear wall”
(1973) volume 18, pp. 11-22 pergamon press.
[2] E. Miranda, M.Asce1; and S. D. Akkar “Generalized Interstory Drift Spectrum” (2006) ASCE P
No: 733-9445.
[3] Gary R. Searer and Sigmund A. Freeman “Design drift requirements for long-period structures”
13th World Conference on Earthquake Engineering, (2004), Paper No. 3292.
[4] Hyo seon park, Chil lim park (1997), “Drift control of high-rise buildings with unit load method” The
Structural Design of Tall Buildings Vol 6, No 1, pp 23–35, March 1997
[5] Hyo seon park, Ji hyun seo, Yun han kwon (2008). “Development of drift design model for high-
rise buildings subjected to lateral and vertical loads”, The structural design of tall and special
buildings, Vol 17, No 2, pp. 273–293.
[6] Kofi g. ayensu and Elin Jensen (2007) “Model Development for the Evaluation of Cost-
Effectiveness of Concrete Framing Systems for Low-, Mid- and High-Rise Buildings”.
[7] Manabu yoshimura, “Control of Seismic Drift Demand for Reinforced Concrete Buildings with
Weak First Stories” Department of Architecture, Tokyo Metropolitan University, Minamiosawa
1-1, Hachioji, Tokyo, 192-03, Japan.
[8] Mohamed a. Azab, “Structural Sustainability Techniques for RC High Rise Buildings” (2010)
World Academy of Science, Engineering and Technology.
[9] S.R.Uma and A.Meher Prasad (2006),” Seismic behaviour of beam -column joints in R.C Moment resisting
frames” A review. ICJ. Vol. 86. -, No.1.
355
Paper No.: 061
PROBABILISTIC ASSESSMENT OF SAFETY, ECONOMY AND

ROBUSTNESS OF STEEL STRUCTURE INCLUDING PARAMETER
UNCERTAINTY IN IS: 800-2007 FORMAT
Soumya Bhattacharjya 1#, Urmi Saha2, Asish Modak2 and Baidyanath Sarker2
1
Assistant Professor, Department of Civil Engineering, Bengal Engineering and Science University, Shibpur
# Corresponding author: soumya@civil.becs.ac.in
2. Masters student, Department of Civil Engineering, Bengal Engineering and Science University, Shibpur
ABSTRACT:
A drastic change in limit states has been observed after the revisions of Indian Standard Code of Practice for
General Construction in steel (IS: 800-2007). Thereafter, there have been numerous studies exploring the
economy and rationality of this new version of code over the previous version (IS: 800-1984). However, most of
these studies are in deterministic format. The effect of uncertainty has been incorporated indirectly by various
partial safety factors. In this study, an attempt has been made to assess economy and safety of design of steel
structures after direct incorporation of uncertainty in system through Monte Carlo Simulation instead of using
code specified partial safety factors. Consequently, Reliability Based Design Optimization (RBDO) and Robust
Design Optimization (RDO) have been carried out using probabilistic measures incorporating uncertainty in the
involved parameters in various limit states of IS:800-2007. The results of Probabilistic design using Statistical
Simulation, RBDO and RDO are compared with the codal partial safety factor based deterministic approach.
Design optimization of tension member, compression member, plate girder and conveyer gallery structure has
been taken up for this study to encompass a broad domain of structure. It has been observed that even after
incorporation of reasonable and high values of coefficient of variations for different parameter uncertainty the
Probabilistic approach using Statistical Simulation, RBO and RBDO produce more economic, safe and robust
design in many cases in comparison to the conventional partial safety factor based codal deterministic approach.
.
KEYWORDS: Steel Structure, IS:800-2007, Reliability Based Design Optimization, Robust Design
Optimization, Statistical Simulation, Parameter Uncertainty
1. INTRODUCTION
A drastic change in limit states has been observed after the revisions of Indian Standard Code of Practice for
General Construction in steel (IS: 800-2007). Thereafter, there have been numerous studies exploring the
economy and rationality of this new version of code over the previous version (IS: 800-1984). However, most of
these studies are in deterministic format. The effect of uncertainty has been incorporated indirectly by various
partial safety factors. In this study, an attempt has been made to assess economy and safety of design of steel
structures after direct incorporation of uncertainty in system through Monte Carlo Simulation (MCS) instead of
using code specified partial safety factors. Consequently, two scheme of optimization under uncertainty, viz
Reliability Based Design Optimization (RBDO) and Robust Design Optimization (RDO) have been employed
to cross-check the deterministically obtained optimum design as per IS 800:2007. A comparison according to
Load and Resistance Factor Design (LRFD) method as per American Institute of Steel Construction (AISC) has
been also performed. The scheme of RBDO and RDO are discussed first in sections 2 and 3, followed by
Numerical study in section 4 and conclusion in section 5.
2. THE RBDO SCHEME
The RBDO is formulated to minimize the expected value of objective function (which is a function of the
design variables X) subjected to satisfying reliability index for various limit states to be more than a target value
of reliability index, β T . Assuming the uncertain Design Variables (DVs) (the specific parameters designer
needs to optimize to achieve the desired performance) and Design Parameters (DPs) (which cannot be
controlled by the designer or are difficult and expensive to control) as normal random and uncorrelated the
problem can be mathematically expressed as (Halder and Mahadevan, 2000; Doltsinis and Kang, 2005)
minimize: f = µ f ( X ,Z )
i i
subjected to: β g −1  1 − p  T
j ( Xi , Z i )
=Φ  f, g ( X , Z  ≥ β j = 1, 2,......, J
 j i i )  (1)
1
N nf  0, if g j ( X, Z) > 0 
where, p f = ∑ I g ( Xi , Z i ) = and I g ( Xi ) =  
N
i =1
N 1, if g j ( X, Z) < 0 
xi ≤ xi ≤ xU
L
i , i = 1, 2......, K .
In the above, β g is the reliability index for jth constraint function g j ( X) , p f is the failure probability for
j ,g j
g j ( X) <0, I g ( Xi ) is the expected value of the indicator function for MCS, N is the number of simulation,
n f is the number of failures, µ f is the expected value of objective function, f by first-order Taylor series
expansion, Φ-1(.) is the inverse of the cumulative density function of a standard normal distribution, X and Z
are the vectors of uncertain DVs and uncertain DPs, respectively and, xiL and xUi are the lower and the upper
bounds of the ith DV, respectively. Default value of β T is taken as 3.0 corresponding to p f of 0.135%. The
optimization problem thus formulated is solved by Sequential Quadratic programming by Matlab.
3. THE RDO SCHEME
The undesirable deviation of system performance due to the presence of uncertainty in the input parameters
indicates a poor quality and added life-cycle cost of structure, including inspection, repair and other
maintenance costs in the perspective of entire design life of the structure. To decrease such deviation, one
possible way is to reduce or even to eliminate the source of uncertainty of the input parameters, which may
either be practically impossible or adds much to the total cost of the structure. Another way is to find a design in
which the structural performance is less sensitive to the variation of parameters without eliminating the cause of
parameter variations. This later one is the task of RDO of structure. The RBDO will bring specified target
reliability of structure performance, but the design may be still sensitive to input parameter variations. Moreover
when the probability density function is unavailable and the uncertain parameter is modeled like uncertain but
bounded type RDO becomes an attractive alternate to RBDO.
The RDO is fundamentally concerned with minimizing the effect of uncertainty in the DVs and the DPs. The
definition of RDO may be described as, “A product or process is said to be robust when it is insensitive to the
effects of sources of variability, even though sources themselves have not been eliminated”. In recent years,
substantial progress has been attained in the field of RDO in the international perspective (Beyer and Sendhoff,
2007). Mathematically, an RDO problem is stated as,
357
µf σf
minimize: φ (u ) = α + (1 − α ) , 0 ≤α ≤1
µf* σ f*
subjected to: µ g + k jσ g ≤ 0 j = 1, 2,......, J (2)
j j
xiL ≤ xi ≤ xU
i , i = 1, 2......, K .
In above, µ f and σ f are the mean and the standard deviation of the performance function, respectively;
φ (u ) is a new objective function, called desirability function; the parameter α serves as a weighting factor;
µ f * and σ f * are the optimal values of the mean and the standard deviation obtained for α equals to 1.0 and
0.0, respectively. Larger value of α implies that the designer puts more significance on the conventional
objective over its variance. In Eq. (2), µ g and σ g are the mean and the standard deviation of the jth
j j
constraint g j , respectively. The designer specified penalty factor, k j is used to enhance the feasibility of the
jth constraint and can be obtained from, k j = Φ-1(Poj ) , where Φ-1(.) is the inverse of the cumulative density
function of a standard normal distribution and Poj =1- p f . Once, the RDO problem is formulated the solution
is done by Sequential Quadratic Programming by Matlab.
4. NUMERICAL STUDY
The comparison is done with the following optimization approaches:

1) The Deterministic Optimization (DO) by IS:800-2007 Limit State Method (LSM) using code-specified safety
factors without incorporation of uncertainty directly,
2) The DO by IS:800-1984 Working Stress Method (WSM) using code-specified safety factors without
incorporation of uncertainty directly,
3) the RBDO incorporating detailed uncertainty directly in the DVs and DPs by MCS instead of using partial
safety factors,
4) the RDO incorporating detailed uncertainty directly in the DVs and DPs by MCS instead of using partial
safety factors,
5) the LRFD approach using code-specified safety factors without incorporation of uncertainty directly in the
optimization process.
The numerical study comprises of optimal design of tension-compression members, plate girders and a conveyer
gallery structure under wind and seismic loading. For the RBDO and RDO study the uncertainty in the DVs (i.e.
the cross-sectional dimensions) is taken with 5% coefficient of variation (cov) and is considered to be normally
distributed unless otherwise specified. The uncertainty in the DPs like yield strength, Youngs modulus, effective
length factors, shear lag factors and other code-specified factors are taken with 10% cov. The uncertainty in the
load related DPs like IRC live load, seismic loading, etc. have been considered to be with 20% cov unless
specified. The default values of β T and α are considered as 3.0 (corresponding to p f of 0.135%) and 0.5,
respectively. In the optimization study, the weight of the structure is considered as the objective function. The
RBDO and the RDO study have been performed with reliability constraints imposed on the limit state
conditions of IS:800-2007. The comparison of optimal designs for tension-compression members are presented
in Fig. 1. It may be observed from Fig. 1 that hollow tube produces the least optimal weight comparison to
double angle and square hollow sections. In all the cases, the RBDO and the RDO approach produce even 20%
lesser optimal weight than the conventional deterministic LSM or WSM approaches even when the load
uncertainty is elevated at 50% level with such a high target reliability index of 3.0. The LRFD predictions are
closer to the RBDO or RDO predictions since the LRFD provisions consider the application of reliability theory
in larger extent than the Indian codes. The optimal weight by the RDO is more than that by the RBDO because
358
RDO ensures lesser deviation in the performance in addition of satisfying target reliability levels of constraint
feasibility. The comparison of optimal designs is further presented in Fig. 2(a) for design of a plate girder
considering IRC 70R live load. The same observations of Fig. 1 are pertinent here, as well. In Fig. 2(b) the
variation of optimal weight by the RDO approach for different settings of α is presented. As can be observed,
smaller α values reckoning more emphasis on minimizing the standard deviation of the performance function
rather than its mean value yield more optimal weight. Fig. 3 presents the variation of optimal weight by the
RBDO approach for varying levels of load uncertainty. The optimal weight by the RBDO approach with 80%
uncertainty in load is lesser than the conventional IS:800:2007 based deterministic approach. This reckons that
the IS:800:2007 design is too conservative in the perspective of uncertainty.
Figure 1: Comparison of optimal designs by WSM, LSM, LRFD, RBDO and RDO approaches for
tension-compression members
(a) (b)
Figure 2: (a) Comparison of optimal designs by WSM, LSM, RBDO and RDO approaches for plate girder
considering IRC 70R loading, and (b) the variation of optimal weight for different settings of weight factor
359
Figure 3: Comparison of the optimal designs by the RBDO and the deterministic LSM design for varying
uncertainty level in the live load on the plate girder
In Fig. 4, the detail of the conveyer gallery structure has been presented. The structure consists of bridges
supported on trestles at 30 m spacing with a transfer tower leading to the melting shop. The vertical trusses 4.8
m apart constitute the conveyor bridge. The vertical trusses are connected at the top and bottom chord levels by
horizontal wind trusses. The bridge has been provided with roof and side coverings. The length of the conveyor
bridge is 150 m. A 25 m length of the conveyor bridge connects the transfer tower to the melting shop at right
angle to the main conveyor bridge. The conveyor structural system comprises of a series of conveyor bridges
simply supported on the trestles and the transfer tower when the bridge changes the direction. The members of
structures are considered hinged at the connections. The trestles and transfer tower are assumed fixed at the
bases. The structure has been analyzed for dead load, live load, wind load and seismic load. The gust component
of wind has been considered in the wind load calculation (IS:875(III),1987). Time history analysis with
El-centro earthquake has been considered as seismic loading as a test case. The DO problem has been
formulated to minimize weight of the structure subjected to the limit states of collapse and serviceability as
defined by IS: 800-2007 considering partial safety factors. The sectional dimensions of the different members
are considered to be DVs. All the involved parameters in design computations and loading have been considered
as uncertain. Thereafter, the RBDO problem is formulated using Eq. (1). The uncertainty in the DVs and DPs
has been considered by MCS directly, instead of using partial safety factors.
A typical result is shown in Fig 5 which compares optimal area of a conveyer trestle by the RBDO approach and
the conventional DO approach, both following stipulations of IS:800-2007. It can be observed that the RBDO
approach consistently produces substantial less sectional area compared to the conventional DO approach. Fig.
6 compares the RBDO results with the DO result for reliability index 3.0. It can be observed that the DO results
correspond to an optimal cross-sectional area of 25076.52 mm2. Whereas, for the RBDO with β T =3.0 and
50% level of uncertainty in seismic load yields an optimal weight of 20235.06 mm2 which is 24% higher than
the conventional partial safety factor based DO approach. Thus, there is enough scope to improve the design
factors, stipulation, partial safety factors of the present latest IS code, which seems to be quite conservative as is
evident from the present RBDO approach. However, this requires further extensive study. The LRFD approach
may be more prudent in this regard.
360
Figure 4: Details of Conveyor Bridge
.
Figure 5: Optimal area of trestle by RBDO and DO
361
30000.00
25076.52 25076.52 25076.52 25076.52 25076.52

25000.00
20235.06 DETERMINISTIC
DESIGN AREA
20000.00
)2 17416.64
m 16099.25
(m 14402.39 14484.59
15000.00 13039.66
A RELIABILITY BASED
E 11476.87
R 10259.97 OPTIMIZED DESIGN
A 9209.06 9583.49 AREA:SEISMIC LOAD
10000.00
5000.00 RELIABILITY BASED

OPTIMIZED DESIGN
AREA:WIND LOAD
0.00
10% 20% 30% 40% 50%
LOAD UNCERTAINTY
Figure 6: Comparison of optimal area of the cross-section by the DO and RBDO approaches for wind and seismic
load for varying levels of load uncertainty and β T =3.0
5. CONCLUSIONS
The optimal probabilistic designs by RBDO and RDO using MCS framework are compared with the codal
partial safety factor based deterministic approach. Design optimizations of tension-compression member, plate
girder and conveyer gallery structure under wind and seismic load have been taken up for this study to
encompass a broad domain of structure. It has been observed that even after incorporation of reasonable and
high values of cov for different parameter uncertainty, the RBO and the RBDO produce more economic, safe
and robust design in many cases in comparison to the conventional partial safety factor based codal
deterministic approach. The RDO yields higher optimal weights than the RBDO since it emphasize on lesser
deviation of performance function in addition of satisfying the user-specified target reliability level. The RBDO
and the RDO results are close to the LRFD results. Thus, the author feel that there is enough scope to improve
the design factors, stipulation, partial safety factors of the present latest IS code, which seems to be quite
conservative as is evident from the present RBDO, RDO and LRFD results. Further extensive research focusing
the RBDO study on other steel structures, connections and the LRFD approach may find more useful future direction
of safe and economic design for the coming era
REFERENCES
[1] Beyer, H. and Sendhoff, B. (2007). Robust optimization- a comprehensive survey. Comput. Methods Appl.
Mech. Engng, 196 (33-34), 3190-3218.
[2] Doltsinis,I, Kang, Z. and Cheng, G. (2005). Robust design of non-linear structures using optimization
methods. Comp. Meth. Appl. Mech. Engng., 194, 1779-1795,
[3] Halder, A.and Mahadevan, S. (2000) Reliability Assessment Using Stochastic Finite Element Analysis, Wiley,
NY,
[4] IS:800:1984, Code of practice for general construction in steel, BIS, India.
[5] IS:800-2007. General construction in steel. Code of Practice, Bureau of Indian Standards, New Delhi.
[6] IS:875 (Part 3) – 1987. Code of practice for design loads (other than earthquakes) for buildings and
structures. Part 3 Wind Loads, Bureau of Indian Standards. New Delhi.
362
Paper No.: 062
A COMPUTER AIDED FEM BASED NUMERICAL SOLUTION FOR

TRANSIENT RESPONSE OF LAMINATED COMPOSITE PLATES WITH
CUTOUTS
AK Nayak1 , RA Shenoi2 , JIR Blake3

1
Author for Correspondence, Reader, Civil Engineering Department, Veer Surendra Sai University of
Technology, Burla, ajaya_nayak_07@yahoo.co.uk, Ph No: 09777460423
2
Professor, r.a.shenoi@ship.soton.ac.uk,
3
Senior Lecturer, james@ship.soton.ac.uk
1,2,3
University of Southampton, Highfield, SO17 1BJ, Southampton, UK.
ABSTRACT :
Cutouts are inevitable in many practical locations in the form of access ports for electro-mechanical systems,
damage inspection and altering the resonant frequency of the structures. A new assumed strain four-node higher
order plate bending element with three translations, two rotations of the normals about the plate midplane, two
warps of the normals and one drilling rotational degree of freedom per node is proposed to carry out the
transient response of composite plates with cutouts. Consistent mass matrix is employed to evaluate the kinetic
energy of the system. Newmark integration scheme is used to integrate the governing equations of motion.
Numerical studies from the present computer aided finite element method based solution for transient response
of composite plates with cutouts could serve benchmark for future investigators.
KEYWORDS: Transient analysis, Higher-order shear deformation theory, plate finite element, Assumed
strain concept, Cutouts.
1. INTRODUCTION
Composite plates with cutouts are extensively used in many strategic applications. Reddy [1] was the first to
carry out investigation of the large amplitude flexural vibration of undamped rectangular plates with rectangular
cutouts using a nine node finite element model based on a Reissner-Mindlin type of shear deformable theory.
Later on Bicos and Springer [2] developed a four node bilinear quadrilateral element based on widely used
Reddy’s higher order theory to obtain numerical results for problems involving damped plates, cylindrical shells
and cylindrical panels with free, clamped, or simply supported edges and with or without circular cutouts. Since
the analytical or experimental results are necessary to validate the numerical computations like the finite
element method, Lee et al [3] proposed an analytical technique based on the Rayleigh Principle along with
Reissner-Mindlin theory to predict the fundamental frequencies of rectangular plates with rectangular cutouts.
Later on Xiang et al [4] used the Rayleigh-Ritz procedure in the energy functional, which includes the effect of
shear deformation and rotatory inertia to derive the governing eigenvalue equation. To investigate the effect of
internal (not a through-thickness) cutouts on the vibration characteristics of fifteen layer cross-ply laminated
cantilever plates, Jenq et al [5] conducted a series of laboratory tests to examine the vibration mode shapes and
natural frequencies by means of holographic interferometry and piezo-electric sensing techniques. Boay [6]
performed experimental investigation on symmetric laminated composite plates (45/02/90/0/−45/0)s containing
a central hole for the plate having clamped, simply supported and free edge conditions to supplement the
theoretical analysis and to provide confidence in the theoretical results. Kong and Cheung [7] proposed a
modified version of the original spline finite strip to incorporate a third-order plate theory to study the effects of
cutouts on static and dynamic response of laminated composite plates. Liew and Feng [8] presented a
three-dimensional free vibration analysis of a perforated plate with rounded corners. A solution method based
on the linear, small strain, three-dimensional elasticity theory and the p-Ritz algorithm is employed. From a
literature review, it is found that there is a scarce of literature on transient response of composite plates with
cutouts to the best of authors knowledge. The formulations [1−8] on traditional plate finite elements without
vortex rotations are satisfactorily accurate. However the membrane response of the four node plate bending
element could be improved by the addition of drilling degrees of freedom as seen in the works of Piancastelli [9]
and Cook [10]. In the present paper, a new four node higher order plate bending element is developed by
incorporating drilling rotations suggested by Cook [10] based on a refined third-order theory. It should be
mentioned that consistent mass matrix is calculated taking into account the shape functions associated with the
vortex rotations. Analytical solutions on higher order shear deformation theory (HSDT), first order shear
deformation theory (FSDT) and classical plate theory (CPT) are implemented for validation studies for plates
without cutouts. The present computer aided finite element method (FEM) based analysis is general and could
be applied to solve wide range of problems dealing with transient analysis of composite plates with cutouts.
2. FINITE ELEMENT FORMULATION
The present finite element displacement field of a new four node plate element based on a higher order shear
deformation theory is stated as
4  4z3 
u1 = ∑  uoi Ni + zψ xi Ni − 2 φxi Ni  + ( y14 N8 − y21N5 )θ z1 + ( y 21N5 − y32 N6 )θ z 2 + ( y32 N6 − y43N7 )θ z 3 + ( y43 N7 − y14 N8 )θ z 4
i =1  3h 
4  3 
u2 = ∑  voi Ni + zψ yi Ni − 2 φ yi Ni  + (x41N8 − x12 N5 )θ z1 + (x12 N5 − x32 N6 )θ z 2 + (x23 N6 − x34 N7 )θ z 3 + (x34 N7 − x41N8 )θ z 4
4z
i =1  3h 
u 3 = wo (1)
Where u1 , u 2 and u 3 are the displacement components in the x, y and z directions respectively, of a generic
point in the laminate space. u o , v o and wo are the in-plane and transverse displacements of a point (x,y) on
the midplane respectively. ψ x and ψ y are the rotations of normal to the mid-plane about y and x axes
respectively. The parameters φ x and φ y are warping functions. θ zi (i=1,4) are the drilling rotations, N i
(i=1,4) are the shape functions, x ij = x i − x j and y ij = y i − y j are corner coordinate differences, and the
shape functions N i (i=5,8) associated with drilling rotations [16] are
N5 =
1
16
( 1
16
)
1 − ξ 2 (1 − η ) N6 = (1 + ξ ) 1 − η 2 N7 =
1
16
( )
1 − ξ 2 (1 + η ) ( ) N8 =
1
16
(
(1 − ξ ) 1 − η 2 )
(2)
Where ξ and η are the usual iso-parametric coordinates.
The strains associated with the displacements in Equation (1) are
ε 1 = ε 1o + zκ 1o + z 3κ 12 , ε 2 = ε 2o + zκ 2o + z 3κ 22 , ε 3 = 0 , ε 4 = ε 4o + z 2κ 42 , ε 5 = ε 5o + z 2κ 52 , ε 6 = ε 6o + zκ 6o + z 3κ 62 (3)
( )
T
T  ∂u ∂v ∂u ∂v 
Where ε = o
ε 1o , ε 2o , ε 6o =  do , do , do + do 
 ∂x ∂y ∂y ∂x 
u do = ∑ (u oi N i ) + ( y14 N 8 − y 21 N 5 )θ z1 + ( y 21 N 5 − y 32 N 6 )θ z 2 + ( y 32 N 6 − y 43 N 7 )θ z 3 + ( y43N 7 − y14 N8 )θ z 4
4
i =1
v do = ∑ (voi N i ) + (x 41 N 8 − x12 N 5 )θ z1 + (x12 N 5 − x 23 N 6 )θ z 2 + (x 23 N 6 − x34 N 7 )θ z3 + (x 34 N 7 − x 41 N 8 )θ z 4

4
i =1
T
4 ∂φ y  ∂φ x ∂φ y  
T
 ∂ψ x ∂ψ y ∂ψ x ∂ψ y
κ =
o
(κ 1o , κ 2o , κ 6o
T
) =  , , +


 κ =
2
( )
κ 12 , κ 22 , κ 62
T  4 ∂φ x
= − 2
 3h ∂x
,− 2
4
,− 2  
 ∂y + ∂x  
 ∂x ∂y ∂y ∂x   3h ∂y 3h  
( ) ( )
T T
T  ∂w ∂w  T  4 4 
ε s = ε 5o , ε 4o =  o + ψ x , o + ψ y  κ s = κ 52 , κ 42 =  − 2 φ x ,− 2 φ y 
 ∂x ∂y   h h 
The generalized mid-surface strains at any point given by Equation (3) can be expressed in terms of nodal
displacements {δ }(e ) as follows:
364
[ ]( ){δ }( ) , κ ( ) = [B ]( ){δ }( ) κ ( ) = [B ]( ) {δ }( ) ε ( ) = [B ]( ){δ }( ) κ ( ) = [B ]( ) {δ }( )
ε o(e ) = Bεo
e
e
oe o e
κ e
2e 2 e
κ e
se s e
ε e
se s
κ
e
e (4)
Where [B ] , [B ] , [B ] , [B ] and [B ] are generated strain-displacement matrices. One basic
ε
o o
κ
2
κ
s
ε κ
s
problem
inherent in the use of standard interpolation of the strains for the transverse shear components is that the element
locks when it is thin. The reason for this locking is that the element, when loaded in pure bending, will exhibit
spurious transverse shear energy. In order to overcome the shear locking, Dvorkin and Bathe [11] proposed
assumed interpolations for the shear strain to develop a four node assumed strain degenerated shell element. In
the present paper same shape functions are chosen to develop our new four node flat shell element with vortex
rotations.
The substitute shear strain fields are chosen as follows:
ε ξςs = ∑ ∑ Pi (ξ )Q j (η )ε ξςsij
1 2
(5)
i =1 j =1
ε ηςs = ∑ ∑ Qi (ξ )Pj (η )ε µς
21 1
sji
(6)
i =1 j =1
Where Q1 (z ) = (1 + z ) / 2 , Q2 (z ) = (1 − z ) / 2 and P1 (z ) = 1(z = ξ , η ) in which ε ξςsij and ε ηςsij are the m× n

unknown substitute shear strain parameters associated with two sets of m× n sampling points ξî ,ηˆ j ( ) and
(
( (
ξ j ,η i . )
In order to eliminate locking, the following equations are obtained:
( ) ( )
εξςs ξî ,ηˆ j = εξςs ξî ,ηˆ j (i = 1,....,m; j = 1,.....,n) (7)
(( (
) ( (
( )
ε ηςs ξ j ,η i = ε ηςs ξ j ,η i (i = 1,...., n; j = 1,....., m) (8)
It is possible to write
ε o 
ε s =  5o  (9)
ε 4 
Where ε 5o and ε 4o are obtained from ε ξςs and ε ηςs given by Equations (5-6) by tensor transformation.
The transformation of the strain tensor in curvilinear coordinates may be written as
δx i δx j
eαβ = ε ij (10)
δε α δε β
where it is assumed that eαβ is the strain tensor in the (ξ , η ) coordinate system and ε ij is the strain tensor in
the (x, y ) system. For further details, see Dvorkin and Bathe [11]. For implementation purpose, ε s in
Equation (4) is replaced by ε s where ε s is the substitute shear strains to remove spurious zero energy modes.
Hence the substitute shear strain ε s is given by
[ ]( ){δ }( )
ε s (e ) = Bεs
e
e (11)
Where [B ] is generated strain displacement matrix.
() s e
ε
For arbitrary values of virtual displacements, the following assembled equation for forced vibration
analysis is stated as:
[M ]{∆&&}+ [K ]{∆} = {F } (12)
Here the unknown vector {∆} is generated by the assemblage of element degrees of freedom {d }Te ,
e=1,….,total degrees of freedom in the region R. The assembled stiffness, mass and buckling matrices for
transient analysis are
BoT ABo + BoT BBo + BoT EB2 + BoT BBo + BoT DBo +
 ε ε ε κ ε κ κ ε k κ 
[K ] = ∑ ∫ BκoT FBκ2 + Bκ2T EBεo + Bκ2T FBκo + Bκ2T HBκ2 + dA

e Ae  
B sT As B s + B sT Ds Bs + BsT Ds B s + BsT F s Bs  (13)
 ε ε ε κ κ ε κ κ 
365
Where Aij , Bij , etc are the plate stiffness, defined by
(Aij , Bij , Dij ) = (

∫ Qij 1, z, z , z , z , z dz(i, j = 1,2,6) )
h/2
2 3 4 6
−h / 2
(A , D , F ) = (
∫ Qij 1, z , z dz(i, j = 5,4) )
h/2
s s s 2 4
ij ij ij (14)
−h / 2
Qij are the transformed plane stress reduced elastic stiffness coefficients, which are given as,
Q11 = Q11c 4 + 2(Q12 + 2Q66 )c 2 s 2 + Q22 s 4 Q12 = (Q11 + Q22 − 4Q66 )c 2 s 2 + Q12 c 4 + s 4 ( )
( 2
( ) ) Q = Q s + 2(Q + 2Q )c s + Q c
Q16 = Q11c + (Q12 + 2Q66 ) s − c − Q22 s cs 2 2 2
22 11
4
12 66
2 2
22
4
Q = (Q s + (Q + 2Q )(c − s ) − Q c )cs Q = (Q + Q − 2Q )c s + Q (c − s )
26 11
2
12 66
2 2
22
2
66 11 22 12
2 2
66
2 2 2
Q = (c Q + s Q ) Q = (c Q + s Q )
55
2
55
2
44 Q = cs(Q − Q ) Q = cs(Q − Q )
44
2
44
2
55 (15) 45 44 45 54 55 44
Where [Q ] is constitutive matrix at lamina level; c = cos θ ; s = sin θ ; θ is the angle between the lamina x
ij k
axis and lamina principal xi axis. Where Q11 = E1 / (1 −ν 12ν 21 ) , Q12 = ν 12 E 2 / (1 −ν 12ν 21 ) ,
Q 22 = E 2 / (1 −ν 12ν 21 ) , Q66 = G12 , Q55 = G13 , Q44 = G23 , E i (i = 1,2) are the Young’s moduli, G12 , G13 and
G23 are the shear moduli. ν 12 and ν 21 are the Poisson’s ratios. The consistent mass matrix [M ] in Equation
(12) can be obtained from the kinetic energy
 
 ∫ ρ (u&&1δu1 + u&&2 δu 2 + u&&3δu 3 )dV  (16)
V 
of the system where ρ (x, y, z ) and V are the density of the plate at (x, y, z ) and volume of the plate
respectively.
A 2 x2 Gauss-Legendre rule (i.e full integration scheme) is employed to integrate bending,
membrane, shear and inertia terms in the energy expressions for the new four node drilling degrees of freedom
plate element. The resulting problem in Equation (6), is solved the Newmark direct integration scheme [12].
3. ANALYTICAL FORMULATION ON HIGHER ORDER SHEAR DEFORMATION THEORY
In this section, a brief description of HSDT [13] is given. For symmetric and anti-symmetric cross-ply laminates,
the following boundary conditions for simply supported edges are assumed to solve the governing equations of
motion:
u o = wo = ψ x = φ x = N yy = M yy = Pyy = 0 … ……….. (17)
On the edge y=constant and
vo = wo = ψ y = φ y = N xx = M xx = Pxx = 0 (18)
On the edge x=constant
The exact form of the spatial variation of mid-surface displacements is given by
∞ ∞ ∞ ∞
u o = ∑ u omn cos αx sin βy vo = ∑ vomn sin αx cos βy wo = ∑ womn sin αx sin βy ψ x = ∑ψ xmn cos αx sin βy
m, n m, n m, n m, n
∞ ∞ ∞ ∞
ψ y = ∑ψ ymn sin αx cos βy φ x = ∑ φ xmn cos αx sin βy φ y = ∑ φ ymn sin αx cos βy q = ∑ Qmn sin αx sin βy
m, n m, n m, n m, n (19)
The Equations of motion of HSDT [13] can be presented as
∂N xx ∂N xy ∂ 2uo ∂ 2ψ x  − 4 I 4  ∂ 2φ x ∂N xy ∂N yy ∂ 2 vo ∂ 2ψ y  − 4 I 4  ∂ φy
2
+ = I1 + I2 +  2 + = I1 + I2 +  2
∂x ∂y ∂t 2 ∂t 2  3h  ∂t
2
∂x ∂y ∂t 2 ∂t 2  3h
2
 ∂t
∂Q xz ∂Q yz ∂ 2 wo ∂M xx ∂M xy ∂ 2uo ∂ 2ψ x  − 4 I 5  ∂ 2φ x
+ + q = I1 + − Q xz = I 2 + I3 +   2
2 
∂x ∂y ∂t 2 ∂x ∂y ∂t 2 ∂t 2  3h  ∂t
∂M xy ∂M yy ∂ 2 vo ∂ 2ψ y  − 4I 5  ∂ φy
2
+ − Q yz = I 2 + I3 +   2
∂x ∂y ∂t 2 ∂t 2  3h
2
 ∂t
366
 − 4  ∂Pxx  − 4  ∂Pxy  − 4   − 4 I 4  ∂ u o  − 4 I 5  ∂ ψ x  16 I 7  ∂ φ x
2 2 2
 2 + 2  +  2  R xz =   + 
 2 
 + 
 4  2 
 3h  ∂x  3h  ∂y h   3h  ∂t
2 2
 3h  ∂t
2
 9h  ∂t
 − 4  ∂Pxy  − 4  ∂Pyy  − 4   − 4 I 4  ∂ v o  − 4 I 5  ∂ ψ y  16 I 7  ∂ φ y
2 2 2
 2 + 2  +  2  R yz =   + 
 2 
 + 
 4  
 3h  ∂x  3h  ∂y h   3h  ∂t
2 2
 3h  ∂t
2
 9h  ∂t
2
(20)
(
Where P = Pxx , Pyy , Pxy ) T
and (
R = Rxz , R yz ) T
are the stress resultants [13].
The elements of the mass matrix [M] can be given as
 − 4I 4 
 I1 0 0 I2 0 2
0 
 3h 
 0 − 4I 4 
I1 0 0 I2 0
 3h 2 
 0 0 I1 0 0 0 0 
 − 4I 5 
 I2 
[M I ] =  0 0 I3 0
3h 2
0

 − 4I 5  (21)
 0 I2 0 0 I3 0 
 3h 2 
 − 4I 4 0 0
− 4I 5
0
16 I 7
0 
 3h 2 3h 2
9h 4 
 − 4I 4 − 4I 5 16 I 7 
 0 0 0 0 
 3h 2 3h 2 9h 4 
The elements of the stiffness matrix [K] are given by
K 11 = A11α 2 + A66 β 2 K 12 = ( A12 + A66 )αβ K13 = 0 K 44 = B11α 2 + B66 β 2 K 15 = (B12 + B 66 )αβ
 −4
(
K 16 =  2  E11α 2 + E 66 β 2 )  −4
K 17 =  2 (E12 + E 66 )αβ K 22 = A22 β 2 + A66α 2 K 23 = 0
 3h   3h 
 −4 
K 24 = (B12 + B66 )αβ K 25 = B22 β 2 + B66α 2 K 26 =  2 (E12 + E 66 )αβ
 3h 
 −4
( )
K 27 =  2  E 22 β 2 + E 66α 2 K 33 = A55α 2 + A44 β 2 K 34 = α [A55 ] K 35 = β [A44 ] K 44 = D11α 2 + D66 β 2 + A55
 3h 
K 45 = (D12 + D66 )αβ

 −4 −4
K 46 =  2  F11α 2 + F66 β 2 +  2  D55 ( )  −4
K 47 =  2 (F12 + F66 )αβ
 3h  h   3h 
K 55 = D66α 2 + D 22 β 2 + A44
 −4
K 56 =  2 (F12 + F66 )αβ
 −4 − 4
K 57 =  2  F66α 2 + F22 β 2 +  2  D 44 ( )
 3h   3h  h 
( )
2 2
 −4 −4
K 66 =  2  H 11α 2 + H 66 β 2 +  2  F55
 3h  h 
( )
2 2 2
 −4  −4 − 4
K 67 =  2  (H 12 + H 66 )αβ K 77 =  2  H 66α 2 + H 22 β 2 +  2  F44 (22)
 3h   3h  h 
The inertias I i (i = 1,2,3,4,5,7 ) are defined by
N zk
(I 1 , I 2 , I 3 , I 4 , I 5 , I 7 ) = ∑
k =1z k −1
(
∫ ρ k 1, z , z , z , z , z dz
2 3 4 6
) (23)
4. NUMERICAL RESULTS AND DISCUSSIONS
In this section numerical results are presented to assess the behavior of a family of plate bending elements
(HSDT4, FSDTC4, FSDTV4 and ECPT4) based on a unified refined higher order shear deformation theory.
HSDT4, FSDTC4, FSDTV4 and ECPT4 denote 4 node plate bending elements based on a HSDT, first order
shear deformation theory with shear correction factor of 5/6 (FSDTC) [14,15], first order shear deformation
367
theory with the shear correction factor from strain energy [16] formulation (FSDTV) and equivalent classical
plate theory (ECPT) with the use of empirical shear correction factor derived from patch test considerations
respectively. It is well known that the CPT [15] under-predicts the deflection than shear deformable plate
theories for thick laminates [14,15]. It is found that the deflection and frequency results from FSDT are
functions of shear correction factors and width to thickness ratios of laminates and CPT results. Hence in the
present analysis, equivalent classical plate theory (ECPT) results are obtained with the use of empirical shear
correction factors which are found to be functions of Reissner’s homogeneous shear correction factors (5/6) and
side to thickness ratio (a/h) from a large number of numerical experiments on a patch test described elsewhere in
the paper.
(
Figure 1: Mesh patterns for a) triangular and b) rectangular elements, the boundary loads Pz , M x , M y ) are indicated
in the brackets
In order to confirm the convergence characteristics of HSDT4, FSDTC4, FSDTV4 and ECPT4, a patch test is
performed on plates with various mesh patterns which can represent constant moment in plates. The rectangular
plate in consideration is simply supported at three corners (i.e w=0) and located with a concentrated load of 2 at
the fourth corner and moments applied at the boundary as shown in Figure 1.The geometrical and material
properties with consistent units are: a=20.0, b=10.0, h=variable, E = 2.1 × 106 , ν = 0.30 . In order to establish
the benchmark for comparison, a widely used DKT plate bending element [17] is implemented presently. The
results from HSDT4, FSDTC4 and ECPT4 are compared with DKT in Table 1. From the results, it is observed
that all the plate models compare well with each other for width to thickness ratio of 10 4 which is also reported
in the works of Dvorkin and Bathe [11]. From numerical experiments, it is found that a value of shear correction
factor K Thin =
K c2
thinparameter
(
10 (6 K c − log 10(thinparameter ))
2
)
K c2 = 5 / 6, thinparameter = length / thicknessratio with
FSDT4 gives close results with respect to DKT results which we will denote as ECPT4 (Equivalent classical
four node plate element) henceforth. The comparative results from ECPT4 are also shown in Table 1. As seen
from the results, ECPT4 compares very well with DKT results for a wide range of width to thickness ratios.
In this example, a free vibration analysis is conducted on orthotropic (0), cross-ply (0/90) and angle-ply
(45/-45) square plates with a square cutout (c/a=0.5, c=size of cutout, a=size of the square plate) to validate the
accuracy of the present finite element formulation. The following material properties are adopted in the present
analysis: E1 / E 2 = 40 , G12 / E 2 = 0.50 , G13 = G 23 = G12 , G13 = G23 = G12 ν 21 = ν 12 E 2 / E1 a/h=100, ρ = 1.0 .
The present results are compared with Chakravorty et al [18] in Table 2 and a close agreement is observed.
Next, transient analysis is performed on simply supported symmetric (0/90/0) laminated composite plates
under time dependent step loading. The following geometrical and material properties [19] are used: a=b=5h
and h=0.1524 m. E1 = 172.369GPa, E2 = 6.895GPa, G12 = G13 = 3.448GPa, G23 = 1.379GPa, ν 12 = 0.25,
368
ρ = 1603.03kg / m3 . The intensity of the transverse load is taken to be qo = 68.9476MPa . The transverse load
q ( x, y , t ) is given by qo F (t )sin (πx / a )sin (πy / b ) where F (t ) is for step loading: 1 0 ≤ t ≤ t1 , 0 t >> t1 .
For this problem, a mesh density of 9 nodes per side in quarter of a plate is adopted. The value of t1 is taken
[19] as 0.006 sec. A time step of 20 µ sec is employed to integrate the governing equations of motion.
Comparative results from HSDT4, FSDTC4, ECPT4, HSDT, FSDTC and CPT are shown in Figure 2. Note that
HSDT, FSDTC and CPT results are obtained from the present analytical solutions. It is seen that both FEM and
analytical results match closely with each other. Again Effect of cutouts on transient response under step and
exponential blast loading q o F (t ) where F (t ) is for blast loading: exp(-t*330) 0 ≤ t ≤ t1 , 0 t >> t1 . is
shown in Figure 3. It is found that the response is higher when the cutout is considered. The findings are in
consistent with earlier publications.
Table 1: Vertical corner displacement of a rectangular plate under constant moment patch test
b/h DKT ECPT4 FSDTC4 HSDT4
5 0.185714* 0.185717 0.196634 0.196817
10 0.00148571 0.00148574 0.00150764 0.00150805
20 0.0118857 0.0118859 0.0119296 0.0119305
30 0.0401143 0.0401149 0.0401801 0.0401814
40 0.0950857 0.0950871 0.0951735 0.0951753
50 0.185714 0.185717 0.185824 0.185826
2
10 1.48571 1.48574 1.48593 1.48594
103 1485.71 1485.74 1485.72 1485.72
104 0.148571** 0.148589 0.148571 0.148571
* Multiplied by 10-3 ** Multiplied by 107
Table 2: Fundamental frequency  ωa 2 ρ a 4 h12(1 − ν 12ν 21 ) / E 2 h 3  for laminated square plates

 
Source Orthotropic(0) Cross-ply(0/90) Angle-Ply(45/-45)
Chakravorty et al 51.230 48.409 82.589
[18]
Present 51.507 48.317 82.216
5. CONCLUSION
In this paper, a new higher order computer aided finite element formulation is proposed to carry out the transient
analysis of laminated composite plates. The novelty of this formulation is that all two dimensional theories are
integrated in a single formulation. Due to the vortex rotations, the membrane response shows improved
performance. Analytical solutions of various two dimensional theories for plates without cutouts are obtained
for validation studies. The finite element formulation is applied to solve the transient response of composite
plates with cutouts which arise in many strategic applications.
REFERENCES
[1] Reddy, J.N. (1982) Large-amplitude flexural vibration of layered composite plates with cutouts, Journal of
Sound and Vibration, 83, 1-10.
[2] Bicos, A.S and Springer G.S. (1989). Vibrational characteristics of composite panels with cutouts, AIAA
Journal, 27, 1116-1122.
[3] Lee, H.P.; Lim, S.P. and Chow, S.T. (1992) Effect of transverse-shear deformation and rotary inertia on the
natural frequencies of rectangular-plates with cutouts. International Journal of Solids and Structures, 29,
1351-1359.
369
[4] Xiang, Y.; Liew, K.M. and Kitipornchai, S. (1993) Transverse vibration of thick annular sector plates,
Journal of Engineering Mechanics-ASCE, 119, 1579-1599.
[5] Jeng, S.T.; Hwang, G.C. and Yang, S.M. (1993) The effect of square cut-outs on the natural frequencies and
mode shapes of GRP cross-ply laminates, Composite Science and Technology, 43, 91-101.
[6] Boay, C.G. (1996). Free vibration of laminated composite plates with a central circular hole. Composite
Structures, 35, 357-368.
[7] Kong, J. and Cheung, Y.K. (1995) A generalized spline finite strip for the analysis of plates, Thin-Walled
[8] Liew, K.M, and Feng, Z.C. (2001) Three-dimensional free vibration analysis of perforated super elliptical
plates via the p-Ritz method, International Journal of Mechanical Sciences, 43, 2613-2630.
[9] Piancastelli, L. (1992) Some considerations on a four-node finite element for composites with the drilling
degrees of freedom, Computers and Structures, 43, 337-342.
[10] Cook, R.D. (1994). Four-node ’flat’ shell element: drilling degrees of freedom, membrane-bending
coupling, warped geometry, and behavior. Computers and Structures, 50, 549-555.
[11] Dvorkin, E.N. and Bathe, K.J. (1984) A continuum mechanics based four-node shell element for general
non-linear analysis, Engineering Computations, 1, 77-88.
[12] Nayak, A.K., Shenoi, R.A. and Moy, S.S.J. (2006) Transient response of initially stressed composite
sandwich plates. Finite Elements in Analysis and Design, 42, 821-836.
[13] Nayak, ,A.K., Moy, S.S.J. and Shenoi, R.A.,(2002) Free vibration analysis of composite sandwich plates
based on Reddy’s higher-order theory, Composites Part B-Engineering, 33, 505-519.
[14] Whitney, J.M. and Pagano, N.J. (1970) Shear deformation in heterogeneous anisotropic plates. Journal of
Applied Mechanics, Trans. ASME, 1031-1036.
[15] Reddy, J..N. (1997) Mechanics of Laminated Composite Plates and Shells, Theory and Analysis, 2nd
Edition, CRC Press, Boca Raton.
[16] Vlachoutsis, S. (1992) Shear correction factors for plates and shells. ,International Journal for Numerical
Methods in Engineering, 33, 1537-1552.
[17] Batoz, J.L. (1982) An explicit formulation for an efficient triangular plate bending element. International
Journal for Numerical Methods in Engineering, 18, 1077-1089.
[18] Chakravorty, D. and, Sinha, P.K. and Bandyopadhyay, J..N. (1998), Applications of fem on free and forced
vibration of laminated shells, .Journal of Engineering Mechanics, ASCE, 124, 1-8.
[19] Khdeir, A.A. and Reddy, J.N. (1989) Exact solutions for the transient response of symmetric cross-ply
laminates using a higher order plate theory. Composites Science and Technology, 34, 205-224.
370
Figure 2: Variation of the center deflection the non-dimensional in-plane stress σ xx (a / 2, b / 2, h / 2 ) / qo as a
function of time, for various plate theories under step loading
Figure 3: Variation of the center deflection for a) step and b) exponential blast as a function of time, for
analytical (19 series terms) and FEM solutions for plates with and without cutouts
371
Paper No.: 063
OPTIMAL DESIGN OF A RETRACTABLE ROOF STRUCTURE USING REAL

CODED GENETIC ALGORITHM
C. Mohanlal 1 and N. Murali Krishna 2
1
Associate Professor, Department of Civil Engineering, Vasavi College of Engineering, Hyderabad,, India
2
Professor, Department of Civil Engineering, University College of Engineering,
Osmania University, Hyderabad, India
Email: c_mohanlal@yahoo.com, nmuralikrishna1956@gmail.com
ABSTRACT:
In this paper, discrete optimization of a retractable roof structural design using real coded genetic algorithm with
hybrid mutation operator is presented. A typical retractable roof structure geometry with few arch frames to
support a set of stationary and movable roof elements is chosen. To achieve opening or closing of the roof, the
movable roof elements slide on the supporting arches. The loads from these movable roof elements get
transferred directly on to the supporting arches at their respective joints. An algorithm using C language is
developed to generate entire geometry of the retractable roof structure including the supporting arches. Design
procedure incorporating real coded genetic algorithm with hybrid mutation operator using discrete design
variables is developed for optimizing the weight of retractable roof structure. The objective function used here
minimizes the weight subjected to the constraints imposed by strength and serviceability requirements. A
benchmark problem that demonstrates the suitability of the hybrid mutation operator is presented. Limit state
method as per IS: 800-2007 is adopted for steel design. The structural design of retractable roof is carried-out
against load combination due to dead loads, live loads and wind loads. While carrying-out analysis against
wind, the permeability of roof structure i.e. roof fully closed, partial open and full open conditions are taken into
consideration.
Key words: Retractable roof, Genetic Algorithms, Optimization, Real coding, Mutation.
1. INTRODUCTION
In a retractable roof structure the roof elements are provided with the facility to move so that the roof can be
kept either open or closed. Such roof structures are desirable for sports stadiums to continue with the activities
even in the situations of adversities. The retractable roofs are in great demand where games of national and
international importance are scheduled and cannot be differed due to unfavourable weather conditions. With
retractable roof, the stadium may be used either as an indoor stadium or an outdoor stadium. The stadium
structure in general occupies extensive area. When the structure is covered with a retractable roof, it consumes
huge quantity of steel in its construction. Therefore the cost of structure is very high. Cost optimization of such
a huge structure can be achieved only when least amount of construction material is used. Even a small
percentage of reduction in the material of such a huge structure would result in enormous savings. Hence,
optimal design of such a structure is essential. Classical methods of optimization are problem dependent, the
solution depends on initial guess value and have the risk of ending up at suboptimal solution and are not suitable
to handle discrete optimization of large structures like retractable roof structures. Genetic Algorithms are
stochastic search and optimization procedures that are inspired by Darwin’s theory of evolution and the natural
law of survival of the fittest. The effectiveness of the genetic algorithm depends on the genetic operators used in
its implementation.
In earlier works the researchers had laid their focus on developing either an effective selection operator,
crossover operator or mutation operator. Only a few researchers have developed hybrid crossover operator. In
the present study a new mutation operator called as hybrid mutation operator is used to investigate its suitability
for structural optimization. The main purpose of the hybrid mutation operator is to increase the population
diversity, so as to avoid premature convergence so that the global optimum is not missed in the search process.
The objective function minimizes the weight of the structure subjected to performance constraints in the form of
stresses and displacements. A benchmark problem that demonstrates the suitability of the hybrid mutation
operator is presented. The validated genetic algorithm is used for optimal design of a retractable roof structure.
2. PROBLEM FORMULATION FOR OPTIMIZATION
The objective function for optimum design of truss structure problems and retractable roof frame problem is to
minimize the weight subjected to performance constraints in the form of stresses and displacements. The weight
may be represented as:
nm Where, W = Weight of the truss structure.
W = f(x) = ∑ γiaili γi = Weight density of member i,
i=1 ai = Cross-sectional area of member i,
Constraints gi(x) for truss structure problems: li = length of member i,
(a). σi ≤ σa for i=1 to n, (b). ∆i , ∆j, ∆k ≤ ∆a for i,j and k=1 to nj. nm = number of members in the truss.
(c). al ≤ ai ≤ au
Where, σi = stress in member i, σa = Allowable stress,
∆i , ∆j, and ∆k = Deflection in three orthogonal directions in space truss, ∆i and ∆j ,deflection in horizontal and
vertical direction in plane truss, ∆a = Allowable displacement of a joint, nj = number of joints in the truss
al = Lower bound cross-sectional area, au = Upper bound cross-sectional area,
Constraints gi(x) for retractable roof framed structure problem as per IS: 800-2007:
(a) Mi < Mdi ( if V < 0.6*Vd), else Mi < Mdvi ( if V > 0.6*Vd) , for i =1 to nm
(b) (Myi/Mndyi)2 + (Mzi/Mndzi)2 ≤ 1.0, (c) ∆i , ∆j, ∆k ≤ ∆a for i,j and k=1 to nj, (d) al ≤ ai ≤ au
Where, Mi = Design moment in member i, Mdi = Design bending strength in member i, Myi, Mzi = Factored
applied moments about minor & major axis of cross-section, Mndyi, Mndzi = Design reduced flexural strength
under combined axial force and the respective uniaxial moment acting alone.
If V < 0.6*Vd: Mdi = βb * Zp * fy/ϒmo ≤ 1.2* Ze * fy/ ϒmo
Where, V = Design shear, Vd = Design shear strength, βb =1 for plastic & compact sections, βb = Ze/Zp for semi
compact sections, ϒmo = partial safety factor, Zp,Ze = Plastic and elastic section modulus, fy = yield stress of the
material.
If V > 0.6*Vd: Mdvi = Mdi – β*(Md – Mfd) ≤ 1.2* Ze * fy/ ϒmo (for plastic & compact sections), Mdvi = Ze * fy /
ϒmo (for semi-compact sections), where, Mfd = plastic design strength of the area of cross-section excluding the
shear area, considering partial safety factor ϒmo.
Genetic algorithms are suitable for solving unconstrained optimization problems. Since the structural design
optimization is a constrained optimization problem, it is required to transform it into an unconstrained
optimization problem. The formulation proposed by (Rajeev and Krishnamoorthy, 1992) based on violation of
normalized constraints is adopted for this purpose. The constraints are represented in normalized form as
follows:
(σi / σa) – 1 ≤ 0 , ( ∆j / ∆a) – 1 ≤ 0 ( for truss structure problems)

(Mi/Mdi) -1 ≤ 0, (Mi/Mdvi) – 1 ≤ 0, (Myi/Mndyi)2 + (Mzi/Mndzi)2 ≤ 0 , ( ∆j / ∆a) – 1 ≤ 0(for retractable roof framed
structure problem)
The total violation coefficient is expressed as,

nc
where, ci = gi(x) , if gi(x) > 0
C = ∑ ci
or, ci = 0, if gi(x) ≤ 0
i=1
Now, the modified objective function φ(x) is written, by including the constraint violation as,
φ(x) = f(x) (1 + KC)
373
where, K is a penalty parameter, generally the value of which is chosen based on the influence of violation
coefficients in each generation. For all design example problems the value of penalty parameter considered is
K=1.
3. HYBRID MUTATION OPERATOR
The effectiveness of the genetic algorithm depends on exploring the search space by using the genetic operators.
In order to develop an effective genetic algorithm, a hybrid mutation operator is introduced for investigation.
The objective of hybrid mutation operator is to widen the population diversity so as to avoid premature
convergence. Hybrid mutation is carried out in two stages. It begins with riff-raff operation and is followed by
double blind mutation operation. Selection of high percentage of strong individuals may cause premature
convergence. Therefore, the riff-raff operation considers a predefined percentage of weak individuals of the
present population and copies them into the next generation. As the individual of the population and the location
from which the discrete design variable is selected are both random, the second stage of hybrid mutation
operator is named as double blind mutation operation.
The working procedure of the hybrid mutation operator is explained in step by step manner as follows:
Stage-I (Riff-raff operation):
1. Arrange all the individuals of the present population in ascending order of their fitness.
2. Let ‘ξ’ be the predefined number in percentage of the strong individuals. Obtain population in the next
generation by removing the last ‘ξ’ number of strong individuals and replacing them with the weakest ‘ξ’
number of individuals of the present population. Figure 1 shows the process of Riff-raff operation.
Stage-II (Double Blind mutation operation):
1. Sort all the discrete design variables under consideration in ascending order and store them in a one-
dimensional array as shown in figure 2 such that x1 < xi < xm , where x1 and xm represents the smallest and
the largest value of discrete design variable and xi being any intermediate value.
2. Randomly select an individual of the population to which mutation is to be applied. From among the genes
of this population find the smallest and largest value of design variable. Let g4 and g5 are the smallest and
the largest gene values.
3. Find the location at which these design variables are stored in the one-dimensional array. Let the locations be
represented as C1 and C2 as shown in figure 2.
Figure 1 Riff-raff operation Figure 2 Double blind mutation operation
4. The search space is increased such that the lower location in the one-dimensional array is shifted down to L1
= β*C1, and the upper location is shifted up to U1 = β*[C2 + (m-1)], where m = total number of discrete
design variables under consideration, β is a factor, the value of which can be between zero and one. A value
of 0.5 is considered for investigation.
5. Randomly select a location between L1 and U1, and fetch the corresponding value of discrete design variable
from this location of the one dimensional array. Let Cs is the selected location and xs is the value at this
location as shown in figure 2.
6. The value of discrete design variable xs so selected is replaced in a gene which is again randomly selected
from the individual of the population. Let g6 is the randomly selected gene. Replace gene g6 with the value
374
xs. To find the optimum fitness the percentage of weak individuals will be incrementally changed and the
algorithm is re-executed. From this process, the value of the percentage of weak individuals to be considered
for optimum fitness is obtained.
3.1. Experiments
To achieve the objective of the hybrid mutation operator, an experimental investigation is carried out by
deliberately permitting different percentages of weak individuals in the population. Range of the weak
individual percentages considered for this purpose varies from 5 to 50 per cent. To carryout investigation the
genetic algorithm is run in two stages listed below:
1. In the first stage the genetic algorithm is executed with a population size and crossover probability under
consideration, but with the percentage of weak individuals ‘ξ’ ranging from 5 to 50 with a step size of 5 per
cent. From this stage of investigation, the percentage of ‘ξ’ that gives optimum fitness value is determined.
2. In the second stage, the percentage of weak individuals ‘ξ’ obtained from the first stage is considered and
the genetic algorithm is re-executed, now by changing the probability of crossover from 30 to 90 per cent
with step size of 5 per cent. From this, the probability of crossover which gives fitness that is global optimum
is determined.
3.2. 25-bar space truss design example
Figure 3 shows configuration of the 25-bar space truss. Each joint of the space truss is numbered. The material
properties: E = 6.89 x 104 MPa; γ = 2,770 kg/m3; the displacements of the joints is restricted to be less than
±8.89mm. The stresses in tension and compression are limited to 275.8Mpa.
The list of discrete design variables is (area of cross-sections in sq.mm.) = (64.52, 129.03, 193.55, 258.06,
322.58, 387.10, 451.61, 516.13, 580.64, 645.16, 774.19, 838.71, 903.22, 967.74, 1032.26, 1096.77, 1161.29,
1225.80, 1290.32, 1354.84, 1419.35, 1483.87, 1548.38, 1612.90, 1677.42, 1741.93, 1806.45, 1935.48, 2064.51,
2193.54) – 30 nos. The details of the loading considered and the grouping of the members are given in table 1
and table 2 respectively.
Table 2 Member groups
Table 1 Details of loading
Figure 3 configuration of 25-bar space truss.
375
Table 3 Comparison of results of 25-bar space truss design problem
Figure 4: 25-bar space truss design problem with population size – 20
3.3. Results and discussion of 25-bar space truss
Results of the 25-bar space truss along with the results of the previous researchers are shown in Table 5.
Optimum design variables that are obtained by execution of genetic algorithm are listed by considering
376
population sizes as 20, 30 and 40. Figure 4 through figure: 6 show the variation of fitness with change in % of
weak individuals (ranging from 5% to 50%) and variation of fitness with change in crossover probability
(ranging from 0.3 to 0.9). With population size as 20, the genetic algorithm permits 30% of weak individuals to
obtain optimum weight with probability of crossover as 0.35. With 30 as the population size, 25% of weak
individuals can be allowed at crossover probability of 0.5. For population size of 40, 30% of weak individuals
can be allowed with crossover probability of 0.45. Table 3 show that the results obtained from the present study
are better than the previous studies.
4. RETRACTABLE ROOF STRUCTURE GEOMETRY
The proposed retractable roof structure is a segment of a hemi-spherical dome. Figure 7a shows the top view of
the retractable roof structure in full open position. The structure comprises of the supporting arch structure and
movable panels that slide on these arches. The supporting arch structure consists of two main arches A and B
and six quarter arches as shown in figure 7a. The ends of the two main arches rest on the ground. One end of the
quarter arches rest on a main arch and the other on the ground. The profile of the top chord of these arches
match with the profile of the dome at respective locations. The main arch and the quarter arches each are a
three-layered box girder space frames. The roof of the stadium consists of six set of movable panels. Two set of
panels’ bi-part over the middle of the playing field by sliding over the main arches. Whereas, the remaining four
set of panels slide on quarter arches. Each set of panels comprises one stationary and two moving panels. When
the roof is open the movable panels rest above the stationary one. When the roof is closed figure 7b shows the
position of panels on the whole structure.
Figure 7 a & b Top view of retractable roof structure in full open & fully closed position
The retractable mechanism consists of wheels at the ends of the panels resting on rails held by supporting
arches. Each movable panel is made up of a set of truss girders. Each truss girder of the panel in turn is
connected by means of rafters. Truss girders and rafters put together forms a grid that supports the sheeting that
makes up the roof. The sheeting may be made up of either PVC-coated sheets or polytetrafluoroethylene (PTFE)
coated glass fibre.
4.1. Size of retractable roof stadium
The size of the retractable roof stadium depends on the type of pitch and the type of sport played according to
Peter Culley and John Pascoe (Peter Culley and John Pascoe, 2005). Depending on the type of game played the
diameter of the viewing area is decided (Geriant John and Rod Sheard, 1994). Here the optimum viewing
distance and the maximum viewing distance are the two criterions which are considered. The height of the
farthest spectator’s location with respect to the surface of the pitch is calculated as recommended by Geriant
John and Rod Sheard (Geriant John and Rod Sheard, 1994). The main arches supporting the retractable roof are
aligned along the length of the pitch. The offset of the main supporting arches from the centre of the pitch
depend on the width of the pitch. The geometrical dimensions of the main arches are found based on these
criteria. The geometry of the supporting arches and the movable panels are generated once the span, the radius
and the height are obtained. Figure 8 show the geometry of retractable roof supporting structure and figure 9
shows the geometry of one of the movable panel generated by the program developed for this purpose.
377
4.2. Design of retractable roof structure
Problem statement of retractable roof structure is as follows:

Stadium diameter: 100 m., Height: 20 m., No. of members in retractable supporting structure: 6628, No. of
joints in retractable supporting structure: 1968, No. of degrees of freedom of retractable supporting structure:
11448. The design is as per IS: 800-2007. The material properties: Elastic modulus, E= 2x105 MPa, ϒ=2767.99
kg/m3. Allowable displacement: 50 mm, No. of member groups for each movable panel – 8 nos. No. of member
groups for the retractable roof supporting structure – 26 nos.
Y
X
Z
Figure 8 Geometry of retractable roof supporting structure Figure 9 Geometry of a movable panel
The discrete design variables (the cross-sectional areas of members in sq. mm) are listed as follows:
121, 155, 181, 184, 202, 241, 258,311, 331, 377, 400, 414, 460, 488, 523, 563, 582, 650, 744, 789, 831, 874,
1010, 1080, 1120, 1240, 1270, 1280, 1470, 1550, 1730, 1850, 1860, 1910, 2050, 2060, 2090, 2250, 2270, 2280,
2320, 2440, 2490, 2710, 2760, 2870, 3190, 3210, 3480, 3760, 3950, 4420, 4950, 6340, 8650.
Loads on the structure:
Movable panels:
1. Dead load: Self weight of the members.
2. Live load: Joint load of 10 KN per each joint of movable panel.
3. Wind load: Uniformly distributed load of 5 KN/m on the members of movable panels.
Retractable roof supporting structure:
1. Dead load: Self weight of the members.
2. Live load: (a). Wheel loads from the movable panels on the respective joints. Wheel load of Panel C:
233 KN, Wheel load of Panel B: 213 KN, Wheel load of Panel F: 234 KN, Wheel load of Panel E: 201
KN (b). Joint load of 10 KN per joint at other joints.
3. Wind load: Uniformly distributed load of 5 KN/m on the members and 5KN joint load acting
horizontal.
Table 4 Results of optimum design of movable panels
4.3. Retractable roof structure results and discussion
Optimal design of the retractable roof movable panels and the supporting structure is carried out using real
coded genetic algorithm with hybrid mutation operator. The population size, percentage of weak individuals
considered and the crossover probability are 20, 30 and 45 respectively. Table 4 shows the results of optimum
378
design of movable panels. The optimum weight of movable panels C and B which move on the main arches is
739.16 KN and 468.25 KN respectively. The optimum weight of movable panels which move on the quarter
arches is 718.18 KN and 465.85 KN respectively. Therefore, the optimal weight of retractable roof (movable
panels) is 7154.54 KN. Table 5 shows the results of optimum design of retractable roof supporting structure
when the roof is fully closed, full open and partial open conditions. The optimum weight of retractable roof
supporting structure when the roof is in the above three positions is 47172.75 KN, 53299.88 KN and 51677.40
KN respectively.
Table 5 Results of optimum design of retractable roof supporting structure
5. CONCLUSIONS
As the retractable roof structure in general requires huge quantity of material for its construction, reduction in
material consumption is highly desirable. An effective real coded genetic algorithm with a new mutation
operator called as hybrid mutation operator is developed. Suitability of the hybrid mutation operator for discrete
optimization problems is demonstrated by using a bench mark problem. The validated genetic algorithm is used
for the optimal design of a retractable roof structure and the results are presented.
REFERENCES
[1] Erbatur F.; Hasancebi O.; Tutuncu L. and Kolic H. (2000). Optimal design of planar and space structures
with genetic algorithm. Comput. Struct., 75, 209-224.
[2] Geraint John and Rod Sheard (1994). Stadia- A Design and Development Guide. Butterworth Architecture
publishers.
[3] Peter Culley and John Pascoe (2005). Stadium Engineering. Thomas Telford publishers.
379
[4] Rajeev and Krishnamoorthy C.S.(1992). Discrete optimization of structures using genetic algorithms.
J.Struct. Engg. ASCE, 118(5), 1233-1250.
[5] Teyfun Dede; Serkan Bekiroglu and Yusuf Ayvaz (2011). Weight minimization of trusses with genetic
algorithm. J. Applied Soft Computing, 11, 2565-2575.
[6] Zhu D.M. (1986). An improved Templeman’s algorithm for optimum design of trusses with discrete member
sizes. Engrg. Opt., 9, 303-312.
380
Paper No.: 064
THE CHANGE IN STRESS DISTRIBUTION PATTERN WITHIN THE

EARTHEN DAM DUE TO THE EFFECT OF SEEPAGE
Dr. Nityananda Nandi1
1
Associate Professor, AE and AM Department, BESUS, Howrah 3, West Bengal.
ABSTRACT:
The present study makes an effort to understand the damage of earthen dams with a high flood level. To make
the investigation more realistic, behaviour of earthen dams considering the occurrence of phreatic line indicating
the submerged zone due to seepage within the dam body is considered. In case of the earthen dams,
homogeneous or nonhomogeneous, the consideration of the occurrence of phreatic line or seepage line through
the dam body is an important part of the earthen dam design methodology. The impervious material properties
in the submerged zone below the phreatic line due to seepage may differ a lot in magnitudes as compared to the
value of the same materials lying above this line. Hence, to have the exact stress distribution scenario within the
earthen dam, the different material properties above and below the phreatic line are considered in this present
study. The study is carried out by two-dimensional as well as three-dimensional finite element analysis under
static loading condition. The study is extended considering the materially nonlinear properties also for these two
conditions. It is observed that in this situation the location of the most of the tension zones is shifted around the
phreatic line. This may lead to a favorable condition for piping. Hence, a slight ground shaking during
earthquake excitation will not only cause the structural failure of the earthen dams but also invite the piping
failure through widened crack which may be generated in the tension zone near the phreatic line due to shaking.
KEYWORDS: earthen dam, seepage, phreatic line, finite element, nonlinear, tension zone.
1. INTRODUCTION:
Consideration of seepage within earthen dams is a very crucial aspect of the dam design because damage of
earthen dams due to piping is quite common. In stability of slopes, the seepage phenomenon plays a very crucial
role by reducing the shear strength of the soil due to the existence of pore pressure. A large number of studies
have been made on the behaviour of dams [Adhikari DP and Chattopadhyay A. 1996, Bhattacharjee SS and
Leger P. 1994, Bhattacharjee SS and Leger P. 1995, Cervera et. al 2000, Chavez JW and Fenves GL. 1995,
Chopra AK and Chakraborty P. 1972, Chopra AK and Zhang L. 1991, Dhariwal A and Purohit DG. 2000,
Dutta et. al. 2002, Ghobarah A, and El-Nady A, Aziz T. 1994, Huang TK. 1996, Hussain et. al 2005, Makdisi
FI and Seed HB. 1977, Majumder DK. 2010, Nandi et. al 2005, Nandi et. al. 2005, Nandi N and Dutta SC.
2002, Nandi et. al 2001, Nandi N and Dutta SC 2003, Nandi et. al 2010, Sarkar et. al 2003, Zeghal M and
Abdel-Ghaffar AM. 1992, Zhao et. al. 1995]. Some of the studies considered the effect of occurrence of
phreatic line in earthen dam design methodology [Majumder DK, Hussain et. al 2005]. One of such studies
[Majumder DK] shows that actual pore water pressure seems to be lesser than hydrostatic pressure. Thus,
consideration of a hydrostatic pressure within the pore may lead most critical condition resulting in
underestimation of factor of safety. However, in this context, the issue of the occurrence of tension zone within
the earthen dam body considering the occurrence of phreatic line is still an area which needs further
investigation. It has been found in some of the previous studies [Nandi et. al 2005, Nandi et. al. 2005, Nandi N
and Dutta SC. 2002, Nandi et. al 2001, Nandi N and Dutta SC 2003, Nandi et. al 2010, Sarkar et. al 2003] that
the occurrence of tension zone at the crest level of the earthen dams may be the crucial issue of the dam design
methodologies. Hence, it is interesting to see whether the analysis considering the existence of phreatic line
within the dam body may also lead to the similar understanding regarding this aspect of the stress distribution
scenario. An earthen embankment surrounding the Ash Pond of Bakreswar Thermal Power Project, West
Bengal, is taken as an example dam for this analysis. Below the phreatic line the soil is assumed to be in
submerged condition due to a hydrostatic pore pressure perhaps implying a somewhat lesser factor of safety
than actual and accordingly the material properties are considered in this region being in more conservative side.
The stress distribution pattern within the dam body is observed under static loading condition. The study is
extended considering the materially nonlinear properties by both two-dimensional and three-dimensional
modeling. It is found in static analysis that the location of the tension zone is mainly concentrated around the
phreatic line and approximately in the middle of the dam section. The observations are finally summarized in
the paper.
The geometrical cross sections of this dam section have been changed to see whether these changed sections
may be effective to change the stress distribution pattern after considering the occurrence of the phreatic line
within the dam section. The original dam section is then modified by the steeper upstream slope, deep cut-off
trench and the large widened lower berm. Both two-dimensional as well as three-dimensional analyses are
carried out to observe the stress zone for these modified sections. It has been found that the geometrical cross-
section may not have any influence to change the stress distribution pattern within the dam body.
2. MODELING AND ANALYSIS:
2.1. Finite element modeling:
A six-noded triangular element is used for two-dimensional finite element analysis. The triangular element in
two-dimensional modeling is particularly used to model the geometry with sufficient accuracy near the corner of
such dams. To ensure accuracy further, convergence study is carried out to arrive at adequate level of
discretization. A brief detail about two-dimensional finite element modeling used in this study has been
presented in the literature [Desai and Abel, 1987]. A ten-nodded tetrahedron element is used for three-
dimensional finite element analysis. This element has quadratic displacement function and is suitable for
irregular meshes. The element is defined by ten nodes each of which having three degrees of freedom, namely,
translations in the nodal x, y and z directions denoted by u, v and w respectively. The details of this element are
shown in Figure 1. The shape functions can be given as follows:
u = uI (2l1-1) l1 + uJ (2l2-1) l2 + uK (2l3-1) l3 + uL (2l4-1) l4 + 4(uMl1l2) + uNl2l3 + uOl1l3 + uPl1l4 + uQl2l4 + uRl3l4
(1)
v = vI (2l1-1) l1 analogous to u (2)
w = wI (2l1-1) l1 analogous to u (3)
where, u, v and w with suffix representing the corresponding nodal points, denote the displacements along x, y
and z directions, respectively.
2.2. Elasto-Plastic analysis:
For this nonlinear analysis, as a yield criterion a generalization of Mohr-Coulomb failure criterion as proposed
by Drucker and Prager is used. This criterion is basically a modified Von-Mises criterion which is dependent on
hydrostatic stress, σm = 1/3 (σx + σy + σz) and can be expressed as
3α (σx + σy + σz) + [(σ1- σ2)2 + (σ2- σ3)2 + (σ3- σ1)2]1/2 = σy (4)
where σx, σy and σz are stress components along x, y and z directions, while σ1, σ2 and σ3 are the principal
stresses. This criterion is considered as a very suitable criterion that governs the shear failure in soil. In this
criterion while being used in ANSYS software, α is taken as 2 sinφ/ √3 (3- sinφ) and σy = 6c cosφ/ √3 (3- sinφ)
where c is the cohesion of soil and φ is the angle of internal friction. However, since soil cannot take tension at
all, an analysis based on this criterion considers the shear action as the only possible way of failure. But the
present study shows that the occurrence of tension cannot be avoided and this can be well pointed-out by a
linear analysis only. Thus, while carrying out nonlinear analysis with the above criterion, propagation of
cracking due to tension perhaps cannot be recognized.
382
2.3. Location of seepage line
The prediction of the position of the seepage line within the cross section of the nonhomogeneous earthen dam
with a very high accuracy is a complex problem. However, the procedure for locating the same the method has
been described in standard literature [Winterkorn HF and Fang HY, 1986]. It has been mentioned in the
literature that while locating the phreatic line of the earthen dams with impervious core, the presence of the
pervious zone may be neglected and water level may be considered unobstructed upto the boundary of the core.
Then the actually curvilinear phreatic line can be located within the impervious core following the standard
procedure. Again on the downstream side horizontal water level will continue from the point where phreatic line
meets the boundary of the core. This consideration has been made in the present study and the curvilinear path
of the phreatic line ABCD has been constructed following the standard literatures [Winterkorn HF, Fang HY.
1986]. The locus of the phreatic line as computed in a typical case is shown in Figure 2. Similar loci of phreatic
line in all other cases have been rigorously calculated but not presented here for the sake of brevity.
3. DETAILS OF THE EMBANKMENT STUDIED
The section of the embankment chosen in the study is considered to have the same geometry as that of the
embankment surrounding the Ash Pond of Bakreswar Thermal Power Project situated at Panuria village in
Birbhum District, West Bengal, India. This is a zoned type earthen embankment which is constructed to make a
reservoir for disposal of fly ash mixed slurry. The height of the embankment is taken as 18m. The width of the
embankment at the base and at the top is considered as 91m and 4m, respectively. Both the upstream and
downstream slopes of the embankment are considered as 2.25H: 1V. At the downstream side, there are two
horizontal berms of width 3m at a height of 6m and 12m from ground level, respectively. A cut-off trench of
depth 2m is provided. It is considered that the maximum height of fluid level is same as the total height of the
embankment. Above the phreatic line, the specific weight of the material is taken as 20.5 kN/m3 with modulus
of elasticity 50,000 kN/m2 and Poisson’s ratio 0.35 for the pervious casing zone whereas for the impervious
core, the specific weight of the material is taken as 19.8 kN/m3 with modulus of elasticity 70,000 kN/m2 and
Poisson’s ratio 0.48, respectively. The geometrical configuration of this original existing embankment is shown
in Figure 3. For impervious core below the phreatic line, the specific weight in saturated condition is taken as 20
kN/m3 with minimum value of Modulus of Elasticity 500 kN/m2 and maximum value of Poisson’s ratio 0.49.
For the pervious casing zone it is assumed that the material properties below the phreatic line are not changed
very much and the specific weight in saturated condition is taken as 21 kN/m3 with modulus of elasticity 50000
kN/m2 and maximum Poisson’s ratio 0.4. The range of variation of material properties was primarily on the
basis of large number of laboratory test results conducted during the construction of the dam. However, the
values of these material properties are well within the range prescribed in the standard literature [Bowles JE
1986, Winterkorn HF and Fang HY, 1986]. The above studies have been repeated for some modified
geometrical cross sections of the earthen dam which are shown in Figures 4, 5 and 6.
Most of the times, the depth of water in the upstream side is somewhat in between the high flood level and
empty condition. So, the analysis is carried out to yield the stress distribution pattern within the earthen dams
corresponding to the different depths of water in the upstream side. It is found that the results obtained for
different depths of water represent more or less same trend regarding the stress distribution pattern within the
dam body. Hence, the representative results corresponding to the reservoir full condition in the upstream side
are presented in the two following subsections, for the sake of brevity.
4.1. Two-dimensional analysis
Analyses under static loading, free vibration as well as response spectrum based CQC analysis are conducted
through two-dimensional modeling. Results are presented in the following sub-sections.
383
4.1.1. Static analysis
Table 1 represents the principal stresses and the maximum shear stress developed in some key points within the
dam body considering the submerged material properties below the phreatic line corresponding to the reservoir
full condition in the upstream side for the original embankment section. Table 2 represents the values
corresponding to the materially nonlinear behaviour. The stress distribution scenario is also attempted to be
represented pictorially. For instance, Figure 7 represents the tension zone and the maximum compression zone
within the dam body corresponding to the reservoir full condition at the upstream side for the original
embankment section. Figure 8 is the pictorial representation of the stress distribution pattern corresponding to
drawdown condition. The study is extended considering the material nonlinear properties of the soil and the
results are presented corresponding to the material properties below the phreatic line as in submerged condition.
Figure 9 is the pictorial representation of the tensile stress zone and the maximum compressive stress zone
within the dam body due to materially nonlinear properties of the soil and submerged condition below the
phreatic line. Figure 10 shows the plastic zone developed within the dam section. The study is also carried out
for the different depths of water at the upstream side and the same trend of stress distribution scenario regarding
the issue of the occurrence of tension zone has been observed in each case. Hence, all the Figures are not
presented here for the sake of brevity.
Figure 7 shows that almost all the tension zone is concentrated at the upper portion of the dam section shifted
slightly towards the upstream side. This tension zone is also observed to extend below the phreatic line. A small
tension zone is also found to occur at the upstream face below the seepage line. Within the tension zones, the
tensile stress varies from 25 to 45 kN/m2. Maximum compressive stress varies from -250 kN/m2 to -350 kN/m2.
Since, the tension zone exists below the phreatic line shifted towards upstream side, soil will be in a very loose
condition in that portion. If a small crack is developed piping will occur through this crack within the dam body
due to the presence of water in that zone. This may lead to failure of earthen dams. This may be a probable
explanation of the piping failure of many earthen dams during flooding. In case of drawdown condition the
stress distribution pattern is only changed slightly otherwise, the same phenomenon occurs for the drawdown
condition as shown in Figure 8 regarding the occurrence of tension zone. It is observed from Figure 9 that the
tension zone also exists within the dam section. Figure 10 points out that the plastic zone initiates below the
phreatic line at the junction of the two materials and extends below the seepage line. Hence, it is clear that the
issue of the occurrence of tension zone within the dam body as well as below the phreatic line is not altered
even after the shear yielding of the materials. The study has further been extended to modify the geometrical
cross section of the earthen dam through changing upstream slope, deep cut-off trench and large widened lower
berm. All the above studies are repeated considering the occurrence of phreatic line for each of the cases to see
whether these changed sections exhibit any considerable alteration in the stress distribution pattern. It is seen
that the stress distribution scenario is only marginally changed as compared to the original section even after the
consideration of the occurrence of phreatic line. The central findings regarding the occurrence of tension zone
more or less remain unaltered for all the above case studies. Hence, these case studies are not presented in
details for the sake of brevity.
4.2. Three-dimensional analysis
Three-dimensional analysis is carried out by ten-nodded tetrahedron element. Dam section is modeled along the
straight geometry as well as curvilinear geometry in plan to have an idea whether the curved geometry in plan
may influence the stress distribution scenario within the dam body. Though there are quantitative changes in the
stress distribution, no qualitative change is observed between the two cases particularly with reference to the
occurrence of tension zone, the results corresponding to the straight geometry in plan are presented in the
following sub-sections.
4.2.1. Static analysis
Table 3 represents the principal stresses and the maximum shear stress developed in some sections within the
dam body corresponding to the material properties below the phreatic line in submerged condition for the
reservoir full condition in the upstream side for the original embankment section. Table 4 represents the values
of principal stresses and strains along x, y and z direction corresponding to the material nonlinear properties.
384
Figure 11 shows the tension zone and the maximum compression zone occurred within the dam body of a
typical intermediate section when analyzed by considering the material nonlinear properties of the soil. Figure
12 shows the plastic zone in a typical intermediate section occurred within the dam body.
Figure 11 implies that almost all the tension zone is concentrated around the phreatic line approximately in the
middle of the dam body between upper berm and lower berm position of the dam cross section. This tension
zone is also observed to extend below the phreatic line. A small tension zone is also found to occur at the crest
level as well as at the upper berm position. In this case, maximum tensile stress varies from 20 to 40 kN/m2
whereas maximum compressive stress varies from -200 to -300 kN/m2. Figure 12 points out that the plastic zone
begins below the phreatic line at the junction of the two materials and extends towards the downward direction.
Hence, it is clear that the issue of the occurrence of tension zone below the phreatic line is still pertinent though
the shear yielding of the materials takes place. This indicates that the shear yielding does not release tension.
Now, the geometrical cross section of the earthen dam is modified with the steeper upstream slope, deep cut-off
trench and large widened lower berm. All the above studies are repeated considering the occurrence of phreatic
line for each of the modified sections to see whether these changed sections may alter the stress distribution
pattern. It is seen that the stress distribution scenario is not changed as compared to the same in the original
section. Hence, these issues are not discussed in details for the sake of brevity.
5. CONCLUSIONS:
The occurrence of tension zone at the crest level of the earthen dams corresponding to different depths of water
at the upstream side may be a possible cause of failure of such dams as observed during earthquakes. The
present study attempts to find out whether the occurrence of phreatic line within the dam body has any influence
on such observations.
a) The occurrence of phreatic line implies the consideration of material properties in submerged condition below
the phreatic line. In this situation the tension zone exists and furthermore the location of the most of the tension
zones is shifted closure around the phreatic line. This may lead to a favorable condition for piping. The dynamic
analysis also shows more or less same stress distribution scenario within the dam body. Hence, a slight ground
shaking during earthquake excitation may not only cause the structural failure of the earthen dams but also
invite the piping failure through widened crack which may be generated in the tension zone near the phreatic
line due to shaking.
b) Due to the tendency of lateral extension at the top as understood from the nature of first few mode shapes, the
vertical crack is also likely to occur at the crest under seismic ground shaking, even if the submerged zone
below the phreatic line is present. The modifications in the configuration of the section of the earthen dam,
namely, the change in upstream slope or provision of deep cut-off trench or provision of large lower berm do
not effectively modify the stress distribution scenario as well as dynamic characteristics of earthen dams even
after the consideration of the occurrence of phreatic line within the dam body. The curvilinear geometry along
longitudinal axis also does not seem to bring any qualitative change in the stress distribution scenario as well as
the dynamic characteristics of the earthen dams.
REFERENCES
[1] Adhikari DP, Chattopadhyay A. (1996). Behaviour of earth and rockfill dam under earthquake. Proc
Symposium on Earthquake Effects on Structures, Plant & Machinery, New Delhi (Nov.).
[2] Bhattacharjee SS, Leger P. (1994). Application of NLFM models to predict cracking in concrete gravity
dam, J Struct Eng, 120(4), 1255-1271.
[3] Bhattacharjee SS, Leger P. (1995), Fracture response of gravity dam due to rise of reservoir elevation. J
Struct Eng 121(9): 1298-1305.
[4] Bowles JE. Foundation Analysis and Design. McGraw-Hill, Inc. 4th International Edition; 1996.
[5] Cervera M, Oliver J, Prato T. (2000). Simulation of construction of RCC dams I: temperature and aging. J
Struct Eng 126(9), 1053-1061.
385
[6] Cervera M, Oliver J, Prato T. (2000). Simulation of construction of RCC dams. II: stress and damage. J
Struct Eng 126(9), 1062-1069.
[7] Chavez JW, Fenves GL. (1995). Earthquake response of concrete gravity dam including base sliding, J
Struct Eng 121(5), 865-875.
[8] Chopra AK, Chakraborty P. (1972). The earthquake experiences at Koyna dam and stresses in concrete
gravity dams. Earthquake Eng & Struct Dyn 1, 151-164.
[9] Chopra AK, Zhang L. (1991). Earthquake induced base sliding of concrete gravity dams. J Struct Eng
117(12), 3698-3719.
[10] Chopra AK. (2001), “Dynamics and structures” Pearson education (Singapore) Pte. Ltd., India Branch,
Delhi 110 092.
[11] Clough RW, Chopra AK. (1997). Earthquake stress analysis in earthen dams. J Eng Mech Div ASCE; 92
(EM 2): 197-210.
[12] Desai CS and Abel JF. Introduction to the Finite Element Method. CBS, India; 1987.
[13] Dhariwal A, Purohit DG. (2000). Failure philosophy of embankment dams. Proceedings of 2000 IGC the
Millennium Conference, Mumbai, India.
[14] Dutta SC, Sarkar P, Nandi N. (2002). An example based study on behavior of earthen embankments.
Proceedings of National Seminar on Recent Challenges of Civil Engineering, Hyderabad, 1: 69-81.
[15] Ghobarah A, El-Nady A, Aziz T. (1994). Simplified dynamic analysis for gravity dams. J Struct Eng
120(9), 2697-2716.
[16] Huang TK. (1996). Stability analysis of an earth dam under steady state seepage. Comput Struct 58(6):
1075-1082.
[17] Hussain Abdul A, Hari Prasad KS and Kashyap D. (2005). Modelling of pore water pressure distribution
in an earthen dam and elevation of its slope stability. Dam Engineering, XV (3).
[18] Makdisi FI, Seed HB. (1977). A simplified procedure for estimating earthquake-induced deformations in
dams and embankments. Earthquake Engineering Research Centre, University of California, Berkeley,
(UCB/EERC-77/19).
[19] Majumder DK. Stability of soil slopes with pore pressure. Indian Geotechnical Journal, 3(4), 334-341.
[20] Murthy VNS. (1989). Soil Mechanics & Foundation Engineering. Sri Kripa Technical Consultants,
Bangalore-560 070.
[21] Nandi N, Dutta SC, Roychowdhury A. (2005). Analysis on shear failure of earthen dams. Dam
Engineering. Xv(4). 255-282.
[22] Nandi N, Dutta SC, Roychowdhury A. (2005). Effect of soil-structure interaction on stress distribution of
earthen dams. Proc. of the International Conference on Structural and Road Transportation Engineering
(START-05), KGP 05 (Jan); 1: 326-333.
[23] Nandi N, Dutta SC. (2002).Tension at crest: a possible reason of seismic failure of earthen dams. Proc. of
12th Int. Symp. of Earthquake Engg. Roorkee, 1: 1000-1007.
386
[24] Nandi N, Sarkar P, Dutta SC. (2001). Study on behavior of a few Indian dams and embankments.
Proceedings of International Conference on Theoretical, Applied, Computational and Experimental
Mechanics (ICTACEM), Kharagpur.
[25] Nandi N, Dutta SC. (2003). A study on shear failure of earthen dams. Proceedings of Int. Conf. on
Structural Engineering Convention (SEC 03), KGP. 1: 636-647.
[26] N. Nandi, S. C. Dutta and A. Roychowdhury, (2010). ‘Explanation of seismic failure possibilities through
dynamic and response analysis of earthen dams’ International journal of Dam Engineering, (ISSN 0-617-
00563-X), XXI (1), 45-67.
[27] Sarkar P, Dutta SC, Nandi N. (2003). A critical review of dam analysis methodologies. Int J App Mech
Eng 8 (3): 461-482.
[28] Winterkorn HF, Fang HY. (1986), “Foundation Engineering Hand Book”. Galgotia Book source, Van
Nostrand Reinhold Comp. Inc.
[29] Zeghal M, Abdel-Ghaffar AM. (1992). Local-global finite element analysis of the seismic response of
earth dams. Comput Struct 42(4): 569-580.
[30] Zhao C, Xu TP, Valliappan S. (1995). Seismic response of concrete gravity dams including water-dam-
sediment-foundation interaction. Comput Struct 54(4): 705-716.
[31] Available in article ‘Dams’ in the website http://geoinfo.use.edu/gees/Recent EQ/India-

Gujarat/Report/dams.
Table 1: Principal stresses and maximum shear stress obtained by 2D finite element based analysis
corresponding in reservoir full condition.
Principal stresses (Smax , Smin ) and

Positions maximum shear stress (τmax), kN/m2
Smax Smin τmax
U/S face -94.74 -168.75 +37.01
Core Heel -133.15 -175.76 +21.31
Ground Level
Core Toe -132.54 -172.46 +19.96
D/S face +0.31 -0.68 +0.50
U/S face -92.43 -168.72 +38.15
Mid section -86.26 -124.53 +19.14
Lower Berm
Inner node -0.92 -110.75 +54.92
Outer node -0.45 -27.20 +13.38
U/S face +16.56 -168.42 +92.49
Mid section +17.06 -174.25 +95.66
Upper Berm
Inner node -12.54 -154.89 +71.18
Outer node -25.76 -152.67 +63.46
U/S face +2.45 -123.00 +62.73
Crest Level Mid section +44.34 +0.03 +22.16
D/S face +39.38 -1.71 +20.55
387
Table 2: Principal stresses and strain along x and y direction obtained by 2D finite element based analysis
corresponding to the nonlinear material properties in reservoir full condition.
Principal stresses (Smax , Smin ) ∈x)

Strain along x (∈
and maximum shear stress ∈y) directions
and y (∈
Positions
(τmax), kN/m2
Smax Smin τmax ∈x ∈y
U/S face -82.03 -118.34 +18.16 -0.86e-04 +0.28e-04
+0.18e-
Ground Core Heel -168.35 -270.65 -0.22e-01
+51.15 02
Level
Core Toe -165.73 -272.55 +53.41 -0.73e-03 -0.19e-01
D/S face -176.14 -317.84 +70.85 -0.14e-17 -0.44e-06
+0.24e-
U/S face -24.25 -81.98 -0.84e-03
+28.87 03
Lower +0.26e-
Mid section -50.72 -55.00 -0.89e-02
Berm +2.14 02
Inner node -1.00 -41.73 +20.37 -0.71e-03 +0.36e-03
Outer node -0.33 -12.62 +6.15 -0.17e-03 +0.62e-04
U/S face -91.87 -126.67 +17.40 -0.41e-03 -0.88e-03
Mid section -45.67 -93.52 +23.93 -0.17e-02 +0.16e-02
Upper +0.40e-
Inner node +5.42 -56.06 -0.16e-03
Berm +30.74 03
+0.78e-
Outer node -12.48 -73.52 -0.4e-04
+30.52 04
U/S face +4.71 -80.40 +42.56 -0.14e-02 +0.98e-03
Crest
Mid section +47.04 +1.02 +23.01 -0.18e-02 +0.78e-03
Level
D/S face +18.50 +1.31 +8.60 -0.11e-02 +0.57e-03
Table 3: Principal stresses and maximum shear stress obtained by 3D finite element based analysis in reservoir
full condition for a typical intermediate section of the original embankment surrounding the Ash Pond.
Principal stresses (S1, S2 and S3) and maximum

Positions shear stress (τmax), kN/m2
S1 S2 S3 τmax
U/S face +21.27 -12.76 -103.62 +62.45
Ground Core Heel -149.87 -149.92 -254.87 +52.50
Level Core Toe -148.41 -147.52 -256.97 +54.28
D/S face +4.75 -2.86 -82.86 +43.81
U/S face +3.00 -12.42 -27.86 +15.43
Lower Mid section -48.76 -49.23 -54.67 +2.96
Berm Inner node -2.73 -33.05 -94.23 +45.75
Outer node -0.64 -2.06 -15.04 +7.20
U/S face -63.38 -64.75 -169.84 +53.23
Upper Mid section -42.16 -94.63 -111.53 +34.69
Berm Inner node +27.55 -0.76 -3.17 +15.36
Outer node -12.65 -65.98 -79.65 +33.50
U/S face +20.10 -25.67 -142.64 +81.37
Crest
Mid section +43.76 -80.87 -121.53 +82.65
Level
D/S face +9.65 -10.62 -107.86 +58.76
388
Table 4: Principal stresses and maximum shear stress obtained by 3D finite element based analysis
corresponding to the nonlinear material properties in reservoir full for a typical intermediate section
of the original embankment surrounding the Ash Pond.
Principal stresses(S1, S2 and S3)

∈x), y (∈
Strain along x (∈ ∈y) and z
and maximum shear stress (τmax),
Positions ∈z) directions
(∈
kN/m2
S1 S2 S3 τmax ∈x ∈y ∈z
+0.27e- +0.16e-
U/S face +9.07 -84.88 -123.00 -0.76e-04
+66.05 04 04
- - +0.17e-
Groun Core Heel -185.73 -0.21e-01 -0.12e-01
159.12 159.34 +13.31 02
d Level
- -
Core Toe -265.80 -0.67e-03 -0.16e-01 -0.17e-01
159.16 159.23 +53.32
D/S face +2.01 -5.07 -14.71 +8.36 -0.24e-17 -0.34e-06 -0.42e-06
+0.25e-
U/S face +2.75 -80.51 -128.78 -0.74e-03 -0.83e-03
+65.76 03
Mid +0.16e-
-57.63 -48.12 -153.87 -0.78e-02 -0.75e-02
Lower section +48.12 02
Berm +0.35e- +0.26e-
Inner node +2.40 -7.77 -24.86 -0.69e-03
+13.63 03 03
+0.57e- +0.59e-
Outer node -0.64 -5.52 -15.30 -0.13e-03
+7.33 04 04
U/S face -63.65 -93.78 -170.81 +53.58 -0.38e-03 -0.79e-03 -0.78e-03
Mid +0.14e- +0.15e-
-43.25 -94.37 -112.34 -0.24e-02
section +34.55 02 02
Upper
+0.37e-
Berm Inner node -12.65 -85.64 -107.52 -0.17e-03 -0.17e-03
+47.44 03
+0.65e-
Outer node +14.64 +0.36 -13.30 -0.35e-04 -0.32e-04
+13.97 04
+0.76e- +0.95e-
U/S face +12.07 -41.05 -130.00 -0.16e-02
+71.04 03 03
Crest Mid +0.72e- +0.75e-
+44.86 -67.75 -119.54 -0.17e-02
Level section +82.20 03 03
+0.57e- +0.40e-
D/S face +17.07 +7.00 -7.28 -0.21e-02
+12.18 03 03
L
R Y
K
P
O Q
X
N
Z
I
M
J
Figure 1: 3D ten nodded tetrahedral element used for this
study.
389
4m
(38.7m, 17.2m) A B (41.33m, 17.2m)
(41.5m, 16.06m)
(43.5m, 14.77m)
(44.5m, 13.35m)
D (65.8m, 11.2m)
17.2m
C (45.5m, 11.2m)
Pervious casing
Impervious
zone
core
Phreatic line
91m
Figure 2: Seepage line (ABCD) within the earthen dam body.
4m
U/s slope of 2.25H: 1V

3m
height
18m
Impervio 3m
Pervious
us core
casing zone
2m deep cut-off trench
Bottom width 91m
Figure 3: Geometrical cross section of the original embankment

surrounding the Ash Pond.

height
18m
Impervio
Pervious
us core
casing zone
85.5m total bottom width
Figure 4: Geometrical cross section of the original embankment section

modified with steeper upstream slope.

height
18m
Impervio
Pervious
us core
casing zone
18m deep
91m bottom width

section modified with deep cut-off trench.
390
50m width
height
18m
50m lower berm
Impervi
ous Pervious
core casing zone
Bottom width 108m

section modified with large lower berm.
LEGEND
Tension zone
Maximum
comp. zone
Seepage line
Figure 7: Maximum compressive and tensile principal stress zone obtained by 2D

finite element analysis for the original embankment section
corresponding to reservoir full condition.
LEGEND
Tension zone
Maximum comp. zone
Seepage line
Figure 8: Maximum compressive and tensile principal stress zone obtained by 2D

finite element analysis for the original embankment section
corresponding to draw-down condition.
391
LEGEND
Tension zone
Maximum comp. zone
Seepage line
Figure 9: Tension zone and maximum compression zone obtained by 2D finite

element analysis taking material nonlinear property for the original
embankment section surrounding the Ash Pond.
LEGEND
Plastic zone
Seepage line
Figure 10: Plastic zone within the dam body as obtained by 2D finite element
analysis taking material nonlinear property for the original
embankment section surrounding the Ash Pond.
LEGEND
Tension zone
Max. compression zone
Seepage line
Figure 11: Tension zone and maximum compression zone obtained by 3D finite
element analysis considering material nonlinear properties for a typical
intermediate section of the original embankment corresponding to the
reservoir full condition.
392
LEGEND
Plastic zone
Seepage line
Figure 12: Plastic zone obtained by 3D finite element analysis considering the material
nonlinear properties of a typical intermediate section for the original
embankment corresponding to the reservoir full condition.
393
Paper No.: 065
FUNDAMENTAL PERIOD OF VERTICALLY IRREGULAR FRAMES

– A MODIFICATION OVER THE CODAL STIPULATION
1 2
Bhaskar Ghosh and Subhashish Roy Chowdhury
1
Post-Graduate Student, Department of Civil Engineering, Jadavpur University, Kolkata – 700032
2
Assistant Professor, Department of Civil Engineering, Jadavpur University, Kolkata - 700032
Email: bgjuce@gmail.com, srceju@yahoo.com
ABSTRACT:
The determination of the fundamental period of vibration of a structure is essential to earthquake design.
Current design codes in India such as IS 1893 (Part 1): 2002 provide formulas for the approximate period of
earthquake-resistant building systems, which are dependent only on the height of the structure or number of
stories. Such a formulation is overly conservative and unable to account for structures with geometric
irregularities. This study investigates the fundamental periods of one different type of concrete
earthquake-resistant building structures – vertically irregular frames. The fundamental periods based on
vibration theory for each example were compared with empirical equations, including current code equations.
Based on the results obtained from Rayleigh equation, equations for the approximate fundamental periods are
put forth for those frames which take into account those vertical irregularities. Through statistical comparison, it
was found that a multi-variable power model which is able to account for irregularities resulted in a better fit to
the time period data than equations which were dependent on height only. The proposed equations were
validated through a comparison of available measured period data. These proposed equations will allow design
engineers to quickly and accurately estimate the fundamental period of moment resisting frame structures by
taking into account these types of vertical irregularities.
KEYWORDS: Fundamental period of vibration, Earthquake-resistant building frame, Vertically irregular

frames, Indian code IS 1893.
1. INTRODUCTION
The sophistication of the structural analysis affects both the detail of the analysis results and the design fee.
Simple methods may provide a reasonable representation of the likely seismic behaviour to enable rapid
assessment of the expected building performance. More complex methods, such as inelastic dynamic time
history analysis provide more information about the response, but take more time and computational cost to
perform properly. Engineers need conceptually simple methods for the following reasons:
(1) To design full structures
(2) To enable a rapid check of likely building performance
(3) To preliminary size members before some more sophisticated studies are undertaken.
The determination of the fundamental period of vibration of structures is essential to design and assessment
against earthquake loading. A reasonably accurate estimation of the fundamental period in such irregular
structures is necessary in both response-spectrum and static earthquake analysis of structures. An accurate
estimation would allow for an improved estimation of the global seismic demands on an irregular structure. As
such, the goal of this research is to investigate the accuracy of existing code-based equations for estimation of
the fundamental period of irregular building structures and provide suggestions in the form of new equations to
improve their accuracy.
2. SETBACK STRUCTURES
All MRF structures are modeled with either 20 stories, 10 stories, or 5 stories (N) and 5 or 10 bays (Nb). All
structures have a uniform storey height of 3.5m. The bays have a uniform spacing of 4m. A total of 66 MRF
structures are evaluated: 11 types of setback structures for each type. For each structure, the criterion for
irregularity is checked to ensure that an irregularity is present as defined by the code as described in Table 5 or
in Figure 4 of IS:1893 (Part 1)-2002, and none falls under the classification of extreme irregularity.
2.1. Analysis of Results
The fundamental periods for all MRFs using codal equations and analysed period are shown graphically in
Figure 1. For all MRFs the fundamental periods given by Rayleigh equation and analysed data range from
identical to differing by 77.72%. The largest differences are found when the level of irregularity is larger, with
an average of 23.87%. The smallest differences are found in the examples without irregularities. As the height
of the structure increases, results from analysed time period approach to those obtained by Rayleigh equation.
For example, five storey structures, the average difference is 23.87%, whereas for ten storey structures, the
average difference decreases to 19.88%.
Figure 1. Fundamental periods of MRFs as a function of height for Setback Structures
2.2. New Equation
Irregularities are incorporated by considering not only the height (H), but also the average height (Hav) of lateral
force resisting MRFs, as well as the depth and the average depth of lateral force resisting MRFs (D, Dav) in the
direction of earthquake forces. Each average value represents a weighted average value. By taking into account
more structural parameters than height or number of stories, these equations will be more accurate than the
current code equations. The proposed equation is developed by nonlinear regression analysis. The program
DataFit, Version 9.0.59 is used to perform the regression analysis. Several equation forms were investigated,
including power models of varying form, quadratic models, polynomial models, and linear models. The
395
equations were modified to account for structural parameters in varying ways, such as including height raised to
a power compared with the ratio of height over depth raised to a power. Based on extensive simulations and
statistical analyses of these options, an equation of the following form was found to match the data the best
statistically:
(1)
where a, b, c, d and e are regression variables. This form of the equation will be referred to as a 4-variable
power model. For a regular structure (Hav/H and Dav/D = 1) this equation takes the form of codal equation. As
such, the proposed 4-variable power model is a generalization of codal equation intended to be used for
structures with irregularities (Hav/H and/or Dav/D ≠1) as well as structures with no irregularities. A direct
comparison is made between Equation 1, a power model (T = aHb) and a linear model (T = aH) in order to
highlight the benefit of using an equation of this form over the forms currently used by the code.
Results are presented for the following regression analyses with following cases:
(1) Unconstrained regression analysis to determine a, b, c, d and e.
(2) Constrained regression analysis with b = 0.75. For a structure with no irregularity, Equation 1 would
simplify to an equation of similar form to codal equation, and b = 0.75 replicates the slope given in that
equation.
(3) Constrained regression analysis with b = 1.0, which for a structure with no irregularity would reduce to the
form of linear equation.
(4) Constrained regression analysis to determine a, with the values of b, c, d and e rounded to the nearest second
decimal from the unconstrained regression case 1.
(5) Unconstrained 1-variable nonlinear regression analysis to determine a & b.
(6) Unconstrained linear regression analysis to determine a.
For each regression, the standard deviation of the residuals (σ) and r-squared (R2) values are found. These
values indicate how well the regression equation fits the sample Rayleigh data. The results of the regression
analyses are given in Table 2.1.
Table 2.1. Regression analysis results for MRFs for Setback Structures
Regression Best-fit Regression
σ R2
Analysis Case Coefficients
a = 0.08323, b = 0.84725
Unconstrained c = 0.00993, d = 0.03795 0.0751 0.992013
e = 0.39742
Constrained a = 0.12195, c = -0.00856
4-Variable 0.1114 0.982127
with b = 0.75 d = 0.03864, e = 0.39728
Power
Constrained a = 0.04521, c = -0.01198
Model 0.1415 0.971158
with b=1.0 d = 0.0369, e = 0.39768
Constrained with
b = 0.85, c = 0.01, a = 0.07715 0.0742 0.991674
d = 0.05, e = 0.40
1-Variable
a = 0.075155
Power Unconstrained 0.1287 0.97536
b = 0.847976
Model
Linear
Unconstrained a = 0.04069 0.173 0.95478
Model
The standard deviation for the constrained 4-variable power (case 4), the 1-variable power (case 5), and the
linear model (case 6) are 0.074, 0.129 and 0.173, respectively. These measures indicate that case 4 provides the
best fit to the data. Case 4 corresponds to an R2 value of 0.992, meaning that 99.2% of the variation in the
396
fundamental period can be explained by this model. Thus, Equation 1 with parameters a = 0.07715, b = 0.85, c =
0.01, d = 0.05 and e = 0.40 is proposed as the best-fit equation for determining the approximate fundamental
period of MRFs:
(2)
Equation 2 represents the best-fit to the period data for MRFs. However, it does not take into account
non-structural elements that add to the stiffness of the structure and thus is a non-conservative estimate. As such,
the periods obtained by this equation represent an upper-bound to the permissible periods used in design as
stipulated by IS:1893 (Part 1)-2002. Therefore, a reduction of the best-fit equation is suggested by lowering
each best fit curve without changing its shape and general pattern by comparing with measured period data
compiled by Kwon and Kim (2010) for an assortment of structures in California without large irregularities.
In order to suggest a conservative approximation for the fundamental period, a reduced form of Equation 2 is
posed by reducing the equation by the standard deviation in order to obtain an equation which yields a
conservative estimate for 80-90% of the measured periods. Equation 2 resulted in a non-conservative estimate
for all measured periods, so an initial modification of minus 0.1 standard deviations (best-fit –0.1σ) is examined,
which resulted in a conservative estimation for 44 of 60 periods (73.3%). Finally, the best-fit –0.2σ was
compared with the measured periods, which resulted in a conservative estimate for 52 of 60 periods (86.67%).
This discussion is shown graphically in Figure 2.
As such, the following equation for the approximate fundamental period for MRFs is proposed for use in
practice:
(3)
Figure 2. Reduction of best-fit Equation 2
397
Comparison with measured data presented in this paper is limited to regular structures. Incorporating additional
structural parameters which account for structural irregularities is important to obtain reasonably accurate
results. Equation 2 with the irregular example structures shows the proposed equation incorporates the effect of
irregularity properly. The comparison of measured data with the results obtained from the proposed Equation 3
validates the merit of the proposed equation.
3. MASS IRREGULAR STRUCTURES
All MRF structures are modeled with either 20 stories, 15 stories, or 10 stories (N) and 5 or 10 bays (Nb). All
structures have a uniform storey height of 3.5m. The bays have a uniform spacing of 4m. A total of 168 MRF
structures are evaluated considering three different types of mass ratio as 1.25, 1.5 and 2. For each structure, the
criterion for irregularity is checked to ensure that an irregularity is present as defined by the code as described in
Table 5 or in Fig. 4 of IS:1893 (Part 1)-2002, and none falls under the classification of extreme irregularity.
From Codal Power Equation, we can easily calculate that the fundamental time period for heights 35m, 52.5m
and 70m is respectively 1.0792s, 1.4628s and 1.815s. And from Codal Linear Equation, we can calculate that
the fundamental time period for 10, 15 and 20 storey is respectively 1s, 1.5s and 2s. Whereas, from the data
obtained, it is clearly seen that Power Equation yields the most conservative estimate of the fundamental period
for all example MRFs having more than 35m height.
It has also been seen that for a fixed number of storey and Mass ratio, time period decreases with increase in bay
number i.e. width. For a fixed Mass ratio and width, time period increases with increase in no. of storey i.e.
height. For a fixed height and width, time period increases with increase in Mass ratio. And finally, for a fixed
height, width and mass ratio, time period increases as the height ratio (Hm/H) increases.
3.2. New Equation
A goal of this research is to develop simplified equation to allow design engineers to quickly and accurately
estimate the fundamental period of MRF structures by taking into account mass irregularity. This goal is
accomplished by considering not only the height (H), but also the height ratio (Hm/H) of lateral force resisting
MRFs, as well as the depth in the direction of earthquake forces and Mass ratio. By taking into account more
structural parameters than height or number of stories, these equations will be more accurate than the current
code equations. The proposed equation is developed by nonlinear regression analysis. The program DataFit,
Version 9.0.59 is used to perform the regression analysis. Several equation forms were investigated, including
power models of varying form, quadratic models, polynomial models, and linear models. Based on extensive
simulations and statistical analyses of these options, an equation of the following form was found to match the
data the best statistically:
(4)
where a, b, c, d and e are regression variables. For a regular structure (Hm/H and Mr = 1) this equation takes the
form of codal equation. As such, the proposed 4-variable power model is a generalization of codal equation
intended to be used for structures with irregularities (Hm/H and/or Mr ≠1) as well as structures with no
irregularities. A direct comparison is made between Equation 1, a power model (T = aHb) and a linear model (T
= aH) in order to highlight the benefit of using an equation of this form over the forms currently used by the
code. For each regression, the standard deviation of the residuals (σ) and r-squared (R2) values are found. The
results of the regression analyses are given in Table 3.1.
398
Table 3.1. Regression analysis results for MRFs for Mass Irregular Structures
Regression Best-fit Regression
σ R2
Analysis Case Coefficients
a = 0.0431, b = 1.047
Unconstrained c = 0.0064, d = 0.0044 0.0187 0.99962
e = 0.01278
Constrained a = 0.1432, c = -0.00536
4-Variable 0.2131 0.92866
with b = 0.75 d = 0.00577, e = 0.0129
Power
Constrained a = 0.0522, c = -0.00621
Model 0.0361 0.99796
with b=1.0 d = 0.00468, e = 0.01272
Constrained with
b = 1.05, c = 0.006, a = 0.04307 0.0186 0.99951
d = 0.004, e = 0.01
1-Variable
a = 0.0422844
b = 1.047155
Model
Linear
Unconstrained a = 0.05124 0.213 0.99767
Model
The standard deviation for the constrained 4-variable power model, the 1-variable power model and the linear
model are 0.0186, 0.0203 and 0.213 respectively. These measures indicate that case 4 provides the best fit to the
data. Case 4 corresponds to an R2 value of 0.99951, meaning that 99.951% of the variation in the fundamental
period can be explained by this model. Thus, with a = 0.04307, b = 1.05, c = 0.006, d = 0.004 and e = 0.01, the
best-fit equation becomes:
(5)
The periods obtained by the equation represent an upper-bound to the permissible periods used in design.
Therefore, a reduction of the best-fit equation is suggested by lowering each best fit curve without changing its
shape and general pattern. A reduction of 0.2 SD to “a” is proposed as done in 2.2 because no real data is found
regarding this type of irregularity that can validate Equation 5. As such, the following equation for the
approximate fundamental period for MRFs is proposed for use in practice:
(6)
4. SOFT-STOREY STRUCTURES
All MRF structures are modeled with either 20 stories or 10 stories (N) and 5 or 10 bays (Nb). All structures
have a uniform storey height of 3.5m. The bays have a uniform spacing of 4m. A total of 90 MRF structures are
evaluated considering three different types of column lengths as 4.5m, 5m and 5.5m making column-length
ratios as 0.778, 0.7 and 0.6364 respectively. For each structure, the criterion for irregularity is checked to ensure
that an irregularity is present as defined by the code as described in Table 5 or in Fig. 4 of IS:1893 (Part 1)-2002,
and none falls under the classification of extreme irregularity.
From Codal Power Equation, we can easily calculate that the fundamental time period for heights 35m and 70m
is 1.0792s and 1.815s. And from Codal Linear Equation, we can calculate that the fundamental time period for
10 and 20 storeys is 1s and 2s. Whereas, from the data obtained, it is clearly seen that Power Equation yields the
most conservative estimate of the fundamental period for all example MRFs having more than 35m height.
399
It has also been seen that for a fixed number of storey and Column-length ratio, time period decreases with
increase in bay number i.e. width. For a fixed Column-length ratio and width, time period increases with
increase in no. of storey i.e. height. For a fixed height and width, time period increases with increase in
Column-length ratio. And finally, for a fixed height, width and Column-length ratio, time period increases as the
height ratio (Hs/H) increases at first, and then it started to decrease gradually.
4.2. New Equation
It is tried to develop an equation to determine the time period of the structure having soft storey irregularity. The
proposed equation is developed by nonlinear regression analysis. Several equation forms were investigated,
including power models of varying form, quadratic models, polynomial models, and linear models. Based on
extensive simulations and statistical analyses, an equation of the following form was found to match the data the
best:
(7)
where a, b, c, d and e are regression variables. For a regular structure (Hs/H and L/Ls = 1) this equation takes the
form of codal equation. As such, the proposed 4-variable power model is a generalization of codal equation
intended to be used for structures with irregularities (Hs/H and/or L/Ls ≠1) as well as structures with no
irregularities. A direct comparison is made between this equation, a power model (T = aHb) and a linear model
(T = aH) in order to highlight the benefit of using an equation of this form over the forms currently used by the
code. For each regression, the standard deviation of the residuals (σ) and r-squared (R2) values are found. The
results of the regression analyses are given in Table 4.1.
Table 4.1. Regression analysis results for MRFs for Soft-Storey Structures
Regression Analysis Best-fit Regression
σ R2
Case Coefficients
a = 0.133, b = 0.743
Unconstrained c = -0.0254, d = -0.0167 0.0187 0.99962
e = -0.0866
Constrained a = 0.129, c = -0.0254
4-Variable 0.2131 0.92866
with b = 0.75 d = -0.0172, e = -0.0675
Power
Constrained a = 0.0458, c = -0.0265
Model 0.0361 0.99796
with b=1.0 d = -0.0172, e = -0.0675
Constrained with
b = 0.75, c = -0.025, a = 0.13 0.0186 0.99951
d = -0.015, e = -0.085
1-Variable
a = 0.1281
b = 0.7432
Model
Linear
Unconstrained a = 0.04372 0.213 0.99767
Model
The standard deviation for the constrained 4-variable power model, the 1-variable power model and the linear
model are 0.04923, 0.07281 and 0.17633 respectively. These measures indicate that case 4 provides the best fit
to the data. Case 4 corresponds to an R2 value of 0. 9927; meaning that 99.27% of the variation in the
fundamental period can be explained by this model. Thus, with a = 0.13, b = 0.75, c = - 0.025, d = - 0.015 and e
= - 0.085, the best-fit equation becomes:
(8)
400
The periods obtained by the equation represent an upper-bound to the permissible periods used in design.
Therefore, a reduction of the best-fit equation is suggested by lowering each best fit curve without changing its
shape and general pattern. A reduction of 0.2 SD to “a” is proposed as done in 2.2 as no real data is found
regarding this type of irregularity that can validate Equation 8. As such, the following equation for the
approximate fundamental period for MRFs is proposed for use in practice:
(9)
5. CONCLUSION
Time Period is a very important parameter that affects the estimated value of seismic load on structure and thus
accurate determination of Time Period is essential. In general, structures without irregularities tend to have a
longer period compared with irregular structures. Measured fundamental period data for irregular structures is
scarce. For the few that such data exist structural properties are not readily available. As such, the validation of
the proposed equations with measured periods were limited to the available data from instrumented structures,
of which did not include information on complete structural dimensions or existing irregularities. Therefore, the
proposed equations were compared with the measured fundamental periods of structures without irregularities.
As more measured data becomes available, further comparison should be conducted with refinement of the
proposed equations as needed.
6. NOMENCLATURE
Hav : Average height of lateral force resisting MRFs as weighted average

Dav : Average depth of lateral force resisting MRFs in the direction of earthquake forces as weighted average
σ : Standard Deviation
R2 : Correlation Ratio
Hm : Height of the heavily loaded floor from ground level in Mass Irregularity
Mr : Mass Ratio i.e. the ratio of heavily loaded floor weight and normal floor weight in Mass Irregularity
Hs : Height of the CG of columns those have longer lengths than other columns from ground level in Soft
Storey Irregularity
L : Length of columns i.e. 3.5m
Ls : Length of the longer columns in Soft Storey Irregularity
REFERENCES
[1] IS 1893 (Part 1): 2002 – Criteria for Earthquake Resistant Design of Structures, General Provisions and
Buildings (Fifth Revision).
[2] Wong, C.M. and Tso, W.K. (1994). Seismic loading for Buildings with setbacks. Canadian Journal of Civil
Engineering, 21(5), 863–871.
[3] Esteva, L. (1992). Non-Linear Seismic Response of Soft First Storey Buildings subjected to Narrow Band
Accelerograms. Earthquake Spectra, 8(3), 373–389.
[4] Goel, R.K. and Chopra, A.K. (1997). Period Formulas for Moment-Resisting Frame Buildings. Journal of
Structural Engineering, 123(11), 1454–1461.
[5] Gong, M.; Sun, J. and Xue, L. (2011). Empirical Formula of Fundamental Period for Steel Structure Based
on Seismic Response Record. International Conference on Civil Engineering, Lushan, China, 283–286.
401
[6] Chopra, A.K. (2000). Dynamics of structures: Theory and Applications to Earthquake Engineering, Prentice
Hall, New Jersey.
[7] Arnold, C. and Reitherman, R. (1982). Building configuration and seismic design, John Wiley & Sons, New
York.
[8] Paulay, T. and Priestley, M.J.N. (1992). Seismic design of reinforced concrete and masonry buildings, John
Wiley & Sons, New York.
402
Paper No.: 066
STRUCTURAL BEHAVIOR OF AN RC BUILDING ON SLOPING GROUND

UNDER EARTHQUAKE LOAD
Maganti Janardhana1 Jogi Naidu P.2 and Hymavathy K.2
1
Associate Professor, JNTUH College of Engineering, Hyderabad, India
2
Former Graduate Student, JNTUH College of Engineering, Hyderabad, India
ABSTRACT
The shape and proportions of the building have a major effect on the distribution of earthquake forces as they
work their way through the building. Earthquake resistant design of reinforced concrete buildings is a
continuing area of research. Geometric configurations, type of structural members, details of connections, and
materials of construction all have a profound effect on the structural-dynamic response of a building. In spite of
all the weaknesses in the structure, either code imperfections or error in analysis and design, the structural
configuration system has played a vital role in catastrophe.
The main objective of this paper is to study the behavior of a six storey building, with different degrees of
ground slope under earthquake load. From presented study, it is concluded that, stiffness of the building is
getting reduced when length of the column is higher, relative to the other extreme end. There is not much
influence of the ground inclination on the base shear irrespective of the both the methods of analyses considered
in the present study. But it is observed that there is a considerable difference in the lateral load distribution with
building height between ‘linear static method’ and ‘response spectrum method’, and hence it is advisable to
adopt ‘response spectrum method’ for building with sloping ground. Presence of ground slope is making the
structure to undergo severe torsion due to considerable variation in the height of ground floor columns. As the
height of the building increases with increase of ground slope, the time period also increases which leads to
increase in flexibility at one side of building (the side where column lengths are high). And it is also concluded
that, presence of ground slope is making one side of the building stiffer than the other side, which leads to
variation in bending moment due to short columns.
1. INTRODUCTION
This paper describes the behavior of a six story building symmetrical in plan ( Fig. 1), resting on different
degrees of sloping ground under earth quake load by adopting both Static method and Linear dynamic analysis
(Response spectrum method). Center line of the columns shown in figure 2. The ground is assumed to be sloped
only in one direction. Seven slope categories ranging from 0° to 30° are considered in this study, which
generally covers the sloped grounds encountered in practice. Elevation of the building with 30° with slope is
shown in figure 3. The building model in the study has six storeys with constant storey height of 5m. Number of
bays and the bay-width along two horizontal directions are kept constant for all the models for convenience.
Soil-structure interaction effects are not considered. The building is assumed to be located in Zone III on a site
with medium soil. The floor slabs are modelled as rigid diaphragms. The modeling and the analysis of the
building frames were carried out using commercial software ETABS.
The buildings were analysed using linear dynamic analysis (Response spectrum method) as per the IS 1893
(Part-1): 2002 loading requirement and compared with the linear static analysis. For each case, time period, base
shear, lateral load distribution, storey displacement and torsion are estimated and studied the effect of ground
slopes on the different.
Figure 1. Plan of typical building Figure 2. Centre line layout of the building
Figure 3. Elevation of the building with 30° ground slope with Tie beam
404
2. RESULT AND DISCUSSIONS
2.1 Time Period:
From the Table 1 and Figure 4, it is observed that a building on flat ground has same natural periods in
both horizontal directions. This was expected that the mass and stiffness distribution is similar in both the
horizontal direction. The Figure 4 also shows that as the ground slope (along X-direction) increases,
natural period in X- direction changing with the increased ground slope we can conclude that
the stiffness of the building is getting reduced where the length of the columns are high relative to the
other extreme end.
Table 1. Comparison of fundamental periods in two horizontal orthogonal directions
Slope of Ground
Tx (s) Ty (s)
(deg)
0 2.2835 2.2835
5 2.3259 2.3087
10 2.3803 2.3177
15 2.4441 2.3211
20 2.5169 2.3231
25 2.5554 2.2926
30 2.6264 2.2943
30 (with Tie Beam) 2.4684 2.2878
And it is also observed that when the ground slope is 30° using tie beams, the time period decreases in
both directions due to increase of stiffness of columns by adding tie beams.
2.7
2.6 FIRST MODE (Tx)

Time period (s)
2.5
2.4
2.3
2.2
2.1
0 5 10 15 20 25 30 30(tie
Slope of the ground(Degree) beam)
Figure 4. Variation of fundamental period with respect to the ground slope
405
2.2 Base Shear
Table 2. Base shear of buildings along X-direction
Computational
Ground Slope (deg)
VB (kN)
0 1329
5 1337
10 1341
15 1345
20 1350
25 1318
30 1323
30(with Tie beam) 1353
From the Table 2, it is observed that the base shear increases along the slope up to 20° and then it
decreases. And the building with slope 30° with tie beams the base shear is higher than the building with
slope 30° without tie beams.
2.3 Lateral load distribution
Table 3. Comparison of lateral load distribution with storey height by Linear Static method and
Response spectrum method (Empirical method)
Storey Lateral loads by linear Lateral loads by Response

Storey level No static method (kN) spectrum method (kN)
Terrace 7 478 173
Fifth 6 863 272
Fourth 5 1115 331
Third 4 1248 383
Second 3 1314 428
First 2 1327 458
Ground 1 1327 458
406
1400
Lateral load distribution
1200
with height 1000 Lateral load by
(kN) 800 Static method
600
Lateral load by
400 Response spectrum
200 method
0
7 6 5 4 3 2 1
Storey Number
Figure 5 Variation of Lateral load with slope using

linear static and Response spectrum
From the Table 3 and Figure 5, it is noticed that there is considerable difference in the lateral load
distribution with building height using linear static method and response spectrum method and therein lies
the advantage of dynamic analysis compared to linear static method.
Table 4 and Figure 6 shows the lateral load distribution with storey height for the buildings lying on 0° to
30° ground slope. From the Table 4 and graph 3, it can be of observe that the lateral loads are decreasing
as slope of the building increases. There is not much difference in lateral load distribution with sloping
ground. These differences are negligible.
Table 4. Comparison of Lateral load distribution with storey height for Different models with slope
variation
Lateral load distribution (kN)

Floor level Storey number Slope of the ground in Degrees
0° 5° 10° 15° 20° 25° 30°
Terrace 7 173 166 162 148 135 125 115
Fifth 6 272 263 263 248 237 227 216
Fourth 5 331 324 326 311 302 296 287
Third 4 383 378 381 366 356 352 342
Second 3 428 423 428 413 405 402 393
First 2 458 448 450 429 415 408 396
Ground 1 458 448 450 429 415 408 396
407
500.00
450.00
400.00
Lateral load distribution (kN)
350.00
0 Deg
300.00 5 Deg
250.00 10 Deg
200.00 15 Deg
20 Deg
150.00
25 Deg
100.00
30 Deg
50.00
0.00
1 2 3 4 5 6 7
Storey Number
Figure 6 Lateral load distribution with storey height for 0⁰ to 30⁰ ground slope
2.4 Storey Displacement

Table Error! No text of specified style in document.. Storey displacements for frame –4 for various
0° to 30° Ground
Storey displacement (mm)

Storey Storey Slope of the ground in Degrees
level no
30° (with
0° 5° 10° 15° 20° 25° 30° Tie Beam)
Terrace 7 91.72 94.97 95.73 98.52 101.80 102.280 108.70 106.85
Fifth 6 83.70 85.76 88.32 91.37 94.90 95.803 102.37 100.49
Fourth 5 69.76 72.28 75.27 78.72 82.63 84.219 90.89 88.95
Third 4 51.55 54.49 57.91 61.75 66.04 68.448 75.07 73.02
Second 3 31.03 34.20 37.92 42.03 46.61 49.818 56.14 53.96
First 2 10.89 13.92 17.62 21.79 26.45 30.329 36.08 33.78
Ground 1 0.46 2.77 6.67 11.46 16.97 21.920 28.17 26.14
Base 0 0.00 0.00 0.00 0.00 0.00 0.000 0 0
408
8
7
0 Deg
6
5 Deg
5
Storey Level
10 Deg
4 15 Deg
3 20 Deg
2 25 Deg
1 30 Deg
30 Deg with T.B

0
0 20 40 60 80 100 120
Storey displacement (mm)
Figure 7. Storey displacement for frame –4 for various 0° to 30° Ground
From Table 5, it is observed that Storey displacement is increasing as the slope of the building increases. It
is observed that slope of the ground will affect the Storey displacement. By observing the cases of building
with 30° slope and the building with 30° slope with tie beams, it is observed that by adding tie beams, the
storey displacement is decreasing. These results are also shown graphically (Fig. 7).
2.5 Modal Participating Mass Ratio
Table 6. Modal participating mass ratio (%)
Mode 1 Mode 2 Mode 3

Slope of
the
ground Ux Uy Uz Ux Uy Uz Ux Uy Uz
0 0.0363 75.7408 0 75.7409 0.0363 0 0 0 0
5 0 76.2251 0 76.3147 0 0 0 0.6153 0
10 0 75.921 0 76.4061 0 0 0 2.3278 0
15 0 75.4963 0 76.2911 0 0 0 4.2377 0
20 0 75.4373 0 76.0928 0 0 0 5.8119 0
25 0 75.6623 0 75.8773 0 0 0 6.9723 0
30 0 76.2266 0 75.6247 0 0 0 7.5985 0
30 (TB) 0 75.1026 0 73.8117 0 0 0 4.4972 0
409
Table 6 shows the modal participating mass ratio (%) for first three modes. This Table shows that the
Mode 1 and Mode 2 (translational modes) are dominant deformation modes. As the ground slope
inclination increases mass participation in Y – direction also increases and mass participation in X –
direction decreases in Mode 1. In Mode 2 as ground slope increases mass participation in X - direction is
increases up to 10° and from 15° onwards it is decreasing. Similarly is the trend in the Y – direction.
Figure 8 shows the variation of mass participation in the rotational mode as a function of building slope.
The figure clearly shows that the building model with zero slope has no mass participating in the rotational
mode of vibration. As the ground slope increases the mass participation in the rotational mode increases
proportionally.
8
Modal participating mass ratio
7
6
5
4
3
2
1
0
0 5 10 15 20 25 30 35
Building Slope (Degree)
Figure 8. Variation of mass participation in the rotational model
2.6 Torsion of the building
The Torsion of the building with ground slope 0⁰ to 30⁰ at bottom of each storey is shown in Table 7 and
represents graphically in Figure 9. It is observed that the torsion of the building decreases as the storey
level increases for all the cases with 0° to 30° ground slope. And also it is observed that in the ground floor
the torsion is increased as the slope of the ground increases. From this it is concluded that as the slope of
ground increases, the torsion will affect the building.
410
Table 7. Torsion at bottom of storey for buildings with ground slope 0° – 30°
Torsion (kNm)
Storey Story
Load Loc Slope of the ground in Degrees
number Level
0° 5° 10° 15° 20° 25° 30°
7 Terrace EQYP Bot 5525 5263 5019 4805 4603 4284 4140
1044 1006
6 Fifth EQYP Bot 8 7 9700 9373 9059 8520 8430
1361 1326 1291 1259 1228 1165 1170
5 Fourth EQYP Bot 1 7 5 4 0 3 2
1540 1518 1493 1470 1446 1384 1410
4 Third EQYP Bot 2 6 9 3 6 8 5
1620 1615 1604 1593 1581 1527 1579
3 Second EQYP Bot 9 1 5 5 8 2 5
1641 1648 1650 1651 1652 1608 1695
2 First EQYP Bot 7 4 6 9 9 6 8
1642 1650 1655 1660 1665 1625 1716
1 Ground EQYP Bot 1 7 8 6 6 1 5
1533 1374 1197 1014
0 Base-3 EQYP Bot 4 5 8 2 1083 6844
1224 1053
Base-2 EQYP Bot 0 8 8800 7074 544 4275
Base-1 EQYP Bot 6979 5664 4435 3344 233 1830
7 0 DEG
5 DEG
STOREY NUMBER
6
10 DEG
5
15 DEG
4
20 DEG
3
25 DEG
2 30 DEG
1
0
0 5000 10000 15000 20000
TORSION ( kNm)
Figure 9. Variation of Torsion for a storey with slope of the ground
411
2.7 Bending moment variation in columns
The bending moment for ground floor columns for ground slopes 0⁰ to 30⁰ are tabulated in Table 8. The
variation in bending moment for the critical columns C1 and C13 is shown in Figure 10.
Table 8. Bending moment for columns with 0° to 30° ground slope.
Moment, M3(kNm)
Story Column Slope of the ground in Degrees
0° 5° 10° 15° 20° 25° 30°
GROUND C1 5.158 21.048 42.348 58.667 71.411 80.086 88.532
GROUND C2 2.146 4.919 1.774 -0.145 0.53 2.889 6.224
GROUND C3 -2.146 9.263 9.703 10.866 12.69 14.169 15.766
GROUND C4 -5.158 -33.238 -41.949 -40.559 -36.5 -31.619 -27.316
GROUND C5 23.361 7.69 8.166 9.673 11.032 12.204 13.02
GROUND C6 23.361 7.689 8.166 9.673 11.031 12.204 13.019
GROUND C7 5.872 2.936 2.258 4.628 7.725 10.488 13.07
GROUND C8 5.872 2.936 2.259 4.628 7.725 10.488 13.07
GROUND C9 -5.873 6.19 8.93 11.566 15.245 18.898 23.047
GROUND C10 -5.873 6.19 8.93 11.566 15.245 18.898 23.046
GROUND C11 -23.361 -11.049 1.299 9.656 15.807 19.756 23.992
GROUND C12 -23.361 -11.049 1.299 9.656 15.806 19.756 23.992
GROUND C13 5.158 21.048 42.347 58.666 71.41 80.086 88.531
GROUND C14 2.146 4.919 1.774 -0.145 0.53 2.889 6.224
GROUND C15 -2.146 9.264 9.703 10.867 12.69 14.169 15.767
GROUND C16 -5.158 -33.238 -41.948 -40.558 -36.499 -31.618 -27.315
100
90
80
MOMENT (kNm)
70
60
50
40
30
20
10
0
0 5 10 15 20 25 30 35
SLOPE OF THE GROUND IN DEGREES
Figure 10. The variation in bending moment for C1 and C13 columns
412
From Table 8 and Figure 10 it is observed that the columns under Grid 1 are critical columns compared to
other grids. The increase in percentage of bending moment from 0⁰ ground slope to 30⁰ ground slope is
1616% for C1 and C13 columns. This is due to short column effect. Poor behaviour of short columns is
due to the fact that in an earthquake, a long column and a short column of same cross section move
horizontally by same amount. However, the short column is stiffer as compared to the long column, and it
attracts larger earthquake force.
If a short column is not adequately designed for such a large force, it can suffer significant damage during
an earthquake. This behaviour is called Short Column effect.
For the ground slope 0⁰ to 30⁰, the area of reinforcement for the ground floor columns are tabulated in
Table 9., and it is shown graphically in Figure 11.
Table 9. The Area of reinforcement for Ground floor columns
AREA OF REINFORCEMENT (mm2)

Col Slope of the ground in Degrees
Story
Line
30⁰⁰(With
0⁰⁰ 5⁰⁰ 10⁰⁰ 15⁰⁰ 20⁰⁰ 25⁰⁰ 30⁰⁰ Tie
beam)
Ground C1 3234 7280 9450 10296 10764 13160 16500 10608
Ground C2 5625 4787 4167 4117 4046 3800 3873 3601
Ground C3 5625 4096 4002 4199 4579 5750 8719 3661
Ground C4 3233 2880 2880 2880 5791 9231 13684 2880
Ground C5 5579 3901 4105 4992 7755 11531 17691 3527
Ground C6 5583 3900 4105 4992 7754 11530 17690 3527
Ground C7 5045 5031 4885 5118 5462 6627 9776 4358
Ground C8 5045 5031 4885 5118 5462 6627 9776 4358
Ground C9 5045 5068 5039 4954 4943 4506 4574 4442
Ground C10 5045 5068 5039 4954 4943 4506 4574 4442
Ground C11 5583 9355 11254 12184 12494 11872 11857 11153
Ground C12 5583 9355 11254 12184 12494 11872 11857 11153
Ground C13 3234 7280 9450 10517 11234 13160 16500 10959
Ground C14 5625 4787 4167 4117 4046 3800 3873 3601
Ground C15 5625 4096 4002 4199 4578 5749 8719 3661
Ground C16 3234 2880 2880 2880 5792 9231 13684 2880
As the bending moment in columns increases with the increase in slope of the ground, obviously the area
of reinforcement also increases with the increase in slope of the ground increases and is shown in
graphically by Figure 8. The increase in percentage of area of reinforcement in columns from 0⁰ ground
slope to 30⁰ ground slope is 410% for C1 and C13 columns. For ground slope 30⁰ (with tie beams), in the
columns it is observed that the area of reinforcement is decreasing compare to the columns of the building
resting on 30⁰ sloping ground.
413
18000
16000
Area of Reinforcement(mm2) 14000
12000
10000
8000
6000
4000
2000
0
0 5 10 15 20 25 30 35
Slope of the ground in Degrees
Figure 11. The variation in area of reinforcement for columns C1 and C13 columns.
3. CONCLUSIONS
The Following conclusions are made from the present study
1. Since the mass is not varying with the increased ground slope, it can be concluded that the stiffness of
the building is getting reduced where length of the columns is higher, relative to the other extreme end.
2. There is not much effect of the ground inclination on the base shear irrespective of both the methods of
analysis (linear static method and response spectrum method) considered in the present study. But it is
observed that there is a considerable variation in the distribution of storey shears. Hence it is advisable
to adopt response spectrum method for building with sloping ground.
3. Presence of ground-slope is making the structure to undergo severe torsion due to considerable
variation, in the height of ground floor columns.
4. As the height of the building increases with increase of ground slope, the time period also increases
which leads to increase in flexibility at one side of building (the side where the column lengths are
high).
5. Presence of ground-slope is making one side of the building stiffer than the other side, which leads to
variation in bending moment due to short column effect. The increase in percentage of bending
moment from 0⁰ ground slope to 30⁰ ground slope is 1616% for C1 and C13 columns. The increase in
percentage of area of reinforcement in columns from 0⁰ ground slope to 30⁰ ground slope is 410% for
C1 and C13 columns.
REFERENCES
[1] IS: 456-2000 “Indian standard Code of practice for Plain and Reinforced Concrete (Fourth revision)”,
Bureau of Indian Standards.
[2] IS 1893(Part 1): 2002 “Criteria for earthquake resistant design of structures”, Part1 General
Provisions and building, Bureau of Indian Standards.
[3] IS: 875(Part 3)-1987, “Indian standard code of practice for design loads (Other than earthquake) for
buildings and structures”, Part 3 Wind loads (Second Revision), Bureau of Indian Standards.
[4] Murty, C.V.R., IIT, Kanpur, BMTPC, New Delhi ‘Earthquake tips’ Feb 2004.
414
[5] A.S.Warudkar, R.P.Sudarsan and G.Visalakshi-“Effect of torsional stiffness on the
dynamic response of RC framed structures” Proceedings of the Tenth Symposium on Earthquake
Engineering ,Nov. 16-18, 1994, Roorkee.
[6] Ashok K. Jain and Satish Annigeri - “Torsional Provisions for Asymmetrical Multi-storey Buildings ”
Proceedings of the Tenth Symposium on Earthquake Engineering, Nov. 16-18 1994, Roorkee.
[7] Satish Kumar and D.K. Paul ``Dynamic Analysis of Step back and Setback Buildings``, Proceedings
of the Tenth Symposium on Earthquake Engineering, Nov. 16-18, 1994, Roorkee.
[8] Paulay,T., and Priestley, M.J.N., ‘Seismic design of reinforced concrete and masonry buildings’,
John-Wiley Publishers, 1992.
[9] Pankaj Agarwal and Manish Shrikhande, “Earthquake Resistant Design of Structures”, Prentice Hall
of India, 2006.
[10] S.K. Duggal, “Earthquake Resistant Design of Structures”, Oxford University Press, 2007.
[11] Bungale S. Taranath “Reinforced concrete design of tall buildings”, CRC Press, 2010.
[12] Bungale S. Taranath “Wind and Earthquake resistant buildings, Structural analysis and design”
Marcel Dekker, 2005
[13] ETABS Non linear version 9.0.4, Extended 3D Analysis of building systems, A product of Computers
and Structures, Inc., 1995 University Ave,email:info@csiberkely.com
[14] Jogi naidu.P “Influence of sloping ground on the seismic analysis of RC Buildings”, A dissertation
submitted in partial fulfilment of the requirements for the award of the degree of Master of
Technology in Structural engineering.
415
Paper No.: 067
THERMAL STABILITY ANALYSIS OF FUNCTIONALLY GRADED PANELS
1 2
V. R. Kar and S. K. Panda
1
Research Scholar, Dept. of Mechanical Engineering, National Institute of Technology, Rourkela, India
2
Assistant Professor, Dept. of Mechanical Engineering, National Institute of Technology, Rourkela, India
Email: visheshkar@gmail.com, call2subrat@gmail.com
ABSTRACT :
The stability behaviour of functionally graded panels under thermal loading is investigated in this article. In this
analysis, the effective material properties of the functionally graded panels have been considered as the
temperature dependent and the gradation is considered in the thickness direction according to a power-law
distribution of the volume fractions of metal/ceramic constituents. The material properties vary gradually from
metal at the bottom surface to ceramic at the top surface. A finite element model is proposed and developed in
ANSYS. The functionally graded panels are modelled based on first order shear deformation theory as in the
ANSYS. The solutions are obtained using ANSYS parametric design language code for the eigenvalue type of
buckling. In order to establish the accuracy of the developed model, a convergence test was made with different
mesh size and compared with the published literature. The study has been extended for different parameters to
prove the efficacy of the developed model for both temperature independent and dependent material properties.
KEYWORDS: Functionally graded material, Buckling, Finite element method, APDL.
1. INTRODUCTION
A new class of advanced composite material has been developed by the gradual variation of material properties
in one/more direction to achieve efficient material properties under elevated thermal environment, namely
functionally graded material (FGM). This material is such that the bottom surface is metal reach and the top face
is ceramic reach. It has great advantage over the existing composite materials to avoid delamination as well as
inter-laminar thermal stress concentration to overcome the lacuna in laminated structures. This is well known
that when the structural components are exposed to high temperature environment they lose their stable
configuration due to the in-plane thermal load and which induces the instability in the geometry of the structure.
Due to the tailor-made material properties of FGM, various researchers have investigated the buckling
behaviour using different classical and shear deformation theories. Some of the important contributions are
discussed here to make the article self-contained.
Thai and Choi [1] investigated buckling of functionally graded (FG) plates for different support conditions
using an efficient and simple refined theory. Ghannadpour et al. [2] applied a finite strip method for analysing
the buckling behaviour of rectangular FG plates under the three types of thermal loadings such as uniform,
linear and nonlinear temperature rise across the thickness. Zhao et al. [3] employed the first order shear
deformation theory (FSDT) in conjunction with the element-free kp-Ritz method to analyse mechanical and
thermal buckling of FG plates. Tung and Duc [4] presented buckling and post-buckling behaviours of FG plates
under in-plane edge compressive, thermal and combined loads based the classical plate theory by incorporating
von Karman type of kinematic nonlinearity and initial geometrical imperfection. Lee et al. [5] employed the
FSDT displacement filed using von Kármán-type strain nonlinearity by adopting the element-free kp-Ritz
method to investigate the post-buckling behaviour of FG plates under edge compression and temperature field.
Woo et al. [6] investigated the post-buckling behaviour of moderately thick FG flat and cylindrical panels under
axial compression using the von Karman theory for large transverse deflections and a HSDT for moderately
thick plates. Na and Kim [7],[8] performed three-dimensional thermal and thermo-mechanical buckling analysis
under uniform and non-uniform temperature rise by adopting an 18-node solid element to analyse more
accurately the variation of material properties and temperature field in the thickness direction.
It is clear that the modelling and analysis of FG panels in commercially available finite element (FE) tools
are not reported in literature in large scale. Today many industries prefer to analyse the structural behaviour
through the ANSYS because of its accuracy and less computational cost. The objective of the present work is to
analyse the buckling behaviour of FG flat panels for different volume fractions, the aspect ratios, the
side-to-thickness ratios. Finite element solutions are obtained by commercially available FE tool using ANSYS
parametric design language (APDL) codes. In the present study, an eight node serendipity shell element
(SHELL281), defined in the ANSYS element library is employed to discretise the present flat panel.
2. EFFECTIVE MATERIAL PROPERTIES OF FGM
Functionally graded materials are inhomogeneous with mechanical and thermal properties varying continuously
from the bottom to the top surface along with the thickness direction. The constituents used are metal/alloy and
ceramic with metal rich at the bottom and ceramic rich at the top surface of the panel. The effective material
properties of FGM can be achieved by mixture rule of composites.
P = Pc (T )Vc ( z ) + Pm (T )Vm ( z ) (1)
where, Pc (T ) and Pm (T ) are the temperature dependent material properties of ceramic and metal, respectively
and can be expressed as the nonlinear temperature function [9] as
P (T ) = P0 ( P−1T −1 + 1 + PT
1 + P2T + P3T )
2 3
(2)
The volume fraction of ceramic Vc ( z ) and of metal Vm ( z ) are the function of thickness coordinate (z)
follow the simple power-law distribution and expressed as
n
 z 1
Vc ( z ) =  + 
h 2
n
 z 1
Vm ( z ) = 1 − Vc ( z ) = 1 −  +  (3)
h 2
where, n is the power-law index, 0 ≤ n ≤ ∞. The variations of volume fractions of the ceramic and metal phase
through the dimensionless thickness (Z=z/h) are shown in the Figure 1 and Figure 2 for different values of
power law indices.
For a functionally gradient material with metal/ceramic constituent materials, the Young’s modulus
E (T , z ) and the coefficient of thermal expansion α (T , z ) can be expressed as
n
 z 1
E (T , z ) = {Ec (T , z ) − Em (T , z )}  +  + Em (T , z ) (4)
h 2
n
 z 1
α (T , z ) = {α c (T , z ) − α m (T , z )}  +  + α m (T , z ) (5)
h 2
where, subscripts c and m refer to the ceramic and metal properties, respectively. The Poisson’s ratio ν is
assumed to be constant throughout the thickness of the FG panel. The temperature dependent material properties
of ceramic and metal are shown in Table 1.
417
Table 1 Temperature-dependent material properties of ceramic and metal (Ref. [9])
Material Properties Po P-1 P1 P2 P3

E(Pa) 3.48E+11 0 -3.07E-04 2.16E-07 -8.95E-11
Si3N4
α (K )
-1
5.87E-06 0 9.10E-04 0 0
E (Pa) 1.23E+11 0 -4.59E-04 0 0
Ti–6Al–4V
α (K-1) 7.58E-06 0 6.64E-04 -3.15E-06 0
n = 0.2
1.0 n = 0.5 1.0
n=1
n=2
0.8 n = 10 0.8
Volume fraction of ceramic
Volume fraction of metal

0.6 0.6
0.4 0.4
0.2 0.2
n = 0.2
n = 0.5
n=1
0.0 0.0 n=2
n = 10
-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6
Non-dimensional thickness coordinate Non-dimensional thickness coordinate
Figure 1 Variations of volume fraction of ceramic Figure 2 Variations of volume fraction of metal
through non-dimensional thickness coordinate through non-dimensional thickness coordinate
3. FINITE ELEMENT MODELLING
In the present study, an eight node serendipity shell element (SHELL281), defined in the ANSYS environment
is used to discretize the flat panel. This element is suitable for analyzing thin to moderately-thick shell
structures. The element has six degrees of freedom at each node i.e. translations and rotations in the x, y and z
directions [10]. A flat FG panel of uniform thickness ‘h’ with rectangular base of sides a and b is considered in
the analysis as shown in the Figure 3.
Ceramic rich b
+h/2
-h/2
Metal rich
a
Figure 3 Geometry and dimension of FG flat panel
The kinematic model of any flat/curved panel is governed by the first order shear deformation theory in
ANSYS. The displacements u , v and w at any point along x , y and z directions are given by
u ( x, y, z ) = u0 ( x, y ) + zφx ( x, y )
v ( x, y, z ) = v0 ( x, y ) + zφ y ( x, y )
418
w ( x, y, z ) = w0 ( x, y ) + zφz ( x, y ) (6)
where, uo , vo and wo are the mid-plane displacements and φx , φ y and φz are the shear rotations with
respect to the corresponding axes.
4. CONVERGENCE AND VALIDATION
The buckling behavior of functionally graded shell panels are analyzed by ANSYS 12.0 using APDL code. The
free vibration solutions are obtained using. The following sets of constraints in the combinations of clamped (C),
simply-supported (S) and free (F) are used as tabulated in Table 2.
Table 2 Sets of support conditions

CCCC u0 = v0 = w0 = φx = φ y = φ z = 0 at x = 0, a and y = 0, b
v0 = w0 = φ y = φz = 0 at x = 0, a
SSSS
u0 = w0 = φx = φz = 0 at y = 0, b
v0 = w0 = φ y = φz = 0 at x = 0 , a
SCSC
u0 = v0 = w0 = φx = φ y = φ z = 0 at y = 0, b
CFFF u0 = v0 = w0 = φx = φ y = φ z = 0 at x = 0
The reference temperature of FG flat panel is assumed to be T0. The temperature is uniformly raised to a
final value T. A clamped FG square plate is analyzed under uniformly distributed thermal load. In present
problem, aluminum (Al) as metal at the bottom surface and alumina (Al2O3) as ceramic at the top surface have
been considered. The Young’s modulus and the coefficient of thermal expansion are 70 GPa and 23×10-6 K-1 for
aluminum and 380 GPa and 7.4×10-6 K-1 for alumina, respectively at 300K. The convergence test has been done
and plotted in Figure 4. It is observed that a (18×18) mesh is sufficient to give the responses.
In order to prove the efficacy of the present model, the above problem has been extended. The results are
computed for five different power-law indices (n = 0, 0.5, 1, 2 and 5) and two different thickness ratios (a/h = 50
and 100) and tabulated in Table 3. The geometrical and material properties are taken same as Ref. [3] for the
computation of results. It is observed that the percentage differences between the results are within 5%. This can
also be seen that the differences are decreases as the thickness ratio increases.
84.0
83.9
Critical buckling temperature rise
83.8 (Present) ∆ T cr=83.709 K

R ef. [3] ∆ T cr=82.357 K
83.7
83.6
83.5
10x10 12x12 14x14 16x16 18x18 20x20
Mesh size
Figure 4 Critical buckling temperature rise of clamped (Al/Al2O3) FG square plate (n=1)
419
In this section, parametric study has been performed for the buckling behaviour of FG panels under uniform
thermal loading. The FG panel comprised of silicon nitrated (Si3N4) as ceramic and titanium alloy (Ti–6Al–4V)
as metal alloy are considered throughout the section, unless otherwise specified. The Young’s modulus and
coefficient of thermal expansion are temperature dependent properties as mentioned in Table 1 and Poisson’s
ratio is taken constant as 0.28.
Table 4 shows the variation of critical buckling temperature rise of simply-supported square FG (Si3N4 / Ti–
6Al–4V) plate for four different thickness ratios (a/h = 10, 20, 50 and 100) and three power-law index values (n
= 0.5, 2 and 10). It is seen from the mentioned figure that the critical temperature decreases with increase in
thickness ratio values and increases when power-law index increases.
Table 5 shows the variation of critical buckling temperature rise of simply-supported FG (Si3N4 / Ti–6Al–4V)
plate (a/h=50) for four different aspect ratios (a/b = 1, 1.5, 2 and 2.5) and three power-law index values (n = 0.5,
2 and 10). It is observed that as aspect ratio increases, the critical buckling temperature also increases thickness
ratio values and increases when power-law index increases.
Table 6 shows the variation of critical buckling temperature rise of square FG (Si3N4 / Ti–6Al–4V) plate
(a/h=50) for four different support conditions (CCCC, SCSC, SSSS, CFFF) and three power-law index values
(n = 0.5, 2 and 10). It is observed that FG plate buckles at lower critical temperature at fully clamped (CCCC)
condition.
Table 4 Variation of critical buckling temperature rise with thickness ratios and power-law indices for
simply-supported square FG plate
Thickness ratio n
a/h 0.5 2 10
10 10634 11292 12947
20 2980 3170.5 3759.9
50 493.19 523.54 629.82
100 123.85 131.28 158.42
Table 5 Variation of critical buckling temperature rise with aspect ratios and power-law indices for
simply-supported FG plate (a/h=50)
Aspect ratio n
a/b 0.5 2 10
1 493.19 523.54 629.82
1.5 966.67 1025.9 1230.4
2 1467.2 1554.9 1861.4
2.5 2120.8 2243.6 2676.3
Table 6 Variation of critical buckling temperature rise of square FG plate for different support conditions
(a/h=50)
n
Support conditions
0.5 2 10
CCCC 170.78 180.62 218.07
SCSC 308.8 326.73 394.21
SSSS 493.19 523.54 629.82
CFFF 2312.7 2444.1 2945.3
420
6. CONCLUSIONS
In this present study the buckling behaviour of FG flat panels are examined and presented. The effective
material properties of FG panels are temperature dependent and graded in thickness direction according to a
volume fraction of power-law distribution. The convergence and comparison studies have been carried out with
the commercially available FE tool (ANSYS12.0) and the results are obtained using APDL coding are showing
good agreement with those reported in published literatures. The influences of different parameters such as
thickness ratio, power-law index, aspect ratio and support condition on the buckling behaviour of FG panel are
analysed. Based on the parametric study, it is concluded that the critical buckling temperature rise increases with
increase in power-law indices and aspect ratios. The thinner and fully clamped FG plat can buckle at
comparatively low temperature.
REFERENCES
[1] Thai, H. T. and Choi D. H. (2012). An efficient and simple refined theory for buckling analysis of
functionally graded plates. Applied Mathematical Modelling, 36, 1008–1022.
[2] Ghannadpour S. A. M.; Ovesy H. R. and Nassirnia M. (2012). Buckling analysis of functionally graded
plates under thermal loadings using the finite strip method. Computers and Structures, 108 (109), 93–99.
[3] Zhao X.; Lee Y. Y. and Liew K. M. (2009). Mechanical and thermal buckling analysis of functionally
graded plates. Composite Structures, 90, 161–171.
[4] Tung H. V. and Nguyen D. D. (2010). Nonlinear analysis of stability for functionally graded plates under
mechanical and thermal loads. Composite Structures, 92, 1184–1191.
[5] Lee Y. Y.; Zhao X. and Reddy J.N. (2010). Post-buckling analysis of functionally graded plates subject to
compressive and thermal loads. Computer Methods in Applied Mechanics and Engineering, 199, 1645–1653.
[6] Woo J.; Meguid S. A.; Stranart J. C. and Liew K. M. (2005). Thermomechanical postbuckling analysis of
moderately thick functionally graded plates and shallow shells. International Journal of Mechanical Sciences,
47, 1147–1171.
[7] Na K. S. and Kim J. H. (2004). Three-dimensional thermal buckling analysis of functionally graded
materials. Composites: Part B, 35, 429–437.
[8] Na K. S. and Kim J. H. (2006). Three-dimensional thermomechanical buckling analysis for functionally
graded composite plates. Composite Structures, 73, 413–422.
[9] Reddy J. N. and Chin C. D. (1998). Thermomechanical analysis of functionally graded cylinders and plates.
J Therm Stresses, 21, 593–626.
[10] ANSYS Inc. (2009). ANSYS 12.0 Reference Manual.
421
Paper No.: 068
PROPERTIES OF RUBBERISED FLY ASH CONCRETE
M. Mishra 1, K. C. Panda 2
1
M. Tech. Student, ITER, SOA University, Bhubaneswar, 751030, Odisha, India
2
Associate profesor, Department of Civil Engineering, ITER, SOA University, Bhubaneswar, 751030, Odisha,
India
ABSTRACT:
This paper presents the results of an experimental investigation to study the effects of partial replacement of
cement with fly ash (FA) in rubberised concrete. The percentage of rubber used in this study was 0%, 5% and 10%
replaced with coarse aggregate and 0% and 5% of FA replaced with cement in conventional concrete. Two size
of rubber chips are used is the present study i.e. 5 mm and 10 mm. 40% of rubber used from 5 mm size and 60%
of rubber used from 10 mm size. The mix design was targeted for M30 grade of concrete. The mix proportion of
the concrete was 1:1.03:2.37 with water to cement ratio of 0.375. The fresh and hardened properties of
rubberised fly ash concrete were compared to those of conventional concrete (CC) without rubber and FA and
without rubber and 5% fly ash. The test results indicate that there is a small reduction in the strength with 5%
increase in rubber content as compared with the CC specimen without rubber and FA. However, the addition of
5% FA in rubberised concrete containing 5% and 10% of rubber chips shows better performance than
corresponding 5% and 10% rubber chips and without FA.
KEYWORDS: Tyre rubber, fly ash (FA), rubberised concrete (RC), compressive strength, flexural strength,
split tensile strength
1. INTRODUCTION
One of the most crucial environmental issues all around the world is the disposal of the waste materials. In
recent decades, worldwide growth of automobile industry and increasing use of car as the main means of
transport have tremendously boosted tyre production. This has generated massive stockpiles of used tyres. The
waste rubber is not easily biodegradable even after a long-period of landfill treatment. It can be used as fuel
material or as raw materials of rubber goods. Unfortunately, not much attention has been paid to use the waste
tyres in portland cement concrete mixtures, particularly for highway use. Limited work was done by researchers
to investigate the potential use of rubber tyres in conventional concrete mixtures {1}. Rubber from scrap tyres is
one of the most recent waste materials investigated for its potential use in the construction field. In general, it is
reported that a reduction in mechanical properties of concrete occurs when the rubber content increases {2}. The
strength and toughness properties of the concrete mixtures containing two types of tyre rubber were
investigated. The aggregate of the control mix was replaced by different amounts of tyre rubber particles of
several sizes. The control mix had a 28-day compressive strength of 35 MPa. Their results indicated that there
was about 85% reduction in the compressive strength when the coarse aggregate was fully replaced by rubber.
However, a smaller reduction in the compressive strength (65%) was observed when sand was fully replaced by
crumb rubber. Concrete containing rubber did not exhibit brittle failure under compression or tension. It was
also concluded that an optimized mix design was needed to provide the optimum tyre rubber content in the
concrete mixture {3}. The workability study has conducted by using two kinds of rubber i.e. crumb and chipped
rubber in preparation of mix {4}. It was found that there was a decrease in slump with increase in rubber content
and mixtures made with fine crumb rubber were more workable than those with coarse tyre chips or a
combination of tyre chips and crumb rubber. {5} investigated on the influence of the rubber content on fresh and
mechanical properties of rubberised concrete starting with the 0% rubber content (i.e., without rubber) and up to
12% rubber content in the M20 grade concrete. With a partial replacement of the coarse aggregate by 3%, 6%, 9%
and 12% of rubber, the compressive strength decreased as the percentage of replacement increased, but rubber
in 3% replacement concrete developed slightly higher compressive strength than those of without rubber
concrete. {6} determined the hardened properties of concrete using different types of tyre rubber particle as a
replacement of aggregate in concrete. The different types of rubber particles used were tyre chips, crumb rubber
and combination of tyre chips and crumb rubber. These particles were used to replace 12.5%, 25%, 37.5%, and
50% of the total mineral aggregate by volume. The results showed that the fresh rubberised concrete had lower
unit weight and workability compared to plain concrete. This study investigates the influence of FA on
mechanical properties of rubberised concrete.
The materials used in this study such as cement, sand, natural coarse aggregate, waste tyre rubber chips and FA.
All the materials are tested as per the specification of Indian Standards. Concrete mix design is carried out for
the preparation of test samples (cubes, cylinders and prisms) as per the recommended guidelines for concrete
mix design IS: 10262-1982 {7}. The Mix Design is targeted for M30 Grade of concrete. Then the concrete mix
was prepared for CC with water cement ratio (W/C) 0.375. Finally the compressive strength, flexural strength
and split tensile strength tests have been conducted.
2.1 Materials used and Properties
In this experimental study, Ordinary Portland Cement 43 grade (OPC), zone I fine aggregate, natural coarse
aggregate (20 mm passing), waste tyre rubber (5 mm and 10 mm), tap water, FA (F class) were used. The rubber
is replaced with coarse aggregate i.e. 5% and 10% of its weight and in rubberised concrete FA is replaced with
cement by 5% of its weight. The chemical composition of FA is presented in Table 1. The properties of fine
aggregate and coarse aggregate are presented in the Table 2. The mix proportion of concrete (as per mix design)
1: 1.03: 2.37, W/C ratio 0.375. The details of concrete mix proportion and identification are given in Table 3.
The details of mix quantity per m3 of concrete are given in Table 4. The rubber chips and FA sample is shown in
Figure 1 and Figure 2 respectively.
Table 1: Chemical composition of FA

Oxides (%) Average Oxides (%) Average
SiO2 58.13 Na2O 0.05
Al2O3 31.00 K 2O 0.90
Fe2O3 4.10 SO3 0.12
CaO 0.60 P2O5 0.39
MgO 0.10 CI 0.011
LOI 2.90 TiO2 1.63
Table 2: Properties of fine and coarse aggregates

Fine aggregate value Coarse aggregate value
obtained obtained
Characteristics
experimentally as Per experimentally as per
IS 383-1970 IS :383-1970
Abrasion value (%) - 27.02
Bulk density(kg/m3) 1568 1418
Crushing value (%) - 28.70
Fineness modulus 3.48 (Zone1) 7.95
Impact value (%) - 24.00
Specific gravity 2.63 2.84
Water absorption (%) 0.30 0.10
423
Table 3: Details of concrete mix proportion along with identification
Concrete Mix Proportion Mix Identity
C Concrete mix with 100% NCA + 100% cement + 0% Rubber chips + 0% Fly Ash CR0F0
Table 4: Details of mix quantity per m3 of concrete
Mix quantity of different constituents per m3 of concrete

Mix Rubber chips Fly
Cement Sand NCA
Identity (kg) Ash Water (kg)
(kg) (kg) (kg)
5 mm 10 mm (kg)
CR0F0 511.00 527.00 1209.27 - - - 191.60
CR0F5 485.45 527.00 1209.27 - - 25.55 191.60
CR5F0 511.00 527.00 1148.81 24.18 36.28 - 191.60
CR5F5 485.45 527.00 1148.81 24.18 36.28 25.55 191.60
CR10F0 511.00 527.00 1088.34 48.37 72.56 - 191.60
CR10F5 485.45 527.00 1148.81 48.37 72.56 25.55 191.60
Figure 1 5 mm and 10 mm rubber chips samples
Figure 2 FA samples
424
2.2 Test Program
The workability of fresh concrete mix is measured by slump cone test and to get the strength properties of
hardened concrete specimens, the compressive strength, flexural strength and split tensile strength of the
concrete specimens were tested after 7, 28 and 90 days. The concrete cubes of size 150 x 150 x 150 mm were
tested for compressive strength, 100 x 200 mm cylinders for split tensile strength and 100 x 100 x 500 mm
prisms for flexural strength. The setup for compressive, flexural and split tensile strength is shown in Figure 3.
The compressive strength, flexural strength and split tensile strength were computed using the expressions
such as for cube, fck = P/B2, for prism, fb = PL/bd2 and for cylinder, fsp = 2P/ ᴨLD respectively. Where, fck is
compressive strength in MPa, P is Maximum applied load in Newton, B is size of the cube specimen in mm, fb
is flexural strength in MPa, b is width of the specimen in mm, d is depth of the specimen in mm, fsp is the split
tensile strength in MPa, L is length of the specimen in mm and D is Diameter of the cylinder in mm.
Figure 3 Test setup for compression, split tensile and flexural strength of concrete specimen
3.1 Fresh Concrete Results
Slump test are conducted to know the workability of fresh concrete mix. The slump test results for this study are
shown in Figure 4. It is observed from the figure that, when FA replaced 5% with cement in CC the slump value
increases as compared with normal CC. The slump value of CC is 36 mm whereas in 5% FA based CC is 38 mm.
In 5 and 10% rubberised concrete with 5% FA having more slump values than 5 and 10% rubberised concrete
without FA. It indicates that FA improves the workability of rubberised concrete.
Figure 4 Slump value for rubberised concrete with and without FA
425
3.2 Hardened Concrete Results
The characteristics of the hardened concrete specimen are obtained by testing the specimens at the time
specified that is after 7 days, 28 days, 90 days. After casting the specimens (cube, cylinder and prism) the curing
has been done properly to achieve the target strength. Compressive strength, flexural strength and split tensile
strength tests were conducted to know the hardened properties of the specimen.
3.2.1 Compressive strength
The summary of compressive strength test results of CC and rubberised concrete with and without FA is given
in Table 5. The compressive strength versus age and the compressive strength versus mix type are shown in
Figure 5 and 6 respectively. From Table 5 it is observed that, as curing period increases the compressive strength
of concrete also increases at all age in all concrete specimens. As rubber particles concerned, a systematic
reduction in compressive strength observed with the increase of rubber content. The addition of 5% FA in CC,
gives more strength than all the concrete mix at all age. The rubberised concrete shows less strength as
compared with CC. When 5% FA added in rubberised concrete the strength increases as compared with the
CR5F0 and CR10F0. The 7, 28 and 90 days compressive strength results of rubberised concrete containing 5%
rubber and 0% FA (CR5F0) are 40.71, 48.54 and 55.35 respectively whereas the results for concrete containing
5% rubber and 5% FA (CR5F5) are 41.75, 49.52 and 56.15 respectively. From figure 5 and 6 it is observed that,
the FA replacement with cement in rubberised concrete increases the strength at all age. As 10% rubber is
concerned, the gain in strength of CR10F5 was more as compared with CR10F0. It is observed from the results,
FA based rubberised concrete gives better performance.
Table 5 Summary of compressive strength test results

7 Days 28 Days 90 Days
Average % change in Average % change Average % change
Mix compressive strength compressive in strength compressive in strength
Identification strength w.r.t control strength w.r.t control strength w.r.t control
(MPa) specimen (MPa) specimen (MPa) specimen
(CR0F0) (CR0F0) (CR0F0)
CR0F0 42.30 0 56.25 0 65.88 0
CR0F5 43.15 2.00 57.85 2.84 67.95 3.14
CR5F0 40.71 -3.76 48.54 -13.71 55.35 -15.98
CR5F5 41.75 -1.30 49.52 -11.96 56.15 -14.77
CR10F0 29.60 -30.02 34.52 -38.63 38.54 -41.50
CR10F5 34.70 -17.97 38.75 -31.11 46.35 -29.65
Figure 5 Compressive strength versus age Figure 6 Compressive strength versus mix type
426
3.2.2 Flexural strength
The summary of flexural strength test results of CC and rubberised concrete with and without FA is given in
Table 6. The flexural strength versus age and the flexural strength versus mix type are shown in Figure 7 and 8
respectively. From Table 6 it is observed that, as curing period increases the flexural strength of concrete also
increases at all age in all concrete specimens. The flexural strength of rubberised concrete decreased with the
increase of the rubber content in similar to that observed in compressive strength. The addition of 5% FA in CC
(CR0F5), gives maximum strength than all the concrete mixes at all age. When 5% FA is replaced with cement
in rubberised concrete it is observed that, FA helps to enhance flexural strength in rubberised concrete. The
flexural strength of rubberised concrete with 10% rubber content (CR10F0) are 4.35, 5.78 and 6.43 at 7, 28 and
90 days respectively, whereas 5% FA replacement in rubberised concrete with 10% rubber content (CR10F5) the
flexural strength are 4.60, 5.92 and 6.65 respectively. It can be concluded that, FA helps to increase the strength
at all age. From figure 7 and 8 it is observed that, the gain in strength is more from 7 to 28 days as compared
with 28-90 days for all concrete mixes.
Table 6 Summary of flexural strength test results

Average % change in Average % change in Average % change in
Mix Flexural strength Flexural strength Flexural strength
(CR0F0) (CR0F0) (CR0F0)
CR0F0 4.70 0 6.25 0 6.80 0
CR0F5 4.78 1.70 6.38 2.08 6.98 2.65
CR5F0 4.55 -3.19 6.20 -0.8 6.71 -1.32
CR5F5 4.60 -2.13 6.22 -0.48 6.75 -0.74
CR10F0 4.35 -7.45 5.78 -7.52 6.43 -5.44
CR10F5 4.60 -2.13 5.92 -5.28 6.65 -2.21
Figure 7 Flexural strength versus age Figure 8 Flexural versus mix type
3.2.3 Split tensile strength
The summary of split strength test results of CC and rubberised concrete with and without FA is given in Table 7.
The split tensile strength versus age and the split tensile strength versus mix type are shown in Figure 9 and 10
respectively. From Table 7 it is observed that, as curing period increases the split tensile strength of concrete
increases at all age in all concrete specimens. The split tensile strength of rubberised concrete decreased with the
427
increase of the rubber content. The addition of 5% FA in CC (CR0F5) gives maximum strength than all other
concrete mixes at all age. In the given mix when rubber replacement percentage increases as compared with
normal CC the strength decreases. When the 5% FA is replaced with cement in rubberised concrete it is
observed that, the FA helps to gain the strength in rubberised concrete. The split tensile strength of rubberised
concrete with 5% rubber (CR5F0) are 3.21, 3.88 and 4.45 at 7, 28 and 90 days respectively, whereas 5% FA
replacement with cement in 5% rubberised concrete (CR5F5) the split tensile strength value are 3.25, 3.92 and
4.51 respectively. It can be concluded that, FA helps to increase the split tensile strength in rubberised concrete.
From figure 9 and 10 it is observed that as age of curing increases the split tensile strength increases. However,
the gain is strength is more from 7 to 28 days as compared to 28 to 90 days.
Table 7 Summary of split tensile strength test results

Average % change Average % change in Average % change
Mix split tensile in strength split tensile strength split tensile in strength
(CR0F0) (CR0F0) (CR0F0)
CR0F0 3.41 0 4.35 0 4.85 0
CR0F5 3.45 1.17 4.44 2.07 4.97 2.47
CR5F0 3.21 -5.87 3.88 -10.80 4.45 -8.25
CR5F5 3.25 -4.69 3.92 -9.89 4.51 -7.00
CR10F0 2.84 -16.72 3.53 -18.85 3.98 -17.94
CR10F5 2.98 -12.61 3.78 -13.10 4.25 -12.37
Figure 9 Split tensile strength versus age Figure 10 Split tensile versus mix type
3.2.4 Modes of failure
Figure 11 shows the failure of specimens after strength test. In rubberised concrete specimens, the cracks
developed at the interface of cement paste and rubber due to weak bonding between rubber and cement paste.
As the rubber particles are much softer than the surrounding cement paste, on loading, cracks are initiated
quickly around the rubber particles, which accelerate the failure of the rubber cement matrix. Also due to the
lack of adhesion between rubber particles and paste, soft rubber particles may be viewed as voids in the concrete
matrix. It was observed in all tests that as rubber content increases, samples tend to fail gradually. The sample
sustained a much higher deformation than control mix. The nature of failure is ductile due to presence of rubber
chips. The rubber chip in rubberised concrete does not allow brittle failure in concrete specimens.
428
Figure 11 Failures in cube, cylinder, and prism specimen after testing
4. CONCLUSIONS
The slight improvement of workability is observed after addition of FA in rubberised concrete.

As age of curing increases the compressive, flexural and split tensile strength of concrete increases.
The compressive, flexural and split tensile strength of rubberized concrete decreases as rubber content
increases.
FA based rubberised concrete has more compressive, flexural and split tensile strength than rubberised
concrete at all age.
In rubberised concrete with 5% and 10% rubber gives less strength than the normal concrete without any
replacement of rubber and FA.
As 5% of FA is added in CC the compressive, flexural and split tensile strength are more than control
specimen at all age.
Due to the presence of rubber chips in rubberised concrete, it does not allow brittle type of failure.
REFERENCES
[1] Guneyisi, E.; Gesoglu, M. and Ozturan, T. (2004). Properties of rubberised concretes containing silica fume.
Cement and Concrete Research, 34, 2309-2317.
[2] Uygunoglu, T. and Topcu, I. B. (2010). The role of scrap rubber particles on the drying shrinkage and
mechanical properties of self-consolidating mortars. Construction and Building Materials, 24, 1141-1150.
[3] Eldin, N. N. and Senouci, A. B. (1993). Rubber tyre particles as concrete aggregate. Journal of Material in
Civil Engineering, ASCE 5 (4), 478– 496.
[4] Khatib, Z. K. and Bayomy, F. M. (1999). Rubberised Portland cement concrete. Journal of Materials in
Civil Engineering, 11(3), 206-213.
[5] Panda, K. C.; Parhi, P. S. and Jena, T. (2012). Scrap-Tyre-Rubber Replacement for Aggregate in Cement
Concrete - An experimental Study. International Journal of Earth Science and Engineering, Vol-05 No 06
(01), 1692-1701.
[6] Khaloo, A. R.; Dehestani, M. and Rahamatabadi, P. (2008). Mechanical properties of concrete containing a
high volume of tyre-rubber particles. Waste Management, 28, 2472-2482.
[7] IS: 10262 (1982). Recommended Guidelines for Concrete Mix Design. Bureau of Indian Standards, New
Delhi, India.
429
Paper No.: 073
INFLUENCE OF OPENING ON THE DYNAMIC STRUCTURAL RESPONSE

OF RECTANGULAR SLENDER SHEAR WALL FOR DIFFERENT DAMPING
RATIOS
1 2
G Muthukumar and Manoj Kumar
1
Lecturer, Dept. of Civil Engineering, Birla Institute of Technology & Science, Pilani, India
2
Associate Professor, Dept. of Civil Engineering, Birla Institute of Technology & Science, Pilani, India
Email: muthug@pilani.bits-pilani.ac.in, manojkr@pilani.bits-pilani.ac.in
ABSTRACT :
Shear wall has been considered to be major lateral load resisting element in multi storied building located in
wind or earthquake prone zone. The behavior of shear wall under various loading conditions has been the
subject of intense research in the last few decades because of its popularity in various countries across the globe.
The behavior of shear walls without openings is completely well understood and well documented in literature.
Nevertheless, due to functional requirement, shear walls are usually pierced with numerous openings of various
sizes to accommodate doors and windows. Even though, it is intuitively known that size of opening has
significant effects on the behavior of shear wall, it is desirable to know the limiting size of opening in shear wall,
beyond which the shear walls may become unserviceable. In order to predict the behavior of shear wall, the
shear wall is subjected time varying ground motion and linear and non-linear finite element dynamic analyses
have been performed using Newmark - Beta method. The shear walls are analyzed for different damping ratios
ranging from 2% to as high as 10%. The displacement response history has been plotted for all cases. The
influence of damping is also examined in this study.
KEYWORDS: slender, shear wall, openings, structural response, damping.
1. INTRODUCTION
The shear walls form the potent lateral load resisting system in tall building design. The shear walls are
classified on the basis of aspect ratio namely short, squat and slender walls. The short shear walls has been the
thing of the past when actually the name shear wall arose in the construction world. Huge concrete walls with
aspect ratio much lesser than one were constructed to withstand the blast loading in United States. Since such
walls mostly deforms in shear, the name shear wall remains valid even today. Most of the shear walls that are
being in use now are of squat and slender type. Nevertheless, the medium rise shear wall (squat and little slender)
with aspect ratio between 2 & 4 are quite popular in the United States [Paulay and Priestley, 1992] where the
non-linear flexural and shear deformations significantly contribute to the lateral response of shear wall. Now
with huge scarcity of land, building goes vertical day by day, people get used to see more and more slender
shear walls being prominently constructed all over the universe. India is no different from that. Nevertheless, it
is suggested to restrict the aspect ratio of shear wall to seven, beyond which the structures behaves with
inadequate stiffness, problems in anchoring the tension side of the shear wall and potentially may undergo large
displacement (P-delta) effects. There are many modeling techniques developed for analyzing shear wall and
invariably with many limitations. The fiber element model, which accounts for axial-flexural-shear effects in
nonlinear range, have not been found suitable for low-rise shear wall. Hence, the most powerful of all methods
is Finite Element Method in spite of its computationally demanding effort. Many research investigations have
been performed in the last few decades on the behavior of such shear walls [Thomsen and Wallace, 2004].
Nevertheless, few investigations have been performed on the behaviour of shear walls with openings. Moreover,
very few investigations are concerned on the exact limiting size of openings in the shear wall. This study
investigates the effect of opening size on the response of shear wall. Since rectangular shear walls are more
common and practicable, only prismatic cross-section rectangular shear wall without boundary elements is
considered. The structural design needs to account for the functional requirement of the shear wall. As the
requirement may be, shear walls are penetrated by numerous openings of varying size to accommodate doors
and windows. Earlier investigations restricted the size of opening to a maximum of 15 % only from the analysis
point of view and not from the structural point of view. Moreover, irrespective of the aspect ratio, the size of the
opening is roughly restricted to 15 %. In this study, a scientific approach has been followed using finite element
analysis to determine the effect of openings on the displacement response of shear wall. The opening size is
varied from 3.5% to as high as 21% but kept it in a symmetric fashion. Usually, structural engineer is
considered to be lucky if the openings are provided in the symmetric fashion. It has been observed that the
opening near the wall boundary impacts both shear and flexural strengths, depending on the size of the opening
[Taylor et al. 1988]. On the other hand, when the opening is demanded in the zigzag fashion, it may not be
normally possible to provide without being located proximal to the boundaries, and is considered to unsafe,
especially at the base region where the potential plastic hinge forms. Hence, it is always safe to avoid zigzag
openings especially at the base region. Slender shear wall enjoys the favourite tag of having inherent ductility as
confirmed through several investigations. The openings may certainly act as a detrimental element to the overall
behaviour if the size exceeds the threshold limit. On the other hand, openings in the shear wall may result in the
stable ductile behaviour for certain percentage of opening sizes. Strut-tie-model has been used in the past to
examine the behaviour of shear wall with openings and proved to be very useful when there is a discontinuity in
the region [Yanez et al. 1992]. However, with the advent in computing facilities it allows us to perform the more
reliable finite element analysis. In this study, a concerted effort has been put up in determining the optimum size
of opening in the slender shear walls without boundary elements using finite element analysis. The next section
briefly describes the finite element geometric modeling.
2. GEOMETRIC MODELING
The use of shell element to model moderately thick structures like shear wall is very well documented in the
literature. Nevertheless, the general shell theory based on the classical approach has been found to be complex
in the finite element formulation. On the other hand, the degenerated shell element [Ahmad et al. 1970] derived
from the three-dimensional element, has been quite successful in modeling moderately thick structures because
of their simplicity and circumvents the use of classical shell theory. However, when the thickness of element
reduces, degenerated shell element has suffered from shear locking and membrane locking when subjected to
full numerical integration. The shear locking and membrane locking are the parasitic shear stresses and
membrane stresses present in the finite element solution. In order to alleviate locking problems, the reduced
integration technique has been suggested and adopted by many authors [Zeinkiewicz et al. 1971; Paswey &
Clough, 1971]. The use of reduced integration may result in spurious mechanisms or zero energy modes. The
selective integration has also been adopted by some researchers with different integration rule for calculating
bending, shear and membrane strains. The assumed strain approach has been successfully adopted by many
researchers [Huang, 1987] as an alternative to avoid locking. In the assumed strain approach, the interpolation
of the transverse shear strains in the natural coordinate system and the membrane strains in the local coordinate
system is done at the assumed sampling points as shown in Figure 2 . A complete survey on the shell element
has been found in the literature [Yang et al. 2000].
η ζ
ξ
u
V
3 k
K hk u 2k
αk
Middle Surface u 1k β k
Figure 1 Geometry of 9-noded degenerated shell element
431
η η
4 1/√3 1/√3 1
2 1
3
1/√3
ξ 2 ξ
5
1/√3
6 2
5
For γξζ 4 For γηζ 3
Figure 2 Sampling point locations for assumed shear/membrane strains
The geometry of the degenerated shell element can be conveniently represented by the coordinates and normal
vectors of the middle surface. In general, the geometry and kinematics of deformations are described by using
different coordinate systems. The Cartesian coordinate system is used to define the geometry of the structure,
nodal coordinates and displacements, global stiffness matrix and applied load vector. The coordinates of a point
within an element are obtained by interpolating the nodal coordinates through the element shape functions.
x  9  xk  V3xk 
    9
ζ hk  y
 y  = ∑ N k ξ ,η  y k  + ∑ N k (ξ ,η )
( ) V3k  [1]
 z  k =1 z  k =1 2  z
   k  mid V3k 
The displacements at any point inside the finite element can also be expressed by
u  n uk  v1xk − v 2xk 
    n
hk  y β
y   1k 
 v  = ∑ N k  vk  + ∑ N k ζ v − v   [2]
2  z z  β 2k 
1k 2 k
w k =1 w  k =1
 v − v
   k  mid  1k 2k 
In the above expression, N k (ξ ,η ) are the element shape functions and hk is the shell thickness at node k.
ξ ,η , ζ are the natural coordinates at any node k under consideration. A nodal coordinate system is defined at
each nodal point with origin situated at the reference mid surface. The vector V3k is constructed from the nodal
coordinates at top and bottom surfaces and is expressed as
 xk   xk 
   
V3k =  yk  −  yk  [3]
z  z 
 k top  k bottom
v3k defines the direction of the normal at any node 'k', which is not necessarily perpendicular to the mid
surface. The major advantage of the definition of V3 k with normal not necessary to be perpendicular to mid
surface is that , there are no gaps or overlaps along element boundaries. The formulation of degenerated shell
element is completely described in Huang, (1987). Once the displacements are determined, the strains and
stresses are calculated using strain-displacement matrix and material constitutive matrix respectively. The next
section briefly deals with the material modeling of concrete and reinforcing steel.
3. MATERIAL MODELING
The modeling of material play may a crucial role in achieving the correct response. The presence of nonlinearity
may add another dimension of complexity to it. Many material models have been developed in the past over the
years such as compression filed theory (CFT) and modified compression filed theory (MCFT) [Vecchio and
Collins (1986)] to model the cracked response of concrete in compression. The nonlinearities in the structure
432
may accurately be estimated and incorporated in the solution algorithm. The accuracy of the solution algorithm
depends strongly on the prediction of second order effects that cause nonlinearities, such as tension stiffening,
compression softening, stress transfer nonlinearities around cracks. These nonlinearities are usually incorporated
in the constitutive modeling of the reinforced concrete. In order to incorporate geometric nonlinearity, the
second order terms of strains are to be included. In this study, only material non-linearity has been considered.
The subsequent sections describe the modeling of concrete in compression and tension, modeling of steel are
discussed briefly.
3.1 Concrete Modeling in Tension
The presence of crack in concrete has much influence on the response of nonlinear behavior of reinforced
concrete structures. The crack in the concrete is assumed to occur when the tensile stress exceeds the tensile
strength. The cracking of concrete results in the loss of continuity in the load transfer and hence the stresses in
both concrete as well as steel reinforcement differs significantly. Hence the analysis of concrete fracture has
been very important in order to predict the response of structure precisely. The numerical simulation of
concrete fracture can be represented either by discrete crack, proposed by Ngo & Scordelis (1967) or by
smeared crack, proposed by Rashid (1968). The objective of discrete crack is to simulate the initiation and
propagation of dominant cracks present in the structure. In the case of discrete crack approach, nodes are
disassociated due to the presence of cracks and therefore the structure requires frequent renumbering of nodes,
which may render in the huge computational cost. Nevertheless, when the structure's behavior has been
dominated by only few dominant cracks, the discrete modeling of cracking seems the only choice. On the other
hand, the smeared crack approach smears out the cracks over the continuum, and captures the deterioration
process through the constitutive relationship and reduces the computational cost and time drastically.
Crack modeling has gone through several stages due to the advancement in technology and computing facilities.
Earlier research work indicates that the formation of crack results in the complete reduction in stresses in the
perpendicular direction, thus neglecting the phenomenon called tension stiffening. With the rapid increase in
extensive experimental investigations as well as computing facilities, many finite element codes have been
developed for the nonlinear finite element analysis, which incorporates the tension stiffening effect..The cracks
are always assumed to be formed in the direction perpendicular to the direction of the maximum principal stress.
These directions may not necessarily remain the same throughout the analysis and loading and hence the
modeling of orientation of crack plays a significant role on the response of structure. Still, due to simplicity,
many investigations have been performed using fixed crack approach, wherein the direction of principal strain
axes may remain fixed throughout the analysis. In this study also, the direction of crack has been considered to
be fixed throughout the duration of the analysis. However, the modeling of aggregate interlock has not been
taken very seriously. The constant shear retention factor or the simple function have been employed to model
the shear transfer across the cracks.
Stresses
ft
α ft
σi
Tension
Ei
Compression εt εi Strains εm
Figure 3 Tension stiffening effect of cracked concrete
Apart from the initiation of crack, the propagation of crack also plays a crucial role on the response of structure.
The prediction of crack propagation is a very difficult phenomenon due to scarcity and confliction of test results.
433
Nevertheless, the propagation of cracks plays a crucial role on the response of nonlinear analysis of RC
structures. The plain concrete exhibits softening behavior and reinforced concrete exhibits stiffening behavior
due to the presence of active reinforcing steel. A gradual release of the concrete stress (Figure 3) is adopted in
this present study (Owen and Hinton, 1980).
The reduction in the stress is given by the following expression

 ε 1
E i = α f t′ 1 − i ; ε t ≤ ε i ≤ ε m [4]
 εm  εi
In the above equation α and ε m are the tension stiffening parameters. ε m is the maximum value reached by
the tensile strain at the point considered, , ε i is the current tensile strain in material direction 1.The coefficient
α depends on the percentage of steel in the section. In the present study, the value of α and ε m are taken as
0.5 and 0.0020 respectively. It has also been reported that the influence of the tension stiffening constants on the
response of the structures are generally small and hence the constant value is justified in the analysis (Owen and
Hinton, 1980). Generally, the cracked concrete can transfer shear forces through dowel action and aggregate
interlock. The magnitude of shear moduli has been considerably affected because of extensive cracking in
different directions.
3. 2 Concrete Modeling in Compression
The theory of plasticity has been used in the compression modeling of the concrete. The failure surface or
bounding surface has been defined to demarcate plastic behavior from the elastic behavior. Failure surface
is the important component in the concrete plasticity. Sometimes, the failure surface can be referred to as
yield surface or loading surface. The material behaves in the elastic fashion as long as the stress lies
below the failure surface. Several failure models have been developed and reported in the literature (Chen,
1982). Nevertheless, the five parameter failure model proposed by Willam and Warnke (1975) seems to
be possessing all inherent properties of the failure surface. The failure surface (Figure 4) is constructed
using two meridians namely, compression meridian and tension meridian. The two meridians are
pictorially depicted in a meridian plane and cross section of the failure surface is represented in the
deviatoric plane.
σ1
τ oct
θ = 60o
θ = 0o
σ oct
σ2 σ3
Meridian Plane (Deviatoric Plane (ii
Figure 4 Failure model for concrete
The variations of the average shear stresses τ mt and τ mc along tensile ( θ = 0 ) and compressive ( θ = 60 )
o o
meridians are approximated by second-order parabolic expressions in terms of the average normal stresses
σ m , as follows:
434
2
τ mt ρt σ  σ 
= = a 0 + a1  m  + a 2  m  θ = 00
f c′ 5 f c′  f c′   f c′ 
2
τ mc ρc σ  σ 
= = b0 + b1  m  + b2  m  θ = 60 0
f c′ 5 f c′  f c′   f c′ 
[5]
These two meridians must intersect the hydrostatic axis at the same point σ m f c′ = ξ 0
(corresponding to hydrostatic tension), the number of parameters need to be determined is reduced to five.
The five parameters (a0 or b0 , a1 , a 2 , b1 , b2 ) are to be determined from a set of experimental data, with
which the failure surface can be constructed using second-order parabolic expressions .
The failure surface is expressed as
τm
f (σ m , τ m , θ ) = 5 −1 = 0 [6]
ρ (σ m , θ )
ρ (θ ) =
( ) [( )
2 ρ c ρ c2 − ρ t2 cos θ + ρ c (2 ρ t − ρ c ) 4 ρ c2 − ρ t2 cos 2 θ + 5 ρ t2 − 4 ρ t ρ c ]
12
( )
[7]
4 ρ c2 − ρ t2 cos 2 θ + (ρ c − 2 ρ t )
2
The formulation of Willam-Warnke five parameter material model is described in Chen (1982). Once the
yield surface is reached, any further increase in the loading results in the plastic flow. The magnitude and
direction of the plastic strain increment is defined using flow rule, which is described in the next section.
3.2 1. Flow rule
In this method, associated flow rule is employed because of the lack of experimental evidence in
non-associated flow rule. The plastic strain increment expressed in terms of current stress increment is
given as
∂f (σ )
dε ijp = dλ [8]
∂σ ij
dλ determines the magnitude of the plastic strain increment. The gradient ∂f (σ ) ∂σ ij defines the
direction of plastic strain increment to be perpendicular to the yield surface; f (σ ) is the loading
condition or the loading surfaces.
3.2.2. Hardening rule
The relationship between loading surfaces (or effective stress) and the plastic work (accumulated plastic strain) is
represented by a hardening rule (Figure 5). The "Madrid parabola" is used to define the hardening rule.
1 E0 2
σ = Eo ε − ε [9]
2 ε0
In the above equation, E0 is the initial elasticity modulus, ε is the total strain and ε 0 is the total strain at peak
stress f c′ .The total strain can be divided into elastic and plastic components as
The elasto-plastic constitutive matrix is given by the following expression
[D ] = [D] − [D]{a}{a} [D]

T
H + {a} [D ]{a} [10]

ep T
In the above equation, a =flow vector, defined by the stress gradient of the yield function; D = constitutive
435
matrix in elastic range. The second term in the above equation represents the effect of degradation of material
during the plastic loading.
3.3 Modeling of Reinforcement in Tension and Compression
In order to incorporate the effect of steel reinforcement, the layered approach is adopted in this study. The steel
is modeled as a smeared layer of equivalent thickness in the natural coordinate system. The properties of the
material is assumed to be constant in that layer. The bi-linear stress strain curve with linear elastic and strain
hardening region is adopted in this study for both compression and tension. The vertical and horizontal
reinforcement in the shear wall is taken as 1 %, however, no ductile detailing is considered near the openings.
4. DYNAMIC ANALYSIS
The dynamic analysis of structure can be performed by three ways namely (i) Equivalent lateral force method
(ii) Response Spectrum method (iii) Time History Method. The equivalent lateral force method determines the
equivalent dynamic effect in the static manner. The response spectrum method aims in determining the
maximum response quantity of the structure. For tall and irregular buildings, dynamic analysis by Response
Spectrum Method seems to be a popular choice among designers. The time history analysis of the structure has
been successfully used to analyze the structure especially of huge importance. Even though, time history
analysis consumes time, it is the only method capable of giving results closer to the actual one especially in the
non-linear regime. In the dynamic analysis, the loads are applied over a period of time and the response is
obtained at different time intervals. The equation of dynamic equilibrium at any time 't' is given by equation (11)
[M ][U&& t ]+ [C ][U& t ]+ [K ][U t ] = [R t ] [11]
In the above equation, M, C and K are the mass, damping and stiffness matrices respectively. The mass matrix
can be formulated either by using consistent mass approach or by using lumped mass approach. Since damping
cannot be precisely determined analytically, the damping can be considered proportional to mass or stiffness or
both depending on the type of the problem. The direct time integration of the equation of motion can be
performed using explicit (central difference scheme) and implicit (Houbolt method, Newmark Beta method and
Wilson Theta method, etc.,) time integration. In the explicit time integration, the formation of complete stiffness
matrix of the structure is not required and hence saves the lot of computer time and money in storing and saving
those data. Moreover, in the case of all explicit time integration schemes, the iterations are not required as the
equilibrium at time t + ∆t depends on the equilibrium at time t. Nevertheless, the major drawback of explicit
time integration is that the time step ( ∆t ) used for calculation of response has to be smaller than the critical
time step( ∆t cr ) to ensure the stable solution.
Tn 2
∆t ≤ ∆tcr = = [12]
π ω
On the other hand, implicit time integration requires the iterations to be carried out within the time step as The
solution at time t+ ∆t involves the equilibrium equation at t+ ∆t . The Newmark β method converges to various
implicit and explicit schemes as shown in Table 1 for different values of Beta, called the stability parameter.
In this study, for β=0.25, the Newmark β method converges to the constant acceleration implicit method, known
as trapezoidal rule. The trapezoidal rule is unconditionally stable and hence allows larger time step to be used in
the calculation of response. Nevertheless, the time step can be made smaller from the accuracy point of view.
The formulation of implicit Newmark Beta method (Trapezoidal rule) is mentioned in Bathe (1996).
4.1. Formulation of Mass Matrix

In any dynamic analysis, the formulation of mass matrix is very important in capturing the correct response of a
structure. The masses can be assumed to be distributed over the entire finite element mesh or can be assumed to be
lumped at nodes. The former is known as consistent mass matrix and later is known as lumped mass matrix. The
mass matrix is said to be consistent if the formulation involves the same shape functions (Ni) as used for the
436
determination of stiffness matrix. The consistent mass matrix contains off-diagonal terms. Nevertheless, the
consistent mass matrix is computationally expensive. The consistent element mass matrix is given by
[M e ] = ∫ ρ [N i ][N i ]T dV [13]
V
For linear or translational motion, resistance of object to a change of state in motion is measured in terms of mass
and given by Newton’s law as follows
F =ma [14]
On the other hand, when a rigid body is rotated, the resistance of object to a change of state in a rotator motion
given by rotational inertia measured in terms of moment of inertia. The influence of rotary inertia in the case of
thick plates is established in the literature (Huang 1989).
N N
1   N
= ∑ mi (ωri ) = ω 2 ∑ mi ri 2 ; I = ∑ mi ri 2 ; T = Iω 2
1 1 1
T= ∑ 2m v 2
i i
i =1 2
2
2  2
i =1  i =1 [15]
where in the above expression of kinetic energy, I is the moment of inertia and ω is the angular velocity; The
above equation can be comparable in linear terms by replacing the moment of inertia by mass and angular velocity
by linear velocity. The Newton’s law of rotational motion is given by
τ = Iα [16]
where α is the angular acceleration. The lumped mass matrix is purely diagonal and hence computationally
cheaper than the consistent mass matrix. Nevertheless, the diagonalization of the mass matrix from the full mass
matrix results in the loss of information and accuracy (Huang 1987). There are many ways of lumping the mass
matrices as described in the literature. Nodal quadrature, row sum and special lumping are the three lumping
procedures available to generate the lumped mass matrices. It has found that all three methods of lumping lead to
the same mass matrix for nine-node rectangular elements. One of the most efficient means of lumping is to
distribute the element mass in proportion to the diagonal terms of consistent mass matrix [Archer and Whalen
2006] and also discarding the off-diagonal elements. This way of lumping has been successfully used in many
finite element codes in practice. The advantage of above special lumping scheme is that the assurance of positive
definiteness of mass matrix. Lumped mass matrix is preferred if not mandatory, over consistent mass matrix in the
case of explicit time integration. The use of lumped mass matrix is mostly employed in lower order elements. For
higher order elements, the use of lumped mass matrix may not be an appropriate option. Hence, in this study, a
consistent mass matrix is employed to formulate the mass matrix.
The linear inertia or translational inertia is given by the following expression
mii = wi ρdV ∫ [17]
The rotational inertia is given by the following expression
I ii = wi ∫ ρ z ′ 2 dV = Wi I ′
Ve
[18]
I ′ = ∫ ρ z ' 2 dV
Ve
Where Wi is the multiplier and ρis the mass density and z ′ is the position of the layer with respect to the axis of
rotation. The multiplier is represented by a following expression.
∫ ρ [N ][N ]
T
i i dV
Wi =
Ve
n
[19]
∑ ∫ ρ [N ][N ]
T
k k dV
K =1 Ve
437
4.2. Formulation of Damping matrix
Mass and stiffness matrices can be represented systematically by overall geometry and material characteristics.
However, damping can only be represented in a phenomenological manner and thus making the dynamic analysis
of structures in a state of uncertainty. Nevertheless, several investigations have been done in making the
representation of damping in a simplistic yet logical manner (Zareian and Medina, 2010). There is no single
universally accepted methodology for representing damping because of the nature of the state variables which
control damping. Rayleigh dissipation function assumes that the dissipation of energy takes place and can be
idealized as the function of velocity. Damping matrix can be formulated analogous to mass and stiffness matrices
(Duggal, 2007). When Rayleigh Damping is used, the resultant damping matrix is of same size as stiffness matrix.
It is also important to note that the damping matrix should be formulated from damping ratio and not from the
member sizes. Rayleigh damping is being used conveniently because of its versatility in segregating each modes
independently. The amount of damping can be set appropriately by setting the values of alpha and beta relevantly
depending on the requirement of including higher modes.
Sometimes, the additional concentrated damping is also incorporated at selected degrees of freedom in addition to
Rayleigh damping. Since damping is a function of velocity, if there is no motion, there will be no damping. It has
been mentioned that visco elastic dampers are being employed in structures to mitigate dynamic effects. Usually,
viscous damping is assumed because closed form solution can be easily available. Very little information is
available about variation of damping of linear and non-linear systems. However, the effect of damping is
generally less than the inertial and stiffness effects in most of the practical situations. Therefore, it is reasonable to
account for damping by a simplified approximation. The representation of damping through viscous damping
coefficient has been in use due to simplicity and accuracy. The damping force is assumed to be proportional to
velocity and the constant of proportionality being the viscous damping coefficient. Whenever the system vibrates
in a fluid, viscous damping is said to occur. The damping forces are proportional to velocity of the medium and is
represented as
F = cu& [20]
[C ] = [M ] ∑ (ak [M ]−1 [K ])
k = p −1
k
[21]
k =0
In the above equation, k=2 yields
[C ] = α [ M ] + β [ K ] [22]
α βω i
ζi = + [23]
2ω i 2
From the above equation, it is essential to note that if β parameter is zero, the higher modes of the structure will be
assigned very little damping. When the parameter α is zero, the higher modes will be heavily damped as the
damping ratio is directly proportional to frequency (Clough and Penzein, 1993). Thus, the choice of damping is
problem dependent. Hence, it is inevitable to perform modal analysis to determine the different frequencies for
different modes. It is also to be noted that the damping is controlled by only two parameters. Thus, in the Cauchy
series, one may include as many terms as possible depending upon the computational efficiency. The use of
proportional damping is implemented in most of the finite element codes. The reason for the use of proportional
damping is justified by the following explanation. In the equation of motion, the coupling of terms usually occurs
which are reflected in the mass and stiffness matrices. Inertia coupling is present when the mass matrix is
non-diagonal and static coupling is present when the stiffness matrix is non-diagonal. The coupling of the modes
usually can be avoided easily in the case of undamped free vibration. The same is not true for damped vibration.
Hence in order to represent the equation of motion in uncoupled form, it is suggested to have a damping matrix
proportional to uncoupled mass and stiffness matrices. Thus, Rayleigh’s proportional damping has the specific
advantage that the equation of motion can be uncoupled when it is proportional to mass and stiffness matrices.
Thus, it is proposed to use Rayleigh damping in this study.
438
4.3 Nonlinear Solution
The numerical procedure for nonlinear analysis employs the iterative procedure to satisfy the equilibrium at the
end of the load step. Once the convergence of the solution is achieved, the algorithm proceeds to the next step. It
is always desirable to keep the load step very small especially after the onset of nonlinear behavior. The
stiffness matrix is updated at the beginning of each load step. The convergence is said to be achieved if the out
of balance forces, calculated as under, is less than the specified tolerance.
ψ ni = f n − pin = f n − ∫ B T σ ni dV < Tolerance (0.0025 ) [24]
V
In order to determine the limiting size of openings, a rectangular reinforced concrete slender shear wall of
dimensions 35 m high, 8 m wide and 0.3 m thick (aspect ratio: 4.375) with various opening sizes is
considered. The entire shear wall is discretized into small finite elements of size (0.5m x 0.5m) using nine
noded degenerated shell element with five degree of freedom. The base of the shear wall is completely
arrested in all degrees of freedom. The discretized shear wall with various opening sizes are shown in
Figure 5. The opening size is varied in such a way that the aspect ratio of the opening is almost square. In
order to incorporate the effect of steel reinforcement, the layered approach is adopted in this study. The
properties of the material are assumed to be constant in that layer. The reinforcement ratio of 0.25 % is
adopted for both vertical and horizontal reinforcement. The steel is modelled as a smeared layer of
equivalent thickness in the natural coordinate system. The bi-linear stress strain curve with linear elastic
and strain hardening region is adopted in this study for both compression and tension. The strengthening
of shear walls around openings is considered as per IS: 13920 requirements. For the dynamic analysis, the
fundamental natural period and frequency have been calculated using the equations mentioned below [Goel
and Chopra 1998]. Only stiffness proportional damping has been considered in the present investigation.
The shear wall has been subjected to the ground acceleration, as shown in Figure 5.5
Based on Dunkerley's method, the fundamental period of a cantilever, considering flexural and shear
deformations, is T = TF2 + TS2
[25]
In the above equation, TF and Ts are the fundamental periods of pure-flexural and pure-shear cantilevers,
respectively. For uniform cantilevers, TF and Ts are given by the following formulas.


2π m 2 m 1 A 
TF = H ; Ts = 4 × H ; Ae = 2 
[26]
3 .516 EI k G Ae H  
1 + 0 .83   
D 
In the above equation, m = mass per unit height; E = modulus of elasticity; G = shear modulus; I = the second
moment of inertia; A = section area; k is the shape factor to account for non-uniform distribution of shear stresses
(= 5/6 for rectangular sections). In the above equation, D is the plan dimension of the cantilever in the direction
under consideration. The unconditionally stable Newmark Beta method of direct time integration has been
employed to determine the response subjected to ground acceleration for a duration of 1.5 seconds. The time step
for the analysis has been taken as 0.015 seconds. The ground acceleration time history has been discretized into
750 steps. The waveform is almost similar to the NS component of the famous EL Centro earthquake 1940. The
material properties are as mentioned in Table 1.
439
15
Acceleration (m/s2)
10
5
0
0.0 0.3 0.6 0.9 1.2 1.5
-5
-10 Time (s)
-15
Figure. 5 Dynamic Ground Acceleration Applied at the Base of the Structure
3.
3.
3.
3.
3.
3.
3.
3.
3.
3.
Case (i) opening Case (ii) opening Case (iii) opening Case (iv) opening Case (v) opening
size 1m x 1m size 1m x 1.5m size 2m x 2m size 2m x 2.5m size 2m x 3m
Figure 6 Discretization of slender shear wall with various opening sizes
Table 1 Material properties used for the analysis
Material Property Magnitude

Young's Modulus of Concrete 2.7 x 1010 kN/m2
Young's Modulus of Steel 2.0 x 1011 kN/m2
Poisson's Ratio 0.100
Uni-axial Compressive Strength of Concrete 30 x 106 kN/m2
Tensile Strength of Concrete 3 x 106 kN/m2
Ultimate Crushing Strain of Concrete 0.0035
Yield stress of steel 50 x 107 kN/m2
Tension stiffening coefficient ( α ts ) 0.6
Tension stiffening constant ( ε ts ) 0.0020
440
For Zero
Case 1 Case 2 Case 3 Case 4 Case 5
Damping 40 50 40
Displacement (mm)
40 40
Displacement (mm)
Displacement (mm)
Displacement (mm)
Displacement (mm)
20 20 20 20
0 0
0 0 0
0.0 0.3 0.6 0.9 1.2 1.5 0.0 0.3 0.6 0.9 1.2 1.5
0.0 0.3 0.6 0.9 1.2 1.5 0.0 0.3 0.6 0.9 1.2 1.5 0.0 0.3 0.6 0.9 1.2 1.5 -20
-20 -20 -20
-40
Time (s) Time (s) Time (s) -40 Time (s) Time (s)
-40 -50 -40
For 2% Case 1 Case 2 Case 3 20 Case 4 Case 5

20 20
Displacement …
Displacement…
20
Displacement …
Displacement …
Damping 20
Displacement
10 10
(mm)
0 0 0 0 0
0.0 0.3 0.6 0.9 1.2 1.5 0.0 0.3 0.6 0.9 1.2 1.5 0.0 0.3 0.6 0.9 1.2 1.5 0.0 0.3 0.6 0.9 1.2 1.5 0.0 0.3 0.6 0.9 1.2 1.5
-10 -10
Time (s)
-20 Time (s) -20 Time (s) Time (s) -20 Time (s)
-20 -20
For 5% Case 1 Case 2 Case 3 Case 4 Case 5

10
Displacement …
Displacement …
Displacement …
damping 10
Displacement …
10 10 10
Displacement
5 5 5 5
(mm)
0 0 0
0 0
0.0 0.3 0.6 0.9 1.2 1.5
-5 0.0 0.3 0.6 0.9 1.2 1.5 -5
0.0 0.3 0.6 0.9 1.2 1.5 -5 0.00.30.60.91.21.5 -5
0.0 0.3 0.6 0.9 1.2 1.5
-10
-10 -10 Time (s)
Time (s) -10 Time (s) Time (s) -10 Time (s)
For 10% Case 1 Case 2 Case 3 Case 4

Displacement…
Case 5
Displacement …
damping 6 12
Displacement …
Displacement …
4 4 10
4
Displacement
2 2 6
5
(mm)
0 0
0.0 0.3 0.6 0.9 1.2 1.5 0 0
-2 0.00.30.60.91.21.5 -2 0.00.30.60.9 1.21.5
-4 0.0 0.3 0.6 0.9 1.2 1.5 0.0 0.3 0.6 0.9 1.2 1.5
-4 -6 -5
-6 -6 -6
Time (s) Time (s) Time (s)
-12 Time (s) Time (s)
441
Figure 7 Influence of openings on the linear structural response of shear walls for different damping ratios
For Zero Case 3 Case 4 Case 5
Case 1 Case 2
Damping 80 80
Displacement (mm)
80 80 80
Displacement (mm)
Displacement (mm)
Displacement (mm)
Displacement (mm)
40 40 40 40 40
0 0 0 0 0
0.0 0.3 0.6 0.9 1.2 1.5 0.0 0.3 0.6 0.9 1.2 1.5 0.0 0.3 0.6 0.9 1.2 1.5 0.0 0.3 0.6 0.9 1.2 1.5 0.0 0.3 0.6 0.9 1.2 1.5
-40 -40 -40 -40 -40
Time (s)
-80 -80 Time (s) -80 Time (s) -80
-80 Time (s) Time (s)
For 2% Case 1 Case 2 Case 3 Case 4 Case 5

Damping 15 15
Displacement (mm)
Displacement (mm)
Displacement (mm)
Displacement (mm)
15 15
Displacement (mm)
15
10 10 10 10
10
5 5 5 5
5
0 0 0 0 0
-5 0.0 0.3 0.6 0.9 1.2 1.5 -5 0.00.30.60.91.21.5 -5 0.00.30.60.91.21.5 -5 0.0 0.3 0.6 0.9 1.2 1.5
-5 0.0 0.3 0.6 0.9 1.2 1.5
-10 -10 -10 -10
-10
-15 -15 Time (s) Time (s) -15 Time (s)
Time (s) -15 Time (s) -15
For 5% Case 1 Case 2 Case 3 10 Case 4 Case 5

Damping 7 10
Displacement (mm)
10
Displacement (mm)
7
Displacement (mm)
Displacement (mm)
5
Displacment (mm)
5
2 5
2
0 0
0.0 0.3 0.6 0.9 1.2 1.5 0.0 0.3 0.6 0.9 1.2 1.5 0.0 0.3 0.6 0.9 1.2 1.5 0
-3 -5 -3 0.0 0.3 0.6 0.9 1.2 1.5 -5 0.0 0.3 0.6 0.9 1.2 1.5
-5
Time (s) -10 Time (s) Time (s) -10 Time (s)
-8 -8 -10 Time (s)
442
For 10% Case 1 Case 2 Case 3 Case 5
Case 4
Damping
Displacement (mm)
10 10 10 10 10
Displacemnt (mm)
Displacement (mm)
Displacement (mm)
Displacement (mm)
5 5 5 5 5
0 0 0 0 0
0.0 0.3 0.6 0.9 1.2 1.5 0.0 0.3 0.6 0.9 1.2 1.5 0.0 0.3 0.6 0.9 1.2 1.5 0.0 0.3 0.6 0.9 1.2 1.5 0.0 0.3 0.6 0.9 1.2 1.5
-5 -5 -5 -5
Time (s) -5
Time (s) Time (s) Time (s)
-10 -10 Time (s)
-10 -10
-10
Figure 8 Influence of openings on the non-linear structural response of shear walls for different damping ratios
443
Table 2 Influence of opening size and damping on the structural response of slender shear wall
Case Opening Opening % Displacement (mm)

Number size Linear Analysis for Nonlinear Analysis for
damping @ damping @
0% 2% 5% 10 % 0% 2% 5 % 10 %
1 1m x 1m 3.57 % 32.68 8.93 5.96 3.58 61.28 8.93 5.91 3.58
2 1m x 1.5m 5.36 % 32.58 8.89 5.88 3.56 56.20 8.90 5.88 3.56
3 2m x 2m 14.28 % 33.51 8.87 5.88 3.56 52.41 8.87 5.88 3.56
4 2m x 2.5m 17.86 % 40.42 9.67 6.36 3.87 57.80 9.71 6.36 3.87
5 2m x 3m 21.43 % 31.93 14.15 8.94 5.56 77.94 14.70 8.91 5.56
Table 3 Influence of damping on the linear structural response of shear wall
Damping Opening Percentage

Ratio (%) 3.56 % 5.36 % 14.28 % 17.86 % 21.43 %
δ (mm) DRR δ (mm) DRR δ (mm) DRR δ (mm) DRR δ (mm) DRR
0 32.68 ----- 32.58 ---- 33.51 ----- 40.42 ------ 31.93 -----
2 8.933 72.66 8.897 72.69 8.876 73.51 9.674 76.07 14.15 55.68
5 5.960 33.28 5.880 33.91 5.883 33.72 6.363 34.23 8.944 36.79
10 3.584 39.86 3.562 39.42 3.565 39.40 3.873 39.13 5.567 37.76
Table 4 Influence of damping on the nonlinear structural response of shear wall
Damping Opening Percentage

Ratio (%) 3.56 % 5.36 % 14.28 % 17.86 % 21.43 %
δ (mm) DRR δ (mm) DRR δ (mm) DRR δ (mm) DRR δ (mm) DRR
0 61.28 ----- 56.20 ---- 52.41 ----- 57.8 ------ 77.94 -----
2 8.936 85.42 8.900 84.16 8.878 83.06 9.721 93.18 14.70 81.13
5 5.916 33.80 5.880 33.93 5.883 33.74 6.363 34.54 8.911 39.38
10 3.584 39.42 3.562 39.42 3.565 39.40 3.873 39.13 5.567 37.53
On the basis of dynamic analysis, the following points have been observed. Figure 7 shows the influence of
opening size on the linear structural response of shear wall with various damping ratios. As expected, shear wall
with no damping has resulted in heavy displacements in comparison to 2, 5, and 10 %. At zero damping, the
shear wall with opening size 17.86 % has resulted in maximum displacement among all cases. Nevertheless, for
damping of 2%, 5% and 10%, the shear wall with opening size 21.86 % has been found to be undergoing more
displacement than other opening cases. Very slight difference has been observed in the displacement time history
waveforms. Figure 8 shows the influence of opening size on the nonlinear structural response of shear wall with
various damping ratios. Out of all cases, the shear wall with opening size 21.43 % has undergone huge
displacement than other opening cases. The influence of damping has been found to be well pronounced on the
structural response as it plays a crucial role in dampening the displacement structural response. Not significant
difference has been observed when the damping has been increased beyond 2%. Tables 3 and 4 show the influence
444
of damping on the linear and nonlinear structural response of shear wall. The impact of damping has been
considered to be the maximum when the damping is introduced from 0% to 2 % with almost 80 % and 70 %
reduction in the displacement response for nonlinear and linear responses respectively. Nevertheless, for larger
openings, the influence of damping has been less pronounced and hence demands for slightly larger damping.
This has been reflected clearly in linear analysis where for an opening size as large as 21 %, the reduction in
displacement has been only 55 % when the damping was increased from 0 % to 2%.
6. CONCLUSIONS
On the basis of above discussion, the following conclusions have been drawn.
1) The damping is considered to be very essential especially for larger openings, especially for openings as large
as 17.86% and 21. 43%.
2) The opening size in the shear wall should not be more than 17.86% beyond which the increase in
displacement is relatively significant.
3) The displacement predicted by non-linear analysis is expectedly more than linear analysis. Nevertheless, the
presence of damping as high as 10% results in the reduction in the non-linear displacement response comparable
to linear response.
REFERENCES
[1] Paulay,T., Priestley, M.J.N., and Synge, A.J. (1982). "Ductility of earthquake resisting shear walls." ACI
Journal, Proceedings, 79(4), 257-269.
[2] Thomsen, J.H., and Wallace, J.W. (2004). "Displacement based design of slender RC structural walls-
experimental verification. " Journal of Structural Engineering, ASCE, 130(4), 618-630.
[3] Taylor C.P., Cote P.A., and Wallace J.W (1998). "Design of slender reinforced concrete walls with
openings", ACI Structural Journal, 95(4), 420-433.
[4] Yanez, F.V., Park, R., Paulay, T. (1992). "Seismic behavior of walls with irregular openings." Proceedings of
the Earthquake Engineering Tenth World Conference, Balkema, Rotterdam.
[5] Ahmad, S., Irons, B.M., and Zienkiewicz, O.C. (1970). “Analysis of thick and thin shell structures by
curved finite elements.” International Journal of Numerical Methods in Engineering, 2(3), 419-451.
[6] Zienkiewicz, O.C., Taylor, R.L., Too, J.M. (1971). "Reduced integration technique in general analysis of
plates and shells." International Journal of Numerical Methods in Engineers, 3(2), 275-290.
[7] Paswey, S.F., and Clough, R.W. (1971). "Improved numerical integration of thick shell finite elements."
International Journal for Numerical Methods in Engineering, 3(4), 575-586.
[8] Huang, H.C. (1987)."Implementation of assumed strain shell elements." Computers & Structures, 25(1),
147-155.
[9] Yang, H.T.Y., Saigal, S., Masud, A., and Kapania, R.K. (2000) "A survey of recent shell finite elements."
International Journal for Numerical Methods in Engineering, 47(1-3), 101-127.
445
[10] Vecchio, F.J., and Collins, M.P. (1986). "The modified compression field theory for reinforced concrete
elements subjected to shear." ACI Journal, 82(2), 219-231.
[11] Ngo, D., and Scordelis, A.C. (1967). "Finite element analysis of reinforced concrete beams." Journal of
American Concrete Institute, 64(14), 152-163.
[12] Rashid, Y.R. (1968). "Analysis of prestressed concrete pressure vessels." Nuclear Engineering and
redesign, 7(4), 334-344.
[13] Owen D.R.G., Hinton E. (1980). Finite Elements in Plasticity: Theory and Practice. Pineridge Press Limited,
UK.
[14] Chen W.F. (1982) " Plasticity in Reinforced Concrete", McGraw-Hill: New York.
[15] Willam, K., and Warnke, E. (1975). "Constitutive model for triaxial behavior of concrete", Proceedings of
the International Association for Bridge and Structural Engineering, 19, Zurich, Switzerland, 1-30.
[16] Bathe, K.J. (1996). Finite element procedures, Prentice-Hall, Upper Saddle River, New Jersey.
[17] Archer GC., and Whalen, TM. (2006). "Development of rotationally consistent diagonal mass matrices for
plate and beam elements." Computational Methods Applied Engineering, 194: 675-689.
[18] Zareian, F., and Medina, R.A. (2010). “A Practical method for proper modeling of structural damping in
inelastic plane structural systems." Computers & Structures, 88(1-2), 45-53.
[19] Duggal, S.K. (2007). Earthquake resistant design of structures, Oxford Higher Education, Oxford
University Press, New Delhi.
[20] Clough, R.W., and Penzein, J. (1993). Dynamics of structures, McGraw-Hill, New York.
[21] Goel, R.K., and Chopra, A.K. (1998) "Period Formulas for Concrete Shear Wall Buildings." ASCE Journal
of Structural Engineering, 124(4), 426-433.
446
Paper No.: 074
STUDIES ON THE CHANNEL FITTING IN AIRCRAFT USING CLASSICAL

AND FINITE ELEMENT METHODS
4
S S S Sastry1, Anasuyeswar K2, Ayush Mathur3 and Shailaja B
1
DGM (Tech), Infotech Enterprises Limited, Hyderabad, India
Email: Subramanya.Sastry@Infotech-Enterprises.com
2
Design Engineer, Infotech Enterprises Limited, Hyderabad, India
Email: Anasuyeswar.Kattamuri@infotech-enterprises.com
3
Senior UG Student, IITK (Kanpur), Uttar Pradesh, India
Email: mayush@iitk.ac.in
4
Senior Design Engineer, Infotech Enterprises Limited, Hyderabad, India
Email: Shailaja.Bhukya@infotech-enterprises.com
ABSTRACT:
Tension Fittings are commonly classified into five types. They are Bathtub fitting, Channel fitting, Angle fitting,
‘PI’ fitting and Double angle fitting. Tension fittings are conservatively sized as their weight is usually small
relative to their importance. In this paper three analysis methods of Channel fitting are compared. A finite
element model is developed and studies are conducted to understand the behaviour of the channel fitting.
Simplified method of approach to the design of channel fitting is compared with finite element model results.
Margin of safety is calculated using three methods and are compare. Aluminium alloy 7050-T7452 is selected
for the study.
KEYWORDS: Tension Fittings, Channel Fittings, FEA, Stress Corrosion Cracking, Exact Method and Angle
Fittings
1. INTRODUCTION
A tension fitting is a fitting in which the bolt is loaded primarily in tension. They are used in the various
applications like removable wing root joints of fighter wings, wing root joints at the four corners of the wing
box, circumferential production joints on assemblies of fuselage sections, axial load transfer of the fuselage
stringer by using tension bolt to go through major bulkhead web rather than by cutting a large hole in the web
and back-up fittings for support pylons, flap track etc., which have to go through wing box [1].
Since the state of stress in a tension fitting is very complex, conventional structural analysis is extremely
difficult to obtain reasonable results. The methods used in this paper have rational basis and give loads agreeing
well with available test data [2]. A fitting factor of 1.15 [1 2] is applied in the analysis while calculating the
margin of safety. The analysis will show a minimum margin of safety of greater than 0.2 [1 2]. This condition is
demonstrated for both yield and ultimate conditions. Eccentricity of the load is inevitable because of the shape
of the fitting and various ruling dimensions. But the resulting stresses are kept to a minimum to give the
required margin of safety. A minimum edge distance of 2.5D and pitch of 6D is considered in the present study.
In this paper, diameter of the rivets is kept to a minimum and only fitting wall thickness is considered as the
variable parameter.
1.1. Analytical Modelling
The various parameters of the Channel Fitting under study are described below.
Type of the fitting: Channel Fitting

Diameter of bolt: 0.562 in
Diameter of rivet: 0.15625 in
Size of the Fitting: 2 inch by 2 inch by 4.5 inch
Ultimate load: 30 Kips
A list of parameters used in this study is provided below.
Pult – Ultimate load on the fitting

MS – Margin of Safety
ta - Fitting wall thickness
tb - Fitting web thickness
tc - Fitting end thickness
Ag – Gross area of web
An – Net area of web
λ – Fitting Factor (1.15)
Pad end – Load application end connected to the load bearing and dissipating member of aircraft
Wall end – End of the outstanding leg of the Channel Fitting connected to the backup structure
Web end – Web end of the Channel Fitting connected to the backup structure
A typical representation of the Channel Fitting under study is shown in Fig. 1.
Figure 1: Pictorial view of Channel Fitting under study
1.1.1 Loads And Boundary Conditions
The Channel Fitting under study is to be connected firmly to the backup structure using Titanium rivets
(Ti-6Al-4V).
The loading on the Channel Fitting is applied at the bolt location. This bolt load is a typical load of Channel
Fittings in fighter planes.
448
1.1.2 Materials
In-service Stress Corrosion Cracking (SCC) failures can be caused by stresses produced from a wide variety of
sources, including solution heat treatment, straightening, forming, fit-up, clamping, and sustained service loads.
These stresses may be tensile or compressive, and the stresses due to Poisson effects should not be ignored
because SCC failures can be caused by sustained shear stresses. Pin-hole flaws in some corrosion protection
coatings may also be sufficient to allow SCC to occur. The high-strength heat treatable wrought Aluminum
alloys in certain tempers are susceptible to stress-corrosion cracking, depending upon product, section size,
direction and magnitude of stress [3].
Avoiding stress corrosion cracking
In order to avoid stress-corrosion cracking, practices such as the use of press or shrink fits; taper pins; clevis
joints in which tightening of the bolt imposes a bending load on female lugs; and straightening or assembly
operations; which result in sustained surface tensile stresses (especially when acting in the Short-Transverse
grain orientation), should be avoided in these high-strength alloys 2014-T451, T4, T6, T651, T652; 2024-T3,
T351, T4; 7075-T6, T651, T652; 7150-T6151, T61511 and 7475-T6, T651 [3].
Where straightening or forming is necessary, it should be performed when the material is in the freshly
quenched condition or at an elevated temperature to minimize the residual stress induced. Where elevated
temperature forming is performed on 2014-T4 T451 or 2024-T3 T351 a subsequent precipitation heat
treatment to produce the T6 or T651, T81 or T851 temper is recommended [3].
It is good engineering practice to control sustained short-transverse tensile stress at the surface of structural
parts at the lowest practicable level. Thus, careful attention should be given in all stages of manufacturing,
starting with design of the part configuration, to choose practices in the heat treatment, fabrication, and
assembly to avoid unfavorable combinations of end grain microstructure and sustained tensile stress. The
greatest danger arises when residual, assembly, and service stress combine to produce high sustained
tensile stress at the metal surface. Sources of residual and assembly stress have been the most contributory
to Stress Corrosion Cracking problems because their presence and magnitude were not recognized. In most
cases, the design stresses (developed by functional loads) are not continuous and would not be involved in
the summation of sustained tensile stress. It is imperative that, for materials with low resistance to stress
corrosion cracking in the short-transverse grain orientation, every effort be taken to keep the level of
sustained tensile stress close to zero.
7050 is an Al-Zn-Mg-Cu-Zr alloy developed to have a combination of high strength, high resistance to
stress-corrosion cracking, and good fracture toughness, particularly in thick sections. The use of zirconium in
lieu of chromium provides a low sensitivity to quench, which results in high strengths in thick sections. Plate,
hand, and die forgings in the T74 temper have static strengths about equivalent to those of corresponding
products of 7079 in the T6 tempers and toughness levels equal to or higher than other conventional
high-strength alloys.
The properties of extrusions should be based upon the thickness at the time of quenching prior to machining.
Selection of the mechanical properties based upon its final machined thickness may be un-conservative.
Therefore, the thickness at the time of quenching to achieve properties is an important factor in the selection of
the proper thickness column. For extrusions having sections with various thicknesses, consideration should be
given to the properties as a function of thickness.
Plate in the T7451 temper has stress-corrosion resistance higher than 7075-T7651, and hand and die forgings in
the T7452 and T74 tempers, respectively, have stress-corrosion resistance similar to 7175-T74 forgings. The
T73 temper provides the highest resistance to stress corrosion for this alloy. The T76 temper provides for good
exfoliation resistance and higher stress-corrosion resistance than T6 tempers of 7075 and 7178. The T74 temper
provides stress-corrosion and strength characteristics intermediate to those of T76 and T73.
449
The materials used in Channel Fitting under study are as follows.
1. Fitting: Aluminum Alloy 7050-T7452 (Hand Forging, AMS 4108)

2. Bolt: Alloy steel (Fsu = 95 Ksi)
3. Washer: Steel
1.1.3 Analysis
Three methods of analysis are described and compared in the below paragraphs for the analysis of Channel
Fittings.
Approximate Method
Analysis of Fitting wall
1. As a first step, the Channel Fitting is made of a ductile material (Aluminum, 7050-T7452).
2. Check the Channel Fitting for limitation on dimensions.
3. Determine the tension on the fitting wall using the relation fta = (P/An), where P is the ultimate load and An
is the net tension area.
4. Calculate the moment of inertia, centre of gravity of the Channel section.
5. Determine the eccentricities of the load from the centre of gravity of the Channel section.
6. Calculate the moment of the load due to the effect of eccentricity on the wall end and web end of the
Channel Fitting.
7. Calculate the bending tensile stress (fnx, fny) in the wall using the moment and moment of inertia of the
fitting.
8. Calculate the total maximum tension stress as ftu = fta +fnx +fny.
9. Calculate margin of safety for ultimate load, MS = (Ftu/λ*ftu)-1 > 0.20, where λ [1] is the fitting factor
(1.15)
10. Check all fasteners between fitting and attached member for failure against shear, tension and shear
tension interaction modes of failure.
Analysis of Fitting end
1. Bending: Compute the ultimate bending stress in the fitting end, fbu = (kc*P/tc2). This bending stress is
function of the aspect ratio of the fitting (a/b).
2. Calculate the margin safety and make sure that its value is greater than 0.2 after using a fitting factor of
1.15.
3. Calculate the ultimate shear stress through the bolt hole, fsu = (P/0.6πDw,otc).
4. Calculate the margin of safety using the formula, MS = (Fsu/λfsu)-1 > 0.2.
Exact Method
The exact method [2] differs from the approximate method as given below.
1. Use of Tension Efficiency Factor to account for the uniform and staggered rivet pattern in determining the
wall tension stress ratio.
2. Calculation of wall tension and bending stress ratio and interaction between the tension and bending.
3. Effect of stress concentration factor on allowable bending stress.
450
Finite Element Method
Modeling: The modeling of the channel fitting is carried out using the CATIA V5 software.
Discretization: The CATIA model is imported into the HyperMesh software. It is meshed using the HEXA and
PENTA solid elements of the element library (Figure. 2). The various quality checks applied and quality
parameters recorded are shown in Table. 1.
Table 1: Quality parameters of elements
Aspect
Elements MPC Jacobian Warpage Length
SPC Ratio
HEXA PENTA RBE3 RBE2 >5 < 0.5 > 300 Min Max
10484 40 1 24 24 0 0 0 0.038 0.177
QUAD TRIA
Min >135 deg Min Max
45.28 48 46.55 78.85
Figure 2: Finite Element Mesh of the Channel Fitting
Analysis
The model is constrained at the 24 fastener locations using the RBE2 / SPC cards. The concentrated load of 30
Kips is applied at the bolt location using RBE3 element. The finite element model is analyzed using the industry
standard analysis software MSC/NASTRAN. The standard material properties for AA7050-T7452 are used in the
analysis [3].
The margin of safety computed using the approximate, exact and finite element method is compared in Table 2.
Table 2: Table of Margin of Safety in the Channel fitting
Method Approximate - Niu FEM Exact - Analytical

MoS (%) 171.0 174.0 179.56
Figures 3 to 7 show the variation of stresses in Channel fitting. Figure 8 shows the deformation pattern of the
Channel fitting.
451
The detailed study of results is presented below.
1. The study is presented for a Channel Fitting of cross section 2 inch * 2 inch.
2. From Fig. 3 it is observed that the von Mises stress in the end wall of the channel fitting exceeds the
ultimate stress (70 ksi) of the material near the hole boundary. This will indicate that a detail plastic analysis
of the fitting in this region is mandatory.
3. From Fig. 4 it is observed that the Sigma-xx (50.4 Ksi) stress is well within the yield stress (60 Ksi) of the
material.
4. From Fig. 5 it is observed that the Sigma-yy (118 Ksi) stress exceeds the ultimate stress (70 Ksi) of the
material near the hole boundary.
5. From Fig. 6 it is observed that the Sigma-zz (76.2 Ksi)stress exceeds the ultimate stress (70 Ksi) of the
material near the hole boundary.
6. From Fig. 7 it is observed that the maximum shear stress (63 Ksi) exceeds the ultimate shear stress (41 ksi)
of the material near the Hole boundary.
7. Fig. 8 shows the deformation of the Channel Fitting. Since the Channel Fitting is not a stiffness critical
structure, this magnitude of the deflection is not of concern.
8. From Table 2 it is observed that the approximate method under predicts the margin of safety by 4.76% as
compared with the exact method. FEM under predicts the margin of safety by 3.09% as compared with the
exact method.
3.CONCLUSIONS
1. Approximate method doesn’t take into consideration the reduction in efficiency of the channel fitting due to
the presence of rivet holes.
2. Approximate method doesn’t take into consideration the effect of bending-tension interaction in the channel
fitting.
452
3. Approximate method under predicts the stress in the end pad by 24.24% as compared to the exact
method.
4. A more realistic modeling of fasteners using the fastener elastic properties will help in understanding
the behavior of the Channel fitting in more detail.
5. By comparing the three methods, it is concluded that detailed study of the channel is fitting is required
using more advanced methods like nonlinear analysis using material nonlinearity.
ACKNOWLEDGEMENT
We sincerely thank and acknowledge the support and encouragement provided by Mr. M. Lakshmana Rao, VP,
Infotech Enterprises Limited, Hyderabad in bringing out this work in the form of a technical paper.
REFERENCES
[1] Niu, Airframe stress analysis and sizing, Second edition 1999.
[2] Lockheed Stress Memo 88a.pdf
[3] Richard C. Rice, Jana L. Jackson, John Bakuckas, and Steven Thompson, MMPDS, Jan 2003
453
Paper No.: 076
MODAL ANALYSIS OF LAMINATED COMPOSITE SPHERICAL SHELL

PANELS USING FINITE ELEMENT METHOD
1 2
P.V. Katariya and S.K. Panda
1
Research Scholar, Dept. of Mechanical Engineering, National Institute of Technology, Rourkela, India
2
Professor, Dept. of Mechanical Engineering, National Institute of Technology, Rourkela, India
Email: pankaj_bhanushali@yahoo.in, call2subrat@gmail.com
ABSTRACT :
In this paper, an effort has been made to model laminated spherical shells using ANSYS and solved vibration
responses of different mode using ANSYS parametric design language code. An eight-noded serendipity shell
element is used for the discretization purpose. The convergence test has been carried out and the results are
compared with those available published literature to check the robustness of the model. Sets of numerical
results are obtained for different parameters such as the boundary conditions, the lay-up schemes, and the
material properties of the structures and their effects on different vibration modes are discussed in details.
KEYWORDS: Free vibration analysis, finite element method, laminated composite panels, APDL Code,
ANSYS
1. INTRODUCTION
Modal analysis is a method to describe a structure in terms of its natural characteristics which are the frequency,
damping and mode shapes–its dynamic properties. The modal analysis is very important in earthquake
engineering; because the obtained periods of vibration helps in checking that a building's natural frequency does
not match with the frequency of earthquakes prone region where the building is to be constructed. If the natural
frequency of the structures equals an earthquake's frequency, the structure undergoes structural damage due to
resonance. Laminated shells made up of composites are being increasingly used today in many engineering
structural components such as aerospace, automotive, sports, biomedical, civil structures and marine due to their
invariable tailored properties. It is necessary to model and analyze these structures for design and
implementation purpose. It is well known that, most of the structural components are of shell type geometry
rather than the flat type and are exposed to large acoustic, aerodynamic, inertia excitation and vibration. The
structural component of aircraft, launch vehicle and missiles are of curved type of geometry and are subjected to
combined loading. Vibrations with high amplitude cause large tensions and the reduction of life due to fatigue. It
is worthy to mention that, the first mode of vibration as well as some higher mode has got important effect on
the structural behavior. The structural members made-up of composite and/or hybrid materials are flexible in
nature as compared to the conventional materials; hence there is a necessity to predict the different modal
frequencies to exploit the exact design strength. In structural engineering by using the structure's overall mass
and stiffness, the modal analysis is applied to find the various periods at which the structure will naturally
resonate. For the analysis of the composite structures many theories are used in past such as, classical laminated
plate theory (CLPT), first order shear deformation theory (FSDT) and higher order shear deformation theory
(HSDT). Many researchers have studied the vibration behavior of laminated composite structures using
numerical and analytical methods. A brief review of the available literature in this field is discussed for the sake
of brevity and problem definition.
Xiang and Kang (2012) have analyzed free vibration behavior of laminated composite plates by using a
mesh less local collocation method based on thin plate spline radial basis function. Mantari et al. (2011) studied
static and dynamic analysis of laminated composite and sandwich plates using a new higher-order shear
deformation theory. Cui et al. (2011) developed a model by using the FSDT mid-plane kinematics and solved by
the discrete shear gap method for bending and the vibration analysis of laminated composite plates. The free
vibration characteristics of laminated composite shells are analyzed by Xiang et al. (2011). Huu-Tai and
Seung-Eock (2010) studied free vibration responses of laminated composite plate which is analyzed by using
two variable refined plate theories. Topal and Uzman (2007) analyzed optimum design of laminated composite
plates by using four node Lagrangian finite element approach and the model is developed using the FSDT.
Matsunaga (2007) have studied the free vibration and stability problems using Hamilton’s principle and solved
based on the power series expansion. Hatami et al. (2007) have analyzed free vibration of moving laminated
composite plates based on the CLPT. Vibration behavior of cross-ply laminated composite plate is analyzed by
Matsunaga (2002). Kant and Swaminathan (2001) used a higher order refined theory for the mathematical
model and solved analytically for the free vibration of laminated composite and sandwich plates. Khdeir and
Librescu (1988) analyzed symmetric cross-ply laminated elastic plates using the HSDT mid-plane kinematics.
Putcha and Reddy (1986) reported stability and natural vibration behavior of laminated plates by using a mixed
element based on a refined plate theory. Reddy and Liu (1985) studied static and vibration behavior of laminated
composite shells based on the HSDT kinematic model.
From the above review, it is clear that there are numerous researchers have already described on vibration
of laminated structures but studies related to free vibration using commercial FEM tool are less in number.
Hence, the present investigation aims to analyze the free vibration behavior of shear deformable laminated
spherical panels using elastic properties. There are many FEM tools are available today for the modelling and
analysis out of which ANSYS is one accepted by many industries. In this article, a general finite element model
of laminated composite is developed in commercial finite element package ANSYS 13.0 and analyzed by the
help of APDL coding. The model has been verified and a set of parametric study is obtained by using developed
model in ANSYS parametric design language (APDL) code to obtain the eigenvalue solutions of vibration cases.
The non-dimensionalised fundamental frequency responses are obtained by varying some design parameters like
curvature ratios (R/a), span-to-thickness ratios (a/h), lay-up scheme and support conditions.
2. MATHEMATICAL MODEL
The doubly curved panel is classified according to its curvature parameter, such as doubly curved (where the
two curvatures are different), spherical (where two curvature is same) and cylindrical (where one curvature is
infinite). A panel is assumed, which is composed of N number of anisotropic layers of uniform thickness h,
length a, and width b which is shown in Figure 1. For the present analysis laminated panel model is developed
using APDL coding in ANSYS. It is well known that, the mathematical model in ANSYS derives based on the
FSDT as follows [Jones (1999)]:
u(x, y, z) = u0 (x, y) + zθx (x, y)

v(x, y, z) = v0 (x, y) + zθy (x, y) (1)
w(x, y, z) = w0 (x, y) + zθz (x, y)
where, u, v and w denote the displacements of any point along the (x, y, z) coordinates. u0, v0 and w0 are the
in-plane and transverse displacements of a point (x, y) on the mid-plane respectively and θ x, θ y and θ z are
the rotations of normal to mid-plane.
The following strain displacements are used to obtain the strains for any kth lamina as follows:
{ε } = {u , x , v , y , w , z , u , z + w,x , v,z + w, y , u , y + v,x }

T
(2)
∂u w ∂v w ∂w
where, u, x = + , v, y = + , and w,z = are the differential form of displacement terms with
∂x Rx ∂y Ry ∂z
respect to the corresponding axis. Rx and Ry are radii of curvatures in the respective axis. The curvature is taken
constant for throughout the analysis and the curved panel termed as flat when the curvature attained to infinite.
455
Figure 1 Geometry data of shallow shell laminated composite
The generalized stress tensor can be obtained via elastic constitutive relation of any kth lamina in the
following fashion.
{σ } = {σ x σ y τ xy τ xz τ yz } = Q  {ε }
k Tk k
(3)
{σ } {Q}
k k
where, is the stress tensor, transfer reduced stiffness matrix.
Total strain energy of laminated structure due to the stresses and strains as discussed in the above can be
conceded as
1
{ε } Q  {ε } dV (4)
2 ∫v
U=
T
The global displacement field can be expressed as
{δ } ={u v w} = [ f ]{δ }
T
(5)
where, [ f ] is the function of thickness co-ordinate.

The kinetic energy of the vibrated laminated panel can be expressed as
T
1  •   •  (6)
T =
2 ∫ [ ρ ]  δ
v


δ  dV
 
{}•
where, ρ is the density, δ and {δ } are the velocity and displacement vector, respectively.
The external work done expression will follow the same steps as in the Eq. (4)
456
1
W = ∫ [ε G ] { N } dV
T
(7)
2
{N} = {N x N xy } is the load vector.

T
where, ε G  is the geometric strain vector and Ny
2.1 Finite Element Modelling
The domain is discretized using the finite element steps and the elemental equations are obtained by using,
an eight noded serendipity shell element (SHELL281) from the ANSYS library. This element is suitable for
analyzing thin to moderately thick shell/plate structures. The nodal displacements can be presented in the form
of their shape functions and their corresponding nodal values as follows:
8 8 8 8 8 8
u 0 = ∑N i ui0 , v 0 = ∑N i vi0 , w0 = ∑Ni wi0 , θ x = ∑Niθ xi , θ y = ∑N iθ yi and θ z = ∑N iθ zi (8)
i =1 i =1 i =1 i =1 i =1 i =1
The above equation can be rewritten in ith nodal displacement as follows:
{δ } = {u
i
* 0
i vi0 wi0 θ xi0 θ yi0 θ zi0 } (9)
{δ } = [ N ]{δ }
i
*
i i (10)
where, [Ni] nodal shape function of a 8 noded element and the values can be seen in Cook (2000).
Substituting the value of nodal displacement in strain, kinetic energy, strain energy and work done, we will
get
{ε } = [ Bi ]{δi } (11)
T=
1
2 ∫A
[ ρ ][ N ]i [ N ]i δ&i dA
T
{ } (12)
1
[ Bi ] {δ i } [ D ][ Bi ]{δ i } dA
2∫
U=
T T
(13)
1
{W } = ∫ [ε G ] [ DG ][ε G ] dA (14)
T
where, [ Bi ] , and [ε G ] are the strain displacement relation matrix and the geometric strain vectors, respectively.
The final form of the equation can be obtained by using Hamilton’s principle as follows:
t
(15)
∫L dt
0
= 0
where, L = T − (U + W )
457
Using Eq. (11)-(14) the Eq. (15) will be modified as follows
[M ] {δ }
••
+ [K ]{δ } = { F } (16)
where, [ M ] and [ K] are the global mass and stiffness matrix.

By dropping appropriate terms from the Eq. (16) the final form of governing free vibration equation can be
written as follows:
{}
••
[ M] δ + [ K ]{δ } = 0 (17)
In this section, the free vibration behavior of laminated composite spherical panel has been analyzed
using APDL code developed in ANSYS environment. The numerical value for the fundamental natural
frequency is calculated using Block-Lanczos method. The developed model is solved in the ANSYS 13.0
environment to find the vibration responses of the laminated composite spherical panel. Some new results are
obtained with different thickness ratios (a/h), curvature ratios (R/a), lay up scheme and support conditions. The
non-dimensional frequency parameter is (ϖ ) = ω a 2 ρ E2 h 2 throughout analysis if it is not stated elsewhere.
The boundary conditions are used for the computation purpose are expressed as follows:
Simply supported (SSSS): v=w=θy=0 at x=0, and a;
u=w=θx=0 at y=0 and b;
Clamped (CCCC): u= v =w=θx= θy= θz=0 at x=0 and a and y=0 and b;
3.1 Convergence and Validation
In this numerical analysis, a square simply supported cross-ply [(00/900), (00/900/00) and (00/900/900/00)]
laminated composite spherical panels have been analyzed for the comparison and validation of free vibration
behavior. The material property used for the vibration behavior of laminated composite panels same as the
reference is tabulated in Table 1. The developed ANSYS model is validated by comparing the results with
references and convergence of laminated composite spherical panel have been plotted in Figure 2. From the
Figure 2 it can be seen that the present results are showing good agreement with the available references.
Based on the convergence study, a (6×6) mesh is used to obtain the responses for the laminated composite
spherical shell and same used throughout the analysis. Some new results for vibration behavior of laminated
composites spherical shell are obtained with same material properties of references and various
span-to-thickness ratios (a/h), curvature ratios (R/a) and different support conditions.
Table 1 Material properties for the laminated composite spherical shell

E1=E2=25 G12= G13=0.6E2 G23=0.5E2
ν12= ν13=0.25 ρ=1
458
a / h =10 (0 0 /90 0 ) ϖ Pre.=9.1589, ϖ Ref.(HSDT)=9.0680, ϖ Ref.(FSDT)=8.9841
40
38
0
a /h =100 (0 /90 ) 0
ϖ Pre.=17.7952, ϖ Ref.(HSDT)=16.7100, ϖ Ref.(FSDT)=16.7060
Non-dimensional fundamental frequency (ϖ)

36 a /h =10 (0 0 /90 0 /0 0 ) ϖ Pre.=12.8811, ϖ Ref.(HSDT)=11.8600, ϖ Ref.(FSDT)=12.2150
34
32
0 0
a /h =100 (0 /90 /0 )
0
30
28
a /h =10 (0 0 /90 0 /90 0 /0 )
0
26 a /h =100 (0 /90 /90 /0 ) ϖ Pre.=20.4888 , ϖ Ref.(HSDT)=20.3800, ϖ Ref.(FSDT)=20.3800
0 0 0 0
24
22
20
18
16
14
12
10
8
6
4
2
0
2×2 3×3 4×4 5×5 6×6 7×7 8×8
Mesh size
Figure 2 Convergence for laminated composite spherical shell panel
For all edges simply supported (SSSS), laminated spherical panel, by varying span-to-thickness ratios
(a/h) and aspect ratios (a/b) for symmetric cross-ply (00/900)s and angle-ply (±450)s results are obtained and
shown in Figure 3 (a) and (b), respectively. Further, for symmetric cross-ply (00/900)s and angle-ply (±45)s
laminated spherical panel results are obtained for different support conditions, i.e., clamped (CCCC) and one
edge clamped and other edges free (CFFF) and plotted in Figure 4 (a) and (b) and Figure 5 (a) and (b),
respectively. It is well known that as the span-to-thickness ratio (a/h) increases the non-dimensional
fundamental frequency of the laminated spherical panel also increases and as the thickness ratio (R/a) increases
the non-dimensional fundamental frequency of the laminated cylinder also decreases which can be seen in
Figure 4 and 5. The first and fourth mode shape of the spherical shell with CCCC support condition has been
showed in the Figure 6.
5000 15000
14000 a/h = 5 a/h = 50
4500 a/h = 5 a/h = 50 a/h = 10 a/h = 100
13000
Non-dimensional fundamental frequency
a/h = 10 a/h = 100 a/h = 20

4000 12000
a/h = 20
11000
3500
10000
3000 9000
8000
2500
7000
2000 6000
5000
1500
4000
1000 3000
2000
500
1000
0 0
-1000
-500
0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16
aspect ratio (a/b) aspect ratio (a/b)
(a) (b)
Figure 3 Vibration response of panel for SSSS support condition R/a=10 (a) (00/900)s (b) (±45)s
459
100 100
95 95
90 90 a/h = 5 a/h = 50
a/h = 5 a/h = 50

85 85 a/h = 10 a/h = 100
80 a/h = 10 a/h = 100 80
a/h = 20 a/h = 20
75 75
70 70
65 65
60 60
55 55
50 50
45 45
40 40
35 35
30 30
25 25
20 20
15 15
10 10
5 5
0 0
0 20 40 60 80 100 0 20 40 60 80 100
Curvature ratio (R/a) Curvature ratio (R/a)
(a) (b)
Figure 4 Vibration response of panel for CCCC support condition a/b=1 (a) (00/900)s (b) (±45)s
5.0 5.0
4.5 a/h = 5 a/h = 50 4.5

a/h = 5 a/h = 50
a/h = 10 a/h = 100 a/h = 10 a/h = 100

4.0 4.0
a/h = 20 a/h = 20
3.5 3.5
3.0 3.0
2.5 2.5
2.0 2.0
1.5 1.5
1.0 1.0
0.5 0.5
0.0 0.0
0 20 40 60 80 100 0 20 40 60 80 100
Curvature ratio (R/a) Curvature ratio (R/a)
(a) (b)
Figure 5 Vibration response of panel for CFFF support condition a/b=1 (a) (00/900)s (b) (±45)s
1st mode shape 4th mode shape

Figure 6 Mode shapes of spherical panel (00/900)s for CCCC support condition with a/h=5 and R/a=5
460
4. CONCLUSIONS
In this present article, a finite element model is developed using APDL code in ANSYS environment and its
comprehensive test has been done. The modal analysis of laminated composite spherical panel has been carried
out for different parameters to check the efficacy of the developed model. The ANSYS model is used to obtain
the frequency responses of laminated composite spherical panels for different span-to-thickness ratios, curvature
ratios, aspect ratios with different support condition. The convergence study is clearly showing that the present
developed model is capable to solve different vibration problems with good accuracy and less computational
cost. The non-dimensional frequency value increases with increase in the aspect ratios and span-to-thickness
ratio and decreases with increase in curvature ratios because as the span-to-thickness ratio increases the
structure becomes thin. Parametric study performed on spherical panel gives a comprehensive understanding of
its behavior under physical conditions. The values of fundamental frequencies under different support condition,
ply configuration and different parameters provide useful contribution in developing an optimum design.
REFERENCES
[1] Cook, R.D.; Malkus, D.S. and Plesha, M.E. (2000). Concepts and applications of finite element analysis,
John Willy & Sons Pte. Ltd., Singapore.
[2] Cui, X.Y.; Liu, G.R. and Li, G.Y. (2011). Bending and vibration responses of laminated composite plates
using an edge-based smoothing technique. Engineering Analysis with Boundary Elements, 35, 818–826.
[3] Hatami, S.; Azhari, M. and Saadatpour, M.M. (2007). Free vibration of moving laminated composite plates.
Composite structures, 80, 609-620.
[4] Huu-Tai, T. and Seung-Eock, K. (2010). Free vibration of laminated composite plates using two variable
refined plate theory. Int. J. of Mechanical Sciences, 52, 626–633.
[5] Jones, R.M. (1999). Mechanics of Composite Materials, Taylor and Francis, U.S.A..
[6] Kant, T. and Swaminathan, K. (2001). Analytical solution for free vibration of laminated composite and
sandwich plates based on a higher refined theory. Composite structures, 53, 75-85.
[7] Khdeir, A.A. and Librescu, L. (1988). Analysis of symmetric cross-ply laminated elastic plates using a
higher-order theory Part II—buckling and free vibration. Composite Structures, 259–277.
[8] Mantari, J.L.; Oktem, A.S. and Soares, C.G. (2011). Static and dynamic analysis of laminated composite
and sandwich plates and shells by using a new higher-order shear deformation theory. Composite
[9] Matsunaga, H. (2002). Vibration of cross-ply laminated composite plates subjected to initial in-plane
stresses. Thin-Walled Structures, 40, 557–571.
[10] Matsunaga, H. (2007). Free vibration and stability of angle-ply laminated composite and sandwich plates
under thermal loading. Composite structures, 77, 249-262.
[11] Putcha, N. S. and Reddy, J. N. (1986). Stability and natural vibration analysis of laminated plates by using a
mixed element based on a refined plate theory. Journal of Sound and Vibration, 285-300.
[12] Reddy, J. N. and Liu, C. F. (1985). A higher-order shear deformation theory of laminated elastic shells. Int. J.
Engineering Science, 23, 319–30.
461
[13] Topal, U. and Uzman, U. (2007). Optimum design of laminated composite plates to maximize buckling load
using MFD method. Thin-Walled Structures, 45, 660–669.
[14] Xiang, S. and Kang, G. (2012). Local thin plate spline collocation for free vibration analysis of laminated
composite plates. European Journal of Mechanics A/Solids, 33, 24-30.
[15] Xiang, S., Bi, Z., Jiang, S., Jin, Y. and Yang, M. (2011). Thin plate spline radial basis function for the free
vibration analysis of laminated composite shells. Composite Structures, 93, 611–615.
462
Paper No.: 077
ENHANCEMENT OF SHEAR DUCTILITY FOR HSC BEAMS WITH

MINIMUM SHEAR REINFORCEMENT
Hipparagi A.K1, Injaganeri S.S2 and Mahadevgouda H3

1
Research scholar, Department of Civil Engineering, Basaveshwar Engg. College, Bagalkot, Karnataka, India
2
Professor, Department of Civil Engineering, Basaveshwar Engg. College, Bagalkot, Karnataka, India
3
Final year BE student, Dept. of Civil Engineering, Basaveshwar Engg. College, Bagalkot, Karnataka, India
Email: ashokhipparagi_05@rediffmail.com, injaganeri@gmail.com
ABSTRACT :
A comparative study of adequacy of minimum shear reinforcement (MSR) provisions as per different national
codes of practice for reinforced high strength concrete (HSC) beams is presented. Five full-scale beam
specimens having concrete compressive strength of 60 MPa, with constant values of shear span-to-depth (a/d)
ratio of 2.78 and longitudinal reinforcement of 1.26% have been tested under three point loading in
displacement control mode. Three beams were provided with MSR as specified by ACI-318-08, Eurocode-2
and IS: 456-2000 and one beam having horizontal stirrups in shear span along with vertical stirrups as per IS:
456 were tested to study the effect on shear strength and ductility. The behaviour of these four beams has been
compared with a beam without shear reinforcement. It is observed that the codal-provisions underestimate the
shear strength of beams for HSC and are to be re-evaluated. However, an increase in shear strength and ductility
has been observed for the beam with horizontal stirrups.
KEYWORDS: minimum shear reinforcement (MSR); diagonal cracking strength; ultimate shear strength;
horizontal stirrups; high strength concrete (HSC) beam.
1. INTRODUCTION
The use of high strength concrete (HSC) has increased considerably during the last few decades due to increase
in construction of high rise structures and bridges, as it can be produced reliably in the field using low
water-cement ratio and admixtures. However, an increase in the strength of concrete increases its brittleness
leading to some concerns about the application of HSC. Further, the behavior of reinforced concrete (RC)
members is a very complex phenomenon mainly because of its non-homogeneity, presence of cracks,
reinforcement and the nonlinearity in its material response.
The transversely loaded RC beams may fail in shear before attaining their full flexural capacity, if they are not
adequately designed for shear. Unlike flexural failures, shear failures are very sudden and catastrophic. A
thorough knowledge of the different modes of shear failures and the mechanisms involved in resisting shear is
necessary to avoid such failures.
The minimum shear reinforcement (MSR) in RC beams has been studied by the researchers over the last few
decades [6,13,15 and 20]. It has been observed that the minimum quantity of shear reinforcement specified by
the codes of practice needs to be increased as the compressive strength of concrete increases [6]. The provisions
of ACI 318-08, BS-8110 and Eurocode2 for MSR are based on this theory, but the limiting reinforcement ratios
provided in IS: 456-2000 are found to depend only on the yield strength of reinforcement and independent of
concrete strength. The current shear provisions incorporated in codes of practice are based on tests carried out
on beams of normal strength concrete (NSC) with compressive strength lower than 40 MPa. The IS: 456-2000
provisions for MSR, if extrapolated for HSC members, may result in compression failure of concrete, rather
than the desired ductile failure of steel reinforcement [18].
2. RESEARCH SIGNIFICANCE
The shear failure in a beam without web-reinforcement is sudden and brittle. Therefore, it is necessary to
provide a minimum amount of shear reinforcement to prevent sudden shear failure on the formation of first
diagonal tension cracking and in addition, to adequately control the diagonal tension cracks at service load
levels. Thus, the minimum area of web reinforcement is intended to ensure that the capacity of the member after
cracking, exceeds the load at which inclined cracking occurs.
In this paper an attempt has been made to compare the adequacy of MSR specified by codes of practice viz.
ACI-318-08, Eurocode-2 and IS: 456-2000 for HSC beams. Also, a beam with the horizontal stirrups along with
the vertical stirrups as per IS: 456 is tested. The study on the provision of horizontal stirrups is to stitch the
diagonal cracks, delay the occurrence of diagonal crack and also to reduce the crack width. The influence of
horizontal stirrups on shear strength of beams made of HSC is investigated. In particular the cracking load,
ultimate shear strength, deflection at mid-span and the propagation of diagonal shear crack for all the five beams
are studied. The enhancement of shear ductility due to horizontal stirrups is also studied.
3. MINIMUM SHEAR REINFORCEMENT REQUIREMENTS
Due to the high tensile strength in HSC members, a higher cracking shear resistance is expected owing to which
large amount of MSR will be required. The provision of an appropriate amount of MSR also helps to control
bond splitting cracks that otherwise could lead to brittle shear bond failures [20].
Shehata et.al. [17] suggested a theoretical approach for minimum shear steel ratios (MSSR) that are required for
ductile behavior at failure in bending, shear and torsion for beams of different concrete strengths. The authors
have presented separate expressions for MSSR from the analysis based on the diagonal cracking load and also
based on truss model for beams cracked in flexure.
The MSSR for a cracked beam in bending is given by Eq.1 based on the tensile strength of concrete proposed by
CEB-FIP MC90 [4]
.
, = 0.06 (1)
Cladera and Mari [3] proposed Eq.2 for the MSR, relating tensile strength of concrete using Artificial Neural
Network (ANN),

, = " (2)
7.5 !
Eq.2. is based on tensile strength of concrete given by Eq.3. and Eq.4. proposed in Concrete Society Technical
Report 49 [5].
*
= 0.3 ' ( MPa if 01
)
≤ 60 MPa (3)
= 0.58 4
5
( MPa if 01 > 60 789 (4)
When the applied shear stress is greater than 0.5 times the concrete strength, the MSR to be provided according
to ACI-318-08 [1] and Eurocode2 [8] are given in Eq.5. and Eq.6 respectively.
464
0.94 ( 1
≥ ≥ (5)
16 ! 3!
0.084 (
≥ (6)
!
MSR specified by CSA A23.3-04 [7] is given by Eq.7
4 ′
= 0.06 (7)
!
As per the provisions of IS:456-2000 [11] and BS: 8110-1-1997 [2], when the calculated nominal shear stress (τv)
is greater than half the design shear strength of concrete (τc) the MSR to be provided are given in Eq.8 and Eq.9
respectively.
0.4 0.46
≥ = (8)
0.87 ! !
A?@ 0.4
≥ (9)
bB s@ 0.95fD
Equations 2, 5, 8 and 9 indicate that the quantity of minimum stirrups required corresponding to a nominal shear
stress resisted by stirrups equal to 0.35 MPa, 0.34 MPa, 0.40 MPa and 0.40 MPa respectively. The shear
reinforcement index (SRI) adopted by different codes of practice (Eq.10), are calculated and the comparison is
shown in Fig. 1.

EFG = ! (10)
( ∗ )
0.9 Euro
0.8 Cladera
SRI = (As*fy)/(b*sv) MPa
ACI
0.7 CSA
0.6 IS & BS
0.5
0.4
0.3
0.2
20 30 40 50 60 70 80 90 100
Compressive strength f'cMPa
Fig. 1 Comparison of SRI values suggested by different codes
It is evident from Fig.1 that there is vast disparity in the provision of MSR provided by different codes of
practice for HSC. The difference between SRI values specified by ACI, CSA codes and that of IS and BS codes
is relatively less for normal strength concrete i.e. for I ranging between 20 to 50 MPa. Whereas the variation
in SRI is large for values of I greater than 50 MPa because the SRI values specified by ACI and CSA codes
are based on members made of normal strength of concrete and small database of HSC beams. However, the
SRI specified by IS and BS codes are independent of the compressive strength of concrete and therefore it
doesn’t seem to be conservative especially for members made of HSC. Fig.1 also indicates that Eurocode2
specifies larger SRI values than that of IS and BS codes. The experimental work referred in this paper is a part
465
of research work carried out on the above issues related to MSR and shear ductility of HSC beams.
4. SHEAR DUCTILITY AND RESERVE STRENGTH OF RC BEAMS
The ductility is a very important parameter for members made of HSC. At present, the studies are limited to the
diagonal cracking and maximum load stage. The available information regarding post peak deformation of RC
members for HSC is limited, which is essential for a better understanding of web reinforcement and failure
mechanisms in situations such as seismic conditions where RC members require higher ductility. The shear
ductility of RC beams has been studied by the researchers [19,14] and has been expressed in terms of shear
ductility index (SDI), reserve strength (RS) and reserve deflection (RD) etc.
Xie et al. (1994) defined member ductility as the ability of member to withstand load while incurring additional
deformation beyond the maximum load stage. The post-peak deformation characteristics were quantified
through shear ductility index (SDI) and it was observed that, the HSC beams with a/d value of 3, exhibit a
plastic post-peak response, when the shear reinforcement provided is twice that of ACI code.
Lee J.Y and Kim (2008) defined reserve strength (RS) as the ratio between the ultimate shear capacity (Vu) of
the beams with the MSR as per ACI 318-05 and that of the beams without shear reinforcement (Vuo). Similarly,
reserve deflection (RD) was defined as the ratio between the deflection corresponding to the ultimate load of
beams with shear reinforcement (∆K ) and the deflection of beams without shear reinforcement (∆KL ).
V
Reserve Strength = V W (11)
WX
∆K
Reserve Deflection = (12)
∆KL
The authors concluded that, both reserve strength and reserve deflections of RC beams increased with the
increase in ρl values but decreased with increase in a/d ratio. The authors are of the opinion that more tests are to
be conducted to confirm the effect of a/d ratio.
5. DETAILS OF EXPERIMENTAL WORK
5.1 Material Properties and Mix Design
The physical properties of materials used for concrete mix viz. ordinary portland cement (OPC-43 grade), fine
aggregate, coarse aggregate and fly ash are given in Table 1. The properties of steel used as reinforcement are
given in Table 2. The target compressive strength of 60.58 MPa for the HSC used in the test was achieved after
mix design trials according to IS:10262-1982 [12]. The final mix proportion adopted was 1:1.25:2.5 with w/c
ratio of 0.32 along with 10% cement replacement by fly ash (Class F). The super plasticizer (Glenium B-233)
was added to improve the workability. The details of mix proportions are given in Table 3. The compressive
strength of concrete cubes of size 150 mm x 150 mm x 150 mm was observed to be 49.84 MPa and 60.58 MPa
after 7 days and 28 days curing respectively.
Table 1. Physical properties of materials used for concrete mix
Cement Fine Aggregate Coarse Aggregate Fly ash

Type Sp.gr. Sp.gr. F.M. Sp.gr. F.M. Max. size mm Sp.gr.
OPC 43 3.10 2.63 3.27 2.80 3.27 20 & 12.5 2.06
Note: Sp.gr. = Specific gravity and F.M. = Fineness Modulus
466
Table 2. Properties of steel
Longitudinal reinforcement Shear reinforcement

! (MPa) ] (MPa) ! (MPa) ] (MPa)
556 624.56 412 504.67
Table 3. Concrete mix proportions
Grade of Type of W/C Unit Weight (kg/m3)

Concrete Cement ratio C S CA FA % SP %
M60 OPC 43 0.32 336.8 421.05 1052.63 10 0.6 - 0.7
Note: OPC = Ordinary Portland Cement, W/C = water to cement ratio, C= cement, S= Fine aggregate (sand),
CA=coarse aggregate, FA= fly ash (class F) content, SP = super plasticizer (Glenium – B233) percent by weight of cement.
5.2 Details of beams
The details of the beams tested are given in Table 4. The overall length and depth are kept constant for all the
beams equal to 1650 mm and 300 mm respectively. However the width of (J-IS H) beam provided with
horizontal stirrups was kept as 250 mm for practical convenience. All the beams were provided with 6 bars of
12 mm diameter longitudinal bars (ρl = 1.26%) having sufficient anchorage at the ends to avoid anchorage
failure and nominal 2 bars of 6 mm diameter on compression side.
Table 4. Properties of reinforced concrete beams

Beam b (mm) d (mm) d/b le (mm) le/d a/d Ast Asv
J – WS 200 269 1.34 1500 5.57 2.78 6 #12 -
J – IS 200 269 1.34 1500 5.57 2.78 6 #12 #6 @ 200 mm c/c
J - ACI 200 269 1.34 1500 5.57 2.78 6 #12 #6 @ 130 mm c/c
J – Euro 200 269 1.34 1500 5.57 2.78 6 #12 #6 @ 160 mm c/c
J – IS H 250 269 1.17 1500 5.57 2.78 6 #12 #6 @ 200 mm c/c
5.3 Test Results
The cube compressive strength of concrete (fck) is obtained by testing control cubes after 28 days curing along
with beams. The beams are tested in flexure testing machine under displacement control and three point loading.
The load when diagonal cracking occurs (Vcr,exp), ultimate load (Puexp) and mid span deflections are observed.
The ultimate shear capacities of the beams as predicted by the codes (Vupred) are compared with experimental
values (Vuexp) in Table 5a and 5b. The mid span deflections of the beams (∆exp) measured during testing, reserve
strength and reserve deflection calculated from Eq. 11 and Eq.12 are listed in Table 6.
5.4 Discussion of Test Results
From the definitions of reserve strength (RS) and reserve deflection (RD) proposed by Lee and Kim (2008) and
referring the test results given in Table 5 and 6, the HSC beams J-ACI and J-Euro exhibit a higher reserve
strength of 1.72 in comparison to beams J-IS (1.40) and J-ISH (1.46) respectively. The reserve deflection for
beams J-IS and J-ACI has been observed to be 1.48 and 1.46 respectively. But RD for beams J-Euro and J-ISH
467
are found to be 1.74 and 1.81 respectively indicating larger ductility compared to beams J-IS and J-ACI.
Comparing the values of RS and RD for beams J-IS and J-ISH, it is observed that the RS value for both beams
is nearly same but J-ISH beam with horizontal stirrups shows an increase of 22.3% in reserve deflection
capacity compared to beam J-IS even though the MSR provided for both beams is same. From the test results it
is observed that use of horizontal stirrups in HSC beams can improve the reserve deflection capacity indicating
a higher ductility. But to quantify it more tests are planned to be conducted in the next part of the research work.
Table 5 (a). Comparison of experimental and predicted shear strength of beams

Experimental Values Predicted Values
Beam ( (^ _ )àb (c _ )àb (8] )àb (VK )efg (vK )efg (VK )efg (VK )ghei (vK )ghei (VK )efg
(V0h )efg (VK )ghei
MPa
kN MPa kN kN MPa kN MPa
J – WS 59.60 80 1.49 161.52 80.76 1.50 1.01 49.92 0.93 1.62
J – IS 59.60 95 1.76 225.41 112.76 2.09 1.19 81.28 1.51 1.39
J - ACI 60.58 98 1.82 277.85 138.92 2.58 1.42 119.21 2.22 1.17
J - Euro 60.58 70 1.30 278.32 139.16 2.59 1.99 90.12 1.68 1.54
J – IS H 60.58 80 1.19 235.00 117.50 1.75 1.47 88.17 1.31 1.33
Table 5 (b). Comparison of experimental and predicted shear strength of beams

Experimental Predicted Values by ^],àb`_ ` jk
Values Standard Codes of practice ^],b_`l `l
Beam (
MPa (^ _ )àb (VK )efg (VK )mn (VK )opm (VK )qKhL (VK )efg (VK )efg (VK )efg
kN kN kN kN kN (VK )mn (VK )opm (VK )qKhL
J – WS 59.60 80 80.76 49.92 70.99 54.85 1.62 1.14 1.47
J – IS 59.60 95 112.76 81.28 102.33 83.07 1.39 1.10 1.36
J - ACI 60.58 98 138.92 98.17 119.21 98.26 1.42 1.17 1.41
J - Euro 60.58 70 139.16 89.12 110.17 90.12 1.56 1.26 1.54
J – IS H 60.58 80 117.50 88.17 120.08 91.87 1.33 0.98 1.28
Table 6. Comparison of mid-span deflections, reserve strength and reserve deflection
(8] )àb (^] )àb ∆àb rs ∆s

Beam ( FE = Fu =
MPa kN kN mm rst ∆st
J – WS 59.60 161.52 80.76 3.60 1.00 1.00

J – IS 59.60 225.41 112.76 5.32 1.40 1.48
J - ACI 60.58 277.85 138.92 5.26 1.72 1.46
J – Euro 60.58 278.32 139.16 6.27 1.72 1.74
J – IS H 60.58 235.00 117.50 6.52 1.46 1.81
6. CONCLUSIONS
Based on the experimental work carried out in the laboratory the following conclusions have been drawn.
1. It can be observed from the comparison of ultimate shear strength between experimental values and the
values predicted by different codes of practice i.e. ACI 318-08, Eurocode2 and IS 456-2000 that the
strength predicted by different codes of practice underestimate the strength when compared to
experimental values for beams made of HSC.
2. The provisions for MSR by ACI-318 and Eurocode2 are related to the compressive strength of concrete
but the predicted values of ultimate shear strength are under estimated by 13% (ACI) and 41% (Euro)
than experimental values.
3. The beam with MSR as per IS: 456-2000 shows that the quantity of MSR suggested by IS code is
inadequate because the failure was sudden and catastrophic.
468
4. The beam with the horizontal stirrups in comparison to the similar beam without horizontal stirrups
shows a delay in the propagation of diagonal shear cracks with smaller crack widths. Hence, a need is felt
to test more number of beams with horizontal stirrups.
5. The use of horizontal stirrups in HSC beams can improve the reserve deflection capacity. Further
investigations are required to quantify the ductility ratio.
7. SYMBOLS USED
a = shear span
= Area of longitudinal reinforcement
= Area of shear reinforcement
, = Area of minimum shear reinforcement
b = width of beam
D = overall depth of beam
d = effective depth of beam
(9/w) = shear span to depth ratio
( = cube compressive strength of concrete
I = cylinder compressive strength of concrete
= tensile strength of concrete
! = yield strength of steel
] = ultimate strength of steel
x` = effective span of the beam
= spacing of vertical stirrups
(8] )àb = Load applied on the beam at failure
c_ = diagonal cracking strength
c] = ultimate shear strength
(^ _ )àb = cracking load applied on the beam
(c _ )àb = cracking strength of the beam
(^] )àb = ultimate shear force on the beam
(c] )àb = ultimate shear strength of the beam
(^] )b_`l = predicted value of ultimate shear force on the beam
(^] )yz = predicted value of ultimate shear force as per IS: 456
(^] ){|y = predicted value of ultimate shear force as per ACI: 318
(^] )}]_~ = predicted value of ultimate shear force as per Eurocode2
(c] )b_`l = predicted value of ultimate shear strength of the beam
(∆)àb = deflection measured at the mid span of the beam
k = percentage of longitudinal reinforcement
, = minimum shear steel ratio
REFERENCES:
[1] ACI Committee 318, (2008) “Building Code requirements for Structural Concrete (ACI 318-08) and
Commentary”, American Concrete institute, Farmington Hills, MI.
[2] BS 8110-1-1997 Structural use of concrete, Part 1: Code of Practice for Design and Construction, British
Standards Institution, London
[3] Cladera.A,Mari A.R., (2005) “Experimental study on high strength concrete beams failing in shear”.
Engineering Structures, 27, 1519-1527.
469
[4] Comitè Euro-International du Bèton, Bulletin d’Information No. 213/214,’CEB-FIP Model Code 90’, May
1993.
[5] Concrete Society Technical Report 49. (1998) Design guidance for high strength concrete. United Kingdom.
[6] Daou Y.A. and Ghanem H., (2003) “Study on Minimum Shear Reinforcement in High Strength Concrete
Beams”, Alexandria Engineering. Journal, 42(4), 601-612.
[7] Design of concrete structures, (2004), CSA A23.3-2004, Canadian Standards Association, Ontario
[8] Eurocode 2 (2002) Design of concrete structures - Part 1-1: General rules and rules for buildings”, EN
1992-1-1:1992, European Committee for Standardization (CEN), Brussels, Final draft December 2003.
[9] G.Appa Rao, and S S Injaganeri, (2011), Evaluation of size dependent design shear strength of reinforced
concrete beams without web reinforcement, Sadhana, 36(3), 393–410.
[10] Hadi M.N.S. and Elbasha.N, (2008) “The effect of helical pitch on the behaviour of helically confined
HSC beams”, Construction and Building Materials, 22(5), 771-780.
[11] IS 456:2000, (2000) Indian Standard Code of Practice for Plain and Reinforced Concrete, Bureau of Indian
Standards, New Delhi.
[12] IS:10262-1982, Recommended Guidelines for Concrete Mix Design, Bureau of Indian Standards, New
Delhi, 1983
[13] Johnson Mark.K, Ramirez J.A. (1989) “Minimum Shear Reinforcement in beams with HSC”, ACI
Structural Journal, Jul.-Aug., 376-382.
[14] Lee J.Y and Kim, (2008) “Effect of Longitudinal Tensile Reinforcement Ratio and Shear Span-Depth ratio
on Minimum Shear Reinforcement in Beams”, ACI Structural Journal, March-April, 134-144.
[15] Ozcebe G., Ersoy U. and Tankut T., (1999) “Evaluation of Minimum Shear Reinforcement Requirements
for Higher Strength concrete”, ACI Structural Journal, May-June, 361-368.
[16] Recommended guidelines for concrete mix design, IS:10262-1982, Bureau of Indian Standards, New Delhi.
[17] Shehata I.A.E.M, Shehata L.C.D, Garcia S.L.G, (2003) "Minimum steel ratios in reinforced concrete
beams made of concrete with different strengths-Theoretical approach", Materials and structures, Jan.-Feb.,
36,pp.3-11
[18] Subramanian N, (2010) “Limiting reinforcement ratios for RC flexural members”, The Indian Concrete
Journal, Sept., 71-80.
[19] Xie Yuliang , Ahmad Shuaib. H., Yu Tiejun, Hino S. and Chung W, (1994) “Shear Ductility of Reinforced
Concrete Beams of Normal and High Strength Concrete”. ACI Structural Journal, March-April, 140-149.
[20] Yoon Young-Soo, Cook William. D. and Mitchell Denis, (1996) “Minimum Shear Reinforcement in
Normal, Medium, and High-Strength Concrete Beams”, ACI Structural Journal, Sept-Oct, 576-584.
470
Paper No.: 083
SPECTRAL STRAIN ENERGY BASED APPROACH FOR SYSTEM

IDENTIFICATION IN STRUCTURES
1 2
J S Ali S Gupta
1
Asst.Professor, Dept. of Civil Engineering, Aliah University, Kolkata, India
2
Student, Dept. of Construction Engineering, Jadavpur University,Kolkata,India
Email: jsa_2005@rediffmail.com
ABSTRACT :
Structural Health Monitoring (SHM) has become significantly important in recent times. Damage in structure
alters the dynamic characteristics of the system. The change of structural parameters is characterized by the
changes in the Eigen parameters. Numerical method incorporating inverse dynamic technique is a global
approach. But measurement at all degrees of freedom does not seem to be feasible in real situation due to
practical constraints. In this paper, a vibration response parameter known as Spectral Strain energy (SSE) is
proposed for structural damage identification in a beam element. First, Displacement Response Power Spectral
Density is used to determine Moment Power Spectral Density (MSD) and Curvature Power Spectral Density
(CSD). A statistical parameter known as Mean-Square Value (MSV) of the Response Spectral Density is then
obtained and used to derive SSE. The level of sensitivity to damage is compared with the results obtained from
the Modal Strain Energy (MSE) method.
KEYWORDS: Structural Health Monitoring, Spectral Strain Energy, Modal Strain Energy
1. INTRODUCTION:
Structural damage is generally considered as weakening of the structural element that negatively affects its
performance and can be defined as any deviation in structure’s original geometric or material properties
that may cause undesirable stresses, displacements or vibrations on the structure. Most civil engineering
structures such as bridges, buildings, chimneys, water and gas pipe lines etc transmission towers are
decayed gradually during their service lives in age due to misuse, lack of repair and in some cases because
they were not designed for present demand. Even if the damage area is too small, the identification of the
location and the degree of damage is significantly important with respect to evaluation of safety & future
usefulness of structure, effective & economical restoration and to exploit better life expectancy of
structure. The structural health can also deteriorate suddenly due to environmental causes such as
earthquake, wind, fire, corrosion, blast etc. In all the cases, damage can severely affect safety and
serviceability of the structure and the interest in the ability to monitor a structure and detect damage at the
earliest possible stage is pervasive throughout the civil, mechanical and aerospace engineering communities.
Thus to get an idea about the strengthening required for reformation of a structure or to decide whether the
structure can safely be used at all or not, identification of damage is a prerequisite. Visual inspection has
always been the most common method used in detecting damage on a structure. But efficiency of the visual
inspections reduces due to increased size and complexity of structure. There are various Non-Destructive
Tests (NDT) available to identify the damage, which can locate damage but cannot quantify the damage
scenarios.
Extensive research has taken place in damage detection using measured response of damage structure and
the development of system identification methods in finite element platform. Senayei et. al. (1994) expressed a
static identification algorithm from static strain data, normally an iterative procedure is used to identify the
element stiffness parameters involving simple static test data, In dynamic identification, damage detection using
modal data have been extensively used in recent past, though transient data were also adopted. A detailed review
on damage detection is provided by Doebling et al (1998) that includes various approaches in system
identification algorithm covering static as well as dynamic data. Maia et al. (2002) expressed a series of
numerical simulation of a simple beam to compare several damage detection techniques based on mode shape
changes. Even though numerous techniques and various response parameters have been used for damage
identification studies, most of them are developed for one-dimensional structures, such as beams, frames and
truss structures. Cawley and Adams were probably the first who developed a damage detection algorithm for
plate structures based on frequency shifts. Lee and Shin presented a methodology for damage identification in
plates using measured modal data from the undamaged state and FRF from the damaged state of the structure
investigated. Li et al. presented a strain mode technique for damage identification in plate-like structures.
In present study, damage detection in different techniques like modal flexibility, curvature mode shapes,
modal strain energy are developed to detect the location and magnitude of damage of an existing structure from
known modal data of the existing structure. In this paper, spectral strain energy (SSE) is proposed as a sensitive
and effective damage indicator for damage identification in frame member. SSE derived from two parameters
known as moment response power spectral density (MSD) and curvature response power spectral density (CSD)
is implemented in the context of a non-model-based approach where damage detection and localization is the
main focus. The method can also be used to monitor both global and local changes in structures
simultaneously .In the past, Cornwell et al. presented a similar technique based on MSE which is derived from
the mode shapes obtained from undamaged and damaged states to identify structural damage using a statistical
hypothesis test .The proposed method which is based on spectral analysis has many advantages: it takes into
account all modal parameters; with the only assumption that the excitation force is stationary, ergodic white
noise; and this method is more flexible since it can also be implemented in the frequency domain as well as the
modal domain.
2. THEORETICAL FORMULATION
The equation of motion of a discreet finite element model can be expresses as,
[ m ]{&&u} + [k]{u} = {f} (1)
where [k] and [m] are the stiffness and consistent mass matrices of the element respectively. {u} and {F}
contain the nodal displacements and force vectors. Assembling the element equations and applying
boundary conditions, one can obtain the following equilibrium equation
..
[ M ]{U} + [K]{U} = {F(t)} (2)
where [M], [K], {U} and {F(t)} indicates the global mass matrix, stiffness matrix, nodal displacement and
force vector respectively. For free vibration, substitution of F(t) = 0 and {U}={φ exp(iωt)} in equation (2)
yields,
( [K]- ω2 [M] ){φ}= {0} i.e., [K]{φ} = λ[M]{φ}, where λ = ω2 . (3)
Equation (3) is a generalized eigenvalue problem that can be solved for a sequence of λ and {φ}. The
eigenvector {φ} is the mode shape and eigenvalue λ is the square of the associated natural frequency in
radians per seconds. The scaled and the un-scaled mode shape are related by the equation:
{ϕ }
{φn } = (4)
{ϕ} [ M ] {ϕ}
*
1
so that the scaling factor is expressed as: α = and {ϕn } = α {ϕ}
{ϕ } [ M ] {ϕ }
*
472
From normalized mode shape vector {φ} one can calculate the Modal Assurance Criterion (MAC) and
Co-Ordinate Modal Assurance Criterion (COMAC) between Damaged and Undamaged model.
2
n
∑ (ϕ
j =1
d ) j (ϕu ) *
j
MAC (d , u ) = (5)
 n * 
n
* 
∑ d j d j   ∑ (ϕu ) j (ϕu ) j 
(ϕ ) (ϕ )
 j =1   j =1 
l
∑ (ϕ
2
d ) jl (ϕu )* jl
COMAC (i ) = l
l =1
l
(6)
∑ (ϕ
l =1
d ) jl 2
∑ (ϕ )
l =1
u jl
2
Here, ϕ denotes the mode shape, the subscript d and u denotes damaged and undamaged state
respectively.MAC is a scalar quantity defines with respect to the pair of modes, and when these modes are
correlated MAC would be close to 1.On the other hand, COMAC is defined with respect to asset of l correlated
mode shapes. If there exist no discrepancies between the two mode shapes, The COMAC would be unity. With
the mode shapes normalized to unity as ϕ M ϕ = I , the stiffness and the flexibility matrices are related to the
T
modal data. Thus Modal Flexibility Matrix is
n
1
F =∑ (ϕiT ϕ )
i =1 wi 2 (7)
Where ϕ is the mode shape matrix ϕ = [ϕ1ϕ2 .....ϕ n ] . From the flexibility matrix the change in the
∆ = Fu − Fd F Fd
flexibility matrix ∆ is obtained as where u and are the flexibility matrices at
δj
undamaged and damaged state. For each degree of freedom j, let be the maximum absolute of the element
δ j = max δ j
in the corresponding column of ∆ i.e., . Sometimes change in flexibility matrix indicate damage
in the beam.
Again from the displacement mode shapes, obtained from the finite element analysis, curvature mode shape is
obtained numerically by using a central difference approximation as
ϕ ii = (ϕi +1 − 2ϕi + ϕi −1 ) / h 2 (8)
where h is the length of the element.
Sometimes the Modal Strain Energy Change Ratio (MSECR) is employed to locate the damage. The basic idea
of Modal Strain Energy (MSE) is defined as the product of the elemental stiffness matrix and the second power
of the mode shape component. For the jth element in the i th mode, the MSE before and after the occurrence of
damage is given by
MSEij = φi* K jφi

MSE d ij = φi d * K jφi d (9)
473
where i denotes the i th mode shape of the undamaged state; K j denotes the undamaged elemental stiffness
of the j th element; superscript * denotes the transpose, and superscript d denotes the damaged state. In the
d
modal strain energy of the damaged case, MSE ij . The stiffness of damaged case K d j is approximated by
K j since the damaged stiffness matrix is not known in advance. The MSECR is defined as
( MSE d ij − MSEij )
MSECRij =
MSEij
In this paper, a response parameter known as spectral strain energy (SSE) is implemented for structural damage
identification. SSE is determined from bending MSD and CSD. The following assumptions are made to obtain
MSDs and CSDs: (i) the structure is assumed to be lightly damped; (ii) modes are well-separated; (iii) sufficient
number of significant modes are identified and used to generate respective response parameters; (iv) the
unknown input PSD is assumed to be stationary ergodic random white noise, i.e., statistical properties of the
input PSD remain constant for all damage conditions. Therefore, the SSE proposed in this paper is given by
1
( ) (σ ) dA
*
statistical properties of MSD and CSD, as follows: U= ∫ σ m2 k
2
2A
(10)
where σ m 2 and σ k 2 are Mean Square Value (MSV)s of Moment Power Spectral Density (MSD) and
Curvature Power Spectral density(CSD), respectively. The statistical properties of σ m 2 and σ k 2 , known as
zero-order spectral moment or mean power of MSD and CSD are computed using the formula widely used in
+α
1
random vibration theory for zero-mean Gaussian spectral density . σm 2
= ∫ S (ω ) dω ,
2π
MM
−α
+α
1
σk2 = ∫ S (ω ) dω (11)
2π
kk
−α
where S (ω ) MM and S (ω ) kk are response MSD and CSD, respectively. ω is the circular frequency. For a
linear system and a stationary random process, input–output relationships for MSD, CSD and MSV are obtained
from the theory of random vibrations as
S (ω ) MM = H (ω ) M S (ω ) ff and S (ω ) kk = H (ω ) k S (ω ) ff
2 2
(12)
+α
( − iϖ t )
where S (ω ) ff = F (ω ) ; F (ω ) = ∫α f ( x, t )e dt . S (ω ) MM and S (ω ) kk
2
are MSD and CSD at grid
−
point j. H (ω ) M and H (ω ) k are the FRFs for moment and curvature responses, respectively. S (ω ) ff is
the input PSD; f ( x, t ) is the input excitation force; and F (ω ) is the corresponding Fourier spectrum.
3. RESULT AND DISCUSSIONS

Numerical Example:
To demonstrate the identifiability and accuracy of the proposed methods as presented in earlier sections a
simply supported beam of uniform cross-section discredited by six beam elements as shown in figure 1 is
taken up. Following numerical data have been used for undamaged beam: Length (L) is 6.0m, modules of
elasticity (E) is32Gpa; mass density is 2500 Kg/m3 width is 0.25m and height is 0.2m.
474
1 2 3 4 5 6
Figure: 1 Simply Supported beam having 6 elements
Table1: Natural frequencies of first five modes for the Intact and damaged beam damaged at second element
(50% reduction in E).
% Change in Frequency for damage at Second

Mode No element
(Damage=5%) (Damage=20%) (Damage=50%)
1 3.14 2.03 7.54
2 0.44 3.49 11.10
3 0.79 1.85 5.61
4 0.43 1.25 4.81
5 0.27 1.88 5.86
Table2: Modal Assurance Criterion (MAC) values for the Intact and the Damaged Displacement Mode Shapes
(at second element)
Damage MAC
Percentage
5% 0.999
20% 0.996
50% 0.997
Table 3: Co-Ordinate Modal Assurance Criterion (COMAC) values for the Intact and the Damaged
Displacement Mode Shapes (50% reduction in E at second element)
Node No 1 2 3 4 5 6 7
COMAC Values 1.0 1.0 1.0 0.99 1.0 1.0 1.0
Figure: 2 Change in Modal Curvature for different modes when damage at second
element (50% Damage)
475
Table 4: MSECR of first three modes for second element for various damages at second element.
Damage(%) MSECR
at Second FIRST SECOND THIRD
element MODE MODE MODE
5% 0.0913 0.0583 0.045
20% 0.454 0.2648 0.2062
50% 2.0244 0.8846 0.7822
Figure: 3 Plot of Spectral Density of Displacement at 6 th DOF for varying damage at second element
4. CONCLUSION
It has been observed that for a simple frame and beam member frequency changes are observed when reduction
of the structural properties like E changes. But for beam MAC and COMAC are almost constant .So, by
observing the frequencies and mode shapes one cannot localize the damage but from the curvature mode shape,
modal flexibility matrix and modal strain energy we can get better result. With the help of spectral strain energy
method elemental damage can be easily identified.
REFERENCES
[1]Sanaye, M., Scampoli, S., Structural Element Stiffness Identification from Static Test Data, Journal of
Engineering Mechanics Division, ASCE, Vol. 117, No. EM5.
[2]Pandey, A.K, et al. Damage Detection from Change in Curvature Mode Shapes, Journal of Sound and
Vibration, (1991) 145(2),321-332.
[3]Pandey, A.K, Biswas,M. Damage Detection in Structures Using Change in Flexibility, Journal of Sound and
Vibration, (1991) 169(1),3-17
[4]Maia,N,M,M et al. Damage Detection in Structures: From Mode Shape to Frequency Response Function
Methods. Mechanical System and Signal Processing, (2003) 7(3) 489-498.
476
[5]W.L. Bayissa, N. Haritos. Structural damage identification in plates using spectral strain energy analysis.
Journal of Sound and Vibration 307 (2007) 226–249.
[6]van Keulen, F., Haftka, R.T. and Kim, N.H., Review of options for structural design sensitivity analysis. Part
1: Linear systems. Comput. Methods Appl. Mech. Eng. v194. 3213-3243.
[7]Adelman, H.M. and Haftka, R.T., Sensitivity analysis of discrete structural system. AIAA J. v24. 823-832.
[8]Design Sensitivity Analysis of Structural Systems, in. 1986. Academic Press, London.
[9]Yan, W.J. and Ren, W.X., A direct algebraic method to calculate the sensitivity of element modal strain
energy. Int. J. Numer. Methods Biomed. Eng. v27. 694-710.
[10]Hesrn, G. and Testa, R.B., Modal analysis for damage detection in structures. J. Struct. Eng. (ASCE). v117.
3042-3063.
[11]Lim, T.W. and Kashangaki, T.A.L., Structural damage detection of space truss structures using best
achievable eigenvectors. AIAA J. v32. 1049-1057.
[12]Doebling, S.W., Hemez, F.M., Peterson, L.D. and Farhat, C., Improved damage location accuracy using
strain energy-based mode selection criteria. AIAA J. v35. 693-699.
[13] Shi, Z.Y., Law, S.S. and Zhang, L.M., Structural damage detection from modal strain energy change. J. Eng.
Mech.-ASCE. v126. 1216-1223.
[14] Shi, Z.Y., Law, S.S. and Zhang, L.M., Improved damage quantification from elemental modal strain energy
change. J. Eng. Mech.-ASCE. v128. 521-529.
[15] Bayissa, W.L. and Haritos, N., Structural damage identification in plates using spectral strain energy
analysis. J. Sound Vib. v307. 226-249.
477
Paper No.: 085
TOWARDS AN ECO-FRIENDLY CONCRETE

WITH WASTE GLASS AND RICE HUSK ASH
A. Syed Mohamed1, C. O. Arun2

1
Post Graduate student, Department of Civil Engineering, B.S.Abdur Rahman University, Chennai.
2
Assistant Professor, Department of Civil Engineering, NIT Warangal.
Email: a.syedmd_civil@yahoo.co.in, co_arun@yahoo.co.in
ABSTRACT
Present world faces lot of environmental threats. One of the biggest environmental problems which are growing
ever more critical is use of waste and non-recyclable materials as landfills. As a solution to this, the scientific
community suggests the use of these materials in construction industry. The present work is an effort in that
direction wherein two waste materials namely, colored glass and rice husk ash (RHA) has been used as partial
replacement for fine aggregate and cement respectively in concrete. Waste glass is found to be one of the better
replacement materials for sand, since glass is rich in silica. Rice husk ash is a by-product obtained from the
combustion of rice husk which consist of non-crystalline silicon dioxide with high specific surface area and high
pozzolanic reactivity. Thus colored glass and rice husk ash are effectively used in concrete. Moreover today’s
construction industry faces a great demand for raw materials. Usage of these waste materials in the concrete
reduces the amount of raw materials required. Furthermore reduction in the usage of cement ultimately reduces
the emission of CO2 from the cement industry. It is found that partial replacement of fine aggregate and cement
in concrete by powdered coloured glass and rice husk ash respectively can be done without loss of workability,
compressive strength, flexure strength or split tensile strength.
KEYWORDS: Rice Husk Ash, Glass, Alkali Aggregate Reaction, Pozzalona
1. INTRODUCTION
Concrete is the second largest consuming material after water by man. Cement is one of the main constituent of
concrete and which is highly responsible for emission of CO2. Nearly 2.35 billion tons of cement gets
manufactured annually. As per studies conducted, for every ton of cement 1 to 1.25 ton of CO2 is produced.
(Habeeb. et al., 2009). Reduction of CO2 emission in cement manufacturing process is one of the hot topics in
the present research work in this area. Moreover construction industry faces scarcity of raw materials like river
sand. Sand mining is becoming a huge environmental threat. The demand for sand is increasing in industry
causing the sand cost to increase enormously and this affects the construction industry as a whole. This
necessitates cut down of the amount of raw materials used in concrete, especially river sand, without affecting
the properties. One of the best solutions can be replacement the raw materials either partially or fully by some
low cost materials which can impart the required properties. Meantime another environmental threats that world
faces is disposal of waste and non-recyclable materials. The result of these materials in construction industry
can reduces the disposing problems to some extent. The different waste or non-recyclable materials that can be
used in concrete are glass, rubber, stone dust, fly ash, bamboo, rice husk ash, slag from steel mills, silica fume.
(M.S. Shetty., 2005).
The recycling of waste glass causes major problem in large municipal areas, especially the colored waste glass
which have low recycle rate and thus dumped as landfill. Generally colored glass scraps are not recycled,
instead they are sent to land fill. Waste glass is found to be one of the better replacement materials for sand,
since glass having silica content of nearly 70% (Zainab and Ismail. 2008). It is found that the addition of glass
material will lead to alkali-silica reaction (ASR), which results in volume expansion and formation of cracks in
concrete (Degirmenci et al., 2011). This can be avoided by adding some high pozzolanic material. Rice husk ash
is one of the good pozzolanic materials which are cheaply available in India. Use of rice husk ash as a
replacement for cement can reduce the ASR. Moreover it also reduces the overall carbon emissions from the
cement industry.
Studies on addition of rice husk ash as pozzolanic material in glass concrete is lacking in present literature. So
for the present work rice husk ash is opted as a pozzolanic material. This research work is an effort to
understand whether the replacement of river sand and cement can be done by powdered glass and rice husk ash
without losing the properties of concrete. Experiments have been designed based on Indian standard codes (IS
456:2000, IS 516:1959, IS 5816-1999, IS10262:2004).
2. GLASS AS FINE AGGREGATE IN CONCRETE.
The waste glasses being reduced to smaller size are used as partial replacement for fine aggregate and are
effective in enhancing the strength of concrete. Usage of glass in concrete lead to many advantages (Meyer et al.,
2001). Glass has basically zero water absorption, and one of the most durable materials known to man. With the
current emphasis on durability of high-performance concrete, it is only natural to rely on extremely durable
ingredients. The excellent hardness of glass gives the concrete an abrasion resistance. Glass aggregate improves
the flow properties of fresh concrete. The aesthetic potential of colour-sorted glass has barely been explored at
all and offers numerous novel applications.
The chemical composition (silica) of the glass determines the major strength properties of the concrete mix.
The chemical composition of different glasses is listed in table 1 (Ahmad S., 2002).
Table-1 Chemical Composition of Glass

Composition
SiO2 Al2O3 TiO2 Cr2O3 Fe2O3 CaO MgO Na2O K 2O SO3
Glass type
Clear Glass 72.42 1.44 0.035 0.002 0.07 11.50 0.32 13.64 0.35 0.21
Brown Glass 72.21 1.37 0.041 0.026 0.26 11.57 0.46 13.75 0.20 0.10
Green Glass 72.38 1.49 0.04 0.130 0.29 11.26 0.54 13.52 0.27 0.07
2.1 Preparation of glass powder from waste glass
Initially, collected waste glasses have to be cleaned from dirt and other debris. After washing, glass pieces are
allowed to dry for an hour. This is followed by crushing them using three different equipments namely, ball mill,
roller crusher and jaw crusher. The finely crushed glass materials are then sieved using IS 4.75, 2.36, 0.85, 0.15
sieves. The size of fine glass adopted for the present work is in the range of 0.85-2.36.
2.2 Alkali Silica Reaction (ASR).
The main disadvantage that the researchers face with use glass in concrete is Alkali-Silica reaction (ASR). It is
the chemical reaction that takes place between the silica rich glass particles and the alkaline solution of concrete.
The build-up of ASR gel leads to volume expansion and results in crack formation in concrete. The particle size
of glass was found to have a major influence on ASR expansion (Ahmad.S, 2002).
Addition of pozzolanic materials is found to be the cheapest and more feasible one to prevent cracking in
479
concrete due to ASR gel. (Gerry Lee, et al., 2011, Ahmad Shayan, 2002). Rice husk ash, being a pozzolanic
material which is cheaply available in India naturally becomes one of the best choices.
3. RICE HUSK ASH (RHA) IN CONCRETE.
Rice husk is the outer covering of paddy grain which is obtained as by-product in rice mill. For every 1000kg of
paddy milled, 20% by weight of rice husk is produced (Habeeb, et al, 2009). If it is heated up to 7000C, RHA
will be formed. The husk is used as fuel in the rice mills to generate steam for the parboiling process.
Industries use rice husk as fuel in boilers and for power generation. In India nearly 20 million tons of (RHA) is
produced every year (Ganesan et al., 2007). The husk contains about 75% organic volatile matter, leaving 25%
to be converted into ash during the firing process, known as RHA. This RHA contains around 85-90%
amorphous silica and non-crystalline silicon dioxide with high specific surface area and high pozzolanic
reactivity (Kartini k, et al., 2006). Moreover RHA is a carbon neutral green product. These properties help RHA
to be used as pozzolanic material in mortar and concrete as a partial replacement for cement.
Cement
RHA
Figure-1 Comparative view of Cement and Rice Husk Ash
The specific gravity of RHA is 2.05, lower than that of cement which is 3.15 and its specific surface area is
about 20 m²/g to 50 m²/g (Satish D.K, et al). In general, the average particle size of RHA is between 5 to 10 µm
while that of cement is approximately 13 µm. The rice husk has a large dry volume due to its low bulk density
(90-150 kg/m³) and possesses rough and abrasive surface that are highly resistant to natural degradation.
(Kartini K., et al 2006). The chemical composition of rice husk ash is listed in table 2: (Habeeb. et al., 2009)
Table-2 Chemical Composition of RHA

Elements SiO2 Al2O3 Fe2O3 CaO MgO Na2O3 K 2O Loss of ignition
Mass Fraction (%) 88.32 0.46 0.67 0.67 0.44 0.12 0.035 5.81
There is a growing demand for fine amorphous silica in the production of special cement and concrete mixes,
high performance concrete, high strength, low permeability concrete, for use in bridges, marine environments,
nuclear power plants etc. Many researchers have studied the effects of RHA on normal concrete by partially
replacing cement. Studies conducted by (Alireze, et al., 2010) concluded that concrete containing 15% of RHA
showed maximum compressive strength. Also they reported that using 30% RHA as a replacement of part of
cement could be considered optimum for all content of W/C ratios in investigated mortars because of its high
value of compressive strength.
But RHA as a partial replacement for cement as well as a technique to reduce the ASR in glass concrete have
not been addressed in available literature. The objective of the present work is to prepare an eco-friendly
concrete by partially replacing fine aggregate and cement with finely crushed coloured waste glass and rice husk
480
ash respectively and to study the properties.
4. EXPERIMENTATION
To understand properties of fresh and hardened concrete, different tests are designed and conducted as per
Indian standard codes. Percentage replacement of river sand by glass is fixed as 25% and that of cement by
RHA is varied from 10% to 20% by weight. Design mix is prepared to proportionating concrete. For present
work M30 grade of concrete to be used. As per IS:10262-1982, mix proportion and the new proposed
ingredients are replacing old as shown in table-3.
Table-3 Mix proportion

Description Mix Binder (kg/m3) Fine Coarse
Notation aggregate aggregate
(kg/m3) (kg/m3)
Cement RHA Sand Glass
0%Glass+0%RHA CC 443 - 606 - 1138
25%Glass+0%RHA GCR0 443 - 454 152 1138
25%Glass+10%RHA GCR10 399 44 454 152 1138
25%Glass+15%RHA GCR15 376 67 454 152 1138
25%Glass+20%RHA GCR20 354 89 454 152 1138
4.1 Workability
Workability is often referred to as the ease with which a concrete can be transported, placed and consolidated
without excessive bleeding or segregation. Increases of water content makes concrete workable at the same time
high amount of water make bleeding of concrete. Hence good workable concrete can be made by choosing
proper water to binder ration.
In present work slump test is carried out to determine workability of concrete. Slump test is conducted on fresh
concrete and results are tabulated in table 4.
Table-4 Slump value for different w/b ratios

Proportion CC GCR0 GCR10 GCR15 GCR20
W/B ratio 0.42 0.42 0.48 0.52 0.58
Slump value (mm) 27.5 45.0 30.0 28.0 20.0
Figure-2 Workability Graph
481
It is observed that as the amount of rice husk ash increases the water binder ratio required to maintain the
workability which increases from 0.42 to 0.58 in order to enhance the fluidity and consistency of the mix.
4.2. Tests on Hardened Concrete
The hardened concrete after curing is tested for its strength using Universal Testing Machine. Cubes of side
150mm, cylinder of diameter 150mm & height 300mm and prism of size 500mm long & 100mm square cross
section were made. Three specimens prepared for each batch, curing is done for 3 and 28 days separately and
test is carried out to evaluate the strength of concrete. The tests conducted are as follows:
4.2.1. Compressive Strength
The compressive strength of the concrete of various mix ratios is shown in the table-5 for 3 and 28 days curing
period. The results shown were mean from 3 specimens in each batch.
Table-5 Compressive Strength Test of Concrete
Specimen Ultimate Compressive

Description
Name Load(kN) Strength(N/mm2)
CC_3 470 20.89
PCC
CC_28 882 39.00
25%GLASS GCR0_3 438 19.27
+0%RHA GCR0_28 1000 44.44
25%GLASS GCR10_3 350 15.56
+10%RHA GCR10_28 816 35.27
25%GLASS GCR15_3 398 17.69
+15%RHA GCR15_28 884 39.29
25%GLASS GCR20_3 444 19.76
+20%RHA GCR20_28 910 40.44
Figure-3 Compressive Strength
It is observed that there is a phenomenal increase in compressive strength of concrete specimen from 3days
482
curing to 28 days of curing. The addition of RHA does not affect the compressive strength of concrete much
compared to PCC. After the additions of rice hush ash even though the strength decreases initially from 44.44 to
35.27 in comparison with glassed concrete, it increases gradually with increase in % of RHA. Thus addition of
10 to 20% RHA will not affect the mean target strength.
4.2.2. Split Tensile Strength Test
Split tensile strength of concrete is evaluated from cylindrical specimen. As same as compressive strength mean
value of each mix proportion is shown as result of 3days and 28days curing period in table-6
Table-6 Split Tensile Strength Test of Concrete
Specimen Ultimate Split Tensile Strength

Description
Name Load(kN) (N/mm2)
CC_3 146 2.07

PCC
CC_28 162 2.30
25%GLASS GCR0_3 154 2.20

+0%RHA GCR0_28 184 2.60
25%GLASS GCR10_3 118 1.67

+10%RHA GCR10_28 168 2.37
25%GLASS GCR15_3 82 1.16

+15%RHA GCR15_28 136 1.92
25%GLASS GCR20_3 48 0.68

+20%RHA GCR20_28 94 1.33
Figure-4 Split Tensile Strength
483
It is observed that split tensile strength gained in initial ages is less for RHA added glass concrete. Test results
show that split tensile strength is decreasing as the proportion of RHA increases. It can be suggested that use of
RHA must be limited to 10 to 15% without much loss of split tensile strength.
4.2.3. Flexural Strength Test
Flexural strength is also known as modulus of rupture, bending strength, or fracture strength. It is the ability of a
beam to resist failure in bending. This is done by supporting the specimen by two bearing surfaces and applying
an axial compressive load at its center by either a third bearing load, or a further two bearing load points. The
flexural strength represents the highest stress experienced within the material at its moment of rupture.
Flexural strength is about 12 to 20% of compressive strength.
A flexure test produces tensile stress in the convex side of the specimen and compression stress in the concave
side. This creates an area of shear stress along the midline.
Table-7 Flexural Strength Test of Concrete
Specimen a Ultimate Flexural Strength

Description
Name cm Load(kN) (N/mm2)
CC_3 19.5 11.6 6.79

PCC
CC_28 22 14.6 7.30
25%GLASS GCR0_3 17.3 12.6 6.59
+0%RHA GCR0_28 18.3 12.2 6.69
25%GLASS GCR10_3 17.0 12.8 6.53
+10%RHA GCR10_28 17.5 11.8 6.19
25%GLASS GCR15_3 20.0 10.4 6.24
+15%RHA GCR15_28 26.0 11.0 5.50
25%GLASS GCR20_3 23 9.6 4.80
+20%RHA GCR20_28 23.2 8.0 4.00
Figure–5 Flexural Strength Test
484
From this result, it was observed that increasing the proportion of rice husk ash have no impact in increasing the
flexural strength of the concrete specimen. As it can be observed from the test results it is advisable to have a
RHA replacement of only 5% to10% by weight of cement without much loss of flexural strength.
5. CONCLUSIONS
From the various tests conducted, it was determined that the workability of the concrete remained same when
25% of fine aggregate is replaced with fine glass. But the workability got reduced when rice husk ash is added
as a partial replacement for cement. This is because fine glass has the ability of absorbing no water where rice
husk ash tends to absorb more. The compressive strength of concrete specimen has shown good results after 28
days of curing. It was observed that the increase in rice husk ash content, improved the compressive strength of
the specimen. But there was no appreciable improvement in both tensile and flexural strength of the specimen.
Study suggests that a replacement of 10 to 15% by weight of cement can be wisely chosen depending upon the
application. Thus to some extent scarcity of raw materials for concrete, the environmental threats caused by
cement manufacturing and disposal of waste glass can be addressed by addition of fine glass and rice husk ash
in concrete without losing the strength properties. Thus this project has emerged as a proving fact of
manufacturing an eco-friendly concrete of low cost.
REFERENCES:
[1] G.A. Habeeb and M.M. Fayyadh, (2009) Rice Husk Ash Concrete: the Effect of RHA average particle size
on Mechanical Properties and Drying Shrinkage, Australian journal of Basic and applied science,
3(3):1616-1622,
[2] C.Meyer, N.Egosi, C.Andela. (2001), Concrete with waste glass as aggregate. International Symposium
Concrete technology unit of ASCE and University of Dundee, March 19-20, 2001.
[3] Nurhayat Degrimenci, Arin Yilmaz, & Ozge Andic Cakir., (2011), Utilization of waste glass as sand
replacement in cement mortar, Indian journal of Engineering and Materials science, l.18, 303-308
[4] Zainab Z. Ismail, Enas A.Al-Hashmi., Recycling of waste glass as a partial replacement for fine aggregate
in concrete, Waste management 29 (2009) 655-659.
[5] Kartini, K., Mahmud, H.B., Hamidah, M.S.,(2006) Strength properties of Grade 30 Rice Husk Ash
Concrete, 31st Conference on Our World In Concrete & Structures, Singapore, 16-17 August 2006.
[6] Satish D. Kene, Pravin V. Domke, Sandesh D. Deshmukh, R.S.Deotale, Assessment of Concrete Strength
Using Fly Ash and Rice Husk Ash, International Journal of Engineering Research and Applications
(IJERA) vol.1, Issue 3, pp.524-534.
[7] Alireza, N. G., Suraya A. R., Farah N. A. A., Mohammad A. M.S., (2010), Contribution of Rice Husk Ash
to the Properties of Mortar and Concrete: A Review, Journal of American Science, 6(3):157-165.
[8] Ahmad S., (2002) Value-added Utilisation of waste glass in concrete. IABSE Symposium Melbourne.
[9] M.S.Shetty., (2005)Concrete Technology Theory and Practice, S.Chand & Company ltd, New Delhi.
[10] Ganesan, K., K. Rajagopal, K. Thangavel, (2008). Rice husk ash blended cement: Assessment of optimal
level of replacement for strength and permeability properties of concrete. Construction and Building
Materials, 22(8): 1675-1683.
485
Paper No.: 086
A STUDY ON ANALYSIS OF TURBO GENERATOR FOUNDATION RESTING

ON PILE FOUNDATIONS FOR VARIOUS EARTHQUAKE ZONES IN INDIA
K.V.S.N.MURTHY 1 and Dr.B.DEAN KUMAR.2

1
PG Student JNTUH College of Engineering, Hyderabad, A.P, INDIA
2
Dept. of Civil Engg. JNTUH College of Engineering, Hyderabad, A.P, INDIA
kvsnmurthy2007@gmail.com, bdeankumar@gmail.com
ABSTRACT
The turbo generator forms the heart of power plant. It is the most vital and expensive equipment of a power
plant complex and is generally housed in a turbo generator building The turbo generator foundation has to with
stand the vibrations caused by running of the machines as well as those due to shaking of ground motion while
ensuring continuous operation of the machine. So far many attempts have been made to analyze such structures
for machine operating conditions only. With the experience of damage of structures in the past earth quakes, it
is necessary to analyze such foundations for earth quake motions also. So, as a case study, in this project, a
turbo generator machine of 2x800MW capacity foundation is considered. This paper presents the study of
analysis of the turbo generator foundation for various seismic zones of India resting on pile foundations. The
analysis for both the static and dynamic conditions is carried out using software. Typical calculations are
presented as a part of study. Comparative study of the analysis on pile foundation forces under normal operating
conditions and failure operating conditions for various seismic zones is carried out. Cost variation analysis is
made for piling for all seismic zones.
KEYWORDS: TG foundation, Static, Dynamic, Seismic analysis.
1. INTRODUCTION:
A turbo generator consists of a turbine, generator, and the other auxiliaries like condenser, pipes carrying super
heated steam etc. Turbo generator falls under high speed rotary machine and its capacity varies from 2 MW to
2000MW.The top deck, column, and bottom raft together constitute turbo generator foundation. The increased
dependence of society on power provides no room for failure and demands equipment and systems with higher
performance and reliability. So, there is an emerging need to take care of the earthquake forces occurring during
the earth quake in the design of TG foundation in addition to the operating conditions of the machine.
[1] Mr. P. St. Fleischer and P.G Trombik have elaborated the practical design approaches, detailed seismic
analysis for different kinds of foundation types (mainly spring mounted, table mounted and raft foundation) and
proposed simplified design principles for large machine foundations.
[2] Mr. Sukanta Adhikari has dealt in his paper with the design of turbo generator foundation for a thermal
power plant and has covered the critical aspects in the design of TG FDNS with respect to IS 2974(part III-
1992) and other international standards.
[3] Mr. K.G Bhatia has covered basic fundamentals for understanding and evaluating dynamic response of
machine foundation system with emphasis on detailed dynamic analysis. He has explained the theory with
number of practical problems.
[4] Mr. P. Sreenivasulu & Vaidyanathan have dealt with various aspects of analysis, design and construction of
machine foundation.
[5] Prof.N.K Samadhiya, IIT Roorkee, presented wonderful video lectures on various types of machine
foundations and its basic fundamentals to understand the detailed dynamic analysis. This video presentation
really helps engineering students and practicing engineers to enrich their knowledge.
2. OBJECTIVE:
The main objective is comparative analysis of TG foundation for turbo generator foundation of 2x800MW
capacity resting on pile foundations for various seismic zones in India. The analysis is carried out for both the
static and dynamic conditions using software. Typical calculations are presented as a part of the study.
Comparative study is carried out for both under normal operating conditions and failure conditions for various
seismic zones. Cost variation analysis is made for piling for all seismic zones.
3. ANALYSIS OF T G FOUNDATION FOR VARIOUS SEISMIC ZONES IN INDIA:
3.1 Description of the Structure:
The STG foundation is a framed RCC structure consisting of base raft, columns and top deck. Top deck
supports the steam turbine and generator. The base raft supports condensers, high pressure drain flash tank, LP
heaters supported on the neck of condensers. Intermediate platforms independent of main TG columns are
provided at EL(+)2.5M, EL(+)5.30M,and EL(+)9.00M.Aproximate dimensions of top deck is 43.40 mx12.50 m
and the elevation of top deck is EL (+) 17.0M(with respect to FFL of EL(+)0.0m).Top level of base raft is kept
at EL(-)2.0m.
3.2 Analysis of TG Foundation:
Static analysis and dynamic analysis of TG foundation was carried out using STAAD PRO software. The
foundation shall be modelled as three dimensional space frame discredited using 3-D beam/plate elements with
six degrees of freedom. The columns are considered along with base raft and foundation piles represented by
springs. No separate model will be used for analysis and design of base raft. Nodes shall be specified at all
bearing points, beam-column junctions, load points and wherever member cross section changes significantly.
For seismic loading response spectrum analysis methods was used as per IS 1893-2002.The response spectrum
values of 5% damping as given in the code are used. Dynamic analysis will be carried out as per IS 2974 with
fixed base. Masses of turbines and generator rotor shall be lumped at their respective bearing points (CG
locations) connected to the deck with rigid members. Masses of beams and columns shall be lumped at all node
points along their lengths. Masses located outside the beam cross section (projections) will be applied at the
respective nodes along with mass moments where significant. Gross properties will be used for calculating
moment of inertia of the members for all load cases. Dynamic elastic modulus as per Cl.9.1.1 (e) of IS
2974(part-3)-1992 was used. Free vibration analysis shall be carried out to calculate the natural frequencies and
mode shapes of the foundation. The highest natural frequency calculated should be at least 10 percent higher
than the operating frequency of the machine. Forced vibration analysis shall be performed at operating speed
and also at frequencies corresponding to certain selected modes (Natural frequencies with 0.95 to 1.05 times the
operating speed) using normal machine unbalance loads. Natural frequencies calculated for different mode
shapes shall be separated from the machine operating frequency by a margin of 20% at least.
Amplitudes calculated shall be ensured to be within permissible limits specified by manufacturer. The
unbalanced centrifugal force at each bearing is calculated. The vertical(Y) and transverse(Z) direction
amplitudes on each of the eight bearings in this case are obtained by applying harmonic excitation forces in Y
and Z directions on each of the bearings separately. The displacements at all the selected nodes (bearing nodes)
are obtained by carrying out a harmonic analysis for the entire frequency range of the TG foundation. The
maximum value at a particular bearing node is tabulated for unbalance for each unit of TG i.e for HP Turbine,
LP1&LP2 Turbine and generator, for both vertical excitation force and horizontal excitation force applied at
487
each of the bearing points .The detailed tabulations for the displacement at each bearing node for each of the 16
excitations analyses was carried out for the entire frequency range. The SRSS values are also furnished for the
entire frequency range. The displacements are post processed in line with the criteria stated by manufacturer and
RMS amplitude values was calculated. The RMS value of vertical and horizontal displacements for+/-5% of the
rated speed (viz.from 47.5Hz to 52.5Hz) was calculated. Detailed static analysis was carried out for various
primary loads including seismic load as one of the primary load case. The foundations was modelled as three
dimensional space frames discredited using 3-D beam/plate elements with six degrees of freedom using STAAD
PRO program. The basic load cases are
Dead load (DL1-Self wt of all structural components of TG Fdn), Dead load (DL2-Loads from machine
components and other permanently applied loads without rotation), Dead load (DL3-Dead load includes weight
of rotating parts. To be used if not included under DL2),Power torque load (PT), Vacuum load (VL), Steam
piping load-VER (SP +V), Steam piping load- VER (SP +V), Frictional load (FL), Temperature
load(TL),Anchor block load(AL),Dynamic load(DYN-V), Dynamic load(DYN-H), Dynamic load(DYN-
A),Short circuit force(SCF),Seismic load(EQX), Seismic load(EQY), Seismic load(EQZ), Erection
load(EL),Live load(LL),Piping load(PL),Condenser operating conditions(COC), Hydro test load(HYD1-Load
from condenser shell during hydro testing provided by manufacture), Hydro test load(HYD2-Load from
condenser tube side during hydro testing provided by manufacture),Condenser abnormal condition(CAC) ,
Normal unbalanced vertical load(NUL-V), Normal unbalanced horizontal load(NUL-H).
4. RESULTS AND DISCUSIONS
After carrying out detailed analysis of TG foundation resting on piles keeping the same soft soil profile for
various earth quake zones in India, a comparative analysis results of the maximum vertical and horizontal
loads along the all grids of TG foundation due to Normal operating and failure operating conditions are
mentioned below.
4.1. Comparison of analysis results of TG foundation for various Earthquakezones in India.
4.1.1 Maximum vertical loads due to normal operating conditions.
The maximum vertical load under normal operating conditions on piles for zone 2,3,4 and 5 are 2189.55 kN ,
2214.11 kN, 2291.83 kN and 2322.09 kN respectively, Which are well within the safe carrying capacities of pile
under compression and the same values are plotted in the graph as shown below.
Under Normal Operating Conditions Maximum Pile vertical Loads in Compression are increased about 1.23%
from Zone 2 to Zone 3, 3.83% from Zone 3 to Zone 4 and 1.44% from Zone 4 to Zone 5.
488
The loads are increased from lower to higher zones as there will be higher seismicity due to increase of seismic
zone and expectation of high frequency ground motions.
4.1.2. Maximum vertical loads due to failure operating conditions
The maximum vertical loads under failure conditions on piles for zone 2,3,4 and 5 are 2420.8
kN,2594.11,2938.5kN and 3254.30 kN respectively, Which are well within safe carrying capacity of pile under
compression and the same values are shown below the graph.
Under Abnormal Operating Conditions Maximum Pile Loads in Compression are increased about 8.33% from
Zone 2 to Zone3, 14.18% from Zone 3 to Zone 4 and 12.28% from Zone 4 to Zone 5.
The increase in percentage of vertical loads are more in case of failure operating conditions because of the
additional loads applied on the structure such as short circuit forces, blade loss forces in addition to seismic
forces.
4.1.3. Maximum horizontal loads due to normal operating conditions
The maximum horizontal load under normal operating conditions on piles for zone 2, 3, 4 and 5 are 87.58kN ,
88.81kN , 88.29kN,89.11kNand respectively, Which are well within the carrying capacities of pile and the
values of same are as shown below in the graph.
Since the shockwaves in the horizontal direction are more intense than the vertical direction, the increase in the
horizontal component of the force will be more than the vertical loads.
489
4.1.4. MAXIMUM HORIZONTAL LOADS DUE TO FAILURE OPERATING CONDITIONS.
The maximum horizontal loads under failure conditions for seismic zone 2, 3, 4 and 5 are 140.95kN, 177.83kN,
245.54kN and 336.1kN respectively, Which are well within the safe lateral carrying capacity of pile and the
values of same are as shown below in the graph.
Due to application of failure operating condition loads in addition to normal operating loads, the
horizontal loads are increasing from lower seismic zones to higher zones than the horizontal loads due to normal
operating conditions.
4.1.5. Percentage variation of maximum vertical loads due to normal operating conditions w.r.t zone ii.
The % variation of maximum vertical load on pile under normal conditions from seismic zone II to III, II to IV
and II to V are increased by 1.23% and 5.1% and 6.6% respectively and the same values are shown below the
graph.
4.1.6 Percentage variation of maximum vertical loads due to failure operating conditions w.r.t ii.
The % of variation of maximum vertical load on pile under failure condition from seismic zone II to III and
zone II to IV and zone II to V are increased by 8.33%, 22.87% and 37.18% respectively.
490
4.1.7. Percentage variation of maximum horizontal loads due to failure operating conditions w.r.t zone ii.
The % of variation of maximum horizontal load on pile under Failure condition from zone II to III, zone II to IV
and zone II to V are increased by 27.37%, 75.40% and 140.53% respectively.
5. CONCLUSIONS
In the present study, we have analyzed and studied the impact on Turbo generator foundation piling forces for
various seismic zones in India. The conclusions obtained from this study are as follows:
1. The variation of forces along the width of the raft is like an inverted parabola in case of the vertical forces
due to both the normal and failure operating conditions of the machine.
2. From this analysis results, it is observed that the maximum pile vertical loads under compression and also
lateral loads are increasing more from lesser seismic zones to higher seismic zones in both the operating
conditions of machine.
3. It is also found from the analysis results that the vertical forces and lateral forces are increasing more in the
case of abnormal/failure operating conditions of machine than normal operating conditions from lesser to
higher seismic zones.
491
4. It is also observed that the lateral forces are increasing more when compared to vertical forces in both the
conditions of operating of machine foundations.
5. The vertical and horizontal forces in piles are increasing because of soft soils which shall have lesser shear
wave velocity as compared to rocks. Due to this property seismic amplification is more with increase of
seismic zone.
6. The cost of piling increases by 45% from zone 2 to zone3, 49.43% from zone 3 to zone 4, and 17.7% from
zone 4 to zone 5.
7. The cost of piling increases by 45% from zone 2 to zone3, 116.7% from zone 3 to zone 4 155% from zone 4
to zone 5.
REFERENCES.
[1] Pst.Fleischer& P.G.Trombik (Oct12-17, 2008). A paper on “Turbo Generator foundations subjected to Earth
quake loadings”, by published in the 14th world conference on Earthquake engineering, Beijing, China.
[2] Mr. Sukunta Adhikari. A paper on” TURBOGENERATOR FOUNDATIONS”

(www.sefindia/forum/download.php).Id-3601.
[3] Mr.K.G.Bhatia. Foundations for industrial machines: Hand book for practicing engineers.
[4]Mr. P.Sreenivasulu, C.V.Vaidyanathan .Hand book of machine foundations .
[5] Prof.N.K.Samadhiya. Video lectures by, I.I.T, Roorkee.
492
Paper No.: 087
TRANSIENT ANALYSIS OF AN ELLIPTICAL JOURNAL BEARING

1 2
Rajiv Verma and Puneet Mathur
1
Assistant Professor, Dept. of Mechanical Engineering, National Institute of Technology, Kurukshetra, India
2
Assistant Professor, Dept. of Mechanical Engineering, MEFGI, Rajkot, India
Email: rajivkkr@yahoo.com, puneet_mathur87@rediff.com
ABSTRACT:
The paper presents the static and dynamic performance characteristics in conjunction with linear and nonlinear
trajectories for an elliptical bearing operating with Newtonian and micropolar fluid. A modified form of
Reynolds equation is used for micropolar fluid and is solved using the Finite difference technique. The static
and dynamic performance characteristics are evaluated for the bearing at different operating conditions. Linear
and nonlinear equations of motion are solved by the classical Runge-Kutta fourth order method to obtain the
trajectories of the journal for a given perturbation. It is concluded that the journal become unstable even when
it’s mass is equal that predicted by the linear theory for both Newtonian and micropolar fluid.
KEYWORDS: Elliptical bearing, Micropolar fluid, Static and Dynamic Characteristics, Linear and Nonlinear
Trajectories.
1. INTRODUCTION
The characteristics of lubricants are often modified using additives in order to meet the specific requirements of
many engineering applications. These oils behave as pseudo-plastic or dilatants fluids with viscosities following
a nonlinear relationship between shear stresses and shear strain rates. Eringen [l] had observed that the motion
of these colloidal fluids is not governed by basic equations of Newtonian fluid and thus resulted in the
modification of momentum and energy equations. Allen and Kline [2] discussed the theory of lubrication with a
micropolar fluid and presented solution for an inclined slider bearing. The analysis showed an increased load
capacity and frictional force for micropolar fluid. Various hydrodynamic characteristics were theoretically
analysed by Zaheruddin and Isa [3] for one dimensional journal bearings, both infinitely long and infinitely
short with micropolar lubrication. Enhanced performance of bearing under micropolar lubrication was observed
in the one-dimensional squeeze film journal bearings operating under a cyclic load [4]. Similar improvements in
terms of load capacity, frictional moment and frictional coefficient were observed in micropolar lubrication
when compared to the Newtonian lubrication with same viscosity for both infinitely long [5] and for a short
bearing [6]. Linear stability theory was used by Huang and Weng [7] to obtain the dynamic characteristics of
finite width journal bearing. Wang and Zhu [8] studied the effectiveness of micropolar fluids in a dynamically
loaded journal bearing. The analysis predicted the importance of the material characteristic length and the
coupling number on the performance of a dynamically loaded journal. In addition to this Guha et al. [9] applied
the first order perturbation of the film thickness and obtained steady state film pressure, the performance
characteristics, critical mass parameter and whirl ratio with respect to the micropolar property for varying
slenderness ratios. Jain et al. [10] determined the journal trajectories using nonlinear equations of motion
including the deformation of the bearing shell. Dynamic performance of the journal bearing system was
evaluated on the basis of the journal motion trajectories to investigate the effect of bearing deformation.
Sinhasan and Goyal [11] observed that the nonlinear trajectories for two-lobe journal bearing system becomes
unstable even at values of non-dimensional journal mass less than the critical journal mass computed using the
linearized equations of motion and Routh-Hurwitz stability criteria.
The linear and nonlinear stability analysis has been carried out in the present paper for an elliptical bearing
using micropolar fluid. Linear stability analysis has been used to evaluate the critical mass parameter and is
validated using the nonlinear transient analysis.
2. ANALYSIS
General form of governing equation for a micropolar fluid is obtained as:
∇. V = 0
Conservation of mass for an incompressible fluid is shown below:
(1)
Conservation of linear momentum for an incompressible fluid
λ + 2µ ∇ ∇. V − ∇∗ ∇∗V + ∇ ∗ υ − ∇. π + F =
µ
(2)
Conservation of angular momentum for an incompressible fluid
α + β + γ ∇ ∇. υ − γ∇ ∗ ∇ ∗ υ + ∇ ∗ V − 2 υ + C = j

υ
(3)
thermodynamic pressure and is to be replaced by the hydrodynamic film pressure, , since, π = − ρ = . E

where,ρ is the mass density, V is the velocity vector, υ is the micro-rotational velocity vector. π is the
is the internal energy and P is to be determined by the boundary conditions. λ and µ are the familiar viscosity
coefficients of the classical fluid mechanics, while α, β and γ are the new viscosity coefficients derived as the
combinational effects of the gyroviscosities for the micropolar fluid as defined by Eringen [1]. is also a new
viscosity coefficient for micropolar fluid, termed as spin viscosity, which establishes the link between velocity
vector and the microrotational velocity vector. FB is the body force per unit mass, CB is the body couple per unit
mass and j is the microinertia constant. D/Dt represents the material derivative. Now consider the flow of thin
lubricant layer in journal bearing and assume the components of velocity vector and microrotation vector have
V = "V# , V$ , V% & and υ = υ# , 0, υ%

the following form:
Using the usual postulations for lubricant theory [13] and ignoring infinitesimal quantities with high order we
obtain the governing differential equations for lubricant flow
+ − (0 = 0
' ( ) *+ (-. (/
(, ) (,
(4a)
− − (2 = 0
' ( ) *1 (- (/
(, ) (2
(4b)
3 + − 2 45 = 0
()- (*1
(, ) (,
(4c)
3 + − 2 46 = 0
( ) -. (*+
(, ) (,
(4d)
78 9 = 0 ∶ ;0 = <5 , ;, = ;2 = 0, 45 = 46 = 0
For the lubricant layer the ordinary boundary conditions are as follows:
A
78 9 = ℎ: ;0 = < , ;, = + < ,; = 0, 45 = 46 = 0
(? (?
(@ (0 2
where h represents the thickness of oil film, <5 , the tangent velocity of surface l, < and ; , the tangent and
normal velocity of surface 2 respectively. Velocity components can be obtained by solving eqns. (4a-4d).
Substituting all these velocity components into the equation (1) and integrating across the film thickness. The
modified Reynolds equation for a micropolar lubricant is obtained as:
494
Φ B, CD , ℎ + Φ B, CD , ℎ = 6<F + 12F
( (/ ( (/ (? (?
(0 (0 (2 (2 (0 (@
(5)
where, Φ B, CD , ℎ = ℎ6 + 12CD ℎ − 6BCD ℎ HI8ℎ J N and B=O , CD = O

K? P
LM ' Q'
Upper lobe
Bearing Journal
Line of center
Obl θ
(XL,YL)
Ep Ωs
c
θ=180 θ=0
X
Ob
-Ep φ Oj
Obu
θ Rj (XL,-YL)
Lower lobe
Figure1. Schematic diagram of an elliptical bearing
N and lm are two important parameters distinguishing a micropolar lubricant from a Newtonian lubricant. N is a
dimensionless parameter called the coupling number, which couples the linear and angular momentum
equations arising from the micro-rotational effects of the suspended particles in the lubricant. lm represents the
interaction between the micropolar lubricant and the film gap or clearance space of a journal bearing and is
termed as the characteristic length of the micropolar lubricant. Non-Dimensional form of modified Reynolds
equation is given as:
(R JΦ B, CD , ℎ N + (2 JΦ B, CD , ℎ N = 6 (R + 12 (@
( (/ ( (/ (? (?
(R (2
(6)
where, S = T ; ℎ = V ; W = T ; = 'XT ;
0 ? 2 /V )
CD = L ; < = YZ; 8 = Y8
V
M
)
6
Φ B, CD , ℎ = ℎ + ) − HI8ℎ J N
5 ? [K? ?KLM
LM LM
3. BOUNDARY CONDITIONS
The following boundary conditions, pertinent to the flow field, are defined in terms of pressure and pressure
gradient.
495
For Elliptical bearing
= 0 at θ = θ5 ] , θ ]

= 0 at θ = θ ]
δ/
δθ
where, θ2 is the location of the trailing edge for mth lobe of the bearing, m=1, 2
4. FILM THICKNESS AND ITS GRADIENT
For an elliptical bearing, the film thickness and its gradient is given by
ℎ = 1 − "^_ − ^` &HIa S + b − "c_ − c` &ade S + b ;
hg = − JXgi − Xgj N cos S + b − JYgi − Ygj N sin S + b ;
where X p and Yp indicates the centre position of the journal. Xgp and Ygp indicates the rate of change of position.
5. STATIC CHARACTERISTICS
5.1 Pressure calculation and Load carrying capacity
A computer programme is developed to compute the pressure by using finite difference method (FDM) and
F# = qr qr ) cosθ dθ dW
afterwards forces are calculated by using
A
θ
F$ = qr qr ) sinθ dθ dW
θ
(7)
Resultant load is given as
F = OF# + F$
where F# and F$ are radial and tangential components of dimensionless hydrodynamic force.
5.2 Attitude Angle
The attitude angle is defined as the angle between the line of action of external force and line of centres. The
attitude angle is established by satisfying the condition of vertical force by using an iterative process for a given
eccentricity ratio.
6. DYNAMIC CHARACTERISTICS
6.1 Stiffness characteristics
− −
The non-dimensional coefficients of stiffness are given as:
K ## K #$
δwx δwx
s u = v z
K $# K $$ − −
δ# δ$
δwy δwy
(8)
δ# δ$
The first subscript denotes the direction of force and the second, the direction of displacement.
6.2 Damping Characteristics
496
− −
The non-dimensional coefficients of damping are given as:
C## C#$
δwx δwx
s u = v z
C$# C$$ − −
δ#g δ$g
δwy δwy
(9)
δ#g δ$g
The first subscript denotes the direction of force and the second, the direction of velocity.
7. STABILITY ANALYSIS
If a journal centre is disturbed from its equilibrium position there is a change in force. The journal centre starts
rotating around its static equilibrium position due to the unbalance forces. The motion trajectories can be
obtained by the linear and nonlinear equations of motion of the journal.
The linearized equations of motion of a journal centre are:

Mp 0 x~ C## C#$ xg K ## K #$ x 0
s u| € + s u syg u + s u |y€ =
0 Mp y~ C$# C$$ K $# K $$ 0
(10)
The nonlinear equations of motion of a journal centre are:

Mp 0 x~ ∆F#
s u| € = | €
0 Mp y~ ∆F$
(11)
The assumption of complementary solution to eqn. (10) of the form x = Xeλ and y = Yeλ leads to following
polynomial
Ar λQ + A5 λ6 + A λ + A6 λ + AQ = 0
where Ar = Mp
A5 = Mp "C## + C$$ &
A = "Mp ∗ K ## + Mp ∗ K $$ + C## ∗ C$$ − C$# ∗ C#$ &
A6 = "K ## ∗ C$$ + C## ∗ K $$ − K $# ∗ C#$ − K #$ ∗ C$# &
AQ = K ## ∗ K $$ − K $# ∗ K #$
making the use of the polynomial equation into Routh-Hurwitz stability criteria, the stability periphery of the
journal bearing system is obtained.
7.1 Critical mass
The critical mass is evaluated by using the stiffness and damping coefficients incorporated with Routh-
Hurwitz stability criteria in following equation:
M„ = ‡
…†
† ˆ„†
(12)
where, ar = "C## ∗ C$$ − C$# ∗ C#$ &
497
br =
"Šxx Šyy & ∗ ‹xx ∗‹yy ˆ‹yx ∗‹xy
"‹xx ∗Šyy Šxx ∗‹yy ˆ‹yx ∗Šxy ˆ‹xy ∗Šyx &
cr =
‹xx ∗Šxx Šyy ∗‹yy ‹yx ∗Šxy ‹xy ∗Šyx
"Šxx Šyy &
For the stability of a journal bearing system, mass of journal (Mp ) should be less than critical mass (M„ ). For any
negative value of M„ the journal bearing system will always be stable.
7.2 Whirl frequency ratio
It is the ratio of the rotor whirl or precessional frequency to the rotor onset speed at instability.
ω = Œ
"‹xx ˆ‹•Ž &∗ ‹yy ˆ‹•Ž ˆ‹yx ∗‹xy
"Šxx ∗Šyy ˆŠyx ∗Šxy &
(13)
where, K •• =
‹xx ∗Šyy Šxx ∗‹yy ˆ‹yx ∗Šxy ˆ‹xy ∗Šyx
"Šxx Šyy &
If K •• ≤ 0, whirl frequency ratio implies that there is an absence of whirl and the system is always stable
7.3 Threshold speed
Threshold speed is the utmost value of speed at which the bearing becomes stable. It is obtained using the
relation given below
Ω’ = O
w∗‹•Ž
“”
(14)
The stability threshold curves divide into two major zones. The zone above this curve is unstable, whereas the
zone below is stable.
8. SOLUTION PROCEDURE
8.1 Linear analysis
A computer programme is developed to solve the Reynolds equation for an assumed eccentricity ratio,
characteristic length (CD ) and coupling number (N) for an elliptical bearing. The modified Reynolds equation (6)
is discretized by using finite difference method and the solution obtained for the boundary condition mentioned
in section 3. Static forces along radial and tangential directions are calculated as shown in eqn. (7). Afterwards
as discussed in eqn. (8) and eqn. (9) we get four stiffness coefficients and four damping coefficients.
Incorporating these values of stiffness and damping coefficients in eqns. (12 - 14). Critical mass, threshold
speed and whirl frequency ratio are determined. Linear motion trajectories of the journal centre are obtained by
solving eqn. (12), by means of classical fourth-order Runge-Kutta method. During the process, initial values of
displacements and velocities of the journal centre are arbitrarily chosen and time step =0.02 is employed to
obtain the journal centre motion trajectories.
8.2 Nonlinear analysis
Nonlinear motion trajectories of the journal centre are obtained by solving eqn. (13), by means of classical
fourth order Runge- Kutta method. After each time step the fluid film thickness is modified using the computed
values of horizontal and vertical components of journal centre disturbed from its steady state equilibrium
498
position. The current velocities are used for calculating the components of the fluid film forces. The procedure
is repeated for each time step.
The linear stability of a journal depends on the stiffness and damping characteristics. Table 1 shows the
table for L/R=2, ϵ=0.5, ellipticity=0.5, CD =10, 40 and N2= 0.1, 0.3, 0.5. The result show that as the coupling
comparison of performance characteristics for a Newtonian and micropolar fluid. The results tabularized in the
number (N) tends to zero or the characteristic length (CD ) of micropolar fluid goes infinitely; the fluid is
converted into Newtonian fluid. It is also observed that critical mass (M„ ) and threshold speed (Ω’ ) increases
with increase in coupling number. It shows that for the same eccentricity the micropolar fluid exhibits better
stability than a Newtonian fluid. The results evaluated in table 1 are compared for Newtonian fluid [11] and the
results agree very well.
Transient responses of an elliptical journal bearing are presented in table 2. For a Newtonian fluid at Mj=Mc,
motion of a journal for linear analysis exhibit limit cycle. However for the same mass, nonlinear analysis shows
instability of journal. When Mj<Mc, journal motion achieve limit cycle for nonlinear analysis while journal is
stable for linear analysis. It has observed that if the mass of journal is taken below to this limit then the journal
shows stability for both the analysis. There are several cases presented in table 2 for micropolar fluid. On the
basis of results obtained in table 2 and to support the above discussion, trajectories are plotted, publicized in fig.
(2) - fig. (4).
Table 1 Comparison of performance characteristics of elliptical journal bearing for ϵ=0.3, ellipticity=0.5
Micropolar Fluid Micropolar Fluid

Coupling Number Newtonian Fluid (•– =10) (•– =40)
(N2) 0.1 0.3 0.5 0.1 0.3 0.5
Attitude angle(ϕ) 82.031 81.875 80.938 79.531 81.719 81.172 80.781
Load (—) 4.2436 4.6927 5.7807 7.2774 4.5272 4.8826 5.1768
Kxx 4.9695 5.5255 7.1514 9.3531 5.3924 6.0812 6.5122
Kxy 11.876 13.087 15.741 18.96 12.588 13.358 13.884
Kyx -16.306 -18.019 -22.039 -27.642 -17.428 -18.785 -19.844
Kyy 24.420 27.041 33.609 43.518 26.238 28.66 30.943
Cxy≈ Cyx
Cxx 15.879 17.505 21.019 25.481 16.852 17.845 18.601
5.8553 6.3879 6.9375 7.584 6.041 5.8779 6.0747
Cyy 47.810 6.0094 6.4266 6.9064 5.6528 5.4498 5.7071
Critical mass (™š ) 52.92 65.364 82.551 51.213 55.335 58.318
Threshold speed (›œ )
63.103
2.3515 70.234 89.702 123.56 68.687 76.442 85.257
ratio (• )
Whirl frequency 2.3601 2.4144 2.5511 2.3749 2.4197 2.499
0.4037
10. CONCLUSION
Important conclusion can be made about the use of micropolar fluid in elliptical journal bearing
499
a. It has been observe that load carrying capacity of a journal bearing with micropolar lubricant at a particular
eccentricity ratio increases when compared with that of bearing with Newtonian fluid.
b. The micropolar fluid approaches to Newtonian fluid as characteristic length of the micropolar fluid grows
indefinitely or coupling number tends to zero.
c. The critical mass and threshold speed for a elliptical bearing under micropolar fluid is increases for high
coupling number. Hence, stability of a bearing increases for high coupling number.
d. The critical mass and threshold speed for a elliptical bearing under micropolar fluid decreases when
characteristic length decreases.
e. The results show that nonlinear analysis gives better results than linear analysis. Linear analysis is necessary
to complete for predicting the value of Mj in nonlinear analysis. It seems that the nonlinear analysis predicts a
lower value of critical mass than linear analysis.
Table 2 Critical mass and Transient response for an elliptical bearing at ϵ=0.3, ellipticity=0.5 and L/R=2
•– ™š žŸ
Transient response
N2
Linear analysis Nonlinear analysis
Mj = Mc Limit cycle Unstable
Newtonian 63.018
Mj = 0.915Mc Stable Limit cycle
Mp = 0.914M„
10 0.1 70.234
Stable Limit cycle
50 0.1 68.036
10 0.5 123.560
50 0.5 81.058
REFERENCES
[1] Eringen A. C., (1966). Theory of Micropolar Fluids, J. Math. Mech.,16 (1) 1.
[2] Allen S. J. and Kline K. A., (1971). Lubrication theory for micropolar fluids, Trans.,ASME, 646.
[3] Zaheeruddin Kh. and Isa M., (1978). Micropolar fluid lubrication of one-dimensional journal bearings,Wear,
502(1) 11.
[4] Isa M. and Zaheeruddin Kh., (1980–1981). Characteristics of squeeze film porous journal bearings with
micropolar lubricant, Trans ASME, 6(3) 146.
[5] Tipei N., (1979). Lubrication with micropolar liquids and its application to short bearings, J Lubr Technol,
Trans ASME ,(101) 356.
[6] Khonsari M.M. and Brewe D.E., ( 1989). On the performance of finite journal bearings lubricated with
micropolar fluids, STLE, Tribology Trans, 32(2) 155.
500
[7] Huang T.W. and Weng C.I.,(1990). Dynamic characteristics of finite width journal bearings with micropolar
fluids, Wear, 140(1) 24.
[8] Wang X. L. and Zhu K.Q., (2004). A study of the lubricating effectiveness of micropolar fluids in a
dynamically loaded journal bearing (T1516), Tribology International 37- 481.
[9] Das S., Guha S.K. and Chattopadhyay A.K., (2005). Linear stability analysis of hydrodynamic journal
bearings under micropolar lubrication, Tribology International, 38- 500.
[10] Jain S. C., Sinhasan R. and Pilli S.C., (1989). A Study on the Dynamic Response of Compliant Shell
Journal Bearings, Tribology Transactions, 32 (3) 297.
[11] Sinhasan R. and Goyal K. C., (1995). Transient response of a two-lobe journal bearing lubricated with non-
Newtonian lubricant, Tribology International, 28(4) 233.
[12] Sinhasan R. and Goyal K. C., (1992). Transient response of a circular journal bearing lubricated with non-
Newtonian lubricant, wear, 156- 385.
[13] Gan Q. Z. and Ji L.Z., (1987). Lubrication theory for micropolar fluids and its application to a journal
bearing with finite length, applied mathematics and mechanics, 8 (7) 655.
and L/R=2 at a) Mi =M„ =63.018 and b) Mi =0.915M„ =57.662

Figure 2 Linear and nonlinear trajectories for an elliptical bearing with Newtonian fluid for ϵ=0.3, ellipticity=0.5
501
ellipticity=0.5, L/R=2, l] =10 and N2=0.1 at a) Mi =M„ =70.234 and b) Mi =0.914M„ =64.193
Figure 3 Linear and nonlinear trajectories for an elliptical bearing with micropolar fluid for ϵ=0.3,
ellipticity=0.5, L/R=2, l] =10 and N2=0.5 at a) Mi =M„ =123.56 and b) Mi =0.927M„ =114.589
Figure 4 Linear and nonlinear trajectories for an elliptical bearing with micropolar fluid for ϵ=0.3,
502
Paper No.: 088
RESPONSE SURFACE BASED ANALYSIS OF STRUCTURES WITH FUZZY

VARIABLES
1 2
A.S. Balu and B.N. Rao
1
Assistant Professor, Department of Civil Engineering, National Institute of Technology Karnataka, Surathkal
2
Professor, Department of Civil Engineering, Indian Institute of Technology Madras, Chennai
Email: asbalu@nitk.ac.in, bnrao@iitm.ac.in
ABSTRACT :
This paper presents a practical approach based on High Dimensional Model Representation (HDMR) for
analyzing the response of structures with fuzzy parameters. The proposed methodology involves integrated
finite element modelling, HDMR based response surface generation, and both, implicit and explicit fuzzy
analysis procedures. The uncertainties in the material, geometric, loading and structural parameters are
represented using fuzzy sets. To facilitate efficient computation, a HDMR based response surface generation
is employed for the approximation of the fuzzy finite element response quantity..
KEYWORDS: Fuzzy analysis, High dimensional model representation, Response surface, Explicit
formulation, Implicit formulation
1. INTRODUCTION
The exponential growth of computational power during the last few decades has enabled the finite element
analysis of many real-life engineering systems which are too complex to be analytically solved in a closed form.
In the traditional deterministic finite element analysis, system parameters such as mass, geometry and material
properties are assumed to be known precisely and defined exactly. However, in practice most of the data used
in the solution process of many practical engineering systems are either collected from experiments or acquired
as empirical data from the past, which are usually ill defined, imprecise and uncertain in nature. The
uncertainty present in the problem can be categorized into randomness and fuzziness.
Probabilistic methods are especially suitable in case of uncertainty due to randomness for which the
information on both the range and the probability density function are available. The probabilistic concept is
already well established for the extension of the deterministic finite element method towards uncertainty
assessment. Fuzzy set theory [1] provides a concept for the description of linguistic or subjective knowledge
and incomplete data in a non-probabilistic manner. The use of fuzzy set theory concept in the finite element
context has led to the development of the fuzzy finite element method for the solution of engineering problems
[2-4]. Its aim is to obtain the membership function of an output quantity, based on the fuzzy description of the
uncertain input parameters. Over the last decade, several techniques have been reported, such as the vertex
method [5−7], the global optimization approaches [8, 9] and the interval arithmetic approaches [10, 11].
However, each of these techniques has its specific limitations when applied to realistic engineering problems,
and none of them has been established as standard interval procedure, and the choice of technique depends on
the type of the conducted FE analysis.
This paper presents a practical approach based on High Dimensional Model Representation (HDMR) [12, 13]
for analyzing the response of structures with fuzzy parameters. The proposed methodology involves integrated
finite element modelling, HDMR based response surface generation, and both, implicit and explicit fuzzy
analysis procedures. The uncertainties in the material, geometric, loading and structural parameters are
represented using fuzzy sets. To facilitate efficient computation, a HDMR based response surface generation
is employed for the approximation of the fuzzy finite element response quantity.
In the proposed implicit formulation, the fuzzy response is expressed as a closed form expression using
HDMR. The binary combinations of the fuzzy variables that result in the extreme responses at an α-level are
obtained by applying combinatorial optimization to the approximated response function using HDMR. The
membership function of the fuzzy response is then obtained by carrying out the full finite element analysis for
the binary combinations that resulted in the extreme responses. In the proposed explicit formulation, using
HDMR the response is expressed in a separable closed-form expression as a linear combination of the fuzzy
variables through the definition of intervening variables, then using the transformation technique [14] the
bounds on the response at each α-level are obtained.
2. APPLICATION OF HDMR IN FUZZY STRUCTURAL ANALYSIS
Let the N−dimensional vector x = { x1 , x2 ,L , xN } , represent the input variables of the model under consideration,
and the response function as g ( x ) . Since the influence of the input variables on the response function can be
independent and/or cooperative, HDMR [12, 13] expresses the response g ( x ) as a hierarchical correlated
function expansion in terms of the input variables as
( ) ( )
N
g ( x ) = g 0 + ∑ gi ( xi ) + ∑ gi1i2 xi1 , xi2 + ... + ∑ gi1i2 Kil xi1 , xi2 ,K , xil + K + g12... N ( x1 , x2 ,K , xN ) (1)
i =1 1≤ i1 < i2 ≤ N 1≤ i1 <K< il ≤ N
where g 0 is a constant term representing the zeroth-order component function or the mean response of g ( x ) .
The function gi ( xi ) is a first-order term expressing the effect of variable xi acting alone, although generally
nonlinearly, upon the output g ( x ) . The function g i i ( xi , xi
12 1 2
) is a second-order term which describes the
cooperative effects of the variables xi and xi upon the output g ( x ) .
1 2
The higher order terms give the
cooperative effects of increasing numbers of input variables acting together to influence the output g ( x ) . The
last term g12,L, N ( x1 , x2 ,L , xN ) contains any residual dependence of all the input variables locked together in a
cooperative way to influence the output g ( x ) . Once all the relevant component functions in Equation 2 are
determined and suitably represented, then the component functions constitute HDMR, thereby replacing the
original computationally expensive method of calculating g ( x ) by the computationally efficient model.
Usually the higher order terms in Equation 1 are negligible such that HDMR with only few low order
correlations, amongst the input variables are typically adequate in describing the output behavior resulting in
rapid convergence of HDMR expansion.
In this work, cut-HDMR procedure is used to approximate the original response function with an explicit
performance function to estimate the response of structural/mechanical systems subjected to fuzzy variables after
defining a reference point c = {c1 , c2 ,L , cN } in the variable space. The first order approximation of the function,
g% ( x ) = ∑∑ φ j ( xi ) g ( c1 ,K , ci −1 , xij , ci +1 ,K , cN ) − ( N − 1) g0
N n
(2)
i =1 j =1
Along each of the fuzzy variable xi having triangular membership function with the triplet number
[ xiL , xiM , xiU ] , n ( = 3,5, 7 or 9 ) sample points xiL , xiM − ( n − 3)( xiM − xiL ) ( n − 1) ,
xiM − ( n − 5 )( xiM − xiL ) ( n − 1) , …, xiM , …, xiM + ( n − 5 )( xiU − xiM ) ( n − 1) , xiM + ( n − 3)( xiU − xiM ) ( n − 1) , xiU are
deployed to obtain the first-order component functions.
If n is the number of sample points taken along each of the variable axis and s is the order of the component
function considered, starting from zeroth-order to l -th order, then total number of function evaluation for
∑ ( N !( n − 1) ) ( ( N − s )! s !)
l
s
interpolation purpose is given by, which grows polynomially with n and s . As
s =0
a few low order component functions of HDMR are used, the sample savings due to HDMR are significant
compared to traditional sampling. Hence uncertainty analysis using HDMR relies on an accurate reduced
model being generated with a small number of full model simulations. The tremendous computational savings
result from just having to perform interpolation instead of full model simulations for output determination.
504
2.1. Implicit Formulation
In explicit formulation, the original response function g ( x ) is approximated using first order HDMR. Once
the binary combinations of extreme left and right values of fuzzy variables at each α -level are formed, then the
responses are obtained using g% ( x ) . The true values of the bounds on the response at each α -level are
obtained by feeding binary combinations that minimize and maximize g% ( x ) , into the original numerical model.
This strategy requires 2 finite element runs at an α -level as opposed to 2 N finite element runs with vertex
method. This is extremely efficient especially for large number of fuzzy input parameters and large finite
element models.
2.2. Explicit Formulation

In explicit formulation, once the original response function is expressed as a linear combination of the fuzzy
variables, then at each α -cut the bounds of the response can be obtained by using interval arithmetic algorithm.
A nonlinear response function can be expressed as a linear combination by the use of first-order HDMR
approximation in conjunction with intervening variables in order to apply an interval arithmetic algorithm.
The first-order HDMR approximation g% ( x ) of the original response function g ( x ) can be transformed into a
linear combination of intervening variables, as follows
g% ( z ) = z1 + z2 + K + z N + β0 (3)
where zi = (βi xi + γ i ) is the relation between the intervening and the original variables with κ being order of
κ
approximation taking values κ = 1 for linear approximation, κ = 2 for quadratic approximation, κ = 3 for
cubic approximation, and so on. If the membership functions of the intervening variables are available, then at
each α -cut, interval arithmetic techniques [15] can be applied to estimate the response bounds at that level.
The use of transformation techniques facilitates the determination of the minimum and maximum values of the
fuzzy variables that correspond to the extreme values of the response at a particular level without the use of
optimization techniques. Moreover, this is an analytical procedure in which there is little room for errors.
3. NUMERICAL EXAMPLES
Three numerical examples are presented to demonstrate the accuracy and efficiency of the proposed
methodology. Example 1 is aimed to demonstrate the accuracy and the applicability of the transformation
technique in the proposed explicit formulation to obtain the bounds on the response. Examples 2 and 3
demonstrate the accuracy and the applicability of the proposed implicit and explicit formulations. If N , n ,
and m respectively denote the number of fuzzy variables, the number of sample points taken along each of
the variable axis, and the number of α -cuts, then using first-order HDMR approximation the total cost of
original function evaluation entails a maximum of N × ( n − 1) + 2 × ( m + 1) and N × ( n − 1) + 1 by the proposed
implicit and explicit formulations, respectively. It can be easily verified that the total computational cost of the
proposed formulations is far less than 2 N m + 1 number of function evaluations required for the crude vertex
method fuzzy analysis in conjunction with combinatorial optimization, especially for large number of fuzzy
input parameters and large finite element models. This clearly demonstrates the computational efficiency of
the proposed formulations. The efficiency and robustness of the proposed formulations is expected to increase
with increase in the complexity of the structure, number of fuzzy variables and number of α -cuts. If
first-order HDMR approximation is not sufficient second-order HDMR approximation may be adopted at the
expense of additional computational cost.
3.1. Closed form example

To illustrate the accuracy and the applicability of the transformation technique in the proposed explicit
formulation to obtain the bounds on the response, the following nonlinear response function is considered.
f ( x ) = 16 x14 − 96 x13 + 216 x12 − 216 x1 − 64 x23 + 240 x22 − 300 x2 + 36 x32 − 84 x3 − 8 x4 + 264 (4)
505
where x1 = [1, 2, 3] , x2 = [−1, 0, 1] , x3 = [−5, 0, 5] , and x4 = [10, 15, 20] are assumed to have triangular
membership functions and the triplet number for each fuzzy variable [ xL , xM , xU ] defines a triangular
membership function, where xL and xU are the lower and upper bounds at α = 0 and xM is the nominal
value at α = 1 . As Equation 4 represents closed form nonlinear function, in the proposed implicit and explicit
formulations, the HDMR approximation of the original response function is skipped. In the proposed implicit
formulation combinatorial optimization is carried out at each α -level. Ten α -levels ranging from 0 to 1 are
chosen. Figure 1 shows the membership function of f ( x ) obtained by the crude vertex method fuzzy analysis
in conjunction with combinatorial optimization. Figure 1 also presents the membership function of f ( x )
obtained by adopting interval arithmetic techniques directly on Equation 4.
As Equation 4 represents closed form nonlinear function, in the proposed explicit formulation, skipping the
HDMR approximation of the original response function, the membership function of f ( x ) is obtained by
suitable transformation of the variables x1 − x4 into a linear combination of intervening variables, as follows
f ( z ) = z1 + z2 + z3 + z4 (5)
where z1 = (2 x1 − 3)4 , z2 = −(4 x2 − 5)3 , z3 = (6 x3 − 7) 2 and z4 = −8 x4 + 9 .
Figure 1 Comparison of membership function of response
Figure 1 includes the membership function of f ( x ) obtained by the proposed explicit formulation. As
expected due to the problem of dependency, the membership function of f ( x ) obtained by adopting interval
arithmetic techniques resulted in wider bounds. The differences in the membership function of f ( x ) obtained
by the crude vertex method fuzzy analysis in conjunction with combinatorial optimization and the proposed
explicit formulation is mainly due to the reason that the membership function obtained by using the vertex
method fails to capture the minimum value of the response because it calculates the minimum and maximum
values of the response only at the lower and upper limits of the variables at each α -level. Therefore, it does not
identify the minimum of the response, which is located inside the design space.
3.2. Three-stepped bar

In this example a three-stepped bar shown in Figure 2 is analyzed to determine the effect of input parameter
fuzzification on the natural frequencies.
L1 L2 L3
A1E1 A2E2 A3E3 P
Figure 2 Stepped bar
506
The fuzzy parameters considered are A1 =[2.99, 3.00, 3.01], A2 =[1.99, 2.00, 2.01], A3 =[0.99, 1.00, 1.01],
L1 =[11.95, 12.00, 12.05], L2 =[9.95, 10.00, 10.05], L3 =[5.95, 6.00, 6.05], E =[2.8e7, 3.0e7, 3.1e7], and
ρ =[0.27, 0.28, 0.29]. Figure 3 shows possibility distributions of the natural frequencies of the system
computed using the proposed first-order HDMR based implicit and explicit formulations. In both implicit and
explicit formulations the effect of number of sample points used for first-order HDMR approximation of the
original response function is examined by varying n from 3 to 9. It is observed that n = 3 and 7
respectively provide the convergent estimation of the bounds of the response using the proposed implicit and
explicit formulations. In addition, in explicit formulation the effect of order of approximation κ is examined
by taking κ = 2 , 3 and 4. It is observed that the proposed implicit and the combinatorial approach yield exactly
the same results, however the explicit formulation estimates a narrow bound for the response.
Figure 3 Possibility distributions of natural frequencies: Modes 1−3
3.3. Edge-cracked plate

This example involves the edge-cracked plate shown in Figure 4, which is fixed at the bottom and subjected to
far-field shear stress of τ∞ unit applied on top. The plate has length L = 16 units, width W = 7 units and
crack length a . The elastic modulus is assumed to follow a hyperbolic-tangent gradation function which can be
expressed with respect to the local ( x1 , x2 ) Cartesian coordinates
E1 + E2 E1 − E2
E ( x1 ) = + tanh ( ηx1 ) (6)
2 2
where E1 = E ( x1 = −0.5W ) , E2 = E ( x1 = 0.5W ) , η is gradation parameter. E1 =[0.5, 1.0, 1.5], E2 =[2.5, 3.0, 3.5],
η =[2.5, 3.0, 3.5], τ∞ =[0.5, 1.0, 1.5], and a =[2.5, 3.0, 3.5] are assumed to be fuzzy variables having triangular
membership functions. A plane strain condition was assumed. A domain size 2b × 2b ( b = 3.0 units) is used
to calculate the mixed-mode SIFs. FEM discretization involved 2711 nodes, 832 8-noded quadrilateral
elements, and 48 focused quarter-point 6-noded triangular elements, as shown in Figure 4.
507
Figure 4 Edge-cracked plate with FEM discretization
The possibility distributions of the mixed-mode SIFs K I and K II are shown in Figure 5. Again, (i) for all
cases the proposed first-order HDMR based implicit and the combinatorial approach yield exactly the same
results for all the possibility distributions of the responses; (ii) however explicit formulation estimates a narrow
bound for the response and with increase in the order of approximation κ the bounds for the response are
observed to become still narrower.
Figure 5 Possibility distributions of KI and KII
4. CONCLUSIONS
In this paper a practical approach based on HDMR for analyzing the response of structures with fuzzy parameters
is presented. The proposed methodology involves integrated finite element modelling, HDMR based response
surface generation, and both, implicit and explicit fuzzy analysis procedures. To facilitate efficient
computation, a HDMR based response surface generation is employed for the approximation of the fuzzy finite
element response quantity. In the proposed implicit formulation, the binary combinations of the fuzzy variables
that result in extreme responses at an α -level are obtained by applying combinatorial optimization to the
approximated response obtained using HDMR. Once the combinations are obtained, full finite element analyses
are then carried out to determine the extreme values of the responses. In the proposed explicit formulation,
using HDMR the response is expressed in a separable closed-form expression as a linear combination of the
fuzzy variables through the definition of intervening variables, then using the transformation technique the
bounds on the response at each α -level are obtained. The proposed explicit formulation is free from the
problem associated with dependency of the fuzzy variables.
508
REFERENCES
[1] L. Zadeh, “Fuzzy sets,” Inf. Contr., 1965; vol.8, no. 3, pp. 338–353.
[2] S. Valliappan, and T. D. Pham, “Fuzzy logic applied to numerical modeling of engineering problems,”
Comp. Mech. Adv., 1995, vol. 2, pp. 213–281.
[3] L. Chen, and S. S. Rao, “Fuzzy finite element approach for the vibration analysis of imprecisely defined
systems,” Finite Elem. Anal. Design, 1997, vol. 27, pp. 69–83.
[4] U. O. Akpan, T. S. Koko, T. R. Orisamolu, and B. K. Gallant, “Practical fuzzy finite element analysis of
structures,” Finite Elem. Anal. Design, 2001, vol. 38, pp. 93–111.
[5] W. M. Dong, and H. C. Shah, “Vertex method for computing function of fuzzy variables,” Fuzzy Sets Sys.,
1987, vol. 2, pp. 201–208.
[6] M. Hanss, “The transformation method for the simulation and analysis of systems with uncertain
parameters,” Fuzzy Sets Sys., 2002, vol. 130, no. 3, pp. 277–289.
[7] S. Donders, D. Vandepitte, V. Peer, and W. Desmet, “Assessment of uncertainty on structural dynamic
responses with the short transformation method,” J. Sound Vibr., 2005, vol. 288, no.3, pp. 523–549.
[8] B. Möller, W. Graf, and M. Beer, “Fuzzy structural analysis using α-level optimization,” Comp. Mech.,
2000, vol. 26, no.6, pp. 547–565.
[9] L. Farkas, D. Moens, D. Vandepitte, and W. Desmet, “Application of fuzzy numerical techniques for
product performance analysis in the conceptual and preliminary design stage,” Comp. Struc., 2008, vol. 86,
no.10, pp. 1061–1079.
[10] S. S. Rao, and L. Berke, “Analysis of uncertain structural systems using interval analysis,” AIAA J., 1997,
vol. 34, no. 4, pp. 727–735.
[11] R. L. Muhanna, R. L. Mullen, “Uncertainty in mechanics problems interval based approach,” J. Engg.
Mech., ASCE, 2001, vol. 127, no. 6, pp. 557–566.
[12] H. Rabitz, and O. F. Alis, “General foundations of high dimensional model representations,” J. Math.
Chem., 1999, vol. 25, no. 2−3, pp. 197–233.
[13] R. Chowdhury, B. N. Rao, and A. M. Prasad, “High dimensional model representation for piece wise
continuous function approximation,” Commun. Num. Meth. Engg., 2008, vol. 24, no. 12, pp. 1587–1609.
[14] P. R. Adduri, and R. C. Penmetsa, “Confidence bounds on component reliability in the presence of mixed
uncertain variables,” Int. J. Mech. Sci., 2008, vol. 50, no. 3, pp. 481–489.
[15] R. E. Moore, “Methods and applications of interval analysis,” SAIM Publication, Philadelphia 1979.
509
Paper No.: 090
SEISMIC DESIGN FACTORS FOR STEEL MOMENT FRAMES WITH

MASONRY INFILLS: PART 1
Shiv Shanker Ravichandran1 and Richard E. Klingner2

1
Design Engineer, Leslie E. Robertson Associates Consulting Engineers (India) Pvt. Ltd, Mumbai, India.
Former Graduate Research Assistant, The University of Texas at Austin, Austin, Texas, USA,
r.shivshanker58@utexas.edu
2
L. P. Gilvin Professor in Civil Engineering, The University of Texas at Austin, Austin, Texas, USA,
klingner@mail.utexas.edu.
ABSTRACT
In this two-part work, seismic behavior and design of steel moment frames with masonry infills are investigated
systematically. In this first part, the “infill strength ratio” (the ratio of the story shear strength of infills to the
story shear strength of the bare frame) is shown to have a fundamental effect on the seismic behavior of an
infilled frame. This fundamental effect is demonstrated using pushover analysis of an example steel moment
frame with masonry infills in uniformly infilled and open ground story configurations. In general, infill strength
ratios greater than about 0.35 are associated with progressive deterioration of seismic performance, leading to
story mechanisms concentrated in the lower stories. Greater infill strength ratios can also lead to local shear
failures in frame members.
KEYWORDS: AAC, Autoclaved Aerated Concrete; Design; Earthquake; Infilled Frames
1. INTRODUCTION
An infilled frame, shown in Figure 1, is a structural system in which a frame of steel or reinforced concrete is
filled with a panel of another material, such as concrete, clay, or autoclaved aerated concrete (AAC) masonry.
The infill is generally a partition wall that also can serve as a thermal, acoustical or fire barrier. Infills can be
located in a frame in many ways, ranging from placement in all bays and stories of the frame (Figure 1 (a)), to
random placement (Figure 1 (d)). The open ground story system of Figure 1 (b) is popular system in many parts
of the world, as it leaves the ground story available for lobby or parking.
Figure 1: Example locations of infills in the panels of a frame

Under lateral loads, infills act as compression struts extending between diagonally opposite corners of the
infilled bay, as shown in Figure 2. The infill stiffens the frame, significantly reducing drift under lateral loads
from service level winds or earthquakes. The infill also strengthens the frame, and if properly designed may be
able to reduce damage and probability of collapse under strong earthquakes.
Figure 2. Structural action of infills
1.1 Past experience with infilled frames during earthquakes
Evidence abounds regarding the beneficial effect of infills on seismic behavior of frames. In numerous
instances, infilled frames have withstood an earthquake that collapsed otherwise similar frames without infills.
Hamburger (2006) discusses the superior structural performance and fire resistance of steel-frame buildings
with clay brick or hollow clay tile masonry infills, compared with that of other structural systems during the
1906 San Francisco earthquake. Murty (2000) notes the excellent performance of reinforced concrete frames
with masonry infills in India during moderate earthquakes.
F cos θ F sin θ
θ
F sin θ F cos θ
(a) (b)
Figure 3: a) Story mechanisms due to irregular placement of infills. b) Shear forces on frame members due to
infill strut action
However, infilled frames are also known to adversely affect the behavior of frame during earthquakes.
Saatcioglu (2001) notes the poor performance of reinforced concrete infilled frames, the primary structural
system used in Turkey, during the Kocaeli earthquake of 1999. Being stiffer, the infills attracted higher seismic
loads, which could only be withstood as long as the infills remained elastic. After the infills started to degrade,
the remaining frames did not have enough lateral load resistance or inelastic deformation capacity, and often
collapsed. Of particular concern is the tendency of infilled frames to form adverse failure mechanisms, such as
those shown in Figure 3 (a), that lead to premature structural collapse. Such failure mechanisms are generally a
result of irregular placement of infills. Even when infills are placed in a regular manner over the height of the
frame, Dolsek (2001) observes that weak-story failure mechanism can still occur. Irregular placement of infills
in plan also leads to torsional effects. Throughout the world, a particularly common irregularity in the placement
of infills is the open ground story (Figure 1 (b)). As shown on the right side Figure 3 (a), inelastic story drifts
511
tend to be concentrated at the open ground story, leading to collapse there. This is extensively presented in
Humar (2001), Scawthorn (2000), Saatcioglu (2001) and elsewhere. Apart from failure mechanisms resulting
from weak stories, infills can also cause local shear failures in adjacent frame members. Development of strut
action in the infill requires that the horizontal and vertical components of the compressive force be resisted by
the bounding columns and beams, respectively (Figure 3 (b)). This can result in local shear failure in frame
members, as observed in lab specimens of Al-Chaar (2002).
1.2 Basic Elements of this Study
In this study, a steel moment frame is selected for investigation, and the performance of that frame is
investigated, without and with infilling, using the ATC-63 methodology (ATC-63 2008), developed as a
systematic means of establishing seismic design factors for structural systems. This paper describes the start of
that study; and the companion paper describes the remainder.
Steel moment frame of this study
30 ft
W24x84
x x x x x x x x
13 ft W24x162
W24x84
x x x x x x x x
W24x162
W30x108
x x x x x x x x
W24x207
W30x108
x x x x x x x x
W24x207
x - Location of RBS
Figure 4: Four-story steel moment frame from ATC-63 (2008) (“ATC-63 steel moment frame”)
This work uses an example steel moment frame in ATC-63 (2008) to study behavior of steel moment frames
with masonry infills. It is a four-story special steel moment frame (R = 8), shown in Figure 4. It is one of the
four perimeter frames providing lateral support to a four-story building. The frame is designed for Seismic
Design Category D and base shear of 309 kips according to the requirements of ASCE7-05 (ASCE 2005) and
AISC (2005). The beams have reduced beam sections (RBS) with 45% reduction in the flanges and are located
15 in. from the face of column flanges. The beam-column connections conform to “strong column-weak beam”
concept. From now on in this work, this steel moment frame is referred to as the “ATC-63 steel moment frame.”
1.3 Fundamental importance of infill strength ratio
Preliminary pushover analyses of the ATC-63 steel moment frame with uniform infills and open ground story
infills indicated that the failure mechanism of the infilled frame depends in general on the ratio of the story
shear strength of the infills to the story shear strength of the bare frame. In this work, this ratio is termed the
“infill strength ratio.” Infill configurations whose total lateral strength in a particular story exceeds about 35% of
the lateral strength of the bare frame in that story are observed to provoke story mechanisms in the frame and
local shear failures in frame members.
The story shear strength of the bare frame is computed using Equation 1, considering a story mechanism that
leads to hinges at the top and bottom of columns at that story (
512
Figure 5). If axial loads in columns are significant, then their effect on plastic moment capacity of column
sections should be considered.
Fstory mechanism =
∑M p
(1)
h
where,
Fstory mechanism = shear strength of the story under consideration
Mp = plastic moment capacity of columns at the story under consideration
h = height of the story under consideration
The story shear strength of infills is calculated as the total expected lateral shear strength of all infill panels at
that story.
Figure 5: Story mechanism used to calculate story shear strength of bare frame
1.4 Analytical modeling of steel moment frames with masonry infills
The analytical model of the ATC-63 steel moment frame with masonry infills is developed in OpenSees©
(OpenSees 2006). Frame members are modeled using elastic beam-column elements and expected locations of
plastic hinges are modelled using point plasticity springs. The moment-rotation properties of the W-sections are
computed using Lignos (2007), which uses the bilinear Ibarra-Krawinkler hysteretic model (Ibarra 2005) to
represent hysteretic behavior of plastic hinges in steel frame members.
Point Plasticity Beam- Column

Spring Joint Element
Frame
element
(a) (b)
Equivalent struts to
Elastic Beam-Column model infill
Lumped Mass
Element
Figure 6: (a) Analytical model of ATC-63 steel moment frame (b) Idealization of infills as equivalent struts
Infill cases considered for study
513
The behavior of panel zones in beam-column joints is modeled using the Joint2D element in OpenSees, based
on Lowes (2003). Infills are idealized as equivalent struts (Figure 6 (b)) (Madan 1997, Negro 1997, Combescure
2000, Crisafulli 2000, Dolsek 2001, Dolsek 2002). In the equivalent-strut approach, the infill is represented as a
combination of two compression-only truss elements, each acting independently. Each equivalent strut element
is assigned the Ibarra-Krawinkler hysteretic model (Ibarra 2005) with pinched hysteretic rules to model the
hysteretic force-deformation behavior of the infill.
The infill cases to be used with the ATC-63 steel moment frame are selected to represent a broad range of infill
strength ratios. In doing so, in any story that is infilled, infills are considered present in all the bays, as shown in
Figure 1 (a) and (b). Further in each panel, the infill is assumed to be tightly fitted against the frame members
without any gaps or connectors. Accordingly, three infill cases are selected: the first, with an infill strength ratio
below 0.35, the second, with a ratio of about 0.5; and the third, with a ratio of about 1. For the first case, infills
are chosen to be made of 8-in. thick Class 4 Autoclaved Aerated Concrete (AAC) units because of its low
strength. The second and third cases correspond more naturally to conventional masonry, with its higher
strength. In particular, the second and third cases can represent clay masonry with specified compressive
strengths of 4 ksi and thicknesses of 8 and 12 in., respectively. These are summarized in Table 1.
Table 1: Infill cases considered for evaluation

Specified compressive
Infill case Infill Material Thickness of infill (in.)
strength (psi)
1 AAC, Class 4 8 580
2 Conventional masonry (clay) 8 4000
3 Conventional masonry (clay) 12 4000
Starting with the ATC-63 steel moment frame, and infilling all bays with the infill cases presented in Table 1,
results in three infilled steel moment frames for the uniformly infilled configuration (Figure 1(a)), and three
more for the open ground story configuration (Figure 1 (b)). This total of 6 infilled steel moment frames is
summarized in Table 2, along with the terminology used to describe each of these infilled frames in the rest of
this work.
Table 2: Infilled steel moment frames used for study
Infill Configuration
Infill material
Uniformly infilled frames Open ground story frames
8-in. thick Class 4
AAC uniformly infilled frame AAC open ground story frame
AAC
8-in. thick 8-in. thick conventional 8-in. thick conventional
conventional masonry uniformly infilled masonry open ground story
masonry frame frame
12-in. thick 12-in. thick conventional 12-in. thick conventional
conventional masonry uniformly infilled masonry open ground story
masonry frame frame
1.5 Properties of infill materials
The stiffness, strength and hysteretic force-deformation behavior of the 8-in. thick Class 4 AAC infill is
modeled as described in Ravichandran (2009) using the Ibarra-Krawinkler hysteretic model (Ibarra 2005). The
strength of the AAC infill for this evaluation is conservatively taken as the maximum shear strength, 97.4 kip,
obtained for the 8-in. thick Class 4 AAC infill in the infilled frame specimen described in Ravichandran (2009).
514
By Table 1 of MSJC (2008a), for conventional masonry the specified strength fm′ of 4 ksi can be achieved using
ASTM C270 Type S mortar and clay masonry units with a tested strength of 11.5 ksi. The corresponding actual
strength of masonry, fm , is assumed as 6 ksi. The initial stiffness, strength and deflection at peak strength are
calculated based on Flanagan (2001). The residual branch of the monotonic curve is modeled using
recommendations by Flanagan (1999) such that the strength drops to 75% of the peak strength at a deformation
1.5 times that at peak strength. The values of pinching parameters kf and kd for the Ibarra-Krawinkler hysteretic
model that best represent the behavior of the AAC infill in the infilled frame specimen of Ravichandran (2009)
are 0.15 and 0.5, respectively and these are used for conventional masonry infills as well.
1.6 Summary of infill strength ratios for the infill cases
The story shear strength of the ATC-63 steel moment frame at the bottom story is computed using Equation 1
corresponding to expected yield strength of steel (55 ksi) as 2137 kip. Table 3 provides a summary of the infill
strength ratios at the bottom stories of the ATC-63 steel moment frame for the infill cases.
Table 3: Infill strength ratio for different infill cases at the bottom story of the ATC-63 steel moment frame
Infill Strength of four

Infill Infill strength ratio at
Infill type thickness infill panels at a
case bottom story
(in.) story (kip)
1 AAC, Class 4 8 390 0.18
2 Conventional masonry 8 1448 0.68
3 Conventional masonry 12 2172 1.01
1.7 Pushover analysis of uniformly infilled frames
Pushover analysis is performed on the analytical models of the ATC-63 steel moment frame as well as the
uniformly infilled frames listed in Table 2. The lateral load profile computed using ASCE7-05 for the ATC-63
steel moment frame is used for the uniformly infilled frames as well so that the results obtained from the
pushover analysis would be comparable for all the frames.
3500
Bare Frame
3000 AAC infill
Conventional masonry infill, 8-in. thick
2500
Base Shear (kip)
Conventional masonry infill, 12-in. thick
2000
1500
1000
500
0
0.00 0.02 0.04 0.06 0.08
Roof Drift Ratio
Figure 7: Pushover curve for the uniformly infilled frames
Figure 7 presents the pushover curves. In Table 4, the stiffness, strength and ductility of the uniformly infilled
frames are compared with that of the ATC-63 steel moment frame. The ductility is calculated as recommended
515
by ATC-63 (2008). In general, increasing infill strength ratio progressively increases initial stiffness and
strength while decreasing the ductility.
Bare Frame
4 AAC infill
Conventional masonry
Height above ground infill, 8-in. thick
3
Story height Conventional masonry
infill, 16-in. thick
0
0 0.02 0.04 0.06 0.08
Horizontal Displacement
Height of frame
Figure 8: Displacement profile of the uniformly infilled frames from pushover analyses
In Figure 8 are compared the displaced profiles of the ATC-63 steel moment frame and the uniformly infilled
frames at the ultimate roof drift ratio of the ATC-63 steel moment frame, determined as prescribed by ATC-63
(2008). From the displaced profile of the uniformly infilled frames, the failure mechanism of the frames can be
deduced. The pushover failure mechanism of the AAC uniformly infilled frame is same as that of the bare
frame. The failure mechanism of the 8-in. thick conventional masonry uniformly infilled frame is concentrated
in the bottom three stories. For the 12-in. thick conventional masonry uniformly infilled frame, the failure
mechanism is limited to the bottom two stories. This was corroborated by increase in the plastic rotation demand
in hinges of beams and columns in these stories compared to the ATC-63 steel moment frame. Further studies
using pushover analysis indicated that an infill strength ratio of 0.35 marks the onset of change in failure
mechanism compared to the bare frame for the ATC-63 steel moment frame in uniformly infilled configuration.
Also, pushover analysis with stronger infills indicated that a weak ground story mechanism would form at an
infill strength ratio of about 2.0.
Table 4: Stiffness, strength and ductility of the uniformly infilled frames from pushover analysis
Initial stiffness with

Strength divided by Displacement
Infill Type respect to ATC-63
design base shear ductility
steel moment frame
Bare frame 1.0 4.0 6.2
AAC, 8-in. thick 2.7 5.3 4.1
Conventional
10.1 8.1 2.6
masonry, 8-in. thick
Conventional
12.5 10.5 2.3
masonry, 12-in. thick
For the uniformly infilled steel moment frames, axial forces in exterior and interior columns were monitored
during the pushover analysis. Compared to the bare frame, the axial forces in exterior columns of the AAC
uniformly infilled frame increase progressively from top to bottom. Because the infills in all stories crushed at
about the same time during the pushover analysis, the differences at each story between the maximum axial
force in a particular column line of the ATC-63 steel moment frame and the same column line in the AAC
uniformly infilled frame, are multiples of 40 kips (the vertical component of the axial strength of the equivalent
516
strut). In Column Line 1 (Figure 9), the difference in maximum axial force between the ATC-63 steel moment
frame and AAC uniformly infilled frame was equal to 40 kips at the fourth story, 80 kips at the third story, and
so on down to the bottom story. For Column Line 5, in axial compression, the vertical components in the
equivalent struts cascade down the columns from top of the third story in increments of 40 kips reaching 120
kips at the bottom story. In interior column lines, in contrast, AAC infills do not cause significant changes in
axial forces compared to the bare frame. This is because the vertical components of the forces in the equivalent
struts framing into opposite sides of an interior column tend to neutralize each other, except at the top story.
Because of the relatively weak AAC infills, the maximum increase in axial force (160 kips in tension and 120
kips in compression) are only 4.8% and 3.6%, respectively, of the concentric axial capacity of the ground story
column. These increases do not significantly reduce the moment capacity or ductility of the columns.
Column lines
1 2 3 4 5
Shear failure in interior columns

Shear failure in beams of exterior bays
Shear failure in beams of interior bays
Figure 9: Equivalent struts active during pushover analysis of archetypical uniformly infilled frames and
illustration of non-simulated collapse modes in infilled frames due to simultaneous shear failure in frame
members
The trend in variation of column axial forces in the conventional masonry uniformly infilled frames is similar to
that observed in the case of the AAC uniformly infilled frame, although with some differences. In contrast to the
case of AAC uniformly infilled frame, however, the increase between two adjacent stories relative to the ATC-
63 steel moment frame is consistently less than the vertical component of the axial strength of the equivalent
strut. The first reason for this is that the infills did not crush completely at the upper stories, and those upper
stories did not participate in the failure mechanism of the frame. The second reason is that axial force in the
columns due to frame action is generally less than in the ATC-63 steel moment frame, because the failure
mechanism of the frame does not involve beam hinging at all stories. The maximum increase in axial force
relative to the ATC-63 steel moment frame in the ground story of Column Line 1, is about 500 kip and 800 kip
for the 8-in. and 12-in thick conventional masonry uniformly infilled frames, respectively. These are 17% and
27%, respectively, of the axial capacity of the ground story columns. Such large increases in axial force may
affect the moment capacity or ductility of the columns.
1.8 Non-simulated collapse modes in infilled frames
When the infill acts as a diagonal strut, it causes local shear forces in frame members, which add to the shear
forces due to frame action (Figure 3 (b)). If the combined shears exceed the shear capacity, the frame member
can fail locally in shear. This possibility could not be directly considered in the analysis, for the following two
reasons:
1) Failure due to shear is not included in the analytical model used for frame members
2) Equivalent struts representing infills connect to the center of beam-column joints, and transfer their
axial forces as point loads to the frame
Using data from the analysis, however, the local shear failures in frame members can be addressed as non-
simulated collapse modes (ATC-63 2008). This was done for infilled frames as explained here. The axial forces
in equivalent struts representing infills were monitored during the analysis. At each instant of the analysis, the
517
appropriate component of the axial force in the equivalent struts is added to shear force in frame members due
to frame action alone. This gives the total shear force in frame members due to frame action plus frame-infill
interaction forces.
At any particular story, because the forces in the equivalent struts and shear forces in interior columns are found
to be nearly uniform, a shear failure in Column Line 3 is considered to indicate a shear failure in Column Line 2
and Column Line 4 as well. Similarly, a shear failure in the beam of one of the interior bays is considered shear
failure in beams of all interior bays, and a shear failure in the beam of one of the exterior bays is considered
shear failure in beams of both exterior bays. Shear failure in such a large number of structural members can
significantly impair the stability of the structure, and can effectively be treated as the failure of the entire
structure. From this, it was detected that the local shear capacity of frame members was exceeded only for the
12-in. thick conventional masonry uniformly infilled frame in the columns of bottom three stories thereby
reducing its ductility from 2.3 to 1.3. Therefore, stronger infills can lead to local shear failure in frame members
due to frame-infill interaction forces.
1.9 Pushover analysis for open ground story frames
Pushover analysis is performed on the analytical models of the open ground story frame listed in Table 2, in the
same manner as was done for the uniformly infilled frames. Results for the open ground story frames are similar
to those of the uniformly infilled frames. However, the formation of hinges in columns progresses faster towards
the ground story for the open ground story configuration reaching it at an infill strength ratio of about 1.0 rather
than 2.0 with the uniformly infilled configuration. Also, for all infill cases occurrence of local shear failure in
frame members did not occur for the open ground story frames.
2 SUMMARY AND CONCLUSIONS
An important parameter defining behavior of infilled frames is shown to be the ratio of story shear strength of
infills to the story shear strength of bare frame. This ratio is termed the “infill strength ratio.” The effect of infill
strength ratio is demonstrated using pushover analysis of an example steel moment frame with masonry infills.
From the results of those analyses, it can be said that for uniformly infilled frames:
1) If the infill strength ratio is less than about 0.35, the presence of infills does not change the failure
mechanism, which involves hinging in beams and at column bases.
2) When the infill strength ratio reaches about 0.35, the presence of the infill begins to change the failure
mechanism of the frame, from hinging in beams and at column bases, to story mechanisms involving
column hinging at multiple levels of the lower stories.
3) When the infill strength ratio reaches about 1.0, infills significantly change the failure mechanism of the
frame, concentrating it in the bottom half of the frame.
4) When infill strength ratio reaches about 2.0, infills consistently lead to ground story mechanisms.
Results from pushover analysis of the example steel moment frame with an open ground story configuration are
similar to those with uniformly infilled configuration except that the formation of hinges in columns progresses
faster towards the ground story reaching it at an infill strength ratio about 1.0 rather than about 2.0 for the
uniformly infilled configuration. Stronger infills also lead to local shear failure in frame members due to frame-
infill interaction forces.
In Part 2 of this work, archetypical steel moment frames with masonry infills of different infill strength ratios
are evaluated using the ATC-63 methodology (ATC-63 2008) to determine seismic design factors and design
guidelines for steel moment frames with masonry infills.
ACKNOWLEDGEMENT
The authors gratefully acknowledge the financial and technical assistance provided by the Autoclaved Aerated
Concrete Products Association, Xella AAC, Xella Mexicana and Trustone AAC for the work described in this
paper.
518
REFERENCES
[1] AISC, 2005. Manual of Steel Construction, 13th edition, printed by American Institute of Steel
Construction, Chicago, Illinois, USA.
[2] Al-Chaar, G., Issa, M., and Sweeney, S., 2002. Behavior of masonry-infilled nonductile reinforced
concrete frames, Journal of Structural Engineering, Vol. 128, No. 8, 1055-1063.
[3] ASCE, 2005. Minimum Design Loads for Buildings and Other Structures, ASCE Standard No. 7-05,
American Society of Civil Engineers and Structural Engineering Institute, Virginia, USA.
[4] ATC-63, 2008. Quantification of Building Seismic Performance Factors - 90% Draft, Report No. ATC-
63, prepared by the Applied Technology Council, Redwood City, California for the Federal Emergency
Management Agency, Washington, D.C.
[5] Combescure, D., and Pegon, P., 2000. Application of the local to global approach to the study of infilled
frame structures under seismic loading, Proceedings of the 12th World Conference on Earthquake
Engineering, Auckland: New Zealand, CD-ROM, Paper No. 505.
[6] Crisafulli, F., J., Carr, A., J., and Park, R., 2000. Analytical modeling of infilled frames structures - a
general review, Bulletin of the New Zealand Society for Earthquake Engineering, 2000, Vol. 33, 30–47
[7] Dolsek, M., and Fajfar, P., 2001. Soft storey effects in uniformly infilled reinforced concrete frames,
Journal of Earthquake Engineering, 2001, Vol. 5, No. 1, 1-12
[8] Dolsek, M., and Fajfar, P., 2002. Mathematical modelling of an infilled RC frame structure based on the
results of pseudo-dynamic tests, Earthquake Engineering and Structural Dynamics, 2002, Vol. 31, 1215–
1230
[9] Flanagan, R., D., and Bennett, R., M., 1999. In-plane behavior of structural clay tile infilled frames,
Journal of Structural Engineering, Vol. 125, No. 6, 590-599.
[10] Flanagan, R. D., and Bennett, R. M., 2001. “In-plane analysis of masonry infill materials”, Practice
periodical on Structural Design and Construction, Vol.6 No.4, 176-182
[11] Hamburger, R.O., and Meyer, J. D., 2006. The performance of steel-frame buildings with infill masonry
walls in the 1906 San Francisco earthquake, Earthquake Spectra, Volume 22, No. S2, S43-S67.
[12] Jag Mohan Humar, David Lau, and Jean-Robert Pierre, 2001. Performance of buildings during the 2001
Bhuj earthquake, Canadian Journal of Civil Engineering, Vol. 28, 979–991
[13] Ibarra, L. F., Medina, R. A., and Krawinkler, H., 2005. Hysteretic models that incorporate strength and
stiffness deterioration, Earthquake Engineering and Structural Dynamics, Vol. 34, 1489–1511.
[14] Lignos, D.G., and Krawinkler, H., 2007. A Database in Support of Modeling of Component Deterioration
for Collapse Prediction of Steel Frame Structures, Proceedings, ASCE Structures Congress, Long Beach,
California, May 18-20.
[15] Lowes, L. N. and Altoontash, A., 2003. Modeling of Reinforced-Concrete Beam-Column Joints
Subjected to Cyclic Loading, Journal of Structural Engineering, ASCE, Vol. 129 (12).
[16] Madan, A., A. M. Reinhorn, and J. B. Mander, 1997. Modelling of masonry infill panels for structural
analysis, Journal of Structural Engineering, ASCE, 1295-1302.
519
[17] Murty, C. V. R. and Jain, S. K., 2000. Beneficial influence of masonry infill walls on seismic
performance of RC frame buildings, Proceedings of the 12th World Conference on Earthquake
Engineering, Auckland, New Zealand, CD-ROM, Paper No. 1790.
[18] MSJC, 2008a. Building Code Requirements for Masonry Structures, Masonry Standards Joint Committee,
The Masonry Society, Boulder, Colorada, USA
[19] MSJC, 2008b. Specification for Masonry Structures, Masonry Standards Joint Committee, The Masonry
Society, Boulder, Colorada, USA
[20] Negro, P., and Colombo, A., 1997. Irregularities induced by nonstructural masonry panels in framed
buildings, Engineering Structures, 19(7), 576-585.
[21] OpenSees, 2006. Open System for Earthquake Engineering Simulation (http://opensees.berkeley.edu/),
developed by University of California Berkeley and Pacific Earthquake Engineering Research Center
(http://peer.berkeley.edu), California, USA.
[22] Ravichandran, S., 2009. Design provisions for Autoclaved Aerated Concrete (AAC) infilled steel moment
frames, PhD Dissertation, December 2009, The University of Texas at Austin, USA.
[23] Saatcioglu, M., Mitchell, D., Tinawi, R., Gardner, N., J., Gillies, A., G., Ghobarah, A., Anderson, D., L.,
and Lau, D., 2001. The August 17, 1999, Kocaeli (Turkey) earthquake — damage to structures, Canadian
Journal of Civil Engineering, Vol. 28, 715-737
[24] Scawthorn, C., and Johnson, G. S., 2000. Preliminary report Kocaeli (Izmit) earthquake of 17 August
1999, Engineering Structure
520
Paper No.: 091
SEISMIC DESIGN FACTORS FOR STEEL MOMENT FRAMES WITH

MASONRY INFILLS: PART 2
Shiv Shanker Ravichandran1 and Richard E. Klingner2
1
Design Engineer, Leslie E. Robertson Associates Consulting Engineers (India) Pvt. Ltd, Mumbai, India.
Former Graduate Research Assistant, The University of Texas at Austin, Austin, Texas, USA,
r.shivshanker58@utexas.edu
2
L. P. Gilvin Professor in Civil Engineering, The University of Texas at Austin, Austin, Texas, USA,
klingner@mail.utexas.edu.
ABSTRACT
In this second of two parts, archetypical steel moment frames with masonry infills are evaluated using the
ATC-63 methodology, which is a systematic procedure for determination of seismic design factors of
structural systems. The ATC-63 methodology is briefly reviewed. Procedures in the ATC-63 methodology
are specialized for application to infilled steel moment frames. Results from the ATC-63 evaluation of the
archetypical infilled steel moment frames are presented, and are used to propose seismic design factors and
design guidelines for steel moment frames with masonry infills. The applicability of the proposed seismic
design factors and guidelines for reinforced concrete infilled frames is discussed.
KEYWORDS: AAC, Autoclaved Aerated Concrete; Design; Earthquake; Infilled Frames
1. INTRODUCTION
The seismic performance of infilled frames has been widely studied (Klingner 1978, Panagiotakos 1993,
Kodur 1995, Dolsek 2001, Dolsek 2004, Dolsek 2008, Krawinkler 2007). The general qualitative insights
from these investigations can be summarized as follows:
1) Increases in strength and ductility of frames due to infilling may offset the increased seismic force
demand as a result of increased stiffness due to infilling.
2) Infills can decrease the force and deformation demand on the frame, even though they shorten the
fundamental period and consequently increase seismic forces. Because infills are generally stiffer than
the frame, they can reduce some actions in frame members and can also reduce lateral drifts.
3) Infills can significantly change the response of a frame; they can change the distribution of damage; and
their effects should be considered in design. Weak story mechanisms are possible even in uniformly
infilled frames, and open ground story frames tend to concentrate failure mechanisms in the ground
story.
4) The most important parameter affecting the seismic performance of infilled frames is the strength of the
infill. Less important parameters include infill ductility, residual strength, and the slope of the
descending branch of their backbone curve.
However, quantitative recommendations regarding seismic performance of infilled frames are still inconsistent
and inconclusive. Design provisions for infilled frames in modern building codes are comprehensively treated in
Kaushik (2006), which reviews national building codes of 16 countries. Kaushik rightly notes the urgent need to
adequately address infilled frames as a structural system in national building codes. Therefore, the goal in
second part of this work is to develop design provisions, including seismic design factors, for steel moment
frames with masonry infills. This goal is broadly addressed by the systematic ATC-63 methodology (ATC-63
2008), which is considered the state of the art in the United States for determination of seismic design factors for
structural systems.
1.1 Brief review of ATC-63 Methodology
The first steps in the ATC-63 methodology involve proposing the general concept of the structural system under
consideration, and identifying its seismic force-resisting system. “Archetypical structures,” which broadly
represent the type of structures that would probably be built using the proposed structural system, are envisaged.
The methodology requires that archetypical structures be designed for the loads prescribed by ASCE7-05
(ASCE 2005), and hence requires basic design rules for determining the proportions and details of structural
components of archetypical structures. The archetypical structures are designed for the seismic loads prescribed
by ASCE7-05 using trial values of seismic design factors, particularly the response reduction factor R.
1
0.9
Probability of collapse
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 1 2 3 4 5 6 7 8
Spectral acceleration ( g )
Figure 1: Collapse fragility curve
The ATC-63 methodology prescribes nonlinear analysis procedures based on pushover and Incremental
Dynamic Analysis (IDA) (Vamvatsikos 2002) to evaluate the seismic performance of archetypical structures.
This requires the development and calibration of nonlinear hysteretic models for structural components,
especially those that contribute to inelastic behavior of the structure. Using the calibrated hysteretic models of
components, analytical models of the archetypical structures are developed in a nonlinear structural analysis
program to assess their collapse capacity. The final step in the evaluation of an archetypical structure is the
determination of the collapse fragility curve (Figure 1), which essentially presents the probability of collapse of
the archetypical structure as a function of spectral intensity. From the collapse fragility curve of each
archetypical structure, its probability of collapse at MCE (Maximum Considered Earthquake) is determined.
The methodology requires that the probability of collapse at MCE of the archetypical structures and average
probability of collapse at MCE of performance groups be less than about 20% and 10%, respectively. If at the
end of the evaluation process the archetypical structures designed using the trial value of R do not meet this
performance criteria, they are required to be re-designed and re-evaluated using either a lower value of R,
improved design methods, improved detailing, or other strategies.
1.2 Structural system addressed by this work
This work primarily addresses steel moment frames with masonry infills up to five stories high, with bay widths
and story heights of 20 to 30 ft and 10 to 15 ft, respectively. The highest seismic design category in which they
522
are intended to be located in the United States is Seismic Design Category D (ASCE7-05). Two different infill
configurations are considered: uniformly infilled frames; and uniformly infilled frames with an open ground
story whose definitions are provided in Part 1 of this work.
1.3 Archetypical steel moment frames with masonry infills
The ATC-63 steel moment frame, presented in Part 1 of this work, is used to develop archetypical infilled steel
moment frames because it is a reasonable choice for an archetypical steel moment frame. Therefore, when the
ATC-63 steel moment frame is infilled with typical infill materials, it is reasonable to expect the resulting frame
to be an archetypical infilled steel moment frame. The infill cases to be used with the ATC-63 steel moment
frame are same as those in Part 1, and the corresponding archetypical infilled steel moment frames are same as
those presented in Part 1. The ATC-63 steel moment frame is evaluated in this work as a bare frame, and is
found to meet the performance criteria of the ATC-63 methodology. Therefore, performance of the bare ATC-
63 steel moment frame can be used as a benchmark to assess the effect of infills on the behavior and
performance of the corresponding archetypical infilled steel moment frames.
1.4 Trial seismic design factors and design
Based on the preliminary observation in Part 1 that for lower ratios of infill strength to frame strength the
original failure mechanism of the bare frame was not altered, it was decided to use the same trial seismic design
factors for design of the steel moment frame with infills, as are currently mandated by ASCE7-05 for the bare
frame. Therefore, the design of the steel moment frame remains unchanged. Thereafter, the infills are simply
placed in the panels of the frame. This is an important assumption, and its validity will later be re-visited.
1.5 IDA for infilled frames
Procedures for IDA are specialized for application to infilled steel moment frames. A fundamental feature of
IDA is scaling of ground-motion intensities, determined as the spectral intensity at the fundamental period of the
archetypical structure, until the ground motions cause collapse of the archetypical structure. However, the bare
ATC-63 steel moment frame and the archetypical infilled steel moment frames have different fundamental
periods, due to their different stiffnesses. To make IDA plots and intermediate results comparable, for all
archetypical infilled steel moment frames, ground motion intensity during IDA is generally scaled at the
fundamental period of the bare ATC-63 steel moment frame. If required, a spectral intensity determined at the
fundamental period of the ATC-63 steel moment frame could readily be converted to the spectral intensity at the
fundamental period of the corresponding archetypical infilled steel moment frame.
Figure 2: Rule for determination of non-simulated collapse of archetypical infilled steel moment frames due to
local shear failure in structural members during IDA
523
1.6 Non-simulated collapse modes
As explained in Part 1 for pushover analysis, local shear failure in frame members due to frame-infill interaction
forces is treated as non-simulated collapse mode during IDA as well. However, during IDA local shear failures
due to frame-infill interaction forces may not occur at all spectral intensities for a particular ground motion. In
this work, if local shear failure is detected at more than one spectral intensity for a particular ground motion
during IDA, then non-simulated collapse is considered to occur at the second (higher) of these spectral
intensities. This is illustrated in Figure 2.
1.7 Application of ATC-63 Methodology to archetypical infilled steel moment frames
In this section, results from the application of the ATC-63 methodology to the bare ATC-63 steel moment frame
and the archetypical infilled steel moment frames are presented.
1.8 Results for archetypical uniformly infilled steel moment frames
The ATC-63 methodology prescribes that the limiting fundamental period of the structure be calculated using
Section 12.8.2 of ASCE7-05. Table 1 presents the fundamental period of the ATC-63 steel moment frame and
the archetypical uniformly infilled frames obtained using ASCE7-05. Table 1 also presents the formulas in
ASCE7-05 that are used to determine the respective fundamental periods.
Table 1: Fundamental period of archetypical uniformly infilled frames
Formula for determination of Code-defined limiting

Infill type
fundamental period by ASCE7-05 fundamental period (s)
Bare frame Steel moment frame 0.93
AAC, 8-in. thick General formula for moment frames 0.56
Conventional masonry, 8- in. thick Masonry shear wall 0.17
Conventional masonry, 12-in. thick Masonry shear wall 0.14
The results from pushover analysis of the archetypical infilled steel moment frames have already been presented
in Part 1 of this work. Figure 3 depicts results including non-simulated collapse modes from IDA performed on
the ATC-63 steel moment frame and the archetypical uniformly infilled frames. Table 2 presents the median
collapse spectral intensities at the fundamental period of the bare frame obtained for the archetypical uniformly
infilled steel moment frames with and without considering local shear failure in frame members as non-
simulated collapse modes. This shows that if only global side-sway failure mechanism of the frame is
considered for collapse evaluation, increasing the infill strength ratio increases the collapse capacity. However,
large increases in infill strength ratio produce only a small increase in collapse capacity. Local shear failure in
frame members was not detected during IDA for the AAC uniformly infilled frame. For the 8-in. thick
conventional masonry uniformly infilled frame, local shear failure generally occurred in the bottom two stories
and sometimes in the third and fourth stories as well. For the 12-in. thick conventional masonry uniformly
infilled frame, local shear failure generally occurred in columns at all stories. The shear capacities of beams
were never exceeded, probably because the vertical component of the infill force was limited by the large aspect
ratio (bay width divided by story height) of the infill.
524
Table 2: Median collapse spectral intensity (at the fundamental period of the bare frame) for the archetypical
uniformly infilled frames with and without non-simulated collapse modes
Considering only Including local shear failure as

Infill type
sidesway collapse non-simulated collapse mode
Bare frame 2.14 2.14
AAC, 8-in. thick 2.34 2.34
Conventional masonry, 8- in. thick 2.53 1.95
Conventional masonry, 12-in. thick 2.63 0.78
7.5 6.5
7 6
6.5 (a) 5.5 (b)
6
5
5.5
5 4.5
4.5 4
4 3.5
3.5 3
3 SCT[T]
2.5
2.5 SCT [T]
2
2
Sa [ TBareFrame ] (g)
1.5 1.5
1 1
0.5 0.5
0 0
0 5 10 15 0 5 10 15
Maximum interstory
Maximum Interstorey drift
Drift (%)
(%)
Maximum interstory drift (%)
6 2
5.5
5
(c) (d)
4.5 1.5
4
3.5
3 1
2.5 SCT[T]
SCT[T]
2
1.5 0.5
1
0.5
0 0
0 5 10 15 0 0.5 1 1.5 2 2.5
Figure 3: Results from IDA including non-simulated collapse modes for: a) ATC-63 steel moment frame b)
AAC uniformly infilled frame c) 8-in. thick conventional masonry uniformly infilled frame d) 12-in. thick
conventional masonry uniformly infilled frame
In Figure 4 are shown the displaced profiles of the bare ATC-63 steel moment frame and the archetypical
uniformly infilled steel moment frames, when the maximum inter-story drift exceeded 15% during IDA for each
of the 44 ground motions of the ATC-63 suite. The displacement profile for the bare frame and the AAC
uniformly infilled frame are similar, and represent a side-sway failure mechanism distributed over the entire
height of the frame. This indicates that uniform AAC infilling generally does not affect the failure mechanism of
the frame or lead to weak-story mechanisms. The displacement profile for the conventional masonry uniformly
infilled frame represents a failure mechanism limited in most cases to the lower three or two stories. Therefore,
it can be said that a high infill strength ratio leads to story mechanisms even in a uniformly infilled frame
designed according to the strong column-weak beam concept, confirming observations by Dolsek (2001). This
is fundamentally attributable to the infill strength ratio. With an infill strength ratio of 1.0, the failure
mechanism is concentrated in the bottom half of the 12-in. thick conventional masonry uniformly infilled frame.
1.9 Uncertainties in ATC-63 evaluation
The various uncertainties prescribed by the ATC-63 methodology in the evaluation of the ATC-63 steel moment
frame and the archetypical infilled steel moment frames are rated as shown in Table 3, along with the
corresponding lognormal distribution standard deviation parameters.
525
Figure 4: Displacement profile from IDA at 15% maximum interstory drift for: a) bare frame b) AAC uniformly
infilled frame c) 8-in. thick conventional masonry uniformly infilled frame d) 12-in. thick conventional masonry
uniformly infilled frame
Table 3: Rating of uncertainties
Record to Design related Test data Modeling Total system

Archetype
β RTR)
record (β β DR)
(β β TD)
(β βMDL)
(β β TOT)
uncertainty (β
Bare frame 0.4 Good (0.3) Good (0.3) Fair (0.45) 0.75
Infilled frames 0.4 Good (0.3) Fair (0.45) Fair (0.45) 0.80
1.10 Collapse fragility curve and probability of collapse at MCE
To plot the collapse fragility curve and compute the probability of collapse at MCE of the archetypical
uniformly infilled frames, the collapse spectral intensities obtained from IDA are converted to those at the
respective fundamental periods of the archetypical infilled steel moment frames. The MCE level spectral
acceleration demand is also determined at the respective fundamental period of the archetypical infilled steel
moment frames. In Figure 5: Collapse fragility curves for archetypical uniformly infilled steel moment frames
Table 4 are presented the Collapse Margin Ratio (CMR), Spectral Shape Factor (SSF), Adjusted Collapse
Margin Ratio (ACMR) and probability of collapse at MCE computed for the archetypical uniformly infilled
frames. In Figure 5 are shown the corresponding collapse fragility curves, plotted using ACMR as the median,
and the total system collapse uncertainty as the lognormal distribution standard deviation parameter, of the
lognormal distribution of collapse spectral intensities. The probability of collapse at MCE of the AAC uniformly
infilled frame is slightly lower than that of the ATC-63 steel moment frame, while the probability of collapse of
the conventional masonry uniformly infilled frames is much greater.
526
1
Probability of collapse
0.9
0.8
0.7
0.6
0.5
0.4
0.3 Bare Frame
0.2 AAC infill
Conv. masonry infill, 8-in. thick
0.1 Conv. masonry infill, 12-in. thick
0
0 1 2 3 4 5 6 7 8
Collapse spectral acceleration
MCE level spectral acceleration
Figure 5: Collapse fragility curves for archetypical uniformly infilled steel moment frames
Table 4: SSF, ACMR and probability of collapse at MCE for the archetypical uniformly infilled steel moment
frames
Infill type CMR SSF ACMR Probability of collapse at MCE
Bare frame 2.2 1.40 3.10 0.066
AAC, 8-in. thick 2.7 1.25 3.37 0.064
Conventional masonry, 8-in. thick 2.4 1.18 2.87 0.094
Conventional masonry, 12-in. thick 0.9 1.16 1.05 0.480
1.11 Results for archetypical open ground story frames
Results from ATC-63 evaluation of the archetypical open ground story frames are generally similar to those of
the archetypical uniformly infilled frames, although with some differences. In Table 5 are presented the median
collapse spectral intensity at the fundamental period of the ATC-63 steel moment frame obtained from IDA for
the archetypical open ground story frames. From consideration of sidesway collapse alone, the decrease in
collapse spectral intensity of the 12-in. thick conventional masonry open ground story frame compared to that of
the 12-in. thick conventional masonry uniformly infilled frame is particularly evident. This is due to onset of the
weak ground story mechanism. Comparison of Table 2 and Table 5 suggests that the decrease in collapse
spectral intensity due to local shear failures is not as severe for the open ground story frames as for the
uniformly infilled frames. One reason for this is that the open ground story frames have no infill at the ground
story, where story shear is highest. Another reason is that the open ground story frames generally have a longer
fundamental period than the corresponding uniformly infilled frames, probably resulting in lower spectral
acceleration demand and lower base shear.
In Figure 6 is shown the displaced profile of the AAC open ground story frame and the 12-in. thick conventional
masonry open ground story frame at a maximum interstory drift of 15% for the 44 ground motions of the ATC-
63 suite during IDA. Displaced profile of the AAC open ground story frame during IDA is similar to that of the
ATC-63 steel moment frame. This means that even with an open ground story, it is possible to design infills that
will not change the native failure mechanism of the frame.
527
Table 5: Median collapse spectral intensity at the fundamental period of the bare frame for the archetypical open
ground story frames, with and without non-simulated collapse modes
Considering only Including non-simulated

Infill type
sidesway collapse collapse mode
Bare frame 2.14 2.14
AAC, 8-in. thick 2.34 2.34
Conventional masonry, 8- in. thick 2.34 2.34
Conventional masonry, 12-in. thick 2.14 1.56
The displaced profile of the 12-in. thick conventional masonry open ground story frame corroborates the
empirical observation that infilled frames with an open ground story often fail in real earthquakes by formation
of a weak ground story mechanism. Infill strength ratios close to 1.0 seem almost to ensure the formation of
weak ground story mechanisms.
Figure 6: Displaced profiles of AAC open ground story frame and 12-in thick conventional masonry open
ground story frame at a maximum interstory drift of 15% during IDA
1.12 Proposed Design Approach for Steel Moment Frames with Masonry Infills
1.12.1 Seismic response reduction factor, R
Archetypical infilled steel moment frames conforming to uniform and open ground story infill configurations
with infill strength ratio less than 0.35, and evaluated using the ATC-63 methodology with a trial seismic force
reduction factor (R) equal to that of the original bare steel moment frame, meet the performance criteria
prescribed by the ATC-63 methodology. Although none of the archetypical infilled steel moment frames
evaluated in this work using the more rigorous IDA have an infill strength ratio of 0.35, there is a fair
correlation between the infilled frame behavior observed during pushover analysis and during IDA. From results
of pushover analysis of the ATC-63 steel moment frame, an infill strength ratio of about 0.35 is found to mark
the onset of story mechanisms. Therefore, the value of R prescribed for steel moment frames by ASCE7-05 is
also applicable to infilled steel moment frames with infill strength ratio up to 0.35.
528
1.12.2 System overstrength factor, Ω0
It has been shown here that if the infill strength ratio is less than 0.35, it does not change the native failure
mechanism of the bare frame. The force resultants in frame members of such an infilled steel moment frame due
to frame action alone will remain the same as for the bare frame. Therefore, the same over-strength factor
recommended by ASCE7-05 for steel moment frames is proposed for the design of frame members of infilled
steel moment frames with infill strength ratio less than 0.35. However, infills induce additional force resultants
on the frame members, which should be accounted for during the design process. This is addressed shortly.
1.12.3 Displacement amplification factor, Cd
The values recommended for Cd by ASCE7-05 for steel moment frames are based on industry-wide consensus
and are generally adequate. In addition, infills with infill strength ratio less than 0.35 have been shown to
consistently reduce drift of steel moment frames under DBE- and MCE-level ground motions. Therefore, the
same value of Cd prescribed by ASCE7-05 for steel moment frames is proposed for design of infilled steel
moment frames with infill strength ratios less than 0.35.
1.12.4 Design guidelines to consider the presence of infills
A simple procedure is proposed for addressing the increased force resultants on frame members due to the
presence of infills. In using the proposed value of R and Ωo , the preliminary design of the bare steel moment
frame can be completed without considering the infills. Each infill panel should then be modeled as a pair of
equivalent diagonal struts, whose properties can be obtained using the draft MSJC infill provisions (MSJC
2008). The frame members bounding the infill panels must be checked for the additional shears produced by the
strut reactions. The frame columns also must be checked for the additional axial forces produced by seismic
overturning due to the presence of the infill panels. This design approach is applicable to both uniformly infilled
and open ground story configurations.
1.13 Applicability of proposed design approach for different infill materials and other types of frames
The proposed design approach is expected to be applicable for all infill materials. However, the design approach
may be easier to implement for weaker infills, because the stronger the infill, the greater the local shear demand
and increase in column axial forces produced in the bounding frame elements by infill strut forces, and the
greater may be the probable necessary modifications in the design of the original bare frame. A weak material,
in contrast, permits the limiting infill strength ratio of about 0.35 to be achieved using a large number of panels,
each of which produces relatively small increases in design shears and axial forces for the original bare frame.
The proposed design approach is expected to be applicable in principle to infilled reinforced concrete moment
frames, because the fundamental principles involved in behavior, analysis and design of reinforced concrete
moment frames are same as those for steel moment frames.
1) Infills of masonry can be designed to improve the seismic performance of an original bare frame by
reducing the probability of collapse under strong earthquakes, and also by reducing drift (damage) under all
levels of earthquakes.
2) In this process, the original seismic design factors used for the bare frame are retained, and hence the
design of the frame is generally unchanged. Nevertheless, infilling is advantageous because it improves
seismic performance.
3) To achieve this objective, infills should be configured and designed so that in each story, their lateral
strength is less than about 35% of the lateral strength of the bare frame in that story. For calculation of this
infill strength ratio, the strength of the infill should be computed as an upper fractile of the expected
strength.
4) If the infill strength ratio, so computed, is kept below about 35%, then the preliminary design of the frame
can be completed without considering the infill, and the same seismic design factors used for the original
bare frame can safely be retained for the corresponding infilled frame. This conclusion is independent of
529
infill configuration (uniform versus open ground story) and infill material. After completion of preliminary
design of the bare frame, shear capacity of frame members and axial capacity of columns are checked for
additional shear forces produced by strut reactions..
5) When the infill strength ratio is maintained below 35%, ATC-63 evaluation shows that the seismic
behavior of the infilled frame is better than that of the original bare frame.
6) If the infill strength ratio exceeds the above limit of about 35%, the ATC-63 evaluation shows that the
seismic behavior of infilled frames becomes worse than that of the original bare frame. This is consistent
with anecdotal experience in earthquakes. With uniformly infilled frames, increasing values of the infill
strength ratio lead to an increasing probability of local shear failures of the frame elements bounding the
infills (referred to in the ATC-63 methodology as non-simulated collapse modes). With open ground story
frames, the same tendency for local shear failures is observed, and is accompanied (for infill strength ratios
above about 1) by an increasing tendency for ground-level, weak-story mechanisms.
7) This work applies to uniformly infilled frames, and uniformly infilled frames with open ground stories.
Infills can be present in other configurations and with openings, or existing configuration may be altered
due to out-of-plane failure of infills. These should addressed adequately by further investigations.
ACKNOWLEDGEMENTS
The authors gratefully acknowledge the financial and technical assistance provided by the Autoclaved Aerated
Concrete Products Association, Xella AAC, Xella Mexicana and Trustone AAC for the work described in this
paper.
REFERENCES
[1] ASCE, 2005. Minimum Design Loads for Buildings and Other Structures, ASCE Standard No. 7-05,
American Society of Civil Engineers and Structural Engineering Institute, Virginia, USA
[2] ATC-63, 2008. Quantification of Building Seismic Performance Factors - 90% Draft, Report No. ATC-63,
2008, prepared by the Applied Technology Council, Redwood City, California for the Federal Emergency
Management Agency, Washington, D.C.
[3] Dolsek, M., and Fajfar, P., 2001. Soft storey effects in uniformly infilled reinforced concrete frames,
Journal of Earthquake Engineering, 2001, Vol. 5, No. 1, 1-12
[4] Dolsek, M., and Fajfar, P., 2004. Inelastic spectra for infilled reinforced concrete frames, Earthquake
Engineering and Structural Dynamics, Vol. 33, 2004, 1395–1416.
[5] Dolsek, M., and Fajfar, P., 2008. The effect of masonry infills on the seismic response of a four-storey
reinforced concrete frame - a probabilistic assessment, Engineering Structures, 2008, Vol. 30, 1991–2001
[6] Kaushik. H. B., Rai, D. C., and Jain. S. K., 2006. Code Approaches to Seismic Design of Masonry-Infilled
Reinforced Concrete Frames: A State-of-the-Art Review, Earthquake Spectra, Vol. 22 No. 4, November
2006, 961–983
[7] Klingner, R. E. and Bertero, V., 1978. Earthquake resistance of infilled frames, Journal of Structural
Engineering, ASCE, Vol. 104, No. 6, 973-989.
[8] Kodur, V. K. R., Erki, M. A., and Quenneville, J. H. P., 1995. Seismic design and analysis of masonry-
infilled frames, Canadian Journal of Civil Engineering, Vol. 22, 576-587.
[9] Krawinkler, H., and Lignos, D. G., 2007. How to Predict and Reduce the Probability of Collapse of Non-
Ductile Building Structures, Proceedings, International Workshop On Measures for the Prevention of Total
530
Collapse of Existing Low-Rise Structures, November 19-20, Istanbul Technical University, Istanbul,
Turkey.
[10] MSJC, 2008a. Building Code Requirements for Masonry Structures, Masonry Standards Joint Committee,
The Masonry Society, Boulder, Colorado, USA
[11] Panagiotakos T. B. and Fardis, M. N., 1996. Seismic response of infilled RC framed structures,
Proceedings, 11th World Conference on Earthquake Engineering, Acapulco, Mexico.
[12] Ravichandran, S., 2009. Design provisions for Autoclaved Aerated Concrete (AAC) infilled steel moment
frames, PhD Dissertation, The University of Texas at Austin, USA, December 2009.
[13] Vamvatsikos, D. and Cornell, C. A., 2002. Incremental Dynamic Analysis, Earthquake Engineering and
Structural Dynamics, Vol. 31 No.3, 491-514.
531
Paper No.: 092
SIMULATION OF BI-METALLIC INTERFACIAL CRACK USING EFGM AND

XFEM
K. Sharma1, V. Bhasin1, I. V. Singh2, B. K. Mishra2 and R. K. Singh1

1
Bhabha Atomic Research Centre, Trombay, Mumbai, India
2
Department of Mechanical and Industrial Engineering, IITR, Roorkee, U.K., India
Email: kamals@barc.gov.in, hello_kamal@yahoo.com
ABSTRACT :
This paper presents the implementation of element free Galerkin method (EFGM) and extended finite element
method (XFEM) for the stress analysis of structures having cracks at the interface of two dissimilar materials.
The material discontinuity at the interface in EFGM has been modeled using a jump function with a jump
parameter that governs its strength. The jump function enriches the approximation by the addition of special
shape function that contains discontinuities in the derivative. The trial and test functions of the weak form are
constructed using moving least-square interpolants in each material domain. An intrinsic enrichment criterion
with enriched basis has been used to model the crack tip stress fields. In XFEM, finite element approximation is
enriched by additional functions using partition of unity. Level set method is used for tracking the location of
crack. The mixed mode (complex) stress intensity factors for bi-material interface cracks are numerically
evaluated using the modified domain form of interaction integral. The numerical results are obtained for centre
cracks lying at the bi-material interface, and are found to be in good agreement with the reference solutions for
the interfacial crack problems.
KEYWORDS: XFEM, EFGM, Bi-material, Interfacial Crack, SIF.
1. INTRODUCTION
The increasing demand of multifunctional materials (having resistance to corrosion, wear, stiffness, thermal and
chemical resistance) in mechanical, aerospace and biomedical applications has imparted the layered materials, a
coveted place in the world of engineering materials. Layered materials are found in a variety of important
structures such as adhesive joints, composite laminates, and various electronics and optic components. The
interaction between materials results in local load distribution and determines the overall strength and fracture
behavior. The abrupt change in properties at the interface is one source of failure of layered materials, unlike the
behavior of homogeneous [1] materials. Unlike crack problems in homogeneous bodies, the bi-material
interface crack always induces both opening and shearing mode behavior for mode-I loading. This coupling of
stress intensification was first demonstrated by Williams [2] who used an Eigen function expansion approach.
The stress singularity in the vicinity of a crack tip of a bi-material interface crack is oscillatory in nature along
with the presence of 1 / r singularity [2]. Various numerical methods for evaluating the stress intensity factors
of bi-material interface cracks have been developed such as finite element method (FEM) and boundary element
method (BEM). Yau et al. [3] applied the M- integral method with FEM. Matos et al. [4] used the virtual crack
extension method in conjunction with the superposition method and FEM. Miyazaki et al. [5-6] applied the
virtual crack extension method and M- integral method with BEM. Ikeda et al. [7] presented the efficient
numerical procedures in conjunction with FEM for the analysis of an interface crack under thermal stresses.
2. REVIEW OF ELEMENT FREE GALERKIN METHOD
2.1 Methodology
h
In EFGM, a field variable u is approximated by moving least square approximation function u (x) [8],
which is given by
m
h T
u ( x) = ∑ p j ( x) a j ( x) ≡ p ( x) a( x) (1)
j =1
where, p(x) is a vector of basis functions, a(x) are unknown coefficients, and m is the number of terms in
the basis. The unknown coefficients a(x) are obtained by minimizing a weighted least square sum of the
h
difference between local approximation, u ( x) and field function nodal parameters u I . The weighted least
square sum L(x) can be written in the following quadratic form:
n T
L ( x ) = ∑ w( x − x I )[p (x)a(x) − u I ]
2 (2)
I =1
h
where, u I is the nodal parameter associated with node I at x I ; u I are not the nodal values of u (x − x I )
because u h ( x) is an approximant and not an interpolant; w(x − x I ) is the weight function having compact
support associated with a node I , and n is the number of nodes in the domain of influence of the point x ,
w(x − x I ) ≠ 0 . By setting ∂L / ∂a =0, a following set of linear equation is obtained as:
A(x) a( x ) = B ( x ) u (3)
By substituting Eq. (3) in Eq. (1), the approximation function is obtained as:
n
u ( x ) = ∑ ΦI ( x) u I
h
(4)
I =1
2.2 Governing Equations of Bi-Material
The treatment of material discontinuity in the EFGM is demonstrated by considering a linear elastostatic
problem. For simplicity, two distinguishable materials separated by a single interface, Γs as shown in Fig. 1 is
considered. This interface is defined by n , the unit outward normal of Ω − along the material boundary. The
governing equilibrium equation is given by
∇. σ + b = 0 in Ω
(5)
+ −
n+ Ω=Ω +Ω
Ω+ y
−
n
Γs
x
− n−
Ω
− − −
Γ = Γt + Γu
Figure 1. Two-dimensional inhomogeneous body
533
along with associated boundary conditions
σ. nˆ = t on Γt
(6)
u=u on Γu
(7)
where, σ is the Cauchy stress tensor and b is a body force vector, t is the specified traction on a surface, u
is the specified displacement field and n̂ is the unit normal to the domain. A perfect interface has been
assumed, and hence the traction and displacement are assumed to be continuous across the interface Γs .
2.3 Jump Function Approach
In this approach, the discontinuities in derivatives are incorporated by using a jump function in the solution. The
enrichment of EFGM approximations is done by adding special shape functions (jump functions) that contain
discontinuities in the derivative [9,10]. The jump shape functions have compact support which results in banded
matrix equations. Consider a two dimensional model having nd lines at which discontinuities in derivatives
exist, then the approximation becomes
n
d
u h ( x) = u EFGM (x) + ∑ q J ( s ) Ψ J (r )
J =1
(8)
where u EFGM (x) is the standard EFGM approximation, which is given as
n
u ( x) = ∑ Φ I ( x ) u I
h
I =1
(9)
q is the jumps amplitude parameters, Ψ J ( r ) are the jump shape functions, and s provides parameterization
J
of the line of discontinuity. q J is discretized as follows:

qJ ( s) = ∑ N (s)q
I
I
J
I
(10)
where, N I are one dimensional shape functions which need to be C 1 so that they do not introduce any
discontinuities in the derivatives except at the discontinuity line. From Eq. (8), it is clear that the constant and
linear fields will still be reproduced exactly when q J = 0 . The distance to a closest point on the line of
discontinuity is denoted by r , which is taken positive on one side of the discontinuity and negative on the other
side. The jump shape function is constructed from the polynomials with a built-in discontinuity in the
derivatives. Let Ψ J be equal to Ψ J ( rJ ) , where rJ = rJ d mI ; rJ is the distance to the J th point of discontinuity
and d mI is the domain of influence. A cubic spline jump function [10] used in the present work is given by,
 −1 r 3 + 12 rJ 2 − 12 rJ + 16 , rJ ≤ 1 
Ψ J ( rJ ) =  6 J 
0, rJ ≥ 1
(11)
3. REVIEW OF EXTENDED FINITE ELEMENT METHOD
The displacement fields u T = [u x u y ] for a four noded quadrilateral element are interpolated from the nodal
534
displacement values u T = [u x u y ] . The enriched trial and test function [12, 13] with the partition of unity [11]
enrichment can be presented in the general form as:
n  
u ( x ) = ∑ N i ( x ) ui + φ ( x ) a i 
h

i =1  123 
 i∈nr 
(12)
where, i is the set of all nodes in the domain, N i (x ) is element shape function associated with node i
satisfying the partition of unity criterion, a J are the additional unknowns associated with the enrichment
function φ (x ) defined for those elements which contains a discontinuity, and nr is the set of nodes belonging
to those elements, which contains a discontinuity.
3.1 XFEM Approximation for Crack
For 2D crack modeling, the enriched displacement trial and test approximation is written in general form as [12,
13]:
 
n  4 
u h (x) = ∑ N i (x)  ui + Η (x) a i + ∑ βα (x) bαi 
1
424 3 α =1
i =1  i∈nr 14243 
 i∈n A 
(13)
where, ui is a nodal displacement vector associated with the continuous part of the finite element solution,
a i are the nodal enriched degree of freedom vector associated with discontinuous Heaviside function, H (x)
defined for those elements, which are completely cut by the crack. Heaviside function, H (x) is used to model
the discontinuity in the displacement due to the presence of a crack, and can be expressed as,
 1 if χ ( x) ≥ 0
H ( x) = 
 −1 otherwise
(14)
where, χ ( x ) is the level set function. bαi are the nodal enriched degree of freedom vector associated with
crack tip enrichment( βα (x) ). βα (x) are the asymptotic crack tip enrichment functions obtained from the
theory of linear elastic fracture mechanics. In Eq. (3), n is the set of all nodes in the mesh; n r is the set of
nodes belonging to those elements, which are completely cut by the crack and n A is the set of nodes belonging
to those elements, which are partially cut by the crack. To model bimaterial interface cracks within the XFEM,
crack tip enrichment functions are used [14].
535
 θ θ 
 r cos ( ε log r ) e −εθ sin , r cos ( ε log r ) e − εθ cos , 
2 2
 
 θ θ 
r cos ( ε log r ) e sin , r cos ( ε log r ) e cos ,
εθ εθ
 2 2 
 
θ θ
 r cos ( ε log r ) eεθ sin sin θ , r cos ( ε log r ) eεθ cos cosθ , 
{F _ l ( r ,θ )}i =1 =  2 2 
12
θ θ 
 r sin ( ε log r ) e −εθ sin , r sin ( ε log r ) e −εθ cos , 
 2 2 
 θ θ 
 r sin ( ε log r ) eεθ sin , r sin ( ε log r ) eεθ cos , 
 2 2 
 θ θ 
 r sin ( ε log r ) e sin sin θ , r sin ( ε log r ) e cos cos θ 
εθ εθ
2 2 
(15)
4. NUMERICAL SIMULATION
A rectangular bi-material plate with an interface crack at the center has also been considered as shown in Fig. 2a.
The width (W ) and height ( H ) of the plate are taken as 100 mm and 200 mm respectively. The material
interface is kept horizontal at a distance of H 2 from the bottom of plate. A tensile load of 9.8 MPa is applied
in a direction normal to the interface. Two different cases of crack length, 2a i.e. 40 and 80 mm are considered.
Young’s modulus of the lower material ( E1 ) is kept constant at 207 GPa.
For both material 1 and 2, the values of Poisson’s ratio are taken as 0.3. The results are obtained for different
ratios of Young’s modulus. The results are normalized as K i σ πa (i=1, 2) so as to obtain non-dimensional
values corresponding to both mode-I and mode-II stress intensity factor. The results are compared with the
available results in the literature.
Due to symmetry, discretization is performed for only half the plate in both the methods. XFEM discretization
of half model involves a mesh consisting of 160X160 elements in the direction of width and length respectively.
For EFGM discretization, a cubic spline weight function was used for generating the EFGM shape functions
along with a domain scaling parameter ( d max = 1.5) . The geometry has been discretized using 25 uniformly
distributed nodes along x -direction and 50 nodes along y -direction i.e. total 1250 (25x50) nodes and 1176
(24x49) background cells.
For a particular crack length and E2 / E1 = 100 , the contours of σ yy and ε yy were plotted in order to have a
clear visualization of stress and strain fields along with establishing the modeling capability of the method.
Contour plots shown in Fig. 2 shows that stress ( σ yy ) are continuous (Fig. 2b) while a significant jump is
present in the magnitude of strain ( ε yy ) at the interface (Fig. 2c). This jump in the magnitude of strain is due to
the sudden change of material property at the interface.
Figure 3 shows the variation of normalized stress intensity factors (SIF) with the variation of E2 / E1 for a
crack length of 40 mm. The values of stress intensity factors have been evaluated at the right tip of the crack.
The analysis shows that the normalized values of K I shows a decreasing trend with the increase in E2 / E1 .
Moreover, the results obtained by XFEM & EFGM are quite close to the reference solutions. Another simulation
is performed for a crack length of 80 mm, and the normalized SIF are plotted with the variation of E2 / E1 in
Fig. 4. It shows the SIF has decreasing trend with the increase in the E2 / E1 . The evaluated values of
536
normalized SIF using both methods are in good agreement with the reference solution available in literature. For
E2 / E1 = 100 , the values of K I are found to be maximum error of 3.5% and 4.9% respectively for a/w = 0.4
and 0.8.
From this simulations of bi-metallic interfacial cracks it was observed that the XFEM and EFGM in capable in
modeling the interfacial cracks in an efficient way for a wide range of crack length and material incompatability.
σo
E 2, ν 2
2a
H
E1 , ν 1 H /2
σo
(a) (b) (c)
Figure 2. Stress contour over the domain
1.3
Ref1
1.2 Ref2
XFEM
Normalised Stress Intensity factor
EFGM
1.1
0.9
0.8
0.7
0.6
0.5
0 1 2
10 10 10
Ratio of Young Modulus
Figure 3. SIF variation of a/w=0.4 for different ration of

E2 / E1
537
2.2
Ref1
Ref2
2 XFEM
Normalised Stress Intensity factor

EFGM
1.8
1.6
1.4
1.2
1
0 1 2
10 10 10
Ratio of Young Modulus
Figure 4. SIF variation of a/w=0.8 for different ration of E2 / E1
5. CONCLUSIONS
In the present work, element free Galerkin method and extended finite element method have been implemented
for the analysis of interfacial cracks lying between two dissimilar materials. The material discontinuity at the
interface has been modeled by a jump function, which enriches the approximation by addition of special shape
function that contains discontinuities in the derivative. The crack has been modeled by an intrinsic enrichment
criterion. In XFEM, finite element approximation is enriched by additional functions using partition of unity.
Level set method is used for tracking the location of crack. The mixed mode stress intensity factors for
bi-material interface cracks are numerically evaluated using the modified domain form of interaction integral
approach. . The numerical results were obtained for two different cases of centre crack length, 2a i.e. 40 and 80
mm, lying at the interface of bi-material using both methods. The numerical results obtained by XFEM and
EFGM were found in good agreement with those available in literature for interfacial centre crack problems.
The evaluated values of normalized SIF, for E2 / E1 = 100 , the values of K I are found to be maximum error of
3.5% and 4.9% respectively for a/w = 0.4 and 0.8. The simplicity and accuracy of this implementation shows
that it can be further applied to study and analyze the complex interfacial failures of multi-layered dissimilar
materials.
REFERENCES
[1] Hutchinson, J.W., Suo, Z., (1991). Mixed mode cracking in layered materials, Advance Applied Mechanics,
29, pp. 63-191.
[2] Williams, M.L., (1959). The stress around a fault or crack in dissimilar media, Bulletin of the Seismology
Society of America, 49, pp. 199-204.
[3] Yau, J.F., Wang, S.S., (1984). An analysis of interface cracks between dissimilar isotropic materials using
conservation integral in elasticity, Engineering Fracture Mechanics, 20, pp.423-432.
[4] Matos, P.P.L., Meeking, R.M., Charalambides, P.G. and Drory, M.D., (1989). A method for calculating
stress intensities in bimetal fracture, International Journal of Fracture, 40, pp. 235-254.
[5] Miyakazi, N., Ikeda, T., Soda T., Munakata, T., (1993). Stress intensity factor analysis of interface crack
using boundary element method (application of virtual crack extension method), JSME International Journal:
Series A, 36, pp.36-42.
538
[6] Miyakazi, N., Ikeda, T., Soda T., Munakata, T., (1993). Stress intensity factor analysis of interface crack
using boundary element method (application of contour integral method)”, Engineering Fracture Mechanics, 45,
pp. 599-610.
[7] Ikeda, T., Sun, C.T., (2001). Stress intensity factor analysis for an interface crack between dissimilar
isotropic materials under thermal stress, International Journal of Fracture, 111, pp. 229-249.
[8] Nguyen, P., Rabczuk, T., Bordas, S., and Duflot, M., (2008). Meshless methods: A review and computer
implementation aspects, Mathematics and Computers in Simulation, 79, pp. 763-813.
[9] Batra, R.C., Porfiri, M., and Spinello, D., (2004). Treatment of material discontinuity in two meshless local
Petrov–Galerkin (MLPG) formulations of axisymmetric transient heat conduction, International Journal for
Numerical Methods in Engineering, 61, pp. 2461–2479.
[10] Krongauz, Y., Belytschko, T., (1998). EFG Approximation with discontinuous derivatives, International
Journal for Numerical Methods in Engineering, 41, pp.1215-1233.
[11] Melenk J., Babuska I., (1996). The partition of unity finite element method: Basic theory and applications.
Computer Methods in Applied Mechanics and Engineering, 139, pp. 289-314.
[12] Moes N., Dolbow J., Belytschko T., (1999). A finite element method for crack growth without remeshing,
International Journal for Numerical Methods in Engineering, 46, pp. 131-150.
[13] Mohammadi S., (2008). Extended Finite Element Method for Fracture Analysis of Structures, Blackwell
Publishing Limited.
[14] Sukumar N., Huang Z. Y., Prevost J. H. and Suo Z., (2004). Partition of unity enrichment for bimaterial
interface cracks, International Journal for Numerical Methods in Engineering, 59, pp. 1075-1102.
539
Paper No.: 093
EVALUATION OF STRESS CONCENTRATION FOR A SIMPLY SUPPORTED

LAMINATED COMPOSITE PLATE WITH A CENTER CIRCULAR HOLE
UNDER UNIFORM TRANSVERSE LOADING
Moon Banerjee1, Dr. N. K. Jain2 and Dr. S. Sanyal3
1, 2
Dept. of Applied Mechanics, National Institute of Technology, Raipur, India
3
Dept. of Mechanical Engg., National Institute of Technology, Raipur, India
Email: moonbanerjee@gmail.com, nkjmanit@rediffmail.com, shubhashissanyal@rediffmail.com
ABSTRACT :
The aim of this investigation is to evaluate the stress concentration factor (SCF) for a simply supported
rectangular plate with a center circular hole under uniform transverse loading over the whole plate. This
research aims to study the layer wise shear stress distribution in the laminate for simply supported boundary
conditions. Laminated composite plate is stacked with symmetric and anti-symmetric cross ply configurations
consisting of four laminas or layers. The analysis has been done for different D/A ratios (where D is the
diameter of the hole and A is the width of the plate) varying from 0.1 to 0.5 and five different composite
materials. Composite materials selected elaborate the dependency of SCF with varying elastic constants. The
analytical treatment of such type of problem is very difficult; hence the finite element method is adopted for
whole analysis. Three dimensional finite element formulation and analysis is done on ANSYS package. The
results are presented in graphical form and discussed.
KEYWORDS: Composites, Finite element method, Stress concentration factor, Transverse loading,
Deflection, Elastic constants.
1. INTRODUCTION
Laminated Composite plates with central circular hole were finding a wide range of applications in structural
design of aerospace, marine, automobile, mechanical engineering especially for light weight structures that have
stringent stiffness and strength requirements. Any abrupt change in geometry of plate under loading give rise to
stress concentration, as a result stress distribution is not uniform throughout the cross section. Composite plates
fail due to the failure of individual component (fibre or matrix) or their interface. It is well known that the
presence of a notch in a stressed member creates highly localized stresses at the root of the notch. Various
researchers analyzed different cases of stress concentration in plate with circular holes.
Hochard et al [1] evaluated behavior up to rupture of woven ply laminate structures under static loading
conditions. A first ply failure model was developed based on continuum damage approach was developed for
balanced woven ply laminates for plates with notches and saw cuts. This study concerns only structures in
tension. Ghezzo et al. [2] performed a numerical and experimental analysis of the interaction between two
notches in carbon fibre laminates. The numerical analysis of the stress distribution in-plane stress assumption
and within the fibrous plate theory framework has been conducted on two symmetric laminates. Ozben et al. [3]
compiled FEM analysis of laminated composite plate with rectangular hole and various elastic modulus under
transverse loads. Ozen et al. [4] presented the failure loads of mechanical fastened pinned and bolted composite
joints with two serial holes. Tsai–Wu failure criterion was used to predict first failure loads by finite element
analysis for the geometrical parameters. Kumar et al. [5] has studied the post buckling strengths of composite
laminate with various shaped cut-outs under in plane shear. Rao et al. [6] evaluated the stress around square and
rectangular cut-outs in symmetric laminates. It has been analyzed that the maximum stress and its location is
mainly influenced by the type of loading. Jain et al. [7] have done finite element analysis for stress
concentration and deflection in isotropic, orthotropic and laminated composite plates with central circular hole
under transverse static loading. Banerjee et al. [8] performed three dimensional parametric analyses on effect of
fibre orientation for stress concentration factor in fibrous composite cantilever plate with central circular hole
under transverse loading.
The present work aims to study the effect of boundary conditions (simply supported at all edges fixed) on stress
concentration factor (SCF) in laminated composite plates with central circular hole subjected to transverse static
loading over the whole plate. The effect of D/A ratio where A is the plate width and D is the hole diameter on
SCF for shear stress (τxy) is investigated by using three dimensional finite element analysis. Results are obtained
for five different materials to find out the variation of SCF on different elastic constants also.
2. DISCRIPTION OF PROBLEM
The model of laminated composite plate of dimension 0.2 m X 0.1 m and 0.01 m thickness with a circular hole
of diameter D under uniformly distributed loading of P (N) in transverse direction is taken for analysis. The
laminated plate with circular hole are stacked in both symmetric [00/900/900/00] and anti-symmetric
[900/00/900/00] cross-ply configuration as shown in figure.1 and 2. The composite laminated plate consists of
four orthotropic layers of constant thickness 0.0025 m for simply supported boundary conditions.
z
D A= 0.1 m
0.2 m
x
y 0.02 N
00 4th Layer Top Layer

900 3rd Layer
0.01 m
900 2nd Layer
0.0025 m 00 1st Layer Bottom Layer
x
Figure 1 Symmetric crossply laminated plate with central hole under transverse loading
541
z
D A= 0.1 m
0.2 m
x
y 0.02 N
00 4th Layer Top Layer

900 3rd Layer
0.01 m
00 2nd Layer
0.0025 m 1st Layer Bottom Layer
x
Figure 2 Anti-symmetric crossply laminated plate with central hole under transverse loading
Table 1 Material properties of composite materials [9].
Material E-glass- Boron/ Graphite/Ep Boron/ Woven-glass

Epoxy Epoxy oxy Aluminium / Epoxy
Properties
Ex 39GPa 201GPa 294GPa 235GPa 29.7GPa
Ey 8.6GPa 21.7GPa 6.4GPa 137GPa 29.7GPa
Ez 8.6GPa 21.7GPa 6.4GPa 137GPa 29.7GPa
Gxy 3.8GPa 5.4GPa 4.9GPa 47GPa 5.3GPa
Gyz 3.8GPa 5.4GPa 4.9GPa 47GPa 5.3GPa
Gzx 3.8GPa 5.4GPa 4.9GPa 47GPa 5.3GPa
µxy 0.28 0.17 0.23 0.30 0.17
µyz 0.28 0.17 0.23 0.30 0.17
µzx 0.28 0.17 0.23 0.30 0.17
Where E, G, µ are representing modulus of elasticity, modulus of rigidity and poisson’s ratio
respectively.
Results evaluated for fixed T/A ratio of 0.10, and for three different D/A ratio of 0.1, 0.2 and 0.5 respectively. A
transverse loading of 0.02 N in the form of uniform distributed loading is applied over the whole plate. Five
542
different composite materials with their respective material properties selected for complete analysis is shown in
table 1.
3. ANALYSIS
Finite element method was chosen for analysis of the model. The model was meshed using a 3-D solid element,
Solid 191 with three degrees of freedom and 60 nodes per element in ANSYS. Typical mesh of the plate using the
above element has been shown in Figure. 2. Mapped meshing is used so that more elements employed near the
hole boundary. Element length is selected as 5mm after running the convergence tests. The example of the
discretized three dimensional finite element model, used in study is shown in Figure 2. Results were then
displayed by using post processor of ANSYS program.
Figure 3 Element used and generated mesh of the plate.
4. RESULT AND DISCUSSION
The present study demonstrates the shear stresses for the symmetric and anti-symmetric laminated plates with
central circular under uniform static transverse loading. Results are evaluated for laminated composite plate
containing four layers of equal thickness each of 0.0025 m. Effect of simply supported boundary conditions on
stress concentration factor for different materials is shown in below figures. Layer wise distribution for SCF (τxy)
in both symmetric and anti-symmetric configuration is compared and shown in below figures. Also results are
obtained for the bottom portion of each of four layers arranged in both symmetric and anti-symmetric
configuration.
Variation of SCF for (τxy) versus D/A for a symmetric cross-ply laminated plate with a central circular hole
loaded transversely over the whole plate for five different composite materials is shown in figure 4. Following
observations can be made:
For all the five materials, SCF in the layers are maximum for fourth layer and minimum in first layer. Maximum
stress concentration within the four layers is in following increasing order for all the materials: first, second, third
and fourth layer. SCF (τxy) continuously decreases with corresponding increase in D/A ratio from 0.1 to 0.5 for all
layers. For e-glass/epoxy, SCF for first layer decreases from 2.57 to 2.20, second layer decreases from 3.8 to 2.5,
third layer decreases from 4.84 to 3.8 and fourth layer decreases from 4.92 to 4.0 on corresponding increase in
D/A ratio from 0.1 to 0.5 respectively. Maximum SCF value among the four layers is 4.92 obtained for fourth
layer at D/A = 0.1 ratio. For boron/epoxy, SCF for first layer decreases from 4.7 to 4.3, second layer decreases
from 10.2 to 7.5, third layer decreases from 12.2 to 7.98 and fourth layer decreases from 15.30 to 8.0 on
corresponding increase in D/A ratio from 0.1 to 0.5 respectively. Maximum SCF value among the four layers is
15.7 obtained for fourth layer at D/A = 0.1 ratio. For graphite/epoxy, SCF for first layer decreases from 18.28 to
17.9, second layer decreases from 20.02 to 19.0, third layer decreases from 26.62 to 22.43 and fourth layer
543
decreases from 60.0 to 48.77 on corresponding increase in D/A ratio from 0.1 to 0.5 respectively. Maximum
SCF value among the four layers is 60.0 obtained for fourth layer at D/A = 0.1 ratio. For boron/aluminum, SCF for
first layer decreases from 1.9 to 1.35, second layer decreases from 2.5 to 1.8, third layer decreases from 2.7 to
1.84 and fourth layer decreases from 3.5 to 2.27 on corresponding increase in D/A ratio from 0.1 to 0.5
respectively. Maximum SCF value among the four layers is 3.5 obtained for fourth layer at D/A = 0.1 ratio. For
woven-glass/epoxy, SCF for first layer decreases from 2.0 to 1.65, second layer decreases from 2.5 to 2.12, third
layer decreases from 2.9 to 1.42 and fourth layer decreases from 3.5 to 2.97 on corresponding increase in D/A
ratio from 0.1 to 0.5 respectively. Maximum SCF value among the four layers is 3.5 obtained for fourth layer at
D/A = 0.1 ratio.
Variation of SCF for (τxy) versus D/A for an anti-symmetric cross-ply laminated plate with a central circular hole
loaded transversely over the whole plate for five different matetrials is shown in figure 5. Following
observations can be made:
For all materials considered, stresses induced in the layers are maximum for third layer and minimum in first layer.
Maximum stress concentration within the four layers is in following increasing order: first, fourth, second and
third layer. SCF (σx) continuously decreases with corresponding increase in D/A ratio from 0.1 to 0.5 for all layers.
For e-glass/epoxy, SCF actor for first layer decreases from 2.34 to 2.0, second layer decreases from 6.0 to 5.9,
third layer decreases from 7.0 to 6.8 and fourth layer decreases from 4.13 to 4.0 on corresponding increase in D/A
ratio from 0.1 to 0.5 respectively. Maximum SCF value among the four layers is 7.0 obtained for third layer at
D/A = 0.1 ratio. For boron/epoxy, SCF for first layer decreases from 5.72 to 5.0, second layer decreases from14.04
to 14.01, third layer decreases from 16.1 to 15.2 and fourth layer decreases from 10.0 to 7.2 on corresponding
increase in D/A ratio from 0.1 to 0.5 respectively. Maximum SCF value among the four layers is 16.1 obtained
for third layer at D/A = 0.1 ratio. For graphite/epoxy, SCF for first layer decreases from 4.69 to 4.0, second layer
decreases from 33.68 to 35.62, third layer decreases from 56.10 to 57.03 and fourth layer decreases from 23.31 to
22.0 on corresponding increase in D/A ratio from 0.1 to 0.5 respectively. Maximum SCF value among the four
layers is 57.03 obtained for third layer at D/A = 0.1 ratio. For boron/aluminum, SCF for first layer decreases from
1.57 to 1.23, second layer decreases from 2.54 to 2.07, third layer decreases from 3.01 to 2.34 and fourth layer
decreases from 2.26 to 1.85 on corresponding increase in D/A ratio from 0.1 to 0.5 respectively. Maximum SCF
value among the four layers is 3.01 obtained for third layer at D/A = 0.1 ratio. For woven-glass/epoxy, SCF for
first layer decreases from 1.73 to 1.65, second layer decreases from 2.54 to 2.42, third layer decreases from 3.12
to 2.97 and fourth layer decreases from 2.22 to 1.12 on corresponding increase in D/A ratio from 0.1 to 0.5
respectively. Maximum SCF value among the four layers is 3.12 obtained for third layer at D/A = 0.1 ratio.
544
SCF
SCF
SCF
Figure 4 Variation of SCF (τxy) versus D/A for a symmetric cross-ply laminated plate for all four layers
545
15 24
E-glass/Epoxy Boron/Epoxy
xy
10 16
5 8
0 0
90 4
Graphite/Epoxy Boron/Aluminum
60 3
30 2
0 1
0.1 0.2 0.3 0.4 0.5 0.1 0.2 0.3 0.4 0.5
D/A D/A
4
Woven-glass/Epoxy
1
0.1 0.2 0.3 0.4 0.5
D/A
Figure 5 Variation of SCF (τxy) versus D/A for a anti-symmetric cross-ply laminated plate for all four layers
546
5. CONCLUSION
Detailed investigations on the stress concentration factor for shear stress in both symmetric and anti-symmetric
cross-ply laminate subjected to transverse loading is conducted for five different composite materials. Following
conclusions can be with drawn:-
• The influence of geometric parameters shows a substantial role for shear stresses in both configuration of
laminate considered.
• SCF (τxy) is seen maximum on the hole boundary along the width direction of the plate.
• Higher Ex/Ey & Ex/Gxy ratios, prominently effect SCF. Anti-symmetric laminates shows maximum
values for all the materials as compared to symmetric configuration of laminates.
Additionally this can also be concluded that the results obtained are in line with other similar works.
REFERENCES
[1] Hochard, Ch.; Miot, S., Lahellec, N., Mazerolle, F., Herman, M. and Charles, J.P. (2009). Behavior up to
rupture of woven ply laminate structures under static loading conditions. Composites Part A, 40, 1017-1023.
[2] Ghezzo, F.; Giannini, G., Cesari, F. and Galigiana, G. (2008). Numerical and experimental analysis of the
interaction between two notches in carbon fibre laminates. Composites Science and Technology, 68, 1057–
1072.
[3] Ozben, T. and Arslan, N. (2010). FEM analysis of laminated composite plate with rectangular hole and
various elastic modulus under transverse loads. Applied Mathematical Modeling, 34, 1746–1762.
[4] Ozen, M. and Sayman, O. (2010). Failure loads of mechanical fastened pinned and bolted composite joints
with two serial holes. Composites: Part B. 42 (2), 264-274.
[5] Kumar, D. and Singh, S.B. (2010). Post buckling strengths of composite laminate with various shaped
cutouts under in-plane shear. Composite Structure, 92, 2966–2978.
[6] Rao, D.K.N.; Babu, M.R., Reddy, K.R.N. and Sunil, D. (2010). Stress around square and rectangular cutouts
in symmetric laminates. Composite Structures, 92, 2845–2859.
[7] Jain, N.K. and Mittal, N.D. (2008). Finite element analysis for stress concentration and deflection in
isotropic, orthotropic and laminated composite plates with central circular hole under transverse static
loading. Material science and Engineering A. 498, 115-124.
[8] Banerjee, M.; Jain, N.K. and Sanyal, S. (2012). Three dimensional parametric analyses on effect of fibre
orientation for stress concentration factor in fibrous composite cantilever plate with central circular hole
under transverse loading. IIUM Engineering Journal, 13, 131-144.
[9] Daniel, I.M. and Ishai, O. (1994). Engineering Mechanics of Composite Materials, Oxford University Press,
New York.
547
Paper No.: 096
EFFECT OF TREATMENT METHODS ON THE STRENGTH

CHARACTERISTICS OF RECYCLED AGGREGATE CONCRETE
Revathi, P1*, Ramesh, R. A2 and Lavanya, K2
1
Assistant Professor, Dept. of Civil Engineering, Pondicherry Engineering College, Puducherry, India
2
PG Student, Department of Civil Engineering, Pondicherry Engineering College, Puducherry
605 014. India
*Email: revathi@pec.edu
ABSTRACT :
With a rising tide of adoption of recycled aggregate for construction, investigation on ways to improve its
quality has been overwhelming. The major reason that affects the quality of recycled aggregate is the large
amount of cement mortar, which remains on the surface of the recycled aggregate. This attached mortar results
in higher porosity, higher water absorption rates and thus a weaker interfacial zone between new cement mortar
and aggregates, weakening the strength and mechanical performance of concrete made from recycled aggregate.
This paper attempts to compare the effect of chemical as well as mechanical treatment approaches in reducing
the mortar attached to aggregate. Six series of concrete mixtures are prepared using natural aggregate, recycled
aggregate, recycled aggregate treated with HCl and H2SO4, recycled aggregate obtained after scrubbing
treatment and heating & scrubbing treatment. The physical and mechanical properties of these aggregates, along
with their strength and static modulus of elasticity are determined. The results show that the treatment with
H2SO4, and heating & scrubbing yield, aggregate with reduced water absorption and other desired properties of
natural aggregate. The concrete made out of these treated aggregates are able to achieve strength characteristics,
on par with the natural aggregate concrete.
KEYWORDS: Recycled aggregate, Chemical treatment, Mechanical treatment, Attached mortar, Recycled
aggregate concrete
1. INTRODUCTION
Crushed concrete is available now-a-days in large quantities, which results from the demolition of old structures
and waste concrete from new structures. The current annual rate of generation of construction waste is 145
million tonnes worldwide. The area required for landfilling this amount of waste is equivalent to the
accumulation of waste. Therefore, recycling construction waste is vital, both in order to reduce the amount of
open land needed for landfilling and to preserve environment through resource conservation. Also from the view
point of sustainable and green building technologies, the use of recycled aggregate (RA) in new concrete
production has increased globally. Many attempts to develop structural concrete with RA are reported in
literature. From these studies, it is worth noting that a certain amount of mortar from the parent concrete remains
attached to the stone particles in RA. This attached/ adhered mortar forms a weak porous and crack layer, which
influences the strength and performance of recycled aggregate concrete (RAC) (Prokopski and Halbiniak,
(2000), Ollivier et al, (1995) and Tam et al, (2005). This is considered to be one of the most significant
differences between RAC and natural aggregate concrete (NAC).
Parekh and Modhera (2011), Berndt (2009), Gonzalez and Martinez (2008), and Yong and Teo (2009) have
studied the fresh properties of RAC and revealed that RA drastically lowers the workability of RAC. It may be
attributed to higher water absorption of RA when compared to natural aggregate (NA). However, these
researchers have proved that the desired workability can be achieved by using suitable mineral and chemical
admixtures. Also Yong and Teo (2009) have recommended the use of recycled aggregate in saturated surface
dry condition to enhance the workability. Fong and Jaime (2002) also reported in their study to pre-wet recycled
aggregate in order to maintain desired workability as that of NAC. Limbachiya et al. (2000) concluded that “It is
suggested not to utilize recycled aggregate for high concrete strength applications due to long-term durability
problems”.
Although reusing RA reduces the consumption of limited resources and thereby saves costs, it also has
shortcomings such as weaker interfacial behaviour between aggregate and cement paste, higher portions of
cement mortar attached and lower quality (Aitcin and Neville, 1993, Alexander, 1996, Bentz and Garboczi,
1991, Buch et al., 2000, Kwan et al., 1999, Mehta and Aitcin, 1990). This limited application of RA is
attributable to their poor quantities. On the other hand, the rising tide of adoption of RA for construction
demands investigation on ways to improve the quality of RA.
1.1 Treatment methods for RA
In order to improve the quality of RA, several techniques have been developed in literature. However, these
methods can be broadly categorized into two. The first category, removes the loose mortar particle on the
surface and the second category modifies the aggregate surface. The loose mortar particle adhered to the surface
can be removed by ultrasonic cleaning method (Katz, 2004), ball milling or by heating at first and then rubbing
(Tsujino et al., 2008). Presoaking RA in acidic environment also removes the adhered mortar from RA.
Recently Tam et al. (2007) tried presoaking approach to remove the attached mortar and obtained RA with
improved qualities.
Modification on the aggregate surface (through coating binder evenly on its surface) increases the bonding force
between RA and new mortar. Pozzolanic materials such as fly-ash and silica fume are used as surface coating
binders, expecting to enhance chemical and physical bonding due to pozzolonic reaction and mechanical action.
Li et al. (2009) found that coating with silica fume and fly ash was most effective for strength growth of RAC
and contributed the improvement to a maximum packing density due to wider distribution of double mineral
additives. Studies by Poon et al., (2004) Kou et al., (2007), Ryu (2002) and Otsuki et al., (2005) clearly
indicated the weaker inter-transitional zone (ITZ) between new cement paste and RA. Therefore, modification
of the microstructure in the ITZ has been one of the great concerns to improve properties of the RAC. By using
a kind of two stage mixing approach (Tam et al., 2005) and triple mixing approach (Kong et al., 2006) ITZ and
thus the properties of RAC can be improved. Kou and Poon (2010) evaluated the effects of poly vinyl alcohol
(PVA) impregnation on the development of strength and durability properties of RAC and showed that there
was not only an improvement in the mechanical properties of the PVA impregnated RAC, but also the shrinkage
decreased, while the resistance to chloride-ion penetration increased.
On thoroughly going through the various literatures pertaining to treatment techniques for RA, it is found
that there is no comparative study reported so far, on the suitability of a particular method of treatment. Hence,
in the present study, it is proposed to experimentally investigate on the effect of treatment method on the
strength and performance of concrete prepared with RA.
2. EXPERIMENTAL INVESTIGATIONS
2.1 Materials
The entire experimental study was conducted at the concrete laboratory of Pondicherry Engineering College,
Puducherry, India. Ordinary Portland cement of 43 grade confirming to IS: 8112 – 1989 was used throughout
the present work. As the present study aims, in predicting the effect of treatment method on the performance of
RAC, the conventional fine aggregate (locally available river sand) is only used as fine aggregate. Both NA
and RA were used as the coarse aggregate in the concrete mixtures. Crushed granite was used as NA and
aggregate obtained from the laboratory crushed specimens was used as RA. The nominal size of the natural and
recycled coarse aggregates was 20 mm and their particle size distributions determined as per IS 383-1970. The
mechanical properties of NA and RA were determined in accordance with IS 2386-1963 and presented in Table
1. RA was subjected to treatment process as explained Section 2.2. The physical properties of RA were also
549
determined after every treatment process, in order to assess the degree of improvement in the quality of RA
through these treatment methods. Potable water available in the college campus was used for the entire study.
2.2 Treatments to Improve the Recycled Aggregate
In the present study, the following four treatment techniques are adopted in order to improve the quality of RA
and taken for comparison:
2.2.1 Chemical treatment:
The recycled aggregates were pre-soaked in an acidic environment at room temperature for twenty-four hours
and then washed with water to remove the acidic solvents afterward. Two acidic solvents are adopted, namely
hydrochloric acid (HCl) and sulfuric acid (H2SO4) with concentration of 0.1 mole. The RA obtained after HCl
and H2SO4 treatment are abbreviated as RAHcl and RAH2SO4 respectively.
2.2.2 Mechanical treatment:
The first technique is called mechanical scrubbing, in which concrete rubbles are scrubbed by one another to
remove adhering cement paste. In the present study, recycled aggregates were subjected to rubbing treatment
with the Los Angeles abrasion machine to remove mortar. The number of revolutions and the weight of charges
to be used in Los Angeles abrasion machine are arrived based on the abrasion results of natural coarse aggregate
and recycled coarse aggregate. From Table 1 it is seen that the abrasion value of RA is 45 % and that of natural
aggregate is 29%. It can be interpreted that the contribution of adhered mortar in RA, in enhancing the abrasion
value is only 15 % approximately (45% -29%) and the rest is due to the inherent property of the coarse
aggregate itself. After some trials, it was decided to rub the RA in Los Angeles abrasion machine for 5 min with
standard charges in order to remove the attached mortar only. The second technology is called
heating-scrubbing in which concrete rubbles are charged in a hot air oven for 24 hours at a temperature of
around 300°C. Heating RA in hot air makes the cement paste brittle and weak. It is then cooled and then
scrubbed in Los Angeles abrasion machine for 5 minutes. RA obtained after scrubbing and heating & scrubbing
techniques are abbreviated as RAS and RAHS respectively.
Table 1 Properties of Coarse Aggregate
Test Results
S.No Properties NA RA RAHCl RAH2SO4 RAS RAHS
1 Specific gravity 2.79 2.38 2.44 2.57 2.49 2.70
2 Water absorption (%) 0.3 1.56 0.92 0.81 0.93 0.78
3 Bulk density (kg/m3) 1508 1239 1438 1453 1462 1483
4 Crushing Value (%) 27 36 33 33 31 30
5 Impact value (%) 23 31 29 27 29 26
6 Abrasion value (%) 29 45 30 28 30 29
2.3 Mix Proportioning and Specimen Preparation
As, there is no standard method, available for proportioning of RA concrete mixes, Indian standard method (IS
10262, 2009) applicable for natural aggregate is adopted. Mix proportioning was done for M 30 grade concrete
with target cube strength of 38 MPa at 28 days. The mix proportion was arrived at as 1: 2.18: 2.82, with a
cement content of 380 kg/m3 adopting 0.45 as water-cement ratio. A concrete mix was prepared in the above
proportion using conventional ingredients such as natural coarse aggregate and natural fine aggregate for
reference and designated as natural aggregate concrete (NAC). Another concrete mix was prepared in the same
proportion replacing NA with RA and designated as RAC. Concrete mixes were also prepared with RAHcl,
RAH2SO4, RAS and RAHS and are designated as RACHCl, RACH2SO4, RACS and RACHS respectively. Hence, totally
550
six series of concrete mixes were prepared and 18 numbers of 150 mm size cubes and three numbers of 150 mm
diameter, 300 mm long cylinders were cast, in each of the mix series. These test specimens were cured in water
under laboratory conditions till the age of testing. Compressive strength of cube specimen was determined at 7,
28, 56 and 90days of age. Static modulus of elasticity, water absorption and drying tests were conducted at 28
days of age.
3. TESTING OF SPECIMENS
3.1 Compressive Strength and Static Modulus of Elasticity
The compressive strength and static modulus of elasticity were determined as per IS 516- 1959, using a
compression machine with a loading capacity of 3,000 kN. The compressive strength was determined with cube
specimens and static modulus of elasticity was determined with cylinder specimens. The results of the
compressive strength test are given in Table 2 and the results of static modulus of elasticity are shown in Fig. 1
for various concrete mixes.
Table 2 Compressive Strength of Concrete Specimens
Slump Compressive strength (MPa)

S.No Mix series
(mm) 7 days 28 days 56 days 90 days
1 NAC 110 25.2 42.4 45.4 48.4
2 RAC 65 20.2 34.6 37.5 40.6
3 RACHCl 80 22.8 38.9 41.3 44.9
4 RACH2SO4 90 24.0 40.3 42.6 46.9
3 RACS 90 22.5 39.9 44.9 46.1
4 RACHS 95 24.9 41.8 46.1 47.3
40
Static Modulus of Elasticity
35
30
25
20
(GPA)
15
10
5
0
NAC RAC RAC-HCl RAC-H2SO4 RAC-S RAC-HS
GPa 33.7 21.9 26.3 30.4 28.8 31.3
Fig.1 Modulus of Elasticity of Concrete Specimens at 28 Days

3.2 Alkalinity test
Alkalinity test was conducted at 28 days of age, for samples RACHCl and RACH2SO4 and compared with the
results of samples from NAC. The specimens were taken out from the curing tank and dried in oven at 105°C
for 24 hours. The dry specimens are cooled to room temperature and broken to separate mortar from it. This
mortar in ground into powder form and sieved in 150 µ sieve. 10 gm of powdered mortar is taken and diluted in
50 ml of distilled water and completely stirred in. Then, the pH meter is immersed into the solution and the pH
value is noted. A pH value greater than 12 indicates low potential for corrosion. The results of the alkalinity test
are given in Table 3, from which, it is clear that all the concrete mixtures have low potential of corrosion.
551
Table 3 Alkalinity of concrete specimens at 28 days of age
Mix Series NAC RAC RACHCl RACH2SO4
pH 12.59 12.46 12.28 12.20
4.1 Properties of Recycled Aggregate
The properties of RA are compared with those of NA, and the salient aspects are discussed below.
4.1.1 Specific Gravity, Water Absorption and Bulk Density:
The specific gravity (saturated surface dry condition) of RA was found to be 2.44, which is lower than the
specific gravity of natural aggregates of 2.79. The water absorption of NA and RA are 0.3 and 1.56%
respectively. The main factors affecting specific gravity and water absorption of RA are the source of aggregate
and the old mortars adhering to it. However, it is clear from Table 3 that desired qualities are achieved in RA
through treatment techniques. The quality of RAH2SO4 and RAHS are comparable to that of NA. The reason is that
these treatment techniques remove the attached mortar adhered to RA efficiently. However, the water absorption
of all recycled aggregates including that of treated ones are higher than that of NA and hence saturated surface
dry conditions of aggregate were maintained before the start of mixing operations. A remarkable improvement
over the bulk density is also seen, indicating the effect of treatment process on removal of loose adhered mortar.
4.1.2 Crushing and Impact Values:
The crushing value and impact value are the general tests, which are prescribed for finding the strength of the
aggregate. The impact value is calculated by recording the fractions passing and retained in a 2.36-mm sieve
after the material has received 15 blows from a standard weight. This is expressed as a percentage of the total
weight. This test is carried out to measure the resistance of a particular aggregate to sudden shock or impact. A
lower percentage indicates tougher and stronger aggregates. From the results, it was observed that the crushing
value of NA and RA is 27 and 36% respectively. The impact value of NA and RA is 23 and 31% respectively. It
is clear that, the resistance against crushing, impact and abrasion of RA are relatively lower than NA, as
observed by other researchers, (Saravana Kumar and Dhinakaran, 2012) due to the separation and crushing of
porous mortar coating. The crushing and impact values of treated RA are improved to the value of 33 to 33%
and 26 to 29% respectively, nearing the values of NA. This remarkable improvement is again attributable to the
effect of treatment techniques adopted.
4.1.3 Hardness of the Aggregate:
Hardness of the aggregate is defined as the resistance to wear obtained in terms of aggregate abrasion value, and
it is determined by the Los Angeles abrasion testing machine. The abrasion value of NA and RA is 29 and 45%
respectively. The effect of chemical and mechanical treatment is significant in improving the hardness of the
aggregate, as it reduces the abrasion value of RA between 28 to 30%, which is as close to NA.
4.2 Properties of Recycled Aggregate Concrete
4.2.1 Workability of Fresh Concrete:
The workability of fresh concrete mixes was determined by slump test and the results are presented in Table. 2.
The slump value of 105 mm was obtained in NAC and only 65 mm was obtained in RAC, in spite of
maintaining RA in saturated surface dry condition, before mixing. The reason is due to the rough texture and
relatively larger surface area of RA demanding higher water, as compared to NAC. Workability of concrete
mixes prepared with treated RA is higher than normal RAC and a maximum workability of 90 mm was
witnessed in RACHS.
552
4.2.2 Compressive strength and Static Modulus of Elasticity:
The compressive strength results presented in Table 2 are the average of three measurements. From this Table, it
is seen that the 28 days compressive strength of RAC is only 81 % of NAC. This reduced strength is due to (i)
the adhered mortar and (ii) the higher initial free water content in the concrete mixture due to the higher water
absorption of RA (Table 1), which was used at the air dried condition along with the moisture content of
aggregate. The 28 days compressive strengths of mixes prepared with chemically treated RA are in the range of
90 to 95% of NAC. Among the chemicals, H2SO4 seems to remove adhered mortar more efficiently than HCl
for the same concentration. It is also seen that the compressive strength of RACHS is 96% of NAC, denoting
the higher efficiency of heating and scrubbing technique than simple scrubbing technique, for obvious reasons.
The static modulus of elasticity values of various concrete mixtures are presented in Fig. 2. Each presented
value is the average of three measurements. The static modulus of elasticity of RAC is almost 35% less than the
value of NAC. However, this reduction is reduced to 22 to 8% and 14 to 7% in concrete mixes prepared with
chemically and mechanically treated RAs respectively. The static modulus of elasticity values of RACH2SO4 and
RACHS are comparable to that of NAC, which is mainly due to the effective removal of loose, cracky mortars on
the surface of aggregate.
5. CONCLUSIONS
The following conclusions can be made based on the results of this study:
1. Quality of recycled concrete aggregate is lower than natural aggregate quality, due to the mortar that remains
attached on the surface.
2. The physical and mechanical properties of recycled aggregates such as specific gravity, water absorption,
bulk density, crushing value, impact value and abrasion value can be improved by suitable chemical or
mechanical treatments.
3. The quality of recycled aggregates obtained after treatment are comparable to the quality of natural
aggregate.
4. The compressive strength and static modulus of elasticity of recycled aggregate concrete are lower than
natural aggregate concrete. However, the compressive strength and static modulus of elasticity of recycled
aggregate concretes prepared with recycled aggregate obtained from sulphuric acid treatment and heating &
scrubbing treatment are as good as natural aggregate concretes.
5. Presoaking recycled aggregate in H2SO4 is more efficient than HCl in removing the attached mortar of
recycled aggregate. Hence, the properties of recycled aggregate as well as the recycled aggregate concrete
are improved.
6. Among the mechanical treatment methods, heating & scrubbing technique seems to improve the quality of
recycled aggregate and hence the property of recycled aggregate concrete also. However, this technique is
suitable, when infrastructures that provide the heat sources for this treatment are available economically.
ACKNOWLEDGEMENT
The authors wish to acknowledge All India Council for Technical Education (AICTE), India for their support in
sponsoring this project.
REFERENCES
[1] Aitcin, P. C. and Neville, A. M. High performance concrete demystified. Concrete International 1993.
[2] Alexander, M. G. The effects of ageing on the interfacial zone in concrete. Interfacial transition zone in
concrete: state-of-the-art report 1996: 150-174.
[3] Bentz, D. P. and Garboczi, E. J. Simulation studies of the effects of mineral admixtures on the cement
paste-aggregate interfacial zone. ACI Materials Journal 1991;September-October: 518-529.
[4] Buch, N., Frabizzio, M. A. and Hiller, J. E. Impact of coarse aggregates on transverse crack performance in
jointed concrete pavements. ACI Materials Journal 2000;May-June: 325-332.
553
[5] Fong, W. F. K, and Jaime, S. K. Y (2002).“Production and application of recycled aggregates.” Proc., Int.
Conf. on Innovation and Sustainable Development of Civil Engineering in the 21st century, China Civil
Engineering Society, Hong Kong.
[6] Gonzalez, B., and Martínez, F., "Concretes with aggregates from demolition waste and silica fume".
Materials and Mechanical Properties Building and Environment, Vol. 43, 2008, pp 429–437.
[7] IS 516 -1959 Indian Standard methods of tests for strength of concrete, Bureau of Indian Standards, New
Delhi
[8] IS 2386 (Part III) – 1963 “Indian Standard Methods of Test for Aggregate for concrete. Bureau of Indian
[9] IS: 383-1970 "Coarse and Fine Aggregates from Natural Sources for Concrete" Bureau of Indian Standards,
New Delhi
[10] Indian standard Code of Practice for Ordinary Portland Cement – 43 Grade IS 8112 -1989 Bureau of
Indian Standards, New Delhi
[11] IS: 10262- 2009 "Guidelines for concrete mix design proportioning" Bureau of Indian Standards, New
Delhi
[12] Katz, A Treatments for the Improvement of Recycled Aggregate ASCE Journal of Materials in Civil
Engineering, Vol. 16, No. 6, December 1, 2004.
[13] Kou SC, Poon CS, Chan D. Influence of fly ash as cement replacement on the properties of recycled
aggregate concrete. J Mater Civil Eng 2007; 19(9):709–18.
[14] Kou,S S and Poon CS “Properties of concrete prepared with PVA-impregnated recycled concrete
aggregates” Cement and Concrete Composites. 32 (2010)649–654
[15] Kwan, A. K. H., Wang, Z. M. and Chan, H. C. Mesoscopic study of concrete II: nonlinear finite element
analysis. Computers and Structures 1999;70(5): 545-556.
[16] Kong DY, Zheng JJ, Zhou XZ, et al. “Strengthening of the ITZ in recycled aggregate concrete. In: Xi’an
international conference on architecture and technology, proceedings – architecture in harmony, Xian, PRC;
September 23– 25 2006. p. 614–7.
[17] Limbachiya M. C., T. Leelawat and R. K. Dhir, “Use of recycled concrete aggregate in high-strength
concrete” RILEM Journal of Materials and Structures. Vol. 33, November 2000, pp 574-580
[18] Li JS, Xiao HN, Zhou Y. “Influence of coating recycled aggregate surface with pozzolanic powder on
properties of recycled aggregate concrete”. Construction and Building Materials 2009;23(3):1287–91.
[19] Mehta, P. K. and Aitcin, P. C. Microstructural basis of selection of materials and mix proportions for high
strength concrete. High Strength Concrete, Second International Symposium, ACI SP121 1990: 265-279.
[20] Ollivier JP, Maso JC, Bourdette B. Interfacial transition zone in concrete. Adv Cem Based Mater
1995(2):30–8.
[21] Otsuki N, Miyazato S, Yodsudjai W. Influence of recycled aggregate on interfacial transition zone, strength,
chloride penetration and carbonation of concrete. J Mater Civil Eng 2003;15(5):443–51.
[22] Parekh, D. N., and Modhera, C. D., “Assessment Of Recycled Aggregate Concrete" Journal of Engineering
554
Research and Studies, Vol.II, Issue I/1-9, January-March 2011.
[23] Poon CS, Shui ZH, Lam L. Effect of microstructure of ITZ on compressive strength of concrete prepared
with recycled aggregates. Constr Build Mater 2004; 18:461–8.
[24] Prokopski G, Halbiniak J. Interfacial transition zone in cementitious materials. Cem Concr Res
2000;30:579–83.
[25] Ryu JS. Improvement on strength and impermeability of recycled concrete made from crushed concrete
coarse aggregate. J Mater Sci Lett 2002; 21:1565–7.
[26] Saravana Kumar P and Dhinakaran G, “Effect of Admixed Recycled Aggregate Concrete on Properties of
Fresh and Hardened Concrete”, ASCE Journal of Materials in Civil Engineering, Vol. 24. (4), April 2012
[27] Tam VWY, Gao XF, Tam CM. Microstructural analysis of recycled aggregate concrete produced from
two-stage mixing approach. Cem Concr Res 2005; 35:1195–203.
[28] Tam VWY, Tam CM and Le KN. “Removal of Cement Mortar remains from Recycled Aggregate using
Pre-Soaking Approaches”, Resources Conservation and Recycling Vol. 50(1) 2007, pp. 82-101.
[29] Tsujino M et al (2008). Completely recyclable concrete of aggregate recovery type by using microwave
heating technology. Proceedings of the third ACF international Conference, Ho Chi Minh, Vietnam.
[30] Yong, P. L and Teo, D. C. L, ‘Utilization of recycled aggregate as coarse aggregate in concrete’, Journal of
civil engineering, Vol. 1 (1), 2009.
555
Paper No.: 097
FINITE ELEMENT ANALYSIS OF SANDWICH PLATES WITH

FUNCTIONALLY GRADED MATERIAL CORE USING A LAYERWISE
THEORY
1 2
Shashank Pandey and S. Pradyumna
Department of Applied Mechanics, Indian Institute of Technology Delhi, New Delhi

Email: 1shashankpandey86@gmail.com,2pradyum@am.iitd.ac.in
ABSTRACT:
Static and free vibration analyses of sandwich plate are presented here. The sandwich plate is made of isotropic
face sheets and functionally graded material (FGM) core. The Material properties of the FGM core are varied
according to a power law distribution in terms of volume fraction of the constituent material. Numerical results
are presented for deflection, free vibration response, in-plane as well as transverse stress. The transverse shear
stresses are obtained through solving the equations of equilibrium. From present analysis it is inferred that the
present formulation is accurate and straight forward to implement.
KEYWORDS: FEM analysis, FGM plate, Layerwise theory, Free vibration, Static analyses
1. INTRODUCTION
FGM was first introduced by Japanese scientists in the year 1984 as a thermal shield (Koizami, 1993). The
ability to withstand mechanical and thermal loads is an important characteristic of FGM plate. Also, there is a
gradual variation of interlaminar stress at core-face sheet interface, hence failure due to delamination could be
reduced, therefore due to above advantages need to manufacture functionally graded sandwich plate was uttered.
The ceramic constituent of the FGM gives the insulation to high temperatures and the metal constituent imparts
the structural strength and toughness to the structure.
Because of sake of brevity only the pertinent pioneering works are referred here. In comprehensive studies by
Zenkour (2005a,b), it was shown that as the constituent of ceramic increases the critical buckling load and
vibration frequency increases whereas deflection and transverse shear stresses decrease. Thus, it was concluded
the gradient index play an important role in determining the response of FGM plate. Ganapathi (2012) presented
bending and vibration of functionally graded material sandwich plates using an accurate theory and a QUAD-8
shear flexible element.Carrera. (2011) concluded that refinements of classical theories that include additional
in-plane variables could lead to meaningless results unless transverse normal strain effects are taken into
account for a functionally graded sandwich plates and shells. Dozio (2013) presented a unified Ritz method for
accurate evaluation of natural frequencies of thick and thin FGM sandwich plates with various boundary
conditions.
In this paper, an eight–noded quadrilateral plate element with 9 degrees of freedom per node is employed to
study the static deflection, free vibration and stresses of a sandwich plate with FGM core. The efficacy of
present formulation for the static analyses i.e. bending under transverse load, the free vibration and stress
analysis subjected to the mechanical sinusoidal load is illustrated through numerical studies.
A sandwich plate of thickness h is made of top ceramic layer, a core with graded elastic properties and a metallic
bottom layer as shown in (Figure 1).The FGM core is of the plate is ceramic rich at the top whereas the bottom
surface is metal rich.
Ceramic
h3
FGM Core
h h2
Metal
h1
Figure 1 The material variation along the thickness of FGM sandwich plate
The volume fraction of ceramic and of metal varying through the core thickness according to a simple
power-law are expressed as Eq. (1) and the material properties with two constituents material is given by Eq. (2),
= where ≤z≤ (1)
z = − + (2)

Where n characterizes the material variation through panel thickness and 0≤n≤∞ and and refer to
corresponding properties of the ceramic and metal constituents.
2.1 Layerwise Theory
The layerwise theory presented in this paper is based on assumption of first-order shear deformation theory in
each layer and the imposition of displacement continuity at layer interface. In each layer similar assumption has
been made as those of first order plate theory. Due to size and complexity of the formulation, we restrict analysis
to a three layer sandwich as shown in Figure 2.However approach is easily extendible to a general laminate.
The displacement equation for middle layer is
, , = , + θ (3)
, , = , + θ 4
" , , == " , (5)
The displacement equation for Top layer is
#
, , = , + θ + $
θ# + #
θ # (6)
#
, , = , + θ + $
θ # + #
(7)
557
" #
, , =" , (8)
The displacement equation for lower layer is
%
, , = , − θ − &
θ% + %
θ % (9)
%
, , = , − θ − &
θ% + %
θ % (10)
" %
, , =" , (11)
Where u and v are the in-plane displacements at any point(x, y, z), and denote the in plane displacement of
the point(x, y, 0) on the midplane.
The strains are derived using Eqs.3-11 as:
+ 0
*ε . * ε/ . χ+
* .
( ( (
( ( ε1 (
+ 0
(ε ( ( χ+ (
γ+ 0 +z(k) χ+ (12)
) + - ) - ) -
= ε/1
(
( + ( (
( (
0
( ( 0 (
/ ( ' ,
0
γ Φ
'γ , ' Φ1 ,
0
In which,
= ,3 3 + (13)
345 365 345 365
ε + ,ε + ,ε + 3 3
: : :
:
3 θ; 3 θ; 3 θ;
[χ+ χ+ , χ+ ] = [ , +
3 θ9
3 3 3 3
, ] (14)
[Φ+ = +θ ,Φ + = +θ
<=> : + <=> : +
< <
] (15)
All the strain terms are expressed as
+ + ? + ? +
*ε . *ε . *ε . *ε .
ε = ε + + ε (16)
+ + + ? + ? +
) - ) - ) ? - ) -
ε
+ + + ? +
'γ , 'γ , 'γ , 'γ ,
+θ
<=> : +
γ+
@ A= B<= : C (17)
<
γ+ >
+θ +
<
Where, superscript m stands for membrane, f for the flexure (bending) and mf for the membrane-bending
coupling components for each layer. It is important to note here that the membrane-bending coupling is zero for
2nd layer. The stresses in the FGM plate (in material global coordinate) can be determined by using,
DEF = [GH ]DεF (18)
Where [GH ] is the effective stiffness coefficient matrix as given by relation [GH ]= (G − G
notations G and G are the effective stiffness coefficient matrices for ceramic and metal constituents
+G , the
respectively. It is worth mentioning that the FG panel becomes ceramic rich and metal rich, respectively for the
value of n=0 and ∞ (Figure 1).
558
2.2 FEM Formulation for FGM Sandwich plate using Layerwise Theory
From Eq. 3-11, the total displacement field is written as

M
u=∑#+J% +
, u= K " θ% θ% θ θ θ# θ# L (19)
The strain relations are given by
DεF=[N]+ DuF where k=1, 2, 3. (20)
For the transverse shear stress the constitutive relation gives the erroneous results. Hence the equilibrium
equations are used to solve the same.
e3 z(3
h3
3
e2 z(2) x
h h2
e1 z(1)
h1
1
layers
Figure 2 1D representation of the layerwise kinematics
In this section results for displacement and stresses of simply supported square (a/b=1) sandwich plate with FGM
sinusoidal transverse mechanical load = sin S sin T , applied at the top of plate. The top face sheet of
core are presented for various side to thickness ratio, power-law exponents. The plate is subjected to bi-
П П
the plate is made of alumina (Al2O3) with Young’s modulus U =380GPa, V = 3800YZ/\# , ν =0.3 and the
bottom face is made of aluminum (Al) with U = 70^ _, V = 2707YZ/\# ,ν =0.3.
3.1. Static analysis of Sandwich with FGM core
alumina and aluminum. The thickness of each face sheet is ℎ% = ℎ# = 0.1ℎ and the core layer thickness
The static analysis of sandwich plate with FGM core is carried out. In the following examples the materials are
is ℎ = 0.8ℎ.This configuration is denoted as a 1-8-1 sandwich.
Convergence studies were performed for different mesh sizes and it was found that mesh of 8 × 8 elements gives
converged results. Table 1 presents out of plane displacement and the value of in plane and transverse shear
stresses. It is seen that with the increase in gradient index, deflection value increases, whereas the in-plane and
along the thickness for a 1-3-1 sandwich plate with ℎ% = ℎ# = 0.2ℎ and ℎ = 0.6ℎ.Transverse shear stresses
transverse shear stress decreases. Figures 3-6 show the plots of variation of in-plane and transverse shear stress
obtained using the present formulation are parabolic in Figures 5 and 6 confirming the elasticity results.
559
The non-dimensional parameters used are
EH = E , evaluated at the ( , , )
S T
"
f= " , evaluated at ( , , 0)
% gh $ S T
S i jk Slk #
EH = E , evaluated at ( 0, 0, ) EH = E , evaluated at the ( , m, )

S
Slk Slk #
Table 1: Non-dimensional deflection and stresses in three layered FGM Sandwich plate under sinusoidal
"f EH EH EH
transverse load
a/h Source
n=1 n=10 n=1 n=10 n=1 n=10 n=1 n=10
4 Brischetto(2009) 0.7629 1.2232 0.6530 0.3627 _ _ 0.3007 0.1412
Carrera(2011) 0.7735 1.2240 _ _ 0.2604 0.1932 _ _
Neves(2012) 0.7746 1.2183 0.6130 0.3247 0.2709 0.1995 0.3301 0.1749
Present 0.7636 1.2235 0.6132 0.3474 0.2608 0.2009 0.3332 0.1789
10 Brischetto(2009) _ _ _ _ _ _ _ _
Carrera (2011) 0.6337 0.8743 _ _ 0.2594 0.1944 _ _
Neves(2012) 0.6357 0.8712 1.5700 0.9214 0.2724 0.2017 0.8453 0.4962
Present 0.6321 0.8696 1.5756 0.9234 0.2612 0.2034 0.8472 0.5003
100 Brischetto(2009) 0.6073 0.8077 15.784 9.5501 _ _ 8.4968 5.1402
Carrera (2011) 0.6072 0.8077 _ _ 0.2593 0.1946 _ _
Neves(2012) 0.6087 0.8075 15.7826 9.4300 0.2743 0.2230 8.4644 5.0672
Present 0.6073 0.8075 15.5107 9.4422 0.2609 0.2025 8.4674 5.0703
Table 2:Comparision of fundamental frequency parameter n =
oS q>
pg
>
of simply supported sandwich
plates with Al2O3/Al FG core
a/h Ref. P=0.5 P=1 P=2 P=5 P=10
5 Li(2008) 1.19580 1.25338 1.31569 1.39567 1.44540
Lorenzo Dozio (2013),ED2 1.20511 1.25927 1.32039 1.40196 1.45402
Lorenzo Dozio (2013),LD3 1.19580 1.25337 1.31569 1.39579 1.44551
Present 1.19565 1.25067 1.31114 1.39221 1.44449
10 Li(2008) 1.29751 1.34847 1.40828 1.49309 1.54980
Dozio (2013), ED2 1.30054 1.35031 1.40970 1.49501 1.55248
Dozio (2013),LD3 1.29750 1.34846 1.40828 1.49312 1.54983
Present 1.29801 1.34804 1.40712 1.49225 1.54971
100 Li(2008) 1.33931 1.38669 1.44491 1.53143 1.59105
Dozio (2013), ED2 1.33934 1.38671 1.44493 1.53145 1.59108
Dozio (2013),LD3 1.33931 1.38669 1.44491 1.53142 1.59105
Present 1.34022 1.38768 1.44585 1.53234 1.59195
Table 3: Fundamental frequency parameter n =

oS q>
pg
>
of simply supported (1-3-1) sandwich plates with
Al2O3/Al FG core
a/h n=0.5 n=1 n=2 n=5 n=10
5 1.1746 1.2092 1.2488 1.3034 1.3383
10 1.2733 1.3003 1.3364 1.3917 1.4291
100 1.3142 1.3374 1.3717 1.4270 1.4653
3.2. Free Vibrations analysis
In this section free vibration analysis of the FGM sandwich plate with metal at top surface (aluminum) and
ceramic (alumina) at bottom surface is presented here. The non-dimensional frequency parameter is taken
as n = pg where r is the natural frequency; V = 1kg/m# U = 1GPa.Table 2 gives the fundamental
oS q>
>
560
frequency of simply supported sandwich plate with FGM core for different a/h ratios and gradient index. The
present results are compared with the second order (ED2),third order (ED3) equivalent single layer theory and
layerwise theory based on Carrera unified formulation(CUF),second order layerwise (LD2)and third order
layerwise (LD3) theory results of Dozio (2013).It is observed from Table 2 that the natural frequency increases
with the increase of gradient index. This can be attributed to the increase in plate stiffness with gradient index.
Table 3 shows the fundamental frequency of FGM (1-3-1) sandwich plate with Al2O3/Al FG core. It is observed
that the stiffness of the plate is increasing with the increase of n.
0.5
0.4
0.3
Normalized thickness, z/h
0.2
0.1
0.0
-0.1
-0.2
-0.3 n = 10
n=2
-0.4
n = 0.2
-0.5
-10 0 10 20 30
stress, σxx(a/2, b/2, z)
Figure 3 Variation of EH through the thickness direction of a simply supported sandwich square plate with
FGM core, a/h=100 subjected to sinusoidal load
0.5
0.4
0.3
0.2
0.1
0.0
-0.1
-0.2
n = 0.2
-0.3 n=2
n = 10
-0.4
-0.5
-10 0 10 20 30
stress,σyy(a/2, b/2, z)
Figure 4 Variation of EH through the thickness direction of a simply supported sandwich square plate with
FGM core, a/h=100 subjected to sinusoidal load
561
0.5
0.4
0.3

0.2
0.1
0.0
-0.1
-0.2
-0.3 n = 0.2
n=2
-0.4
n = 10
-0.5
0.00 0.05 0.10 0.15 0.20 0.25 0.30
stress, σxz(a/2, b, z)
f z{ through the thickness of a simply supported sandwich square plate with FGM core,
Figure 5 Variation of y
a/h=100 subjected to sinusoidal load
0.5
0.4
0.3
0.2
0.1
0.0
-0.1
-0.2
n = 0.2
-0.3
n=2
-0.4 n = 10
-0.5
0.00 0.05 0.10 0.15 0.20
stress, σyz(a, b/2, z)
Figure 6 Variation of EH through the thickness simply supported sandwich square plate with FGM core,
a/h=100, subjected to sinusoidal load
4. CONCLUSIONS
In this paper, bending and free vibration behaviors of sandwich plates with FGM core are presented. The
efficacy of the present layerwise formulation is demonstrated by taking up benchmark problems available in the
literature. It is concluded that present layerwise finite element formulation serves as a powerful tool for static and
free vibration analyses of both thin and thick FGM sandwich plates.
562
REFERENCES
[1] Brischetto, S. (2009).Classical and mixed advanced models for sandwich plates embedding functionally
graded cores. Journal of Mechanics of Materials and Structures Vol. 4, No. 1.
[2] Carrera, E.; Brischetto, S; Cinefra, M.; Soave, M. (2011). Effects of thickness stretching in functionally
graded plates and shells, Composites: Part B 42,123-133.
[3] Ferreira ,A.J.M. ;Fasshauer ,G.E. ;Batra ,R.C.; Rodrigues, J.D. (2008).Static deformations and vibration
analysis of composite and sandwich plates using a layerwise theory and RBF-PS discretizations with
optimal shape parameter. Composite Structures86, 328–343.
[4] Ganapathi M. and Natarajan S. (2012). Bending and vibration of functionally graded material sandwich
plates using an accurate theory. Finite Elements in Analysis and Design 57, 32-42.
[5] Koizami, M. (1993).The concept of FGM ceramic.Transactions Func.Grad.Matter.34, 3-10.
[6] Dozio (2013). Natural frequencies of sandwich plates with FGM core via Variable-kinematic 2-D Ritz
models. Composite Structures96, 561–568.
[7] Zenkour, A. (2005a). A comprehensive analysis of functionally graded sandwich plates: Part 1 - deflection
and stresses, International Journal of Solids and Structures42, 5224–5242.
[8] Zenkour A. (2005b).A comprehensive analysis of functionally graded sandwich plates: Part 2 - buckling
and free vibration. International Journal of Solids and Structures42, 5243–5258.
563
Paper No.: 099
DIAGNOSIS OF MULTIPLE CRACK OF CANTILEVER COMPOSITE BEAM

BY VIBRATION ANALYSIS AND HYBRID AI TECHNIQUE
1 2
Irshad A Khan and Dayal R Parhi
1
Research Scholar, Department of mechanical Engineering, NIT Rourkela, India
2
Professor, Department of mechanical Engineering, NIT Rourkela, India
Email: irshadak85@gmail.com, drparhi@nitrkl.ac.in
ABSTRACT:
In the current investigation numerical and hybrid artificial intelligent (AI) technique methods are adopted for
detection of crack in a cantilever composite beam structure containing transverse cracks. The presence of cracks
a severe threat to the performance of structures and it affects the vibration signatures (Natural frequencies and
mode shapes). The material used in this analysis is graphite fiber reinforced polymide composite. The
Numerical analysis is carried out by using commercially available software package ANSYS to find the relation
between the change in natural frequencies and mode shapes for the cracked and un-cracked composite beam.
Which subsequently used to the design of smart system based on hybrid AI technique for prediction of crack
depths and locations following inverse techniques. The calculated modal frequencies, mode shapes, relative
crack locations and relative crack depths are used to design the fuzzy neural model. The measured vibration
signatures are used as inputs to the fuzzy segment of the hybrid model and initial relative crack depths and
initial crack locations are the output parameters of fuzzy model. The first three relative natural frequencies, first
three average relative mode shape difference and the output from the fuzzy model are used as inputs to the
neural segment of the hybrid model and final crack depths and locations are the output parameters. The
measured vibration signatures are used to formulate series of fuzzy rules and training patterns for the fuzzy and
neural model. Finally, the validation of the proposed method is carried out dynamically by means of
experimental results from the developed experimental setup.
KEYWORDS: Fuzzy, Neural, Natural frequency, Mode shape, Ansys.
1. INTRODUCTION
Health monitoring and the analysis of damage in the form of crack in Beam like dynamic structures are
important not only for leading safe operation but also retraining system performance. Since long efforts are on
their way to find a realistic solution for crack detection in beam structures in this regard many approaches have
so far being taken place. When a structure suffers from damages, its dynamic properties can change. Crack
damage leads to reduction in stiffness also with an inherent reduction in natural frequency and increase in modal
damping. The paper lends a viable relationship between the modal natural frequency and mode shape at the
different crack depth and location. Since last few decade free vibration analysis has become a topic of many
studies therefore attention is focused on it only.
Damage identification on a composite cantilever beam through vibration analysis using finite element analysis
software package ANSYS is established by Ramanamurthy et al. [3]. A composite matrix cracking model is
implemented in a thin-walled hollow circular cantilever beam using an effective stiffness approach by Pawar et
al. [4]. The composite beam model is used to develop a genetic fuzzy system to detect and locate the presence of
matrix cracks in the structure Damage Algorithm and Damage index method used to identify and locate the
damage in the composite beam.Discrete Wavelet Transform based method is presented for the identification of
multiple cracks in polymeric laminate composite beam by Andrzej K., [1]. The valuation of the crack locations
is based on the estimation of natural mode shapes of crack and uncrack beams. The mode shapes were estimated
experimentally using laser Doppler vibrometry. A structural health monitoring methodology is developed for
composite rotor blades based on finite element method by Pawar et al. [2]. An aero-elastic analysis is used to
obtained simulated blade response, loads and strain measurements for a damaged rotor. Genetic fuzzy systems
are developed for global online damage detection using displacement and force-based measurement deviations
between damaged and undamaged conditions and for local online damage detection using strains.. Continuous
wavelet transform is used to identification of crack in beam like structures by analysing the natural frequency
and mode shape of cracked cantilever beam by Rao et al. [5]. The effects of cracks on the dynamic
characteristics of a cantilever composite beam are studied by Kisa M. [6]. The material of composite beam is
graphite fibre-reinforced polyamide containing multiple non-propagating transverse cracks. The effects of the
crack location and depth and the fibre volume fraction and orientation of the fibre on the natural frequencies and
mode shapes of the beam are explored. Two Damage identification algorithms are established for assessment of
damage using modal test data which are similar in concept to the subspace rotation algorithm or best feasible
modal analysis method by Hu et al. [7]. Moreover, a quadratic programming model is set up the two
methodologies to damage assessments. Zhang et al. [8] have suggested a bearing fault detection method based on
multiple scale entropy and adaptive neuro fuzzy inference system (ANFIS) to measure the nonlinearity present in a
bearing system. They have conducted experiments on electrical motor bearing with three different fault categories
and the obtained results from the experimental have been used to design and train the ANFIS system for fault
diagnosis.
2. NUMERICAL ANALYSIS
The numerical analysis is brought out for the cracked cantilever composite beam shown in fig1, to locate the
mode shape of transverse vibration at different crack depth and crack location. The cracked beams of the current
research have the following dimensions.
Length of the Beam (L) = 800mm;
Width of the beam (W) = 50mm;
Thickness of the Beam (H) = 6mm

Relative crack depth (ψ1=a1/H) = Varies from 0.0833 to 0.5;
Relative crack depth (ψ2=a2/H) = Varies from 0.0.833 to 0.5;
Relative crack location (β1=L1/L) = Varies from 0.625 to 0.875;
Relative crack location (β2=L2/L) = Varies from 0.125 to 0.9375;
Properties of graphite fiber reinforced polymide composite material in analysis:

Young’s modulus of fiber = Ef = 275.6 GPa;
Young’s modulus of matrix = Em = 2.756 GPa;
Modulus of rigidity of fiber = Gf = 114.8 GPa;
Modulus of rigidity of matrix = Gm = 1.036 GPa;
Poisson’s ratio = νf = 0.2; Poisson’s ratio = νm = 0.33;
Density of fiber = ρf = 1.9 gr/cm3; Density of matrix = ρm = 1.6 gr/cm3;
Numerical modal analysis based on the finite element modeling is performed for studying the dynamic response
of a dynamic structure. The natural frequencies and mode shapes are important modal parameters in designing a
structure under dynamic loading conditions. The numerical analysis is accepted by using the finite element
software ANSYS in the frequency domain and obtain natural frequencies, and mode shapes.
A higher order 3-D, 8 node element having three degrees of freedom at each node: translations in the nodal x, y,
and z directions (Specified as SOLSH190 in ANSYS) was selected and used throughout the analysis. Each node
has three degrees of freedom, making a total twenty four degrees of freedom per element. The layers stacking in
ANSYS shown in fig2. The results of the numerical analysis for the first three mode shapes for un-cracked and
cracked beam (ψ1=0.166, ψ2=0.333 and β1=0.25, β2=0.5) are shown in the fig3.
565
Fig. 1 Geometry Cantilever beam with multiple cracks
Fig.3a. Relative Amplitude vs. Relative crack location

Fig.2 Layers Stacking in ANSYS from fixed end (1st mode of vibration)
Fig.3b. Relative Amplitude vs. Relative crack location from Fig.3c. Relative Amplitude vs. Relative crack location from
fixed end (2nd mode of vibration) fixed end (3rd mode of vibration)
3. HYBRID FUZZY-NEURAL ANALYSIS
Dynamic behavior of the beam changes due to presence of crack, the first three relative natural frequencies and
first three average relative mode shape differences of the cracked and un-cracked beam for different crack
locations and depths are calculated by numerical and experimental analysis. The calculated modal frequencies,
mode shapes, relative crack locations and relative crack depths are used to fabricate the fuzzy neural model. The
measured vibration parameters are used as inputs to the fuzzy segment of the hybrid model and initial relative
crack depths and initial crack locations are the output parameters. The first three relative natural frequencies,
first three average relative mode shape difference and the output from the fuzzy model are used as inputs to the
neural segment of the hybrid model and final crack depths and locations are the output parameters. The
measured vibration signatures are used to formulate series of fuzzy rules and training patterns for the fuzzy and
neural model. Finally, the validation of the proposed method is carried out dynamically by means of
experimental results from the developed experimental setup. The fuzzy segment of the hybrid model for
566
multiple crack prediction has been developed using triangular membership functions. The term used for the
inputs are as follows; Relative first natural frequency = “rfnf”; Relative second natural frequency = “rsnf”;
Relative third natural frequency = “rtnf”; Relative first mode shape difference = “rfmd”; Relative second mode
shape difference = “rsmd”; Relative third mode shape difference = “rtmd. The term used for the final outputs are
as follows; Relative first crack location = “rcl1” Relative second crack location = “rcl2” Relative first crack
depth = “rcd1” Relative second crack depth = “rcd2”. The neural network segment has ten-layered perceptron.
The chosen numbers of layers are found empirically to facilitate training. The input layer has eight neurons,
three for first three relative natural frequencies and other three for first three average relative mode shape
difference and output of fuzzy controller. The output layer has four neurons, which represents final relative
crack locations and final relative crack depths. The first hidden layer has 12 neurons, the second hidden layer
has 36 neurons, the third hidden layer has 50 neurons, the fourth hidden layer has 150 neurons, the fifth hidden
layer has 300 neurons, the sixth hidden layer has 150 neurons, the seventh hidden layer has 50 neurons, and the
eighth hidden layer has 8 neurons. These numbers of hidden neurons are also found empirically. The hybrid
model shown in fig. 4.
Fifth Hidden
Layer
Third Hidden (300 neurons) Seventh Hidden
Layer Layer
First Hidden
(50 neurons) (50 neurons)
Layer
(12 neurons)
fnf rcl1initial
Triangular fuzzy
model rcl2initial
snf
fnf rcl1final
tnf snf
rcl2final
tnf
fmd rcd1final
fmd smd
rcd2final
tmd
smd
rcd2initial
Output Layer
tmd rcd1initial (4 neurons)
Input Layer
Input Layer Second Hidden Eight Hidden
(6 neurons)
(10 neurons) Layer Layer
(36 neurons) (8 neurons)
Sixth Hidden
Fourth Hidden
Layer
Layer
(150 neurons)
(150 neurons)
Fig. 4 Hybrid Fuzzy-Neuro Model
4. EXPERIMENTAL INVESTIGATION
To validate the numerical analysis result, an experiment on composite beam has been performed shown in fig 5.
A composite beam was clamped at a vibrating table. During the experiment the cracked and undamaged beams
have been vibrated at their 1st, 2nd and 3rd mode of vibration by using an exciter and a function generator. The
vibrations characteristics such as natural frequencies and mode shape of the beams correspond to 1st, 2nd and 3rd
mode of vibration have been recorded by placing the accelerometer along the length of the beams and displayed
on the vibration indicator. The surface specimens were cut with EDM wire cut machine to ensure high accuracy
of cracked model. The experimental results are in close justification with neural analysis results. These results
for first three modes are plotted in fig6. Corresponding numerical results for the cracked and un-cracked beam
are also presented in the same graph for comparison. The comparison of results between Hybrid controller,
numerical analysis and experimental analysis shown in table1.
567
Fig.5 Schematic block diagram of experimental set-up
1. Data acquisition (Accelerometer); 2. Vibration analyser; 3. Vibration indicator embedded with software
(Pulse Labshop); 4. Power Distribution; 5. Function generator; 6. Power amplifier; 7. Vibration exciter;
8. Cracked Cantilever Composite beam
Fig.6a. Relative Amplitude vs. Relative crack Fig.6b. Relative Amplitude vs. Relative crack
location from fixed end (1st mode of vibration) location from fixed end (2nd mode of vibration)
Fig.6c. Relative Amplitude vs. Relative crack

location from fixed end (3rd mode of vibration)
568
5. CONCLUSION
The conclusions derived from the various analyses as mentioned above are depicted below.
1. The results numerical analysis is well agreed with hybrid controller results.
2. The study of vibration signatures of the cracked composite beam shows a variation of mode shapes and
natural frequencies for the cracked and un-cracked beam.
3. The results from numerical analysis, hybrid analysis are compared with the experimental results. They
show good judgment.
4. This method can be employed as a health monitoring tool for vibrating damaged dynamic structures.
REFERENCES
[1] Ramanamurthy, E.V.V. and Chandrasekaran, K. (2012). Vibration analysis on a composite beam to identify
damage and damage Severity using finite element method, International Journal of Engineering Science
and Technology, 3, 5865-5888.
[2] Pawar P. M. and Ganguli R. (2005). Matrix Crack Detection in Thin-walled Composite Beam using
Genetic Fuzzy System, Journal of Intelligent Material Systems and Structures 16,395-409.
[3] Katunin A. (2010). Identification of multiple cracks in composite beams using discrete wavelet transform,
Scientific Problems of Machines Operation and Maintenance, 2(162), 42-52.
[4] Pawar P. M. and Ganguli R. (2007). Genetic fuzzy system for online structural health monitoring of
composite helicopter rotor blades, Mechanical Systems and Signal processing, 21, 2212-2236.
[5] Rao, D. Srinivasa., Rao, K. Mallikarjuna., Raju, G.V. (2010). Crack Identification on a beam by Vibration
Measurement and Wavelet Analysis, International Journal of Engineering Science and Technology, 2(5),
907-912.
[6] Kisa, M. (2004). Free vibration analysis of a cantilever composite beam with multiple cracks”, Composites
Science and Technology, 64, 1391-1402.
[7] Hu, N., Wang, X., Fukunaga, H., Yao, Z.H., Zhang, H.X., Wu, Z.S. (2001). Damage Assessment of
structures using test data, International Journal of solids and structures 38, 3111-3124.
[8] Zhang, L., Xiong, G., Liu, H., Zou, H. and Guo, W. (2010). Bearing fault diagnosis using multi-scale
entropy and adaptive neuro-fuzzy inference, Expert Systems with Applications, 37, 6077–6085.
569
Table 1 Comparison of results between Hybrid model, Numerical analysis and Experiment analysis.
Average Average Average Hybrid Model Numerical Experimental

Relative Relative Relative Relative Relative Relative relative1st crack depth “rcd1” relative1st crack depth “rcd1” relative1st crack depth “rcd1”
first second third first second third 1st crack location “rcl1” 1st crack location “rcl1” 1st crack location “rcl1”
natural natural natural mode mode mode 2nd crack depth “rcd2”, 2nd crack depth “rcd2”, 2nd crack depth “rcd2”,
frequency frequency frequency shape shape shape 2nd crack location“rcl2” 2nd crack location“rcl2” 2nd crack location“rcl2”
“fnf” “snf” “tnf” difference difference difference
“fmd” “smd” “tmd” rcd1 rcl1 rcd2 rcl2 rcd1 rcl1 rcd2 rcl2 rcd1 rcl1 rcd2 rcl2
0.9978 0.9983 0.9878 0.0036 0.0329 0.0141 0.17 0.42 0.16 0.17 0.164 0.24 0.23 0.49 0.161 0.19 0.18 0.44
570
0.9987 0.9972 0.9981 0.2936 0.3428 0.2623 0.412 0.66 0.18 0.18 0.24 0.24 0.414 0.73 0.19 0.20 0.409 0.68
0.9936 0.9976 0.9987 0.0138 0.014 0.0832 0.23 0.622 0.164 0.373 0.159 0.368 0.18 0.618 0.157 0.367 0.16 0.616
0.9976 0.9991 0.9988 0.0014 0.0041 0.0812 0.416 0.49 0.342 0.123 0.327 0.119 0.410 0.45 0.325 0.117 0.409 0.43
0.9849 0.9982 0.9869 0.0134 0.0211 0.0119 0.22 0.622 0.414 0.372 0.410 0.367 0.19 0.618 0.408 0.365 0.18 0.616
0.9989 0.9973 0.9974 0.0017 0.0025 0.0079 0.20 0.71 0.46 0.21 0.44 0.18 0.18 0.68 0.42 0.17 0.16 0.66
0.9979 0.9985 0.9993 0.0087 0.0036 0.0042 0.18 0.42 0.43 0.119 0.48 0.123 0.22 0.46 0.41 0.117 0.16 0.40
0.9962 0.9989 0.9991 0.0036 0.9729 0.2263 0.326 0.868 0.409 0.118 0.414 0.123 0.330 0.873 0.407 0.117 0.324 0.866
Paper No.: 109
PERFORMANCE BASED DESIGN OF RCC BUILDING UNDER

EARTHQUAKE FORCES
Dr. Rehan A. Khan1 and Dr. T. Naqvi2
1. Associate Professor, Department of Civil Engg, Z.H.College of Engineering and Technology,

Aligarh Muslim University, Aligarh (U.P.) – 202002, E-mail: rehan_iitd@rediffmail.com
2. Professor, Department of Civil Engg, Z.H.College of Engineering and Technology,
Aligarh Muslim University, Aligarh (U.P.) – 202002.
ABSTRACT :
The basic concept of Performance Based Seismic Design is to provide engineers with the capability to design
buildings that have a predictable and reliable performance in earthquakes. Performance based Seismic design is
an elastic design methodology done on the probable performance of the building under input ground motions.
The present study is an effort to understand Performance Based Design Approach. In this, a five storey
symmetrical building is designed using STAAD.Pro and the performance based seismic design is performed by
N2 method using a simple computer-based pushover analysis technique using SAP2000, a product of
Computers and Structures International. The procedure compares the capacity of the structure (in the form of a
pushover curve) of a MDOF system with the demands on the structure (in the form of inelastic response spectra
of a single degree freedom system). The method is formulated in acceleration displacement format. The
graphical intersection of the two curves approximates the performance point of the structure. The proposed
method is illustrated by finding the seismic performance point for a five storey reinforced concrete framed
building located in Zone-IV, symmetrical in plan (designed according to IS 456:2002) subjected to three
different PGA levels as input ground motion. An extensive parametric study is conducted to investigate the
effect of many important parameters on the Performance point. The parameters include effect of input ground
motion on performance point, changing percentage of reinforcement in columns, size of columns, beams
individually. The results of analysis are compared in terms of base shear and storey displacements.
KEYWORDS: Performance based seismic design, elastic response spectrum, inelastic response
spectrum, N2 method, pushover analysis.
1. MODELING APPROACH
A 3-D model of five storey RCC building shown in Fig.1 has been created using finite element package SAP
2000 to undertake the non linear analysis. Beams and columns are modeled as nonlinear frame elements with
lumped plasticity at the start and the end of each element. SAP 2000 provides default-hinge properties and
recommends PMM hinges for columns and M3 hinges for beams as described in FEMA-356.
2. ASSUMPTIONS
a. The material is homogeneous, isotropic and linearly elastic.

b. All columns supports are considered as fixed at the foundation.
c. Tensile strength of concrete is ignored in sections subjected to bending.
d. The super structure is analyzed independently from foundation and soil medium, on the assumptions that
foundations are fixed.
e. The floor acts as diaphragms, which are rigid in the horizontal plane.
f. Pushover hinges are assigned to all the member ends. In case of columns PMM hinges (i.e. Axial Force and
Biaxial Moment Hinge) are provided at both the ends, while in case of beams M3 hinges (i.e. Bending
Moment hinge) are provided at both the ends.
g. The maximum target displacement of the structure is kept at 4.0% of the height of the building = (4.0/100) x
17.5= 0.7m = 700mm.
3. SEISMIC INPUT
The elastic response spectra with PGA levels of 0.1g, 0.3g and 0.4g are used as per IS Code 1893: 2002. The
elastic response spectra are changed to inelastic response spectra which are taken as seismic input. The response
spectra for elastic and inelastic response for PGA levels of 0.3g is shown Fig.2.
4. PUSH OVER ANALYSIS
In Pushover analysis, a static horizontal force profile, usually proportional to the design force profiles specified
in the codes, is applied to the structure. The force profile is then incremented in small steps and the structure is
analyzed at each step. As the loads are increased, the building undergoes yielding at a few locations. Every time
such yielding takes place, the structural properties are modified approximately to reflect the yielding. The
analysis is continued till the structure collapses, or the building reaches certain level of lateral displacement.
5. CAPACITY SPECTRUM METHOD
The nonlinear method used here to perform seismic analysis is known as N2 method [1]. It combines the
pushover analysis of a multi degree of freedom (MDOF) system with the response spectrum of equivalent single
degree of freedom (SDOF) system. The method is formulated in the acceleration- displacement format, which
enables the visual interpretation of the procedure and of the relations between the basic quantities controlling the
seismic response. Inelastic spectra rather than elastic spectra with equivalent damping and period are applied.
This feature represents the major difference with respect to the capacity spectrum method [2]. It characterizes
the seismic demand initially using a 5% damped linear-inelastic response spectrum and reduces the spectrum to
reflect the effects of energy dissipation to estimate the inelastic displacement demand. The point at which the
Capacity curve intersects the reduced demand curve represents the performance point at which capacity and
demand are equal.
6. PUSHOVER ANALYSIS: CONSTRUCTING CAPACITY SPECTRUM
Using a pushover analysis, a characteristic nonlinear force - displacement relationship of the MDOF system can
be determined. In principle, any force and displacement can be chosen. In this study, base shear and roof (top)
displacement have been used as representative of force and displacement, respectively.
The capacity spectrum can be developed from the pushover curve by a point by point conversion to the first
mode spectral coordinates. Any point Vi (Base Shear), δi (Roof Displacement) on the capacity (pushover) curve
is converted to the corresponding point Sai, Sdi on the capacity spectrum using the equations:
Sai = (1)
Sdi = (2)
× ∅ ,
Where α1 and PF1, are the modal mass coefficients and participation factors for the first natural mode of the
structure respectively. φ1roof is the roof level amplitude of the first mode.
The modal participation factors and modal coefficient are calculated as:
(3)
572
Where wi is the weight at any level i. As displacement increase, the period of the structure lengthens. This is
reflected directly in the capacity spectrum. Inelastic displacements increase damping and reduce demand. The
Capacity Spectrum Method reduces the demand to find an intersection with the capacity spectrum, where the
displacement is consistent with the implied damping.
7. NUMERICAL STUDY
To illustrate the PBD procedure for finding the performance point, a five storey concrete frame as shown in
Fig.1 is taken as an example. The properties of the concrete frame are shown in Table1. The frame is designed
according to IS 456: 2000 (with the superimposed vertical loads) using STAAD Pro. Structural details and
natural frequencies of the concrete frame are given in Table 2 and 3 respectively. It is seen from the Table 3 that
the natural frequencies of the frame are quite widely spaced, which means that the dynamic response will be
dominated by the first mode only and the contribution due to response of the higher modes can be neglected.
The first two mode shapes are shown in Fig.3 (i) to 3(ii). The frame is subjected to inelastic response spectrum
as per IS Code 1893: 2002 for 5% damping (for medium soil) for PGA levels of 0.1g, 0.3g and 0.4g obtained
from elastic response spectra. The pushover analysis is performed on the RC building (designed according to IS
456: 2000) and redesigning by changing the main reinforcement of various frame elements and again analyzing.
The performance based seismic engineering technique known as N2 procedure has been effectively used in this
regard. The pushover analysis has been carried out using SAP2000, a product of Computers and Structures
International. For parametric studies, a total of 19 cases as per Table 4, for a particular five storey building
frame located in Zone-IV have been analyzed, changing reinforcement and sizes of different structural elements,
i.e. beams and columns, in different combinations as well as at different storey levels. A comparison between
response obtained using elastic and inelastic response spectra for PGA level 0.4g is also obtained.
7.1 Effect of PGA levels on performance of the building for elastic and in-elastic response spectra.
Pushover analysis of the five storey RC framed building subjected to monotonically increasing lateral forces
with an invariant height wise distribution is performed using SAP2000. The pushover curve i.e. curve between
Displacement and Base shear is obtained for inelastic response spectra of 0.1g, 0.3g and 0.4g. The pushover
curve for inelastic response spectra of 0.3g is shown in Fig. 4. It is observed that as the PGA level increases base
shear and roof displacement also increases. Table 5 show the roof displacements for the frame for different PGA
levels at different performance levels for elastic and inelastic response spectra respectively. Table 6 show the
ductility demands for the frame for different PGA levels at different performance levels for elastic and inelastic
response spectra respectively. As is obvious the roof displacement and ductility demand increases as the
performance level goes from operational to collapse prevention level for each PGA level. It is also seen that as
input PGA level increases, the roof displacement and ductility demand also increases.
7.2 Base shear and displacement variation on performance point of the building for elastic and in-elastic
response spectra.
The performance point of the structure is determined by using the Acceleration-Displacement Response Spectra
method (ADRS) pushover curves obtained. The performance point is the point where the capacity and demand
of the structure are equal. Superimposition of the capacity spectrum curve and demand curve are obtained for
inelastic response of different PGA levels. The capacity and Demand curve for inelastic response spectra of 0.3g
is shown in Fig.5. The base shear and displacement variation at performance point is shown in Table 7 for
elastic and inelastic response spectra respectively, for different PGA levels. It can be concluded that as g level
increases, performance point increases.
573
7.3 Effect of change in various parameters on the performance of the building.
7.3.1 Effect of change of reinforcement in Columns (Case B- Case G)
Table 8 shows the effect of change of reinforcement in columns on the performance point. It is seen that as the
reinforcement increases, the base shear increases and the roof displacement decreases and vice versa.
7.3.2 Effect of change of size of the Columns and Beams simultaneously (Case P- Case S)
Table 9 shows the effect of change of size of columns on the performance point. It is seen that as the size
increases, the base shear increases and the roof displacement decreases.
7.3.3 Effect of change of Response Reduction Factor (R)

Table 10 shows that the performance point is slightly affected by variation of Response Reduction Factor (R).
7.3.4 Effect of Soil Behaviour on Different Performance Levels

Table 11 shows comparison of elastic and inelastic response, there is decrease in response due to non-linearity
of soil damping.
Table 1: Properties of five storey two-bay RC frame
1 Size of the Beams 300*450 mm 2
2 Size of the Columns 450*450 mm 2

3 Thickness of Slab 150 mm
4 Bay Width 5.0 m
5 Storey Height 3.5 m
6 Grade of concrete M-20
7 Grade of Steel Fe-415
Table 2: Structural detail as per STAAD. Pro)
S. No. Structural Elements Dimensions(m) Reinforcement

Area
Breadth Depth (mm2)
1 1st Storey Columns 0.450 0.450 2714
2 2nd Storey Columns 0.450 0.450 1608.5
3 3rd Storey Columns 0.450 0.450 1357
4 4th Storey Columns 0.450 0.450 905
5 5th Storey Columns 0.450 0.450 1810
Top Bottom
6 1st Storey Beams 0.300 0.450 822 520
7 2nd Storey Beams 0.300 0.450 897 515
8 3rd Storey Beams 0.300 0.450 930 516
9 4th Storey Beams 0.300 0.450 952 519
10 5th Storey Beams 0.300 0.450 313 256
574
Table 3: Natural frequencies and time periods
Period Frequency
Mode Shapes
(sec) ( cycle/sec)
1 1.10768 0.90279
2 0.34282 2.91698
3 0.18828 5.31124
Table 4: Description of various cases

S.
Case Description Of Cases
No.
1 A Basic Structure 11 K 20% Decrease In Beams Size
10% Increase In Columns
2 B 12 L 10% Increase In Columns Size
Reinforcement
3 C 13 M 20% Increase In Columns Size
Reinforcement
4 D 14 N 10% Decrease In Columns Size
Reinforcement
10% Decrease In Columns
5 E 15 O 20% Decrease In Columns Size
Reinforcement
20% Decrease In Columns 10% Increase In Columns &
6 F 16 P
Reinforcement Beams Size
30% Decrease In Columns 20% Increase In Columns &
7 G 17 Q
Reinforcement Beams Size
10% Decrease In Columns &
8 H 10% Increase In Beams Size 18 R
Beams Size
20% Decrease In Columns &
9 I 20% Increase In Beams Size 19 S
Beams Size
10 J 10% Decrease In Beams Size
Table 5: Roof Displacements for elastic and inelastic response spectra for different Performance Levels
Roof Displacement for Roof Displacement for

Roof Displacement for
S. Performance PGA level 0.1g (mm) PGA level 0.4g (mm)
PGA level 0.3g (mm)
No. Level
Elastic Inelastic Elastic Inelastic Elastic Inelastic
1 Operational 23.4253 23.0635 24.0513 23.1533 24.2999 23.1987
Immediate
2 59.6849 51.9550 59.8973 52.5096 63.3516 52.8775
Occupancy
3 Life Safety 70.4745 62.4786 140.9281 74.2460 171.8685 155.8570
Collapse
4 197.7286 165.0212 236.5626 217.6637 239.7948 224.5969
Prevention
Complete
5 ∞ ∞ ∞ ∞ ∞ ∞
Collapse
575
Table 6: Ductility Demands for elastic and inelastic response spectra for different Performance Levels
Ductility Demand for Ductility Demand for Ductility Demand for

S. Performance PGA level 0.1g PGA level 0.3g PGA level 0.4g
No. Level
1 Operational 1.000 1.000 1.000 1.000 1.000 1.000
Immediate
2 2.548 2.253 2.490 2.268 2.607 2.2800
Occupancy
3 Life Safety 3.009 2.709 5.860 3.207 7.073 6.718
Collapse
4 8.441 7.155 9.836 9.401 9.868 9.682
Prevention
Complete
5 ∞ ∞ ∞ ∞ ∞ ∞
Collapse
Table 7: Performance Point for elastic and inelastic response spectra for different PGA Levels
PGA Level 0.1g 0.3g 0.4g

Performance point
Base Shear (KN) 602.337 70.937 708.364 218.364 755.420 287.784
Roof Displacement
29.331 2.975 86.576 9.295 137.582 12.340
(mm)
Table 8: Effect of change of reinforcement in columns on performance point for PGA level 0.4g
S. Case Roof Displacement % Change in Roof Displacement Base Shear % Change in

No. (mm ) (KN) Base Shear
1 A 137.582 755.420
2 B 134.208 2.4524 789.374 -4.4947
3 C 129.884 5.9268 826.666 -9.4313
4 D 127.365 7.4261 894.974 -18.4664
5 E 139.052 -1.0685 724.102 4.1458
6 F 143.881 -4.5784 692.901 8.2761
7 G 153.439 -11.5255 632.722 16.2424
Table: 9 Effect of change of size of beams and columns of the frame on performance point for PGA Level 0.4g
S. Roof Displacement % Change in Roof Base Shear

Case % Change in Base Shear
No. (m ) Displacement (KN)
1 A 137.582 755.420
2 P 132.186 3.922 819.896 -8.535
3 Q 127.245 7.513 883.651 -16.975
4 R 146.601 -6.555 683.311 9.546
5 S 160.977 -17.004 604.814 19.937
576
Table 10: Effect on performance point by changing the different values of R for PGA Level 0.4g
Response
Spectral Spectral Base
reduction Roof Displacement
S. No. Displacement Acceleration Shear
factor (∆)
(Sd ) (Sa) (V)
(R)
1 2.0 0.117 0.108 662.048 0.153
2 2.5 0.114 0.111 685.180 0.149
3 3.0 0.112 0.114 704.206 0.146
4 3.5 0.110 0.116 720.206 0.143
5 4.0 0.109 0.117 733.933 0.141
6 4.5 0.109 0.119 745.528 0.139
7 5.0 0.107 0.120 755.420 0.138
Table 11: Comparison of target roof displacement and actual displacement observed at various performance
levels for PGA Level 0.4g
Target Roof Actual Displacement for Actual Displacement for
Performance
S.No. Displacement elastic response inelastic response
Level
(% of Height) (% of Height) (% of Height)
1 Operational 0.37 0.14 0.13
Immediate
2 0.70 0.36 0.30
Occupancy
3 Life Safety 2.50 0.98 0.89
Collapse
4 5.00 1.37 1.28
Prevention
Spectral acceleration(Sa/g)
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 1 2 3 4 5
Period(s)
Figure1: 3D view of the five storey two bay RC Figure 2: Response spectra for PGA level 0.3g
frame
577
Figure 3(i): Mode Shape 1 Figure 3(ii): Mode Shape 2
Figure 4: Pushover Curve for Inelastic Response Figure 5: Superimposition of Capacity and Demand
Spectrum 0.3g Curve for Inelastic Response Spectrum 0.3g
8. CONCLUSIONS
Based on the present study, the following conclusions can be drawn:
1. Pushover Analysis is an elegant tool to visualize the performance level of a building under a given
earthquake. It provides valuable information for the performance based seismic design of building frame.
2. Since frequencies are wide apart, thus for pushover analysis higher modes are neglected.
3. Roof displacements increase as the frame is pushed forward for different PGA levels of earthquake.
4. Ductility demand increases as the frame is pushed towards plastic range and ultimately at ∞ demand the
structure collapses due to plastic mechanism for different PGA levels.
5. As the response changes from elastic to inelastic, roof displacements and ductility demands decrease for
different PGA levels.
6. The performance point obtained satisfies the acceptance criteria for immediate occupancy and life safety
limit states for various intensities of earthquake.
7. The increase in reinforcement of columns results in nominal change in base shear and displacement.
8. As the size increases, the roof displacement decreases whereas base shear increases.
9. As the size decreases, the roof displacement increases whereas base shear decreases.
10. Performance point is slightly affected by variation of Response Reduction Factor (R).
578
11. Due to increased damping of soil, maximum response reduces when elastic behaviour changes to inelastic.
REFERENCES
[1] Fajfar, P., (2000). A non linear analysis method for performance based seismic design, Earthquake Spectra,
Volume 16, Issue 3/573- 592.
[2] ATC, (2006). Next-Generation Performance-Based Seismic Design Guidelines: Program Plan for New and
Existing Buildings, FEMA 445, Federal Emergency Management Agency, Washington, D.C.
[3] Computers and Structures SAP2000: Three Dimensional Static and Dynamic Finite Element Analysis and
Design of Structures, Computers and Structures Inc., Berkeley, California, U.S.A.
[4] ASCE, 2000, Pre-standard and Commentary for the Seismic Rehabilitation of Buildings, FEMA 356 Report,
prepared by the American Society of Civil Engineers for the Federal Emergency Management Agency,
Washington, D.C.
579
Paper No.: 111
EFFECT OF STIFFENED CUTOUT ON VIBRATION

AND PARAMETRIC EXCITATION
1
A.K.L. Srivastava
1
Professor, Dept. of Civil Engineering, National Institute of Technology, Jamshedpur, India
Email: aklsriv.nitjsr@yahoo.com, aklsriv.civil@nitjsr.ac.in
ABSTRACT:
Stiffened plates are structural components consisting of plates reinforced by a system of ribs to enhance their
load carrying capacities. The stiffened plates are often subjected to dynamic in-plane loads of varying
magnitude and complexity. The dynamic instability characteristics of eccentrically stiffened plates with cutout
subjected to harmonic in-plane partial edge loadings at the plate boundaries are studied using finite element
method. The instability regions are determined for a wide range of excitation frequencies with different
boundary condition, partial edge loading at different locations, cutout ratios and various parameters of stiffeners
using Bolotin’s method and Hill’s infinite determinants by varying the number, size and location of the stiffeners.
The present paper deals dynamic instability analysis of eccentrically stiffened plates with cutout subjected to
harmonic in-plane partial edge load using Bolotin’s method and Hill’s infinite determinants. Finite element
formulation is applied to study the effects of different boundary conditions, aspect ratios, various parameters of
stiffened plates, cutout size, various partial edge loading position and extent on excitation frequency parameters
and principal instability regions.
KEYWORDS: Stiffened plates, cut outs, vibration, Dynamic stability, Parametric excitation.
1. INTRODUCTION
Stiffened plates are structural components consisting of plates reinforced by a system of ribs to enhance their load
carrying capacities. The stiffened plates are often subjected to dynamic in-plane loads of varying magnitude and
complexity. Cutouts in aerospace, civil, mechanical and marine structures are inevitable mainly for practical and
design considerations. The instability effects are improved with the provision of stiffeners. In the structural
modelling, the plate and the stiffeners are treated as separate elements where the compatibility between these two
types of elements is maintained. The dynamic instability of stiffened plates with cutout subjected to uniform and
non-uniform in-plane edge loadings are of considerable importance. The applied load is seldom uniform and the
boundary conditions may be completely arbitrary in practice. Stiffened plates subjected to dynamic in-plane
loading may undergo unstable transverse vibrations for certain combinations of the values of the load parameters.
Stiffened plates subjected to dynamic in-plane loading may undergo unstable transverse vibrations for certain
combinations of the values of the load parameters, i.e. the magnitude of the mean load, its amplitude and
frequency of the pulsating component of the load, and is called parametric instability. Cutouts in aerospace,
civil, mechanical and marine structures are inevitable mainly for practical and design considerations. Free edges
of the cutout are susceptible to instability behaviour due to local buckling. The dynamic instability of structures
with cutouts poses a tremendous challenge and must be properly understood in the design of such structures. The
instability effects are improved with the provision of stiffeners. The study of static and dynamic stability of
stiffened plate structures with or without cutout is of interest for the present work. In the present analysis, the plate
is modeled with the nine nodded isoparametric quadratic element with five degrees of freedom, where the
contributions of bending and membrane actions are taken into account. The formulation of the stiffener is done in
such a manner so that it may lie anywhere within a plate element. In order to maintain compatibility between plate
and stiffener, the interrelation functions used for the plate are used for the stiffeners also.
Theoretical investigations of the dynamic instability behaviour of stiffened plates with cutout, subjected to
harmonic in-plane concentrated and partial edge loading have been carried out using the finite element
formulation. The dynamic stability strength of stiffened plates with cutout is significantly affected by the static
components of load, with addition of restraint at the edges, with increase of number of stiffeners.
Stiffened plates with cutout are more pronounced in comparison to the unstiffened plates. The onset of
instability occurs with lower excitation frequencies for small cutout. The width of instability region increases
with the increase of cutout size for both simply supported and clamped edge conditions. Dynamic instability
behaviour under non-uniform compressive periodic edge loading shows that the plate is less susceptible to
instability under patch loading near one or both ends of the loaded edges.
Large number of references in the published literature deal with the buckling, vibration and dynamic instability
behaviour of rectangular plates subjected to in-plane uniform loading. The problem becomes complicated when
the loading is non-uniformly distributed over the edges. Parametric instability characteristic of rectangular plates
with localized damage subjected to in-plane periodic load is studied by Prabhakar and Datta [1]. Deolasi and
Datta [2] studied the dynamic stability of plates considering shear deformation and presented results of dynamic
stability of thin, square, isotropic plates for classical simply supported boundary conditions having three degrees
of freedom per node.The parametric instability characteristics of doubly curved panels subjected to various
in-plane static and periodic compressive edge loadings, including partial and concentrated edge loading are
studied by Sahu and Datta [3] using finite element analysis. The onset of parametric resonance of a rectangular
plate reinforced with closely spaced stiffener was studied by Duffield, Merrit and Willens. Duffield and
Willens [4] treated rectangular stiffened and Merrit and Willens [5] investigated the skew stiffened plates. In
these studies, however, a uniformly distributed load has been treated under relatively simple boundary condition.
The effects of stiffener location on the boundaries of the parametric instability regions have been discussed. It
was shown in their investigations that the principal regions of instability could significantly overlap for a
stiffened plate. The dynamic stability of radially stiffened annular plates with radial stiffeners located at equal
angular intervals with both edges subjected to in-plane forces using the finite element method have been studied
by Mermertas and Belek [6]. The instability regions are determined for a wide range of excitation frequency
with different boundary conditions using Bolotin’s method by varying the number of stiffeners.Dynamic
stability of laminated composite stiffened or unstiffened plates and shells due to periodic in-plane forces at
boundaries is studied by Liao and Cheng [7] using the method of multiple scales and finite element method. In
this the instability of laminated composite plates under uni-axial, harmonically varying in-plane loads are
analyzed. Both symmetric cross-ply laminates and antisymmetric angle ply laminates are analyzed.A finite
element analysis of a clamped thin plate with different cutout sizes, along with experiments was carried out by
Monahan et al. [8] using holographic interferometry. Ritchie and Rhodes [9] have investigated theoretically
and experimentally the behaviour of simply supported uniformly compressed rectangular plates with central
holes, using a combination of Rayleigh-Ritz and finite element methods. Paramsivam and Sridhar Rao [10]
developed finite difference method for obtaining the natural frequencies and mode shapes for rectangular plates,
of varying stiffnesses causing re-entrant corners, by assuming average curvature at the corners.
1.1. Description of the problem
The problem considered here consists of a rectangular plate (a x b) with longitudinal stiffeners having a central
rectangular cutout of size (g x d) subjected to harmonic in-plane partial edge loading at the plate boundary.
2. EQUATIONS
The equation of equilibrium for vibration and dynamic stability of a structure subjected to in-plane loads can be
written in matrix form as
[ ]
[ M ] { q&& } + [K b ] − P(t ) [ K G ] { q } = 0 (1)
[ M ] { q&& } + [ [K b ] − α Pcr [ K G ] − β Pcr [ K G ] cos Ω t ] { q } = 0 (2)
The element mass matrix can be expressed in iso parametric coordinate as
581
1 1
[M ] = ∫ ∫ [ N] [m ] [ N] J p dξ dη
T
p e p (3)
−1 −1
The geometric stiffness matrix can be expressed as
[ K G ] P = ∫ ∫ [ BG p ]T [ σ P ] [BG p ]
+1 +1
J p dξ d η (4)
−1 −1
Element stiffness matrix for stiffener
+1
[K S ] = ∫ [BS ] T [DS ] [ BS ] JS dξ , (5)
−1
The geometric stiffness of the stiffener element can be expressed in iso-parametric coordinate as:
+1
[ K G ]S = ∫ [B ] [ σ ] [B ] dξ
T
GS S GS JS (6)
−1
3. ILLUSTRATIONS, DIAGRAMS AND DISCUSSIONS
The simply supported stiffened square plate with two equispaced longitudinal stiffeners is analyzed by varying the
stiffener parameters for c/b=1 which corresponds to a fully loaded plate problem. The analysis is done by partial
load approach in order to ensure the validity of the formulation and programming. The ratio of cross-sectional area
of the stiffener to that of the plate ( δ = AS b t ) has been taken as 0.05 and 0.1. The ratio of the bending stiffness
( γ = E I S b D ) has been taken as 5 and 10. The influences of various parameters like effects of static and
dynamic load factors, cutout sizes, stiffener parameters, boundary conditions, position of loads and load width on
the dynamic instability behaviour of stiffened plates subjected to partial edge load at one end have been studied in
this section. The data used for its geometry are as follows: b = 600 mm, t =1mm, bS = 3.31mm, d S = 20.25mm,
ν = 0.34, E = 6.87 x 10 4 N/mm 2 , ρ =2.78 x 10 − 6 Kg/mm 3 .

The effect of cutout size on the instability region of the simply supported stiffened square plate at load position c/b
= 0.4 is studied from 0.0 (no cutout) to cutout ratio (ratio of size of cutout to length of plate) 0.8 at an interval of
0.2 in figure 1. The figure shows that the instability region and its width depend on the load bandwidth and cutout
size. It can be observed from figure 1 that the width of instability regions increases with the increase of cutout size
(g/a) and the onset of instability occurs with lower excitation frequencies for small cutout. With increase of cutout
size, the onset of excitation frequency increases having wider dynamic instability regions. The effect of load width
(c/b) on instability region of simply supported stiffened square plate having one central stiffener with cutout (g/a =
0.4) subjected to partial edge load at one end is studied taking static load factor ( α = 0.2) and the results obtained
are presented in figure 2. It is observed from figure 2 that for small value of c/b (c/b = 0.2), the width of the
dynamic instability regions is usually smaller in comparison to those for c/b close to unity (c/b = 0.8). This is
because of the fact that the restraint of the edge has a stabilizing effect on the dynamic instability behaviour of the
stiffened plate for small bandwidth of loading. It is also observed that, the instability occurs at lower excitation
frequencies with increase of distance from the edges (c/b).
582
120
Excitaion frequency parameter

80 g/a=0.2
g/a=0.4
g/a=0.6
g/a=0.8
Ω
40
0
0.0 0.2 0.4 0.6 0.8
Dynamic load factor (β )
Figure 1 Effect of cutout size (g/a) on dynamic instability region α = 0.2, c/b = 0.4.
45
Excitation frequency parameter
40
c/b=0.2
c/b=0.4
35
Ω
c/b=0.6
c/b=0.8
30
25
0.0 0.2 0.4 0.6 0.8
Dynamic load factor (β)
Figure 2 Effect of load width (c/b) on dynamic instability region. α = 0.2, g/a = 0.4, a/b = 1.
4. CONCLUSIONS
The parametric excitation behaviour for stiffened plates with cutout is more pronounced in comparison to the
unstiffened plates. The onset of instability occurs with lower excitation frequencies for small cutout. With
increase of cutout size, the onset of excitation frequency increases along higher frequency axis with wider
dynamic instability region. For partial edge load, the excitation frequency increases with increase of cutout size
for simply supported stiffened plate with one central stiffener. The width of instability region increases with the
increase of cutout size (g/a). Parametric excitation behaviour under non-uniform compressive periodic edge
loading shows that the plate is less susceptible to instability under patch loading near one or both ends of the
loaded edges. For small value of c/b (c/b = 0.2), the width of the dynamic instability regions is usually smaller
in comparison to those for c/b close to unity (c/b = 0.8).
583
REFERENCES
[1] Prabhakara, D.L., and Datta, P.K., (1994). Parametric instability characteristic of rectangular plates with
localized damage subjected to in-plane periodic load. Computer and Structures, 49 (1), 825-836.
[2] Deolasi, P. J. and Datta, P.K. (1995) Parametric instability characteristic of rectangular plate subjected to
localized edge loading (Compression or tension). Computer and Structure, 54 (1), 73-82.
[3] Sahu, S. K. and Datta, P.K. (2001). Parametric instability of doubly curved panels subjected to non-uniform
harmonic loading. Journal of Sound and Vibration. 240 (1), 117-129.
[4] Duffield, R. C. and Willens, N. (1972) Parametric resonance of rectangular plates. Journal of Applied
Mechanics, 39(2), 217-226.
[5] Merrit, R. G. and Willens, N. (1973) Parametric resonance of skew stiffened plates. Journal of Applied
Mechanics. 40 (1), 439-444.
[6] Mermertas, M. and Belek, H. T. (1991) Dynamic stability of radially stiffened annular plates. Computer
and Structures, 40(3), 651- 657.
[7] Liao, C. L and Cheng, C. R. (1994) Dynamic stability of stiffened laminated composite plates and shells
subjected to in-plane pulsating forces. Journal of Sound and Vibration, 174 (3), 335- 351.
[8] Monahan, I. J., Nemergut, P. J., and Maddux, G. E.(1970) Natural frequencies and mode shapes of plates
with interior cutouts’. Shock and Vibration Bulletin, 41(1), 37-49.
[9] Ritchie, D. and Rhodes, J. (1975) Buckling and post-buckling behaviour of plates with holes’. Aeronautical
Quarterly, 24 (3), 281-296.
[10] Paramsivam, P and Sridhar Rao, J. K., (1969) Free vibration of rectangular plates of abruptly varying
stiffnesses. Int J Mech Sci, 11 (2), 885-895.
584
Paper No.: 112
MODAL ANALYSIS OF DELAMINATED WOVEN FIBRE COMPOSITE

PLATES IN MOIST ENVIRONMENT
H.S.Panda1, S.K.Sahu1,*and P.K.Parhi2
1
Deptt. of Civil Engg., National Institute of Technology, Rourkela, Odisha-769008, INDIA
2
Deptt. of Civil Engg., College of Engineering and Technology, Bhubaneswar, Odisha, INDIA
Email: sksahu.nitrkl@gmail.com
ABSTRACT :
The present study deals with the modal analysis of delaminated Glass/Epoxy woven fibre composite plates in
moist environment. Based on first order shear deformation theory (FSDT), a composite plate model with
mid-plane delamination is chosen for vibration analysis by using finite element method (FEM). Accordingly,
computer codes are developed for calculation of free vibration analysis which takes into account the moisture
absorption effects of composite plates. The Glass/Epoxy composite plates are cast in laboratory using hand
lay-up method. During casting of the plates, artificial delamination is done by fixing Teflon film on the required
layer. A number of experiments are conducted to obtain natural frequencies of vibration for woven fibre
Glass/Epoxy moist composite plates using B&K FFT analyzer with PULSE Labshop software. The frequencies
of vibration obtained from experimental modal analysis match well with their numerical counterparts. The
results indicate that the vibration behaviour is greatly influenced by the delamination size, boundary conditions
and moisture content of the laminated composite plates. At higher moisture concentrations, complex stresses
develop in the plates and as a result, frequencies of vibration decrease as compared with the ambient conditions.
The size of delamination areas and different boundary conditions play critical role in lowering the frequencies of
vibration due to reduction in stiffness of the composite plate.
KEYWORDS: Free vibration, Moist environment, Delamination, Woven Fiber
1. INTRODUCTION
Laminated composites are becoming the preferred material system in a variety of industrial applications, such as
aerospace vehicles, automotive structural parts and civil engineering structures for design of members. The
increased specific strength, stiffness, toughness, mechanical damping, chemical and corrosion resistance in
comparison to conventional metallic materials with potential for structural tailoring are some of the factors that
have contributed to the advancement of laminated composites. The principal mode of failure of layered
composites is the separation along the interfaces of layers i.e. delamination. Laminated composite structures
are often exposed to moist environments. As the presence of delamination in composite laminates adversely
affects the safety and durability of structures, a comprehensive understanding of the delamination behavior of
composites subjected to moist environment is of fundamental importance in the assessment of structural
performance of laminated composites.
Plenty of studies are available on vibration of laminated composite plates under ambient temperature and
moisture conditions by using different theories [1-2]. Due to its practical applications, researchers have shown
significant interest in determining the natural frequencies of delaminated composite plates. A review of earlier
works on vibration of delaminated composite plates was reported by Della and Shu [3] through 2007. Ju et al.[4]
used a finite element model based on Mindlin’s plate theory for study of free vibration of composite plates with
multiple delaminations. The effects of boundary conditions, delamination size, delamination location and the
number of delaminations on natural frequencies and mode shapes of delaminated composite plates were
investigated. Hou et al. [5] studied the effect of delaminations on natural frequencies of thin composite plates by
conducting free vibration tests. Alnefaie [6] presented 3D finite element models for calculation of natural
frequencies of delaminated composite plates. Shiau et al. [7] investigated the effects of delamination on
frequencies of vibration of simply supported rectangular plate having through width delamination by finite strip
method.
The previous studies on the bending, vibration and buckling of moderately thick plates associated with elevated
temperatures are reviewed by Tauchert [8] through 1991. Sairam and Sinha [9] reported the effects of moisture
and temperature on the natural frequencies of laminated composite plates by FEM for symmetric and
anti-symmetric laminates with simply supported and clamped boundary conditions. Patel et al. [10] presented
the static and dynamic characteristics of composites considering the effects of temperature and moisture on the
natural frequencies, deflection and buckling using a higher order finite element theory. All the above mentioned
works showed the effects of either delamination or hygrothermal effects on free vibration behavior of laminated
composites. Parhi et al. [11] studied the vibration of delaminated composite plates subjected to hygrothermal
environment using FEM. Nanda et al. [12] investigated on the effects of delaminations on the transient response
of composite panels subjected to hygrothermal environment.
To the best of authors’ knowledge, no experimental work has been reported so far in open literature on vibration
of delaminated composite plates subjected to moist environment. The present study deals with combined
numerical and experimental investigation on free vibration behavior of industry driven woven fiber delaminated
composite plates under elevated moisture concentrations.
The constitutive equations for the plate when it is subjected to moisture, are given by
{F }= [D]{ε }− {F N } (1)
Where, {F }= { N x , N y , N xy , M x , M y , M xy , Q x , Q y }, T
The non-mechanical loads due to moist conditions are,
{F }= {
N
N x , N y , N xy , M x , M y , M xy , 0, 0 },
N N N N N N T
{ε }= { ε 0x , ε 0y , γ 0xy , K x , K y , K xy , ϕ x , ϕ y },
T
[D]= Standard stress strain matrix under ambient conditions [13]

The terms with superscript (0) correspond to mid-plane values. N x , N y , N xy are in-plane stress resultants per
unit length, M x , M y , M xy are moment resultants per unit length and Q x , Q y are transverse shear stress
resultants. The terms with superscript (N) represents the corresponding non-mechanical in-plane stress and
moment resultants due to moisture. ϕ x , ϕ y are shear rotations in x-z and y-z planes respectively. ε 0x , ε 0y , γ xy
0
are in-plane strains of the mid plane and K x , K y , K xy are the curvatures of the plate. The non-mechanical
( ) (e) dz ,
in-plane stress and moment resultants due to moist environment are given by,
{N , N Ny , N Nxy} = ∑ ∫ k Qij
n
N T z
x k
k =1 zk −1 k
{M , M , M } = ∑ ∫z (Qij) (e) z dz ,
n
N
x
N
y
N T
xy
k
k i, j = 1, 2, 6 (2)
z k =1 k −1 k
{
Here, {e}= ex , e y , exy } T
k are the non-mechanical strains of the kth lamina, oriented at an arbitrary angle θk ,
expressed as
{e}k = [T ]{β1 , β 2 }Tk (C − C0 ) (3)
586
 m2 n 
2
 2 
Where, [T ] =  n2 m  , m = cos θk and n = sin θk .
− 2mn 2mn 
β1, β2 are the moisture coefficients of the lamina in longitudinal and lateral directions and their values
considered here are 0 and 0.44 respectively. C0 is the reference moisture content in % and here its value is
considered as 0%. C is increased moisture concentration.
2.1. Finite element formulation

An eight-noded isoparametric element is chosen for the present free vibration problem. Five degrees of freedom
considered at each node are u0, v0, w, θx and θy.
2.2 Delamination modeling
A typical composite laminate having ‘p’ number of delaminations is considered for finite element programming.
The displacement field within any sub-laminate‘t’ is denoted by
u = u + ( z − zt ) θ , v = v + ( z − zt ) θ (4)
0 0 0 0
t t x t t y
Here, u0t , vt0 are the mid plane displacements of the tth
sub-laminate and z 0t , the distance between the mid plane of the original laminate and the mid plane of the tth
sub-laminate.
The stress and moment resultants of the sub-laminate are derived which leads to the elasticity matrix of
th
the t sub-laminate in the form
 Aij z t Aij + Bij 0
0
[D ] t =  Bij zt Bij + Dij

0
0

(5)
 0 0 S ij 
Where, [Aij ] t = ∫
h t / 2 + z t0
− ht / 2 + z 0t
[Q ]dz , [B ] = ∫
ij ij t
ht / 2 + z t0
− ht / 2 + z t0
[Q ]( z − z ) dz ,
ij
0
t
[D ] = ∫
ij t
ht / 2 + z t0
− ht / 2 + z t0
[Q ]( z − z ) dz , i, j = 1,2,6 , [S ] = ∫
ij
0 2
t ij t
ht / 2 + zt0
− ht / 2 + zt0
[Q ]dz , i, j = 4,5
ij
Here, ht is the thickness of tth sub-laminate.

The stiffness matrix, the initial stress stiffness matrix, the mass matrix and the load vectors of the element are
evaluated by first expressing the integrals in the local natural co-ordinates, ξ and η of the element and then
performing numerical integration by Gaussian quadrature. The element matrices are assembled to obtain the
respective global matrices [K], [Kσ], [M],{P} and {PN}. The second part of the solution involves determination
of natural frequencies from the condition
[[K ] + [K σ ]]− ω 2 [M ] = 0 (6)
ω is the natural frequency. Eq. (12) is a generalized eigenvalue problem Based on the above FEM
delamination modelling, codes are developed to generate numerical results (FEM) for vibration response of
delaminated woven fibre Glass/Epoxy composite plates exposed to moist environment.
3. EXPERIMENTAL PROGRAMME
Woven glass fiber and Epoxy composite plate specimens were fabricated using weight fraction of 55:45 by hand
layup method. Woven roving Glass fibers (WR 360/100, Owens Corning – 360 grams per square meter) were
used for testing. For the present numerical and experimental study, composite laminates of 16 layers of fibers in
587
balanced form as per ASTM [14] specifications, were chosen. For preparation of Epoxy resin matrix, Hardener
8% (Ciba-Geigy, Araldite HY556 and Hardener HY951) of the weight of Epoxy was used. The laminates were
cured at normal temperature (250C and 55% Relative Humidity) under a pressure of 0.2 MPa for 3 days. After
proper curing of laminates, the release sheets were detached. The specimens were cut for vibration testing into
235mm x 235mm size. The average thickness of 16 layered laminate was measured was found to be 6mm.
Teflon film was embedded (Fig. 1) in the laminate simulating the presence of delamination. Accordingly, Square
size Teflon films were inserted as per the mesh divisions in three different areas such as 6.25%, 25% and 56.25%
of the total area of the laminate. Complete care was taken in casting of the plates with presence of delamination.
Fig.1. Casting of composite plate with Teflon Fig.2. FFT analyzer with set up
film to introduce delamination
The composite specimens were conditioned in temperature/humidity chamber separately before testing to know
the effects of elevated moisture concentrations on natural frequencies of vibration. The dried samples were kept
in humidity chamber maintaining specified temperatures and relative humidity as per requirement for moisture
absorption. To record moisture uptake, the samples were periodically removed from the temperature/humidity
chamber, allowed to cool in a sealed, humid environment; weighed; and then quickly kept back in the chamber,
if the required moisture content was not gained. The moisture contents used for present investigation were
0.25%, 0.5% and 0.75%.
.
3.1 Determination of material constants
The material properties were evaluated experimentally by performing unidirectional tensile tests on specimens
cut in longitudinal and transverse directions and at 450 to the longitudinal directions as described in ASTM
Standard [15]. For determination of Poisson’s ratio (υ12), a different set of specimens were used, along with a
pair of strain gages bonded over the gage length of the sample in two mutually perpendicular directions. The
strains were recorded simultaneously in longitudinal and transverse directions with the help of two strain
indicators. The ratio of transverse to longitudinal strain directly gives the Poisson’s ratio within the elastic range.
The shear modulus (G12) was determined similarly with 450 coupons using the formula.
3.2 Free vibration test

The composite test specimens were fitted properly to the prefabricated iron frame with provision for different
boundary conditions. The connections of FFT Analyzer (Model Bruel and Kjaer, Denmark- B&K 3560 C), laptop,
B&K 4507 transducers, B&K 2302-5 modal impact hammer and cables to the system were done (Fig. 2). The
PULSE Lab Shop Software was used during the vibration measurement. The composite plate was excited at
selected points by means of modal impact hammer and the resulting vibrations of the specimens were picked up
by the accelerometer (B&K 4507). The signal was then subsequently fed to the analyzer, where its frequency
spectrum was also obtained using the PULSE software. Various forms of Frequency response Functions (FRF)
were directly measured and the coherence is observed for each set of measurement. A typical FRF (pulse report)
of the measurement from FFT analyzer is shown in Fig.3. As shown in Fig.3, the different peaks of FRF shows the
different modes of vibrations.
588
Fig.3. Typical FRF of test specimen.

Experimental results of frequencies of vibration for [0]16 woven fiber Glass/Epoxy composite plates are obtained
with increased moisture concentrations for different delamination areas under simply supported (S-S-S-S)
boundary condition. In the present dynamic analysis, moisture concentration is varied from 0% to 0.75%. For
computing the results, centrally located single mid plane delamination of three different sizes like 6.25%, 25%
and 56.25% of the total plate area is considered for composite plates. Figure 4 shows the degree of freedoms per
node considered in S-S-S-S boundary condition.
Fig.4. The degree of freedoms for S-S-S-S boundary condition
4.1 Numerical and Experimental Results
Numerical (FEM) and experimental (expt.) results of fundamental frequencies of vibration for [0]16 woven fiber
Glass/Epoxy composite plates are plotted with moisture concentrations (%) for different delamination areas for
S-S-S-S boundary condition. The geometry and material properties of the composite plates at elevated moisture
concentrations used in this study are shown in Tables 1.
Table 1 Elastic moduli of woven fiber Glass/Epoxy lamina at different moisture concentrations, a = 235mm, b =
235mm, h = 6mm, E1 = E2, G12 = G13 = G23, ‘ρ’ = 1660 Kg/m3, ‘ν 12 ’ = 0.43.
Elastic moduli Moisture concentration, C (%)
0.0 0.25 0.5 0.75
E1(GPa) 15.4 15.38 14.69 13.17
G12(GPa) 3.56 3.53 3.49 3.47
4.1.1 Variation of frequencies of vibration with increased moisture concentrations

FEM and experimental results of lowest four mode frequencies of free vibration for woven fiber [0]16
Glass /Epoxy composite plates are shown in figure 5-8 against moisture concentrations (%) for different
delamination areas with S-S-S-S boundary condition.
589
350 fem-0% mc
Frequency in Hz.
250 expt-0% mc
150 fem-0.25% mc
50 expt-0.25% mc
0 20 40 60 fem-0.5% mc
Delamination area (%) expt-0.5% mc
Fig.5. Effect of moisture content on first mode frequencies of free vibration for woven fiber composite plates
with different delamination areas under S-S-S-S boundary condition.
Figure 5 shows an excellent agreement between the predicted and experimental values of frequencies of
vibration for woven fiber composite plates with different delamination areas under S-S-S-S boundary condition.
At 0% moisture content, 12.73%, 29% and 39.5% decrease in frequencies for 6.25%, 25% and 56.25%
delamination areas of the composite plates respectively from that of the intact plate. At 0.75% moisture content,
21%, 52% and 77% decay in frequencies for 6.25%, 25% and 56.25% delamination areas of the composite
plates respectively from that of the intact plate. Between 0% and 0.75% moisture contents, 18%, 25%, 44% and
70% decay in frequencies for 0%, 6.25%, 25% and 56,25% delaminated plates. The reduction in frequencies is
observed as the plates become soften with moisture concentrations and thereby reduction in the stiffness.
800 fem-0% mc
Frequency in Hz.
700
600 expt-0% mc
500 fem-0.25% mc
400
300 expt-0.25% mc
0 10 20 30 40 50 60 fem-0.5% mc
Fig.6. Effect of moisture content on second mode frequencies of free vibration for woven fiber composite plates
As evident from figure 6, delamination plays a vital role in degrading the frequencies of vibration than the
moisture content of the composite plates. There is a decay of frequencies in the range 8.2%, 9.3%, 18.6% and 29%
for the delaminated plates having 0%, 6.25%, 25% and 556.25% delamination areas respectively between 0%
and 0.75% moisture range.
800 fem-0% mc
Frequency in Hz.
700
600 expt-0% mc
500 fem-0.25% mc
400
300 expt-0.25% mc
0 20 40 60 fem-0.5% mc
Fig.7. Effect of moisture content on third mode frequencies of free vibration for woven fiber composite plates
590
The third mode frequencies of the S-S-S-S woven fiber plates against different delamination areas are shown for
different moisture concentrations in figure 7. Here, as marked from the figure, the delamination becomes
predominant in decaying the frequencies with increase in moisture concentration. Between 0% and 56.25%
delaminations, 42%, 45%, 47.5% and 49.6% degradation in frequencies is marked for moisture uptakes of 0%,
0.25%, 0.5% and 0.75% respectively.
fem-0% mc
Frequency in Hz.
1200
1000 expt-0% mc
800 fem-0.25% mc
600 expt-0.25% mc
0 20 40 60 fem-0.5% mc
Fig.8. Effect of moisture content on fourth mode frequencies of free vibration for woven fiber composite plates
Fourth lowest mode frequencies against the different delamination areas are plotted in figure 8 for various
moisture concentrations. The FEM results match well with their experimental counterparts as obvious from the
figure. As the delamination percentage is increased from 6.25%, sudden decay in frequencies is observed for all
moisture uptakes of the composite plates. Between 0% and 0.75% moisture uptake, the composite plates having
0%, 6.25%, 25% and 56.25% delaminations suffer a decay in frequencies by 5.6%, 5.8%, 9% and 15.6%
respectively because of reduction in stiffness.
5. CONCLUSIONS
Based on FSDT, a dynamic finite element analysis of single mid plane delaminated square composite plates
exposed to moist environment is investigated. The results of natural frequencies of vibration obtained from the
above formulation are also experimentally compared The broad conclusions that can be made from the above
study are as follows:
• The natural frequencies of free vibration of composite plates decrease with increase in uniform moisture
content due to reduction of stiffness for all laminates.
• At higher modes, the decrease of frequency of vibration is more than the fundamental mode.
• The delamination of composite laminates plays an important role in degrading the natural frequencies of the
laminates and in moist environment it offers a challenge to the designer of composite structures.
This property can be utilized to tailor the design of woven fiber delaminated composite plates in moist
environment. The vibration results can be used as a technique for structural health monitoring or testing of
structural integrity, performance and safety.
REFERENCES
[1] Bert, C.W. and Chen, T. L. C. (1978).Effect of shear deformation on vibration of anti-symmetric angle ply
laminated rectangular plates, International Journal of Solids and Structures, 14, 465-473.
[2] Reddy, J. N. (1979).Free vibration of anti-symmetric angle ply laminated plates including transverse shear
deformation theory by finite element method, Journal of Sound and Vibration, 66, 565-576.
[3] Della, C.N. and Shu, D. (2007). Vibration of delaminated composite plates: a review. Applied Mechanics
Reviews, 60, 1-20.
591
[4] Ju, F.; Lee, H.P. and Lee, K.H. (1995). Finite element analysis of free vibration of delaminated composite
plates. Composites Engineering, 5(2), 195-209.
[5] Hou, J.P. and Jerominidis, G.(1999). Vibration of delaminated thin composite plates. Composites Part-A, 30,
989-995.
[6] Alnefaie, K.(2009) Finite Element modeling of composite plates with internal delamination. Composite
[7] Shiau, L.C. and Zeng, J.Y. (2010). Free vibration of rectangular plate with delamination. Journal of
Mechanics, 26(1), 87-93.
[8] Tauchert, T.R. (1991). Thermally induced flexure, buckling and vibration of plates. Applied Mechanics
Review, 44(8), 347-360.
[9] Sairam, K.S. and Sinha, P.K. (1992). Hygrothermal effects on the free vibration of laminated composite
plates. Journal of Sound and Vibration, 158(1), 133-148.
[10] Patel, B.P.; Ganapati, M. and Makhecha, D.P.(2000). Hygrothermal effect on the structural behavior of
thick composite laminates using higher order theory. Composite Structures, 56, 25-34.
[11] Parhi, P.K.; Bhattacharya, S.K. and Sinha, P.K. (2001). Hygrothermal effects on the dynamic behavior of
multiple delaminated composite plates and shells. Journal of Sound and Vibration, 248(2), 195-214.
[12] Nanda, N.; Sahu, S.K. and Bandyopadhyay, J. (2010). Effect of delamination on the non-linear transient
response of composite shells in hygrothermal environment. Journal of Structural Integrity, 1(3), 259-279.
[13] Jones, R.M.( 1975). Mechanics of Composite Materials. Mc Graw-Hill: New York.
[14] ASTM Standard: D5687/D5687M-07, 2007: Standard Guide for Preparation of Flat Composite Panels with
Processing Guidelines for Specimen Preparation.
[15] ASTM Standard: D3039/D3039M-08, 2008: Standard Test Method for Tensile Properties of Polymer
Matrix Composite Materials.
592
Paper No.: 117
SEISMIC ANALYSIS OF REINFORCED CONCRETE VENTILATION STACK

USING SIMPLIFIED MODAL ANALYSIS TECHNIQUE
Abhishek Kumar1, V.S.Phanikanth2 and K. Srinivas3
Scientific Officer(C), Civil Engineering Division (Vizag),Bhabha Atomic Research Centre

Scientific Officer (G), Civil Engineering Division, Bhabha Atomic Research Centre,Trombay,Mumbai-85
Head, Civil Engineering Division, Bhabha Atomic Research Centre, Trombay, Mumbai-85
ABSTRACT:
Wind and seismic forces form the major external loading that can affect tall structures like chimneys. They are
different from that of a normal low rise structure. As the time period of a structure is long, the acceleration
experienced by it is comparatively smaller whereas the displacements are high. In the present paper tapered
chimney of height 100m above ground level and resting on rock has been analyzed for seismic forces.
Simplified model analysis technique is used to evaluate the response under seismic loading in addition to the
detailed dynamic analysis based on FE model as is the practice in many of the design offices. From the detailed
comparison of the methods it is observed that the simplified model analysis technique shows good agreement in
results compared with detailed FE model. Also their results are discussed in detail.
KEYWORDS: Modal analysis, Response spectrum, Simplified model analysis technique
1. INTRODUCTION:
Reinforced concrete ventilation stacks are required to be designed for seismic loads, wind loads in addition to
the self-weight of structure. The seismic design shall include detailed dynamic analysis using site specific
response spectrum for important structures like stacks. The response of a structure to ground vibrations is a
function of the nature of foundation soil, materials and the duration and characteristics of ground motion. In the
present paper it is proposed to evaluate the dynamic response of a reinforced concrete chimney using simplified
modal analysis technique. Also the chimney is analysed using detailed dynamic analysis using response
spectrum method. It is observed that the simplified modal analysis technique is simple, yet accurate to evaluate
the dynamic forces and hence can be readily used by the designers.
2. MATHEMATICAL BACKGROUND
The response of a chimney to earthquake forces is computed by modal analysis techniques. This requires that
natural modes of vibration be obtained, from which the response of the chimney in different modes is computed.
The total response of the chimney is then obtained by the instantaneous combination of the responses in the
different modes.
2.1 Response of Chimney to Earthquakes
If a chimney is subjected to base acceleration ‘a’, then the basic differential equation (ignor ing shear
deformations and rotary inertia effects) with zero damping is given by, Wadi S. Rumman (1985)
∂2  ∂2 y   ∂2 y 
 EI  + m 2 + a  = 0 (1)
∂x 2  ∂x
2
 ∂t 
Where
EI = Flexural rigidity of the beam; m is mass per unit length; y(x,t) are the displacements in the chimney at a
distance x at time t; x = distance along the chimney and t is time
3. SEISMIC ANALYSIS USING SIMPLIFIED MODAL ANALYSIS TECHNIQUE
A Simplified modal analysis technique3 based on energy method and Rayleigh’s principle is used in the present
study for obtaining the dynamic response of tapered chimney. The technique besides its simplicity in estimating
the dynamic response of the structure is also accurate compared to detailed dynamic FE analysis methods. The
solution steps using this method are briefly given below.
The chimney is divided into suitable segments and the mass for each segment is evaluated. The deflection
profile is initially assumed and the inertia forces are computed. Applying these inertia forces, to the structure,
the deflections are computed and are compared to the initially assumed deflection profiles. If found different,
the obtained deflection profile is used to rework inertia forces and the improved deflection profile is computed.
The process is repeated to obtain improved estimations of deflection profile until assumed and calculated
profiles match to any desired degree of accuracy.
For the mode shape so obtained, the fundamental vibration frequency is determined. To obtain the next higher
mode, assume a deflection profile and the mode coefficient and using this, purge the assumed profile of the
earlier modes. With the purified profile, proceed as indicated for the first mode. Iteration will converge on the
next higher mode. Obtain further mode shapes in a similar manner. Using mode shapes the participation factors
for each mode are determined and then modal shears and moments at each mass level can be evaluated. Finally
the modal values are combined to arrive at design values. Some of the concepts inherent in the above steps are
explained below.
3.1 Inertia Forces
Consider a chimney made up of lumped masses (Mr). Assume a deflection profile Φi for the ith mode. Lateral
deflection of mass Mr would be AΦri for mode i, where, A is an arbitrary constant.
The inertia force (Fri) is: Fri = Mr* ωi2 (A Φri ) = (constant)*(Mr Φri), where, ωi = circular frequency
It is not possible to determine the absolute value of Fri however this may not be of any importance. Hence, we
can approximate inertial force as Fri = Mr Φri
3.2 Frequency Determination
Under the effect of forces Fri the characteristic deflection profile of the structure is say Φi. A structure is
assumed to undergo harmonic motion about its original position (Φi = 0) with a maximum amplitude Φi in the ith
mode. Using Rayleigh’s principle, we equate the kinetic energy of the system (masses Mr) at Φi=0 to its
potential energy at amplitude A’Φi’, where A’ is an arbitrary constant, and we get
∑ M ω ∗ A Φ = ∑ F A′ Φ (2)
∑
ωi =
’ ∑
(3)
3.3 Mode Coefficient
Any assumed deflection profile can be built up by a combination of modal shapes and represented
Φra = ∑! ψ Φ
mathematically as:
Where, Φra = assumed deflection of mass Mr, N = number of modes.
594
Multiplying both sides of the above equation by Mr Φrq (where Φrq is the characteristic deflection profile of the
qth mode) and on summation for all masses,
∑ M Φ " Φ # = ∑ . ∑! M Φ#ψ Φ
∑
Utilizing the orthogonality condition, we get, ψ# =
& '
∑
(4)
'
3.4 Higher Modes
If deflection profile for two modes a and b are known and Φc is the deflection profile for next higher mode, then
To obtain the next higher mode, the assumed deflection profile for the higher mode is purified of earlier modes.
Φrap = Φc - ψa Φa - ψb Φb
purified profile will be
(5)
This purified shape is now taken as the assumed profile and further analysis is similar to that explained for the
first mode.
3.5 Participation Factor
∑
∑
Participation of a particular mode in the response is given by a factor Pi such that Pi =
(Pi = participation factor for mode i for a structure with j masses). The modal masses and modal participating
mass ratio’s also can be obtained once the mode shape of the corresponding mode is evaluated1.
3.6 Design horizontal Seismic Coefficient
The design horizontal seismic coefficient Ah is determined using the following expression (IS 1893: 2002):
Z I Sa
Ah = × × (6)
2 R g
Where, Z = Zone factor; I = Importance factor; R = Response reduction factor and Sa/g = Average response
acceleration coefficient.
3.7 Modal Shear and Moment
Modified acceleration is given by:

) *
Sai (mod) = ( × + × S" - × P (7)
The inertia force at each mass level is given as: Fri = Mr Φri Sai(mod)
Shear force and bending moment at each mass level are given by
V = 0∑1
2 V and M = 0∑1
2 M (8)
4. ANALYSIS OF TAPERED CHIMNEY USING SIMPLIFIED MODAL ANALYSIS TECHNIQUE
In the present study an attempt has been made to evaluate the dynamic response of the chimney using simplified
modal analysis technique for a tapered chimney of height 100m above ground level. The input parameters
considered are presented in Table 1.
595
Table 1: Input Parameters
Total height above Raft (base) 103.8 m

Uniform c/s above Raft up to height 3.8 m
Top inner diameter(ID) 3.0 m
Top outer diameter(OD) 3.4 m
Thickness at top 0.2 m
Bottom ID 6.2 m
Bottom OD 7.4 m
Thickness at bottom 0.6 m
Table 2: Geometric Properties at various heights
Height above the base, m ID, m Thickness, m OD, m

0.0 6.20 0.60 7.4
3.8 6.20 0.60 7.4
23.8 5.56 0.52 6.6
43.8 4.92 0.44 5.8
63.8 4.28 0.36 5.0
83.8 3.64 0.28 4.2
103.8 3.00 0.20 3.4
The material properties considered are unit weight of chimney = 25 kN/m3 and Young’s modulus of elasticity Ec
= 3.4×107 kN/m2. The geometric properties at various heights are presented in Table 2.The structure is assumed
to be in seismic zone III as per IS1893-2002. The importance factor is considered as 2.0 and the response
reduction factor is taken as 3.0 in this study. The structure is assumed to be resting on rock and the material
damping is considered as 5% for evaluating the seismic forces.
4.1 Deflection Coefficients
The diameter of the stack is varying linearly along the height and the cross sectional area of stack and Moment
of Inertia at each section are calculated based on the average values of sectional properties for each member.
The height of the chimney is divided into six sections i.e. seven nodes numbered from 1 at the top to 7 at the
base. The mass at node 6 which is at 3.8m just above the base, is assumed not to have significant contribution in
the analysis and thus is ignored in the analysis. Lumping of masses at various nodes are summarised in the table
3. The simplified modal analysis technique requires the deflections to be evaluated for a given set of inertial
loads and for computing this it is better practice to generate the deflection coefficients due to unit load placed at
various node positions. Accordingly the deflection coefficients due to a unit load (1kN) placed at nodes 1 to 5
are presented in table 4.
596
Table 3: Lumped Mass Model Parameters
height avg avg avg avg c/s

ID thickness OD nodal mass
above base node section ID t OD area
m m m kN.s2/m
m m m m m2
6.2
0.0 7 0.60 7.4 0.0000
0
6.2 6.2 0.6 12.817

3.8 6 0.60 7.4 6-7 7.4 615.631
0 0 0 7
5.5 5.8 0.5 11.329

23.8 5 0.52 6.6 5-6 7.0 508.754
6 8 6 8
4.9 5.2 0.4

43.8 4 0.44 5.8 4-5 6.2 8.6255 380.092
2 4 8
4.2 4.6 0.4

63.8 3 0.36 5.0 3-4 5.4 6.2832 269.883
8 0 0
3.6 3.9 0.3

83.8 2 0.28 4.2 2-3 4.6 4.3027 178.128
4 6 2
3.0 3.3 0.2

103.8 1 0.20 3.4 1-2 3.8 2.6842 68.432
0 2 4
Table 4: Deflection coefficients due to unit load (1kN)
Deflection Coefficient
deflection at
Load at
1 2 3 4 5
1 0.33 0.201 0.107 0.046 0.013
2 0.201 0.138 0.079 0.036 0.01
3 0.107 0.079 0.051 0.025 0.007622
4 0.046 0.036 0.025 0.015 0.004978
5 0.013 0.01 0.007622 0.004978 0.002335
Deflection=Coefficient×10-3 m
597
Table 5: FIRST MODE COMPUTATION
First cycle
Deflection,
Load at Mr Φra Fr (kN) Φ'r
×10-3 m
kN.s2/m = MrΦra
1 68.432 1.000 68.432 71.571 1.000
2 178.128 0.800 142.503 48.694 0.680
3 269.883 0.500 134.942 28.700 0.401
4 380.092 0.300 114.028 13.615 0.190
5 508.754 0.100 50.875 4.030 0.056
Second Cycle
Deflection,
Load at Mr Φra Fr Φ'r
×10-3 m
kN.s2/m = MrΦra
1 68.432 1.000 68.432 61.620 1.000
2 178.128 0.680 121.192 41.482 0.673
3 269.883 0.401 108.225 24.166 0.392
4 380.092 0.190 72.306 11.304 0.183
5 508.754 0.056 28.644 3.310 0.054
With these coefficients and with the simplified modal analysis technique described above the deflection profiles
are evaluated and the first mode computation is presented in Table 5. It was observed that reasonable
convergence in mode shape estimation was achieved in the third cycle and hence higher cycle computations are
not included in this table. Using this mode shape, the modal characteristics like frequency, participation factor
etc., are computed and the inertia forces for different nodes for first mode are presented in Table 6.Similarly the
deflection profile is evaluated for the second mode and is presented in Table 7, and it is found that the mode
shape convergence is achieved in the fourth cycle. The converged mode shape and the modal characteristics
such as participation factor, frequency, time period, modal mass etc. are presented in Table 8. The inertia forces
and base moments for different nodes for the second mode are also computed and presented in Table 9.
Table 6: Inertia forces, moments at different nodes and Modal Characteristics - First mode
Node Mass Φ1 Sa1 Inertia Force H Moment
kN.s2 /m m/s2 kN m kN-m
1 68.432 1.000 0.6404 43.824 103.8 4548.93
2 178.128 0.673 0.6404 76.771 83.8 6433.41
3 269.883 0.392 0.6404 67.751 63.8 4322.52
4 380.092 0.183 0.6404 44.544 43.8 1951.03
5 508.754 0.054 0.6404 17.594 23.8 418.73
598
Dynamic Analysis Parameters
∑6
ω1 =0
’ ∑6
= 4.03 rad/s; T1 = 1.5591 s and Sa1/g = 0.6414
∑6
P1 = ∑6 = 1.909; Sa1(mod) = 0.6404 m/s2
Modal mass, M1 = 746884 kg; Modal participating mass ratio = 0.369
Base shear = 250.483kN; Base moment = 17674.62 kN-m
Table 7: SECOND MODE COMPUTATION
First Cycle
Load Deflection,
Mr Φr1 Φra MrΦr1Φra ψ1Φr1 Φrap = Fr = Φr
at ×10-3 m
Φra-
kN.s2 /m MrΦrap
ψ1Φr1
1 68.432 1.000 1.0000 68.432 0.019 0.9810 67.132 4.648 1.0000
2 178.12 0.673 0.1000 11.988 0.012787 0.0872 15.535 0.152 0.0326
3 269.88 0.392 -0.5000 -52.897 0.007448 -0.5074 -136.952 -1.850 -0.3979
4 380.09 0.183 -0.3000 -20.867 0.003477 -0.3035 -115.349 -1.763 -0.3792
5 508.75 0.054 -0.1000 -2.747 0.001026 -0.1010 -51.397 -0.710 -0.1528
∑= 3.909
ψ1 = 3.909/204.897 = 0.019
Second Cycle
Load Deflection,
Mr Φr1 Φra MrΦr1Φra ψ1Φr1 Φrap = Fr = Φr
at ×10-3 m
Φra-
kN.s2 /m MrΦrap
ψ1Φr1
1 68.432 1.000 1.0000 68.4320 -0.0016 1.0016 68.542 4.718 1.0000
2 178.12 0.673 0.0326 3.91053 -0.0011 0.0337 6.004 0.173 0.0367
5 269.88 0.392 -0.3979 -42.0998 -0.0006 -0.3973 -107.227 -1.853 -0.3927
4 380.09 0.183 -0.3792 -26.3764 -0.0003 -0.3789 -144.021 -1.859 -0.3940
5 508.75 0.054 -0.1528 -4.1966 -0.0001 -0.1527 -77.670 -0.764 -0.1620
∑= -0.3302
ψ1 = -0.00161
Table 8: Modal Characteristics- Second mode
Φ'r FrΦ'r MrΦ'r Mr (Φ'r)2

1.0000 68.6221 68.4319 68.4319
0.0391 0.2818 6.958017 0.271793
-0.3926 41.4784 -105.957 41.59911
-0.3974 59.8502 -151.034 60.01547
-0.1643 13.6943 -83.6087 13.74024
∑ 183.9269 -265.21 184.0585
Dynamic Analysis Parameters
599
∑6
ω2 =0
’ ∑6
=14.46 rad/s; T2= 0.4345s and Sa2/g = 2.302;
∑
=-1.440; Sa2(mod) = -1.7347 m/s2
∑
P2 =
Modal mass, M2 = 382140 kg; Modal participating mass ratio = 0.189
Base shear = 460.03kN; Base moment =13319.78 kN-m
Table 9: Inertia forces and moments at different nodes for second model
Node Mass Φ2 Sa2 Inertia Force H Moment

kN.s2/m m/s2 kN m kN-m
1 68.432 1.0000 -1.7347 -118.7089 103.8 -12322
2 178.128 0.0391 -1.7347 -12.0818 83.8 -1012.46
3 269.883 -0.3926 -1.7347 183.8020 63.8 11726.57
4 380.092 -0.3974 -1.7347 262.0239 43.8 11476.65
5 508.754 -0.1643 -1.7347 145.0006 23.8 3451.014
The base shear and moment for the two modes are combined and are given below:
Base shear = 7 250.483 + 460.0357 = 523.81 kN and
Base moment = 7 17674.62 + 13319.78 = 22131.6 kN-m
5. MODAL ANALYSIS RESULTS USING FE MODEL (BEAM ELEMENT)
The tapered chimney is also analysed using finite element method (FEM) using beam3d element. Commercial
finite element software SAP2000 is a comprehensive and integrated finite element analysis and design tool,
including a state-of-the-art user interface, visualization tools, and international design codes. It is capable of
analyzing any structure subjected to static loading, dynamic loads, soil-structure interaction aspects, wind load
analysis, earthquake analysis, and moving loads etc. In the present study the detailed FE analysis using FE
model with the material parameters mentioned above and with input parameters given in Table 1 is also carried
out and their results are presented. The frequencies obtained are presented in Table 10 for different modes. The
periods of vibration for the first and third modes in X direction are 1.56s and 0.436s respectively. The period for
the first mode computed using simplified modal analysis technique was found to be 1.559s and for the second
mode it is observed as 0.434s (beam2d problem). It can be seen that excellent agreement in results is obtained
using the simplified modal analysis technique.
Table 10: Modal frequencies using FE model (beam 3d element)
Modal Periods And Frequencies

Mode Period(Sec) Frequency(cyc/sec) Circ Freq( rad/sec) Eigen value( rad2/sec2)
1,2 1.56 0.641 4.0277 16.223
3,4 0.436 2.294 14.416 207.83
5,6 0.196 5.104 32.072 1028.6
7,8 0.114 8.753 54.995 3024.4
9 0.083 12.027 75.567 5710.4
10,11 0.077 12.957 81.414 6628.2
12 0.038 26.606 167.17 27946
600
13,14 0.032 31.056 195.13 38076
15 0.025 40.201 252.59 63802
16 0.02 50.683 318.45 101410
17 0.018 56.657 355.99 126730
18 0.013 75.561 474.76 225400
Table 11: Base Reactions using FE model:
FX FY FZ MX MY
kN kN kN kN-m kN-m
741.482 741.482 987.873 22874.62 22874.62
Table 12: Response Spectrum Modal Information
Mode Period U1Acc U2Acc U3Acc U1Amp U2Amp U3Amp
1 1.5600 0.3361 0.3361 0.2241 -0.5542 -0.1242 0.0000
2 1.5600 0.3361 0.3361 0.2241 0.1242 -0.5542 0.0000
3 0.4358 1.2291 1.2291 0.8195 -0.0748 0.0885 0.0000
4 0.4358 1.2291 1.2291 0.8195 0.0885 0.0748 0.0000
5 0.1959 1.3072 1.3072 0.8716 -0.0141 0.0103 0.0000
6 0.1959 1.3072 1.3072 0.8716 0.0103 0.0141 0.0000
7 0.1143 1.3072 1.3072 0.8716 -0.0033 -0.0030 0.0000
8 0.1143 1.3072 1.3072 0.8716 -0.0030 0.0033 0.0000
9 0.0831 1.1750 1.1750 0.7835 0.0000 0.0000 0.0048
10 0.0772 1.1282 1.1282 0.7522 -0.0010 -0.0010 0.0000
11 0.0772 1.1282 1.1282 0.7522 0.0010 -0.0010 0.0000
12 0.0376 0.8177 0.8177 0.5452 0.0000 0.0000 0.0003
13 0.0322 0.7754 0.7754 0.5170 0.0000 0.0005 0.0000
14 0.0322 0.7754 0.7754 0.5170 -0.0005 0.0000 0.0000
15 0.0249 0.7180 0.7180 0.4787 0.0000 0.0000 0.0001
16 0.0197 0.6776 0.6776 0.4518 0.0000 0.0000 0.0000
601
17 0.0177 0.6613 0.6613 0.4409 0.0000 0.0000 0.0000
18 0.0132 0.6267 0.6267 0.4179 0.0000 0.0000 0.0000
The base reactions obtained using FE model are given in Table 11.The modal accelerations and the
corresponding amplitudes in each mode are given in Table 12.
Table: 13 Mass Participation factors
Mode UX UY UZ Sum UX Sum UY Sum UZ

1 0.354 0.018 0.000 0.354 0.018 0.000
2 0.018 0.354 0.000 0.372 0.372 0.000
3 0.079 0.111 0.000 0.451 0.483 0.000
4 0.111 0.079 0.000 0.562 0.562 0.000
5 0.061 0.033 0.000 0.623 0.594 0.000
6 0.033 0.061 0.000 0.655 0.655 0.000
7 0.028 0.024 0.000 0.684 0.679 0.000
8 0.024 0.028 0.000 0.708 0.708 0.000
9 0.000 0.000 0.605 0.708 0.708 0.605
10 0.016 0.016 0.000 0.723 0.723 0.605
11 0.016 0.016 0.000 0.739 0.739 0.605
12 0.000 0.000 0.130 0.739 0.739 0.735
13 0.000 0.261 0.000 0.739 1.000 0.735
14 0.261 0.000 0.000 1.000 1.000 0.735
15 0.000 0.000 0.047 1.000 1.000 0.782
16 0.000 0.000 0.018 1.000 1.000 0.799
17 0.000 0.000 0.003 1.000 1.000 0.803
18 0.000 0.000 0.197 1.000 1.000 1.000
Table 14: Comparison of results

Analysis Period Base Shear Base Moment
sec kN kN-m
Mode 1 Mode 2
Simplified Modal Analysis 1.5591 0.4345 523.81 22131.6
Technique
Beam Model using SAP2000 1.560 0.436 741.482 22874.62
The mass participation factors for various modes are given in Table 13. Finally the comparison results between
FE analysis and simplified method are presented in Table 14 and it can be seen that the simplified analysis
602
technique is found to give reasonably accurate results compared to detailed dynamic analysis results and thus
can be readily used by the designers.
6. CONCLUSIONS
A simplified modal analysis technique is used for computation of mode shapes, modal forces/moments and
frequencies. Also detailed dynamic analysis using FE method is carried out for evaluating the response of the
chimney and the results obtained using both these methods are compared. Reasonable agreement was found
using simplified analysis technique and can be readily used by the designers. The method is simple, yet gives
accurate representation of the results in terms of structural response. It is also observed that the fundamental
period of vibration of chimney is quite large and hence acceleration is very low, however the displacements are
high. Therefore such structures have to be carefully designed to safely withstand the forces likely to be imposed
on them by ground motion.
REFERENCES
[1] IS 1893:2002 (Part I). Criteria for earthquake resistant design of Structures. Part I, General Provisions and
buildings.
[2] IS 1893:2005(Part IV). Criteria for earthquake resistant design of Structures. Part IV, Industrial structures
including stack like structures.
[3] Manohar S.N.(1985). Design and construction of Tall Chimneys, Tata McGraw- Hill Publishing Company.
[4] Phanikanth V.S.(2008). “Soil structure interaction analysis of a reinforced concrete stack”, Indian
Geotechnical Conference, Bangalore,India,2008.
[5] SAP2000. “Integrated finite element analysis and design of structure: analysis reference.” Computers and
Structures, Inc., Berkeley, California, 2000.
[6] Wadi S. Rumman, (1985),”Modal Characteristics of Linearly Tapered Reinforced Concrete Chimneys”,
ACI Journal , Title no.82-48,pp.531-536.
603
Paper No.: 118
EFFECT OF TEMPERATURE ON COMPRESSIVE BEHAVIOUR OF

CONCRETE REINFORCED WITH STRUCTURAL POLYPROPYLENE
FIBRES
G Srikar1 B Gopi2 and S Suriya Prakash3
1
B-Tech Student, Department of Civil Engineering, Indian Institute of Technology, Bhubaneswar, India.
2
M-Tech Student, Department of Civil Engineering, Indian Institute of Technology, Hyderabad, India.
3
Assistant Professor, Department of Civil Engineering, Indian Institute of Technology, Hyderabad, India
E-mail: sg15@iitbbs.ac.in1, ce12m1004@iith.ac.in2, suriyap@iith.ac.in3
ABSTRACT:
This paper aims to present the effects of elevated temperatures on the compressive behaviour of concrete
reinforced with structural polypropylene fibres. An experimental program was designed and carried out
involving testing of cubical concrete specimens exposed to different elevated temperatures ranging from Room
Temperature to 250oC. A total of 48 specimens were prepared and tested during the program. The test variables
included temperature of exposure, concrete strength, fibre dosage. A software named Vic-2D was used which
works on the principle of Digital Image Correlation, a non-contacting method for measuring surface
deformations to obtain the stress-strain plots of the specimen. The trends in the variation of relative residual
strengths with respect to exposure temperature and fibre dosage indicated that role played by fibres in
controlling strength depletion is phenomenal. An analysis of the stress-strain curves shows an improvement in
ultimate compressive strengths and post peak behaviour with respect to fibre dosage and exposure temperature.
KEYWORDS:
Structural polypropylene fibres, fibre dosage, Digital Image Correlation, Relative residual strength, Post peak
behaviour.
1. INTRODUCTION:
Concrete has proved to be a versatile material in the construction of structures due to the possibility of moulding
it into virtually any shape and geometry. Utilizing the formable nature of the material, concrete architecture has
made rapid progress in the recent years. The excellence of concrete in compression and at the same time its
inferiority in carrying tensile loads is an axiom in the concrete world. Reinforcement is used to absorb these
tensile forces and arrest the inevitable phenomena of cracking. The capability of a durable structure to resist
weathering action, chemical attack, abrasion and other degradation processes during its service life with
minimal maintenance is equally important as the capacity of the structure to resist the loads applied on it. The
advent of fast track construction techniques lead to the development of Polypropylene Fibre-reinforced concrete
which added a new dimension to the construction industry with its applications to some of the chronic issues of
concrete.
1.1 General:
1.1.1 Fibre-Reinforced Concrete-Types of fibres:

Fibre reinforced concrete (FRC) is a concrete mix containing water, cement, fine & coarse aggregates and
discontinuous fibres of various shapes and sizes. These fibres used in cementitious composites can be classified
with regard to:
• Origin of fibres: E.g. natural organic (cellulose, jute etc.), natural inorganic (asbestos, rock wool etc.), man-
made (steel, glass, synthetic etc.)
• Physical/Chemical properties: On the basis of density, surface roughness, flammability, reactivity etc.
• Mechanical properties: On the basis of specific gravity, tensile strength, elastic modulus etc.
• Shape and size: on the basis of geometric properties such as length, diameter, cross sectional shape
(circular, rectangular, diamond, square, triangular, flat and polygonal).
However, Polypropylene fibres have gained more prominence for reinforcing cementitious materials and are
focused mainly throughout the paper.
1.1.2 Guidelines
Polypropylene fibres are sub divided into micro-synthetic and structural fibres. Micro-synthetic fibres are
typically 12mm long by 18µm diameter whereas the macro ones are significantly larger with 40-50mm lengths
and 1.0-1.5mm width and are used at higher dosages to provide significant level of post-crack control. The
dispersion of these fibres has a crucial effect on the workability of the concrete. Consistency is reduced with an
in increase in the fibre contents and this reduction is even greater with structural fibres. However, workability is
restored with the addition of super plasticizer. The modulus of elasticity of the fibres being higher than the
concrete in plastic state, effectively improve the cracking characteristics of the fresh concrete. Along with these
traditional concreting practices can be typically followed for MIXING, CASTING, DEMOULDING, CURING
and STORING of the specimens.
2. LITERATURE REVIEW:
Fire ranking high among the potential risks, there is a need to study the effects of elevated temperature on FRC.
Tobermorite gel is the primary contributor to the cement paste structural properties. Exposed to elevated
temperatures causes physical changes, including large volume changes due to thermal dilatations, thermal
shrinkage and creep related water loss. The volume changes can result in large internal stresses and lead to
micro-cracking and fracture. High temperatures also cause many changes, such as water migration, increased
dehydration, interfacial thermal incompatibility and chemical decomposition of hardened cement paste. In
general all these changes lead to the decrement of the stiffness of concrete and increment in an irrecoverable
deformation. Various investigations validate that depletion of strength and stiffness of concrete with increasing
temperature, exposure time and thermal cycles. More and more attention has been paid to the mechanical
properties of concrete at high temperature or the residual properties of concrete after exposure to elevated
temperatures as high temperature causes dramatic physical and chemical changes, resulting in the deterioration
of concrete.
In this regard, Ahmed et al (2006) made an attempt to study the effects of addition of various proportions of
polypropylene fibre on the mechanical properties of concrete like compressive, tensile, shear and flexural
strengths. The author reported that the shrinkage cracking was reduced by 83-85% by addition of fibres in the
range of 0.35-0.50%. A notable increase in the flexure, tensile and shear strength was found with increasing
fibre dosages indicating the improvement of the bridging action. Noumowe (2005) developed some important
data relating to the mechanical properties and internal mechanisms that are happening in the high strength
concretes with and without polypropylene fibre reinforcement when exposed to an elevated temperature of
200oC. He observed that a 1.8kg/m3 fibre dosage led to small changes in residual compressive strength, modulus
of elasticity and splitting tensile strength due to melting of fibres due to heating indicating a decrement in the
heat resistance of the mechanical properties. Thermo gravimetric analysis and Differential Scanning
Calorimetric analysis revealed the temperature ranges of the decomposition reactions while Scanning Electron
Microscopy validated the increase in porosity with PP fibre reinforcement at high temperatures which may
result in lowering of vapour pressure hence lowering the risk of spalling due to accident.
605
Poon et al (2004) investigated the effects of elevated temperatures of the compressive strength stress-strain
relationship (stiffness) and energy absorption capacities (toughness) of high strength concretes with different
mineral admixtures like metakaolin (MK), silica fume (SF). Fibre type (steel or Polypropylene (PP)), fibre
dosage being the test variable the results showed that after exposure to 600-800oC, there was 45-23% retaining
of compressive strength, losses in stiffness were much quicker than compressive strength and energy absorption.
The MK concrete showed quicker loss in comparison to the SF. Steel fibres doubled the energy absorption of
unheated concrete and were effective in minimizing the degradation of compressive strength. In comparison
using PP fibres reduced energy absorption capacity of the concretes. Kalifa et al (2001) presented the fact that
PP fibres have been used in the HPC constructions at dosage of 2kg/m3 with the intention of reducing the
propensity of concrete to spall and to understand the physical process through which fibres act against spalling
and to optimize the fibre dosage. Using an electrical analogy to stimulate the evolution of permeability with
fibre dosage and temperature, it was show that fibres act in a serial/parallel system with pores, ITZs and cracks
when they are formed. It appears that the fibre dosage of 2kg/m3 with a fibre length in the order of 20mm is an
efficient solution for preventing spalling in HPC up to 100Mpa.
Type of cooling affects the properties and the effect being more pronounced as the temperature increases.
Husem (2005) validated this with his experimental work made with ordinary and high-performance concrete
when exposed to high temperatures (200,400,600,800 and 1000oC) and cooled differently. The strength loss
curves given in the Finnish code are compared with the experimentally obtained curves and found to be
agreeing with each other. Obtained compressive and flexural strengths of concrete samples were compared
indicating that concrete strength decreases with increasing temperature and decrease in the strength of ordinary
concrete is more than that in high-performance concrete. Fibres have extensively been used to improve the
ductility of concrete. Cheng et al (2004) studied the effects of high temperature on the strength and stress-strain
relationship of high strength concrete (HSC). Results from stress-strain curves of HSC’s say that plain HSC was
brittle till 600oC whereas steel fibre reinforced HSC exhibited ductility after 400oC. The compressive strength
decreased about a quarter of its room temperature strength within the range of 100-400oC. Further depletion of
strength was observed at 800oC.
Recent studies revealed that a number of fibres can also improve the residual properties of concrete after
exposure to elevated temperatures which qualitatively indicates the degree of deterioration caused. An
assessment of the degree of deterioration of concrete structure after exposure to high temperatures can help
engineers to decide whether a structure can be repaired rather than required to be demolished. Xiao et al (2006)
contributed to investigate the residual strength of HPC’s with the test variable being fibre dosage, exposure
temperature, strength of mix. Results portrayed that regardless the presence of PP fibres there is an obvious fall
in the residual compressive strength of HPC’s, however there is an increment in the relative residual strengths.
Unlike compressive strength, the residual strengths drops regardless of PP fibres in case of flexure. Spalling of
concrete was mitigated in specimens containing PP fibres. The author believes that blast furnace slag would be
beneficial to the residual strengths in comparison to silica fume.
Fibre reinforced concrete, an advancement in the concrete technology has become increasingly popular. A
greater understanding of its behaviour under different conditions will improve confidence in its use. With this
perspective Farhad Aslani et al (2013) explored the properties of Self-Compacting Concrete (SCC) with fibre
reinforcement. A test program was designed to develop information about the mechanical properties including
compressive and splitting tensile strengths, moduli of elasticity and rupture, compressive stress-strain curve and
energy dissipated under compression. Their study was performed to develop a simple and rational mathematical
model for the prediction of the mechanical properties which are found to be quite comparable.
With the technical insight gained, the effect of PP fibres on the compressive behaviour of concrete and its ability
to enhance the residual performance of heated concrete is studied in this paper.
3. EXPERIMENTAL PROGRAM:
3.1 Materials:
The cementitious materials used in this study were Ordinary Portland Cement (OPC), Fly ash. The coarse
aggregate used was crushed granite with nominal sizes of 12.5mm and 20mm. the specific gravity of the
606
aggregate was 2.63g/cm3. In concrete mixtures, the 12.5mm and 20mm coarse aggregate were used in
proportion of 2:3. Natural river sand, of specific gravity 2.62g/cm3 was used as fine aggregate. The polymer
fibres used were structural PP fibres with a length, width and thickness being 60mm, 1.68mm and 0.60mm
respectively. Figure-1 shows the photograph of characteristic structural PP fibres.
Figure-1 Structural Polypropylene fibres
3.2 Mix proportioning:
A total of 4 mixes were prepared with varying fibre dosage. All the mixes were prepared with the same
cementitious constitutions, only by varying the fibre dosage (0kg/m3, 4 kg/m3, 5 kg/m3 and 6 kg/m3). All the
concrete mixes were prepared at a water-to-cementitious materials ratio of 0.45. A detailed overview of the mix
proportions are shown in the Table 1.
Table-1 Mix proportions in kg/m3

Coarse Fine
Mix Water OPC Fly-ash River sand PP fibre
Aggregate Aggregate
1 192 298 127 689 519 519 4
2 192 298 127 689 519 519 5
3 192 298 127 689 519 519 6
3.3 Specimen preparation:
The concrete mixes were prepared in a tilting drum type mixer. Each granulometric aggregate was weighed and
placed into the concrete mixer moistened in advance and mixed for 3min with the addition of saturation water,
for 3min with the addition of cement (together with fly ash), and thereafter, mixed for another 3min without
stopping to add the mixing water. For each mix, a total of 12 specimens comprising of cubes of 100 (length) ×
100 (breadth) × 100 mm (height) were cast in steel moulds. The specimens, after removal from the steel moulds
at 1 day, were cured in water at 27oC until the age of 28 days. Figure-2 shows an image taken while casting the
specimens.
607
Figure-2 Concrete moulds
3.4 Test Procedure:
After 28days of curing, the fully saturated specimens were taken out and dried to saturated and surface dry
status. Three specimens from each of the 4 mixes were grouped together and made into a batch. Each batch was
heated in an electric oven to peak temperatures of 150, 200 and 250oC at a definite temperature gradient of
2oC/min and the peak temperature was maintained 1hr. The diagrammatic representation of the temperature load
is shown in Figure-3. An overview of the process is given in Table-2. The heated specimens were then cooled to
room temperature and made ready for the compression testing. The specimens were then coated with a white
paint as a primer on which a speckle pattern is created in order to obtain the strain values based on DIC
technique. The specimens are then subjected to loading operated under load control with the rate of loading
being 2500N/sec and after application of 50-60% of the expected peak load the loading is shifted from load
control to displacement control at a load rate of 0.01µm/min.
Figure-3 Graphical Representation temperature gradient.
608
Table-2 Specimen Chart
Temperature/Dosage 0 kg/m3 4 kg/m3 5 kg/m3 6 kg/m3

Room Temperature 3Cubes 3Cubes 3Cubes 3Cubes
o
150 C 3Cubes 3Cubes 3Cubes 3Cubes
o
200 C 3Cubes 3Cubes 3Cubes 3Cubes
250oC 3Cubes 3Cubes 3Cubes 3Cubes
3.5 Digital Image Correlation:
Digital Image Correlation (DIC) is a full field non-contacting image analysis method based on grey value digital
images. In DIC, the images of a zone of interest are captured at different deformation states and post processed
by tracking a collection of smaller areas known as subsets. DIC has been primarily used as an alternative to the
traditional deformation measuring methodologies as the fracture patterns and gage readings of concrete are not
disturbed while failure progresses.
3.5.1 Working principle:
When a specimen is loaded its surface image deforms accordingly. Two different images represent different
loading stages, one is the initial reference image and other is the deformed image. Calculating the
transformation parameters for images under different loading conditions both the displacement vector and
deformation for each facet can be determined. In order to make the specimens reveal unique image patterns to
apply correlation matching, usually black paint is lightly sprayed on white painted surfaces of the specimens.
Figures 4(a) and 4(b) shows all the pre-processed test specimens and typical speckle pattern respectively.
Fig-4(a) Pre-processed test specimens Fig-4(b) Typical Speckle pattern
609
3.5.2 Stress-strain curves:
The strain values are obtained from the DIC apparatus and the loading values are obtained from the controller of
the compression testing machine. Vic-2D software was used to correlate them accordingly and to obtain the
stress-strain curves.
Peak compressive strength of specimen with different fibre dosage at different temperature are summarized in
Table 3. Stress strain curves of fibre reinforced specimen at different temperature are shown in Figures 5-10.
• A slight colour change from grey to pale white has occurred to the specimens indicating the initiation of
chemical reactions with in the concrete matrix.
• The size of the specimen considered in this study is small (10×10×10 cm). This can be attributed for the
inconsistency in the stress-strain graphs. Due to smaller size of specimen, displacement transducers
were not used in this study. Future work can consider cylinder specimen for establishing stress strain
curves. Additional displacement transducers can be mounted to compare its prediction with DIC data.
Fig-6, 8 shows the inconsistent pattern of the stress-strain curves.
• With an increase in fibre dosage, there is an improvement in the post peak curve at all temperatures.
Fig-5, 6, 8 showing improved post-peak curves.
• There is a slight improvement in the compressive strengths (from Table-3) of the specimens with
increase in fibre dosage.
• However, with an increase in temperature there is decrease in the peak compressive stress accompanied
with an increase in the peak strain at all fibre dosages indicating the depletion of the strength giving C-
S-H gel. So, the specimen deforms indicating the losing the stiffness for the same load.
Stress-Strain at Room Stress-Strain at 150oC

60 70
Temperature
50 60
M25-RT-0 50 M25-150-0
Stress in MPa
40
Stress MPa
M25-RT-4 40
30 M25-150-4
M25-RT-5 30
20
M25-RT-6 20 M25-150-5
10 10
M25-150-6
0 0
-0.002 0 0.002 0.004 0.006 0 0.002 0.004
Strain Strain
Fig-5 Plot of Stress vs. Strain at Room Temperature Fig-6 Plot of Stress vs. Strain at 150oC
610
Stress-Strain at 200oC Stress-Strain at 250oC
60 60
50 50 M25-250-0
Stress MPa
40 40
Stress Mpa
M25-200-0
M25-250-4
30 M25-200-4 30
M25-200-5 20 M25-250-5
20
M25-200-6
10 10
M25-250-6
0
0
-0.005 0 0.005 0.01 -0.002 0 0.002 0.004 0.006
Strain Strain
Fig-7 Plot of Stress vs. Strain at 200oC Fig-8 Plot of Stress vs. Strain at 250oC
Residual Strength vs. Fibre Dosage Relative Residual Strength vs.

1.6 1.4 Temperature (C)
Relative Residual strength
1.4 1.2
Residual Strength
1.2 Room 1
1 Temperature
150C 0.8
0.8
F Dosage-0kg/m3
0.6 200C 0.6
F Dosage-5kg/m3
0.4 0.4
250C
0.2 F Dosage-6kg/m3
0.2
0 F Dosage-4kg/m3
0
0 5 10 0 100 200 300
Fibre Dosage (kg/m3)
Temperature in C
Fig-9 Plot of Residual Strength vs. Fibre Dosage Fig-10 Plot of Residual Strength vs. Temperature
Table-3 Peak Compressive strength
Compressive Room
150oC 200oC 250oC
Strength (MPa) Temperature
0kg/m3 43.00(1.23) 54.00(4.00) 48.67(4.94) 42.33(3.51)
4kg/m3 48.05(8.41) 57.50(2.12) 53.00(1.41) 52.85(0.21)
3
5kg/m 46.00(7.77) 52.67(5.03) 48.00(0.70) 47.33(1.15)
6kg/m3 58.50(4.94) 56.13(0.28) 53.23(1.36) 47.00(1.73)
* The values in the table indicates standard deviation
5. CONCLUSIONS:
From limited experimental study made in this research, generic conclusions cannot be derived. The following
general observations can be made from the test results:
611
• It is observed that with an increase in the fibre dosage there is a slight increase in the compressive
strength at room temperature conditions. The same phenomena holds up with the ones at high
temperatures. However, there is a decrement observed with an increase in the fibre dosage.
• The compressive strength spikes up and falls down (from Fig-9, 10) for a given fibre dosage with
increase in temperature. This may be due to the pseudo autoclaving that is happening during the test
procedure.
• Addition of fibres helps to reduce post-peak degradation in stiffness with increase in the fibre dosage.
6. SCOPE FOR FUTURE STUDY:
Due to time and resource constraints, the current research was confined to compressive behaviour only, but the
following can be considered for future work:
• Influence of fibre reinforcement on other properties of concrete including Flexural strength, tensile
strength, fire resistance, fresh concrete rheology and cracking.
• With fibre type as a test variable an analysis can be done using the idea of hybrid reinforcements with
carbon and steel fibres at various dosages.
• The research can be extended to study the effect of higher temperatures around 1000oC which are more
representative of Fire-accidents.
• The effect of testing conditions (hot and residual) on mechanical properties of Fibre Reinforced
Concrete at elevated temperatures can be studied.
• An analytical model can be developed representing the phenomena in any commercial Finite Element
Software and can be validated with the experimental results. The developed model can be further used
for additional parametric studies as well.
REFERENCES
[1] Cheng, F., Kodur, V. K. R., and Wang, C. (2004). “Stress-Strain Curves for High Strength Concrete at
Elevated Temperatures.” Journal of materials in civil engineering, 16, 0899-1561.
[2] Xiao, J., and Falkner, H. (2006). “On residual strength of high-performance concrete with and without
polypropylene fibres at elevated temperatures.” Fire safety journal, 41, 115-121.
[3] Poon, C. S., Shui, Z. H., and Lam, L. (2004). “Compressive behaviour of fibre reinforced high-performance
concrete subjected to elevated temperatures.” Cement and concrete research, 34, 2215-2222.
[4] Noumowe, A. (2005). “Mechanical properties and microstructure of high strength concrete containing
polypropylene fibres exposed to temperatures up to 200oC.” Cement and concrete research, 35, 2192-2198.
[5] Aslani, F., and Nejadi, S. (2013). “Self- compacting concrete incorporating steel and polypropylene fibres:
Compressive and tensile strengths, moduli of elasticity and rupture, compressive stress-strain curve, and
energy dissipated under compression.” Composites: Part B, 53, 121-133.
[6] Husem, M. (2005). “The effects of high temperature on compressive and flexural strengths of ordinary and
high-performance concrete.” Fire safety journal, 41, 155-163.
612
[7] Kalifa, P., Chene, G., and Galle, C. (2001). “High- temperature behaviour of HPC with polypropylene fibres
from spalling to microstructure.” Cement and concrete Research., 31, 1487-1499.
[8] Ahmed, S., and Imran, A. (2006). “A study on properties of polypropylene fibre reinforced concrete.” Proc.,
31st conference on Our World in Concrete & Structures, Singapore.
[9] Kazuo., and Mugume (2013). “The effect of testing conditions (hot and residual) on fracture toughness of
fibre reinforced high-strength concrete subjected to elevated temperatures”. Cement and concrete Research.,
51, 6-13.
[10] Chan, Y. N., Peng, G. F., and Anson, M. (1999). “Residual strength and pore structure of high-strength
concrete and normal strength concrete after exposure to high temperatures.” Cement and concrete
composites., 21,23-27.
613
Paper No.: 119
BEHAVIOUR OF MASONRY ASSEMBLAGES MADE OF SOFT BRICK

UNDER COMPRESSION
C Sreenivasulu1, R Mehar Babu2, Dr. S Suriya Prakash3 and Prof. K.V.L. Subramaniam4
1
Student, Dept. of Civil Engineering, Indian Institute of Technology, Bhubaneswar, India
2
PhD. Scholar, Dept. of Civil Engineering, Indian Institute of Technology, Hyderabad, India
3
Assistant professor, Dept. of Civil Engineering, Indian Institute of Technology, Hyderabad, India
4
Professor, Dept. of Civil Engineering, Indian Institute of Technology, Hyderabad, India
Email: sc14@iitbbs.ac.in, ce11p1004@iith.ac.in, suriyap@iith.ac.in, kvls@iith.ac.in
ABSTRACT:
This paper documents the results of analysis of stress strain curves of masonry assemblages made of soft bricks
under uniaxial compression. The characteristics of brick masonry mainly depends upon the properties of mortar
and brick. This paper investigates the masonry strength over different mortar strength without varying the brick
strength. Based on the results obtained from experimental program, nonlinear stress strain curves have been
obtained. Effect of mortar strength on the strain corresponding to peak stress was also analyzed. The existing
empirical formula was modified to correctly predict the stress-strain behavior of prisms made of soft bricks.
This modified model requires the compressive strength of prism and mortar which were obtained by
experimental program as available in the codes. Comparison between experimental and analytical models were
made. There was a vertical splitting in the sideways of the specimens which is due to tension development and
central region confinement and vertical splitting in front and backside of the specimens. Vic-2D which uses
digital image correlation (DIC) technique was used to obtain strain measurements. The readings from
conventional LVDT and DIC data also shows close comparison in the pre-peak range. As expected, the strength
of masonry increased with increase in strength of mortar.
KEYWORDS: Stress, strain, prism, DIC, soft bricks, compressive strength
1. INTRODUCTION:
Masonry walls are used in almost all types of building construction in many parts of the world for its
sustainability, eco-friendly nature and its low cost. Brick masonry is building of structures using individual
(bricks) units bonded together by mortar. It has significant importance in India. Brick masonry is
non-homogeneous, non-elastic and anisotropic composite material composed of two components- brick and
mortar. The different mechanical properties of masonry (bricks) units, mortar and interface between them make
the composite system behave differently and difficult to predict the design of masonry structures based on the
individual properties of materials but whereas steel, concrete structures are easy to design based on the
individual properties of materials. Prism properties depends upon several factors-compressive strength of brick
and mortar, shear strength between brick and mortar interface, flexural bond strength, thickness of mortar, type
of bonding etc. This paper focuses on the compressive strength of prism which is the main factor to be
considered in structural design of a structure. Today main focus of the design of structures is based on
sustainability and economical rather than strength which is possible with brick masonry structures. Specimens
were fabricated using three different types of mortar (strength) and single type of masonry (brick) unit. Mortar
influences the strength, durability of masonry structures so in order to have better understanding of mortar
strength effect 1:3, 1:4.5, 1:6 ratio of cement: sand was chosen. Soft bricks are the bricks which have less
strength and they have more tendency to expand i.e. higher poison ratio when compared with stiffer bricks. The
study of masonry made of soft brick was investigated because the properties of non-homogeneous material
composite of brick and mortar needs to understood where does the properties of masonry lie in comparison with
properties of brick and mortar.
2. LITERATURE REVIEW:
Generally it was believed that brick masonry properties lie in between the properties of mortar and brick which
is a contradiction for masonry made of soft bricks and stiffer mortar as the bond between them is weak. McNary
et al. (1985) proposed a theory to determine strength and deformation characteristics of stack bond prisms and
stated that behavior of masonry prism mainly depends on the nonlinear properties of the mortar. Berman (2002)
conducted several tests on triplets to understand the behavior of masonry under bi-axial loading. Berman (2002)
fabricated the triplet specimens by giving a 0.5 inch offset to the middle brick from which he proposed that
triplet, a sub system is a standardized design.
Sarangapani et al. (2005) studied the behavior of masonry made of soft bricks considering parameters mortar
strength, effect of plasticizing additives in mortar and other factors. Sarangapani et al. study revealed that
compressive strength of masonry increased with increase in bond strength without varying the strength of brick
and mortar. Lumantarna et al. (2012) conducted in-situ material testing to investigate the relationships between
compressive strength, shear strength and flexural bond strength. Lumantarna et al. found that flexural bond
strength is better characterized factor using the mortar compressive strength than using the masonry
compressive strength. Venkatarama Reddy et al. (2008) conducted several tests to analyze the influence of shear
bond strength on compressive strength on masonry and concluded that masonry compressive strength increases
with increase in shear bond strength in case of soft brick- stiff mortar. Gumaste et al. analyzed the behavior of
stack bonded prism under compression and found an empirical relation to find prism strength based on the
strength of brick and mortar as
(1)
Kaushik et al. (2007) conducted several tests on brick masonry made of stiffer bricks and soft mortar and by
regression analysis proposed that strength of masonry relays on the strength of brick and mortar as
(2)
Based on the experimental study and regression analysis Dayaratnam (1987) proposed that prism strength
depends upon strength of brick and mortar as
(3)
Priestley based on the experimental data of unconfined masonry made of concrete block proposed a model to
predict the stress strain characteristics by modifying the Kent-Park curve for unconfined concrete which consists
of raising curve, falling curve as
for 0 0.0015 (rising curve)
(4)
615
for 0.0015 0.0025 (falling curve)
(5)
Where
3. RESEARCH OBJECTIVES:
The present research focuses on improving the understanding on
• The effect of mortar strength on the compressive strength on the masonry made of soft bricks.
• The model based on the compressive strength of masonry and mortar to predict the strain corresponding
to peak stress of masonry under uniaxial compressive loading.
• Comparison between experimental and analytical models.
4. EXPERIEMNTAL STUDY:
For the analysis of stress strain curves several tests were conducted on bricks, mortar cubes, masonry prism over
there different grades of mortar. The dimensions of bricks were 220mm×90mm×70mm.
4.1. Properties of brick under compression:

In the present study single type of brick was considered throughout the analysis. The compressive strength of
brick was found as per IS 3495 (Part 1): 1992. Load was increased monotonically at a rate of 14N/mm2 per min.
Six brick specimens were tested using 5000 kN Universal compressive testing machine. The average
compressive strength of brick was found to be 5.09MPa.
4.2. Properties of mortar under compression:

In the present investigation to evaluate the effect of mortar strength on the behavior of masonry under
compression three different grades of mortar were considered by varying the ratio of cement- sand (C:S) in
mortar (1:3 strong mortar, 1:4.5 intermediate mortar, 1:6 weak mortar). The mortar cubes were casted and tested
for 28 days strength using 2000 kN compressive testing machine as per IS 4031 (Part 7): 1988. Load was
increased uniformly until the failure of the specimen occurred. The results are tabulated in the table-1:
Table-1 Compressive strength of mortar cubes

Cement: Compressive Strength
Number of
Sl No. Sand ratio in W/C ratio
Specimens (MPa) ( ) (28 days)
Mortar
1 1:6 0.9 3 7.34(0.21)
2 1:4.5 0.6-0.7 3 37.19(1.75)
3 1:3 0.6-0.7 3 31.38(3.14)

*The values in the brackets indicates the standard deviation values
616
4.3. Properties of brick masonry under compression:
Total of 18 specimens were fabricated using single grade of brick and three different grades of mortar varying
the ratio of cement- sand in mortar (1:3 strong, 1:4.5 intermediate, 1:6 weak). Among the 18 specimens which
were tested, 9 were for 14 day strength and 9 were for 28 days strength.
4.3.1. Prism:
Prism consists of stack of five bricks bonded together by mortar. Prism is a sub assemblage of masonry which
was considered for understanding the behavior of masonry under compression. Prism was considered because
the results obtained from testing the prism can be related to real world structures and it is also very difficult to
test and analyze the full scale equivalent system. Prisms were casted and tested for 14 days strength and 28 days
strength using 250 kN MTS Hydraulic Actuator as per ASTM C1314-12. LVDT was setup between the 2nd
brick and 4th brick to evaluate the relative displacement between 2nd and 4th brick as shown in the Fig-1 (b). DIC
setup was also used to obtain the strain measurements. The results of compressive strength of prisms are
tabulated in the table-2
LVDT
Fig-1(a) Uniaxial Compressive loading Fig-1 (b) LVDT set up of the prism (back view)
Table-2 Compressive strength of Masonry Prism
Cement : Compressive Strength

No of Specimens Modulus of
Sand (W/C ( ) (MPa)
Sl NO elasticity ( )
(C:S ratio)
14days 14days 28 (MPa)
ratio)
28days days
1.75(0.12)
1 1:6 0.9 3 3 267.62
1.46(0.01)
1.92(0.1)
2 1:4.5 0.6-0.7 3 3 229.86
1.65(0.04)
1.54(0.05)
3 1:3 0.6-0.7 3 3 196.43
1.93(0.16)
*The values in the brackets indicate the standard deviation values
617
4.4. Digital Image Correlation:
Digital image correlation is a new innovative idea which can be used for obtaining the strain values on a
planar surface. It involves digital image processing of images before deformation and after deformation from
which it evaluates the deformation at any point. This technique involves strain measurements by considering
the motion of speckle pattern on the surface of the specimen and measuring the intensity data at each pixel
array which is correlation. It uses the principle that applied stresses changes both the thickness and optical
properties of materials. As DIC is strain measurement technique it has desirable properties when compared
with others
• It is a non-contact method, so it can be used for strain measurements of soft materials and as well as
on hard materials
• It gives a whole field displacement and strain measurements in homogeneous and non-homogeneous
materials.
• It gives accurate results in the range of micro scale and nano scale
Vic -2D is a software which uses digital image correlation technique and this was used in the present
investigation.
5. ANALYSIS OF RESULTS:
5.1 Prisms Made of 1:3 Mortar Ratio
The comparison of stress strain curves of central brick, 2nd joint and 3rd joint from bottom was made. From
analysis, it was found that prisms made of 1:3 mortar ratio
• In vertical direction central brick, 2nd joint and 3rd joint were under compression and central brick was
undergoing more deformation than the 2nd, 3rd joints until peak stress as shown in fig-2(b).
• In horizontal direction all the three were first in compression and then in tension. 2nd and 3rd joint were
undergoing deformation in almost same manner.
• So central brick, 2nd joint and 3rd joint were in vertical compression and bilateral tension.
5.2 Prisms Made of 1:4.5 Mortar Ratio

Comparison between central brick, 2nd joint and 3rd joint from bottom was made and it was analyzed that
• In vertical direction central brick, 2nd and 3rd joints were under compression and central brick was
undergoing more deformation than 2nd, 3rd joints until peak stress.
• In horizontal direction all the three were undergoing deformation in the same manner until peak stress.
All the three were in tension as shown in fig-3(b).
• So central brick, 2nd joint and 3rd joint were in vertical compression and bilateral tension.
5.3 Prisms Made of 1:6 Mortar Ratio

• In vertical direction all the three were in compression and central brick was undergoing more
deformation as shown in fig-4(b).
• In horizontal direction all the three were in tension.
618
• So central brick, 2nd and 3rd joints were under uni-axial compression and in horizontal plane under
bilateral tension.
Fig-2 (a) Stress vs Strain Curve (2 to 4 brks) Fig-2 (b) Stress vs Strain
Fig-3(a) Stress vs Strain Curve (2 to 4 brks) Fig-3(b) Stress vs Strain
Fig-4(a) Stress vs Strain Curve (2 to 4 brick) Fig-4(b) Stress vs Strain
619
The compressive strength of prism was found to increase with increase in cement: sand (c/s) ratio. The variation
of compressive strength with cement: sand ratio was shown in the fig-5
Fig-5 Variation of compressive strength with c/s ratio
6. PEAK STRAIN ESTIMATION OF BRICK MASONRY:
In the present investigation it was observed the stress strain curve in the ascending part can be represented by
parabolic curve which depends upon the stress ratio and strain ratio as proposed by Kaushik et al. (2007). The
equation of the parabolic curve can be given by
(6)
Where and Ɛm are compressive stress and strain of brick masonry Ɛ’m is the strain corresponding to peak
stress of masonry. Even though the compressive strength of masonry, brick and mortar is known it has
become a problem to predict the strain corresponding to peak stress which is the factor to be considered for
structural design of masonry structures. The data obtained from experimental program was analyzed and by
regression analysis the following equation was proposed for estimating the strain corresponding to peak stress.
= * (7)
= 0.1* (8)
where is a factor depends on the mortar strength which needs to be considered as the masonry strength
mainly depends on the mortar strength. Comparison of curves is shown in fig-6(a) and 6(b).
620
Fig-6(a) Comparison of Stress- Strain Curves (1:4.5) Fig-6(b) Comparison of Stress- Strain Curves (1:3)
7. COMPARISON BETWEEN LVDT AND DIC DATA:
The data obtained from LVDT and DIC were compared inorder to have better understanding. Comparison of
curves is shown in fig-7(a) and 7(b). LVDT involves is a real time analysis whereas DIC is a non contact
method, post processing of data. The Load vs Displ curves obtained from LVDT and DIC were showing a close
relations before peak load which shows that DIC is more reliable for soft materials and in case of
non-homogeneous materials in order to measure displacement and strain.
Fig-7(a) Comparison of Load vs Vertical Displ Curves (1:3) Fig-7(b) Comparison of Load vs Vertical Displ Curves (1:6)
8. SUMMARY AND CONCLUSIONS:

The effect of mortar strength on compressive strength of masonry by varying the strength of mortar. An
equation was proposed considering the strain corresponding to peak stress for masonry made of soft bricks
under uniaxial compressive loading. The comparison between the stress-strain curves from experimental data
and the proposed equation showed a close correlation. By analyzing the data obtained from experimental
program it was observed that masonry compressive strength mainly varies with the strength of mortar. It was
621
observed that masonry strength (28 days) increased with increase in strength of mortar. Comparison between
middle brick, 2nd and 3rd joint from bottom resulted –in case of uni-axial monotonic loading central brick, 2nd
and 3rd joints from bottom were under uni-axial compression and bilateral tension (unlike bricks in the stiff
brick-soft mortar system). The strength can be increased by improving the bond development using bond
enhancing techniques. Equations were proposed to estimate the peak strain through regression analysis.
Conclusions made in the present investigation was based on masonry made of single type of soft brick over
three different grades (cement: sand) of mortar. In order to have generalized model on masonry made of soft
bricks experimental study is required with various types of soft bricks and mortar strength considering various
factors- elastic modulus, poison ratio, thickness of mortar joint and type of bond between brick- mortar etc.
REFERENCES:
[1] ASTM. C1314-12. Standard Test Method for Compressive Strength of Masonry Prisms, by ASTM
International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959, United States.
[2] Atkinson, R. H. and Noland, J. L. (1983). “A proposed failure theory for brick masonry in compression.
Proc., 3rd Canadian Masonry Symp.” Edmonton, Alta., Canada, 5.1–5.17.
[3] Berman, J. B., Al-Chaar, G. K., and Dutta, P. K. (2002). “Biaxial Loading and Failure Behavior of Brick
Triplets with Fiber- Reinforced Polymer Composite Upgrades.” US Army Corps of Engineers, Engineer
Research and Development Center, CERL.
[4] Dayaratnam, P. (1987). Brick and reinforced brick structures, Oxford and IBH, New Delhi, India.
[5] Gumaste, K.S., Nanjunda Rao, K.S., Venkatarama Reddy, B.V., and Jagadish, K.S. (2007). “Strength and
elasticity of brick masonry prisms and wallets under compression.” Materials and Structures, 40, 241–253.
[6] Hemant Kaushik, B., Durgesh Rai, C.; Sudhir Jain, K. (2007). “Stress-Strain Characteristics of Clay Brick
Masonry under Uniaxial Compression.” J. Mater. Civ. Eng., 19, 728-739.
[7] Indian Standards (IS). (1988). Indian standard methods of test of burn clay building bricks—Part 1:
Determination of compressive strength, IS 3495, 3rd Rev., Bureau of Indian Standards, New Delhi, India.
[8] McNary, W. S., Attkinson, R. H., Abrams, D. P., and Noland, J. L. “Basic properties clay unit masonry
stack bond Prism in compression.” ,711-718.
[9] McNary, W. S., Daniel Abrams, P. (1985). “Mechanics of masonry in compression.” J. Struct. Eng., 111(4),
857-870.
[10] M.J.N. Priestley. (1981). “Ductility of unconfined masonry shear walls.” Journal ASCE, Structural Div.,
14(1), 12-20.
[11] Monjur Hossain, Md., Sekender Ali, Sk., and Azadur Rahman, M. (1997). “Properties of masonry
constituents. Journal of Civil Engineering, 25(2), 135-155.
[12] Ronald Lumantarna, David Biggs, T., and Jason Ingham, M. (2012). “Compressive, flexural bond and
shear bond strengths of in-situ New Zealand unreinforced clay brick masonry constructed using lime
mortar between the 1880s and 1940s.” Journal of Materials in civil engineering, 10. 1061.
[13] Sarangapani, G., Venkatarama Reddy, B. V., and Jagadish, K. S. (2005). “Brick-mortar bond and masonry
compressive strength.” J. Mater. Civ. Eng., 17, 229-237.
622
[14] Venkatarama Reddy, B.V., and Uday Vyas, Ch.V. (2008). “Influence of shear bond strength on
compressive strength and stress-strain characteristics of masonry.” Materials and Structures, 41,
1697–1712.
[15] Venkatarama Reddy, B. V., Richard Lal, and Nanjunda Rao, K. S. (2007). “Enhancing bond strength and
characteristics of soil-cement block masonry.” J. Mater. Civ. Eng., 19, 164-172.
NOMENCLATURE:
=Compressive strength of prism masonry (Mpa)

= Compressive stress of prism masonry (MPa)
= Compressive strength of brick (MPa)
=Compressive strength of mortar (MPa)
= Elastic modulus of prism masonry (MPa)
Compressive Strain of prism masonry
= Compressive strain corresponding to peak stress
= strain in vertical direction
=strain in horizontal direction
623
Paper No.: 120
EFFECT OF TENSION STIFFENING ON BEHAVIOUR OF CONCRETE

COLUMNS UNDER PURE TORSION
1 2
K.V.V. Sumanth and Dr. S. Suriya Prakash
1
Undergraduate Student, School of Infrastructure, IIT Bhubaneswar, Bhubaneswar, India
2
Assistant Professor, Dept. of Civil Engineering, IIT Hyderabad, Medak, India
Email: svk10@iitbbs.ac.in, suriyap@iith.ac.in
ABSTRACT :
Softened Truss Model (STM) is a widely used for determining the torsional response of a reinforced concrete
(RC) element. Generally, the effect of tension stiffening is ignored in the analytical predictions. Inclusion of
tension stiffening in STM can greatly improve the prediction of behaviour of concrete element under pure
torsion. In the present work, an analytical model of the STM is examined with three different tension- stiffening
relations and compared with the experimental results. Existing softened stress-strain relationship of concrete
under compression was adopted in the model. Two circular columns and two square columns were used for
predicting the behaviour of developed model. The analytical and theoretical results were compared and a close
proximity between the predicted and calculated values was observed.
KEYWORDS: Shear flow zone; Torsional strength; Tension stiffening; Diagonal stresses; Softening of
concrete; thin wall tube theory; RC beams.
1. INTRODUCTION
It is necessary to develop a nonlinear torsional model which gives a consolidated output of the behaviour of
the critical parameters like the ultimate torsional strength in order to analyse the mechanical aspects of
reinforced concrete (RC) columns. The different combinations of loadings that are possible on structures are
flexure, moment, shear and torsion. Most commonly, all these loading actions occur in combination. So, to
understand each of these effects, it is necessary to formulate a model of each of these loadings, individually. The
softened truss model (STM) (University of Houston Hsu and Mo 1985a) is one such model which justifies the
torsional response of an RC structure. From a mechanical point of view, torsional behaviour of a rectangular
member is complicated than that of its circular counterpart due to the non-linear distribution of shear stress
across the whole cross-section. Rausch’s space truss analogy (1921) has been the earliest possible theory to
figure out the torsional response of an RC structure. In this analogy, the cross section of a member is assumed to
be effectively hollow. After cracking, the concrete is separated by 45° cracks into a set of helical members
which are then assumed to interact with the longitudinal and hoop steel bars to form a space truss. Each helical
member is idealized into a series of short straight struts, connected at the joints. The compression force in
concrete struts induces outward radial force at each joint which is resisted by the lateral hoop reinforcements.
The chains of diagonal concrete struts and hoop bars form a mechanism that would lengthen at an infinitesimal
applied torque which is resisted by the longitudinal reinforcement. In this way, a space truss is formed with the
diagonal concrete strut in compression and the longitudinal and hoop bars in tension. It is an ingenious
combination of Bredt’s (1896) thin tube theory and plane truss analogy for flexural shear in reinforced concrete.
From a mechanical point of view, the space truss analogy could not justify the effect of shear resistance of
concrete and dowel action of the rebars. Hsu’s (1993) softened truss model estimates the ultimate torque values
to a reasonable extent. The work done here, describes a model which includes the effect of tension stiffening to
make a continuous prediction of the peak torque and the torque-twist curve before and after the cracking. The
inclusion of tension-stiffening was done to improve the prediction of peak torque and twist. The model
effectively assumes that the difference between the experimental and theoretically calculated torque-twist values
is entirely due to tension-stiffening. The model was validated with test data available in literature. Matlab
software was used in the present model to work out with the calculations and iterations.
1.1. Background
The truss-model theory for shear and torsion has undergone four major developments since the late 1960s. First,
the proposal of variable angle truss model and the discovery of diagonal bending of concrete strut was made by
Lampert and Thurlimann (1968, 1969). Second, the compatibility equation was derived to determine the angle
of diagonal concrete strut by Collins (1973). Third, the softening phenomena in concrete struts was discovered
by Robinson and Demorieux (1972), and this theory was quantified by Vecchio and Collins (1981), using a
softening coefficient, fourth, combining the equilibrium, compatibility and softened stress strain relationships, a
softened truss model was developed (Hsu 1988), to predict the pre-cracking and post-cracking behaviour of
torsional loaded RC beams. Once the RC member has cracked in torsion, its resistance is provided primarily by
closed stirrups and longitudinal bars, located near the surface of the member. In the thin walled tube analogy,
the resistance is assumed to have been provided by the outer skin of the concrete cross-section, roughly centred
at the stirrups. Both hollow and solid cross sections are idealized as thin walled tubes both before and after
cracking.
1.1.1 Technical Background
Analysis for torsion is done on a circular column with the following assumptions:
• Plane sections perpendicular to the axis of a circular member remains plane after torque is applied.
• Cross sectional shape remains same.
Observations show that warping occurs after twisting. So the first assumption, so successfully used by Navier
for circular sections, appears not to be applicable for non-circular sections. Whereas the second assumption that
cross-sectional shape remains same appears to be applicable to circular as well as non-circular cross-sections.
Before cracking, the behaviour of an RC member and a plain concrete member are nearly same and can be
predicted by St. Vetant’s theory. After cracking, the material loses continuum and so a new equation taking into
account, the properties of concrete, reinforcements and the interactions between them.
2. TENSION STIFFENING
The property of concrete to offer resistance to diagonal tensile stresses which are developed as a resultant of
shear stresses during a pure torsional loading is called tension-stiffening. In the original STM while considering
the tensile strength to be zero, we are making an assumption that as soon as the concrete is cracked, it has
reached a point from where no further tension cracks can be developed which is practically not the case. In fact,
the cracks develop only at certain locations and the concrete between these cracks is still intact to offer
resistance to the diagonal tensile stresses. The cracked portion of concrete is devoid of any stresses and where
the tensile stresses are carried onto the reinforcements. A length of partial bond breakdown called the transfer
length exists at the portion adjacent to the cracks where the stresses in concrete and reinforcements vary
considerably due to tensile stress relief in concrete. The typical trends of these stresses are shown in fig.1.
625
Fig. 1. Distribution of concrete stress after cracking (TENSION STIFFENING MODEL FOR PLANAR
REINFORCED CONCRETE MEMBERS, Chang-Koon Choi and Sun-Hoon Cheung [1994])
3. SOFTENED TRUSS MODEL FOR TORSION
The softened truss model for torsion is a rational approach towards predicting the torsional response of an RC
member. It is based on the assumption that concrete softens in the process of formation of the tension cracks and
thus, the characteristic properties of concrete undergo a slight modification following the phenomena. The
model is based on satisfying the equilibrium, compatibility conditions and the uniaxial stress-strain relationships
of concrete and steel.
3.1 Equilibrium Equations
The equilibrium equations give a relation between the average internal stresses in the concrete , and in
the reinforcement , to the average applied stresses, , , with respect to the angle between the
longitudinal axis of the beam and the principal compressive direction . A 2-D element in the shear flow zone,
which is subject to shear stresses would satisfy the following equilibrium equations.
(1)
(2)
(3)
2 (4)
(5)
!
(6)
"
, = mild steel ratio in longitudinal and transverse directions respectively.

, = total areas of cross-section of longitudinal and transverse mild steels respectively.
# = perimeter of the centre line of the shear flow zone along a cross-section.
= area of the shear flow zone.
s = spacing between the longitudinal stirrups.
3.2 Compatibility Equations
The compatibility equations give a relationship between the average strains in the $ coordinate
system % , % , and the strains in the & ' coordinate system, % , % . Certain additional compatibility
equations are necessary to completely define a concrete strut. These additional equations are governed by the
out-of-plane warping and bending effects. The bending curvature of concrete struts, ((as in fig. 2.), is related to
626
the angle of twist, ), and the angle of inclination of crack to the longitudinal axis of the section, , by the
relation, ( = ) 2 .
The compatibility equations that show axis transformations are given
by:
% =% +% (7)
% =% +% (8)
* ⁄2 = −% + % (9)
Where %, is the strain in the - − ' &. .

* is the shear strain in the $ − coordinate plane.
Fig. 2. Warping in Diagonal concrete strut
3.3 Constitutive Laws
3.3.1 Softening Coefficient-
The softening of concrete compression stress block is attributed to the reduction in compressive strength of
concrete due to the formation of the diagonal tensile cracks. The strength is assumed to be reduced by a
coefficient, 0, called the Softening Coefficient, which measures the severity of cracking. 0 mainly depends on
the strain in the principal tensile direction, % and to a smaller extent, on% , the strain in the principal
compression direction and other variables.
ζ = 0.9/1 1 + 600% (10)
3.3.2 Average Compressive Stress-
The average compressive stress on the concrete struts is given by

,
= 56 0 7 (11)
Where 56 is a non-dimensional constant which can be defined as the ratio between the average stress and the
peak stress. It is a function of the maximum compressive strain in the concrete strut, % " and the softening
coefficient.
4. INCLUSION OF TENSION STIFFENING:
A close examination of the pre peak behaviour of an RC beam subjected to pure torsion suggests that this region
is dominated by the tension-stiffening effect. The deviation of analytical results from the experimental results
can be attributed solely to the tension stiffening effect i.e. the resistance contributed by the concrete between the
cracked portions to the diagonal concrete stresses. Addition of the tension stiffening effect demands a
modification in the principal tension stress,σ , equation which would be assumed to be equal to ‘zero’ in the
conventional code practices. And in an un-cracked member, the analysis is done based on St.Vetant’s theory.
Several more modifications are to be made to softened truss model to include the tension stiffening effect.
627
4.1. Elastic curve of concrete in tension
The cracking strain for concrete is very small, so most of the deformation in an RC member under tension will
occur at the cracks. By averaging the deformation across several cracks, the cracked concrete can be modelled as
a continuous material. Relationships for the average tensile stress in concrete, , were developed for shear panels
by Vecchio and Collins (1986) and by Belarbi and Hsu (1994). The average tensile stress relationships can be
used in a truss model to predict the behaviour before cracking by assuming that an apparent truss action is
occurring. In the model of an un-cracked panel under membrane stresses, compressive and tensile concrete
stresses will be induced in the principal directions. Before cracking, the stiffness of concrete in tension and
compression are nearly equal (Gopalaratnam and Shah 1985), so the predicted reinforcement stress will be small
and either tensile or compressive in order to satisfy equilibrium. For loads that induce a tensile strain greater
than the cracking strain, the relative stiffness of the concrete in tension is much less than the concrete in
compression, and the reinforcement stresses will tend to be tensile.
The STM was simplified by neglecting the tensile strength of concrete. While concrete acting in tension may
only have a small effect on the prediction of a member’s strength, it has a significant effect on the prediction of
a member’s full load-deformation response. This tension-stiffening response is related to the stress-strain
response of concrete acting in tension. The models for tension stiffening, were typically linear to 7 and ε7 ,
the stress and strain at concrete cracking, and then they decrease rapidly with further increase in tensile strain.
The functions considered in this research are linear, parabolic, and exponential and are out of the form by
Equations (12),(13),(14). The term 8 is a function of ε , and is given by the equation (15). The λ term in
equation (14) is a constant that controls the rate at which the function decays. The term ε7 9 in equation (15) is
the strain at which the tensile stress intersects the strain axis. Also, the parabolic function is tangent to the stress
axis at ε7 9 .
σ 7 1 8 (12)
σ 7 1 28 + 8 (13)
σ = 7 . :λ! ε; :ε<; (14)

ε; :ε<;
8 =ε (15)
<;= :ε<;
The three equations predict nearly the same σ ⁄ 7 upto a strain of 0.003 mm/mm. After that, the linear
equation quickly descends to zero at an ε of 0.0045, the parabolic equation has a slower descent to zero at an
ε of 0.007, and the exponential equation continues to predict a tensile stress of about 10% of the cracking
stress at a strain of 0.007.
The proposed tension stiffened model uses the basic equilibrium and compatibility equations developed from
the STM. Given the dimensions of the cross section, the reinforcement, and the material properties, a ‘’
displacement controlled ‘’ solution to the equations can be made by first selecting % and then assuming trial
values for % and . Next, an iterative procedure is used to find values for the assumed variables that solve the
equilibrium equations, compatibility equations, and stress-strain relationships. This will find one point on
torque-twist curve. Additional points are found by varying the selected value of % from a zero to maximum
value that causes peak torque or that causes to exceed a prescribed value. The limit for % which was
adopted in the program was, 0.1. The angle of inclination of diagonal cracks to the member’s longitudinal axis, ,
is included in many of the basic equilibrium and compatibility equations. However, using can cause
numerical instability in the iterative solution method so it will be eliminated from the solution equations
whenever possible.
628
6. TEST SETUP AND INSTRUMENTATION
Cyclic uniaxial bending, torsion, and combined bending-shear and torsion were generated by controlling two
horizontal servo-controlled hydraulic actuators shown schematically in Fig. 3. The axial load was applied by a
hydraulic jack on top of the load stubs; this jack transferred the axial load to the column via seven un-bonded
high-strength pre-stressed steel strands. Cyclic torsion, flexure and shear, and combined flexure, shear and
torsion were generated by controlling the force or displacement of two horizontal servo controlled hydraulic
actuators. A number of instruments were used to measure the applied loads, deformations, and internal strains.
The axial load was measured by a load cell between the hydraulic jack and the top of the load stub. Electrical
strain gages were attached to the surface of the longitudinal and transverse reinforcement to measure strains
permitting study of the deformation of reinforcement under different loading conditions. Testing of specimens
normally lasted a day or two. Although the data obtained was of high quality, further corrections were needed.
Correcting procedures included filtering the data to remove the noise of the excess data collected during the
pauses in the testing. The data was reduced and analyzed by comparing results for each column based on
various parameters.
Strong Wall Load Cell

Hydraulic Jack
Load Stub
Two Hydraulic
Actuators
Steel Strands
(Inside Column)
Test Unit
Support Blocks
Strong Floor
Fig. 3.elevation Fig. 4.applied forces
7. RESULTS
The TS-STM was validated by comparing the predicted response to the response from specimens TP-90, TP-91
and TP-92 tested by the Tokyo University [Tirasit and Kawashima], and one specimen from Missouri
University [Prakash et al. 2012] Only reinforced specimens were included in the comparison. Separate
predicted values were made using the linear, parabolic, and exponential models for tension stiffening. The
cracking torque and twist values obtained from the analytical results and experimental results, on comparison,
yielded an appreciable correlation. The observed error was due to material imperfections or the discrepancy in
loading steps.
Table 1: Comparison of Results

Specimen Ultimate Torque, >? Ultimate Torque, >?
(kN.m)- Experimental (kN.m)- Analytical
TP 90 70.8 72.9
TP 91 76.6 70.7
TP 92( axial force) 84.5 74.5
Missouri Specimen 328.1 322.9
629
Fig. 5. Specimen1, TP 90 Fig. 6. TP 90 section details
Fig. 7. Specimen2, TP 91 Fig. 8. TP 91 section details
Fig. 9. Specimen3, TP 92(axial stress) Fig. 10. TP92 section details
Fig. 11. Specimen4, Missouri Fig. 12. Missouri specimen section details
630
5. CONCLUSIONS
The proposed tension-stiffened model which is proposed in the paper is an improved version of the softened
truss model to account for the tension stiffened part of the primary curve. The model effectively assumes that
the possible mismatch in the experimental and the STM guided primary curve is due to the tension stiffening
behaviour of concrete. Though theoretically, it wasn’t possible to define a point from which no further tension
cracks could be developed, it was possible to obtain a healthy correlation between the theoretical and practical
data by considering that all the concrete offers resistance to tensile stress. The ultimate strength matched closely
with the experimental results. There was a difference in twist corresponding to ultimate load and the initial
cracking strength. The model can be improved further to predict the cracking strength and ultimate twist and is a
scope for further work.
6. REFERENCES
[1] Rao, G.T.D., Ramaseshu, D. (2005). “Analytical Model for the torsional response of steel fibre reinforced
concrete members under pure torsion.” Cement and Concrete Composites, Elsevier, Vol 27(4),493-501.
[2] ACI Committee 318 (2008). “Building Code Requirements for Reinforced Concrete.” ACI 318-08.,
American Concrete Institute, Detroit. MI.
[3] Hsu, T.T.C (1993). Unified Theory Of Reinforced Concrete, Library Of Congress Cataloguing, C.R.C.
Press., pp. 257-287.
[4] Choi, C.K., Cheung, S.H. (1996). “Tension Stiffening Model for Planar Reinforced Concrete Members.”
Computers & Structures, Elsevier Science Ltd, Vol. 59, pp.179-190.
[5] Xiao-Ning, Peng.,Yuk-Lung, Wong. (2011). “Behavior of reinforced concrete walls subjected to
monotonic pure torsion—An experimental study.” Engineering Structures, Elsevier Ltd, Vol. 33, Issue 9,
pp. 2495-2508.
[6] Lin-Hai,Hana.,Guo-Huang, Yaob, Zhong; Tao(2007); Performance of concrete-filled thin-walled steel

tubes under pure torsion; Thin-Walled Structures, Elsevier Science Ltd, 45 (2007), pp.24–36.
[7] You, Young-Min., Belarbi, A.(2011). “Thickness of shear flow zone in a circular RC column under pure
torsion.” Engineering Structures, Elsevier Science Ltd, 33(2011), pp. 2435-2447.
[8] Yon-Lee, Jung., Woo Kim, Sang.(2010).” Torsional Strength of RC Beams Considering Tension Stiffening
Effect.” Journal of Structural Engineering., ASCE, Vol. 136, No: 11, pp.1367-1378.
[9] Hsu, T.T.C (1984). Torsion of Reinforced Concrete., Library of Congress Cataloguing., Van nonstrand
Reinhold Company Inc., 1984.
[10] Prakash, S., Li, Q., Belarbi, A. (2012). “Behaviour of Circular and Square Reinforced Concrete Bridge
Columns under Combined Loading Including Torsion.” ACI structural journal, Title no: 109-S26, Volume.
109, No.3.
[11] Prakash, S.S. (2009). Seismic Behaviour of Circular RC Bridge Columns under Combined Loading
Including Torsion., PhD thesis., Department of Civil Engineering, Missouri University of Science and
Technology, Rolla, MO.
[12] Lampert, P., Thürlimann, B. (1968).”Torsion tests of reinforced concrete beams.” Rep. No. 6506-2.,Intitut
für Baustatik, ETH, Zurich, Switzerland.
631
[13] Vecchio, F.J., Collins, M. “The Modified Compression-Field Theory for Reinforced Concrete Elements
Subjected to Shear.” ACI Structural Journal. Title no: 83-22, 219-231.
[14] Greene, G.G., Belarbi, A.( 2009). “Model for RC Members under Torsion, Bending, and Shear.” Journal of
Engineering Mechanics, ASCE, V.135, No.9, Parts I and II.
[15] Robinson, J. R.,Demorieux, J. M.(1972). “ Essais de traction—Compression sur modeles d'ame de poutre
en Beton Arme) IRABA Report, Institute de Recherches Applliquees du Beton Arme.” Part 1, June 1968,
Part 2, “Resistance ultimate due Beton de Fame de poutres en double te en Beton Arme.” May (in French).
[16] Bredt, R. (1896). “Kritische Bemerkungen zur Drehungselastizitat,” Zeitschrift des Vereines Deutscher
Ingenieure, 40 (28) 785–790; 40 (29) 813–817 (in German).
632
Paper No.: 121
OPTIMAL EARTHQUAKE RESISTANT DESIGN OF A FIXED BEAM USING A

SIMPLE OPTIMIZATION TOOL
M. R. Das 1 and S. Samal 2
1
Associate Professor, Dept. of Civil Engineering, Institute of Technical Education and Research, SOA
University, Bhubaneswar, India
2
M. Tech Student, Dept. of Civil Engineering, Institute of Technical Education and Research, SOA
University, Bhubaneswar, India
Email: manasdas@soauniversity.ac.in, srikantakumarsamal@gmail.com
ABSTRACT:
The ultimate goal in structural design may be described in terms of a set of predefined functions which the structure
needs to perform satisfactorily and if possible with minimum cost including the initial construction cost as well as
the expected loss during its lifetime. During the past two decades, various mathematical optimization techniques
have been developed and widely recognized to be powerful tools for this purpose. In the present work, an attempt is
made for optimized design of an earthquake resistant fixed R.C.C Rectangular beam. In optimization, Simple
Optimization tool Microsoft Excel Solver has been used. For earthquake resistant design IS codes 456, 1893, 13920
etc are followed.
KEYWORDS: Solver, earthquake resistant, beam, optimization
1. INTRODUCTION
Optimization in general is making the best of anything or the condition being optimized. Mathematically, it is
defined as a mathematical technique for finding maximum or minimum value of a function of several variables
subjected to a set of constraints (Park et al, 2006; Beyer and Sendhoff, 2007; Tsompanakis, 2007). The ultimate
goal in structural design may be described in terms of a set of predefined functions which the structure needs to
perform satisfactorily and if possible with minimum cost including the initial construction cost as well as the
expected loss during its lifetime. During the past two decades, various mathematical optimization techniques have
been developed and widely recognized to be powerful tools for this purpose (Cornell et al., 2002, Beck et al, 1999).
In the practical application of the optimization methods to a structural design, the major considerations may be
stated as (i) Selection of the proper objective function to be minimized or maximized, (ii) Decision on the
constraints, that the structure has to satisfy, (iii) Choice of the proper variables,(iv)Nonlinear/linear programming
techniques, by which the optimal solution may be obtained (Lagaros et al, 2007; Foley et al., 2007; Lagaros et al.,
2004).
2. OPTIMAL DESIGN USING A SIMPLE OPTIMIZATION TOOL
For optimization an illustrative problem of design of a fixed beam having rectangular cross section has been
considered in the present study. For optimization a simple spread sheet based optimization tool has been used which is
elaborated in the subsequent sections. For the study the optimization problem has been formulated.
2.1 Microsoft excel solver as the simple optimization tool:
The Microsoft Excel Solver is being used for the purpose of optimization of modeling system. The Microsoft Excel
Solver combines the functions of graphical user interface (GUI), an algebraic modeling language like GAMS or
AMPL and optimizer for linear, non-linear & integer programs. Each of these functions is integrated into the host
spread sheet program as closely as possible (Schuëller 2006; Jiang et al., 2007).Optimization in Microsoft Excel
begins with an ordinary spread sheet model. The formula language functions of the spreadsheet are used as the
algebraic language to define the model. Through the solver’s GUI the user specifies an objective and constraints by
pointing & clicking with the mouse and filling in the dialog boxes. The solver then analyzes the complete
optimization model and produces the matrix form required by the optimizers. The optimizers employ the simplex
generalized reduced gradient and branch& bound method to find an optimal solution and sensitivity information.
The solver uses the solution values to update the model spreadsheet and provides sensitivity and other summery
information on additional report spreadsheet.
2.2 Problem formulation and methodology
For carrying out Optimization a problem of a rectangular fixed beam has been referred from literature (Duggal,
2007).
The design parameters given in the problem are as follows.
Distributed live load = 20 KN/m = 20 N/mm; Dead load = 25KN/m = 25 N/mm
Maximum bending moment due to earthquake = 60 KN-m = 60 x10 6 N-mm
Maximum shear force due to earthquake = 40 KN = 4X10 3 N
Centre to centre distance between supports = 6 m = 6 X 10 3 mm
M-20 grade concrete and Fe-415 steel has been used.
Design of section-
Assumed dimensions-
Width, b = 300 mm; Overall depth, D = 600 mm
Effective depth, d = D-effective cover = 600-40 = 560 mm,
Distributed load due to self-weight = b x D x 25 = 300 x 600 x 25 x 10 -6 = 4.5KN/m = 4.5 N/mm,
Total dead load, Wd = 4.5 + 25 =29.5 N/mm

Maximum bending moment due to dead load, Md = 2 9.5 x 60002/12 = 88.5 KN m, = 88.5 x10 6 Nmm
Maximum bending moment due to live load, ML= 20 x 60002/12 = 60 x10 6 N-mm,
Given, Seismic moment, SM = 60 x10 6 N-mm
Maximum shear force due to dead load, Vd = (29.5 x 6000) / 2 = 88.5x10 3 N
Maximum shear force due to live load, VL = (20 x 6000) / 2 = 60x10 3 N
Seismic design shear, SH=40 x 10 3 N
Factored moment, Mu = 1.5 (Md + ML) Or 1.2 (Md + ML + SM), whichever is more
634
Therefore, the factored moment is given as
Mu= [1.5x (88.5+ 60) 10 6 =222.75 X10 6 N-mm or 1.2x (88.5+ 60+ 60)] = 250.2X10 6 N-mm
Factored shear force, Vu = [1.5 (Vd+ VL) Or 1.2 (Vd+VL+ SH), whichever is greater;
Therefore, Vu can be calculated as
Vu = 1.5 (88.5 + 60) x 103 = 222.75x 10 3 N Or 1.2(88.5 + 60 + 40) x 103 = 226.2X103 N
For Fe 415 steel and M-20 concrete, fck=20 N/mm2, fy=415 N/mm2, Xulim= 0.48d, Xulim/d=0.48
= 0.36 1 − 0.42 (1)
By putting the above values = Qlim = 2.76
= (2)
!
dreqd= 549.6 mm, This value of dreqd is less than d and thus, it is all right
4.6
"# = 50 &1 − 1 − ( (3)
% ! '
After taking all the values Pt = 0.908%; As required = 0.94 x b x d = 1525.519 mm2
Providing 5 Nos. of 20 mm dia bars =1570 mm2 > 1525.519 mm2 all right
Pt1 = 100As / bd (4)
Pt1= 0.94%
As per IS 13920, for ductility requirement, Minimum percentage of steel
Pt min=0.24 (fck/fy) (5)
Pt min=0.24 (20/415) =0.0526%
Maximum percentage of steel, Ptmax=2.5%, So, 0.0526 %< 0.94 %< 2.5%, which is all right
+
Nominal shear stress
)* = (6)
!
Hence, )* = (226.2 x 1000) / (300 x 560) = 1.35 N/mm2
β = 0.8fck / 6.89Pt1 (7)
Permissible Shear Stress τc,
)1 = 0.85 340.8 5 (15 6) (8)
635
By substituting the values τc = 0.61 N/mm2, Maximum permissible shear stress, τc max = 2.8 N/mm2
from IS 456, Since τv > τc, shear reinforcement will be required.
Spacing required for 8- 2-legged vertical shear stirrups
Area of stirrups Asv
78* =
2 9 ∅8 '
9
4
Asv= 100.56 mm2
Spacing of stirrups=Sv
0.87 % 78
<* 10
)* )1 !
Sv = 163.49 mm c/c
Svmin = 0.87fyAsv / 0.4b (11)
Sv min= 302.47 mm
Maximum spacing, Svmax = 0.75d or 300mm (whichever is less) = 0.75 x 560 or 300mm = 300mm c/c
The spacing of shear stirrups should not be more than
(a) d/4 = 560/4 = 140 mm
(b) 8 times the diameter of the smallest longitudinal bar = 8 x 20 = 160 mm
Provide 8 two-legged vertical shear spacing at L1= 140 mm c/c over a length of 2d = 2 x 560 = 1120 mm at either
end of the beam. For the remaining portion, provide 8 at L2=200 mm c/c
Cost Analysis
Volume of Concrete= b x D x L = 300 x 600 x 6 x 103 = 1.08 x 109 mm3
Unit cost of M20 Concrete= Rs 8000/-
Cost of Concrete = 1 .08 x 109 x 8000 = Rs 8640.00
(12)
Quantity of main bars in Quintal = 0.769 Q
N1 = (2d/L1) + 1 = 18; N2 = (l-2 x 2d / L2) + 1 = 20; N = N1+N2 = 38 (13)
Quantity of stirrups = N Asv 2 (b + d) 78.5 / 109 (14)
Quantity of stirrups = 0.516 quintal
Total quantity of steel = Quantity of main bars + quantity of stirrups = 0.769 + 0.516 = 1.285 quintal
Unit Cost of steel per quintal = Rs 6500/- ; Cost of Steel = 6500 x 1.285 = Rs 8349.82
636
Total Cost of the beam per m3 = Rs 8640.00 + Rs 8349.82 = Rs 16989.82
Cost per meter length = Rs 16989.82 / 6 = Rs 2831.63
2.3 Optimization problem formulation
a. The objective function = Total cost of the beam
b. Design variables are
(1) Overall depth of the beam = 600 mm,
(2) Width of the beam = 300 mm,
(3) Cover of the beam = 40mm,
(4) Area of the steel =As= 1570 mm,
(5) Stirrups area = 100.5 mm2,
(6) Stirrups spacing Sv1= 140mm,
(7) Stirrups spacing Sv2= 200 mm
c. Design constraints are
(1) Sv1 < Svmin,

(2) d reqd < d,
(3) Pt required < Pt1,
(4) Pt1 < Ptmax,
(5) Pt1 > Pt min,
(6) Sv2 < Svmax,
(7) Sv2 > Sv min,
(8) Cover > 25 mm
Objective function cell is entered in appropriate place of solver dialog box. All the cells of design variables and
design constraints are also entered in appropriate places of solver dialog box. Then solver is invoked for
minimization of the objective function. Optimal solutions thus obtained after several runs are considered for
analysis.
3. RESULTS
The results obtained after spreadsheet based optimization both in case of simple design and seismic design are
presented in the Table 1.
637
Table 1: Comparison of design philosophies
RCC BEAM 1 2 3 4
DESIGN SEISMIC DESIGN SEISMIC
WITHOUT DESIGN WITH DESIGN
OPTIMIZATI WITHOUT OPTIMIZATI WITH
ON OPTIMIZATI ON OPTIMIZATI
Design variables Symbol unit ON ON
Overall Depth D mm 600 600 548.918 654.86
Width b mm 300 300 250 250
Cover c mm 40 40 25 25
Bar Dia φ mm 16 20 20 20
Bar Nos n 6 5
Area of steel provided Ast mm2 1205.76 1570 1121.384 1326.006
Stirrups dia φ mm 8 8 6
Stirrups Area(2
legged) Asv mm2 100.53 100.53 56 56
Spacing of stirrups
for L1 Sv1 mm 200 200 200 200
Spacing of stirrups
for L2 Sv2 mm 140 140 130.979 157.466
Design parameter
Characteristics st of
concrete fy N/mm2 415 415 415 415
Characteristics st of
steel fck N/mm2 20 20 20 20
Young’s modulus Es N/mm2 200000 200000 200000 200000
Dead Load DL N/mm 25 25 25 25
Live Load WL N/mm 20 20 20 20
Seismic Moment SM Nmm 0 60000000 0 60000000
Seismic Design Shear SH N 0 40000 0 40000
Cost of concrete Rs 8640 8640.00 6587.01338 7858.359506
Cost of steel Rs 7160.56 8349.82 5251.10 6082.40
Total cost Rs 15800.56 16989.82 11838.11 13940.76
Total cost per m
length Rs 2633.43 2831.64 1973.02 2323.46
638
Figure 3.1 Design philosophy vs. total cost Figure 3.2 Design philosophy vs. cost of concrete
Figure 3.3 Design Philosophy vs. Cost of steel Figure 3.4 Design Philosophy vs. Sectional area
Figure 3.5 Design Philosophy vs. Steel area Figure 3.6 % Steel vs. Total cost
639
4. DISCUSSION
The result obtained after spreadsheet based optimization both in case of simple design and seismic design are
presented in the Table 1. It is observed, in simple design with optimization the total cost per meter length is Rs.
1973.02 over a total cost of Rs.2633.43 of without optimization and thus effecting a saving of 25%. In seismic
design with optimization the total cost per meter obtained is Rs.2323.46 as compared to Rs.2834.64 for seismic
design without optimization which indicates a saving of 17.9%. Corresponding saving in cost of concrete is
obtained as 23.76% and 9% in case of simple design & seismic design respectively. Similarly, saving in cost of
steel is observed as 26.6% and 27.15% in case of simple design and seismic design respectively. Results of the
optimization are also presented in the form of Histograms as given in Figures 3.1, 3.2, 3.3, 3.4 and 3.5. From Figure
3.1, it is clear the total cost obtained after optimization incase of simple design is least as compared to the other
three. Even in case of seismic design with optimization the cost is much less as compared to simple design without
optimization. This will help the policy planner and Engineers to take decision defying the common fear that the
seismic design involves a huge cost. Obviously, in seismic design without optimization the cost involvement is the
highest. From Figure 3.2, it is clear that the cost of concrete in case of simple design with optimization is least. It
may be noted that the cost of concrete for simple and seismic design without optimization is same. It happened
because of same cross sectional area before optimization. Thus the cost of concrete before optimization does not
play any role for different type of design. From Figure 3.3, It is clear that the cost of steel in case of simple design
with optimization is minimum, where as cost of steel in case of seismic design without optimization is the highest
among all four types of design. After optimization the cost of steel reduces in both the cases i.e. simple and seismic
design with optimization. Even the cost of steel in case of seismic design with optimization is lower than simple
design without optimization. So such kind of result will definitely help the engineer to change their mindset at the
time of design. From Figure 3.5, it is clear that the area of steel in simple design and seismic design with
optimization are much lower than their corresponding values i.e. design without optimization. As such we can have
saving in both the cases as described earlier. Results are also presented in form of a graph as given in Figure 3.6.
The total cost is lowest at 0.8% of steel for seismic design with optimization.
5. CONCLUSION
For optimization of rectangular beam section considered in the present study, a simple optimization tool i.e.
Microsoft Excel based solver has been used which minimizes the total cost per meter length and determines the
optimal dimensions along with optimal percentage of flexure and shear reinforcement. In seismic design with
optimization there is a saving of 17.9% in total cost per meter length as compared to that obtained in seismic design
without optimization. In simple design with optimization, there is a saving of 25% in total cost per meter length
over simple design without optimization. Saving in cost of concrete is obtained as 23.765% and 9% in case of
simple design and seismic design respectively. Saving in cost of steel is obtained as 26.6% and 27.15 % in case of
simple design and seismic design respectively. The total cost per meter length obtained in seismic design with
optimization is much less as compared to simple design without optimization with a saving of 11.77 %. This will
help the policy planners and engineers to take a better decision.
REFERENCES
[1] Park, G.J.; Lee,T.H.; Lee, K.H. and Hwang, K.H.(2006).Robust design An overview. AIAA Journal, 44 (1),181-
191.
[2] Beyer, H.G.,;Sendhoff B. (2007). Robust optimization - A comprehensive survey. Computer Methods in
Applied Mechanics and Engineering, 196(33-34), 3190-3218.
640
[3] Tsompanakis, Y.,; Lagaros, N.D. and Papadrakakis,M. (Eds. (December 2007). Structural optimization
considering uncertainties, Taylor & Francis, ISBN: 9780415452601.
[4] Cornell, C. A,; Jalayer, F.,; Hamburger, R.O. and Foutch, D.A. (2002). Probabilistic basis for 2000 SAC
Federal Emergency Management Agency steel moment frame guidelines. J. Structural Engineering, 129(4),
526-533.
[5] Beck, J. L,; Chan, E.,; Irfanoglu, A. and Papadimitriou, C. (1999). Multi-criteria optimal structural design
under uncertainty. Earthquake Engineering and Structural Dynamics, 28, 741-761.
[6] Lagaros, N.D.,; Fragiadakis. M. (2007). Robust performance based design optimization of steel moment
resisting frames, J. Earthquake Engineering,11(5), 752-772.
[7] Foley, C.M.,; Pezeshk, S. and Alimoradi, A. (2007). Probabilistic performance-based optimal design of steel
moment-resisting frames. I: Formulation. Journal of Structural Engineering,133(6) , 757-766.
[8] Duggal, S.K.(2007). Earthquake Resistant Design of structures, Oxford University press publication.
641
Paper No.: 122
UNCERTAINTY PROPAGATION USING HYBRID HDMR FOR

STOCHASTIC FIELD PROBLEMS
1 2
Souvik Chakraborty and Rajib Chowdhury
1
Research Scholar, Dept. of Civil Engineering, Indian Institute of Technology Roorkee, Roorkee, India
2
Assistant Professor, Dept. of Civil Engineering, Indian Institute of Technology Roorkee, Roorkee, India
Email: csouvik41@gmail.com, rajibfce@iitr.ac.in
ABSTRACT:
A new uncertainty propagation method based on Hybrid High Dimensional Model Representation (HHDMR)
for problems with physical properties exhibiting spatial random variation has been proposed. The spatial
random variation of the physical properties has been modelled by random fields. High Dimensional Model
Representation (HDMR) is a general set of quantitative model assessment and analysis tools for capturing the
high-dimensional relationships between sets of input and output model variables. This paper presents a new
HHDMR-based approximation for the response functions of intermediate nature, meaning neither highly
additive nor multiplicative. The proposed approximation of an implicit response function includes both HDMR
and Factorized HDMR expansions through a hybridity parameter. The formalism of the HHDMR is similar to
the spectral stochastic finite element method in the sense that both of them utilize Karhunen–Loève expansion to
represent the input, and lower order expansion to represent the output. The method involves lower dimensional
HHDMR approximation of the system response, response surface generation of HHDMR component functions,
and Monte Carlo simulation. Once the approximate form of the system response is defined, statistics of the
response can be obtained by statistical simulation. The proposed approach decouples the finite element
computations and stochastic computations, and consecutively the finite element code can be treated as a black
box, as in the case of commercial software. Three numerical examples are used to illustrate the features of the
HHDMR and to compare its efficiency and accuracy with full scale simulation.
.
KEYWORDS: High Dimensional Model Representation, Karhunen–Loève Expansion, Random Field,

Stochastic Finite Element, Monte Carlo simulation
1. INTRODUCTION
Numerical computer codes implementing physics based models are the backbone of modern analysis tools of
complex engineering systems. Laboratory based controlled tests are often performed to gain insight into some
specific physics of a problem. Such tests can indeed lead to new physical laws improving a part of the overall
scientific model. Test data can also be used to calibrate a known model. However, neither of these activities are
enough to produce a credible numerical tool because of the several types of uncertainties which exist in the
whole process of physics based computational predictions. Such uncertainties include, but not limited to (a)
parametric uncertainty (e. g, uncertainty in geometric parameters, strength of the materials involved); (b) model
inadequacy (arising from the lack of scientific knowledge about the model which is apriori unknown); (c)
experimental error (uncertain and unknown error percolate into the model when they are calibrated against
experimental results); (d) computational uncertainty (e. g, machine precession, error tolerance and the so called
`h' and `p' refinements in finite element analysis) and (e) model uncertainty (genuine randomness in the model
such as uncertainty in the position and velocity in quantum mechanics, deterministic chaos). These uncertainties
must be assessed and managed for credible computational predictions.
In this paper we focus our attention to the systems with parametric uncertainties. In the parametric
approach, the uncertainties associated with the system parameters, such as Young's modulus, mass density,
Poisson's ratio, and geometric parameters are quantified using statistical methods and propagated, for example,
using the stochastic finite element method (SFEM) [1]. Methods involving perturbation expansion [2],
Neumann series expansion [3], first- and second-order reliability algorithms [4−6], and Monte Carlo simulation
(MCS) [7] including variance reduction strategies have been developed and extensively used for probabilistic
analysis of complex structures [8, 9]. These developments include treatment of spatially varying structural
properties that are modelled as random fields, loads that are modelled as space-time random processes,
structural behaviour covering static and dynamic regimes, linear and nonlinear structural/mechanical behaviour,
characterization of response variability and determination of structural reliability measures. Early development
of Spectral SFEM (SSFEM) by Ghanem and Spanos [1] appears to be a suitable technique for the solution of
complex, general problems in probabilistic mechanics. However, this method requires access to the governing
differential equations of system. Furthermore, the resulting system of equations to be solved for the unknown
response is much larger than those from deterministic finite element analysis. For complicated large system
problems, the system of equations in the SSFEM could be tremendously large. Recently a new implementation
of SSFEM [10-13], which is theoretically equivalent to the original SSFEM, has been developed for the purpose
of utilizing commonly available finite element (FE) software as a black-box. This novel implementation of
SSFEM uses random sampling of the input and consequently a large number of FE analyses to get a stable
estimate of the coefficients in the expansion of the solution.
In this paper, the hybrid HDMR (HHDMR) approach is developed with the view of becoming an efficient
solver for SFEM, incorporating cooperative effects via factorized HDMR (FHDMR). Using this approach, the
cost of computation is expected to be reduced significantly compared to the second-order HDMR [14], while
maintaining the desired accuracy. This methodology uses Karhunen–Loève (K–L) expansion [15] for
discretizing the input random fields and HHDMR to approximate the output responses in terms of functions of
lower dimensions.
Compared to the existing uncertainty propagation methods, the proposed approach has some key advantages:
• The first is reduction of data processing. The exponentially/factorially (e.g., in polynomial chaos based
SFEM) growing amount of function values is represented via polynomially growing tables [16, 17, 18].
This property helps us to cope with high dimensional problems of real world.
• The second advantage is reduction of computational complexity. HHDMR is generated by a family of
projectors [19, 20]. It allows splitting of any problem into easier low-dimensional sub-problems.
• Third advantage is that, HHDMR is a mean square convergent series expansion. This is because of the
inherent properties of orthogonal component functions [20].
• It is optimal in the Fourier sense since it minimizes the mean square error from truncation after a finite
number of cooperative terms in the expansion.
• Also lower dimensional functions stay stable as higher order Taylor series terms are inherently included.
• Finally, compared to existing simulation based random field approach (e.g., stochastic response surface
method [21]), HHDMR based approach does not translate into lower order polynomial approximation of
system response.
The formulation of the HHDMR is similar to the SSFEM in the sense that both of them utilize K–L expansion
to represent the input, and lower order expansion to represent the output. The sample points in the proposed
method are chosen from regularly spaced points along each of the variable axis. This sampling scheme results in
fewer system analysis [16–18].
Organization of the present paper as follows. In Section 2, overview of the theory behind the implementation of
HDMR hierarchies are explained. Section 3 provides a brief overview of Stochastic Finite Element Method
(SFEM). Two numerical examples are illustrated in section 4 to demonstrate the performance of the proposed
method.
2. HIGH DIMENSIONAL MODEL REPRESENTATION HIERARCHIES
The high dimensional model representation (HDMR) of an arbitrary M-dimensional function f ( x ) , x ∈ℜM can
be derived by partitioning the identity operator , I to be called I M in the M-dimensional case and I 1 in the 1D
case hereafter, with respect to the projectors P1 , P2 ,K , PM as follows [19, 20, 39-41]:
643
M
I M = ∏ ( Pm + ( I 1 − Pm ) )
m =1
M M M M
= ∏ Pm + ∑ ( I 1 − Pm ) ∏ Ps + ∑ ∑ ( I 1 − Pm )( I 1 − Ps ) ∏ P p
m =1
{ m =1 s≠m m =1 s = m +1 p ≠ m,n
144 42444 3 144444 42444444 3
1 term M  M 
  terms   terms
1 2
M M
+ L + ∑ Pm ∏ ( I 1 − Ps ) + ∏ ( I 1 − Pm )
m =1 s≠m m =1
144 4 2444 3 14 4244 3
 M  1 term
  terms
 M −1
(1)
composed of 2M mutually orthogonal terms. The orthogonal representation of Equation (1) is a manifestation of
the HDMR and can be rewritten as [41],
M M M
I M = L*x + ∑ L*x xi + ∑ ∑L *
x xi1 xi2 +L+ ∑ L*x xi xi2 L xiα + L*x x1 x2 L xM
1
i =1 i1 =1 i2 =i1 +1 1≤i1 ≤ i2 ≤L≤iα
(2)
where,
L*x = Lx
L*x xi = Lx xi − L*x
L*x xi
1
xi2 = Lx xi
1
xi2 (
− L*x x1 + L*x x2 + L*x )
L*x xi
1
xi2 L xiα = Lx xi
1
xi2 L xiα − ∑α
0 ≤ k ≤ −1
L*x xi
t1
xit
2
L xit
k
1≤ t1 < t2 <L≤α
(3)
and the notation x xi1 xi2 L xiα denotes x xi1 , xi2 ,K , xiα . ( )
The response function f ( x ) is represented in Equation (2) as few lower variate component functions of HDMR
in an additive form. However, when the response function f ( x ) is dominantly of multiplicative nature, HDMR
approximation may not be sufficient to accurately estimate the probabilistic characteristics of the system.
Multiplicative form of HDMR, termed as factorized HDMR (FHDMR), for a given multivariate response
function f ( x ) can be represented as [38, 39]
( ) ∏ (1 + M ) ×L× (1 + M )
M M
I M = M x* ∏ 1 + M x* xi *
x xi1 xi2
*
x x1 x2 L xM (4)
i =1 i1 ,i2 =1
i1<i2
644
*
where, the terms M x* , M x xi can be found as [42, 43]
M x* = L*x
Lx xi − L*x (5)
M *
x xi =
L*x
The hierarchical presentation in Equation (2) and Equation (4) is well suited for the response function which are
dominantly of additive nature and of multiplicative nature, respectively. But in most cases the response function
has neither additive nor multiplicative nature, rather it has an intermediate nature. In this section, hierarchy of
HHDMR methodology is presented for approximation of a response function. HHDMR expansion [44] includes
HDMR and FHDMR expansions in its structure through a hybridity parameter and can be represented as
 M M M 
I M = γ  L*x + ∑ L*x xi + ∑ ∑ L*x xi xi + L + L*x x1 x2 L xM  +
 
1 2
i =1 i1 =1 i2 = i1 +1
  (6)
( ) ( ) ( )
M M
(1 − γ )  M x* ∏ 1 + M x* xi ∏ 1 + M x* xi1 xi2 ×L × 1 + M x* x1 x2 L xM 

 i =1 i1 ,i2 =1
 i1<i2 
where, γ is the hybridity parameter, which controls the general structure of the response function. If γ = 1 , the
function expansion in Equation (6) merges to HDMR presented in Equation (2). Similarly, if γ = 0 , the
function expansion in Equation (6) merges to FHDMR presented in Equation (4). The most important step in
this expansion is to determine the hybridity parameter γ . For this purpose, a functional of the original response
function f ( x ) and the HHDMR approximation f% ( x ) is defined as
F ( x; γ ) ≡ f ( x ) − f% ( x )
2
(7)
Further details of estimating the hybridity parameter γ can be found in Tunga and Demiralp, 2006 [44].
Considering only univariate terms in Equation (6) yields,
   
( )
M M
f ( x ) = γ  L*x + ∑ L*x xi  + (1 − γ )  M x* ∏ 1 + M x* xi  + R 2 (8)
 i =1   i =1 
Therefore, using Equation (8) univariate HHDMR approximation f% ( x ) of the original response function
g ( x ) can be represented as
M   
f% ( x ) = γ  ∑ L x xi − ( M − 1) L*x  + (1 − γ )  ( L*x ) ∏ Lx xi 
M
M −1
(9)
 i =1   i =1 
645
2.1 Computational Flow
Step 1: Determine the term L*x , which is a constant term, representing the mean response. Select the reference
point, x .
Step 2: Interpolate univariate component function Lx xi . If for xi = xij , n function values
Lx xij
= f ( x1 ,K , xi −1 , xij , xi +1 ,K , xM ) ; j = 1, 2,K , n
(10)
the function value for arbitrary xi can be obtained using Lagrange interpolation as
n
Lx xi = ∑ φ j ( xi )Lx xij
j =1
(11)
where the interpolation function, φ j ( xi ) is defined as
∏ (x − x )
n
k
i i
φ j ( xi ) = k =1, k ≠ j
∏ (x − xik )
n
j
i
k =1, k ≠ j
(12)
By using Equation (11), Lx xi can be generated if n function values are given at corresponding sample points.
The same procedure shall be repeated for all univariate component functions, i.e., Lx xi ; i = 1, 2,K , M .
Step 3: Sum the interpolated values of HDMR expansion terms. This leads to univariate HDMR approximation
of the function f ( x ) as
M n
f% ( x ) = ∑∑ φ j ( xi ) Lx xi − ( M − 1) L*x
i =1 j =1
(13)
Step 4: Multiply the interpolated values of HDMR expansion terms. This leads to univariate FHDMR
approximation of the function f ( x ) as
M −1  
f% ( x ) = ( L*x ) ∏ ∑ φ j ( xi ) Lx xi 
M n
 i =1 j =1 
(14)
Step 5: Combine HDMR and FHDMR approximation using the hybridity parameter γ . This leads to univariate
HHDMR approximation of the function f ( x ) as
M n   M −1  
f% ( x ) = γ  ∑∑ φ j ( xi ) Lx xi − ( M − 1) L*x  + (1 − γ )  ( L*x ) ∏ ∑ φ j ( xi ) Lx xi  
M n
(15)
 
 i =1 j =1    i =1 j =1 
646
3. STOCHASTIC FINITE ELEMENT METHOD
Problems of structural mechanics in which uncertainty in specifying the mass and stiffness of the structure is
modelled within the framework of random fields can be treated using the SFEM [22]. Suppose H ( x , θ ) is a
random field with covariance function Γ H ( x1 , x2 ) defined in the probability space ( Θ, F , P ) . Random fields
are a useful tool to model random mechanical properties of engineering systems which are spatially distributed
in nature. This includes Poisson’s ratio, Young’s modulus, thickness, yield stress etc. In the case of the SFEM,
each random field involved is discretized by representing it as a set of random variables. Therefore,
H ( x,θ ) needs to be discretized in order to obtain the associated system of random algebraic equations. The
solution to these equations will enable uncertainty quantification for the particular system in concern. A wide
variety of discretization schemes are available in the literature. The discretization methods can be divided into
three groups: point discretization (e.g., midpoint method [23], shape function method [2, 24], integration point
method [25], optimal linear estimate method [26]); average discretization method (e.g., spatial average [27, 28],
dynamic weighted integral method [29-31], weighted integral method [31–34]), and series expansion method
(e.g., Karhunen–Loève expansion [15], orthogonal series expansion [35], expansion optimal linear estimation
[36]). An advantageous alternative for discretizing H ( x , θ ) is to represent it in a generalized Fourier type of
series as, often termed as K-L expansion,
∞
H ( x , θ ) = H ( x ) + ∑ λi ξi (θ ) ψ i ( x ) (16)
i =1
where {ξ (θ )}
i is a set of uncorrelated random variables. {λi } and {ψi ( x )} are the eigenvalues and
eigenfunctions of the covariance kernel Γ H ( x1 , x2 ) , satisfying the integral equation
⌠
 Γ H ( x1 , x2 )ψ i ( x1 ) dx1 =λi ψ i ( x2 ) (17)
⌡N
ℜ
In practice, the infinite series of Equation (16) must be truncated, yielding a truncated K−L approximation
N
H% ( x , θ ) ≅ H ( x ) + ∑ λi ξi (θ ) f i ( x ) (18)
i =1
which approaches H ( x , θ ) in the mean square sense as the positive integer N → ∞ . Finite element [37]
methods can be readily applied to obtain eigensolutions of any covariance function and domain of the random
field. For linear or exponential covariance functions and simple domains, the eigensolutions can be evaluated
analytically [37].
Once Γ H ( s ) and its eigensolutions are determined, the parameterization of H% ( x , θ ) is achieved by the
K−L approximation of its Gaussian image, i.e.,
 N

H% ( x , θ ) ≅ G  H ( x ) + ∑ λi ξi (θ ) f i ( x )  (19)
 i =1 
According to Equation (19), the K−L approximation provides a parametric representation of the H% ( x , θ ) and,
hence, of H% ( x , θ ) with M random variables. Note that this is not the only available technique to discretize the
random field H% ( x , θ ) . K−L expansion has uniqueness and error-minimization properties that make it a
647
convenient choice over other available methods. See [18] for a detailed study of the cited and other K−L
expansion properties. In particular, engineers are interested in obtaining the probability density function and the
cumulative distribution function of u in order to assess the reliability of the system. However, this objective can
prove to be difficult to achieve and the approach is limited to obtaining the first few statistical moments of u.
There exist several strategies, such as perturbation and projection methods, to deal with this problem. An account
of these methods is given in [35]. In this paper, vector valued lognormal random fields are considered.
3.1 HHDMR for Random Field Problems
A general procedure of HHDMR for random field problems constitutes the following key steps:
1. Representation of random field inputs in terms of standard random variables by Karhunen–Loève expansion.
2. Once the inputs are expressed as functions of the selected standard random variables, the output quantities can
also be represented as functions of the same set of standard random variables.
3. Estimate the hybridity parameter.
4. The model outputs are computed at a set of selected sample points and used to estimate the univariate HDMR
component functions.
5. Calculation of the statistics of the output which are cast as an approximate model in terms of HHDMR
expansion. The statistics of the response can be estimated with univariate HHDMR using either Monte Carlo
simulation or analytical approximation.
The implementation of the HHDMR approach in random field analysis is illustrated with the help of two examples
in this section. When comparing computational efforts in evaluating the statistics of responses, the number of
actual FE analysis is chosen as the primary comparison tool in this paper. This is because of the fact that, number
of FE analysis indirectly indicates the CPU time usage. For full scale MCS, number of original FE analysis is
same as the sampling size. While evaluating the statistics of responses through full scale MCS, CPU time is more
because it involves number of repeated FE analysis. However, in the present methods MCS is conducted within
HHDMR framework. Here, although the same sampling size as in direct MCS is considered, the number of FE
analysis is much less. Hence, the computational effort expressed in terms of FE calculation alone should be
carefully interpreted. Since all variants of HDMR approximation leads to explicit representation of the system
responses, the MCS can be conducted for any sampling size. The total cost of original FE analysis entails a
maximum of ( n − 1) × M + 1 by the present method, where n being the number of sample points of HHDMR and
M represents the number of random variables.
Example I: A Cantilever beam with random properties subjected to a point load
A cantilever beam subjected to a concentrated load P (Figure 1) is considered. It is assumed that the elastic
modulus
E (ξ ) = c1 exp α (ξ )  (20)
and thickness of the beam

t (ξ ) = c2 exp  β (ξ )  (21)
Figure 1. Cantilever beam with random properties subjected to a point load
648
are spatially varying in the longitudinal direction of the beam is an independent, homogeneous, lognormal random
fields. Mean and coefficient of variations of the random fields are µ E = 2.07 × 1011 N m 2 , µt = 0.254 m;
and υ E = 0.4 , υ t = 0.2 , respectively where,
c1 = µ E 1 + υ E2
c2 = µt 1 + υt2
(22)
α (ξ ) , β (ξ ) are zero-mean, homogeneous, Gaussian random field with variance
σ α2 = ln (1 + υ E2 )
σ β2 = ln (1 + υt2 )
(23)
and covariance function
Γα (ξ1 , ξ 2 ) = σ α2 exp ( − ξ1 − ξ 2 b )
Γ β (ξ1 , ξ 2 ) = σ β2 exp ( − ξ1 − ξ 2 b )
(24)
b is the correlation parameter that controls the rate at which the covariance decays. The degree of variability
associated with the random process can be related to its coefficient of variation. The frequency content of the
random field is related to the L/b ratio in which L is the length of the beam. A small L/b ratio implies a highly
correlated random field. L/b = 1 is chosen in this study. In this example, the length of the beam is L = 1 and it is
divided into 10 elements.
The eigensolutions of the covariance function are obtained by solving the integral equation (Equation (17))
analytically. The eigenvalues and eigenfunctions are given as follows:
2σ α2 b
λi =
ωi2 + b 2
2σ α2 b
λi* =
ωi*2 + b 2
cos ( ωiξ )
ψ i (ξ ) = for i = odd
sin ( 2ωiξ )
a+
2ωi
sin (ωi*ξ )
ψi (ξ ) = for i = even
sin ( 2ω ξ )
*
a−
i
2ωi*
(25)
The random fields are discretized into ten SRVs, and the ten-term K–L expansion (Equation (18)) is used to
generate sample functions of the input random fields. The response quantity which is the tip displacement of the
649
beam is represented by HDMR/FHDMR/HHDMR expansion with five sample points. To evaluate the validity of
the results obtained from the proposed method full scale MCS is performed for the same problem. Realizations of
the elastic modulus of the beam are numerically simulated using the K–L expansion method. For each of the
realizations, the deterministic problem is solved and the statistics of the response are obtained. Figure 2 compares
the probability density functions obtained from different variants of HDMR with full scale MCS. Excellent
agreements are observed between the proposed approach and 5×104 full scale MCS results. The cumulative
distribution function is shown in Figure 3.
Figure 2. Probability density for cantilever beam (in Figure 3. Cumulative distribution for cantilever beam
Figure 1) tip displacement. (in Figure 1) tip displacement.
Compared with the first

two response moments obtained using direct MCS
−2 −3
(m1 = 3.778 × 10 , m2 = 7.690 × 10 ) , present method using HHDMR approximation, overestimates the first
two moments by 0.22% (m1 = 3.776 × 10−2 ) and 0.95% ( m2 = 7.654 × 10 −3 ) , respectively. However, HDMR
approximation, underestimates the first two moments by 0.61% (m1 = 3.768 × 10 −2 ) and 1.41%
(m2 = 7.582 × 10 −3 ) , respectively, while FHDMR approximation, underestimates the first moment by 0.61%
(m1 = 3.755 × 10 −2 ) and 1.39% (m2 = 7.583 ×10 −3 ) , respectively.
The present method using different variants of HDMR approximation needs only 41 (M = 10 and n = 5;
(n−1)×M + 1 = 41) original FE analysis, while full scale MCS requires 5×104 number of original FE calculation.
In contrast, polynomial chaos-based SFEM requires 66 and 506 FE analysis using orders of polynomial 2 and 3,
respectively. This shows the accuracy and the efficiency (in terms of FE analysis) of the HDMR approximation
over direct MCS and polynomial chaos-based SFEM.
Example II: A rectangular plate with circular hole subjected to uniform load
A rectangular plate subjected to deterministic uniformly distributed load p is considered. In this example, making
use of symmetry, only a quarter of the plate is analysed, as shown in Figure 4. The elasticity modulus,
E ( y, z ) = c1 exp α ( y, z )  (26)

Poisson’s ratio
ν ( y, z ) = c2 exp  β ( y, z )  (27)
650
and thickness
t ( y, z ) = c3 exp γ ( y, z )  (28)
of the plate are considered as three two-dimensional independent, homogeneous, lognormal random fields.
Figure 4. Rectangular plate with circular hole
Mean and coefficient of variation of the random fields are µ E = 7.0 × 104 N m 2 , µν = 0.25 ,
µt = 0.254 mm and υ E = 0.4 , υν = 0.3 , υt = 0.6 , respectively. Where,
c1 = µ E 1 + υ E2
c2 = µν 1 + υν2 (29)
c3 = µt 1 + υt2
and covariance function
( − y1 − y2 )
by + ( − z1 − z2 bz )
Γα ( y, z ) = σ α2 e
( − y1 − y2 )
by + ( − z1 − z2 bz )
Γ β ( y, z ) = σ β2 e (30)
( − y1 − y2 )
by + ( − z1 − z2 bz )
Γγ ( y, z ) = σ γ2 e
by and bz are the correlation parameters at y and z directions, respectively. The loading is deterministic and p = 25
N/mm2. The plate is discretized into 350 elements in commercial FE code (ADINA) [46].
After obtaining the eigensolutions, the random field is discretized into five SRVs, and ten terms K–L
expansion is used to generate sample functions of the input random field. The response quantity, which is the
vertical displacement at a particular node (Point A), can be represented by HDMR expansion with five (n = 5)
sample points. Figures 5 and 6 show the probability distributions obtained from full scale MCS and variants of
651
HDMR with ten K–L variables. A good agreement is observed between the results from HHDMR-based model
and full scale MCS (1×104 samples).
Compared with the first two response moments obtained using full scale MCS
(m1 = 7.865 × 10−3 , m2 = 1.219 × 10−4 ) present method using HHDMR approximation, overestimates the first
two moments by 0.04% (m1 = 7.868 × 10−3 ) and 4.68% ( m2 = 1.276 × 10 −4 ) , respectively. HDMR
approximation, overestimates the first moment by 3.74% (m1 = 8.159 × 10 −3 ) and underestimates second
moment by 39.34% ( m2 = 7.394 × 10 −5 ) , while FHDMR approximation, underestimates the first two moment
by 0.08% (m1 = 7.859 ×10 −3 ) and 39.63% ( m2 = 7.359 × 10 −5 ) .
Figure 5. Probability density for the vertical Figure 6. Cumulative distribution for the vertical
displacement at point A (in Figure 10) of the plate. displacement at point A (in Figure 10) of the plate.
Further, the present method requires only 21 (M = 5 and n = 5; (n−1)×M + 1 = 21) ADINA FE analysis as
compared to 1×104 number of original FE calculation required for MCS. In contrast, polynomial chaos-based
SFEM requires 26 and 96 FE analysis using orders of polynomial 2 and 3, respectively. This shows the accuracy
and the efficiency (in terms of FE analysis) of the proposed approach over direct MCS and SSFEM.
5.DISCUSSION AND OUTLOOK
In this paper, HHDMR-based approximation methods are developed in a rigorous manner for structural
mechanics problem with random fields. The presented method couples the commercial FE codes and full scale
simulation through approximate models. Formally, Equations (15) can be interpreted as higher order response
surface for the response field, defined by means of input random field(s). In contrast with the usual response
surface methods (e.g. polynomial chaos), HHDMR approximation allows to define it at any order in a consistent
framework. The following points regarding the proposed method may be noted:
• There is a lack of efficient methods within the SSFEM framework to tackle nonlinear variations of random
fields. When the thickness and/or Poisson’s ratio is/are considered, (e. g, Example 1–2) as random field(s),
the nonlinear random field terms in the stochastic stiffness matrix are required to be discretized. This
problem is often simplified by considering an equivalent quantity (such as the bending stiffness of a plate
( )
D = Et 3 12 1 −ν 2 as random field. These types of approximations are done not based on physical data,
but purely for the sake of mathematical simplicity. The proposed approach does not require these often-ad
hoc simplified assumptions.
• HHDMR is a finite sum that contains 2M −1 number of HDMR component functions and N random variables.
In contrast, the polynomial chaos expansion is an infinite series and contains an infinite number of random
652
variables. Therefore, if the component functions in Equation (6) are convergent, then HHDMR
approximation provides a convergent solution.
• The terms in the polynomial chaos expansion are organized with respect to the order of polynomials. In
contrast, HHDMR is constructed with respect to the degree of cooperativity between a finite number of
random variables. If a response is highly non-linear and higher dimensional cooperative effects of multiple
random variables are rapidly diminishing then, HHDMR approximation is effective. This is due to the fact
that, lower-order HHDMR approximation can inherently incorporate nonlinear effects in system
characteristics by HDMR and bivariate cooperative effects via FHDMR. In contrast, many terms are required
to be included in the polynomial chaos expansion to capture high non-linearity.
• The proposed method is independent of the structural analysis (in the context of commercial software) as it
treats the FE model as a black-box. Therefore, the proposed approach can be treated as non-intrusive. If the
material/geometrical non linearity is introduced in the model response, then the appropriate nonlinear
analysis should be selected in FE software, and the underlying HHDMR-based model would automatically
be different.
• The amount of computation required for a given problem in SSFEM grows factorially, whereas, in the
proposed approach it varies polynomially.
In all the formulations developed in this paper, only the K-L expansion is used to discretize the input random
field(s). The use of other discretization schemes would, however, be possible and in some case more practical than
K-L (for instance when other correlation structures than that with exponential decay are dealt with). Future work
will address these issues.
This paper presents a new computational approach for problems in which physical properties exhibit spatial
random variation. The method appears to be efficient, requiring only conditional responses at selected sample
points to accurately compute solution statistics. HDMR approximation works well when the sought multivariate
response function has an additive nature. But, for multiplicative nature of the multivariate response function,
HDMR based method becomes expensive to obtain accurate result. At this juncture, the factorized form of HDMR
can be used for approximating the implicit response function. As an alternative, this paper presents a new
HHDMR based approximation to handle a multivariate response function that has neither additive nor
multiplicative nature but rather an intermediate nature. This method uses both HDMR and FHDMR expansions.
Numerical examples clearly demonstrated the accuracy and computational efficiency of the proposed HHDMR
based approximation. In comparison, full scale MCS may require thousands of realizations or more for converged
statistics. Thus the proposed method substantially reduces the computational effort while maintaining the desired
accuracy. However, K–L expansion requires for discretizing the input random field with many terms, which may
in turn increase the dimensionality of the problem and thus the computational effort of the SFEM. Unlike many
currently available methods, the curse of dimensionality does not arise in the proposed approach as it splits the
whole problem into a number of sub-problems.
Unlike local expansion based methods such as perturbation and Neumann expansion methods, HHDMR can
accurately handle random fields with moderate to large coefficients of variations. The proposed method is
independent of the details of structural analysis (in the context of commercial software) as it treats the FE model as
a black-box. If higher loads introduce nonlinearity in the model response, then the appropriate nonlinear analysis
should be used, and the underlying HHDMR based model would automatically be different. Thus the proposed
method is applicable to the analysis of linear as well as nonlinear structures, provided the right FE software is
used. Good agreements are observed comparing the numerical results between HHDMR-based approaches and
full scale MCS. In all the examples it was observed that the proposed approach uses less number of FE calls
compared to even second-order polynomial chaos based SFEM.
653
REFERENCES
[1] Ghanem, R. and Spanos, P.D. (2002). Stochastic Finite Elements: A Spectral Approach. Dover Publications
Inc., New York.
[2] Liu, W.K.; Belytschko, T. and Mani, A. (1986) Random field finite elements. International Journal of
Numerical Methods in Engineering, 23(10), 1831−1845
[3] Yamazaki, F. and Shinozuka, M. (1988) Neumann expansion for stochastic finite element analysis. Journal
of Engineering Mechanics. ASCE, 114(8), 1335−1354.
[4] Adhikari, S. (2004). Reliability analysis using parabolic failure surface approximation. Journal of
Engineering Mechanics ASCE, 130(12), 1407−1427.
[5] Adhikari, S. (2005). Asymptotic distribution method for structural reliability analysis in high dimensions.
Proceedings of the Royal Society of London, Series - A, 461(2062), 3141–3158.
[6] Ditlevsen, O. and Madsen, H.O. (1996). Structural Reliability Methods. Wiley, Chichester.
[7] Rubinstein, R.Y. (1981). Simulation and the Monte Carlo Method. Wiley, New York.
[8] Schueller, G. I. and Stix, R. (1987). A critical appraisal of methods to determine failure probabilities.
Structural Safety, 4(4), 293–309.
[9] Schuëller, G.I.; Pradlwarter, H.W. and Koutsourelakis, P.S. (2004). A critical appraisal of reliability
estimation procedures for high dimensions. Probabilistic Engineering Mechanics, 19(4), 463–474.
[10] Ghiocel, D. and Ghanem, R (2002). Stochastic finite element analysis of seismic soil–structure interaction.
Journal of Engineering Mechanics, ASCE, 128(1), 66–77.
[11] Nair, P. B. and Keane, A.J. (2002). Stochastic reduced basis methods, AIAA Journal. 40(8), 1653-1664.
[12] Sachdeva, S. K.; Nair, P. B. and Keane, A.J. (2006). Comparative study of projection schemes for
stochastic finite element analysis. Computer Methods in Applied Mechanics and Engineering, 195(19-22),
2371-2392.
[13] Sachdeva, S. K.; Nair, P. B. and Keane, A.J. (2006). Hybridization of stochastic reduced basis methods
with polynomial chaos expansions. Probabilistic Engineering Mechanics, 21(2), 182-192.
[14] Chowdhury, R. and Adhikari, S. (2010). High Dimensional Model Representation for Stochastic Finite
Element Analysis. Applied Mathematical Modelling, 34(12), 3917-3932.
[15] Van Trees, H.L. (1968). Detection, Estimation and Modulation Theory: Part 1, John Wiley & Sons, New
York.
[16] Chowdhury, R. and Rao, B.N. (2009). Assessment of High Dimensional Model Representation Techniques
for Reliability Analysis. Probabilistic Engineering Mechanics, 24(1), 100−115.
[17] Chowdhury, R.; Rao, B. N. and Prasad, A.M. (2009). High Dimensional Model Representation for
Structural Reliability Analysis. Communications in Numerical Methods in Engineering, 25(4), 301−337.
[18] Rao, B.N. and Chowdhury, R. (2009). Enhanced High Dimensional Model Representation for Reliability
Analysis. International Journal for Numerical Methods in Engineering, 77(5), 719 – 750.
[19] Rabitz, H.; Alis, O.F.; Shorter, J. and Shim, K. (1999). Efficient input-output model representations.
Computer Physics Communications, 117(1− −2), 11−20.
654
[20] Rabitz, H. and Alis, O.F. (1999) General foundations of high dimensional model representations. Journal
of Mathematical Chemistry, 25(2−−3), 197–233.
[21] Huang, S.; Mahadevan, S. and Rebba, R. (2007). Collocation-based stochastic finite element analysis for
random field problems. Probabilistic Engineering Mechanics, 22(2), 194–205.
[22] Sudret, B. and Der Kiureghian, A. (2000). Stochastic Finite Element Methods and Reliability: A
State-of-the-Art Report. Tech. Rep. UCB/SEMM-2000/08, University of California, Berkley, USA.
[23] Der Kiureghian, A. and Ke, J. B. (1988). The stochastic finite element method in structural reliability.
Probabilistic Engineering Mechanics, 3(2), 83-91.
[24] Liu, W. K.; Belytschko, T. and Mani, A. (1986). Probabilistic finite elements for non linear structural
dynamics. Computer Methods in Applied Mechanics and Engineering, 56(61-86).
[25] Matthies, G.; Brenner, C.; Bucher, C. and Guedes Soares, C. (1997). Uncertainties in probabilistic
numerical analysis of structures and solids - stochastic finite elements. Structural Safety, 19(3), 283-336.
[26] Li, C. C. and Der Kiureghian, A. (1993). Optimal discretization of random fields. Journal of Engineering
Mechanics, ASCE, 119(6), 1136-1154.
[27] Vanmarcke, E. (1983). Random fields: analysis and synthesis, The MIT Press, Cambridge, Massachussets.
[28] Vanmarcke, E. H. and Grigoriu, M. (1983). Stochastic finite element analysis of simple beams. Journal of
Engineering Mechanics, ASCE, 109 (5), 1203-1214.
[29] Manohar, C. S. and Adhikari, S. (1998). Dynamic stiffness of randomly parametered beams. Probabilistic
Engineering Mechanics, 13(1), 39-51.
[30] Adhikari, S. and Manohar, C. S. (1999). Dynamic analysis of framed structures with statistical uncertainties.
International Journal of Numerical Methods in Engineering, 44(8), 1157-1178.
[31] Adhikari, S. and Manohar, C. S. (2000). Transient dynamics of stochastically parametered beams. Journal
of Engineering Mechanics, ASCE, 126(11), 1131-1140.
[32] Deodatis, G. (1991). The weighted integral method, I: stochastic stiffness matrix. Journal of Engineering
Mechanics, ASCE, 117(8), 1851-1864.
[33] Deodatis, G. and Shinozuka, M. (1991). The weighted integral method, II: response variability and
reliability. Journal of Engineering Mechanics, ASCE, 117(8), 1865-1877.
[34] Takada, T. (1990). Weighted integral method in multidimensional stochastic finite element analysis.
Probabilistic Engineering Mechanics, 5(4), 158-166.
[35] Takada, T. (1990). Weighted integral method in stochastic finite element analysis. Probabilistic
Engineering Mechanics, 5(3), 146-156.
[36] Zhang, J. and Ellingwood, B. (1994). Orthogonal series expansion of random fields in reliability analysis.
Journal of Engineering Mechanics, ASCE, 120(12), 2660-2677.
[37] DiazDelaO, F.A. and Adhikari, S. (2008). Bayesian emulator approach for complex dynamical systems.
49th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics & Materials Conference, AIAA,
Schaumburg, IL, USA.
[38] Adhikari, S. (2007). Uncertainty quantification and propagation in structural dynamics. International
Conference on Civil Engineering in the New Millennium: Opportunities and Challenges, Howrah, India.
[39] Sobol, I.M. (2003). Theorems and examples on high dimensional model representations. Reliability
655
Engineering and System Safety, 79(2), 187–193.
[40] Li, G.; Rosenthal, C. and Rabitz, H. (2001). High dimensional model representations. Journal of Physical
Chemistry A, 105(33), 7765–7777.
[41] Ho, T.S. and Rabitz, H. (2003). Reproducing kernel Hilbert space interpolation methods as a paradigm of
high dimensional model representations: Application to multidimensional potential energy surface
construction. Journal of Chemical Physics, 119(13), 6433–6442.
[42] Tunga, M. A. and Demiralp, M. (2004). A factorized high dimensional model representation on the
partitioned random discrete data. Applied Numerical Analysis and Computational Mathematics, 1(1),
231−241.
[43] Tunga, M. A. and Demiralp, M. (2005). A factorized high dimensional model representation on the nods of
a finite hyperprismatic regular grid. Applied Mathematics and Computation, 164(3), 865−883.
[44] Tunga, M. A. and Demiralp, M. (2006). Hybrid high dimensional model representation (HHDMR) on the
partitioned data. Journal of Computational and Applied Mathematics, 185(1), 107−132.
[45] Huang, S.P.; Quek, S.T.; and Phoon, K.K. (2001). Convergence study of the truncated Karhunen–Loeve
expansion for simulation of stochastic processes. International Journal for Numerical Methods in
Engineering, 52(9), 1029–1043.
[46] ADINA R&D, Inc. (2005). ADINA theory and modelling guide. Report ARD 05−6.
656
Paper No.: 123
DESIGN AND DEVELOPMENT OF HIGH STRENGTH SELF COMPACTING

CONCRETE USING METAKAOLIN.
P. Dinakar1*, Manu S Nadesan2

1
Assistant Professor, School of Infrastructure, IIT Bhubaneswar, Bhubaneswar,India
2
PG Student, School of Infrastructure, IIT Bhubaneswar, Bhubaneswar 751 013. India
*Email: pdinakar@iitbbs.ac.in
ABSTRACT:
Metakaolin is a mineral admixture that appears to have significant potential for the production of high-strength
and high-performance concretes, if appropriately designed. Apart from other mineral admixtures metakaolin
produced under controlled conditions, its composition, white appearance, and performance are relatively
consistent. The use of metakaolin as a supplementary cementitious material (SCM) as well as fine filler is being
increasingly advocated for the production of High Strength High Performance Concrete (HSHPC) and Self
Compacting Concrete (SCC) in recent times. However, for obtaining the required high performance in any of
these concrete composites, metakaolin should be properly proportioned so that the resulting concrete would
satisfy both the strength and performance requirements of the structure. The present paper is an effort towards
presenting a new mix design methodology for the design of self compacting metakaolin concretes based on the
efficiency concept. The methodology has already been successfully verified through a proper experimental
investigation and the self compacting metakaolin concretes were evaluated for their self compactability and
strength characteristics. The results indicate that the proposed method can be capable of producing high quality
SCC.
KEY WORDS: Durability, Efficiency factor, HSC, HSHPC, Metakaolin, SCC, SCM,
1. INTRODUCTION
Self-compacting concrete (SCC) is one of the most important development in the area of construction industry in
recent years, primarily due to many advantages it possess both in the fresh and the hardened state [1, 2]. SCC
reduces honeycombing in the structure primarily because of the excellent filling ability and finally produces high
quality concrete structures [3]. The development of SCC commonly involves the use of industrial wastes such as
fly ash or slag [4–8]. However, metakaolin (MK) is relatively a new material in the concrete industry, which is
used as a mineral admixture in the development of high strength self-compacting concrete, is quite a new
approach in the area of the developments of concrete technology [9]. There are very few studies reported on the
development of SCC using metakaolin. This paper proposes a new mix design methodology for the development
of high strength SCC with metakaolin.
2. PROPOSED MIX DESIGN METHODOLOGY
This paper attempts to evaluate the efficiency of metakaolin in self compacting concrete at various replacement
percentages through the efficiency concept proposed earlier for the design of high strength metakaolin concretes
by using the efficiency factor “k” value [10] and suggest a methodology for incorporating metakaolin in self
compacting concrete. The efficiency factor (k) is generally defined in terms of its strength relative to control
concrete. The efficiency factor (k-value) is defined as the portion of the pozzolanic material such as fly ash,
slag, silica fume, metakaolin etc. that can be considered equivalent to Portland cement [11]. A value of k = 1
indicates that, in terms of the compressive strength performance, the performance of the part of the pozzolanic
material used is equivalent to cement. A value of k less than one indicates that the performance of the
pozzolanic material is inferior to cement. The quantity of the pozzolanic material is multiplied by the k value
to estimate the equivalent cement content, which can be added to the Portland cement content to determine the
resulting water to effective cementitious materials content ratio (w/(c+k*m)), required cement content, etc.
Earlier investigations have shown that metakaolin as a mineral admixture in concrete has the capability to
produce concretes of more than 100 MPa. However, as yet there is no specific mix design procedure available
for designing high strength self compacting concretes incorporating metakaolin based on the strength
characteristics. The present investigation is an effort to incorporate metakaolin into SCC by using a specific
mix methodology. The procedure of the proposed mix design methodology can be summarized in the
following steps:
Step 1: Fix the Total Cementitious or Powder Content for SCC

In the mix proportioning of conventional concretes, the water content is fixed based on the maximum size of the
aggregate and/or aggregate grading. In the case of SCC, the quantity of total fines (powder) is of importance.
In view of this fix the total cementitious materials (TCM) content (preferable to have this around 550 kg/m3).
To understand the behaviour of SCCs one can choose this in the range of 500-600 kg/m3 [12].
Let the TCM = TP kg/m3
Step 2: Fix the percentage of metakaolin and calculate the efficiency of metakaolin
Earlier Babu and Dinakar [10] had evaluated the efficiencies for the metakaolin concretes. The efficiency
factor (k) varied from 6.25 to 2.61 for the percentage replacements varying from 2.5 to 30% as shown in Figure
1. The corresponding relationship for the overall efficiency (k28) at 28 day for replacement levels varying from
2.5 - 30% proposed by Babu and Dinakar [10] is
k28 = 0.005813 p2 – 0.319017p + 7.01479 (1)

Where ‘p’ is the percentage replacement of metakaolin
The maximum compressive strength possible at the different percentage replacements, derived from the
results of Wild et al. [13] was also evaluated by author and presented in Figure.2. It can be seen from the
figure that a maximum compressive strength of about 83 MPa at 28 days is possible at 20% replacement level
and a maximum of 64 MPa at 5% replacement. The efficiency curve (Figure 1) and replacement percentages
possible at particular strength (Figure 2) were used in the present investigation to propose a mix design
methodology for the design of self compacting metakaolin concretes.
Let the metakaolin percentage be p %.
Cement content (cs) = TP (1-p) kg/m3 (2)
Metakaolin content (m) = TP (p) kg/m3 (3)
The efficiency of metakaolin at 28 days (k28) for replacement levels varying between 2.5 - 30% is given in
Figure 1. For a metakaolin replacement of p% the efficiency is calculated using equation (1).
Step 3: Calculation of water content in SCC

Now the water to effective cementitious materials content ratio of self compacting concrete with metakaolin is
calculated using ws / (cs + k28*m), where ‘ws’ is the water content of self compacting metakaolin concrete which
needs to be determined. With respect to any of the standard mix design procedures, the water cement ratio of
normal or conventional concretes (wn/cn) is chosen based on the compressive strength required using the
standard water cement ratio curves. In the present investigation the modified ACI relationship [14] was
utilized to find out water content (wn) required. The water cement ratio obtained for normal concrete shall be
used to determine the water content of self compacting concrete using the following relation
wn/cn = ws / (cs + k28*m) (4)
3
Therefore, ws = (wn/cn) (cs + k28*m) kg/m
658
Step 4: Determination of coarse and fine aggregate contents
Now the total aggregate content can be determined according to the absolute volume method. However, in the
present investigation a combined aggregate grading as recommended by the DIN 1045 standards was utilized.
The fine aggregate content in the total aggregate is generally recommended to be in the range of 48-55% [12].
Three types of coarse aggregates were used for these investigations.
In the combined aggregate grading for SCC let the percentage of fine aggregate in the total aggregate
content be x % and that of the coarse aggregate (CA) content be y% (CA1, mm = y1%, CA2, mm = y2% and CA3,
mm = y3%). This percentage of fine aggregate should be in correspondence with the proposed 48-55% range
for fine aggregate in SCC according to EFNARC standards [12].
Volume of fine aggregate (Vfa) = x% x Vagg
Mass of fine aggregate = Vfa x Gs, where Gs is the specific gravity of sand.
Volume of coarse aggregate (Vca) = y% x Vagg
Mass of CA1 aggregate = y1% x Vagg x Gca1, where Gca1 is the specific gravity of CA1
Step 5: Calculation of superplasticizer (SP) dosage

The chemical admixtures have the most profound impact on the behaviour of fresh SCC. Since for developing
self compacting concretes polycarboxylate ether (PCE) based admixtures are generally used and based on our
experience gained in our laboratory it was found that the dosages levels should be between 0.9 to 1.5% of the
total cementitious or powder content. Similarly, to attain stability or robustness to the mix viscosity modifying
agents (VMAs) are also used; the dosage levels of VMAs should be between 0.1 to 0.3% of the total
cementitious or powder content. If the dosage of SP used is equal to n % and that of VMA used is v % of the
total cementitious content (TP), then the dosages can be obtained as follows:
Dosage of SP used Wsp = n% (TP) (5)
Dosage of VMA used Wvma = v% (TP) (6)
Step 6: Trial mixtures and fresh tests on SCC

Trials mixtures can be carried out using the proportions calculated as above. Fresh property tests such as
slump flow, L-Box, V-Funnel tests should be carried out on SCC and they should comply with the specifications
of EFNARC.
Step 7: Adjustment of mixture proportion

If the results of the fresh tests mentioned above fail to meet the performance required, adjustments should be
made until all the properties of SCC satisfy the requirements according to EFNARC guidelines.
The validity of the mix methodology was carried out on three different concretes of strengths 80, 100 and 120
MPa which have been designed with the mix design methodology explained above for metakaolin replacements
varying between 7.5 to 22.5%. The mix details are presented in Table 1. The applied Ordinary Portland
cement (similar to ASTM Type I) and the metakaolin meet the requirements mentioned in IS: 12269 (53 grade)
and ASTM C618 respectively. Crushed granite with nominal size of 20 mm and good quality well-graded river
sand of maximum size 4.75 mm were used as coarse and fine aggregates, respectively. The different size
fractions of coarse aggregates (20 mm downgraded, 12 mm downgraded and 6 mm downgraded) were taken in
order to get a dense concrete. The specific gravities of aggregates were determined experimentally. The
coarse aggregates with 20, 12 and 6 mm fractions had specific gravities of 2.89, 2.87 and 2.88, whereas the fine
aggregate had specific gravity of 2.65, respectively. The high range water reducer (HRWR) used in this study
was a commercially available polycarboxylate ether (PCE). Commercially available viscosity modifying agent
(VMA) was also used. The fresh and hardened concrete is reported in Table 2 and Table 3 respectively.
659
Table 1. Mix details of the concretes developed
Total Aggregate,
Concrete TCM Cement kg/m3 Water SP VM A
Grade MK (c)
Name (k28) Coarse
(%) Sand
3 3 Aggregate
(MPa) kg/m kg/m kg/m3 kg/m3 (%) (%)
kg/m3
1 NC80 516 0 516 0 1513 348 160 0.40 0
80
2 SCC80 550 7.5 508.75 4.95 870 739 221 0.90 0.10
3 NC100 596 0 596 0 1465 337 155 0.60 0
100
4 SCC100 550 15 467.5 3.54 901 765 197 1.25 0.05
5 NC120 681 0 681 0 1415 325 150 0.90 0
120
6 SCC120 550 22.5 426.2 2.78 935 795 170 1.45 0.05
TCM – Total Cementitious Materials Content (Powder Content), MK - Metakolin

k - Efficiency of metakaolin, SP – Super plasticizer, VMA – Viscosity Modifying Agent
NC – Normal or Conventional Concrete, SCC – Self Compacting Concrete
Table 2. Fresh properties of the concretes investigated
S. Concrete Name Slump T500 Slump V – Funnel L- Box ratio

No Grade flow Flow Time For Gap of
40mm
(MPa) (mm) (sec) (mm) (sec)
1 NC80 110 -- -- -- --
80
2 SCC80 -- 3.5 680 21 0.87
3 NC100 100 -- -- -- --
100
4 SCC100 -- 5.2 660 25 0.81
5 NC120 100 -- -- -- --
120
6 SCC120 -- 6.8 650 28 0.78
Table 3. Compressive strengths of the concretes investigated
S. Concrete Name Compressive Strength

No Grade (MPa)
(MPa) 3 day 7 day 28 day 90 day
1 NC80 54.6 75.70 88.9 97.7
80
2 SCC80 54.9 75.76 94.1 101.2
3 NC100 60.9 84.48 103.2 109.5
100
4 SCC100 53.3 87.07 105.8 112.2
5 NC120 62.0 86.05 101.5 111.6
120
6 SCC120 57.7 94.25 107.5 121.2
660
6
28 day [Babu & Dinakar]
Efficiency of metakaolin at 28 days (k)

5
2
0 5 10 15 20 25 30 35
Metakaolin (%)
Figure 1. Variation of efficiency factor (k) with percentage replacement of Metakaolin
85
80
28 day Compressive strength (MPa)
75
70
65
Wild et al. (1996)
60
0 5 10 15 20 25 30 35
Metakaolin (%)
Figure 2. Maximum possible percentage replacement Vs compressive strength
As seen in Table 2, the highest T500 slump flow time was observed for 22.5% metakaolin content, also T500
slump flow time and V-funnel flow time of the SCC mixtures were influenced significantly by the metakaolin
content. For 80 MPa with 7.5% metakaolin content, T500 slump flow and V-funnel flow times were around 3.5
and 21s, respectively. These values were increased to around 5.2, 6.8 and 25,28s for 15% and 22.5% metakaolin
content respectively.
The compressive strength results of self compacting concretes and normal concretes are presented in Table 3. It
can be noticed that all the SCC containing metakaolin attained a low 3 day strength compared to normal concrete
but the strength gain rate of self compacting metakaolin concrete is higher than that of normal concrete from 7 day
onwards. The result of 80 MPa concrete shows that strength gain rate of SCC at 7.5% replacement was higher than
that of normal concrete from 28 day onwards. The results of self compacting metakaolin 100 MPa concrete
showed same strength gain rate to normal concrete and achieved target strength at 28 days and attained strength
higher than normal concrete at 90 days. Also the SCC120 had achieved its design strength at 90 day but the normal
concrete NC120 has not attained the target strength at 28 days by adopting low water cement ratio along with the
use of superplasticizer. The 120 MPa SCC with 22.5% metakaolin replacement has achieved slightly higher
strength than the corresponding normal concrete at 28 days but achieved strength of 121 MPa at 90 days. From
the results of high strength self compacting metakaolin concretes, it can be seen that, the strength gain rate after 28
days was quite high.
661
5. CONCLUSIONS
Using the proposed methodology and earlier established efficiency values for metakaolin, self compacting
metakaolin concretes of strengths ranging from 80 to 120 MPa, at various replacement levels ranging from 7.5 to
22.5% can be developed. The experimental investigations on self compacting metakaolin concretes designed
with the proposed mix design method, shows that the compressive strengths of the concretes obtained here surpass
very high strengths of 100 MPa at 28 days and 120 MPa at 90 days.
REFERENCES
[1] Parra C, Valcuende M.,Gómez F. (2011). “Splitting tensile strength and modulus of elasticity of
self-compacting concrete.” Constr Build Mater., 25, 201–207.
[2] Nanthagopalan P., Santhanam M. (2011). “Fresh and hardened properties of self compacting concrete
produced with manufactured sand.” Cem Concr Compos., 33, 353–358.
[3] Wang H.Y., Huang W.L.( 2010).” A study on the properties of fresh self-consolidating glass concrete
(SCGC).” Constr Build Mater., 24, 619–622.
[4] Khayat KH.( 1999). “Workability, testing, and performance of self-consolidating concrete.” ACI Mater J.,
96(3), 346–353.
[5] Lachemi M, Hossain KMA., Lambros V., Bouzoubaa N. (2003). “Development of cost-effective self-
compacting concrete incorporating fly ash, slag cement or viscosity modifying admixtures.” ACI Mater
J.,100(5), 419–425.
[6] Hassan AAA., Hossain KMA., Lachemi M.( 2008).” Behavior of full-scale self-consolidating concrete
beams in shear”.Cem Concr Compos.,30 (7), 588–596.
[7] Hassan AAA., Hossain KMA., Lachemi M.(2010).” Strength, cracking and deflection performance of large-
scale self-consolidating concrete beams subjected to shear failure.” Eng Struct., 32(5), 1262–1271.
[8] Uysal M., Yilmaz K. (2011). “Effect of mineral admixtures on properties of self compacting concrete.” Cem
Concr Compos., 33(7), 771–776.
[9] Hubertova M., Hela R. (2007).” The effect of metakaolin and silica fume on the properties of lightweight
self-consolidating concrete.” ACI Publication SP-2433, Detroit: American Concrete Institute, 35–48.
[10] Ganesh Babu K., Dinakar P. (2006d) “Strength Efficiency of Metakaolin in Concrete”, Structural Concrete,
Journal of the fib. Thomas Telford Publishers.,7(1), 27-31.
[11] Papadakis, V. G.and Tsimas, S.(2002). “Supplementary cementing materials in concrete Part I: efficiency
and design.” Cement and Concrete Research., (32), 1525-1532.
[12]. Self Compacting Concrete European Project Group. ” The European Guidelines for Self-Compacting
Concrete.” (2005).BIBM, CEMBUREAU, EFCA, EFNARC and ERMCO, <
http://www.efnarc.org>, (Jan.27, 2009).
[13] Wild, S., Khatib, J. M. and Jones A. (1996). “Relative strength, pozzolanic activity and cement hydration in
superplasticised metakaolin concrete.” Cement and Concrete Research., 26(10), 1537– 1544.
[14] Dinakar, P., (2012) “Design of Self Compacting Concrete with Fly ash.” Magazine of Concrete Research,
Thomas Telford Publishers., 64(5), 401-409.
662
Paper No.: 124
SEISMIC RELIABILITY ASSESSMENT OF R/C STEPPED FRAMES
Monalisa Priyadarshini1, Robin Davis2, Pradip Sarkar3 and Haran Pragalath D C1

1
Research Scholar, Dept. of Civil Engineering, National Institute of Technology Rourkela,
2
Assistant Professor, Dept. of Civil Engineering, National Institute of Technology Rourkela,
3
Associate Professor, Dept. of Civil Engineering, National Institute of Technology Rourkela
Email: monalisa.nitr@gmail.com, robin.davisp@gmail.com
ABSTRACT :
Stepped building’ frames are frames having increasing/decreasing order of number of storeys along the bays.
This type of building is common due to its advantages such as ventilation, architectural view, set back
limitations etc. Stepped buildings are characterised by staggered abrupt reductions in number of stories along
the bay width of the building. There will be change in mass, strength and stiffness (not necessarily at the same
rate) along the height of the building. The dynamic characteristics of these buildings are different from the
`regular' building due to height-wise changes in stiffness and mass. In the present study, typical RC regular
frames and stepped building frames are selected and their performances are evaluated by reliability index.
Reliability index is estimated by combining fragility curve and hazard curve of a selected site. Fragility curves
represents the probability of a system reaching a limit state as a function of intensity measure, such as peak
ground acceleration (PGA) or spectral acceleration corresponding to first natural period Sa(T1). The probability
of exceeding a specific damage state for the building is estimated using a log normal cumulative distribution
function as per Cornell et. al. (2002). Seismic hazard curve represents the probability of occurrence of an
earthquake as a function of PGA for a particular site. In the present study, a location in North eastern region in
India, Darjeeling is selected. The fragility curves and the reliability indices are found out for each building
considered. It is seen that regular frames are more fragile than stepped frames. The storey masses in the stepped
building configuration decreases as the height increases, which reduces the inertia force compared to a regular
frame, this makes the stepped building less vulnerable. Effect of infill increases the performance of the building
significantly.
KEYWORDS: Stepped buildings, non-linear dynamic analysis, fragility analysis, reliability index.
1. INTRODUCTION
Irregular buildings are classified into two groups in seismic design codes (IS 1893: 2002) as; irregular buildings
in plan and irregular buildings in elevation. A common type of irregular building in elevation is setback.
Setback buildings are characterized by staggered abrupt reductions in floor area along the height of the building
which result in drops in mass, strength and stiffness along height [Sarkar et al., 2010]. Several design codes
consider Stepped buildings as a vertical geometric irregular type of building that requires special design
consideration. As per IS 1893:2002, this type of buildings are to be treated as vertically irregular when the
lateral dimension of the maximum offset at the roof level exceeds 25% of the lateral dimension of the building
at the base. When the horizontal dimension of the building in any storey is more than 30% of that in the adjacent
storey, this building should be considered as vertically irregular as per ASCE 7: 2005. The design codes
recommend dynamic analysis for the design of this type of building category. Wood (1992) concluded through
an experimental study that the behaviour of stepped and set-back buildings is not much different from that of
regular ones. The ductility demand for set-back buildings is more, in the storeys above the set-back level and
hence needs special care while designing the top portions of the building, Aranda (1984). Sharooz and Moehle
(1990) found that inter-storey drift damage in the tower-base junction will be higher and need to be considered.
Pinto and Costa (1995) evaluated the non-linear behaviour of set-back buildings of 4, 8 and 20-storeyed
buildings. They observed a greater concentration of ductility demand in the lower storeys. Modern urban
construction has a tendency to provide irregular architectural configurations for buildings. The present study is
an attempt to assess the performance of typical configurations of setback buildings. The reference building is
considered as a regular building (without setbacks). Most of the previous studies employed bare frame analysis
for the study of setback buildings. The performance comparison is carried out first by a fragility curve and
subsequently through a reliability analysis considering seismic hazard at a selected site.
2. METHODOLOGY
2.1. Assessment of Seismic Reliability
A methodology for the assessment of seismic risk of building structures is presented by Ellingwood (2001). This
assessment involves three parts. First part is the identification of the seismic hazard, P [A = a], described by the
annual probabilities of specific levels of earthquake motion. In this study, hazard curve developed by Nath et al
(2013) for Darjeeling region is considered. Second part is the analysis of global response of the structural
system. The response analysis of the structure is carried out by conducting a nonlinear time history analysis for
different earthquakes, and the response is expressed in terms of maximum inter- storey drift at any storey. Third
part is the calculation of limit state probabilities of attaining a series of (increasingly severe) limit states, LSi,
through the expression:
P [LS i ] = ∑ P [LS i | A = a ]P [A = a ] (1)

a
A point that estimate of the limit state probability for state i can be obtained by convolving the fragility FR(x)
with the derivative of the seismic hazard curve, GA(x), thus removing the conditioning on acceleration as per
Eqn. (1).
P [LS i ] = ∫ F R ( x )
dG A (2)
dx
dx
The parameters at the fragility-hazard interface must be dimensionally consistent for the probability estimate to
be meaningful. The reliability index for corresponding probability of failure can be found by the following
standard Equation as shown below.
β = −φ − 1 ( pf ) (3)
Φ-1 is the inverse standard normal distribution.
2.2 Seismic Hazard Analysis
The seismic hazard at a building site is displayed through a complementary cumulative distribution function
(CCDF). The hazard function is the annual frequency of motion intensity at or above a given level, x, to the
intensity. Elementary seismic hazard analysis shows that at moderate to large values of ground acceleration,
there is a logarithmic linear relation between annual maximum earthquake ground or spectral acceleration, and
the probability, GA(a), that specifies values of acceleration are exceeded. This relationship implies that A is
described by the following equation,
G A ( x ) = 1 − exp[ − ( x / u ) − k ] (4)
u and k are parameters of the distribution. Parameter k defines the slope of the hazard curve which, in turn, is
related to the coefficient of variation (COV) in annual maximum peak acceleration.
664
0 0.5 1 1.5 2 2.5
Annual frequency of being

0.1000
exceedence 0.0100
0.0010
0.0001
PGA, g
Figure 1 Hazard curve for Darjeeling region (Nath,2013)
2.3. Development of Fragility curves:
A fragility function represents the probability of exceedance of the selected Engineering Demand Parameter
(EDP) for a selected structural limit state (DS) for a specific ground motion intensity measure (IM). These
curves are cumulative probability distributions that indicate the probability that a component/system will be
damaged to a given damage state or a more severe one, as a function of a particular demand. Fragility curve is a
damaged to a given damage state or a more severe one, as a function of a particular demand. Fragility curve can
be obtained for each damage state and can be expressed in closed form as using Eq. 5
P (C-D ≤ 0|IM) = Φ (5)

|
where, C is the drift capacity, D is the drift demand, Sdis the median of the demand and Scis the median of the
chosen damage state (DS). βd/IMand βc are dispersion in the intensity measure and capacities respectively. Eq.
5 can be rewritten as Eq. 6 for component fragilities (Nielson, 2005) as,
P (DS|IM) = Φ (6)
Where, IMm = exp , a and b are the regression coefficients of the probabilistic Seismic Demand
Model (PSDM) and the dispersion component, β ! is given as,
β ! = |
(7)
The dispersion in capacity, βc is dependent on the building type and construction quality. For βc, ATC 58 50%
draft suggests 0.10, 0.25 and 0.40 depending on the quality of construction. In this study, dispersion in capacity
has been assumed as 0.25.
It has been suggested by Cornell et. al (2002) that the estimate of the median engineering demand parameter
(EDP) can be represented by a power law model as given in Eq. 8.
& = a(IM+
EDP (8)
In this study, inter-storey drift (δ) at the first floor level (ground storey drift) is taken as the engineering damage
parameter (EDP) and peak ground acceleration (PGA) as the intensity measure (IM).
665
3. EXAMPLE FRAMES-DESIGN DATA
For assessing the fragility of vertically irregular buildings, a regular building with 8 storeys and 6 bays is chosen
as a reference (Fig. 2a). Geometries of the stepped buildings considered are shown in the Figs. 2b and 2c. The
performance of buildings is significantly influenced by the presence of infill walls. To study the effect of infills
in the performance of stepped buildings, the same buildings are modelled with infill walls. The infilled frames
and its designations are shown in the Fig. 2d, 2e and 2f. The buildings are assumed to be located in Indian
seismic zone V with medium soil conditions. The design peak ground acceleration (PGA) of this zone is
specified as 0.36g. The frame is designed as per prevailing practice in India. Seismic loads are estimated as per
IS 1893 (2002) and the design of the RC elements are carried out as per IS 456 (2000) standards. The
characteristic strength of concrete and steel were taken as 25MPa and 415MPa. The buildings are assumed to be
symmetric in plan, and hence a single plane frame may be considered to be representative of the building along
one direction. Typical bay width and column height in this study are selected as 3m and 3.2m respectively, as
observed from the study of typical existing residential buildings. The dead load of the slab (3 m x 3 m panel),
including floor finishes, is taken as 2.5 kN/m2 and live load as 3 kN/m2. The design base shear (VB) is
calculated as per IS 1893 (2002).
 Z I Sa 
V =   W (9)
2 R g 
Where, seismic zone factor, Z = 0.36, Importance factor I = 1.0, Response reduction factor R = 3.0.
6 bay @ 3m 6 bay @ 3m 6 bay @ 3m

8 storey @ 3.2m
BF ST 1 ST 2
8 storey @ 3.2m
FF FFST 1 FFST 2
Figure 2 Elevation of building frames considered
4. EARTHQUAKE DATA
The number of ground motions required for an unbiased estimate of the structural response is 3 or 7 as per
666
ASCE 7-05. However, ATC 58(2009) 50% draft recommends a suite of 11 pairs of ground motions for a
reliable estimate of the response quantities. A set of thirty IS 1893 (2002) spectrum compatible ground motions
are generated using WavGen (Mukherjee and Gupta, 2002) from the thirty far field natural time histories
collected from Haselton (2006).
IS-1893(2002)
spectrum
Figure 3 Response Spectra for 30 converted ground motions along with IS 1893 (2002)
design spectrum
5. MODELLING
The designed frames are modelled in the program Seismostruct (2012) for Time history analysis. Seismostruct
uses fiber based spread plasticity elements for frame elements. Mander’s (1988) confinement model is used for
concrete. For reinforcement steel, Menegotto and Pinto (1973) stress-strain relationship with Filippou’s
isotropic hardening rule is used. Rayleigh damping model is used with 3% damping in the first mode and 5%
damping in the third mode. Infill walls are modelled as per Crisafulli (1999) which is implemented in
Seismostruct.
6. MATERIAL UNCERTAINTY
To consider the uncertainty in the material properties, the characteristic strength of concrete, fck, yield strength
of the steel, fy and compressive strength of masonry are taken as the random variable. The statistical details
(Table 2) of the parameters, fck and fy have been taken from Ranganathan (1999). A set of thirty statistically
equivalent analytical models are generated using the Latin Hypercube Sampling (LHS) scheme.
Table 1: Details of random variables used in LHS scheme

Variable Mean COV (%) Distribution Remarks
Concrete fck (MPa) 30.28 21.0 Normal Uncorrelated
Steel fy (Mpa) 468.90 10.0 Normal Uncorrelated
Masonry fm (Mpa) 6.60 20 Normal Uncorrelated
7. CAPACITY MODEL
Limit states define the capacity of the structure to withstand different levels of damage. It can be represented
qualitatively (HAZUS) or quantitatively (FEMA 356). In this study, the inter- storey drift limits suggested by
Ghobarah (2004) has been taken as the median values of the capacity, Sc. The median inter-storey drifts for
light repairable damage (IO), moderate repairable damage (LS) and near collapse (CP) for moment concrete
frames are listed in Table 2. The dispersion in capacity, βc is dependent on the building type and construction
quality. For βc whereas ATC 58(2006) 50% draft suggests 0.10, 0.25 and 0.40 depending on the quality of
construction. In this study, dispersion in capacity has been assumed as 0.25. The details of the capacity model
considered in this study are listed in Table 2. In this present study, maximum of inter-storey drift (δ) at the
667
floor levels is taken as the engineering damage parameter (EDP) and peak ground acceleration (PGA) as the
intensity measure (IM).
Table 2 Damage limits and dispersion associated with various structural performance levels
Limit Median Interstorey
states Performance level Drifts Sc for Dispersion, βc
designation Concrete frames, (%)
IO Light repairable damage 1 0.25
LS Moderate repairable damage 2 0.25
CP Near collapse 4 0.25
6 DISCUSSION AND CONCLUSIONS

Fragility curves for building frames BF, ST1, ST2, FF, FFST1 and FFST2 for three performance levels namely,
IO, LS and CP are generated for reliability assessment. The Target Reliability Indices in accordance with FEMA
356 (2000) is used in the present study. The target reliability requirement for each performance level
(consequences of failure) for each relative cost of measures is mentioned in FEMA 356 (2000). The assessment
of performance of each building is carried out by comparing the reliability indices obtained for each building
with corresponding target reliability indices suggested for moderate level of relative cost measure. The target
reliability indices are taken as 2.3, 3.1 and 3.8 respectively for the performance levels IO, LS and CP.
Fig. 5 display variation of reliability index with the PGA for each frames considered. The PGA for 2%
probability of occurrence in 50years earthquake in the Darjeeling region is 1.6g (refere). The target reliability
corresponding to 2% probability of occurrence in 50 years earthquake is marked in Fig. 5. It can be seen that the
reliability indices of infill frames meet the target reliability at PGA of 1.6g. On the other hand, the bare frames
invariably fail to achieve the target reliability at PGA of 1.6g. This shows that infill frames that are neglected in
the analysis and design stage, comes to rescue the frames at peak PGAs. The reliability indices of stepped
buildings frames are only marginally different. The bare frames are found to be more reliable compared to the
stepped ones.
Occurrence in 50 years
2% Probability of
12 BF
ST 1
10 ST 2
Reliability Index, β
FFST 1
FFST 2
8 FF
.
6
Target Reliability
4
0
0 0.5 1 1.5 2 2.5
PGA, g
Figure 5 Reliability Index versus PGA for Collapse Prevention Limit State
668
REFERENCES:
[1] Aranda, G.R. (1984). “Ductility Demands for R/C Frames Irregular in Elevation”, Proceedings of the
Eighth World Conference on Earthquake Engineering, San Francisco, U.S.A., vol. 4, pp. 559-566.
[2] ASCE 7 (2005). Minimum Design Loads for Buildings and Other Structures, American Society of Civil
Engineers, USA. 2005.
[3] ATC 58 50% Draft, 2009. Guidelines for Seismic Performance Assessment of Buildings, Applied Technology
council, Redwood City, CA. 2002.
[4] B.G. Nielson, (2005). Analytical fragility curves for highway bridges in moderate seismic zones, Ph.D.
Thesis, Georgia Institute of Technology
[5] Bruce R. Ellingwood, (2001). Earthquake risk assessment of building structures, Reliability engineering and
system safety, vol. 74, pp. 251-262.
[6] C. Allin Cornell, Jalayer Fatemeh, O. Hamburger Ronald and A. Foutch Douglas, (2002). The probabilistic
basis for the 2000 SAC/FEMA steel moment frame guidelines., Journal of Structural Engineering, vol. 128
no.4, pp.526-533.
[7] C.B. Haselton and G.G. Deierlein, (2007). Assessing seismic collapse safety of modern reinforced concrete
frame buildings, Blume Earthquake Engineering Research CentreTechnical report no. 156, Stanford
university, pp. 313.
[8] F. J. Crisafulli. (1988). Seismic behaviour of reinforced concrete structures with masonry infills, Ph.D.
Thesis., University of Canterbury, New Zealand.
[9] FEMA 356 (2000). Pre-standard and Commentary for the Seismic Rehabilitation of Buildings, American
Society of Civil Engineers. USA.
[10] H.B. Kaushik, D.C. Rai and S.K. Jain, (2007). Stress-strain characteristics of clay brick masonry under
uniaxial compression, Journal of Materials in Civil Engineering. vol. 19, No. 9.
[11] IS 456 (2000), Indian Standard for Plain and Reinforced Concrete, Code of Practice, Bureau of Indian
[12] IS 1893 (2002). PART 1, Indian standard Criteria for Earthquake Resistance Design of Structures, Bureau of
Indian standards; New Delhi.
[13] J.B. Mander, M.J.N. Priestley and R. Park, (1988). Theoretical stress-strain model for confined concrete.,
Journal of Structural Engineering. vol. 114, pp. 1804-1826.
[14] Kumar Pallav, S T G Raghukanth and Konjengbam Darunkumar Singh, (2012). Probabilistic seismic hazard
estimation of Manipur, India, journal of geophysics and engineering, vol. 9, pp. 516–533.
[15] MatLab (2009), MatLab - Programming software for all kind of problems [online]. < http://
www.mathworks.com/ > (July 20, 2012).
[16] Nath, S. K., Adhikari, M. D., Maiti, S. K., Devaraj, N. “Seismic Hazard and Risk Microzonation of
Darjeeling-Sikkim Himalaya”, Natural Hazards, Under review (2013).
669
[17] Nielson, B.G, (2005). Analytical Fragility Curves for Highway Bridges in Moderate Seismic zones, Ph.D.
Thesis, Georgia Institute of Technology.
[18] Pinto, D. and Costa, A. (1995). Effect of vertical irregularities on seismic response of buildings. Tenth
European Conference on Earthquake Engineering. vol.II: 913:918.
[19] R. Ranganathan, (1999). Structural reliability analysis and design, Jaico Publishing House, Mumbai.
[20] S. Mukherjee and V.K. Gupta, (2002). Wavelet-based generation of spectrum compatible time-histories, Soil
dynamics and Earthquake engineering, vol. 22, pp. 799-804.
[21] Sarkar, P., Meher Prasad, A., Menon, D. (2010). Vertical geometric irregularity in stepped building
frames., J. Engineering Structures, 2175_2182.
[22] Seismostruct (2007) SeismoStruct– A Computer Program for Static and Dynamic Nonlinear Analysis of
Framed Structures [online]. < http://www.seismosoft.com/ >
[23] Shahrooz, B.M. and Moehle, J.P. (1990). Seismic Response and Design of Setback Buildings, Journal of
Structural Engineering, ASCE, vol. 116, No. 5, pp. 1423-1439.
[24] Sykora, M., Holicky, M. And Markova, J. (2011). Target reliability levels for assessment of existing
structures, M.H. Faber, J. Köhler And K. Nishijima, eds. In: Proc. ICASP11, 1-4 August 2011 , CRC
Press/Balkema, pp. 1048-1056.
[25] Wood, S.L. (1992). Seismic Response of R/C Frames with Irregular Profiles, Journal of Structural
Engineering, ASCE, vol. 118, No. 2, pp. 545-566.
670
Paper No.: 126
MAJOR INDIAN EARTHQUAKES AND EARTHQUAKE RISK ASSESSMENT

– A CASE STUDY
P.R. Ravi Teja1 and Sasmita Sahoo2
1
M. Tech (R) student, Department of Civil Engineering, National Institute of Technology, Rourkela
2
Department of Civil Engineering, Govt. College of Engineering Kalahandi, Bhawanipatna
E-mail-prraviteja04@gmail.com, sasmita.sahoo@gmail.com
ABSTRACT
India is considered as one of the most disaster prone countries in the world. It has experienced several
devastating earthquakes in the past resulting in a large number of death and severe property damage. An
earthquake also known as quake, tremor or temblor is the result of a sudden release of energy in earth crust that
creates seismic waves. Earthquakes are measured using observation from seismometer. The seismicity,
seismism or seismic activity of an area refers to the frequency, type and size of earthquake experienced over a
period of time. Earthquake causes ground motion in random fashion, both horizontally and vertically in all
directions from the epicenter. Consequently, structures founded in ground vibrate, inducing inertial forces on
them. It is therefore essential to ensure stability, strength and serviceability. The characteristics of seismic
ground vibration expected at any location depends upon magnitude of the earthquake, its depth of focus,
distance from epicenter, characteristics of the path through which seismic waves travels, and the soil strata on
which the structures stands. The random earthquake ground motions, which cause the structure to vibrate, can
be resolved in any three mutually perpendicular directions. The prominent direction of ground vibration is
usually horizontal. In this paper a study on major earthquakes that occurred in India between 1993 – 2005 are
dealt i.e. 1993 Latur earthquake, 1997 Jabalpur earthquake, 1999 Chamoli earthquake, 2001 Bhuj earthquake
and 2005 Kashmir earthquake and a comparison analysis is done on frequency, magnitude, focal depth, spectral
acceleration, displacement, time period and intensity values of the earthquakes. These values are necessary for
designing the structures. This Paper also deals with earthquake risk assessment of buildings briefly. A building
inventory database is also developed as a part of database preparation for risk assessment that can be used to
collect and store the parameters of buildings in an area. This database fulfills the earthquake assessment surveys
efficiently.
KEYWORDS: epicenter, focal depth, spectral acceleration, displacement, risk assessment and building
inventory database
1. INTRODUCTION
Rocks are made of elastic material, and so elastic strain energy is stored in them during the deformations that
occur due to the gigantic tectonic plate actions that occur in the Earth. But, the material contained in rocks is
also very brittle. Thus, when the rocks along a weak region in the earth’s crust reach their strength, a sudden
movement takes place there opposite sides of the fault (a crack in the rocks where movement has taken place)
suddenly slip and release the large elastic strain energy stored in the interface rocks. The sudden slip at the fault
causes the earthquake, a violent shaking of the earth when large elastic strain energy released spreads out
through seismic waves that travel through the body and along the surface of the earth. And, after the earthquake
is over, the process of strain build-up at this modified interface between the rocks starts all over again. This is
known as the elastic rebound theory.
Most earthquakes in the world occur along the boundaries of the tectonic plates and are called inter plate
earthquakes (e.g., 1897 Assam earthquake). A number of earthquakes also occur within the plate itself away
from the plate boundaries (e.g., 1993 Latur earthquake); these are called intra-plate earthquakes. In both types of
earthquakes, the slip generated at the fault during earthquakes is along both vertical and horizontal directions
(called dip slip) and lateral directions (called strike slip) with one of them dominating sometimes. Earthquakes
are measured using seismographs or seismometers. India lies at the northwestern end of the Indo-Australian
plate, which encompasses India, Australia, a major portion of the Indian Ocean and other smaller countries
(figure 1). This plate when collides with Eurasian plate causes subduction and thus earthquakes. Three chief
tectonic sub-regions of India are the mighty Himalayas along the north, the plains of the Ganga and other rivers,
and the peninsula. So, it is evident from this fact that these three zones are prone to earthquakes and as a result
of this we can notice many earthquakes in these regions only. India is divided into five seismic zones on the
basis of intensity levels they are I, II, III, IV, and V. In this paper ten major earthquakes are considered for
analysis and the correlation is achieved on certain parameters such as intensity, damage, focal depth, etc.
Figure 1: Geographical Layout and Tectonic Figure 2: Indian Seismic Zone Map (IS:
Plate Boundaries of India (Murty C.V.R., 2005) 1893, Part 1-2002)
2. ANALYSIS OF THE FIVE MAJOR EARTHQUAKES
In this paper the study was done on five earthquakes termed major on the basis of their intensities, magnitude
and damage caused. The earthquakes which were studied are-
• 1993 Latur earthquake
• 1997 Jabalpur earthquake
• 1999 Chamoli earthquake
• 2001 Bhuj earthquake
• 2005 Kashmir earthquake.
2.1 1993 Latur Earthquake
The Latur earthquake was one of a very rare type of earthquakes. It was what is referred to as a SCR, or a stable
continental region earthquake. Most earthquakes are a result of interaction between two plates, whether they be
sliding, colliding, or forming a subduction zone. However, in this instance, the cause is very complicated. The
Latur earthquake was an intraplate earthquake, or it occurred in the middle of a plate, as opposed to a plate
boundary. The earthquake’s epicenter was very far from any fault line. The cause of this earthquake is still in
speculation. Some scientists claim that it was a result of the force released from the continuous crumpling of the
672
Indian plate against the Eurasian plate. The theory which most scientists agree on is that the many lineaments, or
mini faults within plates, in that region contributed to the buildup of pressure and its consequent release.
The Latur, India earthquake was the most destructive earthquake in 1993. It occurred on September 30, 1993
and was having a magnitude of 6.4 Ma with mesoseismal area intensity varying between VII-IX. It had a
centroid focal depth of 5 km. Widespread death and destruction in the districts of Latur and Osmanabad,
Maharashtra state; complete destruction of stone /mud structures in about 20 villages covering an area about
15km wide centred 5 km west of Killari. Death toll was about 10,000 and highest relative death toll was 35% in
the villages of Killari and Chincholi-Rebe.
Figure 3:NE-SW trending lineament density contour and isoseismal contour maps in and around Latur region.
Contour interval is 10 km (Mallick et al., 2007)
2.2 1997 Jabalpur earthquake
The 1997 Jabalpur earthquake occurred on May 22, 1997 in Jabalpur District in the Indian state of Madhya
Pradesh. The epicenter of the earthquake was located at 23.18°N 80.02°E near Koshamghat village. The
earthquake has a magnitude of 6.0 Mw with an intensity of VII and it was having a focal depth of 35 km causing
death of 39 people.
The main cause of the earthquake in Jabalpur was due to the presence of Narmada fault in its proximity.The
zone is bounded by Narmada north fault to the north. Other major features include Tapti North fault, Gavilgarh
fault, and Tan shear (Dasgupta et al2000). 1997 Jabalpur earthquake is also associated with the Narmada South
Fault (Mandal et al 2000). The fault has been associated with neo-tectonic activities. The fault-plane solution of
the 1997 main-shock event suggested the tectonic model for the event to reverse-faulting with left-lateral strike-
slip motion.
Figure 4:- Isoseismal map of Jabalpur earthquake (Kayal, J.R.)
673
Figure 5:- Acceleration-time graph of Jabalpur earthquake (Kayal, J.R.)
2.3 1999 Chamoli earthquake
The 1999 Chamoli earthquake was the strongest to hit the foothills of the Himalayas in more than ninety years.
It occurred on 29 March 1999 in the Chamoli district in the Indian state of Uttar Pradesh now in Uttarakhand.
The epicenter of the earthquake was located at 30.408˚N 79.41˚E. The earthquake has a magnitude of 6.8 Mw
with an intensity of VIII and it was having a focal depth of 21 km causing death of 103 people.
Table 1: Characteristic parameters of 1999 Chamoli Earthquake (EERI, 1999)
Origin N (degree) 30-17-82 N

Origin E (degree) 79-33-84 E
Local (Richter) Magnitude (ML) 6.8
Surface Wave Magnitude (MS) 6.6
Body Wave Magnitude (mb) 6.3
Moment Magnitude (MW) 6.8
Fault Type Strike Slip
Fault Depth (km) 15
Fault Length (km) 30
Dip Angle (degree) 9
Circular Source Radius (km) 1.98 to 2.96
Epicenter Distance from Station (km) 80
Hypo central Distance from Station (km) 89.3
The Himalaya Range has been undergoing crustal shortening along the 2,400 km long northern edge of
the Indian Plate which resulted in the formation of several thrust planes including the Main Central Thrust
(MCT), the Main Boundary Thrust and the Main Frontal Thrust. The MCT consists of three sub-thrusts: MCT I,
MCT II and MCT III. Many earthquakes have occurred along these thrust faults. Hence, it is believed the
Chamoli earthquake in 1999 is associated with these fault systems. Liquefaction features formed under normal
and reverse density gradient system like flames, drops, pseudo nodules, ball-and-pillows, slumps and sand dykes
in inter layered sand and silt along with the faults are observed. These deformation features are widely
distributed over the area at different levels of sediment column in the region. This suggests a triggering agent of
more regional nature, like earthquake, was responsible for the formation of these liquefaction features.
674
Figure 6: Isoseismal map of Chamoli earthquake (DEQ-UOR, 1999)
Figure 7: Acceleration vs. Time graph of 1999 Chamoli earthquake at Baghpat (Pandey Y. et al)
2.4 2001 Bhuj earthquake
The Gujarat earthquake occurred on January 26, 2001; India's 52nd Republic Day, at 08:46 AM local time and
lasted for over two minutes. The earthquake reached a magnitude of between 7.6 and 7.7 on the moment
magnitude scale and had a maximum felt intensity of X (Intense) on the Mercalli intensity scale with epicenter
at 23.439°N 70.323°E.
2.4.1 Cause of Earthquake
Bhuj is near the boundary of the Indian and Eurasian plates. The Indian plate is moving northeastward and
collides with the Eurasian plate. Friction between the plates caused rock to break and slip, releasing energy in
the form of waves that shake the ground. About 20,000 people were killed and 167,000 more injured. Property
damage was also substantial and costly.
2.4.2 Aftershocks
A total of 101 aftershocks of magnitude > 3 have been recorded from the vicinity of the main event. Aftershocks
of magnitude > 5 took place within the first month of the main event. The activity died gradually during the
following weeks. All these aftershocks are classified into three categories ranging between M 3, 4, 4.1 to 5, and
> 5 magnitude. Eight aftershocks had a magnitude >5, a large majority had a range between 4.1 and 5, and 24
between 3 and 4.
675
2.4.3 Fault Line
Several new active fault traces were identified along Katrol Hill Fault. A new fault that extends into the Bhuj
Plain was also identified. New active fault traces were recognized about 1km north of the topographic boundary
between the Katrol Hill and the plain area. Primary and secondary surface deformation related to the 2001 Bhuj-
Kachchh earthquake suggests that thrusting movement took place along an E-W fault near the western extension
of the South Wagad Fault, a synthetic fault of the Kachchh Mainland Fault.
Figure 8: Acceleration vs. Time graph of Bhuj Earthquake (Bendick, R. et al)
2.5 2005 Kashmir Earthquake
The Kashmir earthquake occurred on 8th October 2005 was a major earthquake centered in the Pakistan-
administered Kashmir near the city of Muzaffarabad. The magnitude of the earthquake was 7.4 Mw with an
intensity of X-XI, located at depth of 26 km.
2.5.1 Cause of earthquake
This earthquake is associated with the known sub-duction zone of an active thrust fault in the area where the
Eurasian and Indian tectonic plates are colliding and moving northward at a rate of 40 mm/yr., giving rise to the
Himalayan mountain ranges, which has become prone to earthquakes.
3 DISCUSSION
The seismic values such as magnitude, intensity and destruction in terms of casualty are compared in Table 2. It
could be easily reviewed from this table about the most and least powerful earthquake in terms of energy
released and destruction respectively.
Table 2: Comparison of Five Major Earthquake
Max. Intensity
Date Event Time Magnitude (Mercalli index Deaths
scale)
30 Sep 1993 Latur 03:53 6.4 IX 7928
22 May 1997 Jabalpur 04:22 6.0 VIII 38
29 Mar 1999 Chamoli 12:35 6.8 VIII 63

26 Jan 2001 Bhuj 08:46 7.7 X 13805
08 Oct 2005 Kashmir 09:20 7.4 X 20600
676
Table 3: Comparison of intensity, acceleration, velocity perceived shaking and potential damage.
Instrumental Acceleration Velocity Perceived Potential

Intensity (g) (cm/s) Shaking Damage
I < 0.0017 < 0.1 Not felt None
II-III 0.0017 – 0.014 0.1 – 1.1 Weak None
IV 0.014 – 0.039 1.1 – 3.4 Light None
V 0.039 – 0.092 3.4 – 8.1 Moderate Very light
VI 0.092 – 0.18 8.1 – 16 Strong Light
VII 0.18 – 0.34 16 – 31 Very strong Moderate
Moderate to
VIII 0.34 – 0.65 31 – 60 Severe
heavy
IX 0.65 – 1.24 60 – 116 Violent Heavy
X+ >1.24 > 116 Extreme Very heavy
3.1 comparisons of Magnitudes
Magnitude measures the energy released at the source of the earthquake. It is determined from measurements on
seismographs. Figure 9 presents the bar chart for earthquake magnitude for five selected earthquakes.
Figure 9: Comparison of magnitude for five selected earthquakes
3.2 Study on seismic data
Eq. (1) is used to convert spectral acceleration to displacement for a given seismic period (NIBS, 2002)
SD = 9.8 × × (1)
Where
SA = Spectral Acceleration (g)
SD = Spectral Displacement (inches)
T = Time Period (s)
677
In this paper, the ground motion data of 1999 Chamoli Earthquake at 1 HZ ,3 HZ, 5HZ and10 HZ at 5% damping
are used to find the spectral acceleration and displacement. Table 4 presents the values of spectral displacement
calculated using Eq. (1) corresponding to spectral acceleration and fundamental period. Spectral acceleration
and spectral displacement plot is shown in Figure 10.
Table 4: Spectral displacement and spectral acceleration as a function of fundamental period (Ranjan, 2005)
Frequency Spectral Acceleration (g) Spectral Displacement

Time Period (sec)
(HZ) (inches)
1 1.00 0.08 0.784
3 0.33 0.43 0.467
5 0.20 0.31 0.121
10 0.10 0.20 0.020
Figure 10: Spectral Acceleration and Displacement Graph for 1999 Chamoli Eartquake in Dehradun
3.3 Comparison of focal depth
Focal depth refers to the depth at which a slip occurs and earthquakes are also classified on the basis of this.
Slip occurring at the following depth used to classify the type of earthquake:-
• < 70 Km – Shallow focus earthquake

• 70–300 Km – Mid focus earthquake
• > 300 Km – Deep focus earthquake
678
Figure 11: Comparison of ‘focal depth’ of selected earthquakes
3.4 Comparison of Intensity
Intensity measures the strength of shaking produced by earthquake at a certain location. It is determined from
effects on people, human structures and natural environment. Table 5 shows intensity of five selected
earthquakes and its abbreviated MMI for different place of earthquakes.
Table 5: Intensity and its abbreviated MMI for different place of earthquakes
Place Intensity Abbreviated MMI

Damage slight in specially designed structures;
considerable damage in ordinary substantial
Chamoli earthquake, Jabalpur buildings with partial collapse. Damage great in
VIII
earthquake poorly built structures. Fall of chimneys, factory
stacks, columns, monuments, walls. Heavy furniture
overturned.
Damage considerable in specially designed
structures; well-designed frame structures thrown out
Latur earthquake IX
of plumb. Damage great in substantial buildings, with
partial collapse. Buildings shifted off foundations.
Some well-built wooden structures destroyed; most
Bhuj earthquake, Kashmir
X masonry and frame structures destroyed with
earthquake
foundations.
4. EARTHQUAKE RISK ASSESSMENT OF BUILDINGS (BY DATABASE PREPARATION)
For assessment of risk in buildings due to earthquake, database preparation is very vital. In fact this is a tough
task, but, with the advent of new technology and software it is not much difficult. Database preparation involves
two phases- data collection and data preparation. HAZUS methodology is used for database preparation.
The methodology involves four basic steps-

• Review of risk assessment methods in India and other countries
• Identification and generation of dataset such as seismic, ground motion, building response and damage
functions required for using HAZUS methodology in a study area.
679
• Possible modification to use HAZUS based building classification in a study area in India with
consulting research institutes and experts in that area.
• Testing of HAZUS methodology in the study area.
4.1 Data collection
This deals with collection of data required for earthquake vulnerability assessment and also data required as an
input for HAZUS model in Indian context for generating database.
4.1.1 Institutional data collection
For data collection mainly research institutions such as EERI, CBRI, EREC and etc. present in a study area are
surveyed by consulting experts and other institutes such as Survey of India are also consulted for map study.
4.1.2 Field collection
This deals with mainly field surveys, prior to that a field work plan is prepared and survey is done according to
the plan. Data preparation involves - field survey, reconnaissance survey, detailed survey
4.2 Data Preparation
Data preparation mainly deals with Remote sensing data, satellite images, software and Building inventory.
4.2.1 Satellite Images
IKONOS satellite images are used which works on the principle of push broom and simultaneously provides
panchromatic and multispectral images.
Table 6: Description of remote sensing data used for research in Dehradun (Brijesh, 2006)
Ground Resolution Projection

Satellite Image Acquiring Date Study Area
(m) System
IKONOS PAN
05 May 2005 1.0 Dehradun, India UTM, WGS84
2005
IKONOS MS
05 May 2005 4.0 Dehradun, India UTM, WGS84
2005
IKONOS PAN
19 April 2001 1.0 Dehradun, India UTM, WGS84
2001
IKONOS MS
19 April 2001 4.0 Dehradun, India UTM, WGS84
2001
4.2.2 Software
The software that is to be used for processing of data are-
• Arc View 3.2

• Arc GIS 8.2
• ERDAS Imagine
• MS Office-Word and Access
680
4.2.3 Building Inventory
Individual building data or maps can be made for a city using the data collected through the field survey of
study ward. For such works, a database is prepared using MS access as shown in figure 12, which contains
different parameters that are filled by survey of individual building. A laptop can be used in filling the data of
the buildings in data entry form in MS access.
The large printouts of merged IKONOS (satellite) image data with building footprints and building ID for each
block within the city could be taken for fieldwork along with a data entry form prepared in MS Access to
generate the building database. In this way, individual building layers are generated editing the existing digital
footprint map.
Table 6: Building Inventory parameters
Parameters in the inventory Sub parameters/classes
Building ID Address of the building

Group A (Hotels, Temples etc.), Group B (Banks, offices etc.), Group E
Occupancy class (Educational institutions), Group F (factory, industry), Group I (Hospital
and other institutes), Group R (Residential).
Agricultural, commercial, residential, educational, industrial, military,
Occupancy type
religious.
Building shape/plan Rectangular, Circular or any other
Structure type Frame, unframed.
Load path Yes/No
Wall material Bricks, logs, cob, stone cladding, breeze blocks, mud bricks.
Roof material Thatch, shingle, slate, concrete (PCC, RCC), asbestos, brick tiles.
Roof type Flat, terrace, gabbled, tented, hipped, pitched.
Height/base Vary as per building
Soft storey Yes/No
Floor height Vary as per building
Proximity Vary as per building
Maintenance Periodic, Irregular, No Maintenance.
Construction age Very old, old, Recent.
Construction year Vary as per building
Earthquake resistance Yes/No
Set back Vary as per building
Front road width Vary as per building
681
The building parameters like (i) Building Shape (ii) Proximity between two adjacent buildings are determined
through onscreen visual interpretation using Arc GIS 8.2 and Arc View 3.2. All the building inventory data are
stored and assigned to specific buildings and thus a complete database for earthquake vulnerability could be
prepared.
Figure 12 : Building inventory database
682
5. CONCLUSION
In a nutshell, the entire paper deals with two vital points viz. case study on some of the major earthquakes of
India and a bird’s view on earthquake risk assessment of buildings by database preparation. India lies at the
north western end of the Indo-Australian plate, which encompasses India, Australia a major portion of the
Indian Ocean and other smaller countries. This plate when collides against the huge Eurasian plate and goes
under the Eurasian plate, this process of one tectonic plate getting under is called subduction. This subduction
causes earthquakes in Himalayan region and fault planes collision causes earthquakes in peninsular region.
• From the study of earthquakes it can be concluded that 1999 Chamoli Earthquake is the inter-plate
earthquake and 1993 Latur earthquake, 1997 Jabalpur earthquake, 2001 Bhuj earthquake and 2005 Kashmir
earthquake are the intra-plate earthquakes.
• Graphs on comparison of magnitude, focal depth, intensity and spectral acceleration and displacement for
1999 Chamoli earthquake were plotted using MS Excel, these graphs can be interpreted to quickly analyse
the most and least destructive earthquakes respectively.
• Earthquake risk assessment is also a vital component of earthquake study. The overall procedure is also
explained in this paper briefly. One of the important aspects of assessment is database preparation, which
was done by using MS Access that could be filled easily at the survey site.
REFERENCES
[1] Agarwal, P and Shrikhande, M (2012) , ” Earthquake Resistant Design Of Structures ”, PHI Publication,
Edition 2012, PP 50-221.
[2] Benedick, R., et al., “The January 26, 2001 Bhuj Earthquake, India”.
[3] EERI, 1999. The Chamoli, India, Earthquake of March 29, 1999, Earthquake Engineering Research
Institute (EERI), Oakland, CA.
[4] EERI(2002), “Bhuj, India Earthquake of January 26 , 2001 :Reconnaissance Report”, Earthquake Spectra,
Supplement to Vol. 18, July ,2002
[5] Gulati, B. (2006), “Earthquake risk assessment of buildings: Applicability of HAZUS in Deradun, India”
thesis, pp 79.
[6] Jain, S.K., 2000. Historical Developments And Current Status of Earthquake Engineering in India, 12th
World Conference on Earthquake Engineering, Auckland, New Zealand, pp. 13.
[7] Kayal, J.R., Large and Great Earthquakes in the Stable Continental Region of India: Seismotectonic
Perspective.
[8] Mallick, K., Vasanthi, A. et al, (2007) “Demarcation of liquefaction zones by Bouguer Gravity and
Electrical Resistivity method”, Indian Geophysics Union Journal, vol.11, No.4, pp.175-180.
[9] Murty, C.V.R., 2005. “Earthquake Tips: Learning earthquake design and construction.”
[10] NIBS, 2002. HAZUS’99: earthquake loss estimation methodology: technical manual. In: FEMA (Editor),
683
[11] NICEE, 2005. National Information Centre for Earthquake Engineering. www.iitk.ac.in/nicee/. Indian
Institute of Technology, Kanpur.
[12] Pandey, Y., Dharamaraju, R., Chauhan, P.K.S., (2001) “Estimation of source parameters of Chamoli
Earthquake, India”.
[13] Rai, D.C., Narayan, J.P., Pankaj and Kumar, A. (1997).”Jabalpur earthquake of May 22, 1997:
Reconnaissance Report”, Dept. of Earthquake Engineering, University of Roorkee, Roorkee
[14] Ranjan, R., 2005. Seismic Response Analysis of Dehradun City, India. MSc. Thesis, International Institute
for Geo-Information Science and Earth Observation (ITC), Dehradun, 97 pp.
[15] Technical Manual. Federal Emergency Management Agency (FEMA), National Institute of Building
Sciences (NIBS), Washington, DC, pp. 325.
684
Paper No.: 128
ANALYSIS OF PLATE RESTING ON PASTERNAK AND WINKLER

FOUNDATION DUE TO MOVING LOAD
2
P. R. Maiti 1, Bhawesh Madhukar2 and Satyam Mandloi
1
Assistant Professor, Dept. of Civil Engineering, Indian Institute of Technology (Banaras Hindu University)
Varanasi, India
2
Student, Dept. of Civil Engineering, Indian Institute of Technology (Banaras Hindu University) Varanasi,
India
Email: prmaiti.civ@iitbhu.ac.in
ABSTRACT
Dynamic response of pavements on elastic foundation occupies an important place in many fields of structural
and foundation engineering. In this study the behavior of a finite plate resting on a Pasternak foundation and
Winkler foundation subjected to a constant velocity moving load is investigated. Since the Pasternak foundation
provides a rational and computationally efficient model of the sub-soil, it is used in this work as the sub-grade
soil model and comparative study has been carried out with Winkler foundation. The Winkler model neglects
the effect of interconnection among the soil layers and as a result it imposes some limitations in the physical
modeling of the sub-soil system. In this study, stress analysis of plate is carried out on elastic foundation for
static and moving load. The effect of plate aspect ratio, magnitude of load and moving load velocity are
investigated numerically and analytically. The parametric studies include the difference between static and
dynamic response of vehicle speed and soil stiffness on the response of concrete pavements rested on soil due to
single wheel load.
KEYWORDS: Moving Load, Sub-soil system, Pasternak Foundation, Winkler Foundation
1. INTRODUCTION
Pavements are an essential feature of the urban communication system and provide an efficient means of
transportation of goods and services. Depending on its rigidity compared to the subsoil, pavements are classified
as flexible, rigid and semi-flexible. The conventional methods of rigid pavement design are based on the
closed-form solutions obtained from the static analysis of infinitely long plates resting on an elastic foundation.
The actual discontinuous nature of pavement systems is disregarded in this approach. Further, the dynamic
effects of moving vehicles are accounted for indirectly by applying an impact factor. Existing rigid (cement
concrete) pavement design is based on a static theory of thin slab on a Winkler elastic foundation, which
assumes the foundation to behave like independent discrete springs whose stiffness is known as the ‘Modulus of
sub-grade reaction’, usually denoted by ‘k’. However, the Winkler model neglects the interconnection among
the soil layers and as a result it may impose some serious limitations in the physical modeling of the sub-soil
system. Although this assumption may be well justified in the early years when automobiles travel slowly,
engineers are aware of its shortcoming. When the velocity of vehicle loads is high due to the development and
promotion of high-speed surface transportation, dynamic pavement response to a moving load needs to be taken
into account. Dynamic response of plates on elastic foundation occupies an important place in many fields of
structural and foundation engineering. A concrete rigid pavement, an airport runway and a mat foundation
constitute few examples of the plate structures resting on elastic foundation. The dynamic behavior of these
structures under moving masses i.e. the wheel loads from moving vehicles and airplane is completely different
from the static stationary mass acting on these structures. The dynamic response of runways subjected to
moving loads such as passing vehicular traffic, and landing, take-off, or taxiing of aircraft, has received
significant attention in recent years due to the increasing importance of the reliability of the design of runways.
There is also the need for rehabilitating this country’s airport runways for use by planned large transport aircraft.
The tyre pressure generated from the moving vehicle may result in sufficient loads on top of the mine to excite
an explosion. It is of paramount importance to investigate the dynamic displacement and stress field caused by a
moving load. A significant difference between static theory and its dynamic counterpart is that inertial effect,
ignored in the former, is taken into account in the latter. To better understand dynamic response of a thin slab to
a moving load, it is indispensable to analyze dynamic effect of a dynamic load on a slab.
Li et al (2009) presented an improved model for bending of thin viscoe-lastic plate resting on Winkler
foundation. The basic equations for internal forces and stress distribution were derived based on the general
viscoelastic theory under small deformation condition. It was shown that the proposed model reveals larger
flexural rigidity compared to that in classic models, in which the normal stress along the plate thickness was
neglected. Oni and Awodola (2010) investigate the dynamic response under a moving load of an elastically
supported non-prismatic Bernoulli-Euler beam on variable elastic foundation. The elegant Galerkin’s method is
used to reduce the governing fourth order partial differential equations and a modification of the Struble’s
technique is employed for the solutions of the resulting Galerkin’s equations. Wang et al (2010) investigate the
resonance characteristics of a two-span continuous beam traversed by moving high speed trains at a constant
velocity in which the continuous beam has uniform span length. Each span of the continuous beam was modeled
as a Bernoulli-Euler beam and the moving trains were represented as a series of two degrees-of-freedom
mass-spring damper systems at the axle locations. The effects of different influencing parameters, such as the
velocities of moving trains, the damping ratios and the span lengths of the beam, on the dynamic response of the
continuous beam were examined. The two-span continuous beam has two critical velocities causing two
resonance responses, which was different from simple supported beam. The resonance condition of the two-span
continuous beam was put forward which depends on the first and second natural frequency of the beam and the
moving velocity. Sapountzakis and Kampitsis (2011) developed a boundary element method for the
geometrically nonlinear response of shear deformable beams of simply or multiply connected constant cross-
section, traversed by moving loads, resting on tensionless nonlinear three-parameter viscoelastic foundation,
undergoing moderate large deflections under general boundary conditions. The beam was subjected to the
combined action of arbitrarily distributed or concentrated transverse moving loading as well as to axial loading.
Oni and Awodola (2011) further investigate the dynamic behavior under moving concentrated masses of simply
supported rectangular plates resting on variable Winkler elastic foundation. A technique based on separation of
variables was used to reduce the governing fourth order partial differential equations with variable and singular
coefficients to a sequence of second order ordinary differential equations. For the solutions of these equations, a
modification of the Struble’s technique and method of integral transformations were employed. Yang et al (2012)
investigate the dynamic response of an infinite plate on an elastic foundation subjected to a moving traffic load
with variable velocity (accelerating or decelerating) via a triple Fourier transform. Dimitrovová and Rodrigues
(2012) presented An analysis of the critical velocity of a load moving uniformly along a beam on a visco-elastic
foundation composed of one or two sub-domains The case study addressed is related to high-speed railway lines.
A new formulation of the governing equations in the first order state-space form was proposed for the
Timoshenko–Rayleigh beam. Critical velocities were obtained for load travelling on finite and infinite beams,
with and without damping. A new relationship between the viscous damping coefficient and the modal damping
ratio was derived and justified.
In the present study, the stress analysis of a plate rested on elastic foundation is carried out. The subsoil is
modeled in two well known elastic foundation analysis namely Winkler foundation and Pasternak foundation.
An effort is made to found parametric relationship with constant load velocity, soil stiffness and aspect ratio of
the plate. Comparative study is made for different foundation and for assuming different support conditions of
the plate.
2. MATHEMATICAL FORMULATIONS
An elastic foundation is a foundation that follows Hook's law when the structure is constructed on it and
subjected to loading. Commonly, following three types of elastic foundation are considered to model the plate
686
rested on elastic foundation Winkler Foundation, Pasternak Foundation and Kerr Foundation.
2.1 Winkler Foundation
In this type of foundation, the sub grade is assumed to be composed of closely spaced linear springs and sub
grade shear strength is neglected. The sub grade reaction is proportional to the deflection (equivalent to dense
liquid foundation) and the proportionality (spring) constant is called as modulus of sub grade reaction (k). So,
the upward sub grade pressure is P = kw. The equilibrium equation of plate resting on Winkler foundation
becomes
D∇ w + kw = q (1)
According to the classical theory of thin isotropic elastic plates, the transverse deflection w(x,y,t) of a
rectangular plate resting on Winkler foundation satisfies the partial differential equation
( , ,) ( , ,) ( , ,) ( , ,)
D + 2 + +m + kw(x, y, t) = F(x, y, t) (2)

Where, w (x, y, t) = Vertical deflection of the plate at point with coordinates x, y and time t and = ,D is the
!( "# )
stiffness, E is young modulus of the plate, h is the thickness of the plate, ν is Poisson’s ratio, mp is the mass of plate per
unit area, k is sub grade modulus, F(x, y, t) is the moving load.
2.2 Pasternak Foundation
In this type of foundation, the shear interaction between the Winkler spring elements is incorporated and sub
grade shear force is assumed to be proportional to the variation in deflection in the soil layer. The shear effect
Q ,% = G% and Q ,% = G% is incorporated in the formulation. Where, Q ,% and Q ,% are the
subgrade shear forces per unit length acting on the plane orthogonal to X and Y direction respectively and G% is
called as sub grade shear modulus. So, the net upward vertical pressure (P) is
+,,- +.,-
P = kω − * + / = kω − G% ∇! w (3)
The equilibrium equation for plate resting on Pasternak foundation becomes
D∇ ω = q − p = q − (kω − G1 ∇! ω)
D∇ ω − G1 ∇! ω + kω = q (4)
The Winkler foundation is a special case of Pasternak foundation (i.e. when, sub grade shear modulus Gs=0)
In this research work, an orthotropic homogeneous elastic rectangular plate resting on an elastic Pasternak
foundation is considered. According to the classic theory of thin plates, the transverse deflection w(x, y, t) of the
plate satisfies the partial differential equation
( , ,) ( , ,) ( , ,) ( , ,)
D + 2B +D + ρh + γh + k 6 w(x, y, t) − G% ∇! w(x, y, t) = P(x, y, t (5)
Where; w(x, y, t) = transverse deflection, ρ=mass density per unit volume, h=thickness of the plate;
687
t= time, γ = damping ratio, kf =spring stiffness, Gs= shear modulus of the Pasternak foundation,
P(x,y,t)=dynamic load on the plate; Dx, Dy= plate flexural rigidity in x and y direction respectively;
B=effective torsional rigidity.
The dynamic moving traffic load p(x, y, t) modeled as an equivalent concentrated load of harmonically varying
magnitude moving in the x directional axes of the plate can be expressed as follows
P(x, y, t) = P7x(t), y(t), t8 = P(t)δ7x − x(t)8δ7y − y(t)8 = P9 (1 + cos ωt)δ7x − x(t)8δ7y − y(t)8 (6)
!
x(t) = v9 t + ! acc(t ! ) and y(t) = ! b (7)
The deformation of the plate due to moving load in Pasternak foundation may be written as
JMN KMN
w(x, y, t) = C C e"£FGH ( " I ) 7ω9JK cos L(1 − £! )ωJK (t − t 9 )

JM KM
OIGH PFGH IGH

+ sin 7L(1 − £! )ωJK (t − t 9 ) 88 (8)
FGH L( "£ )
Where;
R9ST = initial deflection at t = to; U9ST = initial velocity at t = to; £ = damping ratio
Bending moments and vertical shear forces in the plate can be computed in terms of the deflection and its
derivatives
( , ,) ( , ,) ( , ,) ( , ,)
M = −D * +v / and M = −D * +v /
( , ,) ( , ,) ( , ,) ( , ,)
Q = − *D + B / and Q = − *D + B /
2.3 Kerr Foundation
In Kerr foundation, the sub grade is characterized with two spring constants (k, k0) along with presence of sub
grade shear interaction. The relationship between the upward sub grade reaction (P) and deflection (w) of the
slab as follows
W X
*1 + W / P − Y- ∇! P = kω − G% ∇! ω (9)
I I
The Pasternak foundation is a special case of Kerr foundation (i.e. when k0 = ∞).
Determination of dynamic response of finite plate rested on different type of elastic foundation for different
loading position. A square plate of size 5m and thickness 225mm is placed on different types of foundation.
Analysis is performed for a wheel load of magnitude 20kN that is passed through the centre line of the
pavement and edge of the pavement. The maximum deflection at different critical point is found and presented
in Table 1 and Table 2. The result reveals the significance of dynamic analysis and damping effect of subsoil.
688
Table 1: Maximum Deflection and Stress for a Single Wheel Passed Along the Edge
Model Analysis Deflection at Deflection at Stress at corner Stress at edge

corner (mm) edge (mm) (MPa) (MPa)
Winkler foundation
Static analysis 1.424 0.535 0.919 0.385
Winkler foundation
Dynamic Analysis 1.592 0.606 0.989 0.614
Pasternak foundation
Table 2: Maximum Deflection and Stresses for a Single Wheel Passed Along the Centerline
Model Analysis Deflection at Deflection at mid Stress at edge Stress at mid

edge (mm) point (mm) (MPa) point (MPa)
Winkler foundation
Static analysis 0.575 0.154 0.122 0.145
Winkler foundation
Pasternak foundation
When the wheel load moves with velocity 50 km/h and having 5% damping in the Pasternak foundation the
maximum deflection and stress at different position of the plate is determined. The numerical values are
presented in the Table 3. Analysis is performed taking the moving load value 20kN, Young modulus 50 MPa.
Table 3: Maximum Deflections and Stresses for Moving Load of magnitude 20kN
Wheel along edge Wheel along centerline
Numerical model
[c (mm) \C (MPa) [M (mm) \M (MPa)
1.579 2.117 0.558 1.256
Un-damped condition
1.288 0.064 0.436 0.281
Pasternak Model with
damping 5%
1.592 2.396 0.829 1.362
Winkler model
1.411 0.483 0.477 0.705
Moving Load
Velocity 50km/h
The analysis of the plate is also done for IRC loading conditions. The IRC loading and its position have been
generated by STAAD Pro by itself after selecting the IRC code and vehicle type in Bridge Deck Processor. A plate
of size 8m X 8m has been chosen for the stress analysis without and with elastic foundation and for which the
detail of its IRC loading is shown in Fig 1 to Fig. 4. The plate is analyzed for the values E=2 x 107 kN /m2 ν
= 0.17, ρ = 25 kN/m3, thickness of plate h=200mm and soil sub-grade reaction k = 27.2 MN/m3.
689
Figure 1: Isometric view of Position of IRC Class Figure 2: Isometric View of the Loading
AA Loading on the plate Position for IRC 70 R Loading
Figure 3: Isometric View of the Loading Position Figure 4: Isometric View of the Loading
for Class IRC A Loading Position for IRC Class B Loading
The plate rested on elastic foundation is analyzed for the different static IRC loading and result of the analysis is
stipulated in Table 4. Here the –ve sign is for sagging moment where as +ve sign for hogging moment.
Table 4: Maximum deflection and moment of the plate for IRC loading
IRC Maximum Mx (kNm/m) My(kNm/m) Mxy(kNm/m)
loading Deflection(mm) +ve -ve +ve -ve +ve -ve
Class AA 1.609 3.591 9.375 3.479 8.655 3.470 3.470
Class 70R 1.394 3.015 7.587 2.510 5.249 2.808 2.808
Class A 0.796 0.783 3.841 3.14 7.503 1.266 1.610
Class B 0.403 0.389 1.836 1.389 3.874 0.607 0.575
690
The variations of moment in the plate for different IRC loading cases are presented in Figure 5. Maximum
sagging moment takes place at the centre and it is represented by the color coding given just after the contour
plot.
Figure 5a: Contur of Mx for plate Figure 5b: Contur of Mx for plate
subjected IRC class AA loading subjected to IRC class A loading
Figure 5c: Contur of Mx for plate Figure 5d: Contur of Mx for plate
subjected IRC 70R loading subjected IRC Class B loading
691
Figure 5e: Contur of My for plate Figure 5f: Contur of My for plate
subjected IRC class AA loading subjected IRC Class A loading
Figure 5g: Contur of My for plate Figure 5h: Contur of My for plate
subjected IRC 70R loading subjected IRC class B loading
The moment variation of the plate of size 5 m x 5m due to different IRC loading is shown in the Figure 5 in the
Pasternak foundation. In the pavement design, the design load is recommended by IRC in the bridge design and
rigid pavement design. Here the analysis of Pasternak foundation is performed with moving load with
magnitiude of IRC loading. As expected maximum sagging moment occur in case of Class AA vehicle as shown
in Figure 5. Here +ve moment generates due to the soil sub grade reaction as the loading is symmetric about the
centre the moment distribution is also symmetric about the centre of the plate about x axis and +ve moment
occur near the edge of the plate due to upward reaction from soil sub grade.
4. CONCLUSIONS
A Pasternak foundation model incorporates shear interaction between spring elements, mobilized through a plate
placed on top of the springs, which deforms only by transverse shear. Thus, in this model compressive and shear
deformation of the soil are duly simulated. In the application of the theory of dynamic response of the
orthotropic plate, the continuous elastic foundation modeled as a Pasternak foundation is representing closely
the actual subsoil condition, but requiring advanced analytical treatment in solving the dynamic response
problem. In the present study, the response of a finite plate rested on soil mass that may be treated as elastic
foundation is analyzed for different loading conditions. Initially, the plate is analyzed without any soil mass at
base. The main objective of this study is to introduce the importance of Pasternak foundation. As till now so
many researchers worked on plate on elastic foundation but most of them have chosen Winkler foundation for
their study and ignore Pasternak foundation (i.e. the shear strength of the soil has been neglected). The Winkler
foundation is applicable only for the cases where the foundation is a thin and light layer resting on a rigid
692
support. In such a case, the inertia, the continuity and interaction effects in the foundation as a continuum may
be neglected. In engineering practice, however, this is not always the case and these factors may dominate the
response of the pavement.
REFERENCES
[1] H. Saito and S. Chonan (1980) Response of an Elastically Supported Plate Strip to a Moving Load, Journal
of Sound and Vibration , 71(2), 191-199
[2] G. Pan and S. N. Atluri (1995) Dynamic Response Of Finite Sized Elastic Runways Subjected To Moving
Loads: A Coupled Bem/Fem Approach, International Journal For Numerical Methods In Engineering, Vol.
38, 3143-3166.
[3] B.P. Patel, M. Ganapathi, K.R. Prasad, V. Balamurugan (1999) Dynamic Instability Of Layered Anisotropic
Composite Plates On Elastic Foundations, Engineering Structures, Vol. 21 , 988–995.
[4] M.-H. Huang, D.P. Thambiratnam (2001) Deflection response of plate on Winkler foundation to moving
accelerated loads”, Australia Engineering Structures, Vol. 23, 1134–1141.
[5] Seong-Min Kim, B. Frank McCullough (2003) Dynamic response of plate on viscous Winkler foundation
to moving loads of varying amplitude, Engineering Structures, Vol. 25, 1179–1188.
[6] Seong-Min Kim (2004) Buckling and vibration of a plate on elastic foundation subjected to in-plane
compression and moving loads, International Journal of Solids and Structures, Vol. 41, 5647–5661
[7] Seong-Min Kim (2004) Influence of Horizontal Resistance at Plate Bottom on Vibration of Plates on
Elastic Foundation under Moving Loads, Engineering Structures, Vol. 26, 519–529
[8] Seong-Min Kim (2005) Stability and dynamic response of Rayleigh beam–columns on an elastic
foundation under moving loads of constant amplitude and harmonic Variation, Engineering Structures, Vol.
27, 869–880
[9] J.A. Gbadeyan and M.S. Dada (2006) Dynamic response of a Mindlin elastic rectangular plate under a
distributed moving mass, International Journal of Mechanical Sciences, Vol. 48, 323–340
[10] Lu Sun (2006) Analytical dynamic displacement response of rigid pavements to moving concentrated and
line loads, International Journal of Solids and Structures, Vol. 43, 4370–4383.
[11] P. R. Maiti, R Saha & S W Alisjahbana (2008) Response of plate on elastic foundation under harmonic
moving load, The Icfai University Journal of Structural Engineering, Vol-1, No-1, pp-60.
[12] Zhi-Da Li, Ting-Qing Yang, Wen-Bo Luo (2009) An Improved Model for Bending of Thin Viscoelastic
Plate on Elastic Foundation, Natural science , Vol.1, No.2, 120-123.
[13] S. T. Oni and T. O. Awodola (2010) Dynamic response under a moving load of an elastically supported
Non-prismatic Bernoulli-Euler beam on variable elastic foundation, Latin American journal of solid and
structure, Vol. 7, 3 – 20.
[14] Yingjie Wang, QingChaoWei, Jin Shi and XuyouLong “Resonance characteristics of two-span continuous
beam under moving high speed trains” Latin American journal of solid and structure 7(2010) 185 – 199
693
[15] E.J. Sapountzakis, A.E.Kampitsis (2011) Non-linear response of shear deformable beams on tensionless
nonlinear viscoelastic foundation under moving loads, Journal of Sound and Vibration, Vol. 330,
5410-5426.
[16] Ting Yan, Sritawat Kitipornchai, Jie Yang, Xiao Qiao He (2011) Dynamic behavior of edge-cracked shear
deformable functionally graded beams on an elastic foundation under a moving load, Composite
Structures, Vol. 93 , 2992-3001.
[17] S. T. Oni and T. O. Awodola (2011) Dynamic Behaviour under Moving Concentrated Masses of Simply
Supported Rectangular Plates Resting on Variable Winkler Elastic Foundation, Latin American journal of
solid and structure, Vol. 8, 373 – 392
[18] Z. Dimitrovová, A.F.S. Rodrigues (2012) Critical velocity of a uniformly moving load, Advances in
Engineering Software, Vol. 50, 44-56.
694
Paper No.: 129
DETERMINATION OF DYNAMIC MODULUS OF ELASTICITY OF

CONCRETE
Pardeep Kumar1 and R. K. Sharma2

1
Associate Professor, Dept. of Civil Engineering, National Institute of Technology, Hamirpur, HP, India.
pardeepkumar.nit@gmail.com
2
Professor, Dept. of Civil Engineering, National Institute of Technology, Hamirpur, India
rksnithp61@gmail.com
ABSTRACT:
The dynamic modulus of elasticity of materials is more important while investigating dynamic behavior of the
structures. The limited experimental program was conducted to determine dynamic modulus of elasticity of
concrete and to investigate the effect of strength on the modulus of elasticity. The modulus of elasticity in the
present investigation is determined on the basis of fundamental longitudinal resonant frequency. The tests were
conducted on concrete specimens having five different strengths designated as M15, M20, M25, M30 and M35
grades. All these mixes were designed using indigenously available materials on the basis of IS code method
contained in IS: 10262-2009. Ten cylindrical samples of each grade were cast in the laboratory. These cylinders
were then tested on resonant frequency system and dynamic modulus of elasticity was determined. These values
were computed with available relationship between dynamic modulus and fundamental longitudinal resonant
frequency. The various cylindrical specimens were then tested under compression testing machine to obtain the
stress v/s strain variation and compressible strength under monotonically increasing uniaxial compression. The
static modulus of elasticity was evaluated from the slope of stress-strain variation. Statistical analysis was
performed to obtain co-relation between dynamic and static modulus of elasticity. Correlation between dynamic
modulus of elasticity and compression strength was also obtained. A comparison between dynamic modulus of
elasticity as obtained from the expression available in literature has also been made.
KEYWORDS: Dynamic Modulus of Elasticity, Static Modulus, Concrete, NDT, Resonance.
1. INTRODUCTION
Dynamic testing techniques such as an impact resonance test have an advantage of performing the measurement
non-destructively without breaking the specimens. Non-destructive dynamic methods can be used to estimate
in-place modulus of elasticity, but the meaning of the obtained dynamic modulus is uncertain because Ed is
known to be different (higher) from that obtained by direct static testing of a cylinder drawn from the structure.
The static Young’s modulus (E) is defined as the ratio of the axial stress to axial strain for a material subjected
to uni-axial load [Neville, 1997]. It is important that E of concrete be known because engineers increasingly use
this value in the structural analysis and design process. In order to get the displacement response, it is very
important to have knowledge of modulus of elasticity at different stress levels. For example, elastic modulus is
also needed to analyse the cross-sectional response of a reinforced concrete beam [Leet and Bernal, 1997]. In
concrete as well as steel structure specifications even require a specific elastic modulus of concrete and steel to
be met, mostly to limit excessive deformation and sway in tall buildings. However, once a structure is erected
the in situ elastic properties cannot be measured directly without damaging the structure itself. Most often E is
concrete inferred from the compressive strength (fc) of cylinders cast at site from concrete used in construction,
rather than being measured directly, through the application of established empirical relations. This approach
often leads to overly conservative results because, in order to meet the minimum E requirement, concrete with
much higher compressive strength (fc) is used than the specification requires, which leads to unnecessarily high
material costs. Better understanding of the relation between modulus of elasticity and compressive strength,
with respect to different types of concretes, would improve the efficacy of the estimation of modulus of
elasticity from strength.
Concrete shows a non-linear dependence between stress and strain right from very small values of stresses. The
nonlinear stress-strain relationship for concrete is well described by a quadratic parabola [Shkolnik, 1996]. The
non-linear behaviour provides the base for the conventionally accepted view why Dynamic Modulus is higher
than Static Modulus since concrete is subjected to very small strains in dynamic testing [Neville, 1997].
Although this argument is conceptually satisfying, experimental test results show that this may not in fact be the
case. Furthermore, it is known that Ed values for a given concrete obtained by different dynamic tests do not
agree with each other [Philleo 1955]. Ed computed from vibration resonance of prismatic beams is known to be,
on average, significantly higher than that computed from cylinders for the same concrete mixture.
It is reported in literature that Young's modulus of concrete increases with hardening of the concrete at early
ages. Young's modulus of elasticity measured statically has been widely used to assess the variation of modulus
with curing ages. However, this method is costly because numerous test specimens are required. On the other
hand, dynamic testing techniques have advantage of non-destructive testing which makes method cheaper in
comparison to static method. Furthermore, the perceived difference between static and dynamic moduli is
becoming smaller due to the improvement of accuracy in the static measurement, and static modulus can be
obtained from the dynamic modulus using available correlations.
The objectives of this investigation is to provide the fundamental relationship between modulus and strength
of concrete at early age and to assess the feasibility of using the impact resonance (IR) test for the
non-destructive evaluation of strength of the early age concrete specimens. The differences between dynamic
and static modulus can be explained by the differences in strain levels used in testing, as well as a loading rate
effect. The modulus measured dynamically was found to be about 20 percent greater than the value measured in
the conventional static test which is based on stress and strain at 60 percent of the failure stress.
The advantage of non-destructive testing is to determine the various properties of concrete such as strength,
modulus of elasticity, homogeneity, integrity as well as conditions of strain and stress without damaging the
structure. Selection of the most applicable method or methods of testing requires selection
based on information needed, size and nature of the project and the seriousness of observed
conditions. The dynamic modulus of elasticity is used primarily to evaluate soundness of concrete in
durability tests; more appropriately when concrete used in structures subjected to dynamic loading (i.e. impact
or earthquake).
Mehta [1986] reported that, the dynamic modulus of elasticity is generally 20, 30 and 40 percent higher than the
static modulus of elasticity for high, medium, and low strength concretes respectively. Neville [2000] showed
that, the ratio of the static to dynamic modulus is higher with the higher strength of concrete and increases with
age. In majority of cases, the modulus calculated from resonant frequency tests exceed that calculated from
static loading tests. The British Code [1972] of concrete structures CP: 110-1972 recommends the following
relations between static and dynamic modulus of elasticity.
The British Code of Practice for the structural use of concrete CP: 110-1972 also relates the dynamic modulus
of elasticity of concrete Ed to its strength fc.
Lydon and Balendran [1986] gave an empirical relation between the static and dynamic modulus of elasticity of
concrete. The relation is range bound and it does not apply to concretes containing more than 500 kg/m3 of
cement or lightweight aggregate concrete.
Khalil and Gilles [2002] reported that difference between static and dynamic modulus of elasticity of concrete
exits because of heterogeneity of concrete affects the two moduli in different ways. Dias et al. [1998] showed
that, the dynamic modulus of elasticity determined by a pulse-velocity technique using a pundit instrument
appears to be more sensitive to moisture loss, which indicates that it is more sensitive to temperature changes
than static modulus of elasticity. Aitcin [1998], Mohammed [2006] concluded that there is a simple linear
696
relationship between the static and dynamic elastic moduli. Hansen [2002] showed that, the relationship
between dynamic modulus of elasticity and cube compressive increases with increase in concrete strength,
expressed in form of best fit equation to the observed result as
Where, is expressed in GPa and is expressed in MPa.

Lianxiang Du [2007] established correlations between the dynamic modulus and the compressive strength of
concrete mixes. For a single mix, it is possible to predict the compressive strength development from dynamic
modulus measurement in the course of time. However, if only the dynamic modulus and the coarse aggregate
type is known, it is not possible to estimate the concrete strength development. John S. Popovics [2008]
conducted a study to better understand the relationship between static and dynamic modulus of elasticity and
concluded that Ed values vary, depending on method of measurement. Ultrasonic pulse velocity gives highest
prediction and longitudinal vibration gives lowest. He further concluded that Ed is always greater than Es for
concrete, and the relation between and appears to be linear, or nearly linear. Kwang-Lee et al. [1997]
observed that impact resonant method uses P-wave velocity of concrete specimens; stiffness of concrete
develops rapidly even at early ages and modulus in wet condition is higher that dry condition. Choudhari N. K.
et al. [2002] presented a method to determine dynamic and static modulus of elasticity. The relation to evaluate
Poisson Ratio of a structural member is also derived using thickness of member. Villaini et al. [2009] aimed to
determine the static modulus and dynamic modulus of six different concrete mixes and experimental results
showed that is linear function of porosity. Mohammed et al. [2006] concluded that the compressive strength
and dynamic modulus are significantly influenced by the mineralogy of coarse aggregate.
Thus literature indicates the importance of the study of dynamic modulus. Numerous relationships between
dynamic and static modulus as well as dynamic modulus and strength are available but the accuracy of
prediction is not very high. However, it is observed by many researchers that the dynamic modulus is higher
than the static modulus. This study also focuses to derive an accurate relationship and compare it with the
existing ones.
3. EXPERIMENTAL INVESTIGATION
Concrete mix in present investigation has been done as per IS: 10262-2009 in which the basic assumption made
is that the compressive strength of workable concrete is, by and large, governed by the water- cement ratio.
Another most convenient relationship applicable to normal concretes is that for a given type, shape, size and
grading of aggregates, the amount of water determines its workability. However, there are various other factors,
which affect the properties of concrete, for example, the quality and quantity of cement, water and aggregates;
batching; transportation; placing; compaction; curing; etc. Therefore, the specific relationships that are used in
proportioning concrete mixes should be considered only as a basis for trial, subject to modifications in the light
of experience as well as for the particular materials used at the site in each case.
3.1 Data Required
The following basic data is required to be specified for design of a concrete mix:
a) Characteristic compressive strength of concrete at 28 days.
b) Degree of workability desired in terms of slump.
c) Limitations on the water-cement ratio and the minimum cement content to ensure adequate durability as
per IS: 10262-2009.
d) Standard deviation of compressive strength of concrete depends upon grade of concrete.
e) Properties of the constitute materials of concrete such as specific gravity, moisture content, water
absorption, sieve analysis etc.
The target mean compressive strength as per IS: 10262-2009 is given by:
Where, ft - Target mean strength

fck - Characteristic compressive strength of concrete at 28 days
697
K - Tolerance factor equal to 1.65 (so that not more than 5% result fall below the characteristic strength)
S - Standard deviation for quality control and grade of mix
3.2 Mix Proportion (as per IS: 10262-2009)
In present investigation five mixes of strength 15, 20, 25, 30 and 35 N/mm2 were used. All the mixes were
designed as per the recommendations of IS: 10262-2009. Thus trial mixes were prepared for all the five mixes
based on the calculations. Theses trial mixes were used to cast the cylinders for determination of dynamic
modulus of elasticity for normal concrete. Table 1 contains the final proportions after trial mixes, which were
used for casting the cylinders and cubes.
Table 1 Final mix proportion for five mixes
Mix Cement Water Sand Aggregate
M15 1 0.59 1.96 3.48
M20 1 0.56 1.80 3.29
M25 1 0.50 1.56 2.97
M30 1 0.43 1.23 2.50
M35 1 0.38 1.05 2.22
3.3 Cube Testing
Three cubes each of size 150mm x 150mm x 150mm were then caste for each of trials. These cubes were then
tested on 3000 KN compression testing machine under monotonically increasing uniaxial load to determine their
compressive strength. The testing was done at a slow rate of stress ensuring that the load is acting concentrically
and the stress distribution is uniform throughout the cross section of the cube. The proportions of the trial giving
highest strength were selected for that particular grade. At the time of testing the specimen is placed in a
Compression Testing Machine with the position of cubes at right angles to the position of casting. The load is
applied at a constant rate of stress within the range of 0.2 to 0.4 N/mm2/sec. Under pure uniaxial compression
loading, the failure cracks generated are approximately parallel to the direction of applied load as seen in Figure
1 with some cracks formed at an angle to the applied load. Practically, the compression testing system rather
develops a complex system of stresses due to end restraints by steel plates. A typical stress strain curve obtained
during static testing under monotonic increasing compressive load is shown in Figure 2.
Figure 1Typical cube failure Figure 2 Typical stress strain curve

3.4 Dynamic Modulus of Elasticity
The method used for determining the dynamic modulus of elasticity is the resonant frequency method. The test
is carried out on cylinders. The test is carried out according to ASTM C215-08, using the erudite resonant
698
frequency tester by forced resonance method. The cylindrical specimen cast from the design concrete mixes in
laboratory having 300 mm height and 150 mm diameter were used for dynamic modulus evaluation. In the
forced resonance method, a supported specimen is forced to vibrate by an electro-mechanical driving unit from
one face of cylinder. The specimen response is monitored by a lightweight pickup unit mounted on other face of
the specimen. The driving frequency is varied until the measured specimen response reaches maximum
amplitude. The value of the frequency causing maximum response is the resonant frequency of the specimen.
The fundamental frequencies for the three different modes of vibration are obtained by proper location of the
driver and the pickup unit. The arrangement used for holding the specimen and mounting of electro mechanical
driving units and pickup are shown in Figure 3. Figure 4 shows the display unit of resonant tester.
The dynamic modulus corresponds to a very small instantaneous strain due to suddenly applied stress. Thus, it is
approximately equal to initial tangent modulus but is considerably larger than the static secant modulus. The
longitudinal resonant frequency is measured for each specimen immediately after removal from water curing
tanks. The test is conducted with output voltage of about 1 volt and average value of repetitions is taken for each
sample. Three cylinders are tested at specified age for each batch.
Dynamic modulus of elasticity, Ed, in Pascal from the fundamental longitudinal frequency, mass, and
dimensions of the test specimen can be calculated using the following expression
Ed = DM (n’)2
Where, n' = fundamental longitudinal frequency, Hz, and D = 5.093 (L/d2), per meter for a cylinder.
Figure 3 Cylindrical specimen placed on Figure 4 Typical Input v/s output frequencies
Resonant Frequency Tester in oscilloscope
3.5 Static Modulus of Elasticity
The static-elastic modulus is calculated from the stress-strain diagram. The chord-modulus method can be used.
The chord modulus is the slope of the line drawn between two fixed points on the stress-strain diagram. The
lower point, established to eliminate the effect of crack from the initial portion of the stress-strain curve, and it
is the point where strain is 5x10-5 mm, while the upper point where stress equals 40 percent of the ultimate
stress. IS: 456-2000, gives an approximate value of static modulus as given below which has been used.
Where is cube compression strength in N/mm2.
The dynamic modulus of the specimens was measured by impact resonance method at various
curing ages. The cylinders cast from concretes of grades M15, M20, M25, M30, M35 were tested for
fundamental longitudinal resonant frequency to obtain dynamic modulus of elasticity. As curing time increases,
the frequency of the dominant peak clearly increases, indicating the increasing stiffness of the specimen. Testing
was done on 7th, 14th and 28th day for M15, M20 & M25 and on 14th, 21st & 28th day for M30 & M35 grade
cylinders. For all the test specimens mass, resonant frequency (fr), peak load in compression testing,
compressive strength (fc), static modulus (Es) and dynamic modulus (Ed) were measured. The experimental
results obtained are compared with the analytical expressions available in literature. The dynamic modulus of
699
4.1 Comparison of Dynamic Modulus
A comparison between the dynamic modulus predicted by various relations (Lyndon et al., Hasen, IS code,
British code) is shown in figure 5. Figure 6 further shows the variation of dynamic modulus of elasticity with
respect to the compressive strength. A linear relation between dynamic modulus and Compressive strength is
also proposed on the basis of this limited experimental research programme. Figure 6 shows the correlation and
the R2 value obtained.
Figure 5 Comparison between dynamic Figure 6 Variation of dynamic modulus with

modulus of elasticity (as per different respect to compressive strength.
expressions)
A similar correlation between static and dynamic modulus of elasticity has been developed and is shown in
Figures 7. A third degree polynomial when fitted to the data gives a better confidence and R2 value as shown in
Figure 8. The variation of dynamic modulus with time is also shown in Figure 9.
36000 Es = 0.949Ed - 6245.2 Es = -2E-08Ed3 + 0.0021Ed2 -

34000 R² = 0.7382 73.468Ed + 852376
32000 36000
34000 R² = 0.871
Es (MPa)
30000
32000
Es (MPa)
28000 30000
26000 28000
24000 26000
24000
22000 22000
20000 20000
20000 25000 30000 35000 40000 45000 25000 30000 35000 40000 45000 50000
Ed (MPa)
Ed (MPa)
Figure 7 Linear Correlation between static Figure 8 Third degree Correlation between
and dynamic modulus of elasticity static and dynamic modulus of elasticity
5. CONCLUSIONS
Based upon the results obtained and the analysis carried out, following conclusions were drawn
• Dynamic modulus of elasticity of concrete is higher than Static modulus which is in agreement with
past research done in this field.
• Dynamic modulus is about 13 – 61 % higher than static modulus.
700
• Dynamic modulus is generally higher for high strength mixes as expected.
• 70-80% of the 28th day Dynamic modulus is gained within 7 days from casting.
• Dynamic modulus obtained experimentally is lower than Dynamic modulus (obtained from Static
modulus, as per BS CP 110:1972) by about 17%; on average.
• Dynamic modulus obtained experimentally is lower than Dynamic modulus (obtained from Static
modulus as per formula by Lydon et. al.) by about 35%; on average.
• Dynamic modulus obtained experimentally is higher than Dynamic modulus obtained from the
compressive strength (as per relationship given by Hansen) by 12 %, on average.
Figure 9 Variation of the Ed (dynamic modulus) with time
6. ACKNOWLEDGEMENT
The work presented in this paper was carried out at Department of Civil Engineering at NIT Hamirpur under the
supervision of author. The work carried out by Aman Gupta, Kamal Verma, Ajay Kumar, Pankaj Syal and
Manish Vashisht is duly acknowledged.
REFERENCES
[1] ACI , (1990). Manual of Concrete Test Practice, Part 1,pp. 116R-32.
[2] ACI Committee 318, (1995). Building Code Requirements for Reinforced Concrete and Commentary (ACI
318-M95/318R-95), American Concrete Institute, Detroit.
[3] Aitcin, P. C. (1998). High-Performance Concrete, First Published, London.
701
[4] AlaMalaikah, Khalid Al-Saif and Rajeh Al-Zaid. (2004). Prediction of the dynamic modulus of Elasticity
of concrete under different loading conditions, International Conference on Concrete Engineering and
Technology, University of Malaya, Malaysia.
[5] ASTM C215-08. (2008). Standard test method for longitudinal and torsional resonant frequencies of
concrete specimen”. ASTM International, United States.
[6] CP110:1972. The British Code of Concrete Structures.
[7] Dias, Khoury, and Sullivan. (1998). Mechanical Properties of Hardened Cement Paste Exposed to
Temperatures up to 700oC, ACI Material Journal, pp. 252-261.
[8] Géraldine Villain, Odile Abraham, Loïc Le Marrec, Lalaonirina Rakotomanana. (2009). Determination of
the bulk elastic moduli of various concrete by resonance frequency Analysis of slabs submitted to
degradations, France. IRMAR-UMR 6625.
[9] Hansen, T. C. (2006). Recycled Aggregate and Recycled Aggregate Concrete, Material and Structures
(RILEM), Journal of Engineering and Development, Vol. 10, No. 2.
[10] Ilkerbekir Topçu, Ali Uğurlu. (2007). Elasticity Theory of Concrete and Prediction of Static Elastic
Modulus for Dam Concrete Using Composite Models, Digest 2007, pp. 1115 -1127.
[11] John S. Popovics, (2008). A Study of Static and Dynamic modulus of elasticity of Concrete, University of
Illinois, Urbana, ACI-CRC Final Report.
[12] Khalil, H., and Gilles, P. (2002). Influence of Porosity on Fracture Characteristics in Mortar Structures,
15th ASCE Engineering Mechanics, Columbia University.
[13] Kumar, M. (1986). Concrete Structure Properties and Materials, Englewood Cliffs, New Jersey.
[14] Kwang-M-Lee, Dong-Soo Kim, Jee-Sang-Kim. (1997). Determination of Dynamic Young’s Modulus of
Concrete at Early Ages by Impact Resonance Test, KSCE Journal of civil Engineering, Volume 1, Issue 1,
pp 11-18.
[15] Leet, K.M. and Bernal, D. (1997). Reinforced Concrete Design. 3rd Ed. McGraw-Hill Companies, Inc.
New York.
[16] Lianxiang Du and Elizabeth Lukefahr, (2007). Estimation of compressive strength from dynamic Modulus
of concrete mixture” Concrete Research Engineer, Texas Department of Transportation Austin, Texas,
USA.
[17] Lydon, and Balendran. (1986). Some Observations on Elastic Properties of Plain Concrete, Cement
and Concrete Research, Vol. 16, No. 3.
[18] Mohammed M. Salman and Ali H. Al-Amawee (2006). The Ratio between Static and Dynamic Modulus of
Elasticity in Normal and High Strength Concrete, Journal of Engineering and Development, Vol. 10, No. 2,
pp. 163-174.
[19] N. K. Choudhari, Ashok Kumar, Yudhisther Kumar and Reeta Gupta (2002). Evaluation of Elastic Moduli
of Concrete by Ultrasonic Velocity, National Seminar of ISNT, Chennai, www.nde2002.org.
[20] Neville, A. M. (2000). Properties of Concrete, Wiley, New York, and Longman, London.
[21] P. K. Mehta. (1986). Concrete: Structures, Properties and Materials. Prentice Hall, Inc., Englewood Cliffs,
702
N. J., 450 pp.
[22] Philleo, R.E. (1955). Comparison of Results of Three Methods for Determining Young's Modulus of
Elasticity of Concrete, Journal of the American Concrete Institute, Volume 26(5), pp. 461-469.
[23] Shkolnik, I E. (1996). Evaluation of dynamic strength of concrete from results of static tests,” Journal of
Engineering Mechanics, Volume 122(12), pp. 135-138.
703
Paper No.: 130
CONSTRUCTION OF HIGH LEVEL BRIDGE OVER RIVER MAHANADI AT

6TH KM OF SAMBALPUR-SONEPUR ROAD TO CHADHEIPANK-BINCA
ROAD - A CASE STUDY
S. C. Choudhury, A. Simadri Dora, M. K. Tripathy and P. M. Tripathy
Works Department, Govt. of Odisha, India

Email: saratchoudhury69@gmail.com
ABSTRACT
The paper is to document the construction of the high level bridge over river Mahanadi at 6th km of
sambalpur-sonepur road to chadheipank-binca road. The work is administratively approved for Rs. 39.24 crores
by Rural Development Department, Government of Odisha. The bridge is located on downstream side of
Hirakud Dam. This is a pre-stressed concrete bridge with more than one kilometre of length having 42 spans.
This paper highlights the problem faced at different stages during the construction of the bridge.
KEYWORDS: PSC girder, pier, abutment, pile cap, high flood level
1. OVERVIEW OF THE BRIDGE
The work is administratively approved for Rs. 39.24 crores by Rural Development Department, Government of
Odisha. The survey and investigation work was done in the year 2005.The general arrangement drawing (GAD)
was prepared during the year 2006.The maximum flow was estimated for 47,798 cumecs. The peak velocity
during high flood is 4.7 m/sec. It comes under rural connectivity programme; benefitting 29 numbers of Gram
Panchayats (251 number of villages). The bridge is under construction by Rural Development Department, Govt.
of Odisha. The project implementing unit is Rural works Division, Sonepur (Odisha). The length of the bridge is
1058m and number of spans is 42. Length of each span is 25.174m, number of piers is 41 and number of
abutments is 2. Out of 41 numbers of piers, 17 numbers have pile foundation underneath, while 24 numbers
have open foundation with individual RCC footing. The 2 abutments have open foundation with RCC footing.
The pile foundations consists of 4 numbers of 1m diameter piles under each pier with depth of piles varying
from 4m to10m depending upon the soft or hard rock strata met underneath. The RL of bottom of pile cap is
91.85m; top of pile cap is 93.35m.and depth of pile cap is1.5m.
The bridge design is done by Uttam Buildcon (P) Ltd, Raipur, proof checked by the Department of Civil
Engineering, National Institute of Technology Raipur and approved by Chief Engineer, Rural works-I,
Bhubaneswar, Govt. of Odisha. The executing agency is Uttam Buildcon infrastructure Co. Pvt. Ltd.
2. SUPERSTRUCTURE
It is a composite structure with RCC deck slab of 8.4m overall width including kerbs. The height of hand rail is
0.86m above kerbs. It has 3 number of prestressed concrete (PSC) girders in each span, simply supported on
both ends and placed over elastomeric bearings .The I-section PSC girders are cast in casting yards on both
sides of the river bank, prestressed and carried over to position using cranes and trailers kept over elastomeric
bearings.
9630
Figure 1 Cross section of pier having pile foundations (all dimensions are in ‘mm’)
3. SUB- STRUCTURE
The circular piers are 1.6m diameter, 9.63m height over which flanged pier caps with 6 numbers of pedestals
constructed on top of piers (for both side of spans) and over the pedestals neoprene bearings are fixed in
position.
Three number of prestressed concrete (PSC) girders are provided in each span, simply supported on both ends.
The I-section PSC girders are cast in casting yards on both sides of the river bank, carried over to position using
cranes and trailers, kept over neoprene bearings. The size of PSC girders are as follows. Each PSC girder has
4 numbers of cables, including one dummy cable. So, three numbers of cables are stretched using Dynamic
Company’s multiple jack system. Each cable consists of 19 wires of 11.7mm diameter each.
The clear width of the bridge in between kerbs is 7.5m.The kerbs provided on each side is 450mm width at
bottom, 425mm width at top and 300mm height above deck slab. The height of hand rail over the kerb is 0.86m.
4. PROBLEMS FACED IN THE FIELD:
(1) Initially, it was thought of providing well foundation under each pier. So initial borings are done along
the proposed bridge axis. It was observed that rock strata are met with in all trial bore holes from river
bed or underneath. It is felt that sinking of wells inside rock strata is very difficult, time consuming
and costly. Subsequently after careful consideration, it was decided to go for bored-cast-in-situ pile
foundation for economic and quick completion of the bridge. So pile foundation and open foundation
are adopted as per site condition.
705
(2) The bridge is located on downstream side of Hirakud Dam. During design of the bridge, the HFL was
taken as RL 100.60m. But due to opening of large number of gates of Hirakud Dam during rainy
season, the HFL was recorded as RL 102.96m with an increase of 2.36m of HFL during construction.
Mean while 5 numbers of pier with T-flanged pier caps are already cast. So subsequently, each pier
flange was raised by 2.06m for 5 number of piers already cast to match with the new HFL.
756 mm
1500 mm
Four numbers of cables
756 mm
Figure 2 Cross section of PSC Girders
ACKNOWLEDGEMENT
We are thankful to the Engineers of Uttam Buildcon infrastructure Co Pvt. Ltd, Raipur (Chatisgarh), Junior
Engineer, Assistant Engineers and Executive Engineers, Rural works Division, Sonepur (Odisha), who have
helped us to prepare this case-study of the H.L. Bridge.
706
Paper No.: 131
PERFORMANCE EVALUATION OF TUNED LIQUID DAMPER ON

RESPONSE OF HIGH RISE STRUCTURE UNDER HARMONIC LOAD.
1 2
D.Jena and K.C. Biswal
1
Research Scholar, Dept. of Civil Engineering, National Institute of Technology, Rourkela, Odisha, India
2
Professor, Dept. of Civil Engineering, National Institute of Technology, Rourkela, Odisha, India
Email: deba_datta10@rediffmail.com, kcbiswal2003@yahoo.co.in
ABSTRACT :
Present trend in civil engineering is to build more flexible structures like long-span bridges and high-rise
buildings. These types of structures are at risk during dynamic loading events (e.g., earthquake, strong winds).
Energy dissipating devices are installed for the protection of these flexible structures, their material content and
human occupants, against damage induced by the dynamic actions. One type of passive damping system, called
the tuned liquid damper has been employed in practice. A numerical model based on finite element method has
been developed successfully to evaluate the performance of tuned liquid damper for the control of the
undesirable vibration of high rise structures under harmonic load.
KEYWORDS: High rise structure, Dynamic action, Structural vibration, Tuned liquid damper
1. INTRODUCTION
Structures are vulnerable to the vibrations caused due to earthquake and wind. Structural control method will be
able to make significant contribution to the problem area which is of great economic and social importance.
The evolution of civil engineering field of structural control has been rapid, attracting the interest and attention
of scores of researchers over the past several decades. One type of passive damping system, called the tuned
liquid damper has been employed in practice. The basic principle involved in applying a tuned liquid damper
(TLD) is to provide the restoring and dissipative mechanisms required to reduce the dynamic response of
structure. A TLD, which usually is a rigid tank with shallow water in it, is attached rigidly to a structure. Tuning
the fundamental linear sloshing frequency of the TLD to the structure’s natural frequency causes large amount
of sloshing and wave breaking at the resonant frequencies of the combined TLD-structure system that dissipates
a significant amount of energy. The TLD has several potential advantages over other damping systems, such as
low installation and maintenance cost, fewer mechanical problems, activated at even low excitation levels, and
easy adjustment of the damper parameters even after installation by adjusting the water level in the tank.
Bauer [1] is the first to propose a damping device consisting of a liquid container filled with two immiscible
liquids, in which the motion of the interface is able to dampen the structure. Wakahara et al. [2] and Tamamura
et al. [3] showed the effectiveness of TLDs installed in real structures such as Nagasaki airport tower, Yokohama
Marine tower, and Shin Yokohama Prince (SYP) hotel to reduce the structural vibration. Modi & Seto [4]
conducted numerical study on rectangular TLDs, accounting for nonlinear effects. They included the effects of
wave dispersion as well as boundary-layers at the walls, floating particle interactions at the free surface, and
wave breaking. Koh et al [5] suggested multiple TLDs to reduce multi modal responses of high rise structures to
earthquake ground motion excitations. The results show that it is beneficial to use dampers tuned to several
vibration modes of the structure. Chang & Gu [6] studied experimentally, the control effects of rectangular
TLDs installed on a tall building that vibrates due to vortex excitation. Sun, et al. [7] developed an analytical
model for TLD, based on shallow water wave theory, which proved to be very effective if the wave is non
breaking. They extended this model to account for effect of breaking waves by introducing two empirical
coefficients identified experimentally. Modi and Akinturk [8] focused on the installation of two-dimensional
wedge-shaped obstacles to amplify TLD energy dissipation efficiency. Banarji et al. [9] studied the effectiveness
of a rectangular TLD in reducing the earthquake response of structures for various values of natural time periods
and structural damping ratios. Furthermore, an attempt is made to define appropriate design parameters of the
TLD that is effective in controlling the earthquake response of a structure.
2. OBJECTIVE OF THE PRESENT STUDY:

The objective of the present study is to evaluate the performance of Tuned liquid damper for controlling
structural vibration. It includes development of a numerical code based on Finite element method.
3. NUMERICAL MODEL FOR TLD
The rigid rectangular TLD tank which has a length L, a width b, and a mean liquid depth h, is subjected to a
horizontal base motion xs. The local Cartesian coordinate system (o-x-z) is attached to the tank, and its origin is
placed at the centre of the mean liquid surface. It is assumed that the liquid is incompressible, irrotational, and
inviscid.
Figure 1 Tank geometry
3.1. Governing equation and Finite element formulation
The governing differential equation for the liquid in terms of velocity potential ( φ ) is
∇ φ=0
2
(1)
In which φ = φ (x,z,t)
Equation (1) is solved using finite element technique and with the appropriate time-dependent boundary
conditions as specified below
708
(a) At liquid-structure interface
∂φ / ∂n = v n (2)
On B s
where
n is outwardly drawn normal to the surface of structure boundary
vn is the common velocity of the liquid and boundary in the direction
normal to the surface
(b) At liquid free surface

Considering the effect of small-amplitude waves on a liquid surface, the condition yields
∂ 2 φ / ∂t 2 + g ∂φ / ∂z = 0 on Bf (3)
(c) At the bottom of the tank

∂φ / ∂n = 0 on Bb (4)
3.1.1. Idealization of liquid
The finite element formulation is based on Galerkin weighted residual method. A four noded isoparametric
quadrilateral element is employed to discretize the liquid domain. The velocity potential ( φ ) is approximated
as
N
φ (x,z,t) = ∑N
j =1
j ( x, z ) φ j (t) (5)
Applying divergence theorem to the residual form of governing differential equation for the liquid and
minimizing the energy function, we get
∂N i N ∂N j ∂N i N ∂N j ∂φ
∫( ∂x
∑
j=1 ∂x
φj +
∂z
∑j=1 ∂z
φ j ) dV = ∫N i
∂n
ds (6)
V B
in which B = B s + B f + B b
Substituting equations (2), (3) and (4) in equation (6) we have

∂N i N ∂N j ∂N i N ∂N j 1 N
∫( ∂x
∑
j=1 ∂x
φj +
∂z
∑j=1 ∂z
φ j ) dV = ∫ N i v n ds - g ∫ N i ∑ N j &φ& ds
j=1
(7)
V Bs Bf
The equation (7) can be reduced to
[M f ] {&φ& } + [K f ] { φ } = {F }
p (8)
in which the elements of matrices [M f ], [K f ] and {F }

p are given by
1
M ij =
g
∑ ∫N N i j ds (9)
Bf
709
 ∂N i ∂N j ∂N i ∂N j 
K ij = ∑ ∫  ∂x ∂x
+
∂z ∂z
 dV (10)
V  
Fi =- ∑ ∫ N i v n ds (11)
Bs
4. PROBLEM STATEMENT:
A 10 storey RCC frame structure has been considered for analysis.

Data for analysis of the frame:
a) Type of the structure Multi-storey rigid jointed plane frame.

b) Number of stories Ten (G+9)
c) Floor H 3.5m
d) Bay width 5m
e) Infill wall 250 mm thick in longitudinal direction and 150 mm thick in transverse direction.
f) Imposed Load 3.5 kN/m2
g) Materials Concrete (M25) & Steel (Fe415)
h) Size of columns 250mm x 450 mm
i) Size of beams 250 mm x 400 mm in longitudinal direction and 250 mm x 350 mm in transverse
direction.
j) Depth of slab 100 mm
4.1. Structure idealization:
A multi degree of freedom (MDOF) structure with a TLD attached to it and subjected to a ground motion is
shown in figurre. The equation of motion for this MDOF structure is
M X + C X + K X =− M I X + F (12)
Where,
[M] = The global mass matrix of the frame structure
[C] = The global damping matrix of the frame structure
[K] = The global stiffness matrix of the frame structure
{X} = The global nodal displacement vector
Xg = Ground Acceleration
I = Unit vector
{FTLD} = Resisting force to the structure at corresponding nodes due to TLD
The equations (8) and (12) are coupled, second order, ordinary differential equations. These sets of coupled
equations are solved using Newmark Beta method. First the equation (12) is solved without considering liquid
force. The acceleration of the roof is transferred to the TLD and the developed base shear is applied to the
structure. The displacement of the structure and base shear of the TLD are calculated in a iterative manner
within each time step till the desired level of convergence is achieved.
710
5. RESULTS AND DISCUSSION:
Investigations were conducted to study the dynamic behaviour of a structure with TLD when subjected to
harmonic ground motion. The harmonic ground motion is defined by its excitation frequency and amplitude. As
the harmonic ground motion consists of a single frequency, the analysis of structure with TLD will provide an
understanding of the behaviour of the structure-TLD system for this kind of ground motion. It also gives a
general idea of the behaviour of the structure when it is subjected to more realistic earthquake ground
motions.The dynamic response of the structure is also investigated with and without TLD under the action of
various ground motions recorded earlier. The ground motions are random in nature. For the present study four
different time histories of ground motion are selected to investigate the behaviour of the structure-TLD system.
The top storey deflection of the structure with TLD is found to be reduced with that of the structure without
TLD.
5.1. Structure subjected to sinusoidal loading:
711
6. CONCLUSION:
It is observed that the performance of TLD in the structure considered is satisfactory in dissipating the input
vibration energy and reducing the lateral deflection during dynamic loading events. The basic parameters
affecting TLD effectiveness are discussed on the basis of theoretical studies. The parameters are affected by
several physical quantities such as TLD tank size and liquid depth. The performance of TLD is significantly
affected by greater depth ratio. The contribution of the damper is reduced as major part of the liquid present in
the container does not take part in sloshing. It is concluded that for designing an effective TLD, the ratio of the
structural frequency to that of the TLD should be 1. For larger or smaller frequency ratios, the damper is said to
be mistuned and becomes less effective.
ACKNOWLEDGEMENTS
The support of the Department of Science and Technology, Govt. Of India for this project is gratefully
acknowledged.
REFERENCES
[1] Bauer, H.F. (1984). Oscillations of immiscible liquids in a rectangular container: A new damper for excited
structures. Journal of Sound and Vibration, 93(1), 117-133.
[2] Wakahara, T., Ohyama T., and Fujii K., (1992). Suppression of Wind-Induced Vibration of a Tall Building
using Tuned Liquid Damper. Journal of Wind Engineering and Industrial Aerodynamics, (41- 44),
1895-1906.
[3] Tamura, Y., Fujii, K., Ohtsuki, T., Wakahara, T., and Kohsaka R., (1995). Effectiveness of tuned liquid
dampers under wind excitation. Engineering Structures, 17 (9), 609-621.
[4] Modi, V.J, and Seto, M.L, (1997). Suppression of flow-induced oscillations using sloshing Liquid
dampers: analysis and experiments. Journal of Wind Engineering and Industrial Aerodynamics (67- 68),
611-625.
[5] Koh, C.G., Mahatma, S., and Wang, C.M., (1995). Reduction of structural vibrations by multiple-mode
liquid dampers. Engineering Structures, 17 (2), 122-128.
[6] Chang, C.C. and Gu, M., (1999). Suppression of vortex-excited vibration of tall buildings using tuned liquid
dampers. Journal of Wind Engineering and Industrial Aerodynamics, 83, 225-237.
[7] Sun, L.M., Fujino Y., Pacheco B.M., and Chaiseri P., (1992). Modeling of Tuned Liquid Damper (TLD).
Journal of Wind Engineering and Industrial Aerodynamics, (41- 44), 1883-1894.
[8] Modi, V.J. and Akinturk A.,(2002). An efficient liquid sloshing damper for control of wind-induced
instabilities. Journal of wind engineering and industrial aerodynamics 90, 1907-1918.
[9] Banerji P., Murudi, M., Shah, A.H. and Popplewell N., (2000). Tuned liquid dampers for controlling
earthquake response of structures. Earthquake Engg & Struct. Dyn, 29(5), 587-602.
712
Paper No.: 132
SEISMIC FRAGILITY ANALYSIS OF AXIALLY LOADED SINGLE PILE

1 2
Ganesh R , Haran Pragalath D.C. , Robin Davis3, Pradip Sarkar3 and S.P.Singh3
1
M.tech student, Dept. of civil Engineering, National Institute of Technology, Rourkela, India
2
PhD student, Dept. of civil Engineering, National Institute of Technology, Rourkela, India
3
Dept. of civil Engineering, National Institute of Technology, Rourkela, India
Email: 212ce1018@nitrkl.ac.in, haran5441@gmail.com
ABSTRACT :
Performance-based seismic design is the latest trend to earthquake resistant design for Sub structure and Super
structures. Many parameters involved in seismic design have uncertainty associated with them. Characterizing
the probabilistic nature of these parameters can be done through the use of ‘Fragility Curves’. A fragility
analysis assesses the probability that the seismic demand placed on the structure exceeds the capacity
conditioned on a chosen IM representative of the seismic loading. Demand (D) and capacity (C) are assumed to
follow a lognormal distribution, and the probability of exceeding a specific damage state for a particular
component can be estimated with the standard normal cumulative distribution function as per Cornell et. al. [9].
This paper presents the performance of axially loaded single pile subjected to different earthquakes of varying
PGAs. FEM software called openseesPL is used for Times history analysis. Floating pile is modelled by
beam-column elements, and rigid beam-column elements are used to model the pile diameter. A nonlinearity of
soil is introduced using pressure dependent multiplied model and the pile is modelled as elastic. Proper
boundary conditions, simulating radiation effects are used. To conduct nonlinear dynamic analysis, forty four
natural time history data are selected, and modified to match with Indian response spectrum (IS 1983-2002)
using WavGen (Mukharjee and Gupta, 2002). Uncertainty in soil properties is included. A Latin hypercube
sampling technique is adopted for generation of random variables which follows a normal distribution with
mean and standard deviation. The numerical modelling was done from the generated random values which then
subjected to different PGA for performance evaluation.
KEYWORDS: Fragility, axially loaded pile, openseesPL, Response spectrum
1. INTRODUCTION
Significant number of studies has been reported in the past for dynamic analysis of pile and pile groups in
liquified and non liquified soils. Based on FE analysis, Maheshwari et.al (2004) reported that the soil
nonlinearity increases the pile head’s structural responses at low frequencies and at higher frequencies its effect
is insignificant under for both harmonic and transient excitation. Adak et.al. (2004) reported that the failure due
to buckling instability of the axially loaded pile is the predominant failure mechanism during liquefaction and
also explained the failure mechanism of pile in layered soils. Some researchers like Shahrour et al. (2001)
conducted a 3-D FEM analysis of micro piles using a finite element program, PECPLAS. The soil-micro pile
structure system was assumed to be elastic with Rayleigh material damping. Juran et al. (2001) etc. studied the
behaviour of micro piles under different modelling techniques including numerical modelling. Juran et al. (2001)
used finite difference software namely LPILE and GROUP to study the structure-soil-micro pile behaviour and
also to investigate the response of the micro pile systems subject to earthquake loading. However a limited
number of studies over the past years are reported for the performance evaluation of pile under different limit
states condition under earthquake. In this study, we conduct a finite element simulation of Pile No. 2 of the
Arkansas test series (Alizadeh and Davisson, 1970) using the OpenSeesPL interface. This pipe pile is subjected
to earthquakes of different PGAs. The pile is subjected to axial compressive load of 750 kN. The present work
is to assess the performance of typical axially loaded single pile under earthquakes in probabilistic approach.
Uncertainties in the soil properties are also considered.
2. METHODOLOGY
2.1. Development of Fragility curves
A fragility function represents the probability of exceedance of the selected Engineering Demand Parameter
(EDP) for a selected structural limit state (DS) for a specific ground motion intensity measure (IM). These
curves are cumulative probability distributions that indicate the probability that a component/system will be
damaged to a given damage state or a more severe one, as a function of a particular demand.Fragility curve
damaged to a given damage state or a more severe one, as a function of a particular demand.Fragility curve can
be obtained for each damage state and can be expressed in closed form as using Eq. 1
P (C-D ≤ 0|IM) = Φ (1)
where, C is the drift capacity, D is the drift demand, Sd is the median of the demand and Sc is the median of the
chosen damage state (DS). βd/IM and βc are dispersion in the intensity measure and capacities respectively. Eq.
1 can be rewritten as Eq. 2 for component fragilities (Nielson, 2005) as,
P (DS|IM) = Φ (2)
Where, IMm = exp , a and b are the regression coefficients of the probabilistic Seismic Demand
Model (PSDM) and the dispersion component, is given as,
(3)
The dispersion in capacity, βc is dependent on the building type and construction quality. For βc, ATC 58 50%
draft suggests 0.10, 0.25 and 0.40 depending on the quality of construction. In this study, dispersion in capacity
has been assumed as 0.25.
It has been suggested by Cornell et. al (2002) that the estimate of the median engineering demand parameter
(EDP) can be represented by a power law model as given in Eq. 4.
(4)
In this study, threshold displacement (δ) at the pile top is taken as the engineering damage parameter (EDP) and
peak ground acceleration (PGA) as the intensity measure (IM).
2.2. Ground motion data
The number of ground motions normally required for an unbiased estimate of the structural response is 3 or 7.
However, ATC 58 50% draft recommends a suite of 11 pairs of ground motions for a reliable estimate of the
response quantities. ASCE/SEI 41(2005) suggests 30 recorded ground motions to meet the spectral matching
criteria for NPP infrastructures. A set of forty four Far-Field natural Ground Motions are collected from
Haselton and Deierlein (2007). These are converted to match with IS 1893 (2002) spectrum using a program,
WavGen developed by Mukherjee and Gupta (2002). Figure 2 shows the Response spectrum for converted
ground motions along with Indian spectrum.
714
Figure 2 Response Spectra for 44 converted ground motions along with
IS 1893 (2002) design spectrum axial strain curves
3. NUMERICAL STUDY
3.1 Pile data employed in the OpenSees simulation
The pile considered in the present study is circular in section with a diameter of 0.406m and a wall thickness,
t= 7.925 mm, Table 3.1 Shows the details of pile properties taken from Bowles (1988).
Table 3.1 Properties of Pile

Sl. no Description Value
1 Pile diameter 0.406 m
2 Pile length 16.5 m
3 Moment of inertia 834.2 in2
4 Youngs modulus of pile, (E) 29000 ksi
3.2 Soil data
The pile is embedded in a uniform soil layer and its properties are shown in the Table 3.2. Projection of pile
length above ground surface is 0.3048 m and below the ground surface is 16.195 m. Uncertainty in soil
parameters are taken from Phoon and Kulhawy (1999). Forty four statistical models are developed according to
Latin Hyper cube sampling.
Table 3.2 Properties of soil

Sl no Description Mean Value Cov (%) Reference
1 Reference confinement 80KPa - Lu et al. (2006)
2 Shear modulus, Gs 10.88Ksi - Lu et al. (2006)
3 Bulk modulus, B 29Ksi - Lu et al. (2006)
4 Poisson’s ratio 0.33 - Bowles. (1988)
5 Unit weight, (γ') 62.8pcf 14 Bowles. (1988)
6 Friction angle, φ 320 10 Bowles. (1988)
7 Shear strain, γmax 10% - Bowles. (1988)
715
3.3 Modelling
In view of symmetry, a half-mesh (2,900 8-node brick elements, 23 beam-column elements and 207 rigid
beam-column elements in total) is studied as shown in Figure 2. Length of the mesh in the longitudinal direction
is 520 ft, with 260 ft transversally (in this half-mesh configuration, resulting in a 520 ft x 520 soil domain in
plan view). Layer thickness is 80 ft (the bottom of the soil domain is 27.2 ft below the pile tip, so as to mimic
the analytical half-space solution). The floating pile is modelled by beam-column elements, and rigid
beam-column elements are used to model the pile size (diameter). The following boundary conditions are
enforced:
i) The bottom of the domain is fixed in the longitudinal (x), transverse (y), and vertical (z) directions.
ii) Left, right and back planes of the mesh are fixed in x and y directions (the lateral directions) and free
in z direction.
iii) Plane of symmetry is fixed in y direction and free in z and x direction (to model the full-mesh 3D
solution). The lateral load is applied at the pile head (ground level) in x (longitudinal) direction. The
above simulations were performed using OpenSeesPL (Lu et al. 2006).
Figure 2 FEM Modelling of axially loaded pile in OpenseesPL
4. PERFORMANCE LEVEL
The design threshold displacements at the ground level can be set to approximately 2%, 3% and 4% of the pile
diameter for the serviceability limit state in the Normal Situation and the Frequent Earthquake Situation
(Masahiro Shirato et.al, 2003). These values are generally larger than the past values of 0.01B and 15 mm for a
typical range of pile diameters in highway bridge foundations. Hence the fragility curves for different design
threshold displacement at the ground level are calculated.
5. ANALYSIS
Forty Four Time history analyses are performed for the developed statistical models using openseesPL. Selected
forty four earthquakes are scaled from 0.065g to 0.72g. Displacement at each node is recorded throughout the
analysis. Maximum Displacement in Pile is considered as an Engineering Damage parameter. It is found that
Pile top always shows the maximum response.
716
Figure 3 PSDM for Pile
D – Diameter of pile
Figure 4 Fragility curves for design lateral deformation
6. DISCUSSIONS AND CONCLUSIONS
The response of axially loaded pile under earthquake loading has been studied in probabilistic manner and the
corresponding fragility curve has been formulated for different limit state of serviceability condition 2%, 3%
and 4% diameter of pile as reported in the literatures. Figure 3, shows the Probabilistic Demand model between
Intensity Measure and Engineering Demand parameter. It is fitted through the power law equation having R2
value as 0.9566 and corresponding uncertainty is calculated (βd/IM =0.125). Figure 4, shows the exceedance
probability of pile for various serviceability limit states. It is seen that for a 50 % probability of exceedance for
2%, 3% and 4% limit state condition of serviceability is 0.079g, 0.124g and 0.17g, respectively. It is observed
that the relative displacement at the top of the pile increases as the PGA increases. Pile considered for the
present study is not designed for seismic loading,, it shows the probability of exceedance as high for small PGA.
REFERENCES
[1] Alizadeh, M and Davisson, M. T. (1970). Lateral Load Tests on Piles – Arkansas River Project, JSMFD,
ASCE, Vol. 96, SM5, September, pp. 31-40
[2] Bowles, J. E. (1988). Foundation Analysis and Design, 4th Edition, McGraw-Hill Book Co., New York,
NY 10020.
717
[3] Phoon, K. K. and Kulhawy, F. H. 1999. Characterization of geotechnical variability. Can. Geotech. J. 36(4):
612-624.
[4] IS 456 (2000). Indian Standard for Plain and Reinforced Concrete. Code of Practice, Bureau of Indian
Standards. New Delhi.
[5] FEMA 356 (2000). Pre-standard and Commentary for the Seismic Rehabilitation of Buildings, American
Society of Civil Engineers. USA.
[6] Shahrour, I., Sadek, M., and Ousta, R. (2001). Seismic behavior of micropiles used as foundation support
elements. Transportation Research Record 1772, pp. 84-90.
[7] Juran, I., Benslimane, A., and Hanna, S. (2001). Engineering analysis of the dynamic behavior of micropile
systems. Transportation Research Record, 1772, Paper No. 01-2936, pp. 91-106.
[8] IS 1893 (2002) PART 1. Indian standard Criteria for Earthquake Resistance Design of Structures. Bureau
of Indian standards, New Delhi
[9] Mukherjee, S and Gupta, V.K. (2002). Wavelet-based generation of spectrum compatibtime-histories. Soil
dynamics and Earthquake engineering 22, 799-804
[10] Cornell, C. Allin, Fatemeh Jalayer, Ronald O. Hamburger and Douglas A Foutch (2002). The probabilistic
basis for the 2000 SAC/FEMA steel moment frame guidelines. Journal of Structural Engineering 128(4),
526-533.
[11] Masahiro Shirato1et.al, (2003). Geotechnical Criterion For Serviceability Limit State Of
Horizontally-Loaded Deep Foundations.
[12] Maheshwari, B.k. and Truman, k.z. (2004). 3-D Finite Element Nonlinear Dynamic Analysis For
Soil-Pile-Structure Interaction, 13th World Conference on Earthquake Engineering Vancouver, B.C.,
Canada, Paper No. 1570.
[13] SONG, S. T; CHAI, Y. H. and Tom H. HALE, (2004). Limit State Analysis Of Fixed-Head Concrete Piles
Under Lateral Loads. 13th World Conference on Earthquake Engineering Vancouver, B.C., Canada,
August 1-6, Paper No. 971.
[14] Adak, G. et.al. (2004). Failure Mechanisms of Piled Structures in Liquefiable and Layered Soils.
[15] ASCE 7 (2005). Minimum Design Loads for Buildings and Other Structures. American Society of Civil
Engineers. USA. 2005.
[16] Mazzoni, S., McKenna, F., and Fenves, G. L. (2006). Open system for earthquake engineering simulation
user manual, Pacific Earthquake Engineering Research Center University of California, Berkeley.
[17] Jinchi Lu; Zhaohui Yang, and Ahmed Elgamal (2006). OpenSeesPL Three-Dimensional Lateral
Pile-Ground Interaction, User's Manual, Version 1.00. Report No. SSRP-06/03, Department of Structural
Engineering, University of California, San Diego.
[18] Nielson, B.G.(2005). Analytical fragility curves for highway bridges in moderate seismic zones. Ph.D
Thesis, Georgia Institute of Technology.
[19] Ahmed Elgamal and Jinchi Lu,(2007). finite element analysis of arkansas test series pile #2 using opensees
(with lpile comparison).
718
[20] Haselton, C.B. and Deierlein,G.G (2007). Assessing seismic collapse safety of modern reinforced concrete
frame buildings . Blume Earthquake Engineering Research Centre Technical report no. 156, stanford
university, pp. 313.
[21] Ramin Motamed et.al, (2008). Shaking Model Tests On Behavior Of Group Piles Undergoing Lateral Flow
Of Liquefied Subsoil. 14th World Conference on Earthquake Engineering, 12-17, Beijing, China.
[22] ATC 58 50% Draft, 2009. Guidelines for Seismic Performance Assessment of Buildings, Applied
Technology council, Redwood City, CA.
[23] MatLab (2009) MatLab - Programming software for all kind of problems [online]. < http://
www.mathworks.com/ > (July 20, 2012)
[24] Abderrahmane Kibboua et.al (2011). “Analytical Fragility Curves For Typical Algerian Reinforced
Concrete Bridge Piers”, Structural Engineering and Mechanics, Vol. 39, No. 3 411-425.
719
Paper No.: 133
OPTIMAL DESIGN OF A TUNED MASS DAMPER USING A SIMPLE

OPTIMIZATION TOOL
Priyadarshi Das1 and ManasRanjan Das 2*

1
Assistant professor, Dept. of Civil Engineering, Institute of Technical Education and
Research,Bhubaneswar,India
2
Associate Professor, Dept. of Civil Engineering, Institute of Technical Education and
Research,Bhubaneswar,India
*Email: manasdas@soauniversity.ac.in, priyadarshidas@soauniversity.ac.in,saratdas@rediffmail.com
ABSTRACT:
A Tuned Mass Damper (TMD) is a device consisting of a mass, a spring, and a damper that is attached to
a structure in order to reduce the dynamic response of the structure. The frequency of the damper is tuned
to a particular structural frequency so that when that frequency is excited, the damper will resonate out of
phase with the structural motion. Energy is dissipated by the damper inertia force acting on the structure.
The purpose of adding the mass damper is to limit the motion of the structure when it is subjected to a
particular excitation. The design of the mass damper involves specifying the mass, stiffness, and damping
coefficient. In the present work an attempt has been made for a study on the optimal choice of these
quantities and assessment of optimal behavior of the structure attached with a Tuned Mass Damper
(TMD) using a simple optimization tool, i.e. Microsoft Excel based Solver.
KEYWORDS: Solver, Tuned Mass Damper, SDOF.
1. INTRODUCTION
Tuned mass dampers reduce the effects of harmonic vibration in buildings or other structures. A relatively
small mass is attached in such a way that it can dampen out a very narrow band of vibration of the
structure. The frequency of the damper is tuned to a particular structural frequency so that when that
frequency is excited, the damper will resonate out of phase with the structural motion. Energy is
dissipated by the damper inertia force acting on the structure.TMD is most effective for periodic
excitations and wind loads. The TMD concept was first applied by Frahm in 1909 (Frahm, 1909) to
reduce the rolling motion of ships as well as ship hull vibrations. A theory for the TMD was presented
later in the paper by Ormondroyd and Den Hartog(1928),followed by a detailed discussion of optimal
tuning and damping parameters in Den Hartog‘s book on mechanical vibrations (1940). Hrovat et
al.(1983) presented SATMD, a TMD with time varying controllable damping. To overcome the
frequency-related limitations of TMDs, more than one TMD in a given structure, each tuned to a different
dominant frequency, can be used..Mostaghel and Khodaverdian (1988) wrote a paper on the dynamic
response of base-isolated structures which formed a skeleton for many of the derivations presented in this
study. Their paper focused on friction-based isolation systems .The work presented in their paper is,
however, restricted to unidirectional motion, considering only one horizontal degree of freedom and the
vertical ground motion, which is integral to the frictional effect.In Villaverde(1994), three different
structures were studied, in which the first one is a 2D two story shear building the second is a three-
dimensional (3D) one-story frame building, and the third is a 3D cable-stayed bridge, using nine different
kinds of earthquake records. Numerical and experimental results show that the effectiveness of TMDs on
reducing the response of the same structure during different earthquakes This implies that there is a
dependency of the attained reduction in response on the characteristics of the ground motion that excites
the structure. This response reduction is large for resonant ground motions and diminishes as the
dominant frequency of the ground motion gets further away from the structure's natural frequency to
which the TMD is tuned. James M. Kelly is an influential researcher in the area of base-isolation. His
book, Earthquake Resistant Design with Rubber (1996), discusses the theory and application of base-
isolation in detail.
2. DESIGN PHILOSOPHY FOR OPTIMAL PERFORMANCE OF TMD

2.1 Tuned Mass Damper
The purpose of using TMD is to produce an artificial force originated from the mass of TMD in order to
counteract the structural vibration at which the TMD is installed. TMD is attached to a structure in order
to reduce the dynamic response of the structure. The frequency of the damper is tuned to a particular
structural frequency so that when that frequency is excited, the damper will resonate out of phase with the
structural motion. The mass is usually attached to the building via a spring-dashpot system and
energy is dissipated by the dashpot as relative motion develops between the mass and the
structure. The concept of the tuned mass damper is illustrated using the two- mass system shown in
Figure. Here, the subscript ‘d’ refers to the tuned mass damper; the structure is idealized as a single
degree of freedom system. The Fundamental Equation of motion
Figure 1 SDOF-TMD SYSTEM
The purpose of adding the mass damper is to limit the motion of the structure when it is subjected to a
particular excitation. The design of the mass damper involves specifying the mass , stiffness , and
damping coefficient . The optimal choice of these quantities is discussed in subsequent chapters. Here the
near-optimal approximation for the frequency of the damper is used to illustrate the design
procedure.
2.2Optimized Performance Evaluation
For the present study damping is included in the mass damper, as shown in Figure . The equations of
motion for this case are:
721
− (1)
− − − (2)
The inclusion of the damping terms in Eqs. (1) and (2) produces a phase shift between the periodic
excitation and the response. It is convenient to work initially with the solution expressed in terms of
complex quantities. We express the excitation as
(3)
̂ (4)
Where and ̂ are real quantities. The response amplitudes are
! (5)
!!!! (6)
where the response amplitudes, and , are considered as complex quantities.
Converting the complex solutions to polar form leads to the following expressions:
̂
! "# $%
− "& $'
(7)
̂
!!!! "+ , $-
− ". , $-
(8)
Where the " factors define the amplification of the pseudo-static responses, and the 0 ’s are the phase
angles between the response and the excitation.
The various H and δ terms are as follows:
1(2 & − 3& )& (2235 )²

"# (9)
|7& |
1((1 : )2² − 3²)² (25 32(1 : ))²

"& ( 10)
|7& |
3²
"+ (11)
|7& |
1
". (12)
|7& |
|7& | <((1 − 3& )(2 & − 3& ) − 3²2² :))² (25 32(1 − ρ& (1 : )))² (13)
"& varies with forcing frequency for specific values of mass ratio : and frequency ratio 2, and various
0, there are two peaks with infinite amplitude located
on each side of 3 1 . As 5 is increased, the peaks approach each other and then merge into a single
values of the damper damping ratio,5 . When5
722
peak located at 3 1 . The behavior of the amplitudes suggests that there is an optimal value of 5 for a
given damper configuration.
If we plot a graph between ρ and "& , it is noted that for different values of 5 all the curves are passing
through two common points and location of these two points depend only on : and 2 .
Figure 2 "& versus ρ curves
Again for ρ₁ and ρ₂ Equation for ground motion, can be optimally written as
1 :
("& *@ (14*
|1 ρ₁²(1 : *|
1 :
("& *C (15*
|1 ρ₂²(1 : *|
The above equations show different values for "& at points P and Q. For optimal behavior maximum
amplitude is to be minimized. As a first step, the values of "& for ρ₁ and ρ₂ need to be equal. This
#
#F :
produces a distribution that is symmetrical about 3² . Then, by increasing the damper damping
represents the optimal performance of the TMD system. A further increase in 5 causes the peaks to
ratio,5 , the peak amplitudes can be lowered until the peaks coincide with points P and Q . This state
merge and the amplitude to increase beyond the optimal value.
The relation between the optimal tuning frequency and the mass ratio:
2G √1 0.5 : /1 : (16)
( *G 2G (17)
The corresponding roots and optimal amplification factors are
1 √0.5 :
J3#,& LG M (18*
1 :
The expression for the optimal damping at the optimal tuning frequency is
723
!!!(3
− √0.5 : )
(5 )G M (19)
8(1 : )(1 − 0.5 : )
3. METHODOLOGY IN PRESENT STUDY FOR ASSESSMENT OF OPTIMAL

PERFORMANCE OF TMD.
3.1 Optimization problem formulation
The optimizing parameters are : , ρ₁ , ρ₂ , 2 , 7& ,5 ,"&. As stated above "&has two values, one
maximum and other minimum.For optimal performance maximum amplitude need to be minimized.
3.1.1 Details of objective function and constraints
' '
<J(#F : ) ' ,O' L FJ&PQ O (#F : )L
|R' |
Minimize2(N)
Where (N) "& , Subject to constraints
: ≤ 0.02, 3# ≤ 0.99,3& ≥ 1.01
The design variables are
Mass ratio : , Frequency ratios 3# UV 3& , damping ratio of TMD 5 , Frequency ratio f.
Minimizing the objective function "& different optimal parameters have been observed for all values of ξd.
3.2 Optimization tool used in the present study

In the present study an attempt is made to optimize the performance of a TMD using a simple
optimization tool ‘Solver’. Solver is a Microsoft Excel based built in tool which optimizes the objective
function using either of the three techniques, i.e. GRG method (General Reduced Gradient techniques)
,Simplex LP method (Linear Programming) and GA (Genetic Algorithm). In the present work General
Reduced Gradient (GRG) technique has been used. Solver is run several times to obtain various optimized
values of the objective function and corresponding variables subject to different constraints and the results
have been presented in Tables, Graphs.
Optimal values of different parameters like ξd, ξ, H2, H4, ξ, ξe have been determined using the analytical
values of mass ratio !!!. For same values of mass ratio : optimal values of aforementioned parameters
procedure developed by curve fitting method as given in the literature “JJC Conc.Ch.4.Vol2” for different
have also been obtained in the present study using solver.
724
Table 4.1 Table 4.2
ξd (H2)opt (H2)opt (ξe)opt (ξe)opt (H4)opt (H4)opt (H4/H2)opt (H4/H2)opt

Ref present Ref present ξd Ref present Ref present
0.6 45.189 39.742 0.0110 0.012 0.6 1081.88 38.97 23.941 0.98
0.65 30.863 25.679 0.0162 0.019 0.65 519.28 25.017 16.825 0.97
0.7 23.574 18.624 0.021 0.026 0.7 311.86 18.06 13.228 0.96
0.75 19.094 14.371 0.0261 0.034 0.75 210.73 13.91 11.036 0.968
0.8 15.997 11.501 0.0312 0.043 0.8 152.59 11.15 9.53 0.969
0.85 13.666 9.403 0.0365 0.053 0.85 115.14 9.166 8.42 0.974
0.9 11.774 7.7640 0.0424 0.064 0.9 88.75 7.65 7.53 0.985
0.95 10.2 6.459 0.049 0.077 0.95 69.49 6.49 6.81 1.005
Table 4.1 and Table 4.2 show comparison between referred and present optimal values of H2and ξe and
between H4and H4/H2 respectively. In Table 4.1 it has been observed that the optimum value of ground
motion amplitude H2 obtained in present study is much less than that obtained from referred analytical
procedure. At the same time optimum value of effective damping of the structure-TMD system obtained
different values of ξd and : . This is quite a remarkable improvement in the optimum performance of
in the present study is appreciably greater than that obtained through referred analytical procedure for
TMD observed in present study. Table 4.2 shows optimum values of maximum dynamic amplification
analytical procedure for different values of ξd and :. Simultaneously optimum value of relative
factor for TMD is largely reduced in the present study as compared to that obtained through referred
displacement ratio of TMD is lying between 0.96 to 1.005. This is quite a remarkable improvement
which will solve the problem of accommodating large relative displacement of TMD.
50 0.1
40 0.08
30 m‾ vs H2 0.06 ξe vs m‾ ref
present
20 0.04 ξe vs m‾
m‾ vs H2
ref present
10 0.02
0 0
0 0.02 0.04 0 0.01 0.02 0.03
Figure3 H2 versus : curves Figure 4 : versus ξe curves
725
1200 30
1000 25
H4/H2 ref
800 m bar ref 20 vs m bar ref
vs H4 ref
600 15
m bar H4/H2
400 10
present vs present vs
200 H4 present 5 m bar
present
0 0
0 0.02 0.04 0 0.01 0.02 0.03
".
Figure 5 H4 versus : curves Figure 6 : versus W" curves
&
1
1
0.5
0.5
̅ vs ξd present
̅ vs ξd ref.
0
0
0 0.005 0.01 0.015 0.02 0.025
0 0.005 0.01 0.015 0.02 0.025
Figure 7 : versus ξd ref.curve Figure 8 : versus ξd present curve
Figure 3 shows a non-linear relationship between : and H2 for both the cases. It is important to note that
ground motion amplitude H2 decreases sharply with increase in : up to a value of : 0.005 after which
appreciably less as compared to that obtained from referred analytical method for all values of : and the
the rate of decrease is very less. Moreover ground motion amplitude as obtained in the present study is
difference is more pronounced after the critical value of : 0.005 . Figure 4 shows that optimal value of
effective damping of the system increases with increase in value of : . the optimum value of effective
difference being more pronounced with increase in the value of : . Figure 5 and 6 shows much reduced
damping as obtained in the present study is more than that obtained in referred analytical method with the
referred literature .It must be noted that this ratio is remaining constant for all values of : in the present
value of relative displacement ratio obtained in the present study as compared to that obtained from
study as compared to non-linear variation obtained from referred analytical method. This indicates quite a
ξd with : both in referred literature and present study where ξd increases steadily with :.
remarkable improvement in the optimal performance of TMD. Figure 7 and 8 show identical variation of
5. CONCLUSION
In the present study an attempt has been made for assessment of optimal performance of a damped Tuned
Mass Damper attached to an undamped SDOF structure using a simple optimization tool, Solver. This has
been compared with the optimal performance obtained by a curve fitting method as reported in an
established literature (JJC Conc. vol.2, chapter 4). Following conclusions have been arrived at in the
present study.
726
1. For all values of mass ratio optimal values of ground motion amplitude obtained in present study is
much less than that obtained in referred literature.
2. The output , i.e. optimal value of effective damping ratio of the total system is much better in present
study than that obtained in referred literature with the difference increasing with increasing mass
ratio. Hence the policy planner can consider the limiting value of mass ratio depending on the
corresponding relative displacement that can be accommodated before deciding on the effective
damping of the total system to be achieved.
3. For an undamped SDOF structure attached with a damped TMD the optimal value of ground motion
amplitude in present study is 39.742 whereas the same optimal value as reported in referred literature
is 45.189 which implies a 12% improvement achieved in the present study. This is for 60% damping
of TMD.
4. For same 60% damping of TMD effective damping in present study is 0.0126 as compared to
referred value 0.011 which implies an improvement of 14.5% in effective damping in present study.
Similar quite remarkable improvements have been observed in the present study for other damping
values of TMD.
REFERENCES
[1] JJC Conc.Ch04 vol2 217-250.
[2] Kelly, J. M. (1996), Earthquake-Resistant Design With Rubber, 2nd ed., Springer-Verlag, London.
[3] Pan, T. C., and Kelly, J.M. (1983), “Seismic Response of Torsionally Coupled Base-Isolated
Structures,” Earthquake Engineering and Structural Dynamics, Vol. 11, pp 749-770.
[4] Mostaghel, N., and Khodaverdian, M. (1988), “Seismic Response of Structures Supported on R- FBI
System,” Earthquake Engineering and Structural Dynamics, Vol. 16, pp 839-854.
[5] Chopra, A. K. (2001), Dynamics of Structures: Theory and Applications to Earthquake Engineering,
2nd ed., Prentice Hall, Upper Saddle River, New Jersey.
[6] Chang, C. H., and Soong T T (1980). "Structural control using active tuned mass dampers."
Engineering Mechanics Division., ASCE, 106(6), pp1091-1098.
[7] Skinner R.I, Robinson W.H and McVerry G.H (1993) “An Introduction to Seismic Isolation”, John
Wiley and Sons.
[8] Clough R.W and Penzien J (1975) Dynamics of Structure, McGraw-Hill, USA.
[9] Naeim F., Kelly J.M (1999) “Design of Seismic Isolated Structures: From Theory to Practice”, John
Wiley and Sons. Inc.
[10] Komodromos P.(2000) “Seismic Isolation for Earthquake Resistant Structures” , WIT Press,
Sothhampton, Boston.
[11] Liang Z, Lee G.C, Dargush G.F , Song J. (2012) “Structural Damping” , CRC Press, Taylor And
Francis Group, New York.
727
Paper No.: 134
ANALYSIS OF THICK BEAMS USING FIRST ORDER SHEAR

DEFORMATION THEORY
1 2
Sourabh S. Deshpande and R. L. Wankhade
1
PG Student, Dept. of Applied Mechanics, Government College of Engineering Karad, Maharashtra, India
2
Asst. Professor, Dept. of Applied Mechanics, Government College of Engineering Karad, Maharashtra, India
Email: sourabhdeshpande08@gmail.com, rajanw04@gmail.com
ABSTRACT :
A first order shear deformation beam theory is employed here for the static analysis of thick beams. The
limitations of classical theory of beam bending developed by Euler and Bernoulli forced for the refinement over
these classical theories. After refinement discrepancies of the classical beam theory are eliminated and hence
first order shear deformation theory is developed. The assumption of Euler-Bernoulli beam theory stating that
the plane section remains perpendicular to the neutral axis of the section after the deformation is modified by
Timoshenko. Timoshenko beam theory assumes that the section which was normal to the neutral axis before
deformation does not remain normal to the neutral axis after the deformation. He considers the rotation of the
normal to be the combined effect of the rotation due to bending and shear effects. But, Timoshenko theory does
not remove another discrepancy of the classical theory i.e. normal section remains plane after bending.
Timoshenko also assumes that plane section before bending remains plane after bending also. Thus, He assumes
the constant shear strain across the section and thus rectangular distribution of the shear stress across the section,
which is not in accordance with the actual distribution of the shear stress which is parabolic across the section.
Thus, shear correction factor is required in this theory to modify the result to get the fairly accurate result.
KEYWORDS: Shear deformation theory, Thick beams, Shear stress
1. INTRODUCTION
The elementary theory of beam bending introduced by Euler and Bernoulli is the simplest theory of beam
bending which neglects the effects of the shear deformation. The theory assumes that the plane section which is
normal to the neutral axis before bending remains plane and normal to the neutral axis even after deformation of
the beam. That means shear stresses are not considered in the hypothesis.
Effect of rotary inertia and shear deformation are included in the theory of beam by Rayleigh (1877) and
Timoshenko (1921) as the pioneer investigators. To remove the discrepancies in the classical and first order
shear deformation or Timoshenko theories, higher order shear deformation theories were developed by various
researchers.
Levinson (1981), Bickford (1982), Rehfield and Murty (1982) Krishna Murty (1984), Bhimaraddi and
Chandrashekhara (1993) are some of the investigators who presented parabolic shear deformation theories
assuming a higher variation of the axial displacement in terms of the thickness coordinate for the analysis of
thick beams. Finite element method is used by Kant and Gupta (1988), Heyliger and Reddy (1988) based on
higher order shear deformation of uniform rectangular beams.
Further refined shear deformation theories for thick beams including sinusoidal functions in terms of the
thickness coordinate in displacement field are developed by Vlasov and Leont’ev (1996), Stein (1989). The
drawback of these theories was that shear stress free boundary conditions were not satisfied at top and bottom
surfaces of the beam. Thus, a refined trigonometric shear deformation theory has been developed by Ghugal and
Shimpi (2001) to accurately predict shear stress distribution across the section of the beam and which also
satisfies the free boundary conditions at top and bottom surfaces of the beam.
In this paper, first order shear deformation theory is developed and various non dimensional parameters are
introduced. These parameters are applied to numerical examples to give values of these parameters for the
specific cases and then these values are plotted on the graph to show the variation of these parameters across the
section of the beam.
2. EQUATIONS AND DEVELOPEMENT OF THEORY-
The beam under consideration is shown in figure. It occupies in 0 − x − y − z Cartesian coordinate system in
the region;
0 ≤ x ≤ L; 0 ≤ y ≤ b; −h ≤ z≤h
2 2
Where x, y, z are Cartesian coordinates, L is the length of the beam in x direction, b is the width of the beam in y
direction and h is the thickness of the beam in z direction. The beam is made up of homogeneous, linearly
elastic isotropic material.
2.1 The displacement field

The displacement field of the present beam theory is given by;
u ( x , z ) = − zφ (1)
w( x, z ) = w(x)
Where u is the axial displacement in x direction and w is the transverse displacement in z direction of the beam.
The function φ represents total rotation of the beam at neutral axis, which is an unknown function to be
determined. The normal and shear strains obtained within the framework of linear theory of elasticity using
displacement field given by above equation is as follows-
∂u dφ
Normal strain: ε x = = −z (2)
∂x dx
∂u ∂w ∂w
Shear strain: γ zx = + = −φ + (3)
∂z ∂x ∂x
729
The stress-strain relationship used is as follows-
σx = εx ⋅E τ zx = γ zx ⋅ G (4)
The above relationships are used to obtain the governing differential equations of the present beam theory by
substituting these equations in the principle of virtual work. The procedure is explained in the next section.
2.2 Governing equations and boundary conditions
Using the expressions for strains and stresses (2) through (4) and using the principle of virtual work,
variationally consistent governing differential equations and boundary conditions for the beam under
consideration can be obtained. The principle of virtual work, when applied to the beam leads to;
x = L y =b z = 2
h
x=L
∫ ∫ ∫h (σ xδε x + τ zxδγ zx )dxdydz − ∫ q(x )δwdx = 0

x =0 y =0 z =− x =0
(5)
2
Where, symbol δ denotes the variational operator. Substituting the values of stresses and strains in the above
equation, we get the coupled Euler-Lagrange equations which are the governing differential equations and
associated boundary conditions of the beam. The governing differential equations are obtained as under-
x = L y =b z = 2
h
x=L
∫ ∫ ∫h (Eε xδε x + Gγ zxδγ zx )dxdydz = ∫ qδwdx

x =0 y =0 z = − x=0
2
Putting the values of stresses and strains from the above equations and then integrating and equating the
coefficients of δw and δφ , we get governing differential equations of this theory as follows-
 dφ d 2 w 
GAK s  − 2  = q( x )
 dx dx  (A)
 dw  d 2φ
GAK s  − φ  + EI 2 = 0
 dx  dx (B)
2.3 The general solution of governing equilibrium equations of the beam-
Using above obtained differential equations, we can obtain the general equation of this beam theory in terms of
rotation of the beam.
From equation (B),
d 2φ  dw 
EI 2
− GAK s  φ − =0
dx  dx 
Differentiate w. r. t. x, we get,
d 3φ  dφ d 2 w 
EI 3 − GAK s  − 2  = 0
dx  dx dx 
∴ From equation (A),
730
d 3φ
∴ EI = q(x )
dx 3
This is the general equation of this beam theory in terms of the rotation of the beam at the neutral axis. Using
this equation, we can obtain the expression for transverse displacement of the beam by substituting the value of
rotation of the beam in either of the equations (A) or (B).
3. ILLUSTRATIVE EXAMPLES AND DIAGRAMS-
In order to apply above theory to various boundary conditions, we have considered three examples here. With
the help of these examples, we can predict the distribution of various non dimensional parameters across the
section of the beam. Material properties of the beam are as follows-
E = 210GPa µ = 0.25 ρ = 7800 kg

m3
Where, E =modulus of elasticity of the material, µ =Poisson’s ratio for the beam material and ρ =density of
the beam material
Example 1-simply supported beam with uniformly distributed load over the entire span-
d 3φ
EI 3 = q ( x )
dx
By taking integration of above equation thrice and substituting appropriate boundary conditions in the equations;
we get the expression for the rotation of the beam at the neutral axis. It is given as-
qx 3 qLx 2 qL3
∴ EIφ = − +
6 4 24
This expression gives the rotation of the simply supported beam carrying uniformly distributed load over the
entire span at any section of the beam along its length. Now, this expression is used to find out the expression
for the transverse displacement of the beam at any section along the length of the beam.
From equation (A),

d 2 w dφ q
2
= −
dx dx GAK s
731
Putting the value of φ in the above equation and then integrating twice with respect to x by putting appropriate
boundary conditions in the equations, we get the expression for the transverse displacement of the beam at
neutral axis. It is given by-
q  x 4 Lx 3  qx 2 qL3 x qLx
w=  − 
 − + +
2 EI  12 6  2GAK s 24 EI 2GAK s
After modification, the final expression for the transverse displacement is given by,
qL4  x4 x3 x  qL2  x  x
w=  4 − 2 3 +  +  1 − 
24 EI L L L  2GAK s  L  L
For the maximum transverse displacement, put x = L in above expression, we get,

2
5 qL4  384 EI 
wmax = 1 + 
384 EI  40 L2 GAK s 
= 2(1 + µ ) and simplify the equation, we get the final expression for maximum transverse
E
Put
G
displacement as,
5 qL4   h  
2
wmax = 1 + 1.6(1 + µ ) 
384 EI   L  
Non dimensional transverse displacement is given by,
w h 4 Eb
w=
h L4 q
2
h
∴ w = 1 + 1.6(1 + µ ) 
L
From the above values of transverse displacement and rotation of the beam at the neutral axis, we can find out
the remaining values of axial stress in x-x direction, shear stress in x-z plane, stress in z-z direction and axial
displacement u in the form of non dimensional parameters. These can be found out as follows-
Expression for axial displacement u-
3
zq  L   x3 x2 
u=−    4 3 − 6 2 + 1
2 Eb  h   L L 
732
Expression for axial stress sigma xx-
2
q  z  L   x2 x 
σ x = −6     2 − 
b  h  h  L L
Expression for transverse shear stress using constitutive relationship τ zxCR

It may be noted that it is possible to obtain transverse shear stress τ zx , by using either constitutive relationship or
the equilibrium equation of theory of elasticity. Notion τ zxCR denotes τ zx obtained by using constitutive
relationship.
qL  x 1 
τ zx =  − 
bh  L 2 
Expression for transverse shear stress τ zxEE , and transverse normal stress, σ z , obtained from equilibrium
equation-
The following equilibrium equations of two dimensional elasticity ignoring body forces are used to obtain
transverse shear and transverse normal stresses.
∂σ x ∂τ zx
+ =0
∂x ∂z
∂τ xz ∂σ z
+ =0
∂x ∂z
To get τ zx , substitute the expression obtained for axial stress σ x in equilibrium Equation above and integrate
with respect to thickness coordinate z and impose the following boundary condition at bottom of beam
[τ zx ]z =± h / 2 = 0
To get constant of integration. The expression obtained for transverse shear stress using this procedure in its
final form is as follows:
3 q  L  x 1  z2 
τ zx = −   −   1 − 4 
2 b  h  L 2  h 2 
To get σ z , substitute the expression obtained for shear stress, in the equilibrium Equation above and integrate it
with respect to thickness coordinate z and impose the following boundary condition at bottom of the beam
[σ z ]z =h / 2 = 0
To get the constant of integration. The expression obtained for transverse normal stress σ z is as follows:
q1 3 z z3 
σ zz =  − + 2 3 
b2 2 h h 
733
Example 2-simply supported beam carrying uniformly varying load over entire span-
In this beam, the loading is expressed in terms of maximum intensity of loading q 0 and distance of
section along the length of beam i.e. x . The expression is given by,
d 3φ  x
EI 3
= q 0 1 − 
dx  L
After integrating w. r. t. x thrice by applying appropriate boundary conditions to find constants of

integrations, we get the expression for rotation of this beam at the neutral axis. This expression is given by,
q 0 L3  x4 x3 x2 3 
φ=  − 4 + 3 − 2 + 
6 EI  4L L L 17 
This expression is used to find the expression for transverse displacement. From equation (A), we can
write,
d 2 w dφ q0  x
2
= − 1 − 
dx dx GAK s  L 
Putting the value of φ in above equation and then integrating twice w. r. t. x by applying appropriate
boundary conditions to find out the constants of integration, we get the expression for transverse displacement
of beam at any section along the length of the beam. It is given by,
q0 L4  x5 x4 x3 3 x  q0 L2  x 2 x3 x
w=  − 5
+ 4
− 3
+ 
 − 
 2
− 3
− 0.42 
6 EI  20L 4 L 3L 17 L  GAK s  2 L 6 L L
Non dimensional parameters-
Example 1-
σ xb τ zx b σ zz b
σx = τ zx = σ zz =
q q q
734
Example 2-
σ x bL τ zx bL σ zz bL
P P P
Example 3-
σ xb τ zx b σ zz b
q0 q0 q0
Fig 1-variation of axial stress ( σ x ) through the Fig 2-variation of shear stress τ zx through the
thickness of simply supported beam thickness of the simply supported beam
(At x = L
2
, z ) when subjected to uniformly (x = 0, z ) , when subjected to uniformly distributed
distributed load for aspect ratio 4 and 10. load for aspect ratio 4 and 10, by using equilibrium
equation from 2-D elasticity problem
Fig 3-variation of shear stress τ zx through the

Fig 4-variation of transverse stress σ zz through
thickness of the simply supported beam
(x = 0, z ) , when subjected to uniformly
the thickness of simply supported beam when
subjected to uniformly distributed load
distributed load for aspect ratio 4, by using 1-D
constitutive law of elasticity
735
CONCLUSION-
1. Analysis of thick beams is carried out using first order shear deformation theory. Simply supported beam
carrying uniformly distributed load, carrying point load at the centre and a beam carrying uniformly varying
load are solved using the present theory.
2. The results of axial stress for the beam under given loading, obtained by the present theory are more
accurate than the Euler and Bernoulli theory.
3. The transverse shear stress obtained from constitutive relation and equilibrium using present theory gives
less error than the Euler and Bernoulli theory.
4. The governing differential equations and the associated boundary conditions presented are variationally
consistent.
REFERENCES -
[1] Lord Rayleigh, J.W.S. (1877): The Theory of Sound , Macmillan Publishers, London.
[2] Levinson. M. (1981): A new rectangular beam theory. J. Sound and Vibration. 74(1), 81-87
[3] Bickford, W.B. (1982): A consistent higher order beam theory, In: Proceeding of Dev. In Theoretical And
applied Mechanics ,SETAM,11 137-150
[4] Rehfield, L.W. and Murthy, P.L.N (1982): Toward a new engineering Theory of Bending: fundamentals.
AIAA Journal. 20(5), 693-699
[5] Krishna Murthy, A.V. (1984): Towards a consistent beam theory, AIAA Journal. 22 (6),811-816
[6] Bhimaraddi, A., Chandrashekhara K. (1993): Observations on higher order beam Theory. ASCEJ.
Aerospace Engineering. 6(4), 408-413
[7] Heyliger, P.R, Reddy, J.N. (1988): A higher order beam finite element for bending and vibration problems.
J. Sound and Vibration. 126(2),309-326
[8] Vlasov, V. Z., Leont’ev, U.N. (1966) : Beams, Plates and Shells on Elastic Foundations Moskva, Chapter
1, 1-8. Translated form the Russian by A. Barouch , and T. Plez , Iseral Program for Scientific Translation
Ltd., Jerusalem
[9] Stein, M. (1989): Vibration of beams and plate strips with three dimensional flexibility. ASME J. App.
Mech. 56(1), 228-231
[10] Ghugal, Y.M., Shmipi, R. P. (2001): A review of refined shear deformation theories for isotropic and
anisotropic laminated beams. Journal of Reinforced Plastics And Composites,20(3),255-272
[11] Timoshenko, S. P. (1921): On the Correction for Shear of the Differential Equation for Transverse
Vibrations of Prismatic Bars. Philosophical Magazine, Series 6, Vol. 41, 742-746
[12] Timoshenko, S. P., Goodier, J. N. (1970): Theory of Elasticity, McGraw-Hill, 3rd Int. ed., Singapore.
736
Paper No.: 135
FREE VIBRATION OF MULTIPLE-STEPPED BERNOULLI-EULER BEAM

BY THE SPECTRAL ELEMENT METHOD
1 2
Jitendra Kumar Meher and Manoranjan Barik
1
M. Tech. Student, Dept. of Civil Engineering, National Institute of Technology, Rourkela, India
2
Associate Professor, Dept. of Civil Engineering, National Institute of Technology, Rourkela, India
Email: 212ce2044@nitrkl.ac.in, mrbarik@nitrkl.ac.in
ABSTRACT :
The free vibration analyses of multiple-stepped Bernoulli-Euler beam with various boundary conditions have
been studied by many researchers using different methods of analysis such as Differential Quadrature Element
Method (DQEM), Composite Element Method (CEM), Admonian Decomposition Method (ADM), Differential
Quadrature Method (DQM), Local adaptive Differential Quadrature Method (LaDQM), Discrete Singular
Convolution (DSC) algorithm etc. besides the conventional analytical methods and finite element methods. Here
we use the Spectral Element Method (SEM) which finds out even the higher modes of vibration frequencies
accurately with consideration of least number of degrees of freedom thus promising very high computational
efficiency.
KEYWORDS: Natural frequency, Bernoulli-Euler beam, stepped-beam, Spectral Element Method.
1. INTRODUCTION
The constraints of fabrication, assembly, stiffness, space etc. may lead structures to have multiple stepped beams.
The dynamic behavior of the multiple stepped Bernoulli-Euler beams have been widely studied and reported in
the literature by using different methods of analysis. Jang and Bert [1,2] used the Euler beam theory and
reported the exact solutions for the natural frequencies of the stepped beam for the combination of the classical
boundary conditions. Naguleswaran [3,4] proposed an analytical method for the vibration analysis of stepped
Euler-Bernoulli beam on classical and elastic end supports. As the analytical or exact methods are more
involved, it is preferred to have a numerical method for the solution. Though the finite element method (FEM) is
more versatile for the numerical method, there are few alternative methods which are of better advantage over
the FEM. Recently Differential Quadrature Element Method (DQEM) [5] has been used for the free vibration
analysis of multiple-stepped beams and extensive reviews of literature pertaining to the advantages and
disadvantages of various other methods employed for the stepped-beams vibration analysis have been reported.
Among the different methods reported are the Composite Element Method (CEM) [6], Admonian
Decomposition Method (ADM) [7], Differential Quadrature Method (DQM) [8], Local adaptive Differential
Quadrature Method (LaDQM) [9] and Discrete Singular Convolution (DSC) algorithm [10].
In the present study the free vibration of multiple-stepped Bernoulli-Euler beam with combination of different
classical boundary conditions has been studied using the Spectral Element Method (SEM), This SEM being
similar to the conventional Finite Element Method (FEM) is easy to implement. Further the efficacy of this
method can be realized as the number of elements considered is limited to only the number of change of cross
sections of the stepped beam. For a stepped beam with a single change of cross section, consideration of only
two number of elements suffices the analysis, thus highly reducing the number of degrees of freedom in
comparison to the other methods of analysis.
2. THEORETICAL FORMULATION
The formulation of the spectral element model for a stepped beam is fairly similar to the formulation of the
conventional finite element method. However the major difference is that in general the spectral element
formulation begins with the transformation of the governing partial differential equations of motion from the
time domain to the frequency domain by using the Discrete Fourier Transform (DFT). Here the time variable
disappears and the frequency becomes a parameter to transform the original time –domain partial differential
equations into the frequency-domain ordinary differential equations which are then solved exactly and the exact
wave solutions are used to derive frequency-dependent dynamic shape functions. The exact dynamic stiffness
matrix called the spectral element matrix is finally formulated by using the dynamic shape functions [11]. The
spectral element matrices are then assembled to form the global dynamic stiffness matrix and the boundary
conditions are applied similar to the conventional finite element method thus producing the reduced dynamic
stiffness matrix. The determinant of the reduced dynamic stiffness matrix which is a function of the natural
frequencies when equated to zero and solved gives rise to the required natural frequencies.
2.1. Spectral Element Matrix for Stepped Beam
The spectral element matrix for a stepped beam can be derived by considering one Bernoulli-Euler beam
element of the stepped beam which is of uniform cross section whose free flexural vibration is represented by
EIw′′′′ + ρAw && = 0 (1)
Where w( x, t ) = transverse displacement
E = Young’s Modulus
A = Area of cross-section
I = Moment of inertia
ρ = Mass density
M t ( x, t ) = EIw′′( x, t ) (2)
Qt ( x, t ) = − EIw′′′( x, t ) (3)
Where Qt ( x, t ) =Internal transverse shear force
M t ( x, t ) = Bending moment
Assuming the solution to Eq.1 in spectral form to be
N −1
1
w( x, t ) =
N
∑W
n =0
n ( x; ω n )e iωnt (4)
Substituting Eq.4 in to Eq.1 gives an Eigen-value problem for a specific discrete frequency (such as ω = ω n ) as
EIW ′′′′ − ω 2 ρAW = 0 (5)
Assuming general solution to equation as
W ( x ) = ae − ik (ω ) x (6)
Substituting Eq.4 in to Eq.5 yields a dispersion relation
k 4 − k F4 = 0 (7)
Where k F = wave-number for the pure bending (flexural) wave-mode defined by
1
 ρA  4
kF = ω   (8)
 EI 
Eq.7 gives two pure real roots and two pure imaginary roots as
k1 = − k 2 = k F , k 3 = − k 4 = ik F
(9)
For the finite B-beam element of length L, the general solution of Eq.5 can be obtained in the form of
W ( x; ω ) = a1 e − ik F x + a 2 e − k F x + a 3 e + ik F x + a 4 e + k F x = e( x; ω ) a (10)
738
Where
[
e( x; ω ) = e −ik F x e −kF x e + ik F x e + kF x ]
a = {a1 a4 }
T
a2 a3 (11)
The spectral nodal displacement and slopes of the finite B-beam element can be related to the displacement field
by
W1   W (0) 
 Θ  W ′(0) 
   
d =  1 =   (12)
W2   W ( L) 
Θ 2  W ′( L)
Substituting Eq.10 in to the right-hand side of Eq.12 gives
 e(0; ω ) 
 e′(0; ω ) 
d =  a = H B (ω )a (13)
 e( L; ω ) 
 
e′( L; ω )
 1 1 1 1 
 − ik − kF ik F k F 
Where H B (ω ) = 
F
−ik F L (14)
 e e −k F L e ik F L e kF L 
 − ik F L 
− ik F e − k F e −k F L ik F e ik F L k F e kF L 
The displacement field within the finite B-beam element can be represented in terms of the nodal DOFs vector
d by eliminating the constant vector a from Eq.10 by using Eq.13.
W ( x) = e( x; ω ) H B−1 d (15)
From the Eq. 2 and Eq.3, the spectral components of the bending moment and transverse shear force can be
related to W (x ) as
M ( x ) = EIW ′′( x ) , Q ( x ) = − EIW ′′′( x ) (17)
The spectral nodal bending moments and transverse shear forces defined for the finite B-beam element
correspond to the moments and the forces as given below (Fig. 1)
Figure 1 Bernoulli-Euler spectral beam element with nodal forces and displacements
 Q1   − Q(0) 
 M   − M (0) 
   
fc =  1  =   (18)
 Q2   + Q ( L ) 
M 2  + M ( L)
739
Substituting Eq. 15 into Eq. 17 and the results into the right-hand side of Eq. 18 gives
 EIW ′′′(0)   e′′′(0; w) 
 − EIW ′′(0)   − e′′(0; w) 
    −1
fc =   = EI  H B d (19)
− EIW ′′′( L) − e′′′( L; w)
 EIW ′′( L)   e′′( L; w) 
⇒ f c (ω ) = S B (ω )d
Where S B (ω ) is spectral element (dynamic stiffness) matrix for the B-beam element given by
 e′′′(0; w) 
 − e′′(0; w) 
  −1
S B (ω ) = EI  H B (20)
 − e ′′′( L; w) 
 e′′( L; w) 
2.2. Spectral Global Matrix for Stepped Beam
After spectral element matrix for each element is computed, they are assembled into the spectral global matrix
following the similar convention of the classical finite element method where the inter-element continuity
conditions are automatically maintained. The classical boundary conditions may be applied by eliminating the
corresponding rows and columns of those restrained degrees of freedom thus forming the reduced spectral
global matrix S g (ω ) and the eigenvalue problem is formed as
S g (ω )d g = 0 (21)
Where d g is the global spectral nodal degrees of freedoms vector.
The eigenfrequencies ω i (i=1,2, …..,∞) are determined by equating the determinant of S g (ω ) to zero at
ω = ω i ,i.e.,
S g (ω i ) = 0 (22)
The value of ω is found by following an iterative procedure with a very high degree of accuracy which makes
the value of S g (ω ) close to zero. For practical purpose the tolerance value of 10 −6 is sufficient, however our
solution procedure is able to sustain up to a tolerance limit of 10 −11 uninterruptedly. The solution procedure
followed is detailed below.
In this method we begin with a starting value of frequency ω1 and suitable values of an increment δ
and a tolerance ε are considered so that no frequency may be skipped. The value ‘ ω1 + δ ’ is assigned to
ω 2 and the values of S g (ω1 ) and S g (ω 2 ) are checked for whether they are of the opposite sign or not.
When they are of the same sign each value is incremented by δ otherwise S g (ω 0 ) is determined where
ω 0 = (ω1 + ω 2 ) 2 . In the first case if S g (ω1 ) > 0 & S g (ω 2 ) < 0 , then ω1 = ω 0 if S g (ω 0 ) > 0 or
ω 2 = ω 0 if S g (ω 0 ) ≤ 0 . In the second case if S g (ω1 ) < 0 & S g (ω 2 ) > 0 , then ω 2 = ω 0 if
S g (ω 0 ) > 0 or ω1 = ω 0 if S g (ω 0 ) ≤ 0 . This iterative procedure is repeated as long as (ω 2 − ω1 ) > δ and
740
the corresponding ω 0 value becomes the desired natural frequency. For next mode of natural frequency we
begin with a new ω1 value which is set at slightly greater value than that of previously found natural frequency
(for example 1.0005 times of ω 0 )and the process is repeated till the next mode of natural frequency is obtained
as per requirement.
4.1. Free-Free Beam with single step change
A free-free (F-F) beam with single step change in cross-section schematically shown in Fig.2 is considered. The
geometrical dimensions and material properties are [12] L1 =254 mm, L2 =140 mm, b=25.4 mm, h1 =19.05
mm , h2 =5.49 mm, E=71.7 GPa and ρ =2830 kg / m 3 . The first three numbers of natural frequencies are
computed by considering two numbers of spectral elements and the results are compared with those obtained by
other methods in Table 4.1. There is an excellent agreement of the SEM results with those of DQEM and FEM
methods.
Figure 2 Single Stepped Free-Free Beam
Table 4.1 First 3 natural frequencies (Hz) of the single stepped F-F beam
Mode Present DQEM [5] FEM[5] CEM[6] Experiment
No. Result [12]
(SEM)
1 292.44379 292.44 292.44 291.9 291
2 1181.31992 1181.30 1181.30 1176.2 1165
3 1804.08716 1804.10 1804.10 1795.7 1771
4.2. Simply Supported Beam with three step changes
A simply supported (SS-SS) beam with three step changes in cross-section, schematically shown in Fig.3 is
analysed. The geometrical dimensions and material properties are [7] L1 = L2 = L3 = L4 = 5 m, h1 = h3 = 0.1
m, h2 = h4 = 0.2 m, E=34 GPa, ρ =2830 kg / m 3 . The first ten numbers of natural frequencies are computed
by considering four numbers of spectral elements (one for each cross-section) and the results are compared with
those obtained by other methods in Table 4.2. There is an excellent agreement of the present SEM results with
those of DQEM and FEM.
Figure 3 Three-Stepped Simply Supported Beam
741
Table 4.2 Natural Frequencies (Hz) of three-stepped SS-SS beam
Mode Present Result DQEM [5] CEM[6] FEM [5]
No. (SEM)
1 0.43369 0.43369 0.433 0.43369
2 1.80276 1.80276 1.799 1.80276
3 4.41470 4.41470 4.411 4.41470
4 9.54133 9.54133 9.522 9.54133
5 13.26609 13.26609 13.246 13.26609
6 19.35885 19.35885 19.301 19.35885
7 25.76032 25.76032 25.721 25.76032
8 35.00419 35.00420 34.959 35.00420
9 43.21882 43.21882 43.174 43.21882
10 55.66242 55.66242 55.473 55.66242
4.3. Cantilever Beam with twelve step changes
A cantilever (C-F) beam with twelve step changes in cross-section, schematically shown in Fig.4, is considered.
The material properties are [13] E=60.6 GPa, ρ = 2664 kg / m 3 . The geometrical dimensions shown in Fig.4
are in mm. Rest of the dimensions are: b=3.175 mm, h1 = 12.7 mm, h2 =25.4 mm. Both “Type 1” beam
( I i = hi b 3 / 12 ) and “Type 2” beam ( I i = bhi3 / 12 ) are analysed. The SEM results compares well with DQEM
results.
Figure 4 Twelve Stepped Cantilever Beam(C-F)
Table 4.3.1 Comparison of natural frequency ( Hz) of 12 stepped C-F beam in Flap wise (Type I) bending mode
Mode Present Result DQEM[5] CEM Ritz[13] CMA[13] ANSYS[13] NASTRAN EXP[13]
No. (SEM) [5] [5]
1 10.74507 10.745 10.758 10.752 10.752 10.775 10.745 10.63
2 67.47321 67.473 67.553 67.429 67.463 67.469 67.473 66.75
Table 4.3.2 Comparison of Natural frequency(Hz) of 12 stepped C-F beam in chord wise(Type II)bending mode
Mode Present DQEM[5] CEM[]5 Ritz[13] CMA[13] ANSYS[13] NASTRAN EXP[13]
No. Result [5]
(SEM)
1 54.49652 54.497 54.699 54.795 54.985 54.469 54.497 49.497
4.4. Beam with twelve step changes having different boundary conditions
The beam with twelve step changes of the previous problem is considered for different boundary conditions
and first six natural frequencies obtained by SEM are tabulated in Table 4.3.3 (for Type I) and Table 4.3.4 (for
742
Type II). The boundary conditions are Simply Supported (SS), Clamped(CL), Sliding (SL), Free (FR).
Table 4.3.3 Natural frequency ( Hz) of 12 stepped beam with Flap wise (Type I) bending mode for different
boundary conditions by SEM
Mode SS-SS SS-CL SS-SL SS-FR SL-SS SL-CL SL-SL SL-FR

No.
1 30.33066 47.88263 7.58536 47.00734 7.57989 17.33849 30.35447 17.06272
2 121.50868 155.84275 68.34593 152.33466 68.28672 94.06037 121.62721 92.13502
3 274.14638 326.91131 190.29945 318.42167 190.08724 233.39781 274.50280 227.71211
4 489.35463 562.70885 374.42017 546.82404 373.84259 436.59905 490.27771 424.68969
5 768.38650 865.77188 622.34332 840.56000 620.85088 705.55500 770.94102 685.23496
6 1096.85753 1215.20944 936.63296 1179.71556 931.37324 1048.26917 1136.356 1015.23593
Mode CL-SS CL-CL CL-SL CL-FR FR-SS FR-CL FR-SL FR-FR

No.
1 47.57420 69.85864 17.20854 10.74507 47.23720 10.87556 17.19180 67.93730
2 154.91703 194.08228 93.56830 67.47321 152.95705 68.22281 92.63076 186.85817
3 325.19860 383.82375 232.36983 189.55922 319.34302 191.38826 228.79062 366.35623
4 560.21325 640.84719 435.24088 373.46128 547.53062 376.29277 426.31378 607.67800
5 862.10536 972.37651 704.67591 622.27380 839.81794 625.41163 687.32392 916.81307
6 1222.58119 1337.57776 1052.90263 942.35133 1184.49502 945.08048 1021.36866 1264.24096
Table 4.3.4 Natural frequency ( Hz) of 12 stepped beam with chord wise (Type II) bending mode for different
boundary conditions by SEM
Mode SS-SS SS-CL SS-SL SS-FR SL-SS SL-CL SL-SL SL-FR

No.
1 153.58118 247.36409 38.63037 238.09269 38.35893 89.23032 154.72654 86.35536
2 616.69449 812.52175 348.88449 773.98277 346.11003 488.00942 622.12294 467.26006
3 1396.17734 1724.14679 975.89760 1626.70779 966.35242 1223.58728 1412.01136 1159.79141
4 2496.33906 3001.43775 1932.19548 2811.96773 1906.72651 2316.76980 2536.80644 2176.99142
5 3859.09412 4559.55400 3220.99213 4307.14790 3156.57778 3767.92032 3962.27867 3528.45047
6 4930.95355 6754.77451 4672.16686 5361.80343 4516.71696 6649.31887 7905.37733 5182.77175
Mode CL-SS CL-CL CL-SL CL-FR FR-SS FR-CL FR-SL FR-FR

No.
1 242.65662 364.69000 87.88448 54.49652 239.20688 55.84962 87.70718 344.14366
2 796.87005 1030.21859 482.83080 344.80793 776.54087 353.04653 473.02081 950.04271
3 1689.37625 2078.06938 1212.73063 977.81252 1628.04218 999.81690 1174.53901 1875.87170
4 2931.11694 3541.19406 2302.82271 1951.40933 2801.91432 1991.85638 2206.41607 3147.51455
5 4397.17392 6476.86031 3760.45543 3301.63914 4240.59215 3367.48744 3586.47993 4900.07519
6 7280.80219 7571.29326 7102.96592 5105.41746 5685.56035 5474.28624 5577.46452 5814.99309
743
REFERENCES
[1] Jang, S.K. and Bert, C.W. (1989). Free vibration of stepped beams: exact and numerical solutions. Journal of
Sound and Vibration , 130(2) 342-346.
[2] Jang, S.K. and Bert, C.W. (1989b). Free vibration of stepped beams: higher mode frequencies and effects of
steps on frequency. Journal of Sound and Vibration , 132(1)164-168.
[3] Naguleswaran, S. (2002). Natural frequencies, sensitivity and mode shape details of an Euler-Bernoulli beam
with one step change in cross-section and with ends on classical supports. Journal of Sound and Vibration ,
252(4), 751-767.
[4] Naguleswaran, S. (2002). Vibration of an Euler-Bernoulli beam on elastic end supports and with up to
three step changes in cross-section. International Journal of Mechanical Sciences, 44, 2541-2555.
[5] Wang, X. and Wang Y. (2013). Free vibration analysis of multiple-stepped beams by the differential
quadrature element method. Applied Mathematics and Computation, 219, 5802-5810.
[6] Lu, Z.R., Huang, M., Liu, J.K. et al. (2009). Vibration analysis of multiple-stepped beams with the
composite element model. Journal of Sound and Vibration, 322, 1070-1080.
[7] Mao, Q. (2011). Free vibration analysis of multiple-stepped beams by using Admonian decomposition
method. Mathematical and Computer Modeling , 54,756-764.
[8] Bert, C.W. and Malik, M. (1996). Differential quadrature method in computational mechanics: a review.
Applied Mechanics Reviews, 49, 1-28.
[9] Wang, Y., Zhao, Y.B. and Wei, G.W. (2003). Static analysis of frame structures by differential quadrature
element method. International Journal for Numerical Methods in Engineering, 40, 759-772.
[10] Wei, G.W. (2001). A new algorithm for solving some mechanical problems. Computer Method in Applied
Mechanics and Engineering, 190, 2017-2030.
[11] Lee, U. (2009). Spectral Element Method in Structural Dynamics. John Wiley & Sons (Asia) Pte Ltd, 2
Clementi Loop, Singapore.
[12] Koplow, M.A., Bhattacharyya, A. and Mann, B.P. (2006). Closed form solutions for the dynamic response
of Euler-Bernoulli beams with step changes in cross section. Journal of Sound and Vibration, 295, 214-225.
[13] Jaworski, J.W. and Dowell, E.H. (2008). Free vibration of a cantilevered beam with multiple-steps:
comparison of several theoretical methods with experiment. Journal of Sound and Vibration,312,713-725.
744
Paper No.: 136
ATTENUATION OF SEISMIC RESPONSE OF STRUCTURES USING PASSIVE

DEVICES
Istiyak Khan1, M. K. Agrawal2, S. B. Chafle1

1
Research Reactor Design and Projects Division, 2Reactor Safety Division, Bhabha Atomic Research Centre,
Mumbai-400085, India
Email: ikhan@barc.gov.in, mkagra@barc.gov.in
ABSTRACT:
Biggest challenge for structural engineering community is to improve the performance of new as well as older
structures to enhance their ability to withstand earthquake loads and prevent damages. Seismic response of
structure can be reduced by making use of response reduction techniques. These techniques involve the use of
passive devices. These devices work on different principle either by tuning the frequency through TMD or
providing the base isolation with soft spring at bottom and X-plate damper based on the mechanism of inelastic
energy dissipation. The seismic effectiveness of these passive devices for structures depends mainly on the
dynamic characteristic of structure, device parameters, and characteristics of seismic excitations. The focus of the
paper is to present the sensitivity of different parameters of some of the passive devices and effectiveness for
response reduction subjected to seismic excitation. The devices considered are tuned mass damper (TMD),
base-isolator (BI), and X-plate damper (XPD). Several sets of numerical results for different structural
configurations are presented to demonstrate the effectiveness of these devices under seismic excitation.
KEYWORDS: Seismic effectiveness, Passive devices, Tuned mass damper, Base isolator, X-plate damper
1. INTRODUCTION
Passive devices such as tuned mass damper (TMD), seismic base-isolator (BI) and X-plate damper are widely
used for reducing the seismic demand of structures. Performance of these devices is highly dependent on device
parameters, structural parameters for which response to be reduced and type of excitation. Many studies have been
carried out for assessing the performance of these devices under seismic excitation. A tuned mass damper is the
passive device which reduces the undesirable vibrations by providing control force to the structure and works on
the principle of vibration absorber. The principle of operation and some of the application of tuned mass damper
for seismic response reduction have been discussed in [1-6].
In the recent years, several studies have been carried out in order to understand the performance of base-isolation
devices and systems. Many studies on various base isolation systems, including laminated rubber bearing,
frictional bearing and roller bearing have been conducted to assess the seismic effectiveness [7-12]. The X-plate
damper is an energy dissipating device, consists of an assembly that holds either single or multiple components of
thin metallic or layered plates of ‘X’ shape. For piping systems, these devices satisfy the basic requirement of
thermal expansion without generation of undesired stresses. During strong earthquakes, these devices dissipate
most of the earthquake input energy through the inelastic deformation and thereby reduce the forces transferred to
the piping system. Bakre et al. [13] and Parulekar et al. [14] presented numerical study to investigate the seismic
effectiveness of an X-plate damper for piping systems in industrial units (e.g. chemical and petrochemical
industries). The effect of X-plate damper on the dynamic characteristic of the piping system is also presented [15].
In this paper, sensitivity of different parameters of some of the passive devices and effectiveness for seismic
response reduction has been presented. The devices considered are tuned mass damper (TMD), base-isolator (BI),
and X-plate damper (XPD). Several sets of numerical results for different structural configurations are presented
to demonstrate the effectiveness of these devices under seismic excitation.
2. DESCRIPTION OF STRUCTURES
Three different types of structural configurations have been considered in the present study for the passive devices.
The description of main structure along with passive device is as follows;
2.1. Cantilever beam with tuned mass damper (TMD)
A tuned mass damper (TMD) is essentially a secondary mass, attached to an original system via a spring and
damper in order to reduce the undesired vibration. The natural frequency of the mass damper is tuned to a
frequency near the fundamental frequency of the main system. The structure is modeled as cantilever beam
attached with tuned mass damper at tip as shown in Fig.1 (a). The corresponding finite element model is also
shown in Fig.1 (b). The TMD is characterized by its damping ratio (ξS), frequency ratio (γ) which is defined as the
ratio of the fundamental frequency of the TMD (ωs) to that of the fundamental frequency of primary structure (ωp).
The mass ratio (Mr) is defined as the ratio of mass of the TMD to that of the structural mass.
(a) (b)
TMD system
Figure 1 (a) Structural model, (b) Finite element model of cantilever beam with TMD
2.2. Frame structure with base isolator (BI)
In the seismic base isolation, the effect of the ground acceleration is reduced by interposing structural elements
with low horizontal stiffness between the structure and the foundation. This provides period lengthening (detuning)
and energy-dissipating mechanism to the structures. Fig.2 depicts the 3 D frame structure attached with flexible
member at base. The dimension of columns and beams of frame structure along with material properties are
shown in Table-1.
2m
2m Table 1 Dimensions of beams and columns
2m
Width of beam and column 0.12 m
Height of beam and column 0.12 m
Mass density 2500 kg/m3
Poisson ratio 0.20
Modulus of elasticity 2.236E10 N/m2
Structural damping 5.0%
1m
Base isolator
Figure 2 Three dimensional frame structure with base isolator (BI)
746
2.3. Piping with X-plate damper (XPD)
A simplified piping system with 3 inch NB sch. 40 pipe made of carbon steel (A53 grade A), having Young’s
modulus ( E ) of 2.06E11 N/m2 and Poisson’s ratio 0.3 is considered. Fig.3 shows a schematic and FE model of the
piping system with X-plate damper (XPD). XPD is connected to the point specified in Fig.3 (b) of piping with line
of action in Z-direction. The piping system is modeled using 3-D Beam element, and nonlinear truss element is
used for XPD. The XPD is attached at 1.9 m from the fixed base A. the piping system is supported at 1.5 m from
fixed base by dead weight support with mass of 150kg attached to the points near these supports.
(a) (b)
Figure 3 Piping with X-plate damper (a) piping model, (b) FE model of piping with X-plate damper
3. NUMERICAL STUDY
Three different structural models attached with specific passive device have been considered in the present study.
The finite element models of these structures are shown in Figs. 1, 2 & 3 respectively. To demonstrate the
effectiveness of these devices subjected to seismic excitation, seismic response of structure is investigated under
different ground motions. The earthquake ground motions used in present work are 1940 Imperial Valley
earthquake (Th-1), and 1994 Northridge earthquake (Th-2). The peak ground acceleration (PGA) of Imperial
Valley earthquake and Northridge earthquake ground motions are 0.34g and 0.84g respectively. These typical
earthquake time histories are shown in Fig.4 (a) & (b) respectively.
4 10
(a) (b) Northridge Earthquake (Th-2)
Imperial Valley Earthquake (Th-1) 8
2 6
Accel.(m/s2)
4
Accel.(m/s2)
0 2
0
-2
-2
-4
-6
-4
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
Time(sec.) Time(sec.)
Figure 4 Earthquake time histories (a) Th-1, (b) Th-2
3.1. Tuned Mass Damper (TMD)
The performance of tuned mass damper mainly depends on its parameters and the dominant excitation frequency
ranges, duration and amplitude of earthquake excitation. Therefore, it is necessary to optimize the TMD when
used for multiple frequency excitations such as an earthquake. In the presence of damping in the main structure,
no closed-form expressions can be derived for the optimum damper parameters. However, they may be obtained
by numerical trials with the aim of achieving a system with smallest possible value of its higher response peak.
The methodology for optimization of tuned mass damper for damped structure given in Reference [4] has been
adopted in this paper. The approach for determining the optimum parameters of tuned mass damper for damped
structure is based on suppression of resonance peak of structure and to achieve optimized response under
harmonic excitation. The optimum parameters of tuned mass damper for structure depicted in Fig.1 for 1.0% mass
ratio are shown in Table 2.
747
Table 2 Properties of the optimum TMD system with mass ratio 1.0%
TMD No. γopt ξSopt Mr (%)
1 1.016 0.065 1.0
3.1.1 Seismic effectiveness of tuned mass damper
The effect of detuning of tuned mass damper on structural responses is demonstrated in Fig.5 subjected to seismic
excitation. Fig.5 (a) and Fig.5 (b) show the sensitivity of normalized amplification at the tip of cantilever beam
with respect to frequency ratio and for different mass ratios of TMD subjected to Imperial Valley Earthquake
(Th-1), and Northridge Earthquake (Th-2). Reduction in response of structure is observed as TMD tunes to the
structure. The normalized amplification is defined as the ratio of response of structure with TMD to response of
structure without TMD.
1.8 1.8
(a) Th-1 (b) Th-2
1.6 1.6
1.4 1.4
Normalized Amplification
Normalized Amplification
1.2 1.2
1.0 1.0
0.8 0.8
0.6 0.6
0.4 Mr=0.5 %
0.4
Mr=1.0 %
0.2 0.2
Mr=2.0 %
0.0 0.0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
γ γ
Figure 5 Effect of frequency ratio on normalized amplification for various mass ratios with 2% damping of
structure subjected to ground motions
The maximum attenuation is observed when the frequency of tuned mass damper is such that it tunes with primary
structure. Fig. 5(b) depicts the clear picture of tuning of TMD with primary structure. The effect of mass ratio of
tuned mass damper is also demonstrated. Large attenuation of seismic response is observed with the increases of
mass ratio of tuned mass damper. The time variation of responses of structure under different earthquake loading
is shown in Fig.6 with optimized tuned mass damper given in Table-2. The results obtained show that TMD is
effective in reducing the response of structure subjected to various earthquake loadings. It is noted that the
reduction in response is not uniform over the whole duration of ground motion. However, performance is largely
depends on the dominant frequency ranges, duration and amplitude of earthquake which affect the structural
absolute acceleration and relative displacement responses. It is noted that around 20-30% reduction in the
response of structure is obtained in dominant zone of earthquake.
50
15 Without TMD Without TMD
Absolute tip acceleration (m/s )
Absolute tip acceleration (m/s )
(a) Th-1 (b) Th-2

2
With optimized TMD 40 With optimized TMD

10 30
20
5
10
0 0
-5 -10
-20
-10
-30
-15 -40
0.04
0.1
0.03
Tip displacement (m)
Tip displacement (m)
0.02
0.01
0.00 0.0
-0.01
-0.02
-0.03
-0.1
-0.04
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
Time (sec.) Time (sec.)
Figure 6 Transient responses with 2.0% structure damping under different earthquake loading with optimized
tuned mass damper of 1% mass
748
3.2. Base-Isolator (BI)
A typical three dimensional frame structure is attached with four isolators at the base. The finite element model
coupled with isolators is shown in Fig.2. Isolator is modeled as a two noded linear spring element attached with
concentrated damper. Total mass of frame structure is 1584 kg, and mass of isolator is neglected. Base isolator is
modeled based on linear elastic -viscous model. The isolator frequency is chosen in such a way that to give
effective time period which is away from dominant frequencies of the earthquake ground motions. In this
particular case study the frequency ratio of isolator (rb) is taken as 0.2 which gives the effective time period of
isolator as 1.20 second. Isolators are provided with 5% damping.
Table 3 Natural frequencies and mass participations in X-direction

With fixed-base With isolator ( rb = 0.2, ξ b = 0.05 )
Mode
Number Frequency Mass participation in Frequency Mass participation in X-
(Rad/sec.) X-direction (%) (Rad/sec.) direction (%)
1 25.9604 69.60 5.1193 ~100
2 87.4438 9.70 45.425 2.97E-2
3 182.8608 15.96 103.946 6.42E-4
It is observed from Table-3 that first mode frequency of structure with isolators is more or less same as the isolator
fundamental frequency. Nearly 100% mass is participating in the first mode. It implies that the superstructure is
behaving rigid with respect to isolator, and it can also be demonstrated by Fig.7 (a). The three mode shapes in
transverse direction (X-direction) with Base-Isolator are shown in Fig.7.
(a) (b) (c)
Figure 7 (a) 1stmode in X-direction, (b) 2ndmode in X-direction, (c) 3rdmode in X-direction
3.2.1 Seismic effectiveness of base isolator
The time variations of responses of base isolated structure with frequency ratio 0.2 are plotted in Fig.8. It is noted
that efficiency of seismic isolator is different for earthquake excitations although the parameters of isolators are
same. Structural response depends upon dynamic characteristics of the structure and frequency content of input
motion. However significant response reduction in the response at the top floor of structure is obtained in
dominant zone of earthquake excitation for two different ground motions Th-1- and Th-2.
9 18
Fixed-base Fixed-base
(a)
Th-1 (a) Th-2
With isolator With isolator
6
Abs. top floor accel. (m/s )
12
Abs. top floor accel. (m/s )
2
3
6
0
0
-3
-6 -6
-9
-12
-12
-18
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
Figure 8 Transient responses under earthquake excitation with isolators
749
3.3. X-plate damper (XPD)
X-plate damper is an energy dissipating device, consists of an assembly that holds either single or multiple
components of thin metallic or layered plates of ‘X’ shape. XPD can sustain many cycles of stable yielding
deformation without fatigue, thus dissipating the input seismic energy in the form of hysteretic deformation. The
force-displacement curve shows nonlinear hysteretic characteristics. The material characteristics of the X-shaped
plate such as the elastic modulus, and the strain hardening rate, are expressed by the bi-linear model. The
force-displacement curve for X-shaped plate can be evaluated by beam theory in elastic region, and by using
empirical relation in plastic region [13]. The FE model of piping attached with X-plate damper is shown in Fig.3
(b). Table-4 shows the material and geometric properties of X-plate.
Table 4 Material and geometric properties of X-plate
a (mm) b (mm) σ y (MPa) E (MPa) H (%)
40 40 220 2.06e5 2.58
Based on above material and geometric properties, for different thicknesses of X-plate the load-deflection
characteristics of X-plate for strain hardening rate (H) of 2.58% can be evaluated as shown in Fig.9.
1400 t=2.0 mm
t=3.0 mm H=2.58%
1200 t=4.0 mm
1000
Load (N)
800
600
400
200
0
0.000 0.005 0.010 0.015 0.020 0.025 0.030
Displacement (m)
Figure 9 Force-displacement curves for X-plate with different thicknesses (a = b = 40 mm)
3.3.1 Seismic effectiveness of X-plate damper
There is considerable change in vibration characteristic of piping system at higher thickness of X-plate, which
affects the seismic response. The reduction in the response of structure represents the combined effect of change
in the dynamic characteristic as well as the hysteretic energy dissipation characteristics of X plate damper. The
effect on the vibration characteristic of piping system with different thicknesses of X-plate is demonstrated in
Table-5. It is noted that with increasing thickness of X-plate, the mass participation in first mode is considerably
changed, however the natural frequency is not affected much. The Piping system is provided with 3.0% damping.
Table 5 Natural frequency and mass participation

of first mode for subject piping
Piping system
t (mm) Natural Mass participation
frequency (Hz) in Z-dir. (%)
No XPD 5.18 94.8
2 5.34 62.2
3 5.35 52.9
4 5.36 50.5
Time variation of the relative displacement at the XPD location and the support reaction at the end support of pipe
under earthquakes are shown in Figs. 10(a) and 10((b) for t =3 mm. It is evident from these Figures that there is a
significant reduction in the displacements and reactions for the piping system with an XPD.
750
0.010 0.015
0.008 (a) Th-1 Without X-plate (b) Th-2 Without X-plate

With X-plate 0.010
With X-plate
0.006
0.004 0.005
0.002
xd(m)
xd(m)
0.000
0.000
-0.002
-0.005
-0.004
-0.006 -0.010
-0.008
1600 -0.015
2400
1200
1600
800
400 800
Rx(N)
Rx(N)
0 0
-400
-800
-800
-1600
-1200
-1600 -2400
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
Figure 10 Transient displacement and support reaction of the piping under earthquake excitation with X-plate
thickness ( t ) =3 mm
In Table-6, the peak response quantities of the piping system with and without XPD are compared for different
ground motions. The results are tabulated for an XPD of size a =40 mm, b =40 mm, and for three different
thicknesses of the X-plate. Higher reduction in the peak responses of the piping is noted for XPD with higher
thicknesses.
Table 6 Peak response quantities of piping system without and with X-plate damper
Subjected to ground motions for ( a = b = 40 mm).
With XPD
Without XPD
Excitation t =2 mm t =3 mm t =4 mm
x (mm) R (N) x (mm) R (N) x (mm) R (N) x (mm) (N)
d x d x d x d
Th-1 7.22 1340.4 3.90 833.9 3.00 653.4 1.71 460.80
Th-2 12.02 2275.9 9.46 1836.6 8.42 1604.5 6.14 1184.4
Where, xd - Displacement of piping at XPD location
Rx - Reaction force at the pipe end
4. CONCLUSIONS
The paper investigates the effectiveness of the three different types of devices for attenuation of seismic response.
The tune mass damper is best suitable for existing tall structure whereas the base isolation techniques provides
filtering the ground motion at base of the structure itself, thus standardizing the design of structure for different
ground motion. The X-plate damper is highly suitable for response reduction of piping and equipment. Studies
performed with earthquake ground motion demonstrate that the optimum tuned mass damper is found to be
effective for considerable seismic response attenuation. Reduction in seismic response of structure with base
isolator is significant. However, the results are dependent on the characteristics of earthquake excitation. The
analyses with X-plate damper show that it can significantly influence the vibration characteristics of piping
system depending on its parameters and effective for attenuation of seismic response of piping system.
Effectiveness of X-plate damper and sensitivity to thickness of plate on piping system is demonstrated.
REFERENCES
[1] Brock, J. E. (1946). A note on the damped vibration absorber. Journal of Applied Mechanics/American
Society of Mechanical Engineers, A, 284.
751
[2] Den Hartog, J. P. (1956). Mechanical vibrations, 4th edn, McGraw-Hill, New York
[3] Sadek, F.; Mohraz, B.; Taylor, A.W. & Chung, R. M. (1997). A method of estimating the parameters of tuned
mass dampers for seismic applications. Earthquake engineering and Structural dynamics, Vol. 26, 617-635.
[4] Khan, I.; Agrawal, M. K.; Singh, R.K.; Ghosh, A. K. and Sasidharan, K.(2009). Optimization methodology
for tuned mass damper and seismic effectiveness. 3rd International Congress on Computational Mechanics
and Simulation (ICCMS09), pp.41-42.
[5] Lin, C.C.; Wang, J. F. and Ueng, J. M. (2001). Vibration control identification of seismically excited m.d.o.f.
Structure-PTMD systems. Journal of Sound and vibration, 240(1), 87-115.
[6] Lee, C. L.; Chen, Y. T.; Chungc, L. L. and Wang. Y. P. (2006). Optimal design theories and applications of
tuned mass dampers. Engineering Structures, 28, 43–53.
[7] Gueraud, R.; Noel-Leroux, J.P.; Livolant, M. and Michalopoulos. A. P. (1985). Seismic isolation using
sliding-elastomer bearing Pad. Nuclear Engineering and Design ,84, pp.363-377.
[8] Ahmadi,G. and Lin, Su. (1992). Equipment response spectra for base-isolated shear beam structures. Nuclear
Engineering and Design, 132, pp.287-308.
[9] Marano, G. C. and Greco, R. (2003). Efficiency of base isolation systems in structural seismic protection and
energetic assessment. Earthquake Engng Struct. Dyn., 32, 1505–1531.
[10] Calio, I.; Marletta,M. and Vinciprova. F. (2003). Seismic response of multi-storey buildings base-isolated by
friction devices with restoring properties. Computers and Structures, 81, pp.2589–2599
[11] Kyu-Sik Park, Hyung-Jo Jung, In-Won Lee. (2002). A comparative study on aseismic performances of base
isolation systems for multi-span continuous bridge. Engineering Structures, 24, pp.1001–1013.
[12] Matsagar, V.A. and Jangid, R.S. (2004). Influence of isolator characteristics on the response of base-isolated
structures”, Engineering Structures, 26, 1735–1749.
[13] Bakre, S.V.; Jangid, R.S. and Reddy, G.R. (2006). Optimum X-plate dampers for seismic response control of
piping systems. International Journal of Pressure Vessels and Piping,83,672–685.
[14] Parulekar, Y.M.; Reddy, G.R. and Vaze, K.K. (2006). Passive Control of Seismic Response of Piping System.
American Society of Mechanical Engineers, vol. 128,364-369.
[15] Khan, I.; Agrawal, M. K.; Singh, R.K.; Ghosh, A. K. and Sasidharan, K.(2009). Effect of X-plate damper on
dynamic characteristic of piping system and seismic effectiveness. 3rd International Congress on
Computational Mechanics and Simulation (ICCMS09) , pp.87-88.
752
Paper No.: 137
EXPERIMENTAL AND ANALYTICAL ESTIMATION OF CONCRETE CREEP

FOR LARGE PROTOTYPE STRUCTURE
N. Trivedi and R. K. Singh
Bhabha Atomic Research Centre, Mumbai, India

Email: nehat@barc.gov.in, ntwipro08@gmail.com
ABSTRACT:
A sufficient realistic data on concrete creep for long-time serviceability, durability, long-time stability, safety
against collapse and to perform consistent non-linear analysis of large scale prototype structure-BARC
Containment (BARCOM) test model and small scale specimen is explored. The methodology to separate out the
creep, instantaneous and shrinkage strain on small scale–concrete cylindrical specimens and the large proto type
structure- BARCOM test model is investigated in the presented work. The creep, instantaneous and shrinkage
strain in cylindrical specimen and BARCOM test model is experimentally computed from the installed
embedded sensor of Vibratory Wire Strain Gauge (VWSG). The shrinkage strain value is found to be
insignificant in comparison to the creep and instantaneous strain values and hence present study focuses mainly
on the creep behavior. The experimentally observed creep and instantaneous strain is further validated with the
famous B-3 model for cylindrical specimens and BARCOM test model. The result observed experimentally and
through B-3 is found to be in excellent agreement. Thus the present study invokes the understanding of the
mechanisms of creep, instantaneous and shrinkage behavior on small scale cylindrical specimen and on large
size structure.
KEYWORDS: Creep, BARCOM, Instantaneous strain
1. INTRODUCTION
The time dependent inelastic strain of concrete consist of shrinkage strain which is independent of stress and
additional strain produced by mechanical stress called creep. The time dependent deformation plays a
significant role in serviceability and durability of concrete structures. Creep and shrinkage of concrete is an
intricate phenomenon and it is difficult to formulate a generally applicable and realistic constitutive equation.
Realistic prediction of concrete creep and shrinkage is of crucial importance for durability, long-time
serviceability of concrete structures, simulating the non-linear analysis of any experimental model and in some
cases also for long-time stability and safety against collapse. The concrete specimen shrinks when exposed to
the dry atmosphere by water loss. The creep occurring at the constant moisture content (no moisture exchange
with the environment) is called the basic creep [1]. The aging in concrete refers to the increase in strength and
elastic modulus with age at a gradually decreasing rate [2 and 3]. The concrete specimens loaded at a higher age
creep less. Since the aging property in concrete is different from the other materials where the aging refers to the
gradual degradation of strength and hence the time is always measured from the onset of concrete.
The analyzing and evaluating the concrete creep behavior for most of the concrete structures is an important
phenomenon. The safety of the structure depends on the creep and shrinkage of concrete and relaxation of steel
tendons. From the study on Swedish containment [4], it is concluded that the creep is the phenomenon with the
biggest contribution to the losses. Creep is one of the most important structural factors to be considered for the
safety of a pre-stressed concrete nuclear containment structure during design, construction and maintenance. So
it becomes very important to characterize the concrete creep behavior for BARC Containment (BARCOM) test
model and concrete cylindrical specimens [5 and 6] which is investigated in the present study.
2. BACKGROUND
As part of Nuclear containment structural safety assessment, Bhabha Atomic Research Centre (BARC)
organized a project BARCOM test model [5 and 6] based on very important experimental program in which
concrete creep, shrinkage and instantaneous strain studies are investigated. The test model located in BARC
facilities-Tarapur is a 1:4 scale representation of 540 MWe Pressurized Heavy Water Reactor (PHWR)
pre-stressed concrete inner containment structure of Tarapur Atomic Power Station (TAPS) unit 3&4. Around
1200 sensors installed in BARCOM include vibratory wire strain gauges (VWSG), surface mounted electrical
resistance strain gauges, dial gauges, earth pressure cells, tilt meters and high resolution digital camera systems
for structural response. The BARCOM test model is shown in Fig. 1(a), the free field location on BARCOM test
model does not have any openings so it is ideally suited location to carry out the creep, shrinkage and
instantaneous strain studies and for simplification it is defined as South side (0˚) as shown in Fig. 1(b). The
other location with the openings are designated as MAL (main ail lock/ West) at 90˚, FMAL (fueling machine
air lock/ North) at 180˚ and EAL (emergency air lock/ East) at 270˚ as shown in Fig. 1(b).
Fig. 1 (a) Barc Containment (BARCOM) Test Model at Tarapur site [5 and 6]
Fig. 1 (b) Plan of Inner Containment of Model [5 and 6]

BARCOM’s large cylindrical structure is visualised as horizontal rings with vertical strip equivalent to
754
individual pours of concrete where the creep, shrinkage and instantaneous behaviour of the concrete due to
pre-stress load is studied. In BARCOM, the construction of the cylindrical portion of the test model has been
carried out in various stages lift-wise and the dome construction was carried in the last single lift. The J and H
type cables were used for pre-stressing [5]. The concrete cylindrical specimens were cast from the same mix
used for the construction of BARCOM Dome during the period end of April 2008 [5]. Some of the concrete
specimens from the same pour were subjected to the load of 330 kN and the rest were load free so as to observe
the long term effect in concrete specimen. The size of cylindrical specimen is 150 mm diameter and 300 mm
length. The cylindrical specimens were subjected to load in May 2008 and BARCOM dome was pre-stressed in
August 2008. The cylindrical specimens and BARCOM dome were casted at the same time from the same mix
but time of application of load was different, which results in aging effect [2 and 3].
3. EXPERIMENTAL STUDY
The concrete cylindrical specimens were cast from the same mix used for the construction of BARCOM Dome
during the period end of April 2008 [5 and 6].The measurement of creep, instantaneous and shrinkage strain
requires the two sets of identical specimens subjected to the same environmental condition, one set of specimen
loaded and the other companion set of specimen being load free. Both the sets of concrete cylindrical specimens
were equipped with the strain gauges and reading is taken from the read out meter. The Creep Test Rig apparatus
is limited to concrete in which the maximum aggregate size does not exceed 50 mm, suitable for operation on
220 V, 50 Hz, Single phase AC supply. The difference between the deformations of the two set specimen consist
of creep plus instantaneous strain. The strain values of load free and loaded specimens have been periodically
noted to obtain long term behaviour. The strain value obtained from the load free specimen is purely due to
shrinkage and is observed to be insignificant value. The lower strain value is perhaps due to presence of Tarapur
site which lies in a zone where the average humidity is more than 90 % and hence the expected shrinkage strain
value should be small. The total strain value composed of creep, instantaneous and shrinkage observed
experimentally in the loaded specimens. Sudden rise of strain is the instantaneous strain value, which is
subtracted from the total strain of loaded specimen to separate out creep and shrinkage together as shown in Fig.
2. The average shrinkage strain value obtained from the load free specimens is deducted from the creep and
shrinkage strain for loaded specimens presented in Fig. 2 to obtain the creep component alone as shown in Fig.
3. The L1, L2, L3 stand for the three loaded specimen in Fig. 2 and Fig. 3.
Loaded Specimens
800
Creep & Shrinakge
700
Strain(µε)
600
500
400
300
200 L1
L2
100 L3 Time (days)
0
0 50 100 150 200 250 300
Fig. 2 Creep & Shrinkage strain for 3 loaded samples
755
700 Loaded Specimens
Creep Strain(µε)
600
500
400
300
200 L1
L2
100 L3
Time (days)
0
0 50 100 150 200 250 300
Fig. 3 Creep strain for 3 loaded samples
The strain data for BARCOM test model was obtained from the embedded VWSG sensors starting from the
construction phase, followed by pre-stressing level and beyond pre-stressing to establish long term strain
behavior. The pre-stressing of cables results in instantaneous strain. In similar way as explained for concrete
cylindrical samples Fig. 3, instantaneous and shrinkage strain is deducted from the total strain to separate out
creep component.
3. B-3 MODEL
Once the in-situ observed creep and instantaneous strain are segregated for concrete cylindrical samples and
BARCOM test model, then validation of experimental data is carried by the famous B-3 model [7]. The realistic
prediction for the creep properties of concrete obtained from mathematical B-3 model is based on the theoretical
formulation and statistical optimization with regard to majority of the test data on cylindrical specimens. In
literature mostly the B-3 model has been used for small scale specimen, however in the present work B-3 model
is implemented on large scale prototype structure - BARCOM test model along with small size specimen -
concrete cylindrical specimen. The B-3 model has good accuracy and is more rational. The prediction of the
material parameters of the present B-3 model from strength and composition is restricted to Portland cement
concrete with the following parameter ranges [7]:
0.35 ≤ w/c ≤ 0.85;
2.5 ≤ a/c ≤ 13.5;
17MPa ≤ fc ≤ 70MPa
fc is the 28 day standard cylinder compression strength of concrete MPa (SI) units, w/c is the water-cement ratio
by weight, c is the cement content kg/m3 (SI) and a/c is the aggregate-cement ratio by weight. The formulae are
valid for concrete cured for at least one day.
Since in our study the concrete mix [5 and 6] satisfy the all above mentioned condition on water-cement ratio,
aggregate-cement ration and standard cylinder compression strength of concrete so it has been proposed to
validate our experimental results with B-3 model.
For constant stress applied at age t′ strain as per B-3 model [7]:
ε(t) = J(t, t′)σ + εsh(t) + α∆T(t) (1)
in which J(t, t′) is the compliance function = strain (creep plus instantaneous) at time t caused by a unit uni-axial
constant stress applied at age t′, σ = uni-axial stress , ε = strain, εsh = shrinkage strain ∆T(t) = temperature
change from reference temperature at time t, and α = thermal expansion coefficient.
The material property characterizing the creep and instantaneous strain is the compliance function which is
calculated from the extensive data on the tests conducted on the concrete samples. The mathematical
formulation of creep is as explained:
756
Creep of concrete is an intractable property since it is age-dependent as well as time-dependent, and it is
affected by many other variables both intrinsic and environmental. The creep prediction formula as per B-3
model [7], is given as:
J(t, t’) = q1 + C0(t, t’) + Cd(t, t’ ; t0) (2)
Here J(t, t’) is the compliance function of concrete includes creep and instantaneous part, q1 is the inverse of the
asymptotic elastic modulus = 1/ E0 = instantaneous strain due to unit stress; Cd(t, t’ ; t0) is the additional
compliance due to drying. The creep without moisture exchange, called basic creep, can be closely described for
various ages t' at loading by the double power law.
Above mentioned Eq. (2) reduces to following form after simplification:
J(t, t’) = (1/55900)+(3.4 X 10-4 ln[ 1 + 0.3( t’-.5 + 0.001)(t – t’)0.1 ]) (3)
Where t’ is that time since load start acting on concrete and t is the time at which the strain is computed (t ≥ t’).
4. EXPERIMENTAL AND ANALYTICAL COMPUTATION
Already discussed in the previous sections that the BARCOM test model construction of the cylindrical portion
of the containment has been carried out in various stages lift-wise and the dome construction was carried in the
last single lift. The concrete cylindrical specimens were cast from the same mix used for the construction of
BARCOM dome during the period of April 2008. Some of the concrete cylindrical specimens of size 150 mm
diameter and 300 mm length were subjected to the load and the rest were load free. The computation of creep
and instantaneous strain for cylindrical specimen and BARCOM model is presented below:
Illustrations are to be centered, with the reference and caption printed below the figure. The lettering used in the
illustrations should not be too small. Illustrations are to be referred to as figures, and must be quoted in the text.
One blank line should be left between the figure caption and the next paragraph. Please ensure that all figures
are of the highest quality. Photographs must have a resolution of at least 300 dpi.
4.1 Cylindrical Specimen
Table 1 presents the computed value of compliance function from equation (3) at a few selected period of time
and the segregated creep strain value from the total strain that consist of creep and instantaneous component.
The experimentally observed creep strain on concrete cylindrical samples as shown in the previous section, Fig.
3 is compared with creep strain evaluated from the B-3 model. The result obtained from the experiment and
with B-3 model are shown together in Fig. 4 and found to be in excellent agreement. In Fig. 4, L1, L2, L3
denotes the three loaded concrete cylindrical specimen.
TABLE 1
SEGREGATED CREEP VALUES FOR CYLINDRICAL SPECIMEN
Time = t (days) Compliance function = J(per MPa) Creep strain
-5
119 4.59 X 10 526
-5
179 4.74 X 10 553
-5
300 4.91 X 10 585
757
700 Loaded Specimens
Creep Strain(µε)
600
500
L1
400 L2
L3
300
200
100
Time (days)
0
0 50 100 150 200 250 300
Fig. 4 Creep strain as per the B-3 Model and experimental data
4.1 BARCOM Test Model
Based on B-3 model, depending on the pre-stressing schedule the creep in hoop and longitudinal direction of
BARCOM wall is evaluated for BARCOM wall at El. 0.5m, El. 3.5m and El. 6.5m. The experimental creep
strain at El 0.5m, El. 3.5m, and El. 6.5m is segregated from the instantaneous strain. The creep strain predicted
experimentally and with B-3 model in hoop and longitudinal direction of BARCOM wall at El 0.5m is shown in
Fig. 5. At each elevation, there are many embedded sensors in the BARCOM wall so it becomes important to
select the upper and lower bound strain value which is also shown in Fig. 5. The mean experimental creep strain
in hoop and longitudinal direction is also obtained from the sensors reading and is plotted in Fig 5. The mean
experimental creep values at the above mentioned elevation is found to be in excellent agreement with the B-3
model.
600 700
Creep strain(µε)
Creep strain(µε)
500 600
500
400
400
300
300
200
B-3 Model
200 B-3 Model
100 Lower Bound Experimental value Lower Bound Experimental value
Upper Bound Experimental value 100 Upper Bound Experimental value
Mean Experimental Value Time (days) Mean Experimental Value
0 Time (days)
0
0 100 200 300 400 500 600 0 100 200 300 400 500 600
(a) (b)
Fig. 5. Creep strain of BARCOM wall at El 0.5 m in (a) Hoop & (b) Longitudinal direction
5. CONCLUSION
The present study invokes the understanding of the mechanisms of creep, instantaneous and shrinkage behavior
on small scale cylindrical specimen and on large size structure. The investigated strain data is very useful to
account durability, the long term losses in the concrete material while simulating the non-linear analysis of any
experimental model, long-time serviceability of concrete structures and in some cases also for long-time
stability and safety against collapse.
758
REFERENCES
[1] Bazant Z. P. and Panula L., “Basic Creep”, Matiriaux Et constructions,Vol. 11,N-65.
[2] Bazant Z. P. and Santosh Prasannan, “Solidification Theory For Aging Creep”, Cement and Concrete
Research, 1988; 18, 923-932.
[3] Bazant Z. P., “Prediction of Concrete Creep and Shrinkage: past, present and future”, Nuclear Engineering
and Design, 2001; 203, 27–38.
[4] Lundqvist Peter and Nilsson Lars-Olof, “Evaluation of prestress losses in nuclear reactor containments”,
Nuclear Engineering and Design, 2011; 241, 168-176.
[5] Singh R. K., “International Workshop on Pre-Test Round Robin Analysis of BARCOM”, 2009.
[6] Singh R. K., “Post Test Preliminary Report on Int. Round Robin Analysis for Ultimate Load Capacity
Assessment of BARC Containment (BARCOM) Test Model” Feb07 2011.
[7] Jirasek Milan and Bazant Z. P., “Inelastic Analysis of Structures”, John Willey & Sons Ltd 2002.
759
Paper No.: 138
NUMERICAL SIMULATION FOR VTT IMPACT TEST UNDER IRIS

PROGRAM
N. Trivedi and R. K. Singh
Bhabha Atomic Research Centre, Mumbai, India

Email: nehat@barc.gov.in, ntwipro08@gmail.com
ABSTRACT :
The three dimensional transient dynamic finite element simulations is presented to assess the integrity of
concrete structures impacted by missiles for Nuclear Energy Agency sponsored impact tests for VTT punching
rupture tests. Impact loading condition is characterized by the application of a force of great intensity within a
short duration. Impact on a reinforced concrete (RC) structure like aircraft crash, missile hit and free fall of a
heavy machine part leads to different types of global and local damage such as flexure, penetration, scabbing,
spalling, crushing, cracking, mixed mode and punching shear failure. The evaluation of each of the above
phenomena is highly complex from the point of view of experiments as well as numerical simulations. The
impact load application due to accident condition on nuclear structures is one of the main safety concerns for
Nuclear power plant. It is, therefore, an important issue to invoke the computational systematic approach for
identification of the impact induced failure modes since experimental approach is not cost-effective practicable
solution. Evaluation of concrete targets under missile impact is important for containments and other nuclear
safety related structures. Evaluation of concrete targets under missile impact is important for containments and
other nuclear safety related structures. The numerical simulation of the concrete target impact tests is carried out
to study the target response and identify the failure mechanisms of the benchmark test problems under
IRIS-2012 program
KEYWORDS: Impact, Reinforced Concrete, Failure modes
1. INTRODUCTION
The safety of critical structures such as Nuclear Power Plant (NPP), fuel storage facilities and important
facilities under impact load is of prime concern. The impact load application due to internal/external accident
condition on nuclear structures is one of the main safety concerns for NPP. It is, therefore, an important issue to
invoke the computational systematic approach for identification of the impact induced failure modes since
experimental approach is not cost-effective practicable solution.
The ultimate objective of the IRIS program is to provide recommendation for structural modeling and analysis
as well as design criteria for structure impacted by a missile. The objective of this paper is to present the
Nuclear Energy Agency impact benchmark simulation results for VTT Punching rupture test obtained by
Bhabha Atomic Research Centre (BARC), Trombay. The numerical simulation of the concrete target impact
tests is carried out to study the target response and identify the failure mechanisms of the benchmark test
problems under IRIS, Rambach et. al [1]. The basic emphasis has been to identify the failure modes of
concrete targets through mechanics based identifiable parameters of slab overall deformation pattern, strains and
stresses in concrete and rebar elements. Bhabha Atomic Research Centre (BARC), Trombay has been active in
the area of ultimate load capacity evaluation of nuclear containment structures for static and transient dynamic
impact loads as clear from study by Trivedi and Singh [2] and Trivedi and Singh [3]. Based on the available
geometrical details of the concrete slab and steel missile and the available constitutive data Vepsa et. al. [4], the
VTT punching test is numerically simulated. In the tests performed at VTT, the walls were simply supported at
all of their four sides with the span width being 2 m. All the walls were reinforced with bending reinforcement
in both horizontal and vertical directions with the horizontal rebar being nearer to the surface of the wall. The
walls used in tests for punching behavior were 250 mm thick and they did not include shear reinforcement.
These “punching” walls were tested by impacting a non-deformable missile with mass of roughly 47.5 kg
against them with the target impact velocity being 135 m/s. Main instrumentation in the tests carried out by VTT
consisted of strain gauges placed on the reinforcement rebar as well as on the front surface of the wall, strain
gauges on the supports of the mounting frame of the wall to compute the support forces, five sensors for
measurement of displacements and high shutter speed video cameras to record the impact.
2.CONSTITUTIVE MODEL
2.1 Concrete Damage Plasticity

Continuum damage-plasticity model Tao and Phillips [5] for concrete is based on the mechanism of tensile
cracking and compressive crushing. The evolution of the yield/failure surface is governed by the tensile and
compressive equivalent plastic strains respectively. The constitutive behavior of concrete is described by
introducing a scalar variable d which is designated as dt and dc for concrete in tension and compression
respectively. In continuum damage-plasticity model, the degradation of the elastic modulus induced by plastic
straining both in tension and compression of concrete is given in terms of a scalar degradation variable d as: E=
(1-d)×E0
2.2Post-failure behavior
The post-failure behavior is modeled with tension stiffening by defining the strain-softening behavior for
cracked concrete. The tension stiffening is applied by post-failure stress-strain relation and fracture energy
criterion as shown in Fig. 1(a) and Fig. 1(b) respectively. Tension stiffening in stress-strain criteria is defined in
terms of the cracking strain and damage parameter. The energy required to open a unit area of crack called
fracture energy is designated as Gf which is a material parameter. With this approach the concrete's tensile
behavior is characterized by a stress-displacement response. The cracking displacement at which complete loss of
strength takes place is expressed in the following form; uto = 2Gf / ft’ where Gf is the fracture energy of concrete
as per CEB-FIP model as given by Mehta and Monteiro [6]. The ft’ is the yield tensile strength of concrete (ft’ = σt0).
Fig. 1(a) post failure stress-strain Fig. 1(b) Fracture energy criteria
2.3 Concrete plasticity
The plastic flow potential function and the yield surface defined by the two stress invariants of the effective
stress tensor namely the hydrostatic pressure stress and the Mises equivalent effective stress. The typical yield
surfaces are shown in Fig. 2(a) and Fig. 2(b) on the deviatoric plane and plane stress conditions respectively.
761
Fig. 2 (a) Deviatoric plane Yield surfaces Fig. 2 (b) Plane stress Yield surfaces
3. FINITE ELEMENT (FE) INVESTIGATION AND EXPERIMENTAL SET UP
Three dimensional numerical model of the experimentally tested reinforced concrete slab and solid missile is
shown in Fig. 3(a). Fig. 3(b) presents the reinforcement pattern. The solid slab is modeled as three dimensional
deformable body with 8 noded brick element, the reinforcement as a deformable wire shape element, and
missile as a three dimensional rigid shell element. The reinforcement is defined as the embedded element and
the concrete slab as the host element. The Truss–in-solid type of “embedded elements-in-host elements” is used
in the present study. It is assumed that there is no slip between concrete and steel. Meshing of concrete slab was
done with 3D continuum elements using the reduced integration and reinforcement is meshed as linear truss
element. The three dimensional slab is supported at the edges by providing fully constrained boundary
conditions along the three axes.
A sketch of the testing apparatus is shown on the left in Fig. 4, Vepsa et. al. [4]. The basic idea is to use air
pressure in acceleration of the missile to its target speed. The missile rests on rails which are located on top of
the acceleration tube. In an acceleration phase, the missile is pushed by a fin of the piston. While the missile
continues its flight and ultimately hits the target, the piston is stopped by a piston catcher before it hits the target.
The material behavior of reinforce concrete slab is incorporated using the concrete damage plasticity model for
concrete and the rebar by using conventional biaxial stress-strain plasticity model. The in-house code based on
concrete damage mechanics is used to incorporate the material property. The material constitutive parameters
for concrete and steel are taken as given in experimental input Vepsa et. al. [4].
Fig. 3(a) Finite Element Mesh Fig. 3(b) Reinforcement detail
762
Figure 4. Test apparatus
4. RESULTS & DISCUSSION
Certain specific locations close to the punched area and away from the impacted area are selected in the
experiment to establish the slab displacement and the reinforcement strain. The transverse deflections at the
standard locations on the rear side of the concrete slab are shown in Fig 5. The sensor locations 2 and 3 are very
close to the punching area whereas the sensor locations 4 and 5 are away from the punch area. The displacement
and the velocity variation of the rigid missile causing the impact are shown in Fig. 6 (a) & (b) respectively. The
reinforcement strain variation with time at locations from D-1 to D-7 in the impacted area is shown in Fig. 8.
Fig.5 Deflection at standard location on Rear face Fig. 6(a) Missile Displacement
Fig. 6(b) Missile Velocity Fig.7 Reinforcement strain at standard location
763
A systematic numerical approach to predict the various failure modes uch as crushing, cracking, spalling and
scabbing, under impact load is investigated in this study. The damage failure pattern on the slab interacting with
the rigid missile is shown at particular instant of time in Fig. 8 (a). Fig. 8 (b) shows the cone cracking observed
numerically throughout the slab thickness.
Fig. 8(a) Damage on slab Fig. 8(b) Cone Cracking
4. CONCLUSION
This paper evolved a systematic three dimensional finite element inelastic analysis procedure for reinforced
concrete slabs subjected to missile impact. Numerical simulations have been performed to study the transient
dynamic behavior of slabs for identification of consistent failure modes. Impact test was carried out at VTT in
which a non-deformable missile was impacted against a reinforced concrete resulting in the punching failure. The
displacement and velocity variation of rigid missile is captured. The displacement and strain variation in the
reinforced concrete slab is established. The missile perforation in the reinforced concrete slab is captured and the
radial cracks and damage on the slab is also investigated.
REFERENCES
[1] J M Rambach, N Orbovic and F Tarallo (2011). "IRIS-2010-Part I: General Overview of the Benchmark",
SMiRT-21, New Delhi, India
[2] N. Trivedi and R K Singh (2013)." Prediction of impact induced failure modes in reinforced concrete slabs
through nonlinear transient dynamic finite element simulation ", Annals of Nuclear Energy, 56(2013),
109-121
[3] N. Trivedi and R K Singh (2011)." Identification of Failure Modes in Safety Related Nuclear Structures for
Missile Impact ", SMiRT-21, New Delhi, India
764
[4] A. Vepsa, A. Saarenheimo, F. Tarallo, J.M. Rambach, and N. Orbovic (2011). “IRIS-2010 Part II:
Experimental Data ” SMiRT-21,New Delhi, India
[5] Tao Xiaoya and V. Phillips David. 2004. “A simplified isotropic damage model for concrete under
bi-axial stress states”. Cement and Concrete composites.
[6] Mehta P Kumar and Monteiro Taulo J M. 1997. “Concrete Microstructure Property and Materials”.
765
Paper No.: 139
SYNTHETIC GROUND MOTION GENERATION USING A SEMI

ANALYTICAL MODEL
A. Ravi Kiran1, M. K. Agrawal1, G. R. Reddy1, R. K. Singh1 and K. K. Vaze2

1
Reactor Safety Division, BARC, Mumbai, India
2
Reactor Design & Development Group, Bhabha Atomic Research Centre, Mumbai, India.
Email: arkiran@barc.gov.in
ABSTRACT :
Simulation of earthquake ground motion is necessary for estimating seismic hazard for various systems of
nuclear facilities. The time-history of ground acceleration is required for seismic analysis of structures on
ground and for generation of response spectra at various floor levels. The present work is aimed to generate
synthetic time histories for a given earthquake by inverting the Fourier Amplitude Spectrum (FAS) of
earthquake ground motion which is evaluated using a semi-analytical model. The FAS is based on a point
source of earthquake with a single corner frequency and various seismological considerations. The model allows
the evaluation of FAS of earthquake ground motion in terms of earthquake magnitude and distance between the
source and the site. First, a time history is generated by inverting the FAS with random phase angles distributed
uniformly in (0,2π). Then, the time history is multiplied by an enveloping function to obtain synthetic
accelerogram. A nonlinear optimization algorithm is used to obtain optimum values of the envelope function
parameters. Simulation of synthetic acceleration time histories has been carried out for various earthquakes
whose records are collected from different stations. An empirical correlation has been developed for the site
attenuation factor as a function of hypocentral distance and shear wave velocity. The present paper gives the
details of simulation along with comparison of recorded and computed accelerograms and the corresponding
FAS. This methodology will be useful to generate design basis response spectra for the site of nuclear facilities
with sparse earthquake records.
KEYWORDS: Fourier Amplitude Spectrum, Envelope function, Accelerogram
1. INTRODUCTION
Nuclear facilities have to be designed for normal as well as accidental loads such as earthquake. The design
basis earthquake ground motion is specified in terms of peak ground acceleration (PGA), response spectral
shape and acceleration time history (Ghosh et al. (1990)). Usually a spectrum compatible time-history of ground
acceleration is specified such that the time-history generated response spectrum is compatible with a specified
design response spectrum. The time-history of ground acceleration is required for time-history analysis of
structures on ground and for generation of response spectra at various floor levels.
Ground motions from earthquakes of moment magnitude more than five; at distances enveloping 300 km from
the source (Atkinson et al. (2006) are significant for seismic-hazard analysis for nuclear power plants. In the
present work, a semi analytical model (Atkinson et al. (2006) is used to obtain FAS of the ground acceleration.
This model, based on seismological considerations, allows the evaluation of FAS of earthquake ground motion
in terms of magnitude and distance.
2. METHODOLOGY FOR SIMULATION OF EARTHQUAKE GROUND MOTION
The various steps involved in the generation of earthquake ground motion [2] are as follows. First, a random time
history is generated by inverting the FAS with random phase angles distributed uniformly in (0,2π). Then, the time
history is multiplied by an enveloping function to obtain the synthetic time history. FAS of the ground
acceleration is computed using the seismological model which is described below.
The model given by Atkinson [2] for the Fourier Amplitude Spectrum (FAS) of ground acceleration is,
2
[
A( f ) = CM o (2πf ) exp(− πfκ ) exp(− πfR / Qβ s ) / R 1 + ( f / f o )
2
] (1)
where, M0 is seismic moment and f0 is corner frequency, which is given by
f o = 4.9 x106 β s (∆σ / M o )
1/ 3
(2)
where, ∆σ is stress parameter in bars, M0 is in dyne centimeters, and β s is shear-wave velocity in kilometers
per second. The constant C is given by,
(
C = Rθφ FV / 4πρβ 3 ) (3)
where, Rθφ is radiation pattern (average value of 0.55 for shear waves), F is free surface amplification (2.0), V
is partition onto two horizontal components (0.71), ρ is the density, and R is hypocentral distance. Various
constants in the Eq. (1) are taken from the paper by Atkinsion. The term exp(− πfκ ) is a high-cut filter to
account for near-surface attenuation effects, which describe the commonly observed rapid spectral decay at high
frequencies, exp (− πfR / Qβ ) / R is the attenuation term. The quality factor, Q(f) is an inverse measure of
anelastic attenuation. Through this equation, the spectrum is diminished with distance to account for empirically
defined attenuation behaviour.
The envelope function E(t) is chosen as,
E (t ) = exp(−αt ) − exp(− βt ) (4)
Where, α and β are constants. By equating the first derivative of E(t) to zero, the rise time tR can be obtained as,
1 β
ln − t R = 0 (5)
β −α α
The duration of earthquake is taken as the time interval between 5% of peak acceleration. The times
corresponding to 5% peak acceleration are δt and (tD+ δt). By substituting in Eq. (4), we get
[ ]
0.05 e −αt R − e − βt R − e −αδt + e − βδt = 0 (6)
[ ]
0.05 e −αt R − e − βt R − e −α (t D +δt ) + e − β (t D +δt ) = 0 (7)
The optimum values of α and β are obtained by solving Eqs. (5, 6 and 7) in the least square sense using a nonlinear
optimization algorithm.
3. DETAILS OF STRONG MOTION DATA BASE OF HISTORIC EARTHQUAKES
The strong ground motion data base of historic earthquakes is classified into various groups based on the soil
classification given by Uniform Building Code (UBC (1997)). This classification is given in Table 1. Table 2
gives the details of strong motion data base of historic earthquakes whose records are obtained from Pacific
Earthquake Engineering Research Center (PEER) Strong Motion Database and the European Strong Motion
Database (ESD). The earthquakes are divided into four site classes based upon the shear wave velocity of the
recording station at 30m depth. The earthquakes considered are having magnitude greater than 5.0 and hypo
central distance less than 100 km.
Table 1: Site classification based upon soil profile
Site class Soil profile name Vs30 (m/s)

A Hard Rock >1500
B Rock 760-1500
Very dense soil and soft
C 360-760
rock
D Stiff soil 180-360
767
4. SIMULATION OF GROUND MOTION FOR VARIOUS EARTHQUAKES
In the present work, ground motion simulation has been carried out for various earthquakes listed in Table 2. The
acceleration records are obtained from Pacific Earthquake Engineering Research Center (PEER) and the
European Strong Motion Database (ESD). Figure 1 shows the acceleration record for Kalamata earthquake (S.
No. A1) in horizontal direction. The peak ground acceleration of the record is 0.021 g. The rise time and duration
for the earthquake are 7.44s and 23.98s respectively. The procedure used for ground motion simulation is
described in the following section.
Fourier Amplitude Spectrum (FAS) for this earthquake is computed using Eq. (1). The high cut filter parameter
(kappa) is computed from an empirical correlation given in the subsequent section. A random time history is
generated by inverting the FAS with random phase angles which are distributed uniformly in (0,2π). The optimum
values of the envelope function parameters (α and β) are obtained by solving Eqs. (6, 7 and 8) in the least square
sense using a nonlinear optimization algorithm. The envelope function parameters α and β for the earthquake are
0.1334 and 0.1354 respectively. The random time history is multiplied by the enveloping function to obtain the
synthetic time history. Then FAS of the synthetic acceleration time history is computed and compared with that of
record.
Table 2: Details of various earthquakes
Site Name of earthquake R Vs30

Id. Station name M tR (s) tD (s)
class & date (km) (m/s)
Kalamata earthquake
A A1 Githio-Police Station 6.4 61 7.44 23.98 2300
13/10/97
Aigion earthquake
A A2 Karpenisi-Prefecture 6.5 71 8.3 25.2 2165
16/6/95
Friuli earthquake Tolmezzo-Diga
B B1 6.5 23 3.56 18.2 1021
6/5/76 Ambiesta
Friuli earthquake Tolmezzo-Diga
B B2 5.2 27 2.17 6.48 1021
7/5/76 Ambiesta
Montenegro
B B3 Ulcinj-Hotel Albatros 5.4 15 1.65 11.43 1083
earthquake 9/4/79
Campano Lucano
B B4 Auletta 6.9 25 14.3 66.44 1092
earthquake 23/11/80
KOBE earthquake CUE 99999
C C1 6.9 19.9 3.26 21.1 609
16/1/95 Nishi-Akashi
Borah Peak
C C2 99999 CEM 5.1 25.2 2.94 18.71 659.6
earthquake 29/10/83
C C3 Coalinga 2/5/83 CDMG36416 6.36 47.6 6.73 36.22 376.1
C C4 Coalinga 9/7/83 USGS1608 5.18 11.8 3.46 18.40 376.1
Chi-Chi Taiwan
C C5 CWB99999 5.9 85.2 12.1 34.34 375.3
20/9/99
C C6 Hollister 28/11/74 CDMG47126 5.14 11.2 2.41 18.19 370.8
KOBE earthquake
D D1 CUE 99999 Takatori 6.9 22.2 1.97 30.01 256
16/1/95
Northern California USGS 1023 Ferndale
D D2 6.5 31 7.08 35.31 219
earthquake 21/12/54 City Hall
D D3 Hollister 26/1/86 USGS1656 5.45 17.1 6.52 33.76 215.5
D D4 Northridge 17/1/94 CDMG24612 6.69 36.2 7.96 29.94 270.2
D D5 Big Bear 10/2/01 USGS5328 4.53 41.7 6.45 19.45 271.4
Chi-Chi Taiwan
D D6 CWB99999 6.3 63.5 10.5 65.46 215
25/9/99
Imperial valley
D D7 USGS5115 5.01 20.4 1.97 7.84 188.8
15/10/79
768
5. DEVELOPMENT OF EMPIRICAL CORRELATION FOR KAPPA
In the simulation, near-surface attenuation effects are accounted by the high cut filter term, exp(− πfκ ) which
describe the commonly observed rapid spectral decay at high frequencies. An empirical correlation given in the
literature (Douglas et al. (2010)) is used as initial estimation of kappa. The correlation given by Douglas et al
(2010) is:
κsoil = 0.027+0.000175R
κrock = 0.0207+0.000175R (8)
The numerical simulation of ground motion has been carried out for all listed earthquakes. Later, the values of
kappa (κ) are slightly modified to produce better match of FAS of synthetic accelerogram with that of recorded
accelerogram. As kappa is also strong function of soil condition like rock or soil, effect of shear wave velocity is
also incorporated in the new correlation. Using these values, a new correlation has been obtained for kappa (κ)
in terms of shear wave velocity and hypo-central distance and is given below:
κ = 0.028-0.004Vs30+0.000175R (Vs30 is in km/s and R in km) (9)
The kappa values computed from this equation along with initial values are given in table 3. The envelope
function parameters (α and β) for various earthquakes are also listed in the table.
Table 3: Estimation of kappa as a function of Vs30 and R
Vs30
Id. R (km) κinitial κ (Eqn. 9) alpha beta
(m/s)
A1 2300 61 0.031 0.029 0.1334 0.1354
A2 2165 71 0.033 0.032 0.1204 0.1205
B1 1021 23 0.025 0.028 0.2429 0.3227
B2 1021 27 0.025 0.029 0.4608 0.4609
B3 1083 15 0.023 0.026 0.5504 0.6654
B4 1092 25 0.025 0.028 0.0605 0.0798
C1 609 19.9 0.030 0.029 0.2238 0.4081
C2 659.6 25.26 0.031 0.030 0.2671 0.4241
C3 376.1 47.63 0.035 0.035 0.1151 0.1880
C4 376.1 11.86 0.029 0.029 0.4198 0.5453
C5 375.3 85.2 0.042 0.041 0.0807 0.1489
C6 370.8 11.19 0.029 0.028 0.2821 0.5819
D1 256 22.19 0.031 0.031 0.4601 0.5583
D2 219 31 0.032 0.033 0.1124 0.1747
D3 215.5 17.14 0.030 0.030 0.1222 0.1894
D4 270.2 36.24 0.033 0.033 0.1179 0.1337
D5 271.4 41.72 0.034 0.034 0.0977 0.1662
D6 215 63.5 0.038 0.038 0.0598 0.1079
D7 188.8 20.39 0.031 0.031 0.2754 0.2756
Final simulations have been carried out by using kappa from Eqn. 9. One earthquake from each site category is
chosen for presenting the results. Kalamata earthquake (Id. A1), Campano Lucano earthquake (Id. B4), KOBE
earthquake at Nishi-Akashi (Id. C1) and KOBE earthquake at Takatori (Id. D1) are selected for hard rock, rock,
very dense soil and stiff soil respectively. Except Kalamata earthquake all other earthquakes are having magnitude
of 6.9 and hypocentral distance in between 20-25 kms. Kalamata earthquake has a magnitude of 6.4 and
hypocentral distance of 61 km.
769
Recorded accelerogram for hard rock site (class A) is shown in Figure 1. Corresponding synthetic time history is
shown in Figure 2.The comparison of normalized FAS of synthetic and recorded time histories is shown in Figure
3. In Figure 4, comparison of power spectral density (PSD) of synthetic and recorded time histories is shown.
Figs. 5 to 8 show the recorded, simulated accelerograms and their FAS comparison for Rock sites (class B).
Figures 9 to 12 and Figures 13 to 16 show the results for very dense soil and stiff soil respectively.
1
Simulated Accelerogram
0.8 0.8
Recorded accelerogram
0.6 0.6
0.4 0.4
Normalized Acceleration
0.2
Normalized acceleration
0.2
0
0
-0.2
-0.2
-0.4
-0.4
-0.6
-0.6
-0.8
-0.8
-1
0 5 10 15 20 25
-1 time (s)
0 5 10 15 20 25
time (s)
Figure 1 Recorded accelerogram (Id. A1) - Figure 2 Simulated accelerogram (Id. A1) -
Kalamata earthquake (13/10/97)- Githio-Police Kalamata earthquake (13/10/97)- Githio-Police
Station Station
1
FFT-simulation 10
0.9 FFT-record
1
0.8 Simulated
0.1 Recorded
Normalized Spectral Amplitude
Power Spectral Density
0.7
0.01
0.6
1E-3
0.5
1E-4
0.4
1E-5
0.3
1E-6
0.2
1E-7
0.1
1E-8
0 10 20 30 40 50
0
0 5 10 15 20 25 30 35 40 Frequency (Hz)
Frequency (Hz)
Figure 3 Comparison of spectra for simulated and Figure 4 Comparison of power spectral density
recorded accelerograms (Id. A1) - Kalamata for simulated and recorded accelerograms (Id.
earthquake (13/10/97)- Githio-Police Station A1) - Kalamata earthquake (13/10/97)-
Githio-Police Station
770
1 1
Recorded accelerogram Simulated Accelerogram
0.8 0.8
0.6 0.6
0.4
0.4
0.2
0.2
0
0
-0.2
-0.2
-0.4
-0.4
-0.6
-0.6
-0.8
-0.8
-1
-1 0 5 10 15 20 25 30 35
0 5 10 15 20 25 30 35 time (s)
time (s)
Figure 5 Recorded accelerogram (Id. B4) - Figure 6 Simulated accelerogram (Id. B4) -
Campano Lucano earthquake (23.11.1980)- Campano Lucano earthquake (23.11.1980)-
Auletta Auletta
1
FFT-simulation 10
0.9 FFT-record
1
0.8 Simulated
0.1 Recorded
0.7
0.01
0.6
1E-3
0.5
1E-4
0.4
1E-5
0.3
1E-6
0.2
1E-7
0.1
1E-8
0 0 10 20 30 40 50
0 5 10 15 20 25 30 35 40
Frequency (Hz)
Frequency (Hz)
Figure 7 Comparison of spectra for simulated Figure 8 Comparison of power spectral density
and recorded accelerograms (Id. B4) - for simulated and recorded accelerograms (Id.
Campano Lucano earthquake (23.11.1980)- B4) - Campano Lucano earthquake
Auletta (23.11.1980)- Auletta
0.8 1
0.6 0.8
0.4 0.6
0.2 0.4
0 0.2
-0.2 0
-0.4 -0.2
-0.6 -0.4
-0.8 -0.6
-0.8
-1
0 5 10 15 20 25 30 35 40 45
time (s) -1
0 5 10 15 20 25 30 35 40 45
time (s)
Figure 9 Recorded accelerogram (Id. C1) - Figure 10 Simulated accelerogram (Id. C1) -
KOBE earthquake (16.01.1995)- CUE 99999 KOBE earthquake (16.01.1995)- CUE 99999
Nishi-Akashi Nishi-Akashi
771
1
FFT-simulation 10
0.9 FFT-record 1
0.8 0.1 Simulated

Recorded

0.01
0.7
1E-3
0.6 1E-4
1E-5
0.5
1E-6
0.4 1E-7
1E-8
0.3
1E-9
0.2 1E-10
1E-11
0.1
1E-12
0 10 20 30 40 50
0
0 5 10 15 20 25 30 35 40 Frequency (Hz)
Frequency (Hz)
Figure 11 Comparison of spectra for simulated Figure 12 Comparison of power spectral

and recorded accelerograms (Id. C1) - KOBE density for simulated and recorded
earthquake (16.01.1995) - CUE 99999 accelerograms (Id. C1) - KOBE earthquake
Nishi-Akashi (16.01.1995)- CUE 99999 Nishi-Akashi
1 1
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0 0
-0.2 -0.2
-0.4 -0.4
-0.6 -0.6
-0.8 -0.8
-1 -1
0 5 10 15 20 25 30 35 40 45 0 5 10 15 20 25 30 35 40 45
time (s) time (s)
Figure 13 Recorded accelerogram (Id. D1) - Figure 14 Simulated accelerogram (Id. D1) -
KOBE earthquake (16.01.1995)- CUE 99999 KOBE earthquake (16.01.1995)- CUE 99999
Takatori Takatori
1
FFT-simulation
0.9 FFT-record
100
0.8
10
0.7 1 Simulated
Recorded
0.6 0.1
0.01
0.5
1E-3
0.4
1E-4
0.3 1E-5
0.2 1E-6
1E-7
0.1
1E-8
0
0 2 4 6 8 10 12 14 16 18 20 1E-9
Frequency (Hz)
1E-10
0 10 20 30 40 50
Frequency (Hz)
Figure 15 Comparison of spectra for simulated Figure 16 Comparison of power spectral

and recorded accelerograms (Id. D1) - KOBE density for simulated and recorded
earthquake (16.01.1995) - CUE 99999 accelerograms (Id. D1) - KOBE earthquake
Takatori (16.01.1995)- CUE 99999 Takatori
772
6. DISCUSSION OF RESULTS
It can be observed that the spectra of simulated accelerograms are compared well with those of recorded
acceleration time histories. In the present study, significant frequency band of earthquake is considered as the
frequency range between half of spectral amplitude with respect to peak region frequency. Table 4 gives the
comparison of frequency band for different earthquakes. The frequency band for recorded accelerograms of
Kalamata earthquake (A1) at hard rock site is 0-5 Hz. The corresponding band for the simulated acceleration
time history is 0-5.8 Hz. The frequency band for recorded and simulated accelerograms for Campano Lucano
earthquake (B4) at rock site are 0-5.2 Hz and 0-5.7 Hz respectively. The frequency bands for Kobe earthquake
(C1) at very dense soil site are 0-4.2 Hz and 0-5.5 Hz for recorded and synthetic accelerograms respectively. For
stiff soil site of Kobe earthquake (D1), the band for record and simulation are 0-1.15 Hz and 0-1.4 Hz
respectively. It can be observed that broad band nature of hard rock sites and narrow band feature of soil sites
are simulated well by the semi-analytical model.
Table 4: Comparison of frequency band for different earthquakes
Frequency band (Hz) Frequency band (Hz)

Id. M R (km) Soil class
-Record -Simulation
A1 6.4 61 Hard Rock 0-5 0-5.8
B4 6.9 25 Rock 0-5.52 0-5.7
C1 6.9 20 Very dense soil 0-4.2 0-5.5
D1 6.9 22 Stiff soil 0-1.15 0-1.4
7. CONCLUSION
In the present work, a semi analytical approach is used to simulate the ground motion generated due to
earthquake. The random time histories are generated by inverting Fourier Amplitude Spectrum (FAS) with
random phase angles distributed uniformly. FAS of earthquake ground motion is evaluated using a point source
model. The synthetic time histories are generated by multiplying the random time histories with envelope
function. A nonlinear optimization algorithm is used to obtain the parameters of the envelope function from the
rise time and duration of a given earthquake. Artificial time histories are generated for various earthquakes and
their FAS and PSD are compared with those of acceleration records. A correlation has been developed for kappa
from the strong motion data base of historic earthquakes. The broad band nature of hard rock sites and narrow
band feature of soil sites are simulated well by the semi-analytical model. It is observed that the present
methodology with generated correlation for kappa gives better simulation for strong ground motion. Hence, this
procedure is well suited for generating design basis ground motion for the sites of nuclear power plants with
sparse records.
REFERENCES
[1] Ghosh, A.K. and Rao, K.S. (1990) “Studies on Fourier Amplitude Spectra of accelerograms recorded on rock
sites,” Report BARC-1502, Bhabha Atomic Research Centre, Govt. of India, Bombay.
[2] Atkinson, G.M. and Boore, D.M. (2006) “Earthquake Ground-Motion Prediction Equations for Eastern
North America,” Bulletin of the Seismological Society of America; Vol. 96, No. 6, pp. 2181–2205.
[3] The Uniform Building Code (UBC), (1997).
[4] Pacific Earthquake Engineering Research Center (PEER) Strong Motion Database, http://peer.berkeley.edu.
[5] The European Strong Motion Database (ESD), http://www.isesd.hi.is.
[6] Douglas, J. Gehl, P. Bonilla, L. F. and Gelis, C. (2010) “A k model for mainland France”, Pure and applied
Geophysics; Vol. 167, No. 11, pp. 1303-1315.
773
Paper No.: 140
EXPERIMENTAL AND ANALYTICAL STUDIES OF RATCHETING IN

PRESSURIZED PIPING SYSTEM UNDER SEISMIC LOAD
A. Ravi Kiran1, P. N. Dubey1, M. K. Agrawal1, G. R. Reddy1, R. K. Singh1 and K. K. Vaze2

1
Reactor Safety Division, BARC, Mumbai, India
2
Reactor Design & Development Group, Bhabha Atomic Research Centre, Mumbai, India.
Email: arkiran@barc.gov.in
ABSTRACT :
Experimental and analytical studies have been carried out on pressurized piping system under seismic load with
increasing amplitude. These studies have been carried out to understand progressive strain accumulation known
as ratcheting in piping systems due to large amplitude seismic loads. A three dimensional stainless steel piping
system has been tested under internal pressure and base excitation using shake table. The tested piping system
has been analyzed using Iterative Response Spectrum method to evaluate the strain accumulation. The
experimental and analytical results are presented in the paper.
KEYWORDS: Ratcheting, elbow, Response Spectrum
1. INTRODUCTION
The piping systems are subjected to both static loads such as pressure, temperature and dead weight, and cyclic
loads such as earthquakes. If the maximum load in the cycle is lower than that to cause yield, the response of the
piping system is wholly elastic. If the maximum load exceeds the static load carrying capacity, gross plastic
deformation and plastic collapse will occur. If the maximum load lies between yield and plastic collapse,
accumulation of plastic strain namely ratcheting can occur. Ratcheting leads to incremental plastic collapse in
the pressurized piping systems. In recent years, research has been aimed at improving the seismic design of
piping systems in the nuclear industry. An experimental finding from the tests on shake tables is that very
often the observed failure mode is fatigue-ratcheting. This is a cumulative effect of fatigue and of accumulated
plastic strains. Several experimental and analytical studies are carried out to understand ratcheting phenomenon
in piping systems and components under various loading conditions (Ravikiran et al. (2007, 2013), Suzuki et al.
(2003)). In the literature, two methods are available to predict the inelastic response of the piping systems using
response spectrum as input. First one is incremental hinge technique (Jaquay et al. (1988), Ravikiran et al.
(2008)) and the second one iterative response spectrum method (Singh (2011), Ravikiran et al. (2007)). The
incremental hinge method is based on static collapse and cannot predict the strain accumulation in the piping
system. Hence, iterative response spectrum method which is based on ratcheting is used to analyze the tested
piping system.
2. DESCRIPTION OF PIPING SYSTEM TEST
Shake table test on a three-dimensional piping system is carried out to understand the ratcheting behaviour of
piping systems. The configuration of the piping system used for the test is shown in Fig. 1. The piping system is
made of stainless steel grade SS 304L, Schedule 40. It has 168 mm outer diameter and thickness of 7 mm. The
piping system contains six 900 long radius elbows, one Tee joint and two 250 kg masses. Both ends of the
piping system are anchored to the shake table. Strain gauges are bonded at elbows and anchors. Seismic test
facility at Earthquake Engineering & Vibration Research Centre (EVRC) of Central Power Research Institute
(CPRI), Bangalore is used for the test. The shake table has six degrees of freedom with ten ton payload capacity
in the frequency range of 0.1 to 50 Hz. Fig. 2 shows the finite element model of the piping system. The piping
system is modeled using pipe and elbow elements. Lumped masses are modeled using mass elements. At the
nodes corresponding to the anchor location, all the degrees of freedom are constrained. The piping system is
pressurized with water up to a pressure of 12 MPa to induce hoop stress of 1Sm where Sm is the stress intensity
limit (For SS 304L, Sm =140 MPa). Free vibration characteristics of the piping system are evaluated by sine
sweep test. Further ratcheting test is carried out in which the pressurized piping system is subjected to increasing
seismic load applied in all the three directions.
Elbow-1
Elbow -1 250 kg
Elbow-2 250 kg
250 kg
Anchor-3
250 kg
Elbow -2 Anchor -3
Anchor-2
Anchor -2
Anchor -1 Anchor-1
Fig. 1 Test setup of the piping system Fig. 2 FE model of the piping system
1.1. Free vibration characteristics of piping system
Sine sweep test is carried out at a sweep rate of one octave per minute in the frequency range of 1-50 Hz.
The comparison of natural frequencies of the piping system obtained from experiment and analysis is given
in Table 1.
Table 1: Comparison of natural frequencies of piping system
Natural frequency (Hz)

Analysis on piping Analysis on piping system
Mode no
Sine sweep test system with pipe and with springs replacing
elbow elements critical elbow
1 3.5 3.65 3.63
2 5.5 5.68 5.27
3 15 14.71 13.82
1.2. Seismic Ratcheting Test on the piping system
Seismic ratcheting test is conducted on the pressurized piping system. Test Response Spectra (TRS) in three
directions for 2% damping are shown in Fig. 3 to Fig. 5. Acceleration time history compatible with Test Response
Spectrum (TRS) is applied to the shake table and the response of the piping is recorded. During ratcheting test,
strains at various locations are monitored and strain accumulation is used as the criterion for repeating each
excitation level. If no strain accumulation is observed then the excitation level is increased to next higher level.
Initially the piping system was subjected to base excitation with 1g ZPA applied for five times. In the second
level, the excitation is increased to 1.25g and applied for five times. The details of further excitation are given in
Table 2.
775
Strain time history in hoop direction at crown location of elbow-1 is shown in Fig. 6. During excitation with 2.5g
ZPA, a crack is formed at crown of elbow-1 and water jet has come out. The photograph of water jet and crack at
crown location are shown in Fig. 7 and Fig. 8.
Evaluation of elastic stresses for the piping system under seismic load is carried out to show the exceedance with
respect to the design allowable stress. The evaluation is carried out for the piping system, as per the procedure
given in ASME Boiler and Pressure Vessel Code, Section-III, Div-I, NB-3000 (ASME (2004). Equation (9) of
NB-3600 for level D service condition is given as
PD0 MD
B1 + B2' i 0 ≤ 3S m
2t 2I (1)
For straight pipe, B1= 0.5 and B2' = 1.0
tR
For elbows, B1= -0.1+0.4h and B2' = 1.3/h2/3 where, h =
rm2
The primary stress corresponding to various levels of excitation is given in Table 2. It is observed that the
primary stress intensity has exceeded the design limit of 3Sm at base excitation level corresponding 1.75g ZPA.
5 5
4 4
Acceleration (g)
Acceleration (g)
3 3
2 2
1
1
0
0
0 10 20 30 40 50 60 70 80 90 100
0 10 20 30 40 50 60 70 80 90 100
Frequency (Hz)
Frequency (Hz)
Fig. 3 Test input spectrum in Fig. 4 Test input spectrum in Z-direction
X-direction (horizontal, 1g ZPA) (horizontal, 1g ZPA)
3.5 9000
ZPA2.5g
3.0
ZPA1.75g
2.5 6000 ZPA1.25g
Acceleration (g)
Micro strain
ZPA1.5g
2.0 ZPA1g
1.5 3000
1.0
0
0.5
ZPA2.25g
0.0
0 10 20 30 40 50 60 70 80 90 100 ZPA2g
-3000
Frequency (Hz) 0 250 500 750 1000 1250 1500 1750
Time (s)
Fig. 5 Test input spectrum in Y-direction

Fig. 6 Strain history at elbow-1
(vertical, 0.64g ZPA)
776
Fig. 7 Photograph of water jet through elbow-1 Fig. 8 Photograph of crack at elbow-1
Table 2 Details of various levels of base excitation and primary stress intensity in elbow-1
ZPA Pressure No. of excitation Maximum primary stress
(g) (MPa) time histories Intensity in Sm
1 12 5 1.72
1.25 12 5 2.15
1.5 12 5 2.58
1.75 12 5 3.02
2 12 5 3.45
2.25 12 5 3.88
2.5 12 5 4.31
2.5 14 30 4.31
2.5 16 8 4.31
3. DETAILS OF ANALYSIS USING ITERATIVE RESPONSE SPECTRUM METHOD
The Iterative Response Spectrum (IRS) method is based on envelope moment rotation curve incorporating
ratcheting. Flow Chart for Iterative Response Spectrum method is given in Fig. 9. This technique comprises two
levels of analysis. In the first level, envelope moment rotation curves are generated for the piping components
such as an elbow. These curves are generated from the ratcheting analysis of the pressurized components under
incremental cyclic loading. In the second level of analysis, the whole piping system is modeled using pipe and
elbow elements. Response spectrum analysis of the system is carried out to identify the critical components,
which are replaced with equivalent spring models. These spring models comprise of three translational and
three rotational springs corresponding to each direction. The spring constants are modified from the envelope
ratcheting moment rotation curves. The modification is done in an iterative way such that the resulting moment
rotation response of component from response spectrum analysis matches with the envelope characteristics
obtained from first level of analysis. The details of its application to the tested piping system are given below.
3.1. Level 1: Generation of cyclic strain-moment-rotation curve for the elbow
Ratcheting analysis is carried out on 900 long radius elbow of the piping system. Fig. 10 shows the finite
element model of the elbow which is modeled using shell elements. The ends are attached with straight
pipes of length more than three times diameter.
777
Create FE model of elbow using shell elements.
Perform monotonic analysis on elbow to obtain its
stiffness in all directions. Perform nonlinear cyclic
analysis on elbow to obtain its ratcheting
characteristics and generate envelope cyclic
moment rotation curve
Create FE model of the piping system using pipe

and elbow elements and carry out response
spectrum analysis. Identify the critical component
and replace it with spring model
Carry out response spectrum analysis

and compare the resulting moment
and rotation with cyclic envelope
curve
Is error
between elbow moment No
and cyclic moment
< 1%
Increase level
of excitation Modify rotational
Yes
stiffness of elbow
Evaluate strain corresponds to

moment from the component
results
Is
Yes strain
accumulation
< 5%
Level of excitation corresponding to 5% strain is

the failure level of the piping system
Fig. 9 Flow Chart for Iterative Response Spectrum method
A constant internal pressure of 12 MPa which corresponds to the design stress of 1Sm is applied to the FE model.
Nonlinear static analysis is carried out to estimate yield displacement which forms the basis for ratcheting
analysis. This analysis is carried out by incrementally applying load at the tip of the elbow. The resulting
load-displacement curve is plotted in Fig. 11. The displacement corresponding to yield is calculated by using
intersecting tangent method. The point of intersection of two tangents to elastic and plastic portion of curve is
taken as yield and the yield displacement (δy) is 30 mm. Based on this yield displacement, ratcheting analysis
using Chaboche model is carried out with 10 cycles of different amplitudes with 10% increment. The Chaboche
parameters C1= 1085000, C2=43000, C3=4100, γ1= 280750, γ2=800, γ3=9 are considered for the analysis of
stainless steel (Kulkarni et al. (2004)). Cyclic strain-moment-rotation curve is shown in Fig. 11. The hoop strain
at crown location of elbow is shown in the plot.
778
3.1. Level 2: Iterative Response Spectrum method for piping system
Iterative Response Spectrum analysis is carried out on the piping system with spring model shown in Fig.
12. The details of the spring model in the elbow local coordinate system are shown in Fig. 13. It contains
three translational and three rotational springs. The three translational springs in xl, yl and zl directions are
denoted as kTx, kTy and kTz respectively. Corresponding rotational springs in xl, yl and zl directions are
kRx, kRy and kRz respectively. The spring constants are evaluated numerically from the static analysis on
the elbow. Base excitation with 1g ZPA is applied to the piping model. The resulting in plane moment
(Melb) and rotation (θelb) for elbow-1 are 13.2 kN-m and 0.71o respectively. The envelope moment
corresponds to this rotation is 12.63kN-m and the error is 5.3%. The in plane rotational stiffness of the
elbow is modified to 0.99 MN-m/rad and second iteration of response spectrum analysis is carried out with
1g ZPA. The resulting moment and envelope moment are 13.25 and 13.22 kN-m respectively
corresponding to rotation of 0.74o and the error is 0.23%. Hence, the solution is converged. The strain
corresponding to this moment is 2133 µε, which is obtained from Fig. 11. Further iterative analysis is
carried out for the excitation corresponds to ZPA of 1.25g to 2.5g with an interval of 0.25g. Results of
Iterative Response Spectrum analysis at various stages are given in Table 3. Various levels of this analysis
are plotted on envelope moment rotation curve in Fig. 14. Comparison of predicted strain accumulation
with test results is shown in Fig. 15. The ratcheting strain predicted by Iterative Response Spectrum
method is closely matching with experimental results. Hence, it can be concluded that the Iterative
Response Spectrum method which is based on ratcheting is an improved tool for seismic response
prediction of piping systems.
60000 10
Moment-rotation curve
Hoop strain (Microstrain)
50000 Moment-strain curve

8
Rotation (degrees)
40000
6
30000
4
20000
2
10000
0 0
0 10 20 30 40 50
Moment (kN-m)
Fig. 10 Finite element model of the elbow Fig. 11 Cyclic strain-moment-rotation curves
Rigid link
Node
kTz
kRy
kTy
yl kRz
zl kRx
xl
kTx
Fig. 12 FE model of the piping system Fig. 13 Schematic of spring model

by replacing elbow with springs
779
Table 3 Results of Iterative Response Spectrum analysis
ZPA Iter. kRz Melb θelb Melbc Error

εelb
(g) no. (MN-m/rad) (kN-m) (deg.) (kN-m) (%)
1 1 1.04 13.2 0.71 12.63 5.30
2133
2 0.99 13.25 0.74 13.22 0.23
1.25 1 0.99 16.56 0.93 16.41 0.91 2575
1 0.99 19.5 1.09 18.52 5.29
1.5 3093
2 0.94 19.5 1.15 19.40 0.52
1 0.94 22.9 1.35 21.75 5.29
1.75 2 0.89 22.8 1.42 22.35 2.01 3866
3 0.87 22.78 1.45 22.64 0.62
1 0.87 26.07 1.65 24.54 6.23
2 5080
2 0.82 25.94 1.75 25.70 0.93
1 0.82 28.4 1.92 26.70 6.37
2.25 2 0.77 25.68 1.85 26.10 1.61 5460
3 0.78 26.39 1.87 26.43 0.15
1 0.78 31.96 2.26 29.27 9.19
2.5 6735
2 0.72 28.64 2.21 28.90 0.90
40 8000
Experiment
Iterative response spectrum method
30 6000
Strain in elbow (%)
Moment (kN-m)
2g ZPA
2.5g ZPA
2.25g ZPA
20 1.75g ZPA 4000

1.5g ZPA
1.25g ZPA
2000
10 1g ZPA
0
0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
0 1 2 3 4 5
ZPA of response spectrum (g)
Rotation (degrees)
Fig. 14 Various level of IRS analysis Fig. 15 Comparison of predicted

strain accumulation with test results
4. CONCLUSION
Experimental and analytical studies have been carried out on pressurized piping system under seismic load with
increasing amplitude. A simplified analytical procedure known as Iterative Response Spectrum method is used
for predicting the ratcheting response of a piping system. Two levels of analysis are carried out. In the first level,
envelope moment rotation curve is generated for the elbow from the ratcheting analysis under incremental
cyclic loading. In the second level of analysis, the whole piping system is modeled using pipe and elbow
elements and response spectrum analysis is carried out. The critical elbow is identified and replaced with
equivalent spring model whose constants are obtained from static analysis of the elbow. The spring constants
are modified in an iterative way such that the resulting moment rotation response of the elbow from the response
spectrum analysis matches with the ratcheting envelope characteristics obtained from the first level of analysis.
This procedure is repeated for various levels of base excitation. The strain predicted by this analysis closely
matches with that of experimental results. Hence, it can be concluded that the Iterative Response Spectrum
method which is based on ratcheting is an improved tool for seismic response prediction of piping systems.
780
5. NOMENCLATURE
D Diameter of the pipe

h Elbow parameter
I Moment of Inertia
M Moment
P Internal pressure
t Wall thickness of pipe or elbow
R Bend radius of elbow
Sm Allowable stress intensity
B1, B2' Primary Stress Indices
Melb Moment in elbow
Melbc Envelope moment for elbow
Ci , γ i Chaboche parameters
θelb Rotation in elbow
REFERENCES
[1] Ravikiran, A., Dubey, P. N., Agrawal, M. K., Reddy, G. R. and Vaze, K. K. (2013). “Evaluation of
Inelastic Seismic Response of Piping System using Modified Iterative Response Spectrum Method”,
International Journal of Pressure Vessel Technology, In Press.
[2] Ravikiran, A., Agrawal, M. K., Reddy, G. R., Singh, R. K., Vaze, K. K., Ghosh, A. K., Kushwaha, H. S.
(2007). “Fatigue-Ratcheting Study of Pressurized Piping System under Seismic Load”, Transactions,
SMiRT 19, Toronto, Paper # K12/4.
[3] Suzuki, K., Namita, Y., Abe, H., Ichihashi, I., Suzuki, K., Sakakida, T., Sato, T. and Yokota, H. (2003).
“Seismic Proving Test of Ultimate Piping Strength (Status of Design Method Confirmation Test)”,
Transactions, SMiRT 17, Prague, Paper # K15/2.
[4] Jaquay, K. R., Larson, J. E. and Tang, H. T. (1988). “A Simplified Method for Inelastic Piping System
Seismic Response Prediction”, Nuclear Engineering and Design, 107, 169-181.
[5] Ravikiran, A., Agrawal, M. K., Reddy, G. R., Ramesh Babu, R., Singh, R. K., Vaze, K. K., Ghosh, A. K.,
Kushwaha, H. S. (2008). “Analysis of a Piping system under seismic load using Incremental Hinge
Technique”, Proceedings of ICONE 16, Florids, USA.
[6] Singh, G. (2011), “Development of Incremental Hinge Technique to predict the behavior of piping system
subjected to internal pressure and seismic loading”, M. Tech dissertation, Homi Bhabha National Institute,
Mumbai.
[7] The American Society of Mechanical Engineers. (2004), “ASME Boiler and Pressure Vessel Code Section
III, Division 1”.
[8] Kulkarni, S. C., Desai, Y. M., Kant, T., Reddy, G. R., Prasad, P., Vaze, K. K., Gupta, C. (2004), “Uniaxial
and biaxial ratchetting in piping materials—experiments and analysis”, International Journal of Pressure
Vessels and Piping, 81, 609–617.
781
Paper No.: 141
NUMERICAL ANALYSIS OF SEMI-RIGID JOINTED STEEL FRAME USING

ROTATIONAL SPRINGS
1 2
M. Bandyopadhyay and A. K. Banik
1
PhD Scholar, Dept. of Civil Engineering, National Institute of Technology, Durgapur, India
2
Associate Professor, Dept. of Civil Engineering, National Institute of Technology, Durgapur, India
Email: milanbandyopadhyay@yahoo.co.in, akbanik@gmail.com
ABSTRACT:
This paper investigates the effect of semi-rigid connection and geometric nonlinearity in steel frame analysis. A
general purpose software ANSYS is used for the numerical modeling. Connections are modelled as rotational
springs and frame as one dimensional beam element. Two single storey two dimensional portal frames are
chosen from published literature to validate the analysis results. Effect of beam to column connection flexibility,
support flexibility and geometric nonlinearity are considered. Key features about modeling and necessary input
parameters are discussed in detail. Rotational springs used herein are characterized by linear, bilinear,
multi-linear or nonlinear moment-rotation relationships. For nonlinear moment-rotation relationship four
parameter power model is used. Finally a one bay two storey frame earlier analysed for rigid connections in
referred literature is analysed with different semi-rigid connections. Results in terms of beam moment and nodal
displacement are reported. In both rigid and semi-rigid jointed frames, moments and displacements in general
increase with geometric nonlinearity. Whereas, in both linear and nonlinear analysis beam mid span moments
and displacements increase, while beam end span moments decrease with connection flexibility.
KEYWORDS: Semi-rigid, connections, steel frame, spring, numerical, elastic, nonlinear
1. INTRODUCTION
Investigation on steel beam-to-column connections started a century ago with investigation of the flexibility in
riveted joints in. It was followed by application of slope deflection, moment-rotation, beam-linemethod (Jones
et.al., 1983) till 1950. Computer application facilitated use of matrix method for analysis of large semi-rigid
jointed frames.Several experimental studies were carried out to observe the behaviour of semi-rigid connections.
Data were stored in the form of initial stiffness, initial/ ultimate moment, and geometric properties of
connections. Thus different data banks say Goverdhan data bank, Nethercot data bank, Kishi and Chen data
bank are created. Liew et. al. (1997) conducted some benchmark study by physical testing of several semi-rigid
jointed frames. Ashraf et al. (2004) investigated the sway behaviour of semi-rigid regular steel frames using
ANSYS. Some design charts were developed for prediction of sway of semi-rigid frame if the corresponding
sway of rigid frame is known. Later Ashraf et al. (2007) extended study for semi-rigid irregular steel frames.
The behaviour of semi-rigid connection is expressed in terms of its moment-rotation relationship. In reality the
relationship is always nonlinear but sometimes idealized as linear, bi-linear, multilinear (Diaz et. al. 2011).
Some notable nonlinear models are Frye and Morris polynomials model, Richard and Abott four parameter
power model, Kishi and Chen three parameter power model and Ramberg-Osgood model. The test data are very
useful in development and validation of these connections models. Spring is characterised by moment-rotation
relationship predicted by using these connection models. In the current study Richard and Abott (1975) four
parameter power model is used.
The effect of semi-rigid connection may be considered in numerical analysis of frame by three ways, (i)
modeling of connection as rotational spring and frame as one dimensional beam element (Yan, 2007), (ii)
modeling of connections as three dimensional solid element and frame as one dimensional beam element
(Suparb, 2005) and (iii) modeling of entire frame as three dimensional solid element. In general the different
aspects of a connection are more accurately and efficiently presented by finite element analysis but its use is
limited as it is computationally expensive. This may be the reason why representation of semi-rigid connections
by rotational springs has gained popularity. For analysis of semi-rigid jointed frames, researchers have suitably
modified the conventional analysis methods by replacing connection rigidity with spring stiffness. Some
researchers developed their own computer programs to analyze large frame with semi-rigid connections. But
these computer programs may not be easily available for wide use.
With this background the numerical modeling of rotational springs has been developed to represent the
semi-rigid connections using general purpose software ANSYS. Effect of connection flexibility, support
flexibility and geometric nonlinearity are considered under static loading condition. The Numerical study has
been applied on two single storey two dimensional portal frames available in the published literature to validate
the analysis results. The step-by-step modeling procedure and necessary input parameters are discussed in detail.
Numerical model so developed is also used to analyse a one bay two storey two dimensional portal frame to
study the effect of semi-rigid connections. Results for different beam to column connections for both linear and
nonlinear analysis is compared and presented.
Objective of the present study may be stated as (i) Elastic linear analysis of a two dimensional one bay one story
frame with rigid and semi-rigid connection, (ii) Elastic nonlinear analysis of a two dimensional three bay one
story frame with rigid connection and semi-rigid connections having linear and bilinear moment-rotation
relationship and (iii) Elastic linear and nonlinear analysis of a two dimensional one bay two story frame with
five different semi-rigid connections.
2. ELASTIC LINEAR OR FIRST ORDER ANALYSIS
The elastic linear analysis is performed for normal frames when displacements are very small as compared to
the dimensions of the frame. Both geometric and material nonlinearity are not considered in this method of
analysis. Input parameters required are modulus of elasticity (E) and Poison’s ratio (µ) as material property and
moment of inertia for and cross sectional area (optional) as member property. A two dimensional one bay one
story steel portal frame earlier analysed by Chan and Chui (2000) and Ihaddoudene et. al. (2009) is considered
for validation of current study. Modulus of elasticity (E) and Poison’s ratio (µ) for steel is taken as 2 x 105 MPa
and 0.3 respectively. Geometry of the frame, and member properties are as shown in Table 1. Beam and
columns are modelled as BEAM3 element in ANSYS. Analysis results for rigid jointed frame are shown in
Table 2 and these are in agreement with referred literature.
Ihaddoudene et. al. (2009) adopted mechanical model for the analysis of the frame with semi-rigid joints, where
member end flexibility is considered using rotational springs. They have established modified stiffness matrix
considering flexible connections. Similarly, in ANSYS connection flexibility may be modeled using springs. A
zero length COMBIN39 element is used to simulate the spring. Frame detail is shown in Figure 1(a). To create
Table 1 Geometry, loading and resultof rigid frame chosen for linear analysis
Rigid Connections
Ihaddoudene Present
Quantity Units
et. al. Study
(2009)
M41 kN-m 52.2 52.23

M14 kN-m 127.6 127.5
M52 kN-m 87.1 87.14
M25 kN-m 152.7 152.6
M32 kN-m 260 260
783
the model only five nodes (1 to 5) and four members are required. Two additional nodes 6 and 7 are created at
the location of 1 and 2 respectively (Figure 1b). COMBIN39 element is created between the coincident nodes i.e.
between 1 and 6, 2 and 7. Only in plane rotation is allowed for the springs. Hence, degree of freedom for these
COMBIN39 elementsis ROTZ only and it is necessary to couple UX and UY at 1 and 6, 2 and 7. Spring
constant calculated as kb = 4EI/lb = 10.75 x 103 kN-m/rad. This stiffness is entered as real constant for the spring
element. Now the model has seven nodes, four one dimensional BEAM3 element and two COMBIN39 element.
Support condition is fixed. Results of the analysis are provided in Table 2.
(a) (b)
Figure 1(a) Frame geometry adopted from Ihaddoudene et. al. (2009)
(b) Node and Elements numbering for ANSYS model
The frame is then analysed with both semi-rigid connections and elastic support. Frame details are shown in
Figure 2(a) and 2(b). In this case two more coincident nodes are created for modeling of elastic support.
Rotational springs are created between 1 and 6, 2 and 7 for semi-rigid connections. Whereas rotational springs
created between 4 and 8, 5 and 9 act as flexible supports. The model has nine nodes, four one dimensional
BEAM3 element and four COMBIN39 element. Nodes 1 and 6, 2 and 7 are coupled for UX and UY but at
elastic supports coupling is not applicable. Nodes 4 and 5 are restrained against UX and UY whereas nodes 8
and 9 are restrained against ROTZ. Spring constant calculated as kb = 4EI/lb = 10.75 x 103 kN-m/rad and kc =
4EI/lc = 3.1533 x 103 kN-m/rad.
(a) (b)
Figure 2(a) Frame geometry adopted from Ihaddoudene et. al. (2009)
(b) Node numbering for ANSYS model
Results of the analysis are provided in Table 2. The results are in agreement with those reported in referred
literature. As compared to rigid frame for semi-rigid jointed frame in both fixed and elastic support condition
beam mid span moment (M32) increase and beam end moments (M14 and M25) decrease. As expected support
784
flexibility drastically decreases moment carrying capacity of column at base (M41 and M52).
Table 2 Results of linear analysis for semi-rigid frame

Quantity Units Semi-rigid connections at beam Semi-rigid connections at beam
column joints only. column joints and at base.
kb = 4EIb/lb, kc = 0 kb = 4EIb/lb, kc = EIc/lc
Chan, Ihaddou- Present Chan, Ihaddou- Present
Chui dene et. al. Study Chui dene et. al. Study
(2000) (2009) (2000) (2009)
M41 kN-m 31.7 31.9 31.68 0.3 0.3392 0.32
M14 kN-m 93.6 93.7 93.65 80.3 80.27 80.25
M52 kN-m 71.5 71.8 71.53 24.2 24.1813 24.16
M25 kN-m 113.8 113.9 113.8 116.4 116.4294 116.41
M32 kN-m 296.3 296.4 296.3 301.7 301.6495 301.67
3. ELASTIC NONLINEAR OR SECOND ORDER ANALYSIS
The second order elastic analysis is an extension of first order analysis considering effect of geometric
nonlinearity. Material nonlinearity is not considered as yielding effects are ignored in this analysis. In ANSYS
large displacement option is selected to consider geometric nonlinearity. A two dimensional three bay one story
steel portal frame earlier analysed by Ihaddoudene et. al. (2009) is considered for validation of current study.
Geometry of the frame is shown in Figure 3. Magnitude of uniformly distributed load W is 35 kN/m. Flexural
rigidity per unit length for beams and columns are taken as 15067 kN-m. Thus, moment of inertia for columns
and beams are calculated as 3.767 x 10-4 m4 and 7.533 x 10-4 m4 respectively. Analysis is done in 100 sub-steps.
Modeling technique is similar to that discussed for earlier problems. Springs with linear and bilinear
moment-rotation relationship is used (Figure 5). Typical bending moment diagram is shown in Figure 6 and
corresponding results are given in Table 3.
Figure 3 Frame geometry adopted from Ihaddoudene et. al. (2009)
Figure 4 Node numbering for ANSYS model
785
Figure 5 Linear and Bilinear moment-rotation Figure 6 Bending moment diagram
Results reported in Table 3 for current study are very close to that obtained by Ihaddoudene et. al. While
validating the results of present study, a few variations are seen in respect of MB for rigid connection, MA for
linear and bilinear cases. The value of MB is obtained as 219.04 kN-m as against 118.75 kN-m reported by
Ihaddoudene et. al. Similarly, magnitude of MA for linear and bilinear cases are found as 118.44 kN-m and
121.62 kN-m as against 121.68 kN-m and 118.44 kN-m reported in the literature. As compared to rigid frame
for semi-rigid jointed frame with linear moment-rotation relationship beam mid span moments (MB and MF)
increases and beam end span moments (MA, MC and MD) decrease. It is seen from Figure 5 that for moment
greater than 135.6 kN-m, the rigidity of connection with bilinear moment-rotation relationship is less than that
of connection with linear moment-rotation relation. This results in further increase in mid span moments for
bilinear case as compared to linear one.
Table 3 Results of nonlinear analysis for semi-rigid frame

Quantity Units Rigid Connections Semi-rigid connections Semi-rigid connections
Linear Bi-linear
Ihaddou- Present Ihaddou- Present Ihaddou- Present
dene et. al. Study dene et. al. Study dene et. al. Study
(2009) (2009) (2009)
MA kN-m 148.84 146.10 121.68 118.44 118.44 121.62
MB kN-m 118.75 219.04 296.06 296.50 304.19 304.6
MC kN-m 296.65 291.23 164.36 164.17 144.85 144.81
MD kN-m 194.25 194.31 107.38 107.36 102.33 102.33
ME kN-m 48.61 48.46 14.21 14.24 10.60 10.67
MF kN-m 25.10 24.44 111.37 111.42 116.41 116.47
4. ANALYSIS OF RIGID AND SEMI-RIGID JOINTED FRAME
For linear and nonlinear analysis discussed so far, two separate frames were considered from published
literature. Finally based upon the numerical modeling features learned so far a two dimensional one bay two
storey frame is analysed. Member property, loading and geometry of the frame is shown in Figure 7. Material
property for steelis considered as E = 2.05 x 105 MPa and µ =0.3. The frame is analysed by N. Subramanian
(2012) for rigid joints. Objective of this study is to observe comparative behaviour of a frame with different
semi-rigid connections in both linear and nonlinear methods of analysis. Semi-rigid connections are modeled in
node 2, 3, 7 and 8. Five different semi-rigid connections are chosen. First two of these denoted as C1 and C2,
are the linearand bilinear moment-rotation relationship used for previous problem and shown in Figure 5.
Another three are nonlinear and denoted as C3, C4 and C5. These are adopted from Liu et. al. (2008). C3, C4
and C5 are for extended end plate (EEP), top and seat angle with double web angle (T&S DWA) and double
web angle (DWA) connections. Classification of connections given in Indian steel design code (IS 800: 2007) is a
function of plastic moment capacity of connected member and plastic rotation. But in more general terms, IS 800:
2007 refers EEP, T&S DWA and DWA as rigid, semi-rigid and flexible type connection respectively. Parameters for
these connections are given in Table 4. Richard and Abott (1975) four parameter power model given by
786
Equation 1 is used for generation of moment-rotation curves for C3 to C5. Moment-rotation curves are plotted
in Figure 8.
( )
= + (1)
Where, is initial/elastic rotational stiffness, is strain hardening/softening stiffness and is reference

plastic rotation given by /( ), being reference relative moment.
Table 4 Parameter for specified connection taken from published research paper
Connection parameter Unit C3 (EEP) C4 C5
(T&S DWA) (DWA)
Reference Moment of connection M0 kN-m 137.86 148 55.935
Initial/elastic rotational stiffness Rki kN-m/ rad 18984 24721 20114
Strain Hardening/Softening stiffness Rkp kN-m/ rad 1041.86 151 69.608
Shape parameter of connections n 5.11 0.78 0.964
Figure 7 Frame geometry adopted from Subramanian (2012)
Bilinear-C2 Linear-C1 EEP-C3 T&S DWA-C4 DWA-C5

260
240
220
200
Moment M (kN-m)
180
160
140
120
100
80
60
40
20
0
0 0.01 0.02 0.03 0.04 0.05
Rotation (rad)
Figure 8 Moment-rotation relationship of the connections
787
Results of the analysis are given in Table 5 and Table 6. Moments and displacement for rigid frame are in
agreement with those reported by Subramanian (2012). Connection rigidity is gradually reduced from C1 to C5.
Table 5 Results of linear analysis for rigid and semi-rigid frame

First Order Elastic Analysis
Subra- Present Study
Quantity Units Manian Semi-rigid Connections
Rigid
(2012) C1 C2 C3 C4 C5
M4 kN-m 143.93 143.5 150.58 156.37 168.02 198.07 237.14
M5 kN-m 67.79 67.53 68.86 69.06 70.88 74.14 104.05
M76 kN-m 123.24 123.03 118.66 114.78 107.45 88.06 71.53
M74 kN-m 180.89 181.36 172.86 165.41 150.89 113.88 60.32
M78 kN-m 57.65 58.33 54.19 50.63 43.44 25.82 11.22
M8 kN-m 67.39 67.61 67.71 69.17 70.21 73.91 58.49
V4 mm 41.64 41.36 45.91 49.64 57.51 76.51 101.68
V5 mm 71.54 70.88 74.28 74.78 79.44 87.74 164.06
U2 mm 70.68 69.82 78.54 86.67 104.04 151.57 387.56
U3 mm 104.37 103.03 118.03 131.48 161.37 244.66 692.00
Behaviour of C1 is close to rigid whereas C5 is close to pinned. Connections C2 to C4 fall among these two. In
first order analysis for a particular member increase in beam mid span moment is almost equal to the decrease in
end span moment. It may be seen in beam member 247 that M4 for C1 to C5 increases by 5% to 65% as
compared to rigid connections. Similarly M74 decreases by 5% to 67%. Vertical and horizontal displacements for
C5 connections are nearly 2.5 times and 6 times as compared to rigid connections.
Table 6 Results of nonlinear analysis for rigid and semi-rigid frame

Second Order Elastic Analysis
Subra- Present Study
Quantity Units Manian Semi-rigid Connections
(2012) Rigid
C1 C2 C3 C4 C5
M4 kN-m 143.93 143.53 150.61 158.03 174.10 209.13 159.58
M5 kN-m 67.79 67.85 69.22 69.47 71.48 75.90 69.98
M76 kN-m 132.72 130.75 127.63 124.01 117.12 103.42 75.64
M74 kN-m 190.62 189.18 181.86 173.76 156.82 121.43 58.32
M78 kN-m 57.87 58.43 54.23 49.757 39.69 18.01 16.81
M8 kN-m 68.63 68.74 69.24 71.355 74.56 88.68 56.06
V4 mm 41.64 41.95 49.69 51.69 62.47 87.90 80.81
V5 mm 71.54 71.88 75.58 76.49 82.50 97.33 129.20
U2 mm 70.68 76.78 87.64 99.50 129.34 206.18 352.73
U3 mm 104.37 112.46 130.74 149.99 199.89 331.55 615.74
In second order analysis moments and displacement increases for both rigid and semi-rigid jointed frames in
comparison to first order analysis. This is because of the successive stiffness degradation caused by geometric
nonlinearity of the frame. Effect of connections on moments and displacement are in line with linear analysis.
Time history of horizontal displacement at node 3 is plotted in Figure 9. Solution for C5 connection doesn’t
converge as the frame become unstable at nearly 67% loading. Moments and displacement is reported in Table 6
are corresponding to this 67% loading.
788
1
0.9
0.8
NL-Rigid
0.7
NL-C1
0.6 NL-C2
Time (Sec)
NL-C3
0.5
NL-C4
0.4 NL-C5
Lin-Rigid
0.3 Lin-C1
0.2 Lin-C2
Lin-C3
0.1 Lin-C4
0 Lin-C5
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80
Horizontal displacement at Node 3 (m)
Figure 9 Horizontal displacement at Node 3 for linear (Lin) and nonlinear (NL) analysis
5. CONCLUSION
Numerical model of two dimensional steel frame has been developed using general purpose FEM based
software ANSYS to study the effect of semi-rigid connections and geometric nonlinearity on steel frame. From
the present study it is established that numerical modeling can be easily used to evaluate the above said
behaviour. The results obtained in the present study are shown to have matched with the results reported in
publishedliterature.In first order analysis loss of stiffness is observed for semi-rigid connections, whereas in
second order analysis loss of stiffness is due to combined effect of semi-rigid connections and geometric
nonlinearity. In current study, only elastic analysis under static loading condition is performed. Study may be
extended considering material nonlinearity. Loading condition may also be considered as dynamic or cyclic.
Only in plane moment-rotation behaviour is considered here. Out of plane effect especially the twisting effect
(Jones et. al., 1983) may also be considered as future scope of study. As test results for out of plane behaviour of
connections are rarely available, numerical modeling may be an efficient tool for this study.
REFERENCES
[1] ANSYS. Basic analysis procedure guide, elements reference, commands reference and structural analysis
guide, version 11. ANSYS, Inc., Global Headquarters, Southpointe, 275 Technology Drive, Canonsburg,
PA 15317.
[2] Ashraf, M.; Nethercot, DA. and Ahmed, B. (2004). Sway of semi-rigid steel frames, part 1: Regular frames,
Engineering Structures, 26, 1809-19.
[3] Ashraf, M.; Nethercot, DA. and Ahmed, B. (2007). Sway of semi-rigid steel frames, part 2: Irregular
frames, Engineering Structures, 29, 1854-63.
[4] Chan, SL. and Chui PPT. (2000) Non-linear static and cyclic analysis of steel frames with semi rigid
connections. Elsevier Science Ltd. 336.
[5] Diaz, C.; Marti, P., Victoria, M. and Querin, O. M. (2011). Review on the modeling of joint behaviour in
789
steel frames, Journal of Constructional Steel Research, 67, 741-758.
[6] Ihaddoudene, ANT.; Saidani M and Chemrouk M. (2009). Mechanical model for the analysis of steel
frames with semi rigid joints. Journal of Constructional Steel Research, 65(3),631–40.
[7] Indian steel design standard, IS 800 (2007). General Construction in Steel - Code of Practice. Bureau of
[8] Jones, S.W.; Kirby P.A. and Nethercort D.A. (1983).The analysis of frames with semirigid connections-A
state-of-the-art report, Journal of Constructional Steel Research, 3(2), 2–13.
[9] Liew, JYW.; Yu, CH.; Ng, YH. and Shanmugam, NE. (1997).Testing of semi-rigid unbraced frame for
calibration of second-order inelastic analysis. Journal of Constructional Steel Research,41(2–3), 159–95.
[10] Liu, Y.; Xu, L. & Grierson, D. (2008). Compound-element modeling accounting for semi-rigid connections
and member plasticity. Engineering Structures, 30(5), 1292-1307.
[11] Richard, R. M. and Abbott, B. J. (1975). Versatile elastic-plastic stress-strain formula. Journal of the
Engineering Mechanics Division, ASCE, 101(4), 511–15.
[12] Subramanian, N. (2012). “Design of Steel Structures”, Oxford, University Press, India.
[13] Suparb, S. (2005). Behaviour and inelastic analysis of steel frames with semi-rigid connections. PhD Thesis,
Department of Civil Engineering. Graduate School, Kasetsart University,
[14] Yan, L. P. (2007). Non-linear behaviour of one-bay steel frameswith semi-rigid connections. Masters Thesis.
Civil Engineering, Universiti Teknologi Malaysia.
790
Paper No.: 142
PASSIVE CONTROL OF STRUCTURES SUBJECTED TO EARTHQUAKE

EXCITATION USING PARTICLE SWARM OPTIMIZATION
Kotla Shiva Kumar1 and Diptesh Das2

1
Assistant Executive Engineer, Dept. of I&CAD, Andhra Pradesh
2
Assistant Professor, Dept. Of Civil Engineering, NIT Durgapur
Email: kotlashiva124@gmail.com, Dipteshdas@yahoo.co.in
ABSTRACT:
Tuned mass dampers (TMD) have been widely used to attenuate undesirable vibrations in structural
engineering. In the present paper, the effectiveness of passive TMDs in controlling vibrations induced in
structures subjected to earthquake excitations is studied. Two example ten-storey shear buildings, with different
natural frequencies are considered and the problems are solved in state space using MATLAB. Optimal
parameters of the TMDs are first designed following the conventional method of tuning and time history
analysis with actual earthquake ground motions are performed to study the effectiveness of the TMDs in
reducing different response quantities of the structures. A parametric study is carried out to demonstrate the
effect of TMD mass on structural response reduction. Most optimization problems of TMD are solved by either
numerical iteration technique or conventional mathematical methods that require substantial gradient
information. In the present study, PSO is applied to search optimal parameters of the TMD that achieve
maximum response suppression for the building. Apart from its inherent simplicity and desirable convergence
behaviour, the main advantage of PSO algorithm is that it usually provides a globally optimum solution, which
means that it will find global minima without falling in the trapping of local minima. Sum of maximum absolute
relative displacements of all floors in the building are taken as performance index for the optimization criteria in
PSO algorithm. The results indicate that the TMDs designed using both the conventional method and the PSO
algorithm are able reduce the different response quantities of two structures significantly. However, the TMDs
designed using PSO are found to be more effective in controlling the responses of the structures subjected to
earthquake ground excitation when compared with those designed by the conventional method.
KEY WORDS: Earthquake; Tuned Mass Damper; Particle Swarm Optimization.
1. INTRODUCTION
Over the past three decades,a great interest has been generated by the use of protection systems to mitigate the
effects of dynamic environmental hazards on civil engineering structures such as earthquakes and strong wind.
In recent years,it has been paid a considerable attention to new concepts of structural control including a large
variety of techniques, that can be defined infour classes:passive,active,hybrid andsemi-active.Passive control
techniques,such that base isolation and passive control devices,are the first of them implemented from historical
point of view. Passive tuned mass damper is a viscous spring-mass unit, when attached to a vibrating main
structure, provides a frequency dependant hysteresis that increases the damping in the structure. The efficiency
of TMD for controlling structural response is sensitive to its parameters i.e. mass, frequency, and damping ratio.
TMD acts as a secondary vibrating system when connected to primary vibrating system. While, the basic
principles of TMDs on reducing structural response have been well established, optimal TMD configurations
are a quite a different problem. In the design of any control device for the suppression of undesirable vibrations,
the aim would be to provide optimal damper parameters to maximize its effectiveness. The chief design
response oriented parameters of the TMD are its tuning ratio (the ratio of the damper frequency to the natural
frequency of the structure) and damping ratio. The other important design parameter is the mass ratio (the ratio
of the damper mass to the mass of the structure) generally considered between 0.01to0.10.
In this paper, study includes the finding of the best optimum TMD parameters by conventional method (tuning
to fundamental frequency of the structure) and also using numerical technique such as PSO to design and
evaluate the effectiveness of tuned mass damper (TMD) for response control of lumped mass model of two
multi-storey shear buildings, with different natural frequencies under seismic excitations. A parametric study is
carried out to demonstrate the effect of TMD mass on structural response reduction.
2. STRUCTURAL MODEL
The equations of motion of the structural system under earthquake loading can be written in the form of Eq.1.
Mx + Cx t + x t) =E . (1)
Where M, C, and K are mass, damping, and stiffness matrices, respectively. E is the influence vector (or matrix)
due to the earthquake acceleration and is the ground acceleration respectively. X is the relative displacement
with respect to the ground, M is the mass matrix for a high rise building structure, with the assumption of
masses lumped at floor levels, is a diagonal matrix in which the mass of each storey is sorted on its diagonal, as
given in the following Eq.2.
M=diag (m , m , m … … m , (2)
Where, is mass of ith floor (i=1, 2… n) in building, is the damper mass. The structural stiffness matrix K
based on the individual stiffness, of each floor,
C=a M + a K, a = 2ε ω , a = 2ε /ω
Where, a and a are the proportional coefficients, ε is the damping of the structure and ω is the frequency of
structure of n#$ mode.
3. CONVENTIONAL METHOD OF TMD DESIGN
The natural frequency of a TMD is tuned to a first natural frequency for the first mode of the main system
should be controlled. Thus, there are three main parameters in a TMD system, mass, stiffness and damping.
Consequently, the objective is to find the optimum value of these parameters such that responses of the structure
get reduced to maximum extent. To find the optimum parameters, mass of TMD is considered from a range of
1-10%.Stiffness is calculated for the corresponding mass of the TMD so that the frequency of the TMD is equal
to the frequency of the structure and the damping of the TMD is assumed equal to the damping of the
structure.TMD is installed at the top floor of the structure.
3.1. Parametric Study
In this paper study on mass parameter of TMD is carried out for one of the two structures considered for the
simulation with the six different earthquake ground excitations.
Tuned mass damper mass is varied between 0.01 to0.1 (structural mass) and tuned to the natural frequency of
the structure and the percentage control of displacement response is observed, from the results it is clear that as
TMD mass increases the percentage control of displacement is maximum.
4. PSO BASED TMD DESIGN
4.1 Particle Swarm Optimization
Particle swarm optimization is based on a model of social interaction between independent agents and uses
social knowledge or swarm intelligence in order to find the global maximum or minimum of a function. Particle
swarm optimization uses social rules to search in the parameter space by controlling the trajectories of a set of
independent particles. The position of each particle, representing a particular solution of the problem, is used to
792
compute the value of the fitness function to be optimized. Each particle may change its position, and
consequently may explore the solution space, simply varying its associated velocity.
The main PSO operator is, in fact, velocity update, that takes into account the best position, in terms of fitness
value, reached by all the particles during their paths and the best position that the agent itself has reached
during its search Pi resulting in a migration of the entire swarm towards the global optimum. At each iteration
the particle moves around according to its velocity and position, the cost function to be optimized is evaluated
for each particle in order to rank the current location. The velocity of the particle is then stochastically updated,
according to the formula.
V& = ωV& + ∁ φ P& − X & + ∁ φ G − X &
The term ω < 1 is known as the inertial weight and it is a friction factor chosen between 0 and 1 in order to
determine to what extent the particle remains along its original course unaffected by the pull of the other two
terms, it is very important to prevent oscillations around the optimal value. The other two terms are known as
self-knowledge and social knowledge and they are balanced by the scaling factors C1, 2 that are constants and
by φ , that are random positive numbers with a uniform distribution and a value that goes from 0 to 1.
4.2 TMD design using PSO
This study employs PSO to find the global optimum value of fitness function. It should be noted that choosing
of the properly objective function is very important in synthesis procedure for achieving the desired level of
system robust performance. Because different objective functions promote different PSO behaviours, which
generates fitness value providing a performance measure of problem considered.
For our optimization problem, the following objective function is.
J=∑01 -./
AMD= 234
Where, 234 is maximum relative displacement of ith story building response under base earthquake excitation
AMD is absolute maximum displacement. PSO is used to solve the above optimization problem that search for
optimal or near optimal set of TMD parameters. Optimum parameters of TMD to the ten storey building
(T==2sec) obtained using PSO for Elcentro earthquake excitation are m5 = 0.0543 β6 = 0.8783, ξ6 = 1.
In this study the design of a TMD using conventional method and PSO is carried out. For this purpose two ten
storey structures with different natural frequencies are considered. A time- history analysis is performed on the
structures considered under six different earthquake excitations. Properties of the two structures are given below
in table1 and Table 2
Table 1 Properties of the first ten-storey building Structure used in the analysis
Mass(kg) Stiffness(N/m) First Mode Natural Effective Percentage

Stories *10^3 *10^6 mode no frequencies(9) modal of effective
shape mass(kg)*10^6 modal mass
1 179 62.47 0.0002 1 3.143 1.1089 80.06
2 170 52.26 0.0003 2 8.335 0.1630 11.76
3 161 56.14 0.0005 3 13.516 0.0571 4.12
4 152 53.02 0.0007 4 18.434 0.0267 1.92
5 143 49.91 0.0008 5 22.953 0.0141 1.01
6 134 46.79 0.0010 6 26.967 0.0077 0.55
7 125 43.67 0.0011 7 30.382 0.0042 0.30
8 116 40.55 0.0012 8 33.123 0.0021 0.15
9 107 37.43 0.0013 9 35.126 0.0009 0.06
10 98 34.31 0.0013 10 36.357 0.0002 0.01
793
Table 2 Properties of the second ten-storey building Structure used in the analysis
Stori Mass(kg) Stiffness(N/m) First Mode Natural Effective Percentage

es *10^3 *10^6 mode no frequencies( modal of effective
shape 9) mass(kg)*10^ modal mass
6
1 1380 39.88 0.0554 1 0.8035 1.1701 84.79
2 1380 39.88 0.1095 2 2.3295 0.1261 9.14
3 1380 39.88 0.1612 3 3.9281 0.0427 3.09
4 1380 39.88 0.2093 4 5.3760 0.0197 1.42
5 1380 39.88 0.2527 5 6.7037 0.0103 0.74
7 1380 39.88 0.3217 7 8.8837 0.0031 0.22
8 1380 39.88 0.3458 8 9.6871 0.0015 0.11
9 1380 39.88 0.3622 9 10.2742 0.0006 0.04
10 1380 39.88 0.3705 10 10.6318 0.0001 0.01
6. RESULTS
6.1Structure one
Table 3. Responses of the ten-storey building using conventional method with and without TMD
with a mass ratio of 0.10
Earthquake data Displacement(m) Abs. Acceleration(g) Interstorey
drift(m)
UC C %Controlled UC C %controlled UC C
Elcentro 0.0228 0.0291 7.2 7.2 0.3773 10.91 0.0014 0.002
Kobe 0.0237 0.0139 41.21 41.21 0.4033 21.74 0.0009 0.0001
Taiwan 0.0197 0.0254 -28.89 -28.8 0.188 12 0.0006 0.0014
Capemendicono 0.0243 0.0220 9.53 9.53 0.526 7.3 0.0017 0.0011
Duzce 0.0093 0.0128 -37.72 -37.7 0.159 0.40 0.0003 0.0011
Mammoth lakes 0.0193 0.0124 35.80 35.80 0.3172 -1.44 0.0008 0.0002
6.2 Structure two

Table 4. Responses of the ten-storey building using conventional method with and without TMD with a mass
ratio of 0.10
Earthquake data Displacement(m) Abs. acceleration(g) Interstorey
drift(m)
UC C %Controlled UC C %controlled UC C
Elcentro 0.0555 0.0302 45.61 0.0879 0.0771 12.28 0.0024 0.0009
Kobe 0.0191 0.0174 9 0.0817 0.0798 2.28 0.0021 0.0013
Taiwan 0.0711 0.0510 28.17 0.1008 0.087 13.72 0.0009 0.0014
Capemendicono 0.0419 0.0295 29.53 0.0899 0.0812 9.68 0.001 0.0012
Duzce 0.0706 0.0480 31.99 0.0961 0.0967 -0.562 0.0025 0.0009
Mammoth lakes 0.0115 0.0108 6.14 0.1877 0.1863 0.733 0.0003 0.0003
794
50
Percentage control of diplacements

40
Elcentro
30 Kobe
20 Taiwan
10 Capmendocino
Duzce
0
Manmothlakes
0 2 4 6 8 10 12
Mass ratio
Fig.1. Percentage control of displacement in variation of mass ratio of a ten storey building (T=7.819sec)
0.03 uc 0.06
Top floor displacement(m)

3rd floor displaceent(m)
uc
0.02 c 0.04
c
0.01 0.02
0
0
-0.02
-0.01
-0.04
-0.02
-0.06
-0.03 0 10 20 30 40 50 60
0 10 20 30 40 50 60
Time(sec) Time(sec)
(a) (b)
0.15 0.1
uc uc
0.08
c
Top floor absolute
c
3rd floor absolute
0.1 0.06
acceleration(g)
acceleration(g)
0.04
0.05
0.02
0
0
-0.02
-0.05 -0.04
-0.06
-0.1 -0.08
0 10 20 30 40 50 60 0 10 20 30 40 50 60
(a) Time(sec) (b) Time(sec)
Fig.2. Displacement and Absolute acceleration of third and top floor of a ten storey building (T=7.819sec)
with and without TMD using conventional method
795
10 uc 10 uc
c c
9 9
8 8 (b)
7
(a) 7
6 6
Floors
Floors
5 5
4 4
3 3
2 2
1 1
0 0
0 0.02 0.04 0.06 0 0.002 0.004 0.006 0.008 0.01
Displacement(m) Inter storey drift(m)
Fig.3. Displacements 5(a) and Inter storey drift 5(b) in floor wise with and without TMD
of a ten storey building (T=7.819sec) using conventional method.
Table.5. Percentage control of responses of the ten-storey building (T=2sec) with and without TMD
(CM=conventional method; PSO = Particle swarm optimisation method)
Earthquake Structure one Structure two

data Displ Abs.accl Displ Abs.accl
(%controlled) (%controlled) (%controlled) (%controlled)
CM PSO CM PSO CM PSO CM PSO
Elcentro 7.63 20.53 10.91 18.25 45.61 45.18 12.28 3.99
Kobe 40.21 39.86 21.74 11.58 9 6.55 2.28 -0.60
Taiwan -28.89 34.01 12 35.78 28.17 28.73 13.72 6.93
Capeme- 9.53 6.27 7.3 1.75 29.53 28.63 9.68 0.59
ndocino
Duzce -35.72 24 0.40 0.92 31.99 32.55 -0.56 -0.49
Mammo- 35.80 46.44 -1.44 1.07 6.14 5.91 0.73 0.12
thlakes
Fig. 4. Minimum fitness convergence curve using PSO for a ten storey structure (T=2sec) generated for
Elcentro earthquake excitation
796
0.01 UC 0.025 UC
10th Floor Displacement(m)

3rd Floor Displacement(m)
0.008 TMD_C 0.02 TMD_C

TMD_PSO_C 0.015 TMD_PSO_C
0.006
0.004 0.01
0.005
0.002
0
0 -0.005
-0.002 -0.01
-0.004 -0.015
-0.006 -0.02
-0.008 -0.025
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time(sec) Time(sec)
(a) (b)
0.4 UC 0.4 UC
TMD_C 0.3 TMD_C
10th Floor Absolute

0.3
3rd Floor Absolute
TMD_PSO_C
Acceleration(g)
Acceleration(g)
0.2 TMD_PSO_C
0.2
0.1
0.1
0
0
-0.1
-0.1 -0.2
-0.2 -0.3
-0.3 -0.4
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time(sec)
(c) (d) Time(sec)
Fig.5. Displacement and absolute acceleration of third and top floor of a ten storey building (T=2 sec)
with and without TMD using conventional method and PSO
10 10 UC
TMD_C
9 9 TMD_PSO_C
8 8
7 7
6 6
Floors
Floors
5 5
4 4
UC
3 TMD_C 3
2 TMD_PSO_C 2
1 1
0 0
0 0.005 0.01 0.015 0.02 0.025 0 0.0005 0.001 0.0015 0.002 0.0025 0.003
Displacement(m) Displacement(m)
(a) (b)
Fig.6. Displacements 8(a) and Inter storey drift 8(b) in floor wise with and without TMD of a ten storey building
(T=2sec) using conventional method and PSO
7. CONCLUSIONS
In this study, the effectiveness of the TMD for control of the two example ten storey shear building, with
different natural frequencies are considered and time history analysis with six past actual earthquake ground
motions are performed. In the time history analysis, using the Elcentro, Kobe, Taiwan, capemendocino, duzce,
mamothlakes, The TMD parameters designed using conventional method and by PSO technique, based on the
numerical results obtained in this paper, overall TMD is effective in reducing displacement and absolute
acceleration and, thereby, can be used as an effective control measure for structures under earthquake. The
following observations and conclusions can be made from this study.
797
1. It is observed that percentage control of displacements is increasing with increase in mass ratio; the
parameters obtained from a mass ratio of 0.10 are the best optimum parameters for the TMD to control the
responses of the structures with low frequencies.
2. Conventional method results conclude that TMD is effective in reducing the responses of the structures
which has low natural frequency.
3. From the results it is observed clearly that TMD designed using PSO is more effective in controlling the
response of the structures subjected to earthquake excitation when compared to the conventional method.
4. It can also be concluded from the above results, TMD is more effective in controlling the flexible structure.
REFERENCES
[1]. Y.P.Gupta and Chandrasekaren, Absorber system for earthquake excitation, Proc.4th world conf.on
earthquake engineering, Santiago, Chile, Vol.2, 1969, pp.139-148.
[2]. A.M.Kaynia, D.VeneZiano and J.M.Biggs,’Sesimic effectiveness of tuned mass dampers

J.struct.div.ASCE107,1465-1484(1981).
[3]. J.R. Sladek and R.E.Klinger, Effect of tuned-mass dampers on seismic response; J.struct.div.ASCE 109,
2004-2009(1983).
[4] .P.H.Wirsching and J.T.P.Yao, safety design concepts for seismic structures;Comput.struct.3,809-
826(1973).
[5]. R.G.Dong, Vibration-absorbing effect under seismic excitation; J.struct.div.ASCE 102, 2021-2031(1976).
[6]. S.Ohno,A.Watari and I.Sano, Optimum tuning of the damper to control response of structures to
earthquake ground motion;Proc.6th world conf.on earthquake engineering,New Delhi,India,Vol
3,1977,pp.157-161.
[7]. M.Abdullah, A.Richardson, placement of sensor/actuators on civil structures using genetic algorithms,
Earthquake Engineering and Structure Dynamics, 2001, pp.1167- 1184.
[8]. H. Shayeghi, H. Eimani, H.A. Shayanfar, A. Shayeghi, PSO based TMD design for vibration control of tall
building structures , Proc. of the Int. Conf. on Artificial Intelligence (ICAI 09), Las Vegas, U.S.A., 2009.
[9]. R. Randa, T.T. Soong, Parametric study and simplified design of mass dampers.
Engineering Structures, Vol. 20, No. 3, 1998, pp. 193 -204.
[10]. Symans MD, Constantinou MC. Semi-active control systems for seismic protection of
structures: a state-of-the-art review. Engineering Structures 1999; 21:469–87.
798
Paper No.: 143
PROPERTIES OF HIGH STRENGTH CONCRETE CONTAINING

ULTRAFINE SLAG
Jagdish Malav1, Swapnil Takle1, Kali P Sethy2, Dinakar. P3 and U. C. Sahoo3

1
B.Tech Graduate, School of Infrastructure, Indian Institute of Technology, Bhubaneswar-751013, India
2
PhD Scholar, School of Infrastructure, Indian Institute of Technology, Bhubaneswar-751013, India
3
Assistant Professor, School of Infrastructure, Indian Institute of Technology, Bhubaneswar-751013, India
*Email: pdinakar@iitbbs.ac.in
ABSTRACT:
This study presents the effect of incorporating ultrafine slag (UFS) on the mechanical and durability properties
of high strength concrete for a constant water/binder ratio of 0.3. Ultrafine slag (UFS) mixtures with cement
replacement of 5, 10, 15 and 20 % were designed for target strength and slump of 90 MPa and 100 ± 25 mm.
From the results, it was observed that upto 20% replacements the incorporation of UFS had shown a significant
improvement in terms of the mechanical properties of the concrete compared to control mixture. Compressive
strength of 105 MPa was achieved at 20% replacement. Splitting tensile strength and elastic modulus values
have also followed the same trend. In durability tests UFS concretes have exhibited high resistance compared
to control and the resistance increases as the UFS percentage increases. This investigation has shown that the
UFS has the capability to produce high strength and high performance concretes.
KEYWORDS: Ultrafine slag, Compressive strength, durability, water permeability, UFGGBS.
1. INTRODUCTION
The necessity for the development of high strength and high performance concretes has increased gradually
because of the demands imposed by the construction industry in recent times. In the last three decades,
supplementary cementitious materials such as fly ash, silica fume and ground granulated blast furnace slag have
been judiciously utilized as cement replacement materials as these can significantly enhance the strength and
durability characteristics of concrete in comparison with ordinary Portland cement (OPC) alone, provided there
is adequate curing [1]. Hence, high strength and high-performance concretes can be produced at lower w/b
ratios by incorporating these supplementary materials. However, when high strength is desired, use of silica
fume is desirable [2].
When concretes were designed at a very low water/binder ratio, silica fume provides a very good particle
packing and, because of its strong pozzolanic activity increases the resistance of the concrete to aggressive
environments also [3]. However, this product is rather expensive. In India, most of the good quality silica
fume is imported and the cost is 9 to 10 times the cost of ordinary Portland cement (OPC). The substitution of
silica fume by another ultra fine material of lower cost, whose availability would not raise particular difficulties,
would support the development of concrete industry. There are numerous interesting types of ultra fine
materials with various characteristics and performance levels according to their origin. In the present
investigation ultrafine slag (UFS) was used as a replacement for silica fume for developing high strength
concretes of more than 90 MPa. The objective of this study was to investigate the effect of using ultra fine slag
with a fineness of 1200 m2/kg obtained commercially as pozzolan on the development of high strength and
permeability characteristics of concrete designed for a very low w/b ratio of 0.3.
2.1 Materials
An experimental program was designed to produce a high strength concrete by adding several combinations of
ultrafine slag. The materials used and the experimental procedures are described in the following sections
• The cement used in all mixes was normal Ordinary Portland cement (53 grade) conforming to IS 12269. It
has got a specific surface of 360 m2/kg and a specific gravity of 3.15. Ultra fine slag (UFS) used as a
mineral additive has a specific gravity of 2.94, and a specific surface of 1200 m2/kg with average particle
size of 2.9 microns. The properties of UFS were given in Table 1.
• As for the aggregates, good quality aggregates have been procured for this investigation. Crushed granite
with nominal grain size of 20 mm and well-graded river sand of maximum size 4.75 mm were used as coarse
and fine aggregates, respectively. The specific gravities of aggregates were determined experimentally.
The coarse aggregates with 20, 12.5 mm fractions had specific gravities of 2.91 and 2.80, whereas the fine
aggregate had specific gravity of 2.73, respectively.
• Commercially available polycarboxylic ether based superplasticizer (SP) was used in all the concrete
mixtures.
Table- 1 Chemical and Physical Composition of Ultrafine slag
Chemical Analysis Mass % Physical Analysis Range

CaO 35.2 Bulk density 650 kg/m3
Al2O3 17.4 Surface area 12000 cm2/gm
Fe2O3 2.6 Particle shape Irregular
SiO2 31.2 Particle size, d10 < 2u
MgO 7.5 d50 < 5u
SO3 2.3 d90 < 9u
2.2 Mixture proportions
Trials mixtures were prepared to obtain target strength of more than 90 MPa for the control mixture at 28 days
and the water/binder ratio for all the mixtures were kept constant at 0.30. The details of the mixtures for the
study are presented in Table 2. Five different mixtures (UFS0, UFS5, UFS10, UFS15 and UFS20) were
employed to examine the influence of low water to binder ratio on concretes containing UFS on the mechanical
and durability properties. The control mixture (UFS0) does not contain ultrafine slag. In mixtures UFS5,
UFS10, UFS15 and UFS20, cement content was partially replaced with 5%, 10%, 15% and 20 UFS (by mass)
respectively. The binder consists of cement and UFS. Trial mixtures were conducted to determine the
optimum dosage of superplasticiser for each of the mixtures in order to achieve the target slump of 100 ± 25 mm.
With respect to the aggregate grading, in the present investigation a combined aggregate grading as
recommended by the DIN 1045 standards was utilized. The aggregates 20, 12.5 and 4.75mm were combined
in such a way, so that it meets nearly the combined grading specification of DIN ‘A’ curve. The percentage
fractions of aggregates used for 20, 12.5 and 4.75mm are 31%, 32% and 37% of the total aggregate content.
Blending aggregates in this fashion and designing concretes at very low water binder ratios will yield high
strength concretes, because of the good packing density [4].
The mixing of the concrete ingredients was carried out in a pan mixer of 40-litre capacity. Initially, the
aggregates were put into the mixer, followed by the cement and mineral additives. Mixing of dry ingredients
was carried out for a period of 2 min. The required superplasticiser dosage together with the mix water was
added to the concrete mixer and the mixing was continued for 3 min. The fresh concrete was placed into the
steel cube moulds and compacted on a vibrating table. Finally, surface finishing was done carefully to obtain a
uniform smooth surface. 100 mm cubes, 100mmX200mm cylinders, 150X300mm cylinders and 15mm cubes
were used to evaluate the compressive strength, splitting tensile strength, modulus of elasticities and water
permeabilties of the concrete respectively. Tests are conducted at different test ages; 3,7,28 and 56 days of age.
800
All experiments were performed on three specimen replicates. The average values are used in the discussion of
the test results. The specimens were demolded after 24 hours of casting and were then cured in water at
approximately 270 C until the testing day.
Table- 2 Mix Proportion per one cubic meter (SSD Condition)
Constituents/Mi
UFS300 UFS305 UFS310 UFS315 UFS320
x
Cement 533.33 506.67 480 453.33 426.67
UFS 0 26.67 53.33 80 106.67
Water 160 160 160 160 160
20mm Aggregate 605.03 604.48 603.94 603.39 602.85
10mm Aggregate 600.94 600.4 599.85 599.32 598.77
Natural Sand 677.46 676.85 676.24 675.63 675.10
Density 2576.76 2575.06 2573.36 2571.67 2569.97
2.3 Test procedures
The workability of the fresh concrete is measured by using the standard slump test apparatus. The unconfined
compressive strength was obtained, at a loading rate of 2.5 kN/s at the age of, 3, 7, 28, and 56 days on 3000kN
machine. The average compressive strength of three specimens was considered for each age. The split tensile
strength was also tested on the same machine at the age of 28 and 56 days.
The elastic modulus was determined at the age of 56 days. The specimens were fixed with a longitudinal
compressometer, placed vertically between the platens of the compression testing machine and tested. This
test conforms to ASTM C 469 (ASTM 2002) for static modulus of elasticity of concrete in compression. All
the specimens were tested on saturated surface dry condition.
The permeability characteristics were determined by measuring the the water penetration depths under pressure
that were performed on 150mm cubes as per EN 12390-8 (BS 2000) at 28 and 56days. The test method
involves the study of water penetration on 15 cm cubes over a 5 bar pressure for a period of 72 hours. The
experimental setup used for this study was shown in Figure 3. After the test the specimens were split exactly
into two halves and the water penetration front was marked on the specimen as shown in Figure 4. The
maximum depth of penetration under the test area was determined using vernier caliper and recorded it to the
nearest millimetre.
3.1 Fresh Properties
In this study, different superplasticiser dosages were added to the different mixtures in order to obtain the
consistency or workability in terms of target slump of 100±25 mm. It can be seen from Table 3 and Fig. 1 the
superplasticiser demand decreased with increase in the ultra fine slag replacements. For example, the 20, 15,
10 and 5% UFS mixtures require 62.5, 50, 37.5 and 25 % less superplasticiser dosage in comparison with that of
the control mixture for obtaining almost the same workability. This is mainly because the ultra-fine slag
particles fills the space among cement particles and make the particle size distribution better than the cement
alone and finally making the concretes more fluidifying than the control at a very low admixture dosages
compared to control mixture. Moreover, the lower the dosage of superplasticizer needed for the UFS concretes
also indicate better compatibility of UFS with the superplasticizer used.
As far as the workability is concerned, in fact all the concretes the control and the UFS mixtures have obtained
their design slumps and performed almost similarly as shown in Table 3. According to these results, concretes
801
made had a high slump values, highly cohesive and can be easily pumped. No wide variations in the slump
values for the mixtures containing increased amounts of UFS were observed.
Table- 3 Superplasticizer demand and the workability test results
Admixture
Slump
Mix Dosage
(mm)
(%)
UFS300 0.40 110
UFS305 0.30 100
UFS310 0.25 100
UFS315 0.20 105
UFS320 0.15 105
0.5
Admixture dosage (%)
0.4
0.3
0.2
0.1
0
0 5 10 15 20 25
UFS (%)
Fig.1. Admixture dosage demand with respect to UFS replacement.
3.2 Compressive strength
The compressive strength results of samples presented in Table 4, shows that all the concretes made in this study
are high strength, as even the seven day compressive strength varied between 80 and 88 MPa. The 28-day
strength varied between 92 and about 97 MPa, and the 56-day strength varied between 93 and 105 MPa. The
20% replacement UFS20 mixture had exhibited higher strengths comparatively than the other UFS percentages.
All the concretes including the control had achieved their target strength of 90 MPa at 28 day and at 56 days the
UFS20 achieved strength of more than 100 MPa. Figure 2 presents the relation between compressive strength
and ultrafine slag percentages at 28 and 56 days. It can be seen that the compressive strength was the highest
for the UFS20 mixtures achieving strengths of 97 and 105 MPa at 28 and 56 days. Beyond 28 days the UFS
concretes demonstrated the best strength development showing continuous strength improvement and higher
strengths than control concretes.
Ultra fine slag belong to active mineral admixtures, they can improve the microstructure and mechanical
properties of hardened concretes. When they were utilized with cement, the reaction mechanism can be
divided into physical and chemical aspects. The physical effect is that the ultra-fine particles fill the voids in
cement, which makes the microstructure of cement paste denser. The chemical effect is the reaction of ultra
fine slag in the presence of cement hydrates. However, since the slag was ground into ultra-fine, it had
improved the initial hydration as well as the strength development considerably compared to control concretes.
On the other hand, the average particle size of ultra fine slag is smaller than that of cement. It filled the space
among the cement grains, improving the particle distribution of cementitious system and forming a dense
microstructure leading to a significant increase in the matrix strength of the concretes. The fineness of ultra
fine slag concretes makes the pozzolanic reaction stronger. Moreover, it was verified by Ping et al [5] that
adding fine slag can increase the density and offer a unique strengthening mechanism for the transition zone
which is potentially a weak link with respect to strength of concrete. The current observations agree with this
point.
802
Table- 4 Mechanical properties of the concretes developed
fsp/fck Elastic Elastic Elastic

Compressive strength (MPa) modulus modulus modulus
Mixes
(GPa) (GPa) (GPa)
3day 7day 28day 56day 56 day ACI[4.73√fck] BIS[5√fck]
UFS300 66.65 81.52 92.64 93.28 4.17 44.86 41.16 43.51
UFS305 65.42 79.75 92.30 94.15 5.23 45.10 41.93 44.32
UFS310 73.75 81.64 93.19 96.58 5.38 45.88 44.12 46.64
UFS315 74.97 84.16 95.60 98.30 5.35 46.55 45.71 48.31
UFS320 67.49 88.70 97.30 105.80 5.16 47.34 52.95 55.97
120
Compressive strength (MPa)
100
80
60
28 days
40
56 days
20
0
300 305 310 315 320
Mixes
Fig.2. Comparison of compressive strengths.

3.3 Split Tension strength
The tensile strength results of UFS concretes with varying amounts of ultrafine slag are shown in Fig. 3. The
average value of the 28-day tensile strength for the concretes made was about 5.21 MPa, which corresponds to
5.28% of the compressive strength for the same concretes. Table 4 shows that the average ratio between the
tensile strength (fsp) to cube compressive strength (fck) of concrete at 56 days was lower than the range (of about
9-10%) for medium strength concrete reported earlier [1][6][7]. This indicates that as the compressive strength
increases lower would be the ratio, which is consistent with the data published by other investigators earlier
[6][7][8]. From the results it can be seen that similar to compressive strength the splitting tensile strength also
exhibited the highest strength at UFS20 mixture. Figure 7 presents the comparison of the split tensile strength
values of all the concretes developed both at 28 and 56 days. From the figure it can be seen that as expected as
the replacement of UFS content increases the corresponding split tensile strength values also increases.
803
6
Split tensile strength (MPa)

5
3
28 days
2 56 days
1
0
300 305 310 315 320
Mixes
Fig.3. Comparison of split tensile strengths.
3.4 Elastic Modulus
The modulus of elasticity is mainly related to the compressive strength of concrete. However, due to the
existence of non-linear relationship between them [1][9], the increase in the modulus of elasticity is not in
proportion to the increase in compressive strength as noted in Table 4. The modulus values presented in Table 4
indicate that the rate of increase in the modulus is lower than the rate of increase in the compressive strength.
The elastic modulus (E) values with respect to the UFS contents are presented in Table 4 and in Figure 4. The
trend is similar to that obtained for compressive strength. In addition, the predicted values according to the
American Concrete Institute (ACI) model (E = 4.73√fck ) and BIS model (E = 5 √fck) are also calculated and
shown in the same table. The figure shows that the values obtained experimentally for UFS mixtures lie
slightly above the predicted modulus of ACI model but the BIS model overestimates the UFS values obtained
by actual testing.
48
47
Elastic Modulus (Gpa)
46
45
44
43
42
41
40
300 305 310 315 320
Mixes
Fig.4. Comparison of Elastic modulus.
3.5 Water Penetration depth
One of the main factors of concrete durability is permeability. Concrete with lower permeability shows better
resistance against chemical attacks. When water penetrates into the concrete, some soluble salts including
chloride ions penetrate into concrete and cause corrosion. Generally, it seems that lower permeability causes
804
higher durability in concretes [10]. Water penetration test was used to evaluate the permeability of concretes
and validity of these tests has been approved [11]. Fig. 5 shows the results of the water penetration depths of
all the mixtures studies in this investigation. The water penetration depths decrease as the ultrafine slag
content increases very significantly up to 20% replacement. The trend is the same both at 28 and 56days.
Water penetration depth (mm) 2.5
1.5
28 days
1
56 days
0.5
0
300 305 310 315 320
Mixes
Fig.5. Comparison of water permeability depths.
From the above investigation it is obvious that the judicious choice of type of mineral admixture is essential
when producing high strength concretes. In the last two decades silica fume has been used as a material of
choice for developing very high strength concretes in the concrete industry. But because of its very high
fineness, of the order 10,000 to 20,000 m2/kg, will naturally be highly reactive and can give the advantages of
reduced curing time needed to obtain a desired level of strength and water tightness. But, abnormal fineness
will definitely bring in many practical field problems when designed at low water to cementitious materials ratio.
On the other hand, moderately fine or reactive mineral admixtures with a fineness of 1200 to 1500 m2/kg may
create a highly durable cement matrix with improved resistance to thermal cracking whilst maintaining
continued strength development as reflected in the present study with ultra fine slag [12]
4. CONCLUSIONS
The following conclusions can be made based on the results of this study:
• UFS has supplementary effect on water reducing property. UFS concretes demanded less superplasticizer
dosages as the replacement percentages increases.
• UFS concretes possessed excellent strength improvement at all the ages studied. This may be attributed to
the average particle size of UFS, which is smaller than cement leading to the formation of dense matrix and
interface bonding property of hardened concretes. Strengths of more than 100 MPa can be realized using
UFS. Similar trends were also noticed on other mechanical properties such as split tensile strength and
elastic modulus. As far as durability is concerned UFS concretes possessed good resistance in terms of the
water permeability studied compared to the control concretes.
• Thus, the results obtained with ultra fine slag used in this study appear very promising for the manufacture of
the high strength concretes. As this ultra fine is produced industrially, its availability and its low price
would favor its development. The study of microstructure and other durability represent the next research
work in order to complete the behavior of ultra fine slag in high strength concrete.
805
REFERENCES
[1] Neville, A.M. (1997) Concrete with particular properties. In Properties of Concrete, Longman, Harlow,
Essex, 653 – 672.
[2] Basu, P.C. (2003) High Performance Concrete, Proceedings INAE National Seminar on Engineered
Building Materials and Their Performance, 426-450.
[3] Abdul, R.H., & Wong, H.S. (2005) Strength estimation model for high-strength concrete incorporating
metakaolin and silica fume. Cement Concrete Research, 35 (4), 688–695.
[4] Dinakar, P. (2012) Design of self compacting concrete with fly ash. Magazine of concrete research, 64(5),
401-409.
[5] Ping, Gu., Xie, P and Beaudoin, J. J 'Microstructural characterization of the transition zone in cement
systems by means of A.C. impedance spectroscopy', Cement and Concrete Research. 23 (3) (1993),
581-591.
[6] Rasiah, A.R. (1983) High strength concrete for developing countries. In: Proceedings of the First
International Conference on Concrete Technology in Developing Countries. Amman, Jordan.
[7] Haque, M.N., & Kayali, O. (1998) Properties of high strength concrete using a fine fly ash. Cement &
Concrete Research, 28(10), 1445-1452.
[8] Yogendran, V., Langan, B.W., Haque, M.N. & Ward, M.A. (1987) Silica fume in high strength concrete.
ACI Materials Journal, 84, 124.
[9] Mehta, P.K., & Monteiro, P.J. (1999) Concrete: microstructure, properties, and materials. Indian Concrete
Institute, India.
[10] Ramezanianpour AA, Pilvar A, Mahdikhani M, Moodi F. Practical evaluation of relationship between
concrete resistivity, water penetration, rapid chloride penetration and compressive strength. Constr Build
Mater 2011; 25(5):2472–9.
[11] BS EN-12390-8. Depth of penetration of water under pressure. British Standards Institution; 2000.
[12] Dinakar, P (2010) Effect of ultrafine slag on the fresh and hardened properties of ultra high strength
concrete. 3rd Asian Conference on Ecstasy in Concrete, ACECON 2010, 527-532, Dec, Chennai, India.
806
Paper No.: 144
PROGRESSIVE COLLAPSE STUDY OF SEMI-RIGID JOINTED FRAME

STRUCTURES: A STATE-OF-THE ART REVIEW
1 2
A. K. Banik and M. Bandyopadhyay
1
Associate Professor, Dept. of Civil Engineering, National Institute of Technology, Durgapur, India
2
PhD Scholar, Dept. of Civil Engineering, National Institute of Technology, Durgapur, India
Email: akbanik@gmail.com, milanbandyopadhyay@yahoo.co.in
ABSTRACT:
Progressive collapse of structures is caused by sudden removal of one or more load bearing members. It may be
partial or total collapse of a structure depending upon the layout and type of frame. Significant study has been
carried out in the area of progressive collapse of structures in last few decades. Collapse of Ronan Point
Building in UK in 1968 and terrorist attacks against the Alfred P. Murrah Federal Building, Oklahoma City,
1995, and the World Trade Center (WTC), New York, 2001 attracted the structural engineers and researchers
around the globe towards this field. Based upon the research findings, design standards and guidelines are also
modified to take into account progressive collapse of structure. UK is the first nation to incorporate this
phenomenon in their national design code followed by US, Canada. India also included disproportionate
collapse in the recent revision of steel structure design code IS 800-2007. But it is worthwhile to mention that
most of the research conducted in the area of progressive collapse analysis is for moment resisting framed
structure, considering the joints as fully rigid. For the ease of analysis connections are generally considered to
be rigid or pinned but in reality, between these two extreme types of connections there exists another type of
connections known as semi-rigid connection. A limited study is available in the area of progressive collapse of
semi-rigid jointed frame structures because of its wide and complex behaviour. This paper covers the
state-of-the art review of relevant works done in last four decades in the areas of (i) progressive collapse of
structures, (ii) modeling and influence of semi-rigid connection and (iii) the combination of both (i) and (ii). The
potential areas of future research in progressive collapse of semi-rigid jointed frames are also identified.
KEYWORDS: Progressive collapse, semi-rigid connections, moment-rotation curves, beam-to-column

connections, steel frame.
1. INTRODUCTION
Sudden removal of load bearing members, whether accidental or intentional, increases the loading on the nearby
members. Modern buildings designed using sophisticated numerical tools may not have reserve capacity to
carry this additional sudden load (Smilowitz, 2006) and thus may fail. The share of load is then transferred to
next adjacent members which may in turn exhibit loadings more than their capacity and fail. The process goes
on until stability is obtained or the structure undergoes partial or complete collapse. The catastrophic
progressive collapse events of last decades marked it as a topic raising widespread interest. Probably one of the
most famous progressive collapse failures was the 1968 collapse of the Ronan Point apartment building. The
building was a 22-story precast concrete bearing wall system. An explosion in a corner kitchen on the 18th floor
blew out the exterior wall panel and failure of the corner bay propagated up and down to cover almost the
complete height of the building (Figure 1). After this event, England was the first nation to address progressive
collapse explicitly in their building standards. Renewed attention to the problem was given due to terrorist
attacks against the Alfred P. Murrah Federal Building, Oklahoma City, 1995, and the World Trade Center
(WTC), New York, 2001 (Figure 1). Starting from the seventies of the last century, the strategies to evaluate and
provide structural robustness, i.e. safety in presence of extraordinary accidental events like explosions or
impacts, have improved remarkably. Mohamed (2006) has reviewed some literature, technical papers, standards
and guidelines related to progressive collapse of structure till 2005-2006. The publication discussed the gist of
19 important shortlisted papers and 6 Codes of practices and standards i.e. British, Canadian and four US
standards (ACI 318, ASCE 7, GSA 2003, DoD 2005). He raised some doubt over sufficient clarity of the design
guidelines that help design engineers to carry out design of progressive collapse resistant structures.
Figure 1 Progressive collapse, clockwise from left top corner, (a) Ronan Point building after 16 May 1968
collapse, (b) Murrah Federal Office building after 19 April 1995 attack, (c) World Trade Center 1 and 2 on 11
September 2001 and (d) Schematic diagram of Ronan Point building progressive collapse
Substantial studies are carried out in last decades in the progressive collapse assessment of structure, but most of
the researcher has investigated the phenomenon for moment resisting frame i.e. rigid jointed frame. Liu et. al.
(2010) mentioned that semi-rigidly connected frames are more vulnerable than rigidly connected frames in
resisting progressive collapse. For the ease of analysis we consider the frame to be rigidly connected but in
practice it is very difficult to attain such rigidity, making the frame as semi-rigid jointed. The review papers
prepared by Jones et. al. (1983) and Diaz et. al. (2011) covers the research and development in the analysis of
semi-rigid jointed frames and joint behavior of frame respectively since 1930’s.
This paper covers the state-of-the art review of relevant works done in last four decades in the areas of (i)
progressive collapse of structures, (ii) modeling and influence of semi-rigid connection and (iii) the combination
of both (i) and (ii). Due to space limitation findings of not all the researcher but only the relevant and widely
referred papers are reported herein .The paper also identifies the potential areas of future research.
2. PROGRESSIVE COLLAPSE OF STRUCTURES
ASCE/SEI-7 (2010) defines progressive collapse as the spread of an initial local failure from element to element,
eventually resulting in the collapse of an entire structure or a disproportionately large part of it. General Services
808
Administration (GSA 2005) describes progressive collapse as a situation where local failure of a primary structural
component leads to the collapse of adjoining members which, in turn, leads to additional collapse. The guidelines
issued by the US Department of Defense (DoD 2009) include detailed methodologies to reduce the potential for the
progressive collapse of structures. It employs three design/analysis approaches to evaluate the progressive collapse
resistance of buildings: one direct method called Tie Force (TF) and two indirect methods known as Alternate Path
(AP) and Enhanced Local Resistance (ELR).
The GSA Progressive Collapse Analysis and Design Guidelines (GSA 2005) adopted a threat independent or
alternate path approach in evaluating and designing structure against progressive collapse. With this
methodology, the designer is required to systematically remove key elements carrying gravity loads. These key
elements may be columns or load-bearing walls around the perimeter of the building. The remaining structure is
then designed to sustain redistributed loads without collapse.
2.1. Primary causes of Progressive Collapse
The root cause of progressive collapse i.e. failure of a primary vertical support might occur due to extreme
loadings such as a bomb explosion in a terrorist attack, a car colliding with supports in a parking garage, an
accidental explosion of explosive materials, failure of inadequate members caused by design flaw and severe
earthquake.
Blast is treated as a primary reason behind progressive collapse. In most of the studies blast is also simulated by
removing one or a few load-carrying structural members with static and zero initial conditions. But this
assumption is not valid, as it blast will damage adjacent structural members and initiate a collapse with non-zero
initial condition. Shi et. al. (2010) proposed a new method for progressive collapse analysis of reinforced
concrete (RC) frame structures by considering non-zero initial conditions and initial damage to adjacent
structural members.
Though it has been discussed by many researchers (Gurley, 2008) that earthquake or severe earthquake may
cause progressive collapse, a very limited study has been carried out considering earthquake as a cause of
progressive collapse. This may be because of difficulties in characterizing the loading and burden of performing
nonlinear analysis. Most of the study assessed the progressive collapse resistance of seismically designed
structure. Gurley (2008) discussed some differences and similarities between earthquake damage and
progressive collapse of structure. With reference to records of earthquake damage the author mentioned that
earthquakes can also remove supports, often corner columns causing two-way cantilever mechanisms.
Earthquake engineering does need to include recognition of “lost column” events and to incorporate design
against progressive collapse.
Li and Ellingwood (2006) investigated the potential for aftershocks to cause additional damage to steel moment
frame buildings and provided a probabilistic description of structural damage states prior to and following those
aftershocks. They proposed enhanced uncoupled modal response history analysis (EUMRHA) method which is
a modification of Modal Pushover Analysis (MPA) developed by Chopra and Goel (2002). Damage
accumulation is analysed in terms of a normalized ‘damage ratio’ and maximum inter-storey drift.
Seffen (2008) discussed the factors responsible for the onset of collapse of WTC towers. He mentioned that
despite localized and substantial horizontal impacts by fuel-laden aircraft, both the towers initially survived. But
gradually the intense fire compromised the load carrying capability of the remaining intact columns close to the
aircraft impact zones. Lien et al. (2010) studied the behaviour of steel frame under fire induced by earthquake.
They concluded that the deformation of structure is significantly affected by the aftershock, fire and fracture of
structural element.
A structure could be subject to more than one critical loading events during its service life. Asproneet. al. (2010)
proposed the multi-hazard approach for consideration of the blast actions in the form of blast fragilities in
addition to the seismic fragilities.
809
2.2. Progressive Collapse Analysis Procedure
Progressive collapse is a highly nonlinear dynamic event. But apart from nonlinear dynamic (ND) analysis, the UFC
(Unified Facilities Criteria under Department of Defense) also allows linear static (LS) and nonlinear static (NS)
analysis with suitable load or dynamic increase factor. As the simplest analysis option, the linear static (LS)
procedure offers preliminary and often conservative assessment of the progressive collapse potential. It doesn’t
take care of either geometric and material nonlinearity or the dynamic effects of loading. It has gained the
popularity in engineers’ community because of its simplicity and as it can be performed in any design office
with less effort. This method is useful for simple structure where both nonlinear and dynamic load effects can
easily be predicted and compensated by a suitable load increase factor. The nonlinear dynamic (ND) procedure
is the most accurate and also most computationally expensive option, as it necessitates sophisticated numerical
modeling and considerable computing efforts to carry out the time history analysis. As an intermediate analysis
option, the nonlinear static (NS) procedure involves modeling of both material and geometrical nonlinearities
while it does not perform time history analysis to simulate the load redistribution behavior. To approximately
compensate for the dynamic effects, a dynamic increase factor is used to increase the gravity loads acting on the
areas that are immediately affected by the removed structural elements.
Marjanishvili (2004) studied the advantages and disadvantages of the above procedures when applying them in
the progressive collapse analysis. Marjanishvili and Agnew (2006) compared the four analysis methods (LS, NS,
ND and linear dynamic, LD) by analyzing a nine-storey steel moment-resistant frame building under GSA
guidelines. They plotted the vertical displacement at the node just above removed column against the applied
load for all four methods. The respective demand capacity ratio and hinge rotations are also compared. To
further contribute to the elaboration of design codes for progressive collapse Menchelet. al. (2009) compared
four different methods: linear static GSA procedure, the linear static DoD procedure, the nonlinear static DoD
procedure, and the Load History Dependent (LHD) Procedure.
Many researchers investigated the adequacy of these load and dynamic increase factors. McKay et. al. (2012)
found these factors to be conservative whereas Ruth et. al. (2006) concluded that a dynamic multiplier of 1.5
better captures the dynamic effects when a static analysis is performed, and will result in more economical
designs.
In order to overcome the limitations of the static approach, Kaewkulchai and Williamson (2004) formulated a
beam element for computing the dynamic response of frame structures during progressive collapse. They (2006)
also investigated the impact of failed members and found that impact velocity plays an important role on
predicting progressive collapse. Kwasniewski (2010) presented the nonlinear dynamic simulations of
progressive collapse for an existing eight-story steel framed structure and identified influence of modeling
parameters.
Kim (2009) investigated progressive collapse resisting capability of steel moment frames using push-down
analysis and compared the results with nonlinear incremental dynamic analysis (IDA). It was found that push
down analysis might overestimate the inherent capacity of structures against progressive collapse. As
recommended in the guideline, each member end was assumed to become the “hinge” when the maximum
rotation reaches 0.035 rad.
Khandelwal and El-Tawil (2011) proposed three pushdown methods that can be used to investigate the
robustness of a damaged building system in terms of residual capacity and associated collapse modes. These
were uniform pushdown (UP), bay pushdown (BP) and incremental dynamic pushdown (IDP). They found IDP
to give the most realistic estimate of residual capacity and collapse modes.
Asgarian and Rezvani (2012) introduced a new algorithm, namely the extended progressive collapse analysis
(EPCA) algorithm to find the element removal impact factor and the most critical locations of such removals.
Their failure modes were determined by using pushdown and vertical incremental dynamic analyses (IDA).
Kokot et. al. (2012) conducted static and dynamic analysis of a reinforced concrete flat slab frame building for
810
progressive collapse both numerically and experimentally. Numerically they considered instantaneous (less than
5 ms) column failure case using SAP2000. Merits of using linear static or dynamic and nonlinear dynamic
analyses were discussed. They found that the nonlinear dynamic analysis (taking into account the capability of
redistribution of internal forces) predicts no mechanism which might lead to progressive collapse. Several
plastic hinges were formed but the propagation of failure was arrested.
The consideration of effect of sudden column removal where we ignore the residual capacity of damaged
element may be a conservative assumption. Furthermore, the beneficial effects of arching or catenary action
following local damage are also generally ignored. The value of the dynamic increase factor specified in the
UFC does not consider factors, such as axial forces in beams (i.e. catenary effect) and structural configuration
which are known to have a significant effect on the nonlinear dynamic response. To address some of the above
concerns, Xu and Ellingwood (2010) proposed an energy-based partial pushdown analysis procedure for
analyzing a structural system following local damage.
Fu (2012) conducted nonlinear dynamic analysis of a three dimensional twenty storied composite steel frame
under consecutive column removal condition using ABAQUS. The author analysed the frame for five different
one by one consecutive column/bracing removal conditions and a simultaneous consecutive column removal
condition. It was found that columns being removed one by one caused plasticity in one adjacent beam but
columns being removed simultaneously cause plasticity in both the adjacent beams. And also the structure
designed as per current design code sustains one column removal.
2.3. Stability Consideration in Progressive Collapse Analysis of Structures
Ettouney et. al. (2006) urged the need for a structure to remain globally stable for the development of catenary
action, which plays an important role in resisting progressive collapse. Moment resisting sway and non-sway
frame, both single story 2D frame and multistory 3D frame were considered to assess the global effect in
progressive collapse potential. In widely used alternate path method typically a component design strategy is
employed. The adequacy of the system is based on individual structural components successfully satisfying the
acceptance criteria. The author mentioned that neglecting global consideration may be unsafe as component
level check is not always sufficient. After loss of stiffness due to local failure, system needs to be stable with
degraded properties. Linear static analysis doesn’t consider stability, nonlinear static may consider but nonlinear
dynamic considers both stability and strength.
2.4. Relationship between Seismic Design and Design of Structure against Progressive Collapse
Several studies have been undertaken to investigate the relationship between the seismic design and the
progressive collapse capacity of the buildings (Bao et al., 2008 and Tsai and Lin, 2008). Based upon these
studies, Hadi and Alrudaini (2012) mentioned that buildings designed for high seismic regions performed better
and are less vulnerable to gravity induced progressive collapse than buildings designed for low to moderate
seismic risk.
Khandelwal et al. (2008) investigated the progressive collapse resistance of seismically designed steel moment
frame buildings by using macro-model based simulation. The simulation results show that the frame designed
for high seismic risk has somewhat better resistance to progressive collapse than the frame designed for
moderate seismic risk. The better performance is attributed to layout and system strength rather than the
influence of improved ductile detailing. Macro-model based simulation of progressive collapse was also
extended for RC frame structure in another research paper by Bao et al. (2008).
Whereas, study done by Liu (2011) reveals that buildings designed as per the current seismic provisions do not
necessarily have sufficient capacity to resist progressive collapse, unless it is intentionally designed against
progressive collapse. This is because of the change in buildings behaviour when withstanding an earthquake and
when resisting progressive collapse. Progressive collapse design is more concerned with gravity loads acting on
the structure but seismic design primarily focuses on lateral loads.
811
2.5. Development of Codes and Guidelines for Design of Structure against Progressive Collapse
Recently, design guidelines have been developed to explicitly take progressive collapse into account for
building design. With the passage of time different countries have developed their design codes or modified the
existing codes for design of structure against progressive collapse with some explicit and implicit guidelines. A
brief overview regarding these developments may be discussed as follows.
British Code
To prevent progressive collapse of structure, the major changes in British codes was initiated in 1976. It
demands constructions of five storeys and above to consider the effect of accidents. British code focuses in
development of horizontal and vertical tying of building and where tying is not possible it recommends
developing bridging capability through effective catenary actions of horizontal members. It also prescribes for
the design of connections and for improving efficiency of moment resisting connections.
Canadian Code
National Building Code of Canada (NBCC) included regulations to ensure integrity in structure after the Ronan
Point apartment building collapse. The current code includes provisions for minimum tensile forces in ties and
provisions for local resistance of structural members. It also recommends considering in design all the extreme
loads with probability of occurrence of 10−4/year or more.
United States
Several guidelines released by different US authorities are available for design of structures against progressive
collapse. ASCE/SEI 7 (2010) recommends sustaining local damage to avoid any disproportionate collapse. This
shall be accomplished by providing sufficient continuity, redundancy, or energy-dissipating capacity (ductility),
or a combination thereof, in the members of the structure. Requirement here is threat independent and focus is
on redundancy and alternate load paths. The American Concrete Institute (ACI 2002) includes extensive
requirements for structural integrity. Though there is no explicit mention of redundancy or alternate load paths,
the code focuses on reinforcement detailing to improve integrity and continuity.
(GSA, 2003) recommends to carry progressive collapse analysis for a typical non-exempt buildings (for instance,
any building of over ten stories is nonexempt) in steel or concrete. The nonexempt structure should not fail
when one column or one 30 ft length of load bearing wall is removed.
U.S. Department of Defense (DoD 2005) guidelines try to provide satisfactory continuity, ductility, and
redundancy in buildings so that the spread of local damage can be arrested. Subsequent research and findings
raised questions about the criteria described in the DoD (2005) and hence necessitated a revision (Stevens et. al.
2011) to form DoD (2009). Notable revision were use of load increase factor (LIF) in linear static method and
dynamic increase factor (DIF) for nonlinear static method of analysis instead of a constant load factors used in
DoD 2005 for both linear and nonlinear static analysis.
Indian Code
Recent revision of Indian steel design standard - General Construction in Steel - Code of Practice (IS 800, 2007)
necessitates the structure to be stable and have alternate load paths to prevent disproportionate overall collapse
under accidental loading. Explicit requirements regarding tie force and alternate path method for prevention of
disproportionate collapse are in line with other international standards.
812
3. SEMI-RIGID JOINTED FRAME STRUCTURES
Fully rigid connection transfers the entire moment from beam to column and hence doesn’t allow any rotation
whereas pinned connection cannot transfer any moment as it allows full rotation. Semi-rigid connection partially
transfers the moment allowing certain rotation of connection. Theoretical representation of these connections is
shown in Figure 2.
Figure 2 Joint types according to their behaviour, where θ is the angular rotation between the beam and the
column: (a) pinned, (b) rigid and (c) semi-rigid. (Diaz et. al., 2011)
Figure 3 Types of semi-rigid connections Figure 4 Moment-rotation curves of semi-rigid connections
813
Several types of semi-rigid connections (Figure 3 & 4) are possible, these may be arranged with descending
order of rigidity/ rotational stiffness as T-stub, extended end plate (EEP), flush end plate (FEP) with or without
column stiffener, top and seat angles with or without double/ single web angles, header plate, double web angle
(DWA) or single web angle (SWA) etc.
The most referred code i.e. Eurocode3 (2005) is exclusively dedicated to all types joints including semi-rigid. It
is a collection of decades of research in steel structures and connections. Report published by European
Convention for Constructional Steel Research (ECCS) in 1978, forms basis of the first draft of Eurocode3
published in 1984. Other important code that considers design with semi-rigid connection is AISC-LRFD.
Indian steel design standard IS 800 (2007) also allows design of structures with semi-rigid connections.
Researches done with the semi-rigid jointed frame may be categorized into (i) direct modification of
conventional analysis methods to consider the effect of semi-rigid connections, (ii) numerical modeling of
semi-rigid connections and (iii) physical experiment of semi-rigid connection.
3.1. Analysis Methods for Semi-Rigid Connections and their application
Investigation on steel beam-to-column connections started a century ago when Wilson and Moore (1917) first
investigated the flexibility in riveted joints in 1917. But actual work started with a little gap in 1930’s. Baker
and Rathbun both independently applied slope deflection and moment rotation method in analysis of semi-rigid
jointed frames in 1936. Both of them used connection factor “Z” i.e. rotation per unit moment. The assumption
makes these methods strictly applicable for low value of rotation (Diaz et. al., 2011). In almost same time, Batho
and Rowan proposed beam-line method, where beam line is plotted on actual experimentally calculated
nonlinear moment-rotation relationship curve. Intersecting point represents the end restraint condition of
concerned member. Effect of semi-rigid connection on sway of frame was first investigated by Young and
Jackson in 1934.
Till 1950 focus of research was application of these methods. But analysis of frames with semi-rigid connection
becomes very tedious and cumbersome for even medium sized frame. Development of digital computer in
1950s and its application in structural analysis in 1960s promoted the use of stiffness matrix methods in analysis
of large structural frame with both rigid and semi-rigid connections. In 1961 Lightfoot and Baker developed a
computer program to analyse frame with elastic connection using generalized slope deflection equation in
matrix form. Initial assumption of fully rigid connection was amended by use of correction matrix for semi-rigid
connection. Monforton and Wu were first to consider the effect of semi-rigid connection into matrix stiffness
method in 1963. Romstad and Subramanian investigated the buckling capacity of frame with semi-rigid end
restraint in 1970. Weaver in 1969 and Suko and Adams in 1971 investigated the dynamic behavior of semi-rigid
connection. They used equivalent spring to model the elasto-plastic bevaviour of connection.
Ashraf et al. (2004) investigated the sway behaviour of semi-rigid regular steel frames. A new term ‘flexibility
factor (FF)’ was introduced to consider the ratio of the maximum lateral sway of the semi-rigid frame to that of
the ‘corresponding rigid frame’. Some design charts were developed for computing this flexibility factor which
depends upon the value of beam to connection stiffness ratio, beam to column stiffness ratio, no of storey, beam
length to column height ratio. These design charts are useful in prediction of sway of semi-rigid frame if the
corresponding sway of rigid frame is known. Later Ashraf et al. (2007) extended study for semi-rigid irregular
steel frames.
Hayalioglu and Degertekin (2005) performed a minimum cost design of semi-rigidly connected steel frames.
Study reveals that frames with connections like header plate, DWA and SWA costs 33.8% more as compared to
frames with rigid connections. Whereas frame with other type of semi-rigid connections like T-stub, EEP, FEP
costs 30.5% less as compared to frames with rigid connections. Depending upon the required rotational stiffness,
cost of semi-rigid connection varies from 20% to 59% of the cost of steel quantity for members. Whereas, cost
of rigid connections is nearly 70% of the cost of steel quantity for members. The study shows the importance of
optimizing both the connections and members over only the members.
814
3.2. Experimental Investigation of Semi-Rigid Connections
Starting from 1917, several experiments were carried out by different researcher to know the connection
behavior. Till 1950, the maximum focus was on riveted connections and thereafter it shifted to high strength
bolts. To store these test results for future use, different data banks were created. Information from these tests
viz. geometric and mechanical properties of each component forming the connection, the rotational stiffness,
moment carrying capacity and name of researcher were stored in data bank. A few most important data banks
are Goverdhan data bank (1983), Nethercot data bank (1985), Kishi and Chen data bank (1986).
Goverdhan (1983) collected 230 test results conducted in USA during 1950 to 1983 to form a data bank.
Connections tested were double web angles, single web angles/plates, header-plate, end-plate and top and seat
angle with or without web angles.
Nethercot (1985) developed the first European data bank after examining 700 test results from 70 experimental
tests done by other researcher. This data bank includes the test results of Goverdhan data bank and some test
results on T-stub connections with or without web angles.
Kishi and Chen (1986a and 1986b) extended the work done by Goverdhan and collected the test results for 303
tests conducted all over the world during 1936 and 1986. They developed a computerized data bank system
known as Steel Connection Data Bank (SCDB) program to recover all these test results. A mathematical
formulation for curve fitting of moment-rotation relationship of different connections is also provided. Abdalla
and Chen (1995) extended the data base by adding test results from 46 additional experiments. Jones et. al.
collected and reviewed 323 test results from 29 different studies. Several other small scale experiments followed
for mainly end plate and T-stub connections.
These data banks are widely used in validation of different moment rotation relationship models developed by
different researcher rather than day to day design work. Due to wide variety of geometrical parameters of these
connections there is a very low probability of finding the data for exact connection required by a designer.
Liew et. al. (1997) conducted some benchmark study by physical testing of several semi-rigid jointed frames.
3.3. Modelling of Moment-Rotation Behaviour of Semi-Rigid Connections
The accuracy of the analysis result depends upon how the connection behavior i.e. moment-rotation relationship
is modeled. The relationship may be expressed as linear, bilinear, multi-linear, nonlinear, polynomial, cubic
B-spline, power model, exponential etc. Using these models semi-rigid joints are modeled either by modifying
member end rigidity or by representing this moment-rotation relation by rotational springs (Jones et. al., 1983).
For analysis of 2D portal frames with semi-rigid joints, Ihaddoudene et. al. (2009) adopted mechanical model
where member end flexibility is considered using springs. They used both linear and bilinear relationship.
Figure 5 Linear, bilinear and multi linear moment rotation relationship behaviour
815
Modification of slope deflection and moment rotation method discussed in 3.1 assumes the moment rotation
relationship as linear (Figure5-a) given by Equation (1),
M= Kθ (1)
Where, M is connection moment, θ is connection rotation and K is initial stiffness.
Lionberger in 1969 and Romstad and Subramanianin in 1970 used bilinear relationship model to obtain better
results. A shallow second line representing plastic rotational stiffness was added (Figure 5-b). This model was
further used by Tarpy and Cardinal in 1981, by Melcher and Kaur in 1982 and by Lui and Chen in 1986 (Chen
et.al. Handbook on semi-rigid connections)
Piecewise linear or multi-linear model was first proposed by Razzaq in 1983 (Figure 5-c). These linear models
are easy to use but the inaccuracies and jumps at transition points are undesirable (Abdalla and Chen, 1995).
Polynomial model was first proposed by Frye and Morris in 1975. It is expressed as Equation (2),
θ = C1(KM) + C2(KM)3 + C3(KM)5 (2)
Where, C1, C2 and C3 are curve fitting constants and K is a standardisation parameter. Frye and Morris basically
extended the Somar’s polynomial series function θ = f(KM) developed in 1967, which was introduced for
header plate. Frye and Morris standardized the equation for double web angle, single web angle, header plate,
top and seat angle, end plate with and without web angles and T-stub connections. IS 800 (2007) also refers this
model for generation of moment-rotation relationship of semi-rigid connections. Recommended value of C1, C2,
C3 and K are given in tabular form. Prediction equations for these required parameters developed by Picard et.al
in 1976, Altman et. al. in 1982 and Goverdhan in 1983 are well captured in the semi-rigid connection handbook
prepared by Chen, Kishi and Komuro. Though this polynomial model generates reasonably accurate
moment-rotation behaviour, it gives negative stiffness for some value of moments which may cause problem in
analysis process.
To avoid this problem of negative stiffness B-spline curve fitting technique was developed, where the
polynomial was divided into certain ranges and each ranges was then fitted with cubic curve. Though B-spline
curve describes the connection behavior as very close to actual, it requires a lot of experimental data.
Apart from these linear models several power models were developed by different researcher to define the
nonlinear behavior of moment-rotation relation, which is very close to reality. The two parameter power models
were developed by Batho and Lash in 1936 and by Krishnamurthy et. al. in 1979. Three parameter power
models were developed by Colson and Louveau in 1983 followed by Kishi and Chen in 1987. Whereas, Richard
and Abott proposed four parameter power model in 1975. Analytical equations of these models are given in the
semi-rigid connection handbook prepared by Chen, Kishi and Komuro.
Two parameter power model may be defined by Equation (3),
θ = aMb (3)
Where, a and b are curve fitting parameters with a condition a > 0 and b > 1.
Three parameter power model of Colson and Louveau may be expressed by Equation (4),
| |
= | / |
(4)
Where, Rki is the initial connection stiffness, Mu is the ultimate capacity of connection moment, and n is the
816
shape parameter.
Three parameter power model of Kishi and Chen may be given by Equation (5),
= / / or, = / (5)
/
Where, is reference plastic rotation given by / (Figure 6-a)
Four parameter power model of Richard and Abott (1975) is given by Equation (6),
= / / + (6)
Where, is strain hardening/softening stiffness and is reference plastic rotation given by / ,

being reference relative moment (Figure 6-b).
Figure 6(a) Three parameter power model Figure 6(b) Four parameter power model
Three parameter power model of Kishi and Chen may be obtained by considering strain hardening/softening
stiffness Rkp as zero. Using three parameter power model θ is obtained directly from equation, whereas four
parameter power model involves iterative procedure for finding θ. This is the reason why three parameter model
is used by most of the researcher in their analysis, though it is unable to reflect the strain hardening/softening
nature of connection. Similarly four parameter power model reduces to linear when Rkp tends to Rki and it
becomes bilinear when shape factor n tends to infinity.
Another important and effective nonlinear model in Ramberg-Osgood (1943) model. It was originally proposed
for nonlinear stress-strain relationships by Ramberg and Osgood in 1943 and later standardized by Ang and
Morris in 1984. The M–θ curve of this model is represented as Equation (7),
| | | | #
= 1+! " $ (7)
where (KM)0 and θ0 are constants defining the position of the intersection point A (Figure 7), n is a parameter
defining the sharpness of the curve, and K is a dimensionless factor depending on the connection type and
geometry.
Lui and Chen first proposed multi parameter exponential model in 1975 and it was then modified by Kishi and
817
Chen in 1986 to accommodate sharp changes in the curve. Yee and Melcher also proposed an exponential
equation which allows positive, zero and negative rotational stiffness. This negative stiffness represents post
local buckling failure of the joints (Diaz et. al., 2011).
Figure 7 Ramberg-Osgood model
Generation of the moment rotation curve using these models is possible if we have the value of requisite
parameters i.e. connection stiffness, ultimate moment capacity, shape parameter etc. which are dependent upon
the geometry of the different component of a connection. Kishi and Chen (1990) developed equations to predict
initial stiffness, ultimate moments for single and double web angle connections and top and seat angle
connections with or without double web angle. Shape factor was obtained by least-square curve fitting with
experimental tests. Similar prediction equations for extended end plate connections were provided by
Christopher and Sriramulu (1995). Yee and Melchers (1986) developed equations for end-plate moment
connections.
3.4. Finite Element Modelling (FEM) of Semi-Rigid Connections
FEM tool is used by different researcher for parametric study, developing moment-rotation behaviour for
different connections, validation of prediction equation, analytical and experimental results. Most of the tests
conducted are for in-plane moment rotation relationship; there is hardly any data for twisting moment-rotation
behavior of connections. Also the investigation towards effect of shear and axial force has hardly been
considered in experimental works. Finite element modeling (FEM) has been done to judge these effects.
Urbonas and Daniunas (2006) investigated the effect of axial forces on rotational capacity of end-plate bolted
joint using Cosmos/M FEM code. They loaded the joints in tension or compression with up to 25% of maximum
axial load carrying capacity and found that tension reduces the rotational stiffness of joint whereas it is
increased in compression. FEM analysis of Pirmoz et. al. (2009) on bolted top and seat angle connections using
ANSYS also confirmed that axial tension reduces initial connection stiffness and moment carrying capacity.
Pirmoz et. al. (2008) studied 31 FEM of top and seat angle with double web angle using ANSYS to verify the
effect of web angle dimension on moment-rotation behaviour. The results were verified with 18 experiments
done by Azizinamini. Effect of shear force was considered and based upon the parametric study, they proposed a
second order equation for estimation of rotational stiffness deterioration due to shear force.
3.5. Classification System for Beam-to-Connections
Different guidelines allow design of partially restrained connections (AISC-LRFD, 1994) or semi-rigid
connections (Eurocode 3, 2005), but whether a connection falls under this category may be judged based upon
the classification systems developed over the years. Eurocode 3 has clear demarcation for rigid, semi-rigid and
818
pinned or flexible connections. The nondimensional classification of Eurocode3 for both braced and unbraced
frame is given in Figure 8. Y-axis represents % &= ⁄ whereas X-axis represents ̅ = ⁄ , and
are full plastic moment and plastic rotation of connected beam.
Figure 8 Eurocode 3 classification System: (a) Unbraced frame (b) Braced frame
Bjorhovde et. al. (1990) developed another important classification system in terms of strength, stiffness and
ductility using tests and theoretical data. The nondimensional connection classification diagram of Bjorhovde et.
al. (1990) given in Figure 9(a). The classification system adopted by IS 800 (2007) after Bjorhovde is given in
figure 9(b). Limits used in IS 800 (2007) for strengths and stiffness are also given in Table 1.
Figure 9 (a) Nondimensional Connection classification System by Bjorhovde

(b) Connection classification System of IS 800 (2007) after Bjorhovde
Table 1 Connection classification limit as per IS 800 (2007)

Sl. No. Nature of the connection In terms of strength In terms of stiffness
1 Rigid connection m1> 0.7 m1> 2.5θ1
2 Semi-rigid connection 0.7 > m1>0.2 2.5θ1> m1>0.5θ1
1
3 Flexible connection m < 0.2 m1< 0.5θ1
Where, m1 = ratio of ultimate moment to plastic moment, θ 1= Normalized rotation with

respect to plastic rotation
819
4. PROGRESSIVE COLLAPSE OF SEMI-RIGID JOINTED FRAME STRUCTURES
Most of the researcher has studied progressive collapse behavior of frame structure with rigid connections or a
combination of rigid and pinned connection. Liu et. al. (2010) mentioned that semi rigidly connected frames are
more vulnerable than rigidly connected frames in resisting progressive collapse. They have studied two planar
steel frames to measure the influence of semi-rigid connections and joint damage. Study earlier carried out for
rigid connection (Grierson et. al., 2005) has been extended for semi-rigid connections. They have used a
compound element model developed earlier (Liu et. al., 2008). This compound element accounts for member
nonlinearity and semi-rigid connections by incorporating two springs in series, one each for member
nonlinearity and semi-rigid connections. Stiffness for member plasticity is as mentioned in (Grierson et. al.
2005). Stiffness of connection is adopted from four parameter power model originally proposed by Richard and
Abbott (1975) and further confirmed by validating with experimental data for flush end plate and extended end
plate (Kishi et. al. 2004). Liu et. al. (2010) urged that the study adopted a simple approach but much work is
needed before the phenomenon is understood. In the study a simplified quasi-static approach has been employed
instead of dynamic analysis. 3D model study is required to consider the effect of out of plane moment and
torsion effect, presently being ignored by 2D model. Lim (2004) conducted nonlinear dynamics analysis using
ABAQUS to study the progressive collapse behavior of different 2D and 3D steel frame with rigid, semi-rigid
and reinforced semi-rigid connections. All semi-rigid connected 2D frames collapsed partially by joint failure
whereas in 3D frames rigid and semi-rigid connections did not make a much difference in collapse pattern.
Parametric study for with different spans and stories are also conducted. Valipour and Bradford (2012)
formulated 1D frame compound element for progressive collapse analysis of frame with bolted flush end plate
and extended end plate connections. They used FORTRAN for coding and compared the results with numerical
analysis results done using ANSYS, which were in good agreement.
5. CONCLUSION AND FUTURE SCOPE OF RESEARCH
Development in the field of analysis and design of structure against progressive collapse and study in the area of
semi-rigid jointed frames are discussed. Several analyses are based upon widely used alternate path method as
per GSA or DoD in linear or nonlinear and static or dynamic procedure. Starting point of most of the analysis is
removal of single load bearing element. In the field of design there are certain recommendations from different
codes or guidelines. Notable studies have been carried out for analysis of semi-rigid jointed frames but
progressive collapse of semi-rigid jointed frames still remains an untouched area. In this field, the research
community being in minority, there exist certain areas which are to be covered to know the complex behavior of
the phenomena. This will help to form design guidelines in a better way. Identified potential area of research
may be discussed as below:
a. The load and dynamic increase factors used in codes and guidelines should be a function of allowable
displacement and rotation. Rotation depends upon member nonlinearity and connection behavior. Hinge
property is considered generally as elastic perfectly plastic and connections as fully rigid or somewhere as
pinned. All the progressive collapse studies done with rigid jointed frame should be repeated with semi-rigid
connection to observe the actual behaviour.
b. Research review of semi-rigid connections reveals that most of the researcher considered only in-plane
moment rotation behaviour of connections. To get the actual 2D behaviour we need to incorporate combined
moment-shear-axial deformation. Whereas for actual 3D behaviour three more out plane degree of freedom
which are lateral shear, lateral bending and twisting moment are required. As it is difficult get these
relationships from experimental data FEM may be a good help in this regard.
c. After the failure of load bearing element the structure should remain globally stable to find an alternate path
for resistance against progressive collapse. To satisfy codal requirement the structure is checked at
component level and thereby a limited study is there in the global stability consideration area. Hence a
820
reasonable study should be devoted in global system consideration for both rigid and semi-rigid jointed
frames.
d. Semi-rigid connections are less costly but rotation at joints shall be more. For cost optimisation in moment
resisting frames, an attempt should be made to judiciously use semi-rigid or combination of both rigid &
semi-rigid connections.
e. Notable research has been done in performance based seismic design of structure. But performance based
design guidelines for blast and progressive collapse is yet to be developed. Though not discussed here DoD
(2009) may be treated as starting point for performance based design of progressive collapse. DoD (2009) is
applicable for defense buildings. In the same line criteria for civic buildings may be prepared.
REFERENCES
[1] Abdalla, KM. and Chen, WF.(1995). Expanded database of semi-rigid steel connections. Computers and
Structures, 56(4), 553-64.
[2] ACI 318-08.(2008). Building code requirements for structural concrete and commentary. American
Concrete Institute (ACI).
[3] AISC (2001). Manual of Steel Construction-Load and Resistance Factor Design (LRFD). 2(2), American
Institute of Steel Construction, Chicago, IL.
[4] ASCE/SEI 7. (2010). Minimum Design Loads for Buildings and Other Structures Structural Engineering
Institute. American Society of Civil Engineers, Reston, VA, 424.
[5] Asgarian, B. and Rezvani, F. H. (2012). Progressive collapse analysis of concentrically braced frames
through EPCA algorithm. Journal of Constructional Steel Research, 70, 127-136.
[6] Asprone, D.; Jalayer, F.; Prota, A. and Manfredi, G. (2010). Proposal of a probabilistic model for
multi-hazard risk assessment of structures in seismic zones subjected to blast for the limit state of collapse.
Structural Safety, 32, 25-34.
[7] Ashraf, M.; Nethercot, DA. and Ahmed, B. (2004). Sway of semi-rigid steel frames, part 1: Regular
[8] Ashraf, M.; Nethercot, DA. and Ahmed, B. (2007). Sway of semi-rigid steel frames, part 2: Irregular
[9] Bao, Y.; Kunnath, SK.; El-Tawil, S. and Lew, HS. (2008). Macromodel-based simulation of progressive
collapse: RC frame structures. Journal of Structural Engineering (ASCE), 134(7), 1079-1091.
[10] Bjorhovde, R.; Colson, A. and Brozzetti, J. (1990). Classification system for beam-to-column connections.
Journal of Structural Engineering, ASCE, 116(11), 3059-3076.
821
[11] Chen, W. F.; M. Kishi, N. and Komuro. Semi-rigid connections handbook.
[12] Chopra, A. K. and Goel, R. K. (2002). A modal pushover analysis procedure for estimating seismic
demands for buildings. Earthquake Engineering and Structural Dynamics, 31, 561-582.
[13] Department of Denfense (DoD), (2009). Design of buildings to resist progressive collapse, Unified
Facilities Criteria (UFC). UFC4-023-03, Washington, D.C. (approved for public release).
[14] Department of Denfense (DoD), (2005), Design of buildings to resist progressive collapse, Unified
Facilities Criteria UFC 4-023-03, Department of Defense, USA.
[15] Diaz, C.; Marti, P.; Victoria, M. and Querin, O. M. (2011). Review on the modeling of joint behaviour in
steel frames. Journal of Constructional Steel Research, 67, 741-758.
[16] Ettouney, M.; Smilowitz, R.; Tang, M. and Hapij, A. (2006). Global system considerations for progressive
collapse with extensions to other natural and man-made hazards. Journal of Performance of Constructed
Facilities, (ASCE), 20(4), 403-17.
[17] European Committee for Standardisation (CEN). Eurocode 3 (2005). Design of steel structures, part 1–8:
design of joints (EN 1993-1-8:2005). Brussels.
[18] Frye, M. J. and Morris G. A. (1975). Analysis of flexibly connected steel frames. Canadian Journal of
Civil Engineering, 2(3), 280–91.
[19] Fu, F. (2012). Response of a multi-storey steel composite building with concentric bracing under
consecutive column removal scenarios. Journal of Constructional Steel Research, 70, 115-126.
[20] Goverdhan, AV. (1984). A collection of experimental moment–rotation curves and valuation of prediction
equations for semi-rigid connections. Master thesis. Nashville (TN): Vanderbilt University.
[21] Grierson, D. E.;Xu, L. and Liu, Y. (2005). Progressive-failure analysis of buildings subjected to abnormal
loading. Journal of Computer-Aided Civil and Infrastructure Engineering. 20(3), 155-71.
[22] GSA (2005). Progressive collapse analysis and design guidelines for new federal office buildings and
major modernization projects. U.S. General Services Administration USGSA, Washington, D.C.
[23] GSA (2003). Progressive Collapse Analysis and Design Guidelines for New Federal Office Buildings and
Major Modernization Projects, Office of Chief Architect, General Services Administration, Washington,
DC.
[24] Gurley, C. (2008). Progressive collapse and earthquake resistance. Practice Periodical on Structural
Design and Construction, (ASCE), 13(1), 19-23.
822
[25] Hadi, M. N. S. and Alrudaini, T. M. S. (2012). A new building scheme to resist progressive collapse.
Journal of Architectural Engineering, (ASCE), 1(44).
[26] Hayalioglu, M.S. and Degertekin, S.O. (2005). Minimum cost design of steel frames with semi-rigid
connections and column bases via genetic optimization. Computers and Structures, 83, 1849-1863.
[27] Ihaddoudene, ANT.; Saidani M and Chemrouk M. (2009). Mechanical model for the analysis of steel
frames with semi rigid joints. Journal of Constructional Steel Research, 65(3), 631-40.
[28] Indian steel design standard, IS 800 (2007). General Construction in Steel - Code of Practice. Bureau of
[29] Jones, S.W.; Kirby, P.A. and Nethercort, D.A. (1983). The analysis of frames with semi-rigid
connections-A state-of-the-art report. Journal of Constructional Steel Research, 3(2), 2-13.
[30] Kaewkulchai, G. and Williamson, E. B. (2006). Modeling the impact of failed members for progressive
collapse analysis of frame structures. Journal of Performance of Constructed Facilities, (ASCE), 20(4),
375-383.
[31] Kaewkulchai, G. and Williamson, E. B. (2004). Beam element formulation and solution procedure for
dynamic progressive collapse analysis. Computers and Structures, 82(7–8), 639-651.
[32] Khandelwal, K.; El-Tawil, S.; Kunnath, S. K. and Lew, H. S. (2008). Macromodel-based simulation of
progressive collapse: steel frame structures. Journal of Structural Engineering, (ASCE), 134(7),
1070-1078.
[33] Khandelwal, K. and El-Tawil, S. (2011). Pushdown resistance as a measure of robustness in progressive
collapse analysis. Engineering Structures, 33, 2653-2661.
[34] Kim, T.; Kim, J. and Park, J. (2009). Investigation of progressive collapse-resisting capability of steel
moment frames using push-down analysis. Journal of Performance of Constructed Facilities, ASCE, 23(5),
327-335.
[35] Kishi, N. and Chen, WF. (1990). Moment-rotation relations of semirigid connections with angles. Journal
of Structural Engineering, ASCE, 116 (7), 1813-1834.
[36] Kishi, N. and Chen, WF. (1986a). Steel connection data bank program. In: Structural engineering. 2nd ed.
School of Civil Engineering, Purdue University, West Lafayette, Report no. CE-STR86-18.
823
[37] Kishi, N. and Chen, WF. (1986b). Data base of steel beam-to-column connections. In: Structural
engineering. School of Civil Engineering, Purdue University, West Lafayette, 1/2, Report no.
CE-STR-86-26.
[38] Kishi, N.; Komuro, M. and Chen, W. F. (2004). Four-parameter power model for M-θ curves of end-plate
connections. ECCS/AISC Workshop Connections in Steel Structures V: Innovative Steel Connections,
Amsterdam, The Netherlands.
[39] Kokot, S.; Anthoine, A.; Negro, P. and Solomos, G. (2012). Static and dynamic analysis of a reinforced
concrete flat slab frame building for progressive collapse. Engineering Structures, 40, 205-217.
[40] Kwasniewski, L. (2010). Nonlinear dynamic simulations of progressive collapse for a multistory building.
Engineering Structures, 32(5), 1223-1235.
[41] Li, Q.; Ellingwood, B. R. (2006). Performance evaluation and damage assessment of steel frame buildings
under main shock–aftershock earthquake sequences. Earthquake Engineering and Structural Dynamics,
DOI: 10.1002/eqe.667 (in press).
[42] Lien, K.H.; Chiou, Y.J.; Wang, R.Z. and Hsiao, P.A.(2010). Vector form intrinsic finite element analysis of
nonlinear behavior of steel structures exposed to fire. Engineering Structure, 32,80-92.
[43] Liew, JYW.; Yu, CH.; Ng, YH. and Shanmugam, NE. (1997). Testing of semi-rigid unbraced frame for
calibration of second-order inelastic analysis. Journal of Constructional Steel Research, 41(2–3), 159–95.
[44] Lim, J. (2004). Progressive Collapse Analyses of Steel Framed Moment Resisting Structures. Ph.D. Thesis.
The Graduate School, College of Engineering, The Pennsylvania State University.
[45] Liu, M. (2011). Progressive collapse design of seismic steel frames using structural optimization. Journal
of Constructional Steel Research, 67, 322-332.
[46] Liu, Y.;Xu, L. and Grierson, D. (2008). Compound-element modeling accounting for semi-rigid
connections and member plasticity. Engineering Structures, 30(5), 1292-1307.
[47] Liu, Y.; Xu, L. and Grierson, D. (2010). Influence of semi-rigid connections and local joint damage on
progressive collapse of steel frameworks. Computer-Aided Civil and Infrastructure Engineering, 25,
184-204.
[48] Marjanishvili, S. and Agnew, E. (2006). Comparison of various procedures for progressive collapse
analysis. Journal of Performance of Constructed Facilities,(ASCE), 20(4), 365-374.
824
[49] Marjanishvili, S. (2004). Progressive analysis procedure for progressive collapse. Journal of Performance
of Constructed Facilities, (ASCE), 18(2), 79-85.
[50] McKay, A.;Marchand, K. and Diaz, M. (2012). Alternate path method in progressive collapse analysis:
variation of dynamic and nonlinear load increase factors. Practice Periodical on Structural Design and
Construction, (ASCE), 1, 55.
[51] Menchel, K.;Massart, T. J.; Rammer, Y. and Bouillard, P. (2009). Comparison and study of different
progressive collapse simulation techniques for RC Structures. Journal of Structural Engineering, (ASCE),
135(6), 685-697.
[52] Mohamed, O. A. (2006). Progressive collapse of structures: annotated bibliography and comparison of
codes and standards. Journal of Performance of Constructed Facilities, (ASCE), 20(4), 418-425.
[53] Nethercot, D. A. (1985). Steel beam-to-column connections: a review of test data and its applicability to the
evaluation of joint behaviour in the performance of steel frames. CIRIA report, RP338.
[54] Pirmoz, A.; Daryan, A. S.; Mazaheri, A. and Darbandi, E. H. (2008). Behavior of bolted angle connections
subjected to combined shear force and moment. Journal of Constructional Steel Research ,64,436-46.
[55] Pirmoz, A.; Khoei A. S.; Mohammadrezapour, E. and Daryan, A. S. (2009). Moment-rotation behavior of
bolted top seat angle connections. Journal of Constructional Steel Research, 65, 973-984.
[56] Ramberg, W. and Osgood, WR. (1943). Description of stress–strain curves by three parameters. National
Advisory Committee for Aeronautics, Technical report 902.
[57] Richard, R. M. and Abbott, B. J. (1975). Versatile elastic-plastic stress-strain formula. Journal of the
Engineering Mechanics Division, ASCE, 101(4), 511–15.
[58] Ruth, P.; Marchand, K. A. and Williamson, E. B. (2006). Static Equivalency in Progressive Collapse
Alternate Path Analysis: Reducing Conservatism While Retaining Structural Integrity. Journal of
Performance of Constructed Facilities,(ASCE), 20(4), 349-364.
[59] Seffen, K. A. (2008). Progressive Collapse of the World Trade Center: Simple Analysis. Journal of
Engineering Mechanics, (ASCE), 134(2), 125-132.
[60] Shi, Y.; Li, Z.X. and Hao, H. (2010). A new method for progressive collapse analysis of RC frames under
blast loading. Engineering Structures.32,1691-1703.
[61] Smilowitz, R. (2006). Progressive Collapse: Emerging Challenges for the Design Professional, Journal of
Performance of Constructed Facilities, (ASCE), November Editorial, 307-308.
825
[62] Stevens, D.et.al. (2011).DoD Research and Criteria for the Design of Buildings to Resist Progressive
Collapse. Journal of Structural Engineering, ASCE, 137(9), 870-880.
[63] Tsai, M. H. and Lin, B. H. (2008). Investigation of progressive collapse resistance and inelastic response
for an earthquake-resistant RC building subjected to column failure. Engineering Structures, 30,
3619-3628.
[64] Urbonas, K. and Daniunas, A.(2006). Behaviour of semi-rigid steel beam-to-beam joints under bending and
axial forces. Journal of Constructional Steel Research, 62, 1244-1249.
[65] Valipour, H.R. and Bradford, M. (2012). An efficient compound-element for potential progressive collapse
analysis of steel frames with semi-rigid connections. Finite Elements in Analysis and Design, 60, 35–48.
[66] Wilson, W.M. and Moore, H. F. (1917). Tests to determine the rigidity of riveted joints in steel structures.
University of Illinois, Engineering experiment station, Bulletin 104, Urbana (USA).
[67] Xu, G. and Ellingwood, B. R. (2010). An energy-based partial pushdown analysis procedure for assessment
of disproportionate collapse potential. Journal of Constructional Steel Research,
doi:10.1016/j.jcsr.2010.09.001.
826
Paper No.: 145
RE- EVALUATION OF INDUSTRIAL EQUIPMENTS FOR EARTHQUAKE

AND WIND LOADS
M. K. Agrawal1, A. Ravi Kiran1, G. R. Reddy1, R.K. Singh1, K. K. Vaze1,

D.K.Sakhrodia2, B. Biswas2, A. Bhowmick2
1
Reactor Safety Division, BARC , 2Heavy water Board, Mumbai
Email: mkagra@barc.gov.in
ABSTRACT :
Qualification of the equipments of a industrial plant Kota which is commissioned in 1985 is carried out. Industrial
and nuclear facilities are required to undergo re-qualification when any change in input load due to external load
like earthquake and wind or codal limits on allowable stress are changed. In the initial phase of design seismic
load was not considered due to zone coefficient is null at that point of time. In the year 2002, IS code (IS
1893:2002) [1] has been revised and the zone factor for maximum considered earthquake (MCE) of Kota site
which has been classified as Zone-II is 0.1g. Another extreme loading conditions is wind load for which various
tall towers are required to be qualified. As per the regulatory requirement, wind load with 1000 year return period
are considered for qualification. Analysis has been performed for the wind load and earthquake load for MCE or
DBE as per the codal requirements for qualification. Maximum deflection due to wind and earthquake are
checked against permissible limit.
KEYWORDS: Re-qualification, Earthquake, Wind, Vortex shedding frequency
1. INTRODUCTION
Plant at Kota is built and commissioned in 1985. During the original design period, the applicable national code
was IS-1893-1966 in which the seismic coefficient for zone-I was null. Hence, the load due to earthquake was not
considered in design phase. Industrial and nuclear facilities are required to undergo re-qualification when any
change in input load due to external load like earthquake and wind or codal limits on allowable stress are changed.
In the year 2002, IS code (IS 1893:2002) [1] has been revised and the zone factor for maximum considered
earthquake (MCE) of Kota site which has been classified as Zone-II. The new PGA values for the design basis
earthquake (DBE) is 0.05g and MCE level of earthquake 0.1g . Another extreme loading condition is wind load
for which various tall towers are required to be qualified. As per the regulatory requirement, wind load with 1000
year return period are to be used for qualification. Analysis has been performed for the wind load (for 1000/100
years return period) and earthquake load for MCE or DBE as per the codal requirements for qualification.
2.0 RE-QUALIFICATION PROCEDURE
2.1 Seismic Walkthrough:
Prior to analysis of the equipment, seismic walkthrough on the plant was performed to understand the different
safety related system and its function and role for safety of the plant. This exercise is carried out with team of
experts for classification and categorization of various equipment of the plant.
2.2 Category of Structure as per IS 1893
The IS 1893 code has classified the structures into four categories for seismic qualification. Structures, systems
and components are categorized based on the inventory of H2S [2] according to the definitions of IS1893-2005
[1].
2.3 Allowable Stress Basis:
Allowable stress basis have been used for qualification of equipment and its supports anchorage based on latest
codal limits used for the design purpose. For pressure boundary, stress is limited to 1.5S as defined in ASME
Section VIII, Div-2 [3] for normal load and extreme load combination. For non-pressure boundary parts like skirt
and base plates maximum stress is limited to yield stress under extreme loadings like wind or earthquake for
either of material SA 516, Gr-70 or IS 2062 material. For anchor bolts made up of A 193 -B7 material are
permitted up to 2.4S as per the ASME section III For extreme loading condition.
2.4 Ground Motion Input
The cold tower is classified under category-I structure and hence the damping to be considered is 4%. Structures
under catergory-1 have been checked for maximum considered earthquake (MCE) with 0.1g PGA as well as for
design basis earthquake (DBE) with 0.05g PGA [1]. The response spectral shape used for the design purpose is
based on the RAPS 3&4 is shown in figure 1.[4]
0.50
0.45
2%ζ
0.40
Acceleration (g)
4%ζ
0.35 5%ζ
0.30
0.25
0.20
0.15
0.10
0.05
0.00
0 5 10 15 20 25 30 35
Frequency (Hz)
Figure 1: Response spectra for different damping ratio
2.5 Wind Load Input
Basic Wind Speed (Vb):

Basic wind speed is based on peak gust velocity averaged over a short time interval of about 3 seconds and
corresponds to mean heights above ground level in an open terrain. Figure 1 of IS-875 (Part-3) [5] gives basic
wind speed map of India. For the purpose of this analysis a basic wind speed of 47 m/s is used.
Design Wind Speed (Vz):
The basic wind speed for any site shall be modified to get design wind velocity at any height ( Vz) for the chosen
structure. It can be mathematically expressed as follows:
Vz = Vb*k1*k2* k3 (1)
Where,
Vz = design wind speed at any height z in m/s;
k1 = probability factor (risk coefficient);
k2 = terrain, height and structure size factor; and
k3 = topography factor.
Dynamics effects of wind:
Slender structures whose frequency is less than 1 Hz are checked for dynamic effects of wind load. To account for
828
the dynamic effects of wind load gust factor is calculated for the equipments whose natural frequency is less than
1 Hz. Gust factor is calculated using the formula.
G= 1+gf*r * sqrt( B*(1+phi)*2 + S*E/ (beta))) (2)
Design Wind Pressure:

The design wind pressure at any height above mean ground level shall be obtained by the following relationship
between wind pressure and wind velocity:
Pz= 0.6 Vz2 (3)
Where, Pz = design wind pressure in N/m2 at height z, and Vz = design wind velocity in m/s at height z.
2.6Analysis of Equipment
Analysis has been performed for the wind load and earthquake load as per the codal provision given in IS-1893 for
seismic load and IS-875 for wind load.
3.0 DESCRIPTION OF A TYPICAL TANK
A typical cold tower is considered for the qualification. The cold tower has a height of 53.59 m and diameter of
4.25m. The tower is made of carbon steel, SA516 Gr 70 and is supported on skirt with base plate. The upper
cylindrical portion has 38 mm, middle portion has 40 mm and the lower portion has 42 mm thicknesses. The Top
and bottom hemi-spherical portions have a thickness of 25 and 28 mm respectively.
3.1 Static Analysis:
Tank is model using PVElite[6] software and analyze for static and wind load. PVElite is a PC-based pressure
vessel design and analysis software program. PVElite analyze each element to determine the required wall
thickness for internal and external pressure based on the ASME Section VIII Division-1 or Division-2 rules.
Analysis is performed for thickness subtracted after corrosion allowance. However, all the loads are calculated
considering full thickness of vessel. The analysis results are compared with present thickness obtained by latest in
service inspection (ISI) report. Periodic ISI is carried out to check for any thickness depletion due to corrosion.
3.2 Wind Analysis
Wind load profile is generated along the height in accordance with IS 875 part-3. Wind load profile is used as an
input to calculate element stresses and wind deflection. Deflection of cold tower for 100year return period and
1000year return period wind load are given in Table 1. The maximum allowable deflection is calculated based on
the codal provision[7].
Table 1: Tower deflections under wind loads
Deflection
Wind
Case Actual Allowed
Return
mm mm
Period(RP)
100 28
268
1000 44
829
3.3 Dynamic Analysis Of Cold Tower
3.3.1Finite element model
The finite element model of the Cold Tower is made using cosmos/m [8] shown in Figure 3. There are 6440
elements and 3670 nodes in the model. The tower is modeled using four noded shell elements. The mass of trays is
accounted by providing lumped mass elements at corresponding locations. The base of the model is considered
fixed as boundary condition as the shear wave velocity of the founding strata is 1600 m/s. [4]. The corresponding
material properties of SA516 Gr70 are used for analysis.
3.3.2 Free vibration analysis of cold tower
Analysis has been carried out for the cold tower to obtain the natural frequency. Natural frequencies and mass
participation corresponding to different modes are listed in Table 2. It has been analysed for 100 modes to obtain
the participation more than 90% of mass. The first and seventh flexural mode shape is shown in figures 2 & 3.
Table 2: Natural frequencies and mass participation for cold tower
Y- mass Z- mass
Freque Natural X- mass participation
participation participation
ncy no. Frequency (Hz) (% mass)
(% mass) (% mass)
1 1.34E+00 0.265 9.48E-20 0.364
2 1.34E+00 0.364 7.24E-24 0.265
3 5.40E+00 1.41E-16 1.17E-20 2.06E-18
5 7.78E+00 8.63E-02 4.00E-20 0.114
6 7.78E+00 0.114 9.74E-21 8.63E-02
22 2.11E+01 2.08E-19 0.835 1.50E-19
Cumulative MPF(100 Mode) 0.957 0.835 0.957
Figure 2: First mode shape (1.34 Hz) Figure 3: Seventh mode Shape (7.78 Hz)
830
3.3.2 Response spectrum analysis of cold tower
The cold tower is classified under category-I structure and hence analysed for MCE and DBE level of earthquake
and corresponding damping 4% and 2% are used. The peak ground acceleration considered for horizontal
direction is 0.1g while in vertical direction, 2/3rd of 0.1g is taken. Complete quadratic combination (CQC) method
is used for the modal combination and SRSS method is used for combining the responses in three directions. The
response spectrum analysis is performed for 100 modes. This covered frequencies up to 60.7 Hz. The total mass
participation from all 100 modes is 95.7% in X and Z directions. No missing mass calculation is required [9]. The
Contour plot of resultant Von-Mises stress and resultant displacement is shown in Figure 4 &5 respectively. The
resultant maximum displacement of the tower is 51.7 mm. This displacement is lesser than the permissible
displacement of 268 mm.
Figure 4: Contour plot of resultant Von-Mises Figure 5: Contour plot of resultant

stress displacement
4.0 RESULT AND OBSERVATIONS
The stresses due to internal pressure, operating weight and earthquake load are combined and the combined
stresses are checked with the allowable stresses. Table 3 gives the tensile combined stresses at various locations of
cold tower along with the permissible stresses. The combined stresses due to internal pressure, operating weight
and seismic load for the cold tower are found to be well within the permissible values. The resultant displacement
due to seismic load is well within the allowable limits. Table-4 gives the stress combinations for the earthquake
load with wind load. In one of the case earthquake of MCE level is combined with design wind of 100 return
period and other case wind load of 1000 year return period with DBE level of earthquake. Stresses are found to be
well within the permissible limit.
831
Table 3 : Combined tensile stresses at various locations of cold tower
Combined Tensile
Longitudinal Stress
Stresses
Permissible
PART pressure Operating
±EQ(MCE) Stress IP+OW+EQ(MCE)
(IP) Wt (OW)
min(Sm*1.5,Sy)
kg/cm2 kg/cm2 kg/cm2 kg/cm2 kg/cm2
SKIRT 0 -100 215 2388 115
Bottom
1083 -113 1 2388 971
Head
Shell1 657 -68 145 2388 735
Shell2 696 -46 64 2388 714
Shell3 740 -21 3 2388 722
Top Head 1255 -2 2 2388 1254
Table 4: Combinations for extreme loads

IP+OW+WI100 + IP+OW+WI1000 +
±WI100 ±WI1000 ±EQ(DBE) ±EQ(MCE)
PART EQ(MCE) EQ(DBE)
kgf/cm2 kgf/cm2 kgf/cm2 kgf/cm2 kgf/cm2 kgf/cm2
SKIRT 224 357 127 215 339 384
Bottom
225 359 1 1 1196 1330
Head
Course
138 220 86 145 872 895
(22-15)
Course
68 108 38 64 782 796
(14-07)
Course
15 24 2 3 737 745
(06-01)
Top
0 1 1 2 1255 1255
Head
4.1 Base Plate and Anchorage Check
The Base plate and anchorage are checked for maximum load due to seismic and wind using PVElite. Thickness
requirement for base plate and area requirement for foundation bolts are compared with thickness and area
available.
4.2 Comparison of Vortex Shedding Frequency with Natural Frequency
Cold tower is checked for vortex shedding frequency and compared with natural frequency as shown in Table 5
below. Vortex shedding frequency is defined by Eq-4 below, which is function of velocity of fluid (wind),
diameter of object and Strouhal number (S).
fs = S*U/D (4)
Where, fs is the vortex shedding frequency in Hz, U is the free stream velocity approaching the cylinder and D is
cylinder diameter, Strouhal number is 0.25 for cold tower [5].
832
Table 5: Comparison of vortex shedding frequency with natural frequency
Frequency (Hz)
Natural frequency (Hz) 1.34
Vortex shedding frequency (Hz) 1000 year 4.4
It is observed that natural frequency of tower is away from the Lock-in frequency range. Response is not amplified
due to wind induced frequency.
5.0 CONCLUSION
Re-qualification methodology is described for the industrial plant. Analysis has been performed for the wind load
and earthquake load as per the codal requirements for qualification of equipment. Maximum deflection obtained
from the seismic load and wind is well within the permissible limits. Maximum Stress obtained from the seismic
load is well within the permissible limits. Adequacy of foundation bolts in terms of number and size is also
checked and meeting the requirement for wind load of equipments. Stress combinations for two extreme load due
to wind and earthquake are also provided and found to be safe.
REFERENCES
[1] IS criteria for Earthquake Resistant Design of Structures, part 4 Industrial structures including stack like
structures, IS 1893 (part 4): 2005.
[2] AERB categorization letter, Ref AERB/IPSD/PG/25/2012/dated 03 Sep 2012.
[3] ASME SECTION VIII Division 1 and 2: Rules for Construction of Pressure Vessels, 2010
[4] Dynamic analysis for service building of RAPP 3&4 Units, Vol-2, RAPP 3&4 / 24000/7004/SR/A-(Vol-2)
Nov 1992.
[5] IS 875 : Part 3 : Code of Practice for Design Loads (Other than Earthquake) for Buildings and Structures:
Wind Loads
[6] PVElite Version 2007, Developer - COADE, Inc.
[7] COSMOS/M, 1997 COSMOS/M 2.0, A Complete Finite Element Analysis System Structural Research &
Analysis Corporation Los Angeles, USA.
[8] IS 6533 (Part 2) : 1989 Code Of Practice For Design and Construction Of Steel Chimney
[9] ASCE standard 4-98, Seismic analysis of safety related nuclear structures.
833
Paper No.: 146
UTILIZATION OF DREDGED MATERIAL FROM RESERVOIR OF HYDRO

PROJECT AS FINE AGGREGATE IN CONCRETE
R. K. Sharma1 and Pardeep Kumar2

1
Professor, Dept. of civil Engineering, National Institute of Technology, Hamirpur, India
rksnithp61@gmail.com
2
Associate Professor, Dept. of Civil Engineering, National Institute of Technology, Hamirpur, HP, India.
pardeepkumar.nit@gmail.com
ABSTRACT:
In this experimental study, samples of dredged material from three locations were used as fine aggregate in
different proportions varying from 50 % to 100 % with two local sands (coarse and medium) along with crushed
stone aggregate and Portland Pozzalana cement in ratio of 1: 2: 4 to cast concrete cubes. From these test results,
it can be concluded that the dredged material can be suitably used as a partial substitute for fine aggregate when
mixed with local sands in ratio of 1: 1 for mass concreting works under quality controlled conditions. Further,
the geotechnical properties like compaction characteristics, permeability and shear strength of dredged material
were also studied to explore the possibility of its other uses. It can be concluded that dredged material can be
used as a fill material in reclamation of low-lying areas and for filling eroded portions around hilly river in
conjunction with geotextile materials such as tubes and rip-raps.
KEYWORDS: Compressive strength, dredged material, fine aggregate, geotechnical properties.
1. INTRODUCTION
The Beas Sutlej link project in India diverts 4716 million cubic meters of Beas river waters into river Sutlej
through 38-km long water-conductor system terminating in the balancing reservoir from where the water gets
into Sundernagar-Slappar tunnel. Hundreds of hectare-metre of dredged material dredged by the Bhakra and
Beas Management Board (BBMB) from the balancing reservoir built to feed the Dehar Power House at
Sundernagar, is dumped into the Suketi Khad which flows through the Balh Valley to fall into the Beas river at
Mandi. The dredged material gets spilled over onto the farms on both sides of at least 18-kilometre stretch of the
khad, damaging vegetation and crops as shown in Figure 1. Residents face threat to their lives due to silt and
some lost their lives due to sinking into the dredged material. Some dredged materials can be used for
construction purposes. Because of the growing demand for construction materials and dwindling inland
resources, this may be an important beneficial use. In many cases, dredged material consists of a mixture of
sand, silt and clay fractions, which requires some type of separation process. Depending on the sediment type
and processing requirements, dredged material may be used as concrete aggregates (sand and gravel); backfill
material or in the production of mortar (sand). Murray et al., (1999) illustrated the use of dredged material
excavated from US Naval Station, Mayport for the construction of compressed blocks, which passed ASTM
building block specifications. O'Donnel and Henningson (1999) demonstrated the utilization of dredged material
to produce a cementous grout to seal the abandoned Pennsylvania coal mines to reduce acid mine drainage and
placing fill at Port Liberte golf course. Zeller et al (1999) reported the effective utilization of dredged recycling
material for making economical concrete blocks for the construction of security wall. Millrath et al., (2001)
highlighted the beneficial use of dredged material for different construction purposes. Tribout et al., (2011)
addressed the application of treated dredged sediments as base material for the construction of roads. The
analyses have shown a low bulk density of the dredged material in comparison to that for standard sand.
Fig. 1 Dredged material clogging Suketi khad, Mandi (HP) India.
In the present experimental study, Portland pozzalana cement along with samples of dredged material from three
locations was used as fine aggregate in different proportions varying from 50 % to 100 % with two local sands
(coarse and medium) along with crushed stone aggregate in the ratio of 1 : 2 : 4 (cement : fine aggregate : coarse
aggregate) to cast concrete cubes. The compressive strength tests were performed on cubes in universal
compression testing machine to determine their compressive strength at 7 days and 28 days respectively. The
geotechnical properties like compaction characteristics, permeability and shear strength of the dredged material
have been studied to explore the possibility of its other uses.
In order to explore some engineering solutions to the problems created by silt deposition, samples were
collected from the following three sites: (i) near the dredger outfall where the dredged material is in its original
composition without any finer particles removed, (ii) near Dadaur bridge where substantial portion of the finer
particles have been removed due to washing away by flowing water in Suketi khad and (iii) near Ratti Bridge
(Dhangu village) where dredged material is mixed with local sand in the Suketi and other tributories.
Undisturbed and disturbed soil samples were collected from above three sites to determine geotechnical and
material properties.
3. PROPERTIES OF DREDGED MATERIAL
The geotechnical properties of three silt samples determined as per IS 1498: 1970 , IS: 2720 (Part IV): 1975 and
IS 2720: 1980 for the three samples of the dredged material are given in Table 1. From the permeability values,
it is observed that the dredged material is moderately permeable.
Further, it can be observed from the specific gravity and void ratio of the samples that the critical hydraulic
gradient for the three samples are 0.90, 1.03 and 0.95 at maximum dry density and 0.82, 0.94 and 0.85 at the
field density respectively indicating that in the dredged material quick sand conditions occur. The particle size
distribution curves for the three samples are given in Figure 2. The fineness modulus of the three samples is low
and, therefore, it cannot be used independently as fine aggregate in concrete. But, if mixed with local sand, it
can be used as fine aggregate in preparation of concrete. Maximum dry density and optimum moisture content
of the samples determined from standard Proctor tests is found to be low. The compaction characteristics of the
dredged material are not good as can seen from Figure 3. The maximum dry density for dredged material
samples is very low indicating its poor performance as a construction material in embankments. The angle of
shearing resistance of soil varies from nearly 32˚ to 34˚ with no cohesion indicating that the shear strength of the
material is low. Therefore, the material alone as such cannot be effectively used as a construction material.
835
Table .1 Geotechnical properties of dredged material
S. No. Property Sample 1 Sample 2 Sample 3
3
1 Field density (g/cm ) 1.435 1.613 1.531
2 Water content, w (%) 8.2 12.5 10.3
3 Dry density (ρd) (g/cm3) 1.326 1.433 1.388
Co-efficient of permeability
4 Falling head method x10-3(cm/s) 1.208 2.686 2.003
Constant head method x10-3(cm/s) 1.225 2.668 1.998
5 Specific gravity (G) 2.636 2.668 2.569
6 Particle distribution size curve Figure2 Figure2 Figure2
Compaction characteristics
7 Optimum moisture (%) 4.0 3.8 4.5
Maximum dry density (g/cm3) 1.455 1.642 1.547
Shear strength (at MDD)
8 Cohesion, c 0 0 0
Φ (degrees) 32.1 33.2 34.2
9 Fineness modulus 0.836 0.772 0.755
Fig. 2 Grain size distribution for dredged material
Fig. 3 Compaction curve for dredged material
836
3.1Properties of Fine Aggregate
With the possibility of using the coarser portion of the dredged material as a partial substitute for fine aggregate in
low/normal strength concrete; local sand samples have been collected from two sites in the Suketi Khad. The
properties of sand samples collected from two sites as per IS 383: 1970 are given in Table 2. The grain size
distribution curves for the two sand samples are shown in Figure 4 and the two samples can be classified as SP and
SW.
Table 2 properties of sand samples
S. No. Property Medium sand Coarse sand
1 Fineness modulus 2.02 3.01
2 Silt content 1.40 0.60
3 Classification (IS:1498-1970) SP SW
Fig. 4 Grain size distribution for medium and coarse sand
3.2 Properties of Coarse Aggregate
Coarse aggregate of 20 mm and 10 mm sizes was used. The physical properties of coarse aggregate obtained
below as per IS 2386: Part 4 1963 are given in Table 3.
Table 3 Properties of coarse aggregate

Specific Fineness Impact Abrasion Elongation Flakiness Water
gravity modulus value value index index absorption
2.67 6.61 19.5% 21.4% 12.8 14 0.79%
Fig. 5 Grain size distribution for coarse aggregate
837
The particle size distribution curve for coarse aggregate is given in Figure 5. The fineness modulus of crushed
stone coarse aggregate used in concrete is 6.61. The coarse aggregate and sand samples have been used with
different proportions of dredged material to cast the test concrete cubes.
3.3 Properties of Cement
Portland Pozzalana Cement (PPC) conforming to relevant Indian standard [IS 1489: 1967] was used in the
present study. The test results obtained for Portland Pozzalana Cement (PPC) are given in Table 4. The variation
of compressive strength of cement with age is shown in Figure 6.
Table 4 Properties of cement

Initial setting time Final setting time Fineness Normal consistency
39 minutes 9.5 hours 3.2% 29.5%
Fig. 6 Compressive strength of cement
3.4`Testing of Concrete Cubes
The experimental study includes preparation of samples of desired grading in air-dried conditions as per IS 456:
2000. Each batch of the concrete was tested for consistency immediately after mixing according to the relevant Indian
standard [IS 1199: 1959]. The concrete was filled in 3 layers and each layer was compacted. The test specimens were
stored on the site at a place free from vibration under damp matting for 24 hours and then stored in clean water at a
temperature of 24-30°C. Specimens stored in water were tested for compressive strength immediately after removal
while they were still in the wet condition.
The samples of dredged material have been used as fine aggregate in different proportions with local sands
varying from 50 % to 100 % along with crushed stone aggregate and Portland pozzolana cement (PPC) to cast
concrete cubes. The cement, fine aggregate (comprising of dredged material and local sand) and coarse
aggregate are used in the ratio of 1 : 2 : 4 to cast the test cubes of size 15 cm x 15 cm x 15 cm. Compression
strength tests were performed on cubes in universal compression testing machine to determine their compressive
strength at 7 days and 28 days respectively. The water-cement ratio is kept as per Indian standard IS 456: 2000.
4. DISCUSSION
Figures 7 shows the variation of 7 days compressive strength with percentage of dredged material varying from
50 % to 100 % used to replace the coarse sand. It is observed that the compressive strength at 7 days increases
838
from 8.0 MPa to 14.05 MPa for sample 1 when the percentage of coarse sand in the concrete increases from 0 to
50 %. Similarly, the compressive strength at 7 days increases from 8.05 MPa to 14.40 MPa for sample 2 when
the percentage of coarse sand in the concrete increases from 0 to 50 %. Again, the compressive strength at 7
days increases from 7.05 MPa to 14.30 MPa for sample 3 when the percentage of coarse sand in the concrete
increases from 0 to 50 %. The 7-day compressive strength for the control cube (1 : 2 : 4) is (with 100% coarse
sand as fine aggregate) is 14.90 MPa.
Fig. 7 Compressive strength of concrete at 7 days (coarse sand)
Fig. 8 Compressive strength of concrete at 7 days (medium sand)
50 % to 100 % used to replace the medium sand. It is observed that the compressive strength at 7 days increases
from 7.6 MPa to 12.4 MPa for sample 1 when the percentage of medium sand in the concrete increases from 0
to 50 %. Similarly, the compressive strength at 7 days increases from 7.8 MPa to 15.20 MPa for sample 2 when
the percentage of medium sand in the concrete increases from 0 to 50 %. Again, the compressive strength at 7
days increases from 7.75 MPa to 15.10 MPa for sample 3 when the percentage of medium sand in the concrete
increases from 0 to 50 %. The 7-day compressive strength for the control cube (1 : 2 : 4) is (with 100% medium
839
Fig. 9 Compressive strength of concrete at 28 days (coarse sand)
50 % to 100 % used to replace the coarse sand. It is observed that the compressive strength at 28 days increases
from 10.7 MPa to 20.5 MPa for sample 1 when the percentage of coarse sand in the concrete increases from 0 to
50 %. Similarly, the compressive strength at 28 days increases from 14.4 MPa to 19.1 MPa for sample 2 when
the percentage of coarse sand in the concrete increases from 0 to 50 %. Again, the compressive strength at 28
days increases from 10.4 MPa to 19.20 MPa for sample 3 when the percentage of coarse sand in the concrete
increases from 0 to 50 %. The 28-day compressive strength for the control cube (1 : 2 : 4) is (with 100% coarse
Fig. 10 Compressive strength of concrete at 28 days (medium sand)
Figures 10 shows the variation of 28 days compressive strength with percentage of dredged material varying
from 50 % to 100 % used to replace the medium sand. It is observed that the compressive strength at 28 days
increases from 10.3 MPa to 19.8 MPa for sample 1 when the percentage of medium sand in the concrete
increases from 0 to 50 %. Similarly, the compressive strength at 28 days increases from 13.1 MPa to 20.6 MPa
for sample 2 when the percentage of medium sand in the concrete increases from 0 to 50 %. Again, the
compressive strength at 28 days increases from 10.6 MPa to 21.0 MPa for sample 3 when the percentage of
medium sand in the concrete increases from 0 to 50 %. The 28-day compressive strength for the control cube (1 :
2 : 4) is (with 100% medium sand as fine aggregate) is 22.10 MPa. From these test results, it can be concluded
that the strength of 1 : 2 : 4 (cement : sand : coarse aggregate) concrete cubes is within acceptable limits as per
Indian standards for using this concrete as a construction material particularly in mass concreting works when
fine aggregate (coarse or medium sand) is partially replaced with dredged material. Therefore, it can be
concluded that the dredged material particularly near the NH-21 bridge over Suketi khad can be suitably used
840
when mixed with the local sand in ratio of 1 : 1 for mass concreting works under quality controlled conditions.
4.1 Other Uses of Dredged Material
The dredged material can be utilized for reclamation and fill foundation purposes. The dredged material has low
placement density as such it cannot be advisable to use it as a reclamation material. However, in combination
with river-bed material (coarse aggregates) and appropriate usage of geotextiles, it can be used for reclamation
purposes. Since dredged material has poor compaction characteristics (Figure 3); therefore, it may be used with
other soils i.e., sandy clays (SC), etc. having good compaction characteristics to construct embankments.
5. CONCLUSIONS
From the compressive strength test results on concrete cubes at different percentages of dredged material from
three locations mixed with local fine aggregates and coarse aggregates, it can be concluded that the dredged out
material may be effectively used as a partial substitute for fine aggregate in the construction of structures having
moderate importance such as retaining walls, check dams, parapets, drains, boundary walls etc., where the
strength is not a deciding factor but the bulk of concrete is needed. Also, in proper proportions (50 : 50 ratio)
with local sand, it can be used for making concrete of moderate strength as can be seen that 1 :2 : 4 cement
concrete possesses 28 day compressive strength more than 20 MPa. Considering the shear strength test results, it
is observed that the material is having negligible amount of cohesion and the angle of internal friction lies
between 32º-34º which permits its use in backfills in combination with river-bed material. Also, the dredged
material can be used in the fill foundation in appropriate proportions with other soils.
REFERENCES
[1] IS 383: 1970. Specification for coarse and fine aggregates from natural sources for concrete, Bureau of
Indian Standards, Manak Bhawan, New Delhi.
[2] IS 456: 2000. Plain and reinforced concrete - Code of practice, Bureau of Indian Standards, Manak Bhawan,
New Delhi.
[3] IS 1199: 1959. Methods of sampling and analysis of concrete, Bureau of Indian Standards, Manak Bhawan,
New Delhi.
[4] IS 1489: 1967. Specifications Portland Pozzalana Cement (PPC), Bureau of Indian Standards, Manak
Bhawan, New Delhi.
[5] IS 1498: 1970. Classification and identification of soils for general engineering purposes, Bureau of Indian
Standards, Manak Bhawan, New Delhi.
[6] IS: 2720 (Part IV): 1975. Methods for test for soil grain size distribution, Bureau of Indian Standards,
Manak Bhawan, New Delhi.
[7] IS 2720: 1980. Methods of test for soils, Part VII: Determination of water content-dry density relation
using light compaction, Bureau of Indian Standards, New Delhi.
[8] IS 2386 (Part 4): 1963. Methods of test for aggregates for concrete Part 4 - Mechanical properties, Bureau
of Indian Standards, Manak Bhawan, New Delhi.
[9] Millrath, K. S., Shimanovich Kozlova, S. and Meyer, C. (2001). Beneficial use of dredged material 2,”
Second Progress Report prepared for Echo Environmental, Inc., Columbia University, New York, NY.
841
[10] Murray, N. K., and Associates. (1999). Compressed blocks from dredged material from US Naval Station.
Mayport. FL CDF, Contract Report. US Naval Station, Mayport, FL.
[11] O'Donnel, S. and Henningson, J. (1999). The beneficial use of dredged material to mitigate acid mine
drainage, Proc. of the 19th Western Dredging Association (WEDA XIX) Annual Meeting and Conference
and 31st Texas A&M University Dredging Seminar (TAMU 31), Louisville, Kentucky.
[12] Tribout, C., Husson, B. and Nzihou, A.(2011). Use of treated dredged sediments as road base materials:
Environmental assessment, Waste Biomass Valor, 2, pp-337–346.
[13] Zeller, G. Murray, N. K. and Lee, C. R.(1999). Recycling materials: Eco-blocks, American Society of
Agronomy Abstracts, p13.
842
Paper No.: 147
EFFECT OF SILICA FUME ON STRENGTH OF

PARTIALLY USED RECYCLED COARSE AGGREGATE CONCRETE
1 2
Dr. D. V. Prasada Rao and G. V. Sai Sireesha
1
Associate Professor, Dept. of Civil Engineering, Sri Venkateswara University College of Engineering,
Tirupati, Andhra Pradesh, India
2
PG Student, Dept. of Civil Engineering, Sri Venkateswara University College of Engineering,
Tirupati, Andhra Pradesh, India
Email: dvpr@rediffmail.com
ABSTRACT:
This paper presents the results of an experimental investigation on the strength characteristics of concrete
prepared by the partial use of recycled coarse aggregate (RCA) and silica fume (SF). The recycled coarse
aggregate partially replaces the natural coarse aggregate and the silica fume partially replaces the cement. In the
present investigation, recycled coarse aggregate was used as 25% and 50% replacement of natural coarse
aggregate in addition to silica fume partially replaces 5%, 8% and 12% of cement by weight. To evaluate the
M20 grade concrete prepared by the partial application of RCA and SF for structural applications, various
strength characteristics of concrete were obtained. The compressive strength, split tensile strength and modulus
of rupture test results of concrete prepared using various percentages of recycled coarse aggregate and silica
fume are presented. Experimental investigations indicated that the strength properties of partially used recycled
coarse aggregate concrete can be improved by the addition of certain percentage of silica fume. Based on the
test results, it can be concluded that a combination of 25% recycled coarse aggregate and 8% silica fume can be
recommended for the structural concrete, because its strength characteristics are better than the natural coarse
aggregate (NCA) concrete.
KEYWORDS: Recycled Coarse Aggregate, Silica Fume, Partial Replacement, Compressive Strength, Split
Tensile Strength and Modulus of Rupture.
1. INTRODUCTION
Concrete is the most common construction material. The main constituents of concrete are cement, natural
aggregates and water. Due to increase in the various construction activities the natural resources are fast
depleting. During the natural calamities like earthquakes the collapse of reinforced concrete buildings causes a
serious concern with respect to the disposal very large quantities of demolished concrete waste material. For the
disposal of the demolished concrete waste material the availability of open land is also a problem. To preserve
the natural resources for our future generations and also to overcome the problem of disposal of the demolished
concrete waste material, the concrete waste can be used as coarse aggregate known as recycled coarse aggregate.
Recycling of the concrete waste can be considered as an excellent source of coarse aggregate for the production
of new concrete. It has been observed that the recycled coarse aggregate obtained from the concrete waste is
rough-textured, angular and has higher water absorption than the natural coarse aggregate. The density of
recycled coarse aggregate is less than the natural coarse aggregate because of the porous and less dense residual
mortar is adhering to its surface. The preliminary studies indicated that the parial use of recycled coarse
aggregate in the concrete preparation results in the reduction of its strength characteristics. An attempt has been
made to study the effect of silica fume on the strength of partially used recycled coarse aggregate concrete. The
present experimental investigation is focused on the effect of silica fume as partial replacement of cement and
recycled coarse aggregate as partial replacement of natural coarse aggregate on the strength characteristics of
concrete. Compressive strength, split tensile strength and modulus of rupture tests were conducted on concrete
specimens prepared with partial use of recycled coarse aggregate and silica fume and then the results were
compared with the natural coarse aggregate concrete.
2. MATERIALS
2.1. Cement
In the present investigation Ordinary Portland cement (OPC) of 43 Grade confirming to IS specifications was
used. The specific gravity of the cement is 3.15.
2.2 Fine aggregate
Locally available river sand (Zone - II) confirming to IS specification with fineness modulus of 2.7 was used as
the fine aggregate in the concrete mix.
2.3 Recycled Coarse Aggregate
The recycled coarse aggregate used in the present investigation consists of a mixture of coarse aggregate
obtained by crushing the tested concrete cubes of three different grades, M 20, M 25 and M 30, cast with natural
coarse aggregate. The crushing process affects the characteristics of the resulting recycled coarse aggregates.
The nominal sizes of recycled or natural coarse aggregate used in the present investigation were 20 mm and 10
mm. The properties of the coarse aggregate are given in the table. 1. The tests were conducted as per IS
specifications. The recycled coarse aggregate was found to be angular and porous compared to the natural
coarse aggregate. Further, RCA is to be cleaned and should be free from detrimental levels of impurities and
other harmful constituents.
Table1. Properties of NCA, RCA and Sand

Property Natural Coarse Recycled Coarse Fine aggregate (Sand)
Aggregate Aggregate
Specific Gravity 2.67 2.48 2.52
Water Absorption 0.56 1.88 0.44
2.4 Silica Fume
Silica fume is a by-product of ferro-silicon industry. Compared to the cement, the particle size of silica fume is
very finer. It acts as an excellent pore-filling material. Silica fume is usually considered as a supplementary
cementitious material. Generally, it can be used in the proportion of 5 – 12% of the cement content. The
properties of silica fume used in the present investigation are shown in the table. 2.
Table 2. Properties of Silica Fume

S.NO. Characteristics Actual Analysis Result
1 Sio2 90.20%
2 Moisture 0.20%
3 Pozzolana Activity Index 127%
4 Specific Surface Area 21m2/gm
5 Bulk Density 604
844
2.5 Water
Potable water fit for drinking was used for the preparation of concrete and also for curing the test specimens.
In the present experimental investigation, the effect of combined use of recycled coarse aggregate as partial
replacement of natural coarse aggregate and silica fume as partial replacement of cement on various strength
characteristics of M20 grade of concrete is studied. Concrete samples were prepared with 25% and 50% of
recycled coarse aggregate content in addition to 5%, 8% and 12% of silica fume.
3.1 Concrete Mix Proportions
M20 grade natural coarse aggregate concrete was designed based on IS method of mix proportioning. In the
concrete preparation, 10 mm and 20 mm size coarse aggregate were used in the ratio of 1:1.5. No
superplasticiser was used in the concrete mix. The mix proportion of M20 grade concrete is shown in table. 3.
Concrete mix was prepared with cement content of 320 kg/m3. After obtaining the concrete mix proportion, the
recycled coarse aggregate was used as 25% and 50% by weight replacement of the natural coarse aggregate in
addition to 5%, 8% and 12% of silica fume. The quantities of various ingredients of the concrete mix were
obtained by assuming the aggregate were in a saturated-surface dry condition and appropriate adjustments were
made to cater for the different water absorption properties of the aggregates.
Table 3. Mix Proportion of M 20 Grade Concrete

Cement Fine Aggregate Coarse Aggregate Water-Cement Ratio
1 1.97 3.79 0.55
3.2 Concrete Specimens Preparation
A total of four sets of concrete specimens were prepared for different curing periods, 3, 7, 28 and 56 days. Each
set of concrete specimens consists of 150 mm × 150 mm × 150 mm cubes, 150 mm × 300 mm cylinder and 100
mm × 100 mm × 500 mm prisms. Concrete cubes and cylinders were used to get the compressive strength.
Concrete cylinders were also used to obtain the split tensile strength and the concrete prisms were used to get
the modulus of rupture.
3.3. Tests on Concrete Specimens
Compressive strength, splitting tensile strength and modulus of rupture tests were conducted on concrete
specimens. The compressive and split tensile strengths of concrete were obtained using a compression testing
machine. The rate of loading for the tests is in accordance with the Indian Standard specifications. The
compressive strength tests on concrete cubes and cylinders were carried out at the age of 3, 7, 28 and 56 days.
The split tensile strength and modulus of rupture tests were conducted at the age of 28 days.
The compressive strength, split tensile strength and modulus of rupture test results of concrete prepared with
25% and 50% of recycled coarse aggregate in addition to 5%, 8% and 12% of silica fume content indicates
exactly the same trend.
845
4.1 Compressive Strength
The test results of the cube compressive strength of M20 grade concrete for various combinations of recycled
coarse aggregate and silica fume at different curing periods are shown in figure 1. Each value of the cube
compressive strength indicates the average value of three test results. It can be observed that as the age of
partially used recycled coarse aggregate concrete with silica fume concrete increases its cube compressive
strength is also increases. Improved cube compressive strength of recycled coarse aggregate concrete can be
observed at 8% of silica fume.
(a) 25 % RCA
(b) 50 % RCA
Figure 1.Variation of cube compressive strength with age of M 20 grade partially used
recycled coarse aggregate concrete with different percentages of silica fume.
846
The variation of 7 and 28 day’s cube compressive strength of 25% and 50% recycled coarse aggregate
concrete with increase in the percentage of silica fume is shown in figure 2.
(a) 7 Day’s
(b) 28 Day’s
Figure 2. Variation of cube compressive strength with age of M20 grade partially used
recycled coarse aggregate concrete with different percentages of silica fume.
4.2 Split Tensile Strength
The variation of split tensile strength of M20 grade concrete with 25% and 50% of recycled coarse aggregate
content with increase in the percentage of silica fume is shown in figure 3. It is also observed that the split
tensile strength of recycled coarse aggregate concrete with silica fume is around 10% of its 28 day’s cube
compressive strength. The split tensile strength of M 20 grade natural coarse aggregate concrete is 3.66 N/mm2.
847
Figure 3 .Variation of split tensile strength of M20 grade partially used
recycled coarse aggregate concrete with silica fume.
4.3 Modulus of Rupture
The variation of modulus of rupture of M20 grade concrete with 25% and 50% of recycled coarse aggregate
content with increase in the percentage of silica fume is presented in figure 4. It is also observed that the
modulus of rupture of recycled coarse aggregate concrete with silica fume is varying between 12% and 20% its
28 day’s cube compressive strength. The modulus of rupture of M20 grade natural coarse aggregate concrete is
4.2 N/mm2.
Figure 4.Variation of Modulus of Rupture of M20 grade partially used

recycled coarse aggregate concrete with silica fume.
848
The variation of various strength characteristics, compressive strength, split tensile strength and modulus of
rupture, of recycled coarse aggregate concrete with silica fume indicates the same trend. Using the test results, it
can be concluded that for any given percentage of recycled coarse aggregate, as the silica fume content is
increased the various strength characteristics concrete are also increases up to 8% of silica fume content and
then the strength of concrete decreases. The increase in the strength of partially used recycled coarse aggregate
concrete and silica fume can be attributed to the availability of additional binder in the presence of silica fume.
As the silica fume content is increased beyond 8% the strength of the additional binder decreases leading to the
reduction in the strength of partially used recycled coarse aggregate concrete.
The experimental investigations indicates that the improved strength characteristics of partially used recycled
coarse aggregate concrete can be obtained by the addition of silica fume compared to the concrete prepared with
the partial use of recycled coarse aggregate only. The influence of varying percentages of recycled coarse
aggregate and silica fume on M20 grade concrete indicates that the combined application of 25% replacement of
natural coarse aggregate by recycled coarse aggregate and 8% replacement of cement by silica fume possess
improved the strength characteristics than natural coarse aggregate concrete of same grade.
REFERENCES
[1] Bairagi N K, Vidyadhara H S and Kishore R (1990) Mix design procedure for recycled aggregate concrete,
Construction and Building Materials, 4 (4), 188-193.
[2] Fouad M. Khalaf and Alan S. DeVenny (2004) Recycling of Demolished Masonry Rubble as Coarse
aggregate in Concrete, Journal of Materials in Civil Engineering, ASCE, 16(4), 331-340.
[3] IS: 383:1970 (Reaffirmed 1997) Specification for Coarse and Fine Aggregates from Natural Sources for
Concrete. Bureau of Indian Standards, New Delhi.
[4] IS 456:2000. Plain and Reinforced Concrete Code for Practice, Bureau of Indian Standards, New Delhi, India.
[5] IS 516:1959. Methods of Tests for Strength of Concrete, Bureau of Indian Standards, New Delhi, India.
[6] IS: 8112:1989. 43 Grade Ordinary Portland Cement – Specification. Bureau of Indian Standards, New
Delhi.
[7] IS: 10262-2009. Concrete Mix Proportioning – Guidelines (First Revision). Bureau of Indian Standards,
New Delhi.
[8] Kou S C and Poon C.S (2008). Mechanical Properties of 5-year old concrete prepared with recycled
aggregates obtained from three different sources. Magazine of Concrete Research, 60(1), 57-64.
[9] Padmini, A.K; Ramamurthy.K. and Mathews, M.S. (2009)Influence of parent concrete on the properties of
recycled aggregate concrete. Construction of Building Materials, 23, 829-836.
[10] Yajun Ji and Jong Herman Cahyadi (2003) Effects of densified silica fume on microstructure and
compressive strength of blended cement pastes. Cement and Concrete Research, 33, 1543-1548.
849
Paper No.: 148
SEISMIC RESPONSE OF VERTICALLY IRREGULAR STRUCTURES:

ISSUES TO BE ADDRESSED
Pratika Preeti1
1
1st Year, M. Tech, School of Infrastructure, Indian Institute of Technology, Bhubaneswar, Odisha, India
Email- pratika.preeti@gmail.com
ABSTRACT:
Structures are vulnerable under seismic loading. However, for safe seismic design of structures, height based
period formulas are prescribed in various codes. In reality, most of the structures have irregularity because of
functional, architectural and other reasons. The present study makes it a point that neither a single height based
period formula may not be applicable for such structures, nor the complexity of mode shape resulting in local
stress concentration is adequately accounted in design provisions. Further, performance based design for such
structures also pose a special challenge while applying pushover analysis as a relatively practical alternative to
rigorous non-linear dynamic analysis for predicting inelastic seismic behaviour. While pushing the structure
maintaining its first mode shape may yield an acceptable results predicting inelastic behaviour of structures with
regular configuration an adequate combination of various initial mode shapes may be a viable alternative for
predicting inelastic performance of such irregular structures. In view of very limited research effort available on
this topic, targeted as well as systematic research in these directions is the need of the hour.
KEYWORDS: Vertical Irregularity, Non Uniform Mass Distribution, Non Uniform Stiffness Distribution,
Seismic Behaviour
1.0. INTRODUCTION:
Irregular configuration of buildings is inevitable due to various architectural, functional and economic
constraints. Irregularity exists in either or both horizontal and vertical direction. The present study aims to
analyse the works in field of seismic behaviour of vertically irregular structures.
Vertical discontinuities in distribution of mass, stiffness or strength as well as geometrical inconsistency leads to
vertical irregularity in structures. A horizontal offset in the plane of the exterior wall of a structure called as set
back is an example where both mass and stiffness vary with height.
The research in this area gained momentum from 1970. Previous studies like research efforts of Al-Ali and
Krawinkler (1998), Chopra and Chintapakdee (2004) consider different kinds of structural irregularities at
varying heights with separate combination of mass, stiffness and strength. Some studies (Valmundsson and
Nau,1997) show various seiemic responses of vertically irregular buildings. Other set of research (Cheung and
Tso, 1987; Tso and Yao, 1994) have explored force distribution in setback structures at bent locations where as,
(Duan and Chandler, 1995 examined the adequacy of building codes for safe design for a kind of eccentric
setback buildings). However, this area of seismic response of vertically irregular structure has not been studies
extensively. Many parameters like evaluation of lateral period for setback structures, analysis of stress
concentrations in setback structures, adoption of most accurate method for analysis are still not evaluated
accurately. In this context, the paper is an attempt to highlight such topics which need further research.
2.0. CODAL PROVISIONS:

Earlier versions of IS 1893 (BIS, 1962, 1966, 1970, 1975, 1984) do not mention about vertically irregular
structures. The same has been mentioned in detail in the recent edition of IS 1983 (Part 1) – 2002 (BIS 2002). It
categorizes the same to 5 types as shown in the Figure 1 namely stiffness irregularity (soft story), mass
irregularity, vertical geometric irregularity (set-back), in-plane discontinuity in lateral-force-resisting vertical
elements, and discontinuity in capacity (weak story).
In other codes like NEHRP (BSSC, 2003), similar classification as above has been suggested according to
which irregularity is defined when one of the parameters (mass, stiffness or strength) exceeds a certain
minimum prescribed value in adjacent stories. Further, various building codes suggest the use of dynamic
analysis (elastic time history or elastic response spectrum analysis) for calculation of lateral force distribution in
place of equivalent lateral force methods.
(a) (b)
A
A/L > 0.1

A A
A A
A/L > 0.25
A/L > 0.15 L
L
L
(c)
a
b
Upper floor Lower floor

(d)
(a) Figure 1 Some common forms of vertical Irregularities; (a) Elevation of multistoreyed building showing
irregularity in stiffness distribution; (b) Elevation of multistoreyed building showing irregularity in mass
distribution; (c) Vertical Geometric irregularity or setback (d) In plane discontinuity in lateral force resisting
vertical elements when b>a: plan view (Soni, D.P. and Mistry B.B. (2006): Qualitative review of seismic
response of vertically Irregular building frames)
851
3.0. IMPORTANT ISSUES :
As mentioned earlier, this field of seismic response of irregular structures along vertical direction requires a
more exhaustive study to establish the analysis methods. This section aims to highlight certain issues which
need to be addressed for easier study of the same.
3.1. Suitable Lateral Period Relation:
One of the previous studies (Valmundsson and Nau,1997) concluded that the change in period in case of
structures with mass irregularity along the vertical direction. The study considered earthquake response of 5-,
10- and 20- storey framed structures modelled as two-dimensional shear buildings. Considering un shifted
period leads to overestimation of storey shears. The study estimates relative shift in period by the relation:
3
= / −1 /
4
Where Mnw = mass of non-uniform structure, Mu= mass of uniform structure and N= no of stories. The period of
the non-uniform structure Tnu is then calculated in terms of the period of uniform structure T as
= 1 +

However, distribution of mass can vary vertically with different proportions. Further study needs to be
conducted considering different combinations of mass irregularity. Some cases can be presented as
accumulation of higher masses in lower 2 or 3 continuous storey and lower masses above them or vice versa.
Figure 2 illustrates few possible conditions.
(a) (b)
Figure 2 Irregularity in distribution of mass along vertical direction: a) Lower masses are heavier b) Upper
masses are heavier
Setback structures also face an anomaly in consideration of lateral period. Indian Standard IS 1893 (Part 1 –
2002) gives an empirical relation to find the lateral period of various types of structures:
Ta = 0.075 h 0.75 for RC framed building
Ta = 0.085 h 0.75 for Steel frame building
where Ta = Lateral period of structure, h = height of the structure.
This equation creates confusion in case of setback structures where the height varies along the two
sides of the building. Some typical cases are illustrated in Figure 3.
852
H1
H1
H2
H3
H2
H4
Figure 3 Height Variation in a Setback structure
In detailed study is needed to be conducted so as to finalise the height or else it would lead to over or under
estimation of storey shear.
3.2. Stress Concentration At Various Locations Of Irregularity:
Several studies have been conducted on setback structures (Humar and Wright, 1977; Aranda, 1984; Moehle
and Alarcon, 1986; Shahrooz and Moehle, 1990; Wong and Tso, 1994). A setback structure has a sudden
change in the vertical distribution of mass, stiffness or strength. It is made up of two parts: a base i.e. the lower
part with many bays and the tower i.e. the upper part with fewer bays.
A study (Humar and Wright,1997) studied seismic response of steel frames with setbacks under a single ground
motion and found that the story drift is larger in tower parts of setback as compared to that of regular structures.
Similarly, smaller storey drifts were observed in base parts as compared to regular structures. They observed
that the difference in elastic and inelastic story drifts between set-back and regular structures depends on the
level of story considered. Altered displacements and high ductility demands were found in the vicinity of
irregularities.
Several studies offer contradicting views over the response of setback structures under seismic excitations. In a
comparative study (Aranda,1984) between ductility demands of setback and regular structures while using
ground motions on soft soil, setback structures were found to have a higher ductility demand, with a substantial
increase in the tower portion. An analytical study (Shahrooz and Moehle, 1990) concluded higher concentration
of damage in the tower portion owing to high rotational ductilities. They also inferred from an experimental set
up that the fundamental mode dominates the response in the direction parallel to the setback. However, the
study (Wood,1992) offered contradictory result that the response of setback structures did not deviate much
from that of regular structures. Further analysis is required to find a consistent result regarding seismic
behaviour of setback structures.
A study (Wong and Tso, 1994) on the seismic response of setback structure using elastic response spectrum
analysis observed higher modal masses of higher modes to be larger for setback structure leading to deviation in
lateral load distribution from the static code provisions. Detailed study in this regard is necessary to account for
varying mass of setback structures and thus the increased seismic load.
As mentioned above, change in mass, stiffness or strength along the height leads to vertical asymmetry. This
generates eccentricity (on each floor). This presents as a complex problem considering the rotational aspect as
well. In addition to this, the discontinuity leads to sudden change in strength and stiffness along the line of
setback or notch. These notches act as a re-entrant corner where the stresses have to circle around it, thus
hindering the direct passage. This creates stress concentration along the discontinuity joints. Much research has
not progressed in this area considering the effect of stress concentration on lateral forces.
853
Various types of setbacks within the limitation of 2 storeys 2 bay frames to 3 storey 3 bay frames have been
studied in a research endeavour (Dutta et al, 2010). The study highlighted that the stress concentration created
due to the presence of setbacks may lead to higher accumulation of stress quantities at some structural elements-
that may be twice as the amount in case of its regular counterpart. It was also observed that the setback
buildings have lesser base shear as compared to their regular counterparts due to lesser mass concentration in
case of setbacks. However, these issues need further explorations considering multi storeyed frames wit larger
number of stories and bays.
These days, the structures are designed for multiple limit states (Operational Level, Immediate Occupancy
Level,, Life Safety Level, Collapse), where they are allowed to undergo a stipulate amount of inelastic
deformation corresponding to a particular limit state in accordance with the socioeconomic and functional
requirements. This section presents a brief explanation about the philosophy of such performance based design.
However, such approach needs evaluation of inelastic response with accurate adequacy. Non Linear Dynamic
analysis is an exhaustive method. On the other hand, Modal Pushover Analysis serves as an easier yet
acceptably accurate alternative. Inelastic response of regular structures, through the Non Linear Dynamic
Analysis and Pushover Analysis is simpler, while for vertically irregular structures, the same posses as a
challenging issue. Following section is an effort to highlighting these aspects.
3.3. Application Of Philosophy Of Performance Based Design:
The concept of Performance based design emerged recently which promotes to strike a balance between
economy and safety of the structures. The design process allows the structure to undergo a fixed amount of
inelastic deformation when subjected to the reversible seismic loading. However, a detailed non linear dynamic
analysis is necessary to estimate the extent of inelastic deformation undergone by the structure and the
corresponding level of seismic ground excitation. The process involves exhaustive computational approach and
rigor even for the steel frame structures which exhibit an elasto-plastic or bilinear behaviour. In case of
Reinforced concrete structures, strength and stiffness degradation and pinching is observed which adds to the
complexity of the problem. Owing to these limitations, a relatively simpler approach – Pushover Analysis is
proposed to accomplish performance based design.
3.3.1. Pushover Analysis

Un
Capacity Sa Capacity
Vb Curve Spectrum
Vb
Un Sd
Performance Point
Capacity Spectrum
5% damped Response
spectrum
Reduced Response considering

equivalent increased damping
Figure 4 Overall Procedure of Modal Pushover Analysis (Source : Dutta, S.C., Presentation for Advanced
Structural Dynamics and Earthquake Engineering Course Lecture Series)
854
Pushover Analysis combines the processes of Non-linear Static analysis and Linear Dynamic Analysis with
increased damping to account for the hysteretic energy dissipation as shown in the Figure 4.
In this process, Non-linear static analysis is carried out by applying n external force which deforms the structure
maintaining the first mode shape. The process gives an accurate result in case of uniform distribution of mass
and stiffness along the height of the structure. However, in case of vertically irregular structures( with
discontinuity in mass, stiffness or strength along the height), the issue emerges in form of finalising the shape
for which Nonlinear Static analysis is to be carried.
A limited number of study (Al-Ali and Krawinkler, 1998; Chintanapakdee, and Chopra, 2002, Chopra, and
Goel,2002 etc) have been made in this context.
It has been observed in one such study (Chintanapakdee and Chopra, 2004) that for vertically irregular
structures, first mode does not give an accurate account of the structural behaviour. A refined methodology -
Modal Pushover Analysis was suggested in which a combination of first few modes is considered for an
accurate analysis. However, the study observed a considerable variation in the finalising the combination of the
primary modes, depending upon the nature of irregularity. This emerges as a challenging issue to outline a
guideline regarding the participation of these primary modes to arrive at an accurate result through the pushover
analysis. Detailed investigation needs to be conducted in this regard.
4.0. CONCLUSIONS:
The present paper is an attempt to review the previous researches in the area of seismic behaviour of vertically
irregular structures. Following salient issues may be noted regarding the further study to be carried out in this
direction.
1. It has been found that period formula available in codes may be difficult to be applied to structures with
vertical irregularity due to uneven distribution of mass along height in different combinations. The shift in
period may affect the response of the structure. It is thus a need of the hour to propose an adequate
formulation for calculating period.
2. It appears that lateral period of setback structures should account for variation in height of structures in
some form.
3. Irregular structures have higher stress concentration at the points of discontinuity. Present methods of
analysis don’t take into account these effects. The methods need to be modified to account for such effects.
4. The present method of pushover analysis needs to be adequately modified to be applicable for vertically
irregular structures. The issue needs attention to arrive at dominant combination of modes under the seismic
excitation.
5. Non-linear dynamic analysis may give more accurate results for performance based design of such irregular
structures. But, due to the complexity and rigor of the problem, much study has not been done in the area.
Researches are needed to be carried out for structures with vertical irregularity encompassing their elastic
as well as inelastic behaviour. This may help to validate the results of relatively simpler methods like
Pushover Analysis.
ACKNOWLEDGEMENT:
The author acknowledges the suggestions, help and technical guidance rendered by Dr. Sekhar Chandra Dutta,
Professor of Civil Engineering, & Head of School of Infrastructure, & Dean, Alumni Affairs, International
Relations and continuing education, Indian Institute of Technology, Bhubaneswar, Odisha, India and Sanket
Nayak , Research Scholar, School of Infrastructure, Indian Institute of Technology, Bhubaneswar, Odisha,
India.
855
REFERENCES:
[1] Al-Ali, A.A.K. and Krawinkler, H. (1998). “Effects of Vertical Irregularities on Seismic Behavior of
Building Structures”, Report No. 130, The John A. Blume Earthquake Engineering Center, Department of
Civil and Environmental Engineering, Stanford University, Stanford, U.S.A.
[2] Aranda, G.R. (1984). “Ductility Demands for R/C Frames Irregular in Elevation”, Proceedings of
theEighth World Conference on Earthquake Engineering, San Francisco, U.S.A., 4, pp. 559-566.
[3] BIS (1962). “IS 1893-1962: Indian Standard Criteria for Earthquake Resistant Design of
Structures”,Bureau of Indian Standards, New Delhi.
[4] BIS (1966). “IS 1893-1966: Indian Standard Criteria for Earthquake Resistant Design of Structures (First
Revision)”, Bureau of Indian Standards, New Delhi.
[5] BIS (1970). “IS 1893-1970: Indian Standard Criteria for Earthquake Resistant Design of Structures
(Second Revision)”, Bureau of Indian Standards, New Delhi.
[6] BIS (1975). “IS 1893-1975: Indian Standard Criteria for Earthquake Resistant Design of Structures (Third
Revision)”, Bureau of Indian Standards, New Delhi.
[7] BIS (1984). “IS 1893-1984: Indian Standard Criteria for Earthquake Resistant Design of Structures
(Fourth Revision)”, Bureau of Indian Standards, New Delhi.
[8] BIS (2002). “IS 1893 (Part 1)-2002: Indian Standard Criteria for Earthquake Resistant Design of
Structures, Part 1 – General Provisions and Buildings (Fifth Revision)”, Bureau of Indian Standards, New
Delhi.
[9] BSSC (2003). “The 2003 NEHRP Recommended Provisions for New Buildings and Other Structures” Part
2:Commentary (FEMA 450)”, Building Seismic Safety Council, National Institute of Building Sciences,
Washington, DC, U.S.A.
[10] Cheung VW-T, Tso WK. 1987. “Lateral load analysis for buildings with setback”. J. Struct. Engng ASCE,
113: 209–227.
[11] Chintanapakdee, C. and Chopra, A.K. (2004). “Seismic Response of Vertically Irregular Frames: Response
History and Modal Pushover Analyses”, Journal of Structural Engineering, ASCE, 130(8), pp. 1177-1185.
[12] Chintanapakdee, C., and Chopra, A. K. (2002). ‘‘Evaluation of the modal pushover procedure using
vertically ‘regular’ and irregular generic frames.’’ Rep. No. EERC 2003-03, Earthquake Engineering
Research Center, Univ. of California at Berkeley, Berkeley, Calif., 224.
[13] Chopra, A. K., and Goel, R. K. (2002). ‘‘A modal pushover analysis procedure for estimating seismic
demand or buildings.’’ Earthquake Eng. Struct. Dyn., 31(3), pp.561–582.
[14] Duan XN, Chandler AM. 1995. “Seismic torsional response and design procedures for a class of setback
frame buildings”. Earthquake Engng. Struct. Dyn. 24: 761–777.
[15] Dutta, S.C., Saha, P., Sinha, S., Goswami, K.(2010). “Seismic Behaviour of Structures with Vertical
Irregularities”, Civil Computing, Computer Application in Civil Engineering, ACECOMS, M37-0315-
0310,March, pp. 14-19.
[16] Humar, J.L. and Wright, E.W. (1977). “Earthquake Response of Steel-Framed Multistorey Buildings with
Set-Backs”, Earthquake Engineering & Structural Dynamics, 5(1), pp. 15-39.
856
[17] Wong, C.M. and Tso, W.K. (1994). “Seismic Loading for Buildings with Setbacks”, Canadian Journal of
Civil Engineering, 21(5), pp. 863-871.
[18] Shahrooz, B.M. and Moehle, J.P. (1990). “Seismic Response and Design of Setback Buildings”,Journal of
Structural Engineering, ASCE, 116 (5), pp. 1423-1439.
[19] Soni, D.P. and Mistry B.B. (2006). “Qualitative review of seismic response of vertically Irregular building
frames”, ISET Journal of Earthquake Engineering, Technical Note, 43 (4), pp. 121-132.
[20] Valmundsson, E.V. and Nau, J.M. (1997). “Seismic Response of Building Frames with Vertical Structural
Irregularities”, Journal of Structural Engineering, ASCE, 123 (1), pp. 30-41.
[21] Tso WK, Shu Yao. 1994. “Seismic load distribution in buildings with eccentric setback”. Canadian J. Civil
Engng. 21: 50–62.
[22] Wong, C.M. and Tso, W.K. (1994). “Seismic Loading for Buildings with Setbacks”, Canadian Journal of
Civil Engineering, 21(5), pp. 863-871.
[23] Wood, S.L. (1992). “Seismic Response of R/C Frames with Irregular Profiles”, Journal of Structural
Engineering, ASCE, 118(2), pp. 545-566.
857
Paper No.: 149
SLOPE STABILITY ANALYSIS BY STATIC AND DYNAMIC METHOD
Tarapada Mandal1, Sanjay Sengupta2

1
Research Scholar, Dept. of Civil Engineering, National Institute of Technology Rourkela,
Email: 612ce3009@nitrkl.ac.in
2
Assistant Professor, Dept. of Civil Engineering, Biet suri, Birbhum, west Bengal,
Email: sanswati2004@gmail.com
ABSTRACT :
Slope stability is an important consideration in the construction of artificial slope & natural slope The
fundamental criterion for stability analysis is the factor of safety against sliding which should be always greater
than or equal to one for the slope to be safe. This is true when the slope the slope experiences elastic
deformation only. But in case of a strong earthquake motion, the deformation of slope may exceed elastic limit
and hence permanent displacements take place. In such a case factor of safety as stability criterion becomes
immaterial and magnitude of the permanent displacements becomes measure for estimating the overall stability
of the slope. The objective of the study is to find out the factor of safety of the given slope under static and
dynamic loading (earthquake) condition using Swedish Circle Method, Bishop Stability Method. This study also
involves seismic stability analysis of the slope using Pseudo-Static Method. The probability of permanent
displacement of the earthen slope in various earthquake zones and in different soil considered is compared in
this study.
KEYWORDS: Stability analysis, factor of safety, earthquake zones, soil strata.
1. INTRODUCTION
There are so many method of finding out factor of safety of slopes under static and dynamic loading (earthquake)
condition. In all the methods, stability analysis is based on limit plastic equilibrium. A failure surface is assumed
first and then factor of safety is calculated from the ratio of the shear stress along the failure surface. In methods
of limit plastic equilibrium the concepts of static is applied. Although most problems in slope stability are
statically indeterminate, some simplified assumptions are always made to make the problem the problem
statically determinate to have a unique factor of safety. Under static loading condition the methods are Swedish
Circle Method, Friction Stability Method, Bishop Stability Method (Bishop 1955) and Janbu’s Method (Janbu
1954, 1973). In case of seismic analysis of slopes the methods are Pseudo-Static Analysis and Newmark’s
Theory. Here in our study we are solving the problem by Swedish Circle Method, Bishop Stability Method,
Pseudo-Static Analysis and Newmark’s Theory. In recent years a basic point of reference has been the U.S.
Army Corps of Engineers manual for seismic design of new dams (which are generally considered to be among
the more critical civil engineering facilities). This requires use of a seismic coefficient of 0.1 in Seismic Zone 3
and 0.15 in Seismic Zone 4, in conjunction with a minimum factor of safety of 1.0. In California, many state and
local agencies also require the use of a seismic coefficient of 0.15 but impose the slightly more conservative
requirement that the minimum computed factor of safety be not less than 1.1. Clearly, however, engineering
judgment must still be applied as to the applicability of pseudo-static analyses and the acceptable factor of
safety might be varied with the uncertainties involved in a particular analysis.
2. Methodology
A case study is taken with a problem statement such as: An embankment of sandy clay having cohesion of 28
KN/m2.angle of internal friction Ø = 220 & a unit weight of 18kN/m3 .The slope height of the embankment are
1.7:1(H: V) & 10 m respectively.
2.1. Static method:
For our case study two static methods are considered, Swedish circle method and Bishop’s stability method.
For Swedish circle method the formula is:
( ɸ ∑ )
F.S. = ∑
(1)
Bishop’s stability method is a trial and error method. However, convergence of trial is very rapid and also it
gives more accurate results. The formula is:
∑ ( ɸ ) [ ]
( θ )
F.S. (Fs) = ∑
(2)
θ
2.2. Dynamic method:
Analyses of seismic slope stability problems using limit equilibrium methods in which the inertia forces due to
earthquake shaking are represented by a constant horizontal force (equal to the weight of the potential sliding
mass multiplied by a coefficient) are commonly referred to as pseudo-static analyses. In this method when we
find out factor of safety, we use the seismic coefficient. Two different approaches are considered in this study.
Design spectrum method (IS 1893-2002 code method) and Newmark’s method.
3. Data analysis
Swedish circle method (method of slices) gives the factor of safety to be 2.425 and Bishop’s stability method
gives a factor of safety 3.27 after five trial and error method.
Table 1 values obtained for the parameters in Swedish circle method

Slice Width Avg. Area Weight N T
no. (m) length A (W = γ.a) Α (Wcosα) (Wsinα)
(m) (m2)
1 2.5 1.4 3.50 W1 = 63 300 N1=54.56 T1 = 31.50
0
2 2.5 4 10.00 W2 = 180 20 N2=169.14 T2 = 61.56
3 2.5 6.2 15.50 W3 = 279 120 N3=272.90 T3 = 58
0
4 2.5 8 20.00 W4 = 360 3.5 N4=359.33 T4 = 22
5 2.5 9.5 23.75 W5=427.50 -40 N5=426.46 T5 = 29.82
0
6 2.5 10.7 26.75 W6=481.50 -12 N6=471.00 T6 = -100.10
7 2.5 11.25 28.125 W7=506.25 -200 N7=475.72 T7 = -173.15
0
8 2.5 10.60 26.50 W8 = 477 -30 N8=413.10 T8 = -238.50
9 2.5 8.85 22.125 W9=398.25 -390 N9=309.50 T9 = -250.63
0
10 2.5 6.3 15.75 W10=283.5 -50 N10=182.23 T10 = -217.17
11 2.1 2.4 5.04 W11=90.72 -650 N11=33.34 T11 = -82.22
ΣN = 3167.28 ΣT = -918.53
Factor of safety for above four earth quake zone for various time periods T1 = 0.05 sec; T2 = 0.10 sec; T3 = 0.40
sec; T4 = 0.55 sec; T5 = 0.67 sec & T6 = 4.0 sec is calculated. The design horizontal seismic coefficient Ah for a
structure shall be determined by the following expression:
859
Ah = !
(3)
And Factor of safety is given by:
" (#$% &'(Σ)) Ø

Factor of safety = # &'(Σ
(4)
The factor of safety for different time periods and different zones are calculated. Table2 shows the calculated
values for different zones and time periods for rocky or hard soil sites.
Table 2 Factor of safety for rocky or hard soil

For Rocky or Hard Soil Sites
Zone Period (Seconds) Factor of Safety

Zone-(ii) T1 = 0.05 2.176
T2 = 0.10 1.580
T3 = 0.40 1.580
T4 = 0.55 1.730
T5 = 0.67 1.820
T6 = 4.00 2.183
Zone-(iii) T1 = 0.05 1.520
T2 = 0.10 1.320
T3 = 0.40 1.320
T4 = 0.55 1.500
T5 = 0.67 1.600
T6 = 4.00 2.150
Zone-(iv) T1 = 0.05 1.290
T2 = 0.10 1.078
T3 = 0.40 1.078
T4 = 0.55 1.270
T5 = 0.67 1.390
T6 = 4.00 2.086
Zone-(v) T1 = 0.05 1.040
T2 = 0.10 0.856
T3 = 0.40 0.856
T4 = 0.55 1.020
T5 = 0.67 1.140
T6 = 4.00 1.988
860
A graphical approach for different values of factor of safety obtained for different earthquake zones and for
different soil strata is given in figure 1. We consider maximum risk,
So, taking =1
Where,
I = Importance factor = 1
R = response reduction factor = 1
2.5
1.5
Factor Of Safety →
Rocky or Hard Soil

1
Medium Soil
Soft Soil
0.5
0
0 1 2 3 4 5
Time →
Figure 1 FOS Variation for Soil Strata at Zone-(V) w.r.t. Time Period
Variations of factor of safety with respect to different zones with different time periods are obtained and plotted
in the figure 2, figure 3.
861
1.8
1.6 1.58
1.4
1.32
1.2
1.078
1
0.856 Rocky or Hard Soil
0.8
Medium Soil
0.6
Soft Soil
0.4
0.2
0
II III IV V
Zone →
Figure 2 Variation of FOS w.r.t different Zone (T3 = 0.40 Sec)
2
1.8
1.6
1.4
1.2
1 Hard or Rocky Soil
0.8 Medium Soil
0.6 Soft Soil
0.4
0.2
0
II III IV V
Zone →
Figure 3 Variation of FOS w.r.t different Zone (T5 = 0.67 Sec)
Newmark’s method for determining factor of safety in dynamic conditions is an approximate but rational
connection between the seismic coefficient that is used in a pseudo-static stability analysis and the expected
amplitudes and duration of ground motion by working backwards through the method for computing
displacements of slopes that was originally suggested by Newmark (1965), first explored by Seed (1979).
Newmark has given a simple derivation for movement of a rigid plastic mass to give a quick estimate of the
magnitude of the motions to be expected in a sliding wedge of an earth when subjected to the influence of
dynamic forces from an earthquake. The calculations are based on the assumptions that whole moving mass
moves as a single rigid body with resistance mobilized along the sliding surface.
862
According to Newmark’s method “it is only necessary to perform a pseudo-static analysis for a seismic
coefficient of 0.1 for magnitude 6.5 earthquakes or 0.15 for magnitude 8.25 earthquakes and obtain a factor of
safety of the order of 1.15 to ensure that displacements will be acceptably small. For magnitude 6.5 earthquake
seismic coefficient, k y = 0.10.
Therefore, using the formula:
CRθ + (∑ N − ky × ∑ T ) tan φ
FOS =
(∑ T + ky × ∑ N ) (5)
And the value of factor of safety obtained is 1.691 and for magnitude 8.25 earthquake seismic coefficient, k y =
0.15 and the value obtained for factor of safety is 1.486. Figure 4 shows the variations of factor of safety with
variations in the earthquake magnitude.
1.75
1.7
1.65
1.6
FOS
1.55
1.5
1.45
0 1 2 3 4 5 6 7 8 9
Magnitude of Earthquake →
Figure 4 variations in FOS with different magnitudes of earthquake
4 CONCLUSIONS
From the above study following things can be concluded:

The slope defined in the problem statement is quite stable under static condition.
Using Swedish circle method the FOS is found to be less than the Bishop’s method. Although Bishop’s method
is considered to be more accurate the FOS found using the Swedish circle method will be on the more
conservative side.
The slope is stable under static condition. However, the slope becomes gradually unstable with the more
earthquake prone zone. It is found from the analysis that the FOS becomes less than 1 in zone V for all types of
soil strata. So the permanent displacement is bound to occur in most cases of earthquake in zone V. It is found
that more the earthquake prone zone the probability of permanent displacement increases from hard soil strata to
soft soil strata.
863
Using Newmark’s method of pseudo-static method of analysis the slope is found to be stable in 8.5 magnitude
earthquake also. However, the effect of liquefaction has not been considered in this case.
REFERENCES
[1] V.N.S Murthy “Textbook of Soil Mechanics and Foundation Engineering”, Geotechnical Engineering
Series. 1st Edition, 2007. Robert Pyke “Selection of Seismic Coefficients for Use in Pseudo-Static Slope
Stability Analysis”, 2011. IS: 1893 (part 1): 2002.
864
Paper No.: 150
DAMAGE DETECTION IN SIMPLY SUPPORTED BEAM - AN ANN

APPROACH
Karthik Subhash .A1, Mishra .U.K 2 and Ganesh .R1

1
M.Tech Student, 2Assistant Professor
Dept. of civil Engineering, National Institute of Technology, Rourkela, India
Email: 212ce2042@nitrkl.ac.in,212ce1018@nitrkl.ac.in, haran5441@gmail.com
ABSTRACT:
Identification of crack location and depth in structural elements such as beams, columns, and slabs etc.,
subjected to time varying loads will help them to retrofit before failure thereby its life and structural capabilities
can be greatly improved. Since it is not always possible to detect the initial cracks by visual inspection. This
paper presents the various ANN models for detection of initial crack location and depth based on laboratory test
results of simply supported beam considering frequencies as inputs and cracked location as output for an
isotropic material. The performance of the developed models is accessed in terms of R- square value and root
mean square error. Comparison has been made between developed ANN models and regression models shows
that the ANN models can perform well than regression models. Further the simple equation has been presented
for evaluation of crack depth based on performance of the models developed here.
KEYWORDS: Crack Location, Crack Depth, Damage Detection, ANN Models
1. INTRODUCTION
Various structural health monitoring techniques have been proposed to obtain a reliable and efficient approach
in order to increase the safety of civil engineering structures. Non-destructive techniques (NDT) are the
commonly used techniques for monitoring Structural health. One of the non-destructive techniques is ultrasound
plays an important role for quality assurance of structural members and engineering materials during
manufacturing stage and operating life. However it requires rich propagation and scattering features of
ultrasonic waves for detecting and characterizing the defects in a wide range of structural materials [1]. When
the structure or structural members subjected to dynamic loads the presences of damages such as crack change
the physical characteristics of a structure which in turn alter its dynamic response characteristics. By observing
the changes in the dynamic characteristics such as natural frequencies the crack can be detected successfully [2].
Application of popular soft computing technique such as (artificial neural network) ANN have been used
successfully for detecting structural damages by various researchers like Wu et al. [3] explored the use of an
ANN to detect member damage in a 3-storey frame. Pandey and Barai [4] provided a more detailed treatment of
ANN architecture in their study identifying damage in a 21-bar bridge truss. Zhao et al. [5] applied a
counter-propagation neural network to detect damage and support movement in a continuous beam. Zapico [6]
developed a procedure for damage assessment of steel structures based on ANN. Present work involves: (1)
development of various ANN model taking popular transfer functions such as log sigmoid, tan sigmoid and also
with purlin shown in Figure 1, (2) evaluating performance of developed model in terms of R-square value and
root mean square error (RMSE), (3) comparison of ANN models with regression model.
Figure 1: Various Transfer Functions Used In Present Study
2. METHODOLOGY
2.1. Network Data Preparation
In the present analysis ANN models are developed by taking input parameters as changes in natural frequency
ratio (1st mode, 2nd mode & 3rd mode) and crack depth ratio (CDR) and crack lacation ratio (CLR) as output
parameters for a data set given in [1]. The minimum and maximum values of inputs and outputs are given in
Table 1 and the properties of the material tested are given in Table 2.
Table 1: Minimum And Maximum Values Of Data Set
INPUTS OUTPUTS
Fundamental Natural Frequency Ratio (Ωc/Ω) Crack Length Crack Depth Ratio
1st Mode (f1) 2nd Mode (f2) 3rd Mode (f3) Ratio (L1/L) (A/H)
0.7065 0.8175 0.8014 0.0625 0

1 1 1 0.875 0.7
Pre-processing of data’s helps in generalization of ANN models. Hence all data in the set has been scaled in the
range of -1 to +1 by a simple linear normalization function:,
 X − X min 
X = 2 n  − 1 (1)
 X max − X min 
Where Xn, Xmax, Xmin are the normalized, maximum and minimum values of data in a vector.
Table 2: Properties Of The Beam Tested
Descriptions Value
Material Aluminium
Width of the beam (mm) 25.4
Depth of the beam (mm) 25.4
Length of the beam (mm) 650
Elastic modulus of the beam (Gpa) 70
Poisson’s Ratio 0.35
Density (g/cc) 2.696
866
2.2. Development Of ANN Models
An Artificial Neural Network (ANN) consists of a set of interconnected elements called neurons that provide a
response or output from a series of inputs. In the present study ANN models are developed using Levenberg-
Marquardt training algorithm as this algorithm is often the fastest back propagation algorithm in MATLAB
Neural Network Toolbox and is highly recommended as a first choice supervised algorithm. First nine models
for prediction of crack depth ratio (CDR) and next nine for crack length ratio (CLR). The details network basic
structure of ANN models developed and activation function used are given in Table 3. The number of hidden
neurons depends primarily on the nature of the data. However, there are no strict rules for choosing the number
of hidden layers and the number of neurons in each hidden layer. Most back-propagation neural networks will
have one or two hidden layers, with the number of neurons in the hidden layers usually falling somewhere
between the total number of input and output neurons. In our approach, the number of hidden layers and the
corresponding neurons were decided by the corresponding learning performance of the neural network during
the training and testing of network. In this case, a network having one hidden layer with maximum of 3 neurons
is used. Therefore, the basic structure of the back propagation neural network adopted here is 3–1–1. It has been
checked to have the best performance by varying the numbers of hidden layers and neurons [2]. In the training
phase, 80% inputs from training set along with corresponding desired outputs used to train the network. The
weights in the neural network were changed iteratively according to the generalized delta rule. 15% of total set
were used for testing and 5% data are used for validation. The data’s are randomly selected for model training,
testing and validation. A major difference between training and testing is that in the test case, the weights in the
network are not updated [2]. Since the earliest work by [7], dynamics parameters such as natural frequencies,
and mode shapes have been widely applied for damage detection, as the modal parameters are functions of
structural properties. This implies that any degradation of the structural properties results in changes of the
modal parameters [3]. The general mathematical equation relating the input variables and the output can be
written as,
 h
  m
  
CDR n = f out b0 + ∑  wk × f  bhk + ∑ wik X i    (2)
 k =1   i =1   
where, CDRn is the normalized CDR value, (in the range -1 to 1 in this case); b0 = bias at the out put layer; wk
= connection weight between kth neuron of hidden layer and the single output neuron; bhk = bias at the kth neuron
of hidden layer; h = number of neurons in the hidden layer; wik = connection weight between ith input variable
and kth neuron of hidden layer; Xi = normalized input variable i in the range [-1, 1] and f , fout= transfer function
in hidden and output layers.
2.3 Development Of Regression Models
The multiple linear regression models has been developed which is extension of a simple linear regression
model to incorporate three explanatory variable such as Fundamental natural Frequency ratio (1st mode, 2nd
mode, and 3rd mode). Multiple regression modelling is now a mainstay of statistical analysis in most fields
because of it's power and flexibility. The predicted and actual values for both CDR and CLR are shown in
Figure 1. The general form of multiple linear regressions is shown below,
Y = b0 + b1 × x1 + b2 × x 2 + .......... + bn × x n
(3)
Where Y is output, b0,b1,..,bn are constants and x1,x2,….,xn are the explanatory variables.
3. RESULTS AND DISCUISSIONS
Table 3 shows the performance of various model created for prediction of both CDR and CLR. It is found that
the network structure of 3-1-1 with log sigmoid activation function can better predict the CDR value with a
testing R-value of 0.967. comparison of RMSE of various developed ANN models for CDR prediction is given
in Figure 2 shows that the ANN model with tansig activation function gives lesser RMSE values than others.
867
However considering overall R-values of all model the one with tansig gives higher value as shown in table 3
with increase of hidden neuran. For the shake of simplicity a simple equation (4) is presented for prediction of
CDR.
Table 3: Performance Of Ann Model Developed For Cdr And Clr Prediction
R- Values for CDR predictive models

Network Tansig Logsig Purelin
structure Training Testing Validation Training Testing Validation Training Testing Validation
3-1-1 0.965 0.968 0.95 0.963 0.967 0.961 0.865 0.929 0.83
3-2-1 0.962 0.966 0.979 0.959 0.989 0.874 0.856 0.948 0.965
3-3-1 0.989 0.980 0.963 0.975 0.966 0.891 0.847 0.961 0.985
R- values for clr predictive models
3-1-1 0.369 0.561 0.729 0.352 0.596 0.32 0.307 0.699 0.223
3-2-1 0.28 0.56 0.542 0.29 0.586 0.22 0.129 0.58 0.377
3-3-1 0.463 0.674 0.2316 0.256 0.846 0.6 0.183 0.528 0.044
Rmse of cdr predictive models Regression
Tansig Logsig Purelin
Network
structure Training Testing Training Testing Training Testing
3-1-1 0.170 0.203 0.195 0.173 0.339 0.381 0.335
3-2-1 0.175 0.200 0.190 0.178 0.335 0.372
3-3-1 0.094 0.142 0.148 0.196 0.355 0.347
RMSE Of CLR predictive models
3-1-1 0.68 0.59 0.68 0.56 0.63 0.65
3-2-1 0.59 0.96 0.64 0.62 0.67 0.44 0.7674
3-3-1 0.62 0.80 0.67 0.51 0.67 0.63
0.4
0.355 0.347
0.35 0.335 0.335 0.335
0.3
0.25
RMSE
0.193 Regression
0.2
0.142 0.152
0.15 ANN (Training)
0.094 ANN (Testing)
0.1
0.05
0
Tansig Logsig Purelin
Activation Function In The Hidden Neuron
Figure 2: comparison of RMSE of various developed ANN models for CDR prediction
The ANN model for prediction of CDR as follows,
868
− 868.231
CDR = + 0.6837 (4)
1 + e −n
where n = −0.6104 × f1 + 18.179 × f 2 + 24.2973 × f 3 − 49.0953

the above equation gives R-value of 0.967.
1.2
ANN Non Linear Regression
Multiple Linear Regression Line of Equality
1
Predicted Value of CDR
0.8
1.3x Line of Equality
0.6
0.4
0.2 0.65x Line of Equality
0
0 0.2 0.4 0.6 0.8 1 1.2
Actual Value of CDR
Figure 3: Actual And Predicted Values Of CDR through all model
Figure 4: Actual And Predicted Values Of CLR And CDR (regression)
The non linear regression equation has been presented for the prediction of CDR having R-value of 0.933 as
869
follows,
ln(0.044 × f1 − 21.344 × f 2 − 21.119 × f 3 + 43.641)

CDR = (5)
2.022
Where f1, f2 and f3 are the frequency ratios (1st mode, 2nd mode, and 3rd mode).
Multiple linear regression for prediction of both CDR and CLR has been caried out and the actual and predicted
response is shown in figure 3. It is noted that the predicted response of CDR to actual CDR showing high
correlation coefficient of 0.731 when fitted with linear model. Using a simple non linear function, the linear
model has been transformed, model giving high correlation coefficient of 0.871 when fitted with exponential
function. The non linear regression equation is presented in equation 5. The predicted values of CDR through all
model with actual value is presented in figure 4 which shows that the ANN model with network structure of
3-1-1 can predict the CDR value in more realistic than other models followed by non linear regression model.
CONCLUSION:
An attempt has been made to develop various models (ANN and regression) considering frequencies as inputs
for prediction of damages i.e crack depth and location for a simply supported beam. The performance of
developed model is accesed by R-value and RMSE. Comparison between ANN and regression model has been
made. This developed ANN models perform better than other developed models here for crack depth prediction.
A simple equations (equation 4 and 5) has been presented for prediction of CDR with relatively high R-value.
This developed model can also be used for different varied applications of engineering, given the predefined
values as input to the system.The networks here produced reasonable performances for the identification of
cracks.
REFERENCES
[1] S.W. Liua, Jin H. Huanga,, J.C. Sungb, and C.C. Lee. (2002), “Detection of cracks using neural networks
and computational mechanics”. Comput. Methods Appl. Mech. Engrg. 191 (2002) 2831–2845.
[2] Lokesh Kondru and M. R Narasinga Rao, (2013). “Damage Detection in Cantilever Beams Using Artificial
Neural Networks”. International Journal of Mechanical and Production Engineering Research and
Development (IJMPERD). ISSN 2249-6890. Vol. 3, Issue 1,259-268.
[3] Wu X, Ghaboussi J, Garrett JH. (1992).”Use of neural networks in detection ofstructural damage”. Journal
of Computers & Structures; 42(4):649–59.
[4] Pandey PC, Barai SV. (1995).” Multilayer perceptron in damage detection of bridge structures”. Journal of
Computers & Structures; 54(4):597–608.
[5] Zhao J, Ivan JN, DeWolf JT. (1998).” Structural damage detection using artificial neural network”. Journal
of Infrastructure Systems; 4(3):93–101.
[6] Zapico JL, Worden K, Molina FJ.(2001).” Vibration-based damage assessment in steel frames
using neural networks”. Journal of Smart Materials and Structures; 10:553–9.
[7] Owolabi, G.M., Swamidas, A.S.J. and Seshadri, R. (2003). “Crack detection in beams using changes in
frequencies amplitudes of frequency response functions”. Journal of Sound and Vibration, Vol.265,
pp.1-22.
[8] Cawley P, Adams RD.( 1979), “ The location of defects in structures from measurements of natural
frequencies”. Journal of Strain Analysis; 14(2):49–57.
870
Paper No.: 151
BEHAVIOUR OF RECYCLED AGGREGATE CONCRETE

Kirtikanta Sahoo1, Pradip Sarkar2, Robin Davis P.2
1
Research Scholar, Dept. of Civil Engineering, N.I.T Rourkela, India
2
Associate Professor, Dept. of Civil Engineering, N.I.T Rourkela, India
Email: sahoo.kirti@gmail.com
ABSTRACT:
Recycled coarse aggregates (RCA) are different in properties and composition than natural coarse
aggregate. So it is very difficult to expect the same behaviour of RCA during mix design of concrete.
This paper attempts to reveal the probable relationship between RCA properties and compressive
strength of RCA concrete through regression analysis based on 30 numbers of available data collected
from published literature. The parameters considered are water, cement, natural coarse aggregate,
sand, RCA content in the design mix, saturated surface dried density, and maximum size of RCA. The
regression analysis is run through SPSS software.It is observed that proposed relation predicts the
compressive strength of RCA concrete that closely matches the experimental results.
KEYWORDS: Recycled coarse aggregate, regression analysis, concrete, compressive strength
1 INTRODUCTION
Due to rapid urban development along with trade and industry activities, construction debris creates a
problem for environment now a day. In construction waste concrete is the major source which makes
disposal problem with more transportation cost. Such situation occurs in many countries like in India.
As per Central Pollution Control Board it has been assessed that the solid waste generation is about 48
million tonnes per annum of which 25% are from the construction industry. Simultaneously due to
rapid construction early exhaustion of natural aggregates (fine and coarse) occurs and put the impact
on the environment.
Since the last two decades, many researchers have prepared use of various methods to predict the
properties of concrete prepared by different mix proportions. Among these methods, methods based
on regression analysis are quite popular. The main advantage of this technique, it is totally a
mathematical regression process. The intend of this paper is to account the results of a study to
evaluate whether or not the compressive strength of concrete made with different types and sources of
recycled aggregates can be predicted by regression analysis. For this data sets are collected and
compared with each other by linear regression analysis. So in the present work for regression analysis
(Linear Regression System) IBMSPSS software is used. In this analysis the variables are divided into
two categories such as dependent and independent variables. By considering these variables a
mathematical formulation is formulated. The constructed model had 9 input parameters and one
output parameter. The predicted results were compared with the experimentally determined results.
2. RECYCLED AGGREGATE
The common difference between RCA and natural aggregate (NA) is that RCA is adhered with mortar
from the parent concrete which is about 20-30% of the total volume. This is normally depends on the
properties of the original concrete from which the RCA is obtained and the making process of the
RCA. RCA has properties like more flaky in shape, lower bulk and SSD specific densities, higher
water absorption, inferior strength, presence of contaminants (e.g. ceramic) and lower resistance to
mechanical and chemical actions when compared with NA and known as the primary features.
Parameters like specific density, water absorption, porosity, percentage of elongated particles,
fineness modulus, strength indices (10% fines value, crushing value, impact value, and Los Angeles
abrasion) and impurity content are often adopted to evaluate the properties of the RCA.
It is observed that the concrete made from RCA often substandard to those concrete prepared with
only NA due to the high water absorption and low density of RCA. Normally RCA are used as
recycled aggregate because recycled fine aggregate has more water absorption quality. So this paper
discussed the use of recycled coarse aggregate in new concrete. Also it is examined that the strength
of RCA concrete depends on the replacement ratio of RCA. It is reported that with the increment in
the ratio of replacement RAC the strength simultaneously decreases.
From the previous literatures, the mechanical properties of RAC concrete are in general low-grade to
those of NAC concrete; mainly because of the presence of old cement paste. To sustain the same
workability without the use of chemical admixtures, RAC requires more water to maintain
workability than conventional concrete, which affects the quality and strength of the concrete,
resulting in lower concrete strength. On the basis of literature survey it is found that the attached
mortar content was inversely proportional to the size of the RCA, and the higher amount attached
mortar would direct to lower density, and the trends were similar for values of water absorption, Los
Angeles abrasion and sulphate content of the RCA.
3. REGRESSION ANALYSIS
As previously describe that this paper aims to produce a mathematical equation, by using this we
easily predict the relation between the compressive strength RCA concrete and RCA properties. In
this present work, an efficient technique is used for forming a mathematical equation. So mathematics
like Statistics Data Editor System (SPSS) is proposed for prediction of the relation.
The term “regression” originates from the 14th century, where it had a biological mean in gas “the act
of going back”. It was first adapted to a more general statistical context by the well-known
statisticians Udny Yule and Karl Pearson. Regression is a generic term for all methods attempting to
fit a model to observed data in order to quantify the relationship between two groups of variables. The
fitted model may then be used either to merely describe the relationship between the two groups of
variables, or to predict new values. The first widely studied form of regression analysis has been
linear regression, due to the simplicity of the model and the statistical properties of the estimators.
Linear regression is usually used for the purpose of hypothesis testing or for the purpose of prediction
and forecasting. Many statistical methods and techniques have emerged from its study and one of
them is the simultaneous confidence band. This paper provides a general review of linear regression
and presents some preliminary results necessary for the construction and comparison of simultaneous
confidence bands.
3.1 Linear Regression Analysis
A common problem in experimental science is to observe how some sets of variables affect others.
Some relations are deterministic and easy to interpret, others are too complicated to understanding or
describe in simple terms, that possibly having a random component. In the present study, by using
relatively simple empirical methods we estimated these actual relationships by simple functions or
random processes. Among all the methods used for calculating such complex relationships, linear
regression possibly is the most useful. In this methodology, it is necessary to assume a functional,
parametric relationship between the variables in question, and also unknown parameters which are to
be estimated from the available data. Two sets of variables can be well-known at this stage such as
Predictor variables and response variables. Predictors variables are those that can either be set to
872
controlled or else take values that can be observed without any error. Our objective is to discover that
how changes occurs in the predictor variables that affect the values of the response variables.
Linear regression analysis means it is a statistical technique used to model data comprising of a
dependent random variable and one or more independent variables, so as to evaluate the relationship
between the dependent variable and the independent variables. In statistics, simple linear regression is
the least squares estimator of a linear regression model with a single helpful variable. In other words,
simple linear regression fits a straight line through the set of n points in such a way that makes the
sum of squared residuals of the model (that is, vertical distances between the points of the data set and
the fitted line) as small as possible.
Table 1 Inputs and outputs for the analysis
Parameters Minimum Maximum

3
W (Kg/m ) 132 241
3
C (Kg/m ) 275 501
S( Kg/m3) 467 868
3
NA Kg/m ) 226 1301
RA Kg/m3) 422.2 1196
DCA (mm) 5 31.5
W/C 0.43 0.5
Wm (%) 0.4 9.25
Fcu(Mpa) 27 54.8
Table 2 Ingredients of mix collected from available literature
Cement Water Sand Natural Recycled

Size of
Literature Content Content W/C Content Aggregate Aggregate
CA (mm)
(kg/m3) (kg/m3) (kg/m3) (kg/m3) (kg/m3)
Ref. (1) 379 190 - 623 1237 1171 10,20
Ref. (1) 410 225 - 642 1048 1017 10,20
Ref. (1) 392 132 - 823 960 888 25
Ref. (1) 275 179 - 868 226 830 -
Ref. (1) 401 - 0.43 574 1261 1128 -
Ref. (1) 390,501 185 0.47 558,467 1301 1109 5,31.5
Ref. (1) 380 190 0.5 714.56 882.2 874.04 -
Ref. (1) 353 214,241 - 667,625 1080 990 10,20
Ref. (1) 325 - 0.5 710.5,710.5 346.5 422.2 -
Ref. (1) 395 198 - 563 1196 1196 -
Ref. (1) 380 190 0.5 687 1145 955 10,20
873
Table 3 Properties of the ingredients of mix collected from available literature
Water absorption (%) Char. Oven dried density (kg/m3)

Literature Strength Natural
Natural Recycled Natural Natural fine
(MPa) recycled
Aggregate Aggregate aggregate aggregate aggregate
Ref. (1) 1.25 8.82 41.5,32.6 2.62 2.41 2.39

Ref. (1) 1.12 4.26 48.6,36.8 2.62 2.49 -
Ref. (1) 0.69 2.99 27 - - -
Ref. (1) 1.8 5.3 51.3,42.7 2.51 - 2.32
Ref. (1) 1.13 3.1 49.45,40.8 1.58 1.62 1.34
Ref. (1) 0.4 9.25 35.6,28.7 1.45 - 1.29

Ref. (1) 0.98 5.19 54.8,45.25 2.65 - 2.34
Ref. (1) 1.25 7.56 48.3,46.8 2.62 - 2.33

Ref. (1) - - 29,28 - -
Ref. (1) - - - 1.26 - 1.41
Ref. (1) 0.9 7.6 54.3,42.65 2.61 2.6 2.34
5.1 Input Parameters
For regression analysis total 9 parameters are taken into consideration. These are listed above table
no.2 respectively.
5.2 Concrete parameters
The concrete constituents in a concrete mix such as amount of water, W/C ratio, cement, sand, natural
coarse aggregate, recycled coarse aggregate has a effects on concrete for its strength. Simultaneously
the quality of coarse aggregate such as maximum particle size, water absorption, density also affects
the strength of the concrete.
5.3 Output parameters
In this paper 28 days compressive strength is taken only output value. Because all the parameters are
related to compressive strength of the concrete.
5.4 Regression analysis
For regression analysis 9 number parameters are taken into consideration from 20 literatures.
874
6. RESULTS
From regression analysis the obtained standard error between actual and compressive strength of the
concrete provided in the table no. 4 & 5 respectively.
Table 4 Natural Aggregate concrete Table 5 Recycled coarse aggregate
Fcu Std Error Fcu Std Error

Predicted Residuals Predicted Residuals
actual Prediction actual Prediction
42 43.59 -1.59 4.19 32.6 31.53 1.07 1.53
49 49.75 -0.75 4.34 36.8 37.72 -0.92 1.57
27 27.53 -0.53 4.43 36.5 36.83 -0.33 1.48
51 48.86 2.14 4.13 42.6 43.08 -0.48 1.67
49 49.21 -0.21 4.44 40.8 41.51 -0.71 1.51
36 36.12 -0.12 4.44 28.7 28.60 0.10 1.67
55 53.10 1.90 3.36 45.25 43.88 1.37 1.39
48 49.83 -1.83 4.02 46.8 46.50 0.30 1.55
29 31.70 -2.70 4.01 38 38.20 -0.20 1.69
49 48.88 0.12 4.44 48.51 47.79 0.72 1.58
36 31.88 4.12 3.33 46 46.24 -0.24 1.66
54 54.54 -0.54 3.29 42.65 43.31 -0.66 1.40
60
55
R² = 0.9607
50
Predicted Values
45
40
predicted
35
Linear (predicted)
30
25
25 30 35 40 45 50 55 60
Actual Values
Fig. 2: Predicted vs Observed Value for Natural aggregate Concrete
(FCU ) NA = −3.4C + 1.33W − 1.4W / C − 2.5NA − 2.3S − 3.56DCA + 0.25δ a + 0.36δ s − 6.5Wm
875
50
R² = 0.9855
45
Predicted Value
40
Series1
35 Linear (Series1)
30
25
25 30 35 40 45 50
Actual value
Fig. 3: Predicted vs Observed Value for Recycled aggregate Concrete
(FCU )RCA = −1.24C + 0.22W − 1.1W / C + 0.9RA − 0.28S − 0.52DCA − 1.05δ b + 0.32δ s − 0.26Wm
C: ordinary Portland cement; W: water; W/C: water–cement ratio; RA: recycled aggregate; S: sand;
DCA: maximum size of coarse aggregate; δa: oven dried density of NA; NA: natural aggregate; δb:
oven dried density of RA; δs: oven dried density of Sand; Wm: water absorption of coarse aggregate;
fcu: 28-day compressive strength
6 CONCLUSIONS
The salient conclusions from the present study are as follows:

• The above two equations formulated for the prediction of compressive strength of the
concrete has a high capacity for prediction.
• Oven dry density, water absorption and size mainly govern the analysis.
• The regression analysis still requires more parameters for improving the advance exploration.
REFERENCES
[1] Won-Chang Choi, Hyun-Do Yun. (2013). Long-term deflection and flexural behaviour of
reinforced concrete beams with recycled aggregate. Materials and Design, 51, pp.742–750.
[2] Vivian W.Y. Tam, et ,al. (2005). Microstructural analysis of recycled aggregate concrete produced
from two-stage mixing approach. Cement and Concrete Research, 35, pp.1195–1203.
876
[3] Salomon M. Levy, Paulo Helene. (2004). Durability of recycled aggregates concrete: a safe way to
sustainable development. Cement and Concrete Research, 34, pp.1975–1980.
[4] F.T. Olorunsogo, N. Padayachee. (2002). Performance of recycled aggregate concrete monitored
by durability indexes. Cement and Concrete Research, 32, pp.179–185.
[5] Benito Mas, Antoni Cladera,et.al. (2012). Influence of the amount of mixed recycled aggregates
on the properties of concrete for non-structural use. Construction and Building Materials, 27,
pp.612–622.
[6] Zoran Jure Grdic, et.al. (2010). Properties of self-compacting concrete prepared with coarse
recycled concrete aggregate. Construction and Building Materials, 24, pp.1129–1133.
[7] S.C. Kou, C.S. Poon. (2009). Properties of self-compacting concrete prepared with coarse and fine
recycled concrete aggregates. Cement & Concrete Composites, 31, pp.622–627.
[8] Shi-Cong Kou, et.al. (2011). Influence of recycled aggregates on long term mechanical properties
and pore size distribution of concrete. Cement & Concrete Composites, 33, pp.286–291.
[9] Benito Mas, et.al. (2012). Concrete with mixed recycled aggregates: Influence of the type of
ement. Construction and Building Materials, 34, pp.430–441.
[10] C.S. Poon, et.al. (2004). Influence of moisture states of natural and recycled aggregates on the
slump and compressive strength of concrete. Cement and Concrete Research, 34, pp.31–36.
[11] M. Etxeberria, et.al. (2007). Influence of amount of recycled coarse aggregates and production
process on properties of recycled aggregate concrete. Cement and Concrete research, 37,
pp.735–742.
[12] Patrícia S. Lovato, et.al. (2012). Modeling of mechanical properties and durability of recycled
aggregate concretes. construction and Building Materials, 26, pp. 437–447.
[13] Yeong-Nain Sheen, et.al. (2013). Assessment on the engineering properties of ready-mixed
concrete using recycled aggregate. Construction and Building Materials, 45, pp.298–305.
[14] Won-Chang Choi, Hyun-Do Yun. (2012). Compressive behaviour of reinforced concrete
columns with recycled aggregate under uniaxial loading. Engineering structures, 41 pp.285–293.
[15] Z.H. Duan, S.C. Kou, C.S. Poon. (2013). Prediction of compressive strength of recycled
aggregate concrete using artificial neural networks. Construction and Building Materials, 40,
pp.1200-1206.
[16] M. Chakradhara Rao, et.al. (2011). Behaviour of recycled aggregate concrete under drop weight
impact load. Construction and Building Materials, 25, pp.69–80.
[17] S.Manzi, et.al. (2013). Short and long-term behavior of structural concrete with recycled
concrete aggregate. Cement & Concrete composites, 37 pp.312–318.
[18] C. Thomas, et.al. (2013). Durability of recycled aggregate concrete, Construction and Building
materials, 40, pp.1054–1065.
[19] A.K. Padmini, , et.al. (2009). Influence of parent concrete on the properties of recycled aggregate
concrete. Construction and Building Materials, 23, pp.829–836.
[20] Akash Rao, et.al. (2007). Use of aggregates from recycled construction and demolition waste in
concrete resources, Conservation and Recycling, 50, pp. 71–81.
877
[21] Dash Saine S. (2013). Stage-Discharge Modeling for Meandering Channel. Thesis submitted to
the National Institute of Technology, Rourkela.
878

E-Proceeding ICSEM 2013

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

E-Proceeding ICSEM 2013

Caricato da

Copyright:

Formati disponibili

e-proceeding

INTERNATIONAL CONFERENCE ON STRUCTURAL

December 20-22, 2013

Co-Editors: Pradip Sarkar, Robin Davis P. and Srinivas Sriramula

DEPARTMENT OF CIVIL ENGINEERING

Paper Authors Title Page

R. P. Khandelwal; Calculation of inter-laminar shear stresses in laminated

004 S. N. Patel Nonlinear response of laminated composite stiffened plates 19

Revived technique for compressive strength of CFRP

M.N.A. Gulshan Taj and

V. Chitra and Effect of lay-up sequences and imperfections on the

Evaluation of shear strength of RC members using different

Chhabirani Tudu and Study of torsional behaviour of rectangular RC beam

Benazir F. Ahmed and Influence of location of staircase on seismic behaviour of

Arijit Acharjya and Influence of staircase and elevator core on twisting

P. Bhattacharya and Random vibration analysis of RCC structure using pseudo-

N Murli Krishna and Evaluation of performance point using energy based

Arghya Sengupta and

Pawan Thakur and Seismic behaviour of plan-asymmetric structures under

Syed H. Basha and Influence of masonry properties on lateral load response of

Komathi M. and Strengthening of columns for shear in reinforced concrete

Shakti P. Jena and

Saswati De and Behaviour of two-span continuous steel box RC deck

Prediction of hysteretic energy demand in steel moment

Performance evaluation of concrete against rebar corrosion

Potentiodynamic polarization study on bare steel in

Snehal Kaushik and Seismic behaviour of slab-structural wall junction in rc

Sanjay Goswami; Partha

Tushar Kanti Dey ,

C. Mohanlal and Optimal design of a retractable roof structure using real

The change in stress distribution pattern within the earthen

Bhaskar Ghosh and

G Muthukumar and Influence of opening on the dynamic structural response of

P.V. Katariya and Modal analysis of laminated composite spherical shell

Spectral strain energy based approach for system

K. V. S. N. Murthy and A study on analysis of turbo generator foundation resting on

Rajiv Verma and

Revathi P.; Ramesh, R. A. Effect of treatment methods on the strength characteristics

Irshad A Khan and Diagnosis of multiple crack of cantilever composite beam

Effect of stiffened cut-out on vibration and parametric

H. S. Panda; S. K. Sahu Modal analysis of delaminated woven fibre composite

G Srikar; B Gopi and Effect of temperature on compressive behaviour of concrete

K.V.V. Sumanth and Effect of tension stiffening on behaviour of concrete

Optimal earthquake resistant design of a fixed beam using

P. Dinakar and Design and development of high strength self compacting

Pardeep Kumar and

Jitendra Kumar Meher and Free vibration of multiple-stepped Bernoulli-Euler beam by

N. Trivedi and R. K. Experimental and analytical estimation of concrete creep

Kotla Shiva Kumar and Passive control of structures subjected to earthquake

Jagdish Malav; Swapnil

D. V. Prasada Rao and Effect of silica fume on strength of

Seismic response of vertically irregular structures: issues to

Tarapada Mandal and

Karthik Subhash A.; U. K. Damage detection in simply supported beam - an ANN

Kirtikanta Sahoo; Pradip

CALCULATION OF INTER-LAMINAR SHEAR STRESSES IN LAMINATED

3. NUMERICAL RESULTS AND DISCUSSIONS

3.1. Convergence study

Normalized Thickness (z/h)

0.3 HSDT_LSE 0.3 h/a = 0.1

Normalized Thickness (z/h)

0.3 R/a =5 0.3 R/a =10

HSDT_LSE (R/a = 20)