Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
by
Sydney, Australia
February 2001
CERTIFICATE OF- ORIGINALITY
I hereby declare that this submission is my own work and to the best of my knowledge
it contains no materials previously published or written by another person, nor material
which to a substantial extent has been accepted for the award of any other degree or
diploma at UNSW or any other educational institution, except where due
acknowledgement is made in the thesis. Any contribution made to the research by
others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged
in the thesis.
I also declare that the intellectual content of this thesis is the product of my own work,
except to the extent that assistance from others in the project's design and conception
or in style, presentation and linguistic expression is acknowledged.
/~/'/
f 'hw-~
:t-1/•1/fl"'
Yadav Raj Khwaounjoo
:JJ rulivairvd ~ rrvlf ~~
Q/YL~
.}( 53 aluuLU/f'-- ~ y uba/l'VrVa :£0/X/rrUi
ACKNOWLEDGMENTS
The research presented in this thesis was undertaken in the School of Civil and
Environmental Engineering at the University of New South Wales, Australia.
The author is highly indebted to Dr. Stephen J. Foster for continuous guidance and
support he has given throughout this research. Dr. Foster showed special interest in
this research project and provided information, valuable suggestions, encouragement
during all phases of the research program. The author is also grateful to
Prof. R. Ian Gilbert for his suggestions, comments and supports throughout the period
of author's PhD study.
The computation work of this research was carried out in the computer laboratory of
the School of Civil and Environmental Engineering. The assistance of the computer
support staff is appreciated. The experimental programme described in this thesis was
undertaken at the Randwick Heavy Structures Laboratory. The successful completion
of the test programme was made possible by dedicative efforts of Messrs Chris
Gianopoulos, Tony Macken, Frank Scharfe, Ron Moncay, Paul Gwynne, and William
Terry.
Finally I wish to thank my wife Lila, daughter Prashannata and son Prashanna without
whose love, patience, encouragement and understanding this dissertation would never
have been completed.
ABSTRACT
The finite element program DIANA was used to analyse punching type problems
using 20-node brick elements to model the concrete and 2 or 3-node embedded bar
elements to model the reinforcing and prestressing steel. The model was calibrated
using a wide range of experimental tests and with the results of the pullout tests
undertaken in this study. It was concluded that the finite element models used in
DIANA give reasonable results provided that the studies are accompanied with
parameter sensitivity analyses. The softening slope of the concrete tension stress-strain
curve was shown to affect the flexural response of the slabs while the shear retention
factor was shown to be highly influential on the failure load for slabs failing in
punching shear.
In the main experimental investigation three one-third scale edge column connections
and one comer column connection were tested with three of the four specimens being
partially prestressed. Punching shear failure was observed in all of the specimens
tested with the strength and behaviour of the specimens influenced by the degree of
the applied prestress. The results of the experimental testing and subsequent numerical
modelling showed that the capacities of flat plate slab-column connections increase
approximately linearly with increasing slab prestress.
Shear force and moment data obtained from a wide range of experimental tests were
compared with predictions from five code models. There was a significant degree of
scatter between the test data and the predicted strengths with only the Australian code
giving a lower than five percent probability that a single specimen will fail below the
capacity given by the code model. The ACI code, Eurocode 2 and CEB-FIP Model
code fall slightly below the 95 percent characteristic level. The British code model
gave an unacceptably high probability of failure for the design of slabs failing in a
punching mode.
CONTENTS
Acknowledgments iv
Abstract v
Contents vi
Nomenclature xiii
3.4.2 Sensitivity to the Concrete Tensile Strain at Zero Stress, Eu. 3-41
REFERENCES R-1
c cohesion
a parameter
equivalent cohesion
elasticity matrix
d effective depth
f potential function
Gt fracture energy
overall depth
K spring stiffness
direction cosines for the strain normal to the crack plane (n-axis)
relative to the global co-ordinate system
XV
N transformation matrix
shear force
q projection vector
UJ perimeter
control perimeter
shear force transferred to the column centre at the collapse limit state
xvi
V, Vc shear stresses
w intensity of load
a threshold angle
ratio of long side dimension to the short side dimension of the column
cracking strain of concrete
q, angle of friction
v effectiveness factor
dilatancy angle
CHAPTER 1 - INTRODUCTION
1.1 General
A flat plate floor system involves the direct transmission of forces from the slab to the
columns without the use of beams as shown in Figure 1.1. Reinforced and prestressed
concrete structures with flat plate construction are popular structural systems. This is
because of their architectural versatility, easier formwork construction and easier steel
building heights and the ease of placement of services which ultimately leads to lower
overall construction costs. These structural systems are widely used in offices and car
park construction. Since the 1960's, increasing demand for longer spans has lead to
stiffness. The use of prestressing can effectively increase the stiffness of the member,
control cracking and deflections and may improve the punching resistance of the
Although considerable attention has been given to studying flat plate and flat slab
developed in flat plates gives rise to complex load distributions in the vicinity of the
connection. High bending moments together with high shear forces may be developed
criteria. Figure 1.2 shows different types of shear problems in flat plate structures. The
1-2
Edge Corner
column column
..,X
Shear
D
ol
I
x~~D X
1Shear and moment
Or only shear transfer
D D D
D D D
extent of the influence of the prestress in preventing punching failure remains unclear
especially for the connections with edge and comer columns. Limited experimental
and theoretical studies have been undertaken on edge column connections of post-
Different structural design codes use different models to incorporate the effect of
prestress in the design of slab-column connections. For example the American code
ACI 318 (1999) neglects the effect of prestress if the slab projection beyond the edge
column does not exceed four times the slab depth while CEB-FIP Model code (1993)
suggests a reference to specialist literature. The diversities in the design models used
by various codes highlight the need for further research in the field.
The research described in this thesis comprises of three main parts. Part one
for the analyses of the punching type problems. In part two an experimental
columns. Finally in part three the finite element model is used to extend the findings of
tested, three edge column connections and one comer column connection. The
specimens are sized to be representative of a full scale structure and the slabs are
uniformly loaded to failure. Two of the edge column specimens and the comer column
specimen are prestressed with unbonded tendons, the aim being to study the influence
1-4
pullout tests are also undertaken to provided data for the calibration of the finite
element model.
In part three ofthis thesis the fmite element program DIANA (1997) is used to extend
the range of the research with modelling of various levels of prestress on comer and
edge connections. Lastly the numerical model is used to determine the likely influence
1.2 Objectives
edge and comer column-slab connections that are punching shear critical.
1.3 Scope
slab-column connections. Experimental and numerical models are used for the
investigation of edge and comer column connections under statically applied uniform
vertical loading. The outcomes of this research are applicable to the ultimate strength
1-5
design of flat plate slab-column connections failing in punching shear under gravity
load.
The effects of overhangs, holes and shear reinforcement are not dealt with in this
research. Also time dependent effects like creep and shrinkage, thermal effects,
dynamic effects, rate of application of load and size effects are beyond the scope of
this research.
effect of mesh grading and the sensitivity of different modelling parameters needed for
number of results from experimental studies. A range of studies was selected for a
wide variety of flexural, torsional and shear failures. The effects of boundary
In Chapter 5 results of plain concrete pullout specimens are given. The tests give
Chapters 6 gives details of the main experimental programme and the results are given
three of the specimens are edge column connections and one specimen is a comer
column connection. The results of the tests presented are loads, deflections, reactions
and the strains measured at key locations of a number of steel reinforcing bars. The
crack patterns and the failure mechanisms are discussed and the results are compared
In Chapter 8, the fmite element models developed in Chapters 3 and 4 are used to
fmite element modelling is used to gain further insight into the behaviour of the tested
The thesis has four appendices. In Appendix A the design details for prototype slab
used to size the model specimen are given. In Appendix B the prestressing losses in
the test slabs are calculated and the results of relaxation tests on the strand are given.
Appendix C provides the basis for the correction to the horizontal reaction measured
2.1 Introduction
The earliest flat slabs are reported to have been built by Turner in the United States of
America in 1906. In Europe, Robert Ma.illart pioneered the construction of flat slabs
(Regan, 1981, Furst and Marti, 1997). Since this time numerous flat slabs have been
In flat slab concrete structures the shear force and a part of the unbalanced moment are
transferred from the slab to the columns at the slab-column interface. Transfer of shear
and moment for a typical internal slab-column connection is shown in Figure 2.1. Shear
stresses are developed in the critical section because of the shear force and the
For the case of a slab in one-way bending, the slab behaves as a beam. After the
formation of diagonal tension cracks in the vicinity of the critical section of the slab
(around the perimeter of the loaded area) the shear force is carried by a combination of
shear across the compression zone, aggregate interlock and dowel action. However, for
a slab in two-way bending, the nominal ultimate shear stress that can be developed at
the assumed critical section is much higher than in a beam or slab in one-way bending.
/
Vu~~
/ Mu1
Forces to be transmitted
by slab-column connection
Shear stresses on
~~!
~ andMux1
critical section due
Note:
Mu1 =Mux1 +Mux2
Mux1 and Mux2 are parts of
total unbalanced moment M
transmitted by shear and
flexure, respectively
Under the action of load on a slab in two-way bending, the first crack is usually formed
by negative bending in the radial direction (van den Beukel, 1976, Park and Gamble,
2000). This crack is approximately circular and forms tangential around the perimeter
of the loaded area (Figure 2.2). Cracks radiating from the column are formed due to
negative bending moments in the tangential direction. The radial moment, which is the
moment causing the circumferential cracks, decreases rapidly away from the loaded
portion and, thus, to propagate the tangential cracks around the loaded area some
distance further from the first cracks, the load must be increased significantly. The
diagonal tension cracks in the slab normally originate near the mid-depth and are
~-·l:--d/21
o~J
d/2-l t~l
= d slab effective depth
Vu
Figure 2.2 - Assumed critical section and failure mode for punching shear
failure of reinforced concrete slab-columns with concentrically
loaded columns (Park and Gamble, 2000).
cracks are dependent on the stiffness of the slab surroundlng the cracked region and on
the slabs ability to maintain the shear transfer (at high loads) by aggregate interlock and
yielding of the negative moment slab reinforcement in the vicinity of the punch. The
diagonal tension (web shear cracks) are usually formed at about one-half of the load
required for punching shear failure. Equilibrium of the three-dimensional system can be
written in the generalised form for the horizontal forces acting on the section near the
critical diagonal tension crack (Figure 2.3). For the slab-column connection,
2-4
no unique value for the concrete compressive force C1 (refer Figure 2.3) at one face of
compression and increased dowel action. However, an increase in steel content in the
connection region has a limited effect on the shear strength (Park and Gamble, 2000).
The top cover concrete starts to split from the slab after a threshold value resulting in a
reduction in the dowel action and tensile force resisted by the steel.
The critical sections of the slab for both negative moment and punching shear are at or
close to the perimeter of the loaded area resulting in moment-shear interaction. The
characteristics of the failure mode and of the load-deflection curve for the slabs are
van Den (1974) (Figure 2.4) ductile flexural failure and brittle punching shear failure
are observed for slabs with both small and large slab steel ratios (refer Figure 2.5).
When a small amount of flexural steel is used punching shear failure was often observed
as a secondary failure after the slab reached its flexural strength limit.
Test data summarised in the ASCE-ACI Committee 426 (1974) report show that the
shear strength is independent of the ratio of tension reinforcement in the slab, p, and on
the yield strength of the steel, fsy· However, concentrations of top and bottom
reinforcement are recommended as the top steel improves the flexural behaviour of slab
in the service load range and the bottom steel acts as a suspension net holding the slab
to the column. Kinnunen and Nylander (1960) and Zaghlool and de Paiva (1973)
observed some increase in punching shear strength with an increase in flexural top
reinforcement but it is small and generally ignored. Once the cover to the top steel is
lost due to spalling, the top steel tears from the slab providing little to no shear carrying
Edge, comer and internal slab-column connections with non-uniform load, or unequal
spans on each side of the connection (see Figure 2.6) require the transfer of both shear
and unbalanced moment. Also any horizontal loading due to earthquake, wind, or other
Experimental
Column
load at .
- - -Idealized
,___!_.....!.._.!...,__.:__. -
failure
vtest
• /
•/
Punching shear
failure
f
• /Flexure V
• / failure
/
Slab steel ratio
Column
load
Column
load
Figure 2.5 - Effect of slab reinforcement ratio on shear strength, ductility and
load deflection curves of slab-column connections with
concentrically loaded columns (Dragosvic and van Den 1974, Park
and Gamble, 2000).
2-7
the slab and the columns. In such cases, the transfer of unbalanced bending moment
causes the distribution of shear in the slab around the column to become non-uniform,
resulting in a reduction of shear capacity of the connection. The shear force and
unbalanced moment are transferred by combined bending, torsion, and shear at the
faces of the critical section in the slab as shown in Figure 2. 7. When the shear stress
developed at any section of the slab exceeds the shear capacity, the slab fails in diagonal
tension at the side where the high shear stress is developed resulting in a punching
failure. The location of the critical section and the proportions of the moments
transferred from shear and flexure depend on the model being considered. The different
Flat slabs are susceptible to failure due to punching shear under the large early age
loads that may be imposed during construction. Generally these slabs are constructed
using proprietary flying or table forms and as a result the construction is relatively
rapid. Loads coming on the supporting slabs during construction can equal or exceed
the supporting slab's service load and load is transferred to the ground through the
immature slab and slab-column connections. In some cases slabs have been reported to
Multistorey flat plate buildings have limited use as seismic-resistant structures without
the provision of other moment resisting and stiffening elements such as frames and
cyclic bending moments causing failure of the slab due to degradation of the shear
2-9
C2
c
Load transfer detail
¢
A
Critical sections
Note: MARx and VAB are x direction moment and shear for of side AB, similarly for the
other sides.
capacity of the slab around the column. Such failure is highly brittle in nature and, as
such, the use of the flat plate construction is not recommended in high seismic regions
(ACI-ASCE Committee 352, 1989). Shear reinforcement have been found effective in
increasing the strength and in changing the failure pattern from brittle to ductile. The
amount of increase in strength and the extent of change in the failure mechanism
slab-column connections. Early tests reported (Elstner and Hognestad, 1956, Kinnunen
and Nylander, 1960, Moe, 1961, Dragosvic and van Den, 1974) were on simple
isolated specimens with the specimens extending close to the first line (nominal) line of
contraflexure (refer Figure 2.8). The effect of specimen size and boundary conditions
were not considered. Isolated small specimens were tested and observations were on
the basis of the results obtained. The design provisions in ACI-318 (1999) are based
largely on the results of empirical studies made using such models. It was observed by
Rankin and Long (1988) that these specimens do not represent the real structure well.
After the influence of specimen size and boundary conditions were better understood,
experimental models were improved to incorporate these effects into the test. ·In later
studies (Rangan and Hall, 1983, Robertson and Durrani, 1992 and Durrani et al., 1995)
statically indeterminate specimens were used for the investigations and sized to
incorporate the laws of similitude to simulate the behaviour of real sized slab-column
connections. In the following sections the literature on the tests conducted on various
1.83 metre square, 152 mm thick slabs. The slabs were loaded through centrally located
column stubs. The variables considered were concrete strength, amount and
0
Second line of
First line of ----::----;-,...-----------1 contraflexure
contraflexure 1I
I
I
I
I
1
.----+- - - - - - 1 - - - Portionscovered in more
·..
Jl
· .1 I
recent tests
f, ' ,.,. ' t
+--------~-+-~----
Portions covered ------+--..1 · - l
in early tests __J J
:
-·1I
'I
1
_________________ j
:
Second line of
l~lt<--
UFkst Une of
contraflexure
of the column support and loading conditions, the amount and the position of shear
reinforcement and loading eccentricity. In most of the cases punching failure was
were found to have no effect on the ultimate shear strength of the slab subjected to a
concentrated load through the column. However, the results showing that load
eccentricity has no effect on the punching shear strength was later contradicted in tests
2-12
undertaken by Moe (1961), Rangan and Hall (1983a, 1983b), Rangan (1990), and
Falamaki and Loo (1992) which showed clearly that the moment caused by load
For the connections without shear reinforcement Elstner and Hognestad (1956)
V2 _ V 333 0.046
-+-- (2.1)
+'- Lbd+'
Jc 8 'Jc ~~ l/Jo
where b is the circumference of the loaded area (in inches); l/Jo is the ratio of the shear
capacity to flexural capacity; f~ is the cylinder strength (in psi); dis the effective depth
of the slab in inches; Vis the shear force (in pound); and v2 is ultimate shearing
Kinnunen and Nylander (1960) conducted a series of tests on 150 mm thick circular
reinforcement. The slabs were loaded via a circular stub column located at the centre of
the specimens. The test variables were the type and quantity of flexural steel and the
ratio of the stub column diameter to slab thickness. They found an increase in the
Moe (1961) undertook an investigation to study the basic mechanisms of shear failure
by testing twelve 1.83 metre square slabs with thickness of 152 mm. The specimens
2-13
were simply supported allowing the comers to lift freely. The loads were applied at
different eccentricities through a centrally located square column stub. Holes were
provided in the vicinity of the column of the specimens to enable observation of crack
formation. During the test, cracks were observed at loads as low as 50 percent of the
ultimate load. Moe found the shear strength to be dependent on the flexure strength of
the slab. Concentration of flexural reinforcement in the narrow bands over column had
a significant influence on increasing the yield load of the slabs. However, the punching
resistance was not affected by the arrangement of the flexural reinforcement. Based on
the test results of 43 slab-column specimens, Moe developed a method for an ultimate
strength analysis for moment transfer in the slab-column connections under combined
shear and unbalanced moment. Moe assumed the critical section for shear to be at the
Moe found that the punching capacity is reduced with increasing eccentricities. Based
on the results of his tests, Moe concluded that approximately one-third of the total
unbalanced moment was transferred to the column by shear stresses and is independent
of the quantity of the flexural reinforcement. Moe also concluded that the shear
concrete.
Kinnunen (1963) improved the mechanical model of Kinnunen and Nylander (1960) to
make the model applicable to slabs with two-way reinforcement but without polar
symmetry. This modification was based on the further assumption that the
reinforcement cutting across the shear crack carries a part of the shearing force. The
2-14
modified model is as shown in Figure 2.9. In the modified mechanical model shown in
Figure 2.9, r is the radius, h effective depth, y distance from the soffit of the slab to the
root of the shear crack at failure, z distance from the neutral surface to the top surface
of the slab in the middle part of a panel in a flat slab, P the column load, R1, R2, R3 and
R4 the force resultants from the reinforcement and T the oblique compression force in
The Kinnunen (1963) model was found to be in good agreement with experimental
results for the slabs with ring and radial reinforcement. However, for the slab with two
way reinforcement the average calculated punching load and average angle of rotation
of the slab portion outside the crack at failure by punching were smaller than the
were attributed to dowel action. In the study of Kinnunen the effect of reinforcement
cutting across the shear crack was taken into consideration in transmitting a part of the
column load to the slab. The resultant force consisted of two components (i) membrane
force and (ii) a dowel force. This revised model gave a better agreement with the
experimental results for the slabs with two-way reinforcement although it is complex.
including one exterior column connection. Their main aim was to study the behaviour
of moment transfer at interior columns. The variables used in their investigation were
the loading arrangement and the location of voids adjacent to the column. Hanson and
Hanson used their test results to evaluate the design methods proposed by Di Stasio
2-15
p~.f
2n
.. .... ... ..
. . .. ... . .
~
.. . .. 1.
Note: R2 = R3 for the connections with shear reinforcement and R3 = 0 for the
connections without shear reinforcement.
and Van Buren (1960), Moe (1961), the ACI-ASCE Committee 326 (1962)
recommendations and those given in ACI 318 (1963). A modified version of the model
proposed by ACI-ASCE committee 326 (1962) was found to estimate the strength of
Stamenkovic and Chapman (1974) tested internal, edge and comer columns with
varying axial load and moment applied to the loading columns. In all 52 tests were
undertaken on 914 mm square by 76 mm thick slabs. Two column sizes were used,
127 mm square columns and 152x76 mm rectangular column with the column
extending above and below the slab. The parameters considered were the type of
loading (vertical or horizontal and combined vertical and horizontal), the column
location and the column shape. The slabs were supported along the four edges using tie
method to calculate the strength at the column head under axial force and bending.
Interaction curves were presented for all types of connections under combined vertical
'
and horizontal loading. For internal connections Stamenk.ovic and Chapman proposed a
V M
-+-=1 (2.2)
Vu Mu
where V and Mare the failure shear force and moment, respectively, Vu is the shear
capacity for zero moment and Mu the flexural capacity for zero shear.
2-17
0 M Mu
Dragosavic and Van Den (1974) conducted tests on 24 small scale slab-column
column connections (Figure 2.4). The test results showed that the punching shear
Van Den (1976) proposed an analytical model for predicting the punching shear
strength of different slab-column connections. The effects of the eccentricity in the edge
and comer column were included by introducing an eccentricity factor based on the
geometrical properties of the connection. The model was compared with the test results
integral beam and slab reinforcement pattern were the main variables of the test
programme. The 1/3 scale specimens consisted of 2.1 m wide square slabs supported
on a central 305 mm square column. Hawkins et al. found that concrete splitting and
dowel effects for the slab reinforcement passing through the column reduced the
moment transfer capacity. Also shown was that the rotational stiffness of connections
decreased continuously with the increase in load and the equivalent column stiffness
procedure of ACI 318-1983 was satisfactory only in the linear range. However, after
the propagation of cracks due to shear and torsion, the rotational stiffness was
two times the slab thickness on either side of the column. The ACI 318-1983 model
gave reasonable results for the strength of the connections provided that flexural
reinforcement within a width of 1.5 times the slab thickness on either side of the
column did not exceed 0. 7 percent. The model yielded conservative results when the
calculate the shear strength and deflection for slab-column connections. Broms
hypothesised that the failure occurs when the concrete in compression near the column
compression zone of the slab near the column face was considered to be under biaxial
assumed to occur when the stress in the conical shell reached the value 10 percent
higher than the uniaxial strength of the concrete. The critical strain of 800 J..l£ was
assumed as the strain for developing critical macro-cracks parallel to the compression
2-19
direction. These cracks enabled the shear crack to propagate through the compression
zone of the slab. The method was compared with data from a number of studies and
was shown to give a reasonable correlation with the test data for interior flat
Based on the tests of eight isolated internal square slab-column connections with
different flexural as well as shear reinforcement, Broms (1990b) concluded that with
the use of a combination of bent bars and multiple U-stirrups a good ductile
reinforcement in the form of stirrups enclosing only the tension flexural reinforcement is
ineffective in improving ductility. Broms tests showed that openings in the slab inside
the critical perimeter do not effect on the strength or stiffness of the slab.
Alexander (1990) tested 12 isolated interior flat plate-column connections. The test
200 mm x 200 mm x 200 mm column stubs above and below the plate. Four of the
specimens were reinforced with steel fibres. The main parameters considered in the test
were the boundary restraints, fibre content, clear cover to the reinforcing steel and the
spacing of the reinforcement through the column. From the results of the tests,
Alexander observed that the truss model proposed by Alexander and Simmonds (1986)
modified model referred to as the bond model. The bond model was compared against
116 test results and was shown to correlate well with the experimental data. With the
2-20
inclusion of steel fibres in the concrete mix the cracks in the specimen were reduced
Pan and Moehle (1992) tested four slab-column sub-assemblages under combined
gravity and lateral loads. Substantial reduction in strength, ductility and drift capacity
small gravity loads, the ductility and load carrying capacity were significantly improved
when lateral loads were applied. However, with the increase in gravity load significant
reductions in the shear capacity and the lateral deformation performance were
observed. Bottom reinforcement in the slab, which passed through the column,
improved the post punching behaviour of the joint. Finally, the post punching shear
strength was found to be dependent on the detailing of the top and bottom bars in the
slab. Similar observations are given by the ASCE-ACI Committee 426 Report (1974),
Park and Gamble (2000), Durrani et al. (1995) and Gardner and Shao (1996).
Chana and Desai (1993) tested a total of 14 specimens of two different geometry (see
Figure 2.11) to study the effect shear reinforcement on punching of the interior
the specimens failed in brittle punching shear mode. The maximum crack widths before
failure were reported to be approximately 0.3 mm and 0.15 mm for the small and large
action in controlling the crack widths. Based on the test results a number of design
t-0.4L-j r-o.4L---I
~ !
¥ I
f
I. I
1.5L
f
.I
I
type I type II
Farhey et al. (1995) studied the effects of attaching steel plates in the repair of the
damaged slab-column connections. The plates were bolted and glued to the slab. They
concluded that the method is effective in resisting both lateral and vertical loads. The
response of the repaired connection was found to be better than the original.
Chow and Selna (1995) developed a method for modelling the seismic response of
fit with the experimental data of others. From their analyses Chow and Selna concluded
that although flat-plate systems can withstand moderate earthquakes, they would be
subjected to the large drift causing heavy damage to nonstructural elements. The
primary mode of failure of the structure was predicted to be the punching of the slab.
Shear stresses in the slab that accompany the yield moment were sufficient to cause low
cycle fatigue failure with vertical collapse of the slab induced by punching failure due to
sway. However, with continuous bottom reinforcement in the slab the likelihood of this
mode of failure is reduced (also reported by ASCE-ACI Committee 426, 1974, Durrani
resting on rectangular hollow steel tubes to study the flexural and punching shear
failure of the flat plate column connections. A vertical downward load was applied to
the slabs via a 120 mm diameter steel tube at the slab centre. It was observed that the
load carrying capacity of the connection decreased with the increase of punching crack
inclination. Providing one ring reinforcement in the slab in addition to the minimum
orthogonal reinforcement controlled the inclination of the punching crack. The angle of
inclination of the punching crack decreased with an increase in the diameter of the ring
reinforcement. It was concluded that by changing the punching crack inclination angle,
which was achieved by providing the ring reinforcement, the mode of failure could be
Dechka et al. (2000) proposed a shear friction based mechanical model to determine
the capacity of the slab-column connections. The model was calibrated using 91 test
data on isolated interior slab-column connections with and without shear reinforcement.
The model was shown to give a reasonable prediction of the strength of the
connections.
In their 17 tests, Hanson and Hanson (1968) tested one edge column connection. The
150 mm square column at the edge of one of the longer sides. The slab was reinforced
with 1.67 % of reinforcement in each direction at both the top and bottom of the slab.
The slab is reported to have failed in punching shear. Hanson and Hanson
2-23
flexure and 40 % by the eccentricity of shear about the centroid of the critical section.
Zaghlool (1971) tested 11 edge column connection specimens. All specimens were
1.83 m x 0.97 m x 0.15m. Square columns of 178 mm, 267 mm and 356 mm were
used. The slab edges without the column were rested on the test frame through
12.5 mm thick neoprane pads. The loads were applied to the test specimens through the
column stubs. The parameters considered in the tests were the moment shear ratio, slab
reinforcement content and the column width to effective depth ratio. From the results
of their tests Zaghlool concluded that the punching of the column through the slab at
the compression zone due to the combined action of flexure and shear. It was
concluded that besides the concrete compressive strength, the quantity of reinforcement
in the vicinity of the column, the ratio of column size to the effective depth of the slab
and the ratio of the bending moment to the shear force significantly influence the
Stamenkovic and Chapman (1974) investigated the effect of axial load and moment
specimens with a square column of 130 mm. In their tests, the edge adjacent to the
column was kept free with the other three edges supported along their length using
12.5 mm and 38.1 mm diameter bars. The outline of the typical edge column specimen
r--915-----1 r--915-----1
130mm square
column
l t lJ915
p
M-P j -l,_,r--_M
130mm square
column
T
76
788 788
'-----,_[76
--rT
l 130mm square
column
........... ~p
FRONT ELEVATION
l 130mm s;;::-T
column
"""''"""~P
SIDE ELEVATION
·--·- T
Figure 2.12- Outline of Stamenkovic and Chapman (1974) specimens for the
edge column connections (a) Moment in x-direction; and
(b) Moment in y-direction.
Stamenkovic and Chapman modified the expression they developed for internal column
unbalanced moments are assumed to be transferred from the column to the slab by the
combined action of bending and torsion of the vertical slab-column interfaces normal to
the moment axis. For edge connections under normal moments (developed from
horizontal forces perpendicular to free edge, refer Figure 2.12a) a circular interaction
relation was developed and is shown in Figure 2.13. For edge connections subject to
tangential moments caused by horizontal loading parallel to the free edge (refer
Figure 2.12b) the straight line interaction formula shown in Figure 2.13 was proposed
0 M/Mu
Kanoh and Yoshizaki (1979) tested 8 edge slab-column connections to measure the
torsional shear stress on the side faces of slab-column connections. Using a beam
analogy similar to that of Park and Islam (1976), Kanoh and Yoshizaki concluded that
the ultimate torsional shear stress capacity is 0.636 .[1: MPa. This is considerably
greater than was given in ACI 318 ( 1977) with the increase in the strength attributed to
restraint of the critical section of the slab. The forces at the critical section of the slab
are shown in Figure 2.14. To calculate the ultimate strength of the slab-column
connections transferring the moment and shear, Kanoh and Yoshizaki proposed the
following equations
(2.3a)
(2.3b)
(2.3c)
2-26
Column
Figure 2.14- Forces acting at the critical section (Kanoh and Yoshizaki, 1979).
where Mo is the moment transferred to column; Mt is the flexural strength at the critical
section; Ms is the moment transferred by the shear applied at the critical section; Mt is
the torsional moment at the critical section; Mtest is the measured moment transferred to
column about the centroidal axis of column; Vo is the maximum shear capacity of the
connection; Vu is the ultimate vertical shear stress; Vtest is the measured shear force
Rangan and Hall (1983a, 1983b) tested five half-scale models of edge panels of a flat
plate floor systems. They increased the size of the specimens over those previously
tested by moving the boundary to the second line of contraflexure from the connection
(refer Figure 2.8). Two of the five models tested had a spandrel beam of
300 mm x 250 mm along the free edge. In the other four specimens a spandrel section
2-27
was provided within the depth of the slab over a width of 250 mm. The spandrel
contained closed ties at 60 or 70 mm centres. Rangan and Hall showed that most of the
floor load entered the column via the spandrel, even in case of the slabs without the
spandrel beams. Thus, the strength of the spandrel in combined shear and torsion was
From their tests on edge-column-slab connections, Rangan and Hall (1983b) observed
applied load. The restraint to the spandrel against longitudinal expansion considerably
increased its torsional strength. Once cracking occurred in the connection, the stiffness
decreased significantly compared to that of the uncracked state. Based on their test
data and the concept of lateral restraint put forward by Onsongo and Collins (1972),
Rangan and Hall developed expressions for the moment and shear transfer in the
The AS 3600 (1994) model for the design of the flat slabs against punching shear
strength is based on the results of Rangan and Hall (Rangan and Hall, 1983a and
Rangan, 1987). The physical model used to develop these models is as shown in
Figure 2.15, where C1 and C2 are the column size normal and parallel to the free edge,
respectively; bt and bz are the corresponding widths of the critical section of the slab
and are given by b1 = C1 + d/2 and b2 = C2 + d; dis the effective depth of the slab; V,
V1 and V2 are shear forces transferred to column centre and at the front or at a back
face of the critical section, respectively; M* and M1 are the unbalanced moment
transferred to the column section and the yield moment of the slab steel at the front or
at back of the critical section; and T1 is the torsional moment at a side of the critical
2-28
SLAB
Front face of
critical section
section. The critical section was assumed to be at a distance of d/2 from the face of the
column. It is assumed that the shear stress at the critical section is uniformly
transmitted to the slab by shear unlike the ACI 318-1999 model, where the proportion
of the unbalanced moment transferred by shear is a function of the column size and the
effective depth of the slab. Rangan compared the model against 104 different test
results of interior, edge and comer column connections. The average test to calculated
ratio was 1.6 with the standard deviation of 0.4 and coefficient of variation of 25 %.
2-29
Thus the design equations of AS3600-1994 are seen to be conservative. Rangan (1987)
stated that shear strengths measured in the laboratory studies were significantly
influenced by the boundary conditions adopted for the test specimens giving such
conservative results.
Rangan (1990a) used the model developed by Rangan (1987) to compare the 117 test
data covering wide range of interior, edge and comer slab-column connections. The
average of the test to calculated value was shown to be 1.59 with a coefficient of
variation of 25 percent. It was claimed that AS3600-1988 was simpler to use and a
useful altemativ~? to the ACI building code model for the design of punching shear in
Rangan (1990b) tested four specimens to study the effect of spandrel beams, closed ties
and different concentrations of slab steel on the moment and shear transfer strength in
flat-plate floors in the vicinity of edge columns. It was observed that shallow spandrel
beams with nominal closed ties are effective in preventing the punching shear failure
with the closed ties in the spandrel strip increasing the shear strength. Top
reinforcement provided within the width of column strip plus the adjacent half middle
strips controlled the transfer of bending moment to the column and was only slightly
affected by the presence of the torsion strips with closed ties. Rang an further concluded
that the ratio of unbalanced moment to the shear transmitted to the column significantly
connections with and without openings and shear reinforcement in the slabs and loaded
2-30
under high moment to shear ratios. The basic dimensions of the specimens are as shown
in Figure 2.16. All the specimens except those with stud shear reinforcement
r-1020-----j
25Dmm square
column
l
Slab
Simply
TI J1540
Column supported
H Slob
ELEVATION
failed in punching shear mode, whereas the two specimens with shear studs failed in
flexure and punching-flexure. The openings in the slabs had a significant effect in
reducing the capacity of the connections with the greatest effect for slabs with higher
moment to shear ratios. The shear reinforcement significantly improved the ductility
connections. Together with the El-Salak.awy et al. (1998) study five specimens tested
had square openings located in the neighbourhood of the connections. The main
parameters of the study were the locations and size of the openings in the slabs. The
size and location of the slab openings influenced the capacities of the connections. The
2-31
larger the openings the lower the strength of the connections. The connection capacity
also reduced when the opening was at the front of the column compared to similar
Zaghlool et al. (1970) tested four specimens of comer column connections with the
slabs resting on square columns located at the comers. All specimens consisted of a
3.05 m x 3.05 m x 0.165 m slab (see Figure 2.17) supported by 140 mm or 165 mm
square columns. Three specimens failed in shear and one failed through a support. A
linear shear moment equation was proposed similar to that proposed by Moe (1961).
However, the test data of Zaghlool et al. suggested a factor of 0.04 instead of 0.33 for
the moment reduction factor as proposed by Moe for the interior slab-column
connections. Similarly, the test results compared best with the moment-shear
interaction model of Hanson and Hanson (1968) with a moment reduction factor of
Zaghlool et al. (1970) developed a model for calculating the strength of comer column
connections considering the axial force transferred to the column alone. The model is
based on the principal diagonal tensile strength of the failure cone with an angle of
inclination of the failure plane of 45 degrees. The mean test to calculated strength for
his test results was found to be 1.03 with the standard deviation of 0.09. In the model
Zaghlool (1971) and Zaghlool and de Paiva (1973) reported tests on eleven comer
column connections. All specimens consisted of a 150 mm thick by 1.07 m square slab.
2-32
Square columns of different sizes 178 mm, 267 mm and 356 mm were used. As in the
----
/t
Reactions
edge column connections, in these tests the parameters studied were the slab
reinforcement content, the moment to shear ratio and column size to slab effective
depth ratio. The edges of the slabs without the column were rested on the test frame
through 12.5 mm thick neoprane pads. The load was applied to the specimens
symmetrically through the column stubs using hydraulic loading systems. Zaghlool and
de Paiva (1973) reported that the straight line shear hypothesis adopted in the 1971
ACI Code to find the strength of the column-slab connections gave extremely
conservative results compared to the results of their tests. Punching of the column
occurred following the destruction of the compression zone due to the combined action
of flexure and shear. It was also concluded that besides the concrete compressive
stress, the amount of reinforcement in the vicinity of the column, the ratio of column
size to the effective depth of the slab and the ratio of the bending moment to the shear
Stamenk:ovic and Chapman (1974) tested six comer column connection specimens with
the main variable being the ratio of applied axial moment. The arrangement of the
comer column specimens of Stamenk:ovic and Chapman is shown in Figure 2.18. From
their test results, Stamenk:ovic and Chapman proposed an interaction curve for comer
Walker and Regan (1987) tested eleven comer column specimens with the slab resting
on four columns. A typical specimen is shown in Figure 2.19. The bottom of the
columns were connected by steel ties (see Figure 2.19) with pinned ends. The tie forces
were determined using the strain gauges fitted to the rods. Except the specimen SC6,
provided in the diagonal directions. Most of the specimens failed in punching shear.
Comparisons of the results with ACI 318-1983 provided a good assessment of the
column moments. The results of specimen SC6 showed that the special arrangement of
negative reinforcement increased the stiffness of the slab and had some influence in the
r----915~
130mm square
l_l 915
column
p ~p
l 130mm square
column
lzzzi- ~ p
ELEVATION
-.
T
,.. 2000-------~, s 80
y---t-;:=====::::=:::;~ t
8 0
250 250
I I 5oo I 5oo I 5oo I I
Column size:
SC8, SC9, SC11: 160mmx160mm
ELEVATION SC12: 300mmx300mm
PLAN
(SC8 AND SC9)
250
300
300
300
250
250 250
500 500 500
I I I I I I
PLAN
(SC11 AND SC12)
Hammill and Ghali (1994) tested five full scale reinforced concrete flat plate
connections with comer columns subjected to shear and moment transfer. Three of the
specimens were without any shear reinforcement while the remaining two contained
shear reinforcement. The specimens used consisted of 1075 mm wide square slab by
150 mm thick and 250 mm square column. Hammill and Ghali reported that the
Canadian code (CAN3-A23.3-M84, 1984) and the ACI 318 (1989) code are
Hammill and Ghali showed that the equation developed by Elgabry (1991), based on
the linear finite element analysis, is more realistic than the code models for comer
columns. Finally, they suggested that a nonlinear finite element analysis is required to
Rankin and Long (1988) observed that the results of the tests using the conventional
specimens with the slabs extending only to the first line of contraflexure
(refer Figure 2.8) lead to erroneous conclusions. In such conventional models the
influence on the results of the boundary conditions was significant. Rankin and Long
to verify this behaviour with the reinforcement content and span to depth ratio of the
slabs varied.
outward expansion of the slab at the boundaries and if this expansion is restrained the
2-37
compression membrane action). Rankin and Long observed that in the punching of
interior slab-column connections, the slab beyond the nominal line of contraflexure
(that is beyond the first line of contraflexure, refer Figure 2.8) provides lateral restraint.
the slabs extending up to nominal line of contraflexure. Rankin and Long proposed a
was applied through the columns with all slabs failing in a punching shear mode. The
variation in the reinforcement content was found to have little effect on the ultimate
punching load of the heavily reinforced specimens, but it was effective for lightly
reinforced specimens. Johanson's (1962) yield line theory was found unsuitable in
predicting the punching load capacity for highly restrained and/or lightly reinforced
slabs. Compression membrane actions were developed depending on the degree of end
Kuang and Morley also observed that serviceability conditions such as deflections and
crack widths were affected by the degree of the boundary restraint. As may be expected
deflections were reduced when a compression membrane action develops. Crack widths
in the lightly reinforced slabs were larger than the more heavily reinforced slabs. Also
the cracks were fine and large in numbers in the slabs that were highly restrained.
2-38
Kuang and Morley concluded that compression membrane forces play a significant role
The performance of flat slab structures subjected to seismic loading has attracted
increasing attention especially because of the poor performance of this type of structure
under dynamic loading. In early studies the testing was related specially to the
subjected to a combination of moment and shear. Rangan and Hall (1983a, 1983b), and
Robertson and Durrani (1992) showed that the tests studied did not take into account
the restraining effect of the boundaries and the shift of contraflexure lines with the
deterioration of the structure as the test progresses, both of which affect the strength.
redistribution of forces within the slab. Given that the behaviour of slab-column
the column.
Regan (1978, 1981) conducted a series of tests comprising of different types of single
and multi-panel specimens with interior, exterior and comer columns. Regan tested 12
statically indeterminate slabs with interior columns and 2 slabs mixed with wall and
Gosselin (1984) tested four slab-column connections consisting of exterior and interior
panel extending laterally to the centre lines between the column. The overall size of the
slabs was 3 m x 2 m x 0.063 m. The exterior and the interior columns were of
225 mm x 150 mm and 225 mm x 225 mm, respectively, and drop panels were
provided in the column regions. The columns were post-tensioned to simulate a gravity
load equivalent to a stress of five storeys and combinations of gravity and lateral load
Lamb (1984) tested four one third scale specimens of reinforced concrete slab-column
connections with and without the spandrel beams. Of the four specimens two were
constructed with the spandrel beams. The specimens consisted of interior and exterior
columns extending above and below the slab a distance equivalent to mid-height of the
column used in the scaled prototype. The tests were undertaken by applying
combinations of lateral and gravity loads over a number of cycles. Of the specimens
lateral load. The panel dimensions were 2.9 m by 2.0 m and the slabs were 115 mm
thick as shown in Figure 2.20. Robertson and Durrani reported that an increase in
gravity load at the interior slab-column connection caused significant reduction in the
was also reported with the increase in the gravity load and was attributed to accelerated
cracking in the slab around the connection. Robertson and Durrani (1990)
2-40
12900
Lateral
0
"
L(')
..-
0
~--254mm square columns
"
Slab width = 2.0m
Reactions
Figure 2.20 - Details of a typical test specimen of Robertson and Durrani (1992).
tested altogether nine specimens including two isolated (one interior and one exterior
Falamaki and Loo (1992) tested to failure a series of nine half-scale models of
reinforced concrete flat-plate structures. They developed their specimens from the flat
plate structural system shown in Figure 2.21 with their model representing two
continuous panels of a building floor. The models consisted of six columns including
three slab-column connections included with a spandrel beam or torsion strip. The
columns consisted of rectangular steel sections with stiffness matched to that expected
from concrete columns. The steel columns were connected to the concrete slab via steel
plates and bolts. Their aim was to study the ultimate behaviour of reinforced concrete
Z•Q
.... ~ ....
~
X
I
X
c
·a
~
Jl
.....
"";:::... ;:::...
were the strength and stiffness of the spandrel beam, the slab reinforcement ratio in the
vicinity of the columns, the volume of closed ties, the width of the edge columns and
Based on their test data, Falamaki and Loo concluded that the failure mechanisms of
the comer and edge connections were greatly influenced by the distribution of slab
flexural reinforcement through the column regions and the strength of the spandrel
Whilst the tests of Falamaki and Loo (1992) is a study in the area, the use of steel
section connected to the slab via bolts must be questioned. As has been discussed
2-42
previously in this thesis, the boundary conditions are of vital importance in the
interpretation of the experimental results. In this respect the moment transfer to the
connection detail. The tests of Falamaki and Loo are for a different floor construction
Durrani et al. (1995) tested four two-bay slab-column sub-assemblies designed and
detailed for gravity load as was traditionally used in the 1950's and 1960's. The tests
were carried out under earthquake type loading with the aim to study the behaviour of
slab-column subassemblies of the buildings constructed in 1950's and 1960's. With the
increase of gravity load, the interior connections were more vulnerable in punching
shear than the exterior connections due to the lack of continuous bottom reinforcement
through the columns. The moment transfer capacity of the slab at the interior
connections was found to depend on the amount of, and the detailing of, the top and
Durrani et al. observed that under combined lateral and gravity loading the drift
capacity was reduced with an increase in the gravity load. Cracking of slab under the
I
gravity load decreased the initial lateral stiffness of the slab-column sub-assemblies
under lateral loading. As may be expected, sub-assemblies with spandrel beams were
found to have larger initial stiffness compared to those without spandrel beams. The
reduction in stiffness was greater in the exterior connection in the negative moment
region. With the loss of stiffness in the exterior connection, the gravity shear was
2-43
The behaviour of the interior connection of the Durrani et al. (1995) tests generally
agreed with the eccentric shear model of ACI 318 (1989). The shear strength of the
connections decreased with the increase of lateral drift under cyclic loading. Durrani et
al. concluded that the moment transfer capacity of the exterior connections was
good prediction could be made using the equivalent beam method, that is by combining
the flexural strength of the slab over the width and depth of the columns and adding the
Gardner and Shao (1996) tested to failure a half scale two-bay by two-bay (9 columns)
slab with span lengths of 2743 mm (see Figure 2.22). The support system consisted of
7 columns of 254 mm square and two circular columns of 254 mm diameter (one edge
column and one comer column). In their test arrangement Gardner and Shao used
supplementary supports as shown in Figure 2.23. However, because of the use of the
supplementary supports the true failure load could not be obtained for the external
columns. Based on the strength enhancement model proposed by Shehata (1990) and
the CEB (1990) size effect expression for the interior connections, Gardner and Shao
(2.4)
2-44
5867
254
D
D=254mm/
2 43
254mmx254mm
Columns
5867 [] D
~--27431--~~--2743,--~
127 254
~------5867------~
•· Ill
Ill
II II II Ill Ill
5867 CJ• Ill··'·
Ill
II II II II II
127
2743 2743
127 254
where Vu is average shear stress (MPa); Vc is a nominal shear stress of the slab (MPa);
Vu is the shear force (N);fyis yield stress of the reinforcement (MPa); u is the length of
the critical perimeter (mm); dis average effective depth of tension reinforcement (mm);
p is average of the top and bottom reinforcement content ratio in the slab; /em is mean
Although Gardner and Shao's investigation focused mainly on the behaviour of the
interior column, for comer and edge connections they suggested that designers should
follow the ACI linear interaction formula or to use simple expressions independent of
eccentricity ofload and steel content (similar to BS 8110, 1985). From the test it was
observed that the interior slab-column connection was more critical in punching shear
than edge or comer columns. Similar results were found in the study of Durrani et al.
(1995) and were attributed to the redistribution of forces from external columns to the
internal column. It was concluded that as punching shear is associated with high
strut and tie model can be used to explain better the behaviour but did not give any
further details on this aspect. Gardner and Shao concluded that isolated punching tests
can be used to study the behaviour of the interior slab-columns connections but did not
specimens. One specimen was designed for gravity load only whereas the second
specimen was designed for combined seismic and gravity load. Both specimens had a
150 mm thick by 5 m wide by 10.25 m long slabs. Each specimen had 3 columns of 250
2-46
mm square (1 interior and 2 edge columns) projecting 1.5 m above and 1.5 m below the
slab. Shear studs were provided in both specimens. The test results showed that shear
studs were effective in increasing not only the punching shear strength of these
connections under seismic load, but also in improving the ductility of the system. Shear
studs along with other reinforcement maintains a high level of concrete integrity to
allow the structure to form a plastic hinge at the connections, thereby making the
It has been reported~ the literature (Brown and Dilger, 1994, ACI 408.2R, 1992) that
concrete shear strength under cyclic moment transfer is less than the strength under
connections subjected to cyclic loading may require shear reinforcement, which may
not be required under monotonic loading. Tests have been conducted using mainly four
types of shear reinforcement (shown in Figure 2.24), conventional closed ties, bent up
Islam and Park (1976) carried out tests on concentrically loaded square columns with a
rectangular slab with and without shear reinforcement. The tests were performed under
gravity and lateral loads. The main conclusions arising from the tests were:
• Specimens without shear reinforcement failed suddenly with the splitting of the
concrete along the bars at the top of the slab on the side of the column where
1-1-
I LL
I u
1-t-
{~}
Closed ties
0
Bent up bars
I I
• Cranked bars used for the shear head reinforcement did not increase the
• Closed stirrups in the slab around the longitudinal bars passing through the
• Two legged and four legged stirrups performed equally well and were effective
in torsion and flexure and improved the strength and ductility of the system.
• The provision of closed ties formed a small in-built beam inside the slab
Elgabry and Ghali (1987) reported the test results of five full-scale reinforced concrete
flat-plate connections with interior columns. All the specimens were subjected to
shear-moment transfer and with the exception of one specimen stud shear
reinforcement was used in the column heads. The test results showed that the
stud-shear reinforcement was effective in increasing not only the shear strength of the
Elgabry and Ghali (1990) described the design and detailing of different flat
ACI 318-1989 model. Based on their test results a new design method was
recommended.
Dilger and Cao (1991) reported that the reversed cyclic loading of slab-column
connections reduces the capacity to transfer the unbalanced moment. This observation
was further verified in the study of Megally and Ghali (1994). Megally and Ghali
2-49
ratio of 1.5 percent without punching failure. Generally in seismic zones lateral load
resisting structural elements, such as shear walls, should be combined with flat plates to
ensure the lateral drift ratio does not exceed 1.5 percent.
Mortin and Ghali (1991) tested six full scale edge slab-column connection using SSR.
The use of SSR was effective in improving the shear strength by about 30 to 50 percent
and the ductility was also significantly improved. Deflections at failure were found to be
double that of companion tests on panels without shear studs. It was observed that SSR
could change the failure mode from a brittle punching shear failure to a more
The effect of the type of shear reinforcement and the shear reinforcement ratio on the
et al. (1992). Yamada et a1. tested 13 slab-column specimens. The slabs were supported
by concentrically located column stubs and were tested under symmetrically distributed
monotonic loading. Yamada et al. used hat and hook type shear reinforcement (shown
in Figure 2.25). The hat type reinforcement was prefabricated into a cage and welded
to two horizontal straight bars. The hook type shear reinforcement consisted of
individual bars with a 180 degree hooks at each end. The ratios of the shear
reinforcement varied from 0.0 to 1.98 %. The use of both types of shear reinforcement
showed a superior performance in the post peak region of the load-deflection response.
In addition a significant increase in the punching shear resistance with the hook type
shear reinforcement was reported. This increase was due to the good anchorage of the
shear reinforcement.
2-50
ottom flexural
reinforcement
H ok type boro-
rro
Every node
lL Flexural barsl!.VEvery second node
Rangan and Lim (1992) reported tests of three large scale edge slab-column
connections with one specimen containing no shear reinforcement and two specimens
containing SSR. The tests consisted of 4950 mm by 3170 mm by 110 mm thick slabs
with a 250 mm square columns located at the middle of the free (4950 mm) edge.
The slab was loaded via 24 point loads evenly distributed across the slab simulating a
uniform loading. The SSR was effective in increasing the punching strength and the
ductility of the concrete slab-edge column connection with the greatest capacity from
Megally and Ghali (1994) reported on the design of slab-column connections in seismic
ductility and drift capacity was reported with the use of shear stud reinforcement. It
2-51
was concluded that SSR is better than the conventional stirrups in improving the
Brown and Dilger (1994) tested four interior flat slab-column connection specimens
containing SSR. The specimens consisted 150 mm by 1900 mm square slabs with
250 mm square columns at the centre protruding 700 mm from each face of the slab.
The parameters investigated were the size of the shear reinforced zone and the
distribution of the main steel in the slab. The specimens were subjected to shear due to
gravity loading combined with reversed cyclic lateral displacements imposed at the
columns to represent seismic loading. Two specimens failed around the column after
the SSR yielded and the other two failed outside the shear reinforced zone before the
yielding of the SSR. All the specimens had a ductile failure mode with the final storey
drift ratio of more than 5 percent. It was concluded that the shear stress resisted by the
concrete in shear zone reinforced with SSR was about 0.15.fj[, where f~ is the
Lim and Rangan (1994, 1995) reported tests of nine large scale reinforced concrete
slabs to further study the effectiveness of SSR. Of the nine specimens seven specimens
were edge column connections, which includes the three specimens reported in Rangan
and Lim (1992) and two specimens were comer column connections. The four edge
column connection specimens not reported in Rangan and Lim (1992) consisted of a
4950 mm by 3120 mm, 110 mm thick slab with a column located at the middle of its
free (4950 mm) edge. Rectangular columns of different sizes were used in these
110 mm thick, slab with a 250 mm square column at one comer. It was observed that
the use of SSR was effective in increasing the shear strength of the slab in the
slab-column connections as was found in the earlier studies ofMortin and Ghali (1991),
Rangan and Lim's tests showed that the punching shear failure with SSR is initiated as
a combined torsion-shear failure in the torsion strips with closer spacing of the SSR
Lim and Rangan modified the truss theory of Simmonds and Alexender (1987) to
incorporate the effect of SSR in slab-column connections. The average ratio of test to
calculated shear strength was 1.02 with the coefficient variation of 10 percent. Lim and,
Rangan concluded that the strength of the slabs correlated well with the truss theory.
Gomes and Regan (1999a, 1999b) used short offcuts of steel 1-beams in the twelve
punching load was increased from 560 to 1227 kN with the use of shear reinforcement.
The mode of failure was also changed from a brittle punching mode near the column to
a ductile failure mode with failure outside of the shear reinforcement region when shear
reinforcement was added. Gomes and Regan (1999b) proposed a theoretical model
based on the results of the tests of Kinnunen and Nylander (1960), Andersson (1963)
concluded that shear capitals are not as effective as providing stud shear reinforcement
slab-column connections. Elastic plate theory, beam analogy, linear variation of shear
stress model, truss model, finite element methods are but some of the approaches used
(Aalami, 1972, Park and Islam, 1976, Simmonds and Alexander, 1987). These models
Based on elastic plate theory, Mast (1970) concluded that the straight line shear
distribution hypothesis, which considers the linear effect of the unbalanced moment in
torsional effect in punching shear. Hence, the model is only suitable for the design of
connections having columns with large depth to width ratios. Mast developed an
analogy for behaviour of the edge column connections based on the behaviour of the
interior column connections. His analogy assumes that for edge column connections
only the supporting column width is significant in carrying the reaction. Based on this
assumption various equations to find the stresses due to the applied moment at the edge
were developed. Mast compared the results of his analytical model with the
stresses determined using Mast's theory did not agree with the experimental results for
2-54
rectangular columns having side dimensions cl > c2, where c] is the side dimension
perpendicular to the axis of bending and c2 is the dimension parallel to the axis of
Aalami (1972) presented a method of obtaining the lower and upper bounds to the
rotational stiffness of flat floor plates at the column-slab connection. The method is
based on the elastic theory of isotropic thin plates. In the lower limit the column
stiffness does not add to the bending stiffness of the floor. For the upper limit the
column is assumed to make the slab-column interface infinitely stiff. Aa1ami showed
column dimensions. He concluded that the rotational stiffness of the torsion strips is not
greatly influenced by the span or the boundary conditions. This, however, contradicts
the findings of other investigators (Rangan 1987, Rankin and Long 1988) and is in
Based on elastic plate theory and on Levy's equation, Pecknold (1975) proposed an
expression for the effective width of the interior panel of column-slab connection
systems for various column sizes and slab aspect ratios. The results from the proposed
method compared well with numerical results using 16 degrees of freedom plate
bending elements.
Wong and Coull (1980) developed an influence co-efficient method to determine the
effective width and coupling stiffness of a floor slab in a laterally loaded flat
2-55
Column
plate-column structure using the classical thin plate theory based on Levy's method of
better than the finite element solution in terms of computational effort. Pavlovic and
Poulton (1991) developed an approximate closed form solution for computing the
effective slab width of the flat plate-column structure under vertical and lateral load
Pavlovic and Poulton compared their results with the methods of Wong and Coull
(1980) and Pecknold (1975) and claimed it to be more accurate and efficient.
Various beam analogies have been proposed to study the behaviour of slab-column
connections (Hawkins and Corley, 1971, Park and Islam, 1976). In the beam analogy
method the slab is idealised as beams running over the column from two directions at
2-56
right angles (see Figure 2.27) and the strength of the connection is taken to be that of
the beams. In Figure 2.27, C1 and C2 are the column dimensions and dis the effective
depth of the slab. The beam analogy transforms a three-dimensional problem into
involving the transfer of shear and unbalanced moment from the slabs to the columns.
The slab strips adjacent to the column and parallel to the column edges act as beams
under the action of flexure, torsional moment and shear force similar to that shown in
Figure 2. 7. The strength of each beam depends on the combinations of the stress
resultants and the strength of the section with respect to these stress resultants.
Assuming a sufficiently ductile system the method gives a lower bound solution to the
capacity of the connections. A detailed summary of the various beam analogy models
Hawkins and Corley (1971) developed an ultimate strength design procedure for
internal and external slab-column connections using the beam analogy. In their model
the strength of the edge connection is governed by either combined flexure and torsion
or combined shear and torsion. The applied moment is initially resisted at the front face
of the column. The flexural strength of this face in negative bending is nearly exhausted
before significant moments are distributed via the side faces by torsional action.
Hawkins and Corley assumed that once the shear on the front face is close to its
ultimate value, additional shear is transferred to the column via the side faces until a
Idealised beam
Column~
"D
+
~
[I'
I ~. . ...·lrI
G1' -
T
Slob
, . C1 +d/2 ·I
Ideolised beam
Slob
1 Idealised beam
1-
. ·. llI 1
"D
+
s C::l :~C1 s
+
fj
t-
. .l
O.Z,
• •
ctt
• r
·l -
T lt"column
T Column
Slob
CORNER COLUMN CONNECTION
Figure 2.27 - Beam analogy for the design of edge column connections (Park and
Gamble, 2000).
Park and Islam (1976) developed a beam analogy for interior slab-column connections.
They assumed that sufficient ductility is available for bending, torsion and shear at the
critical faces to allow the development of the ultimate capacities. In their model the
critical section was located at one half of the effective depth from the outside of the
' -
column. The unbalanced bending moment and the shear forces transferred by the
connection were expressed in terms of flexural moment, torsion and the shear force
transferred at each face of the critical section. An interaction diagram plotting values of
2-58
the ultimate moment (Mu) and ultimate shear force (Vu) based on the equations derived
gave slightly improved results to that of the ACI (1971) model. Park and Islam
extended their approach to cases with shear reinforcement. The shear reinforcement
considered was bent up bars, closed ties and structural steel column heads.
Other beam analogy studies include those ofZaghlool (1971) and Kanoh and Yoshizaki
(1979). These studies have been discussed in Sections 2.2.2 and 2.2.3 of this thesis.
DiStasio and Van Buren (1960) proposed a working stress method of analysis for the
moment. Di Stasio and Van Buren assumed that the shear stresses at the critical section
vary linearly with the distance from the centroidal axis of the perimeter, where the
critical section was taken at the distance h-1.5 inches from the face of the column
(h =total depth ofthe slab in inches). In considering punching shear, DiStasio and Van
Buren limited the maximum vertical shear stress to 0.0625 J; when the critical section
Moe (1961) had also used the concept of linear variation of shear stress to develop a
moment-shear interaction equation. Moe used a reduction factor of 1/3 to transfer the
unbalanced moment in the form of shear stress. Moe recommended a limiting vertical
shear stress to allow the flexural failure of the connections to govern the ultimate
In the ASCE-ACI Committee 326 (1962) report, based mainly on the studies of
DiStasio and Van Buren (1960) and Moe (1961), the concrete strength, the magnitude
of bending moment near the column and the relative size of column compared to slab
thickness were recognised as the main parameters governing the punching shear
column, where J; is the compressive strength of the concrete and d effective depth of
the slab. Based on the evaluation of 25 test results the Committee further recommended
that at d/2 from the face of the column the shear stress to be limited to 4.[1{ . The
Hanson and Hanson (1968) also developed a model based on the linear variation of
shear stress. They found a better correlation between the test and theoretical results
when the value of the moment reduction factor was taken asK = 0.4.
The American Concrete Institute gave design recommendations for the first time in its
1971 edition of the code (ACI 318-1971). For rectangular columns the moment
(2.5)
where C1 and C2 are the dimensions of the column measured in the direction of the
moment and in the direction normal to the moment, respectively and d is the effective
2-60
depth of the slab. For a square column (C1 = C2) equation 2.5 gives K = 0.4 as
recommended by Hanson and Hanson (1968). In the 1971 edition of the code a limit
was placed on the shear stress of 4..Jf: (psi) at the critical section.
ASCE-ACI Committee 426 published a comprehensive report on the state of art on the
shear strength of slab-column connections in 1974. Design methods for the shear of
reinforced concrete slab-column connections without shear reinforcement and with the
transfer of shear and unbalanced moment were introduced in ACI 318 (1977). The
design shear model adopted in ACI 318-1977 took the critical section at a distance of
d/2 from the column. The shear stress on a critical perimeter was taken to vary linearly
with the distance from the centroidal axis of the critical section and that the shear stress
is developed because of the shear force and a component of the unbalanced bending
(2.6)
The remainder of the unbalanced moment is carried by flexure at the front face of the
(ACI 352.1R-89). The recommendations were limited to columns with an aspect ratio
less than 4 and with conventionally reinforced connections without prestress or shear
constructed in regions of low seismic risk. The reduction factor rv recommended was as
given by equation 2.6 and a simplified approach based on the shear force was given for
the design of comer and the edge connections transferring moments perpendicular to
the slab edge. Guidelines for the detailing of reinforcement in slab-column joints were
Megally and Ghali (1999) proposed a model to design all type of slab-column
connections for shear. The model is based on the linear eccentric shear model of ACI
318-1995. In this model equations for the moment reduction factors Yvx and ')Ivy were
proposed in terms of the projections of the critical section on the principal axes of the
critical section, instead of the width of the critical section as used in ACI 318 (1995).
The proposed modifications are based on the results of the linear finite element analyses
of Elgabry and Ghali (1996). A further proposed change is the replacement of terms lx
and ly, which are defined as the terms analogous to polar moment of inertia, by the
moment of inertias lx and ly of the critical section about the x and y axes, respectively.
Design procedures with some examples were presented for a range of slab-column
connections.
2-62
Alexander and Simmonds (1986) and Simmonds and Alexander (1987) developed a
punching was described as failure of the slab to confine the concrete compression
forces out of the plane of the slab. The model was calibrated with the results of 48
interior slab-column connection tests under shear load and verified against the test data
investigators. The model was shown to predict well the failure of edge column-flat slab
strength of 1.09 for slabs with no unbalanced moment and standard deviation of 0.11.
Similarly the mean for 23 shear-moment tests was 1.0. The truss model for the edge
In the eight tests isolated interior column connections reported by Alexander (1990),
Alexander and Simmonds (1992a) observed that in most of the specimens anchorage
failure of many reinforcing bars occurred before the actual punching failure although
visual inspection suggested a punching failure. It was also observed that arched struts
(see Figure 2.29b) form to carry the compression from the beam strips to the column
support, rather than straight struts (see Figure 2.29a) as was assumed in the earlier
Strain measurements in the reinforcing steel in the Alexander (1990) tests showed a
curved arch profile (as shown Figure 2.30) rather than the straight line compression
2-63
Uplift
bar
jd jd
jd
Figure 2.29 - Struts assumed and observed by Alexander (1990) and Alexander
and Simmonds (1992b) (a) compression strut model; and (b) test
measurements defining compression strut.
2-65
jd
1---Column face
profile assumed in the model Alexander and Simmonds (1986). Therefore, Alexander
and Simmonds (1992b) modified the 1986 model to better describe the curved strut
profile observed in the tests. The model was developed by combining the radial arching
action effect with the critical shear stress on the critical section of the connections (also
a limiting factor for beam-action shear for punching shear failure. The punching
capacity of the slab is determined by summing the contribution of all radial strips, with
the radial strips as shown in Figure 2.31. The total uniformly distributed load in the
radial strip 2w for the length lis determined from the primary shear, where wand l are
intensity of load for half of the strip and the loaded length of the radial strip,
respectively. Three different approaches were proposed to find the value of w, which
has a significant influence on the strength calculation. The model was compared with
115 data from eight investigations. The best mean test strength to predicted strength
was obtained with the value of w determined using the ACI code model. With this
model the mean of the test data to model prediction was 1.29 with a standard deviation
of0.16.
2-66
Remote end of
radial strip
Reinforcemnt
direction
of
radial strip
Alexander (1999) further developed the bond model given in Alexander and Simmonds
(1992b). The loaded length of each radial strip was treated as a deep beam with arching
action, while the quadrant of the two-way slab was treated as for beam action
delivering an internal shear to the side faces of the supporting radial strip. The model
was described as a strip model. The comparison of the strip model with the
ACI 318-1989 code model suggested that ACI model was conservative with increasing
reinforcement ratios. Comparing the strip model with experimental data showed that, in
general, the model performed well. The use of this model for the design of interior
Shehata and Regan (1989) proposed a model for axisymmetric reinforced concrete
slab-column systems under concentrated loads. The model is based on test observations
2-67
and numerical analyses. The slab was divided into rigid segments, with each segment
bounded by two radial crack lines (refer Figure 2.32). The angle of inclination of the
internal crack surface was taken as 20° and the concrete in compression at the column
face was taken to be in a plastic state. The proposed model describes the slab behaviour
beyond the cracking stage and calculates both the punching and flexural capacities of
the slab. Comparing the results of their model with experimental data, Shehata and
Regan concluded that the model correlated well with the data.
In flat plate structures it is preferable that flexural failure takes place before the shear
connection can be determined from the yield line analysis. Under combined shear and
unbalanced moment there are primarily two yield line patterns to be investigated. These
are 1) local mechanism in the slab forming fans of yield lines surrounding the column
and 2) folding type collapse mechanism with the yield lines extending the length or
Gesund and Kaushik (1970) examined different punching shear tests to study the extent
of the validity of yield line theory in the outcomes of these tests. Expressions were
derived to find the strengths of the connections based on yield line pattern. The
arithmetic mean of the P.YieldlineiPtest calculated for 106 slab-column specimens was
found to be 1.02 with the standard deviation of 0.25, where PYieldline and Ptest are load
carrying capacity of a slab predicted by yield line theory and the load at which the
2-68
~------------------
(d-x4
a~
dO P 2n
specimen was reported to have failed, respectively. Gesund and Kaushik developed a
(2.8)
where p is the reinforcement ratio; h is the yield stress of reinforcing steel; d is the
For a value of Q < 2 bending capacity controls the strength of the slab-column system.
For 2 : : ; Q : : ; 4 the strength is controlled by either flexure or shear and for Q > 4 the
punching shear strength controls the design. It was recommended that the
Post-tensioned prestressed concrete slabs are often used for the flat slab construction.
The design of a post-tensioned slab must ensure that the concrete stresses are not
excessive at service loads, the slab must be strong enough to resist the factored design
loads and the deflection of the slab should be within allowable limits.
Typical flat slabs have inherent ductility in flexure. A two-way slab is highly capable of
redistributing local loads by finding alternative load paths. Tests on real structures
(Ockleston 1955, Vecchio and Collins 1990, Collins and Mitchell 1991) indicate that
2-70
Early tests on prestressed flat plates were undertaken on square slabs resting on
supports without monolithically cast columns. Scordelis et al. (1956) conducted a test
each direction. The tendons were arranged to give a uniform prestress of 2.8 MPa. The
slab was supported on roller supported base plates at its comers and a ductile flexural
concrete flat slab with a parabolic cable profile. The slab was 75 mm thick and
consisted of two bays in each direction. The overall dimension of the square slab in
each direction was 4.6 m. The slab was supported on nine supports with rocker and
roller arrangements at each support to allow for sufficient rotations and horizontal
movements without any restraint. The average prestress was 1.1 MPa. After extensive
flexural cracking the slab failed at an applied load of 17.3 kPa with the centre support
punching through the slab at failure angle of approximately 45 degrees. This was the
first time punching shear was identified in a test series on prestressed slabs.
Since the early tests of Scordelis et al. (1956) a number of similar studies were
undertaken (Scordelis et al., 1958, Gerber and Bums, 1971, Smith and Burns, 1974).
However, tests undertaken in the earlier period did not have the column monolithically
cast with the flat plate. Some of these tests were for the simulation of lift slabs, which
2-71
are slabs cast on the ground and lifted up the columns, placed in position to the final
height and locked into place using a shear key inserted in the column.
Gerber and Bums (1971) reported tests on ten prestressed concrete slab specimens.
These specimens consisted of four cast in place and six lift slab specimens. All slabs
were 3.7 m square and 180 mm thick representing a half scale model of an interior
prototype flat slab. The slabs rested on 300 mm square columns. The total lengths of
the columns were 1380 mm including a projection of 600 mm above the slab. The
objectives of the tests were to study the shear and flexural capacities and to investigate
the effect of various types of reinforcement and the arrangement for crack control and
load carrying capacity. All the specimens were reported to fail in shear followed by
flexural failure with the yielding of the bonded reinforcement. The bonded
reinforcement played a significant role in the control of crack growth and in increasing
the ductility. Passing post-tensioning tendons over the column made the connection
capable of transmitting load to the column after the failure of the concrete adjacent to
the column. Analyses of the results by Gerber and Bums show that the ACI 318 (1963)
Smith and Bums (1974) tested three one-third scale post-tensioned flat plate specimens
to failure. The slab consisted of a 2. 74 m square plate with a 200 mm square column
stub at the middle. The slabs were 70 mm thick with parabolic prestressing cables with
an average prestress of 2.2 MPa. The tests were conducted under symmetrically applied
line loads placed at 600 mm from the centre, as shown in Figure 2.33, via a
whiffle-tree. The main aim of the study was to investigate the effect of adding
2-72
channel sections
2 00
1200
I. 2700
.. I
c====L=in=e==·lo=a=d==~.-==~==L=in=e=~=lo=a=d==~~
t
Figure 2.33- Loading system for the specimens of Smith and Burns (1974).
2-73
bonded reinforcement to prestressed concrete flat plate systems. The shear capacity, the
flexural strength and general behaviour of the structures were analysed. All specimens
were reported to have failed in a combination of flexure and shear with the final mode
of failure being punching shear. In all tests the failure surface outlined a truncated
pyramid. The failure surface extended conically from the perimeter of the column at the
bottom surface of the slab to the perimeter on the top surface of the slab. The slope of
the surface was observed to be significantly less than 45 degrees. The inclusion of
Bums and Hemakom (1977) conducted a series of multiple panel tests to study the
behaviour of prestressed concrete flat plate structures up to collapse for a range of load
cases. The safety of prestressed concrete slabs against punching shear failure was also
investigated. The variables studied were the distribution of cracking, the strength gain
due to inclusion of bonded non-prestressed reinforcement and the stress increase in the
Trongtham and Hawkins (1977) report the results of tests on five unbonded
post-tensioned flat plate slabs. Four of these slabs represented interior slab-column
connections and one a typical exterior slab-column connection. The model slab
representing the exterior connection was loaded incrementally with shear and moment.
From their study Trongtham and Hawkins concluded that the ACI 318-77 approach
could be applied for calculating the capacity of connections transferring moment as well
as shear.
2-74
slab-column subassemblages. Out of the total six specimens tested, four were to
simulate interior column connections and one each to represent edge and interior lift
slab-column connections. The sizes of all interior column specimens and the edge
column specimen were 3.96 mx2.13 mx0.15 m and 2.13 mx 2.16 mx0.15 m,
respectively. All the specimens except the lift slab specimen consisted of 356 mm
square columns. The main objective of the study was to determine the strength and
Hawkins reported that in their tests all the interior column specimens failed in punching
shear. The bonded reinforcement at the edge column connection had yielded before the
failure. Based on the test results, Hawkins recommended the provision of bonded
connections when the shear stress in the critical section exceeds 0.17 .fj{. Hawkins
also recommended that tendons bundled through the column, or over the lifting collar,
Cooke et al. (1981) published the results of tests on 12 prestressed concrete slabs, nine
with unbonded tendons and three with bonded tendons. The span to depth ratio was
varied in the study. The results were compared with the predicted values using
Pannell's method (Pannell, 1969) for calculating the flexural strength of slabs having
taken into account the effect of the span to depth ratio with a wide scatter of results
observed. The reasons for the scatter were attributed to the differences in the loading
arrangements, the duct material used for the tendons, the deviation of actual cable
2-75
profile from the proposed profile and inaccuracies in measuring techniques for the
ACI 318 (1963) model to estimate the ultimate strength in prestressed concrete
structures.
plate-edge column connections. They carried out tests to failure on four, two-third
scale, models of unbonded post-tensioned two way edge slab-column connections. The
ACI-ASCE committee 423 (1974). Each of the specimens consisted of a 1.52 m square
prestressed concrete slab by 100 mm thick. A 305 mm square column was located at
the middle of one edge of the slab with the top and bottom ends of the column held in
position. The overall length of the column was 2.03 m. The specimens were reinforced
with 11 by 9.5 mm diameter tendons with an average prestress level of 1.65 MPa in the
slab. All the tendons were grouped in a narrow band over the column line in one
direction while the tendons in the orthogonal direction were uniformly distributed
In the Sunidja et al. tests the closely spaced tendons were assumed to create beams in
one direction while the :flat plate acted as a one-way slab between these beams in the
transverse direction. In the first two specimens the banded tendons were perpendicular
to the direction of the exterior edge of the slab, while in the other two specimens the
bonded tendons were parallel to the exterior edge of the slab. The loads were applied
using a whiffle-tree at four points lying in a line parallel to the free edge and opposite
2-76
2-f_ 1520
-;Io
-t- 405
1 h _J_ 1
L I 15o 152~--- X
f-J I
-+5 280
-----.!-
1030
115
+
2_j
Figure 2.34 - Typical tendon arrangement for Sunidja et al. (1982) tests
(shown for specimens S3 and 84).
to the column. The distance of the line from the column was varied to vary the moment
to shear ratio transferred from the slab to the column. The results showed that the
ultimate bending moment transmitted to the column and the ductility of the specimen
decreased as the shear between the slab and column increased. The cracking load and
the strength of the connection were increased when the prestressing tendons were
Sunidja et al. (1982) compared the strength of their experimental connections with
different models such as the ACI 318-1977 model, the modified beam analogy of Park
2-77
and Islam (1976), the yield line method and with finite element analyses. The beam
analogy proved the best approach for calculating the strength of the tested prestressed
concrete slab-edge column connections. With the increase of shear transmission from
the slabs to the columns the ductility, as well as the ultimate bending moment
transmitted to the column, decreased. The crack development and its effect on ultimate
problem.
Sunidja et al. remarked that punching remains a problem with no simple solution and
recommended that further research be undertaken with models that take into account
the three-dimensional nature of punching shear behaviour. Sunidja et al. concluded that
additional data and new experiments were required to understand the true behaviour of
Martinez-Cruzado et ai. (1994) conducted tests on four interior, two edge and two
earthquake loading. The tendons in the slab were banded in one direction and uniformly
distributed in the orthogonal direction. The slabs were reinforced with mild steel bars
satisfying the ACI 318-1989 minimum requirements for bonded reinforcement. It was
observed that the ACI 318-1989 and the Park and Islam (1976) models were
conservative for the design of interior connections. Repaired connections were also
2-78
tested and the efficacy of various retrofit and repairs strategies were identified. Using a
number of repair technique the retrofitted connections were found to be stronger than
Based on their test data, Loo et al. (1995) developed formulae for the punching shear
strength of post-tensioned flat plates with comer and edge columns with and without
spandrel beams. The shear strengths measured from the laboratory tests were compared
with various predicted strengths from various design models. Loo et al. claimed that
their procedure gave satisfactory and consistent results while the British standard
procedure was described as being conservative and the Australian and ACI approaches
as being unsafe.
Many investigations have been carried out to analyse the behaviour and the strength of
these studies the use of finite element method was limited to elastic section analysis due
power through the 1990s, 3D non-linear modelling of concrete structures has become a
powerful tool in analysis and design. In the case of slab-column connections, non-linear
load transfer mechanisms and behaviour of connections. In this section, finite element
studies undertaken to date on the load transfer from slabs to columns are summarised.
2-79
Long (1967) reported the results of theoretical analyses of punching shear problems of
slabs having flexural reinforcement in two directions and for slabs without shear
reinforcement. Elastic theory was used to formulate the finite elements with the failure
load determined based on an orthogonal shear criterion for the concrete. Several
approximations and correction factors were employed in the analyses. Materson and
Long (1974) carried out further studies on Long's work using finite element analysis
for plate bending. However, the analyses were limited to slabs in which the flexural
Mehrain and Aalami (1974) evaluated the lower and upper limits to the rotational
stiffness of the junction of the slab and the column of square flat slabs with centrally
located square columns. They conducted finite element elastic plate analyses of a
and effectively expressed the rotational stiffness behaviour of flat slabs. Furthermore,
Mehrain and Aalami concluded that the boundary conditions parallel to the direction
considered did not significantly affect the slab stiffness and hence might be neglected in
Sunidja et al. (1982) carried out a linear finite element analysis of one of their
prestressed slab specimens. Sunidja et al. used four node flat rectangular elements
capable of handling flexural and membrane forces. Each node had six degrees of
freedom consisting of three translations and three rotations. A finer mesh was used in
the immediate area around the column and a coarser mesh away from the column. The
prestressed forces were defined to be uniformly distributed along the outer edge of the
2-80
elements. Since the analysis was limited to a linear analysis, only qualitative information
could be obtained about the behaviour of the tested specimen. Sunidja et al. stated that:
"As more sophisticated elements are developed that allow cracking due
It was also concluded that punching shear is a complex three dimensional problem
using 3D linear elastic finite element analyses. The pattern of one of the finite element
meshes used by Barzegar et al. is shown in Figure 2.35. As the column size was
increased, the connection became more rigid and the load was found to concentrate
more near the edge of the slab and comer of the slab-column interface. Barzegar et al.
also concluded that in the elastic range, the transfer mechanism of unbalanced moment
from the column to the floor slab may be assumed to be independent of slab aspect
Zhou and Jiang (1991) applied the non-linear finite element method to analyse an
stud. The failure criterion and the equivalent stress-strain relation adopted for the
(!)
-
E
c
.....
Symmetry lines
+ -'
c ~----,,____ X
~
c 2u square
>
·:;
o-
w
-.--
.0
- ... ·-
Plane of symmetry
PLAN
Colu mn
Plane of symmebw
Plane of symmetry
Mesh pattern
Figure 2.35- Plan of the slab-column connection analyses (Barzegar et al., 1991).
2-82
stress levels and post peak characteristics were incorporated using a secant modulus
formulation. The steel reinforcement was assumed to be elasto-plastic and perfect bond
between the steel and the concrete was assumed. Both radial and circumferential
reinforcement were considered as bar elements, with the radial bars along the sides of
the triangular elements and the circumferential reinforcement at the nodes. Although it
is said that bar elements were used to model the circumferential reinforcement, it is
Zhou and Jiang modelled the slab-column connection of Kinnunen and Nylander (1960)
using the finite element mesh shown in Figure 2.36. Displacement control was used to
solve the problem with sufficiently small displacement increments used to allow for
stress redistribution within the slab. The predicted crack pattern and ultimate load were
close to the experimental results. The punching shear failure was concluded to be due
to the interaction of normal (compressive) and shear stresses over the failure surface. A
sketch of the analysed stress distribution pattern along the punching failure surface is
Guan and Loo (1994) implemented a non-linear finite element model for the failure
Eight-node degenerate Mindlin shell elements were used having 5 degrees of freedom
per node (3 translation and 2 rotations). Shear deformations were not considered in the
study. The failure behaviour of the slab was studied using the three-dimensional
constitutive laws and the four-parameter failure criterion of Ottosen (1977). Each layer
~------------920------------~
~-----------865-----------.~
0
LO
450
Figure 2.36- Dimension and finite element mesh (Zhou and Jiang, 1991).
Applied load
thickness with perfect bond being assumed. The results of the analyses implemented for
the comer column specimens of Zaghlool and de Paiva (1973) and for other slabs were
Guan and Loo (1995) modelled a plane stress problem using a layered finite element
model with transverse shear deformation. Cracking was modelled in the concrete using
2-84
a tension cut off model Cracking was taken to occur when a specified principal tension
was reached with the cracks forming in the direction normal to the principal tension
stress. After the crack, the principle stresses are released and the remaining stresses are
distributed. A maximum of two sets of orthogonal cracks were allowed to form at each
sampling (Gauss) point. Comparisons of the results using the deflections and the crack
pattern for the reinforced concrete slab were concluded to be satisfactory. The
proposed method was claimed to be capable of predicting the transverse and punching
Guan and Loo (1996a, 1996b) used a 3D non-linear finite element model to study
punching shear failure behaviour of reinforced concrete flat plates with spandrel beams
were used and a strain-hardening plastic material model (Owen and Figueiras, 1984)
used to represent the concrete. The reinforcing steel was smeared through layers within
the mesh. Shear deformations associated with the Mindlin hypothesis were also
considered. The closed ties were simulated in the analysis considering the contribution
of out-of-plane shear reinforcement. Thus, a full interaction between the spandrel beam
Guan and Loo (1996b) compared 11 half-scale reinforced concrete flat plate specimens
and 4 single slab-column specimens. The results were compared with the
semi-empirical model of Loo and Falamaki (1992) as well as different design code
procedures. It was concluded that the Loo and Falamaki model is accurate and more
Elgabry and Ghali (1996a, 1996b) proposed a new model for the calculation of the
In developing this model linear 3D finite element analyses were used to fix the
nodes on the middle surface were used for the finite element analyses. Each node had
six degrees of freedom (3 translational degrees of freedom and 3 rotations) with both
shear and bending deformations considered in the element formulation. The finite
element mesh shown in Figure 2.38a was used to model the comer column connection
shown in Figure 2.38b. In Figure 2.38 cy , cy and L are the dimensions of column in
the x and y directions and centre to centre distance between columns, respectively. The
nodal translation in the direction normal to the slab (w) and rotations ( Oy and By)
were prescribed for the column. Finer elements were used in the vicinity of the columns
in all models.
Elgabry and Ghaliexamined a total of 189 critical sections for interior, edge and comer
columns with different slab aspect ratios. The distance between the critical section and
the column face was varied between half the effective depth of the slab to eight times
the effective depth. From the analyses the elastic values for the component of
To arrive at more realistic results, however, a reduction factor of 0.15 was applied to
the value for distribution factor rv. that is rv =0.15rv,elastic. Curves were selected to
envelop the results of the finite element analyses for the plots of 'Yv verses
2-86
®y
wL®x
Nodal degree of freedom
"' F
.,"'~ ®y-=0_
.,"
E
"-
H
J I
- G
Edge column Interior column
(a) (b)
Figure 2.38 - Elgabry and Ghali (1996) (a) Finite element mesh and boundary
conditions for corner slab-column connections; and (b) top view
of flat plate showing corner panel.
l/lx where ly and lx are the length of the projections of critical section perimeter on the
principal axes. For internal columns ~ was determined to be a function of the shape of
the critical section. These results indicated that ACI 318-95 and CAN3-A23.3-M94
method used to calculate ~ at d/2 from column faces for all cases is incorrect.
Megally and Ghali (1996) developed a nonlinear finite element model to study the
20-node isoparametric brick elements. The element had three translational degrees of
2-87
freedom at each node. The concrete constitutive model used was capable of
considering concrete cracking, crushing and post cracking shear degradation. Perfect
bond was used between the concrete and steel with an elastic-plastic stress-strain
model Based on their parametric studies it was shown that the equation developed by
Elgabry and Ghali (1996b) for %, combined with the linear equation given by
CSA-A23.3-94, was suitable to calculate the maximum shear stress distribution in edge
column connections.
Loo and Guan (1997) developed a non-linear layered finite element model to analyse
cracking and punching shear failure of reinforced concrete flat plates with spandrel
beams or torsion strips. In the model the full interaction between the spandrel beams
and the adjoining slab was considered. The transverse shear deformations associated
with the Mindlin hypothesis were also considered. A plasticity based incremental
stress-strain model (Owen and Figueiras, 1984) with smeared cracking was adopted for
the concrete with the contribution of the out of plane shear reinforcement and the
Loo and Guan compared their results of the numerical model with test results and a
good correlation was claimed. It was also reported that the performance of the
proposed analyses was satisfactory and consistent for flat plates with either spandrel
beams or torsion strips. The approach of ACI 318-1989 and BS8110-1985 were
Polak (1998) examined the applicability of non-linear layered finite element modelling
based on the shell formulation, on the punching shear problem of reinforced concrete
2-88
slabs. The elements used had five degrees of freedom at each node (3 translations and 2
rotations). The uncracked concrete was treated as a linear elastic material and the
model formulation was based on the compression field theory of Vecchio and Collins
(1986). A smeared rotating crack approach was adopted. The model was implemented
on the slabs of Yamada et a1. (1992) and Elstner and Hognestad (1956) failing in shear
and/or flexure. Although the finite element model predicted the behaviour well, the
actual strength of the slab was underestimated by about 20 percent. This was attributed
Hueste and Wight (1999a, 1999b) developed a 2-dimensional nonlinear finite element
dynamic loading. The model could account the loss of rotational stiffness caused by the
punching damage as well as predicting the punching shear failure. The model was
applied to a four-story reinforced concrete framed office building that was damaged
during the Northridge earthquake. It was observed that the inclusion of punching shear
failure in the model modified the overall response of the building in terms of drift,
Megally and Ghali (2000a) observed that the fraction of unbalanced moment
transferred between the column and slab given as~ in ACI 318-1995 is adequate when
the critical section is considered at d/2 distance from the face of the column, where d is
the effective depth of the slab. Further, it was concluded that the expressions for ~
developed by Elgabry and Ghali (1996), using linear finite element analysis are
connections using nonlinear finite element analyses. The concrete was modelled with
20-node isoparametric brick elements with three degrees of freedom per node. The
constitutive relation was based on the classical incremental theory of plasticity relating
the increments of plastic strain to the state of stress. Tension softening and shear
retention were included in the model for the cracked concrete. The steel reinforcement
was modelled using bar sub-elements embedded within the concrete elements.
Ozbolt et al. (2000) implemented a three dimensional nonlinear finite element model
based on the microplane model with a smeared crack approach to analyse interior
slab-column connections. The parametric studies showed that fracture energy and the
reinforcement ratio significantly influenced the punching load. However, the influence
2.6 Conclusions
behaviour of different types of flat plate-column connections. These studies have been
focussing on laboratory tests with some analytical modelling. The size of the test
specimens varied from the very small (Dragosavic and van den Beukel, 1974) to the
relatively large multi-panel specimens (Robertson and Durrani, 1990, Gardner and
Shao, 1996).
2-90
In many studies the test specimens are developed as scale models of a prototype
structure ignoring size effects. However, the scale effects are yet to be verified. The
bu1k of the research has been for gravity loaded structures with a few more recent
studies considering both gravity and lateral loading. Besides the theoretical approaches
based on plate theories and beam analogy some research has been undertaken using
Code models have been developed mostly from research on interior column
connections with significantly fewer studies for edge and the comer column
are limited in number. The literature review has highlighted a lack of data on
prestressed edge connections and no studies were found on the testing of prestressed
Most of the previous studies are based on the laboratory tests on scaled specimens.
From these studies and other analytical investigations it is well understood that the
highlighted. However, only a few studies have been carried out using finite element
modelling.
CHAPTER 3 - FINITE ELEMENT MODELLING OF PUNCHING PROBLEMS
3.1.1 Introduction
Different material models are used to represent the behaviour of the concrete and the
reinforcing steel at various stress states. In this thesis the model developed by Rots
(1988) and implemented in DIANA version 6.2 (1997) is used for numerical
modelling. A smeared tension cracking model for the concrete with a plasticity model
For small displacements, without an appreciable change in geometry, th~ total strain e
(3.1)
behaviour. The total stress at any time (t) is a function of the total strain e and it is also
a function of stress and strain history. The stress and strain history of the material is
of the concrete at different levels of stress. The elastic relationship between the total
(3.2)
where
T
<1 = [axx <Yyy <Yzz <Yxy <Yyz azx] ; (3.3a)
ee = ~ix e
Eyy ee
zz ri;, e
ryz r:X] (3.3b)
E* E*
1 0 0 0
E* 1 E* 0 0 0
E(l- J.l) E* E* 1 0 0 0 (3.4)
D=
(1 + J.l)(1- 2f.l) 0 0 0 G* 0 0
0 0 0 0 a* 0
0 0 0 0 0 a*
(1- 2J.l)
where E is the elastic modulus, J1 the Poisson's ratio, E* = f.l and a*
(1- J.l) 2(1- J.l)
The yield condition governs the state of stress at the initiation of plastic flow. The
yield condition is a fu~ction of the stress vector and the internal state parameter (a
hardening parameter). If the value of the yield condition is less than zero, the system is
in an elastic state. A state with the yield condition greater than zero is not admissible
f(o;7()=0 (3.5)
The inelastic strain rate vector is a function of the state of stress and the j-th plastic
·P _ ~ 1 ag;
e -""A,·-
j=l J aa
(3.6)
3-3
The proportionality constant (ij) is a scalar such that ij ~ 0. If the yield function is
The evolution of the internal state parameter 7C is given as a function of the stress
vector cr and the plastic strain rate vector eP, ie. 7C = 7C(a,eP). Using the consistency
condition j =0 the relationship between the stress rate vector and the strain rate
vector is derived as
(3.7)
indicate that under compressive loadings, inelastic volume contraction occurs at the
ultimate stress (Chen and Han, 1987). Inflection points are usually observed and this
sort of behaviour violates the associated flow rule. A plastic potential function g other
than the loading function/is therefore required to define the flow rule. In such cases a
non-associated flow rule f '# g is adopted. For f '# g the tangent stiffness matrix. on
constitutive law. The uniaxial compression curve, shown in Figure 3.1, is represented
compressive stresses (if any) are modelled using the Drucker-Prager yield criterion.
Stress
t:u = C:X1 ~r
Figure 3.1- Stress-strain curve for concrete in uniaxial compression and tension.
(3.8)
-1 -1 0 0 0
2
-1 2 -1 0 0 0
-1 -1 2 0 0 0
P= (3.9)
0 0 0 6 0 0
0 0 0 0 6 0
0 0 0 0 0 6
where if> is the angle of internal friction, l/>o the initial angle of internal friction,
plastic strain.
3-5
The Drucker-Prager yield condition with tension cut-off is shown in Figure 3.2. In this
figure it is shown that the deviatoric stress vector r ( r = ~2 J 2 ) makes an angle (J with
the first principal stress in the state of stress projected on the deviatoric plane. The
length along the hydrostatic axis (g) is given by~ = I d..f3 . The first invariant of the
stress tensor, /1 and the second invariant of the deviatoric stress tensor, hare given by
where Oi (i=1, 2, 3) are principal stresses and r', f are values of r and ~
After yielding of the concrete a non-associated flow rule is used with the plastic
(3.12)
with
(3.13)
and where 1fl( r<) is the dilatancy angle required to model volume changes due to shear
distortions. The dilatancy angle represents the ratio of plastic volume change over
plastic shear strain. For three-dimensional structures under low levels of confmement
lf/(7<) was found to affect the load carrying capacity (Vermeer and de Borst, 1984) but
its effect vanishes with high confining pressures and ranges from 0 to 20 degrees. The
relationship between the friction angle (l/J) and the dilatancy angle (lfF) is taken as
3-6
(a) (b)
Figure 3.2 - Drucker-Prager yield condition with tension cut-ofT (a) 1t-Piane
(3.14)
The relationship between the state variable and the rate of plastic strain in the minor
(3.15)
_ 1-a
1 (3.16)
c = fcp f3o
where fcp is the uniaxial compression strength of the concrete and CXJ and f3o are as
For concrete, the angle of internal friction is reported to be within the range of 30 to 35
degrees (Vermeer and de Borst, 1984), although in some cases angles as high as
l/J = 37 o have been used (Nielsen et al., 1978). In this thesis the angle of internal
friction was taken as l/J = 30 o • The cohesion-plastic strain curve for the concrete
(1951) stress-strain model with a strain in the concrete at the peak uniaxial
compression stress of 0.002. In this thesis the dilatancy angle was taken as
The multi-directional smeared crack model of de Borst and Nauta (1985) and
Rots (1988) is used for the cracked concrete. The total strain e is decomposed into an
(3.17)
The strain vectors in equation 3.17 relate to the global coordinate axes and for the
three dimensional configuration they have six components. The global crack strain
vector ~r is given as
ecr _
-
r~xx
cr ecr
yy
ecr
zz
ycr
xy
ycr
yz
ycr] T
zx (3.18)
where x, y, and z refer to global coordinate axes. The crack stress-strain laws are
defmed with the local coordinate system aligned with the crack plane as shown in
where e~~ is the normal strain for the mode I crack and r~~, r~[, respectively, are the
shear strains for the mode ll and mode m cracks. The relationship between local crack
(3.20)
z2 lxly lz lx
X
m2 mxmy mzmx
X
n2 nxny nznx
N= X (3.21)
2lxmx lxmy+lymx lzmx +lxmz
2mxnx mxny+mynx mznx +mxnz
2nxlx nxly+nylx nzlx +nxlz
and where lx, mx and nx are the direction cosines for the strain normal to the crack
plane (n-axis) relative to the global co-ordinate system (similarly the direction cosines
3-9
with the subscript y refer to the s-axis and those with the subscript z to the local
t-axis).
(3.22)
where t~' is the mode I normal traction and t;' , tf are mode II and mode III shear
tractions, respectively. The relationship between the global stress and the local
(3.23)
point, the crack strain ecr is further decomposed as the sum of the global primary
crack ef' and the global crack strain increment owing to the secondary cracks, that is
Each (fixed) crack is assigned its own local crack strain vector ef', its own traction
vector tf' and its own transformation matrix Ni according to equations 3.19 to 3.22.
The relationship between the crack-strain and traction for the multi-crack system is
(3.25)
T
and ecr = ~f' e~r ...]
Finally the relationship between global stress a and global strain£ is given by
(3.26)
where De is stiffness matrix corresponding to elastic strain (Rots, 1988). The cracking
stiffness vcr is the corresponding matrix assembled for all the cracks formed at a
sampling point with Dcr for the n-th crack denoted as D~r and given by
EEt
0 0
E-Et
vcr- {3 E
n - 0 0 (3.27)
1 - {3 2(1 + Jl)
0 0 __L E
1- {3 2(1 + Jl)
where E and Et are the tangent moduli of elasticity for the ascending and descending
branches of the concrete stress strain relationship for tension. The Poisson's ratio is
given by Jl and {3 is a shear retention factor to account for the effects of aggregate
(3.28)
For concrete in tension a linear tension cut-off model was used with smeared cracking
added to the plasticity model discussed above. A crack arises if the major principal
tensile stress (a1) exceeds either the uniaxial tensile strength of the concrete or if
(3.29)
3-11
where C13 is the minor principal stress and f cp is the strength of the concrete. Multiple
cracking at a gauss point can occur provided that the principal tensile stress exceeds
the maximum stress condition and the angle between the principal tensile stress and an
With the use of a threshold angle the number of possible crack formations at a point is
controlled between the fixed single crack model (a = 90 degrees) and the rotating
crack model (a= 0 degree). In this investigation a value of a= 15 degrees was used.
The uniaxial stress-strain curve used for modelling the concrete is shown in
Figure 3.1. In the case of plain concrete, crack propagation and fracture are primarily
dependent on the behaviour of the material in tension. It has been well established that
1985). Upon loading, a limited numbers of such cracks can develop anywhere in the
specimen but if, at a point in the specimen, the tensile stress reaches the limiting
strength (/t) all additional deformation due to micro-cracking localises within the
fracture zone. In the fracture zone, the stress gradually decreases while the strain
After cracking the stiffness of the reinforced concrete structure is reduced. However, a
concrete block between two adjacent cracks is capable of resisting tensile forces by the
bond between the concrete and the reinforcement. Due to the tension stress between
observed in the reinforced concrete as compared with the effect of the reinforcement
A number of models have been developed to consider the tension stiffening effect. The
Scanlon used a modified stress-strain curve for the concrete in tension by assigning a
component to the descending branch to account for the retained stiffness upon
cracking. Gilbert and Warner (1978) artificially increased the stiffness of the
concrete panels was not always as successful (Hu and Schnobrich, 1988). Lin et al.
(1975), Cope (1986) and Prakhya and Morley (1990) developed various approaches
such as Petersson and Gustavasson (1980), Bazant and Oh (1983, 1984) and Sih and
In this investigation the tension softening and tension stiffening effects in the cracked
concrete were simulated using the simple linear tension-softening model, as shown in
Figure 3.1. The slope of the descending branch is adjusted to allow for tension
softening and tension stiffening using the model of van Mier (1987). The strain for
zero stress, post cracking ( eu) is expressed in terms of the cracking strain (ecr) such
that
where Bcr = ft / Ec ; Gt is the fracture energy; h is the crack band width (equivalent
length of the mesh); ft is the tensile strength of the concrete and Es and Ec are moduli
3-13
Different investigators have selected various shapes for the softening response of
concrete in tension and different strain levels for eu . In most cases, the selection of the
multiplier az depended on the experience of the analyst with the specific problem. For
example, Lin (1973) used a value of about az = 5, Gilbert and Warner (1978) and
a1 =az Ec fsy where a2 is in the range of 0.3 to 0.8. Barzegar and Schnobrich (1990)
Es ft
presented different expressions for different problem types for ultimate cracking strain
using a nonlinear tension stiffening model similar to that used by Scanlon (1971). In
In this study, equation 3.30 is used with a2 = 0.5 for reinforcing bars with
fsy $; 550 MPa and a2 = 0.4 for bars with fsy > 550 MPa. Sensitivity tests for the
The development of cracks in concrete reduces the shear stiffness but (not to zero)
Iyengar 1988, Barzegar 1989). This effect is taken into account by the introduction of
a shear retention factor (/3). In this study a constant shear retention factor is used with
3-14
f3 = 0.2 unless stated otherwise. Sensitivity testing for the influence of f3 is presented
in Chapter 4.
A bi-linear or tri-linear elasto-plastic model was adopted for the conventional and
Figure 3.4b, fsyi is the steel stress corresponding to the strain of Bsyi and Ep is the
tangent modulus of the steel stress-strain diagram in the post elastic region. Perfect
bond was assumed between the concrete and all conventional reinforcement and for
Many solution procedures are available to solve the nonlinear finite element problems.
The solution procedure adopted is important as it influences the time required for
solution, disk space and the accuracy of the results. In nonlinear problems, the solution
procedure has to be carefully chosen in order to get the results efficiently without
In this study, a modified Newton Raphson iteration procedure was used with the
stiffness matrix updated at the start of each load step. Load control was used in the
early (elastic) part of the analyses and load-displacement control via an updated
normal arc-length procedure was used in the latter part of the analyses. Except where
mentioned specifically convergence was set at 1 percent of the energy norm with a
Stress, a
Stress, a
Figure 3.4 -Material model for steel (a) Elasto-plastic bi-linear stress-strain
curve; and (b) Elasto-plastic tri-linear stress-strain curve.
3-16
3.2.1 Introduction
The accuracy of the solutions obtained using finite element modelling depends on
many factors, with mesh grading, the element aspect ratio and the integration scheme
being some of the most important. The effects of the element aspect ratio and the
number of Guassian integration points, both in plan and through the section, for
punching problems are discussed in this chapter. The analyses were carried out using
3 x 3 x 3 quadrature.
To study the effects of aspect ratio on punching type problems a plate of uniform
thickness was modelled under a centrally applied load. Meshes of different aspect ratio
are considered. The slab, shown in Figure 3.5, consisted of a concrete plate of 1.7 m
square in plan and with a 100 mm square rigid plate embedded at a depth of 100 mm
from top surface of the slab. The depth of concrete was taken to be large compared to
the depth of the plate from the top and, thus, the base was modelled as restrained in the
z-direction. The rigid plate action was modelled using constant displacement for the
nodes representing the plate and it was taken that the plate is not bonded to the
concrete. The problem was investigated using 20 node isoparametric brick elements
with the aspect ratios shown in Figure 3.6. One quarter of the plate was modelled due
to symmetry. The meshes (shown in Figure 3.6) were developed to investigate the
effect of increasing the mesh size away from the critical zone. Details of the meshes
y
j----850--t-f-----850-----j
... .. .. .... .. p
. .. .. ...
q~adre~~t :.us~d.
jfor. Ft' ·~nafysj~
I 1700 ~
....~ ·. :::.....
''..:.. X I
--T- oo- _j
L
1
Faoo----j r-aoo
100
SECTION 1-1
~ 50,•.------------------800----------------~~
<..: (8 @ 100)
.....
Q)
E I
~~~----+---~~---+-----r----+-----~---+----~
Ul
b ~~~--*-~~~?r-*~~~~--*-~~~?r~~~7
Q)
:§ p
Constant
displacement
TYPICAL SECTION FOR FE MODEL (Mesh 1E2)
..< >-
>..
~ X ~
><><
~~'I: ::t:-.
~~ ~ w
/'
~
/'
~
Mesh 1E2 Mesh 1F4
Figure 3.6 - Finite element meshes used to study the effects of aspect ratio.
3-19
Table 3.1 - Properties of the different FE models having 2 and 4 elements through
the thickness.
Maximum Minimum
Model Total Total Total
Nodes Elements DOF x/y Max(x, y)/z x/y Min(x, y)/z
Note: * x/y is the aspect ratio of an element in the plan of the slab.
Max(x ,y)/z and Min(x,y)/z are the maximum and the minimum aspect ratios
through the thickness respectively.
For all the analyses undertaken in this study, the concrete is modelled with an in-situ
uniaxial compressive strength offcp = 25 MPa, a tensile strength offt = 1.65 MPa and
an initial modulus of elasticity of Ec =25300 MPa. The values of the tension softening
parameter a 1 (as defmed by equation 3.30) were calculated for each model and are
given in Table 3.2. The fracture energy was based on the CEB-FIP (1993) model code
32 mm for plain concrete. The other parameters needed to define the concrete material
where J-l is the Poison• s ratio and other parameters are as defmed in Section 3.1 of this
thesis.
The results of the finite element analyses are presented in Table 3.3 and the load
versus deflection curves are shown in Figure 3.7. The responses in the elastic region
are similar for all cases, however. with the propagation of cracks the aspect ratio
adopted in the critical region significantly influences the results. In models 1C2 and
1D2, with larger meshes in the region away from the critical zone, the pattern of
load-deflection curves are similar to that of the uniformly meshed model 1E2. In all
cases the responses of the two layer mesh slabs are seen to be in the same range as
predicted by the refined 4 layered mesh 1F4 showing that some of the other meshes
are somewhat stiffer than mesh 1F4. The ultimate load for the two layered meshes are
close to each other with the difference between the extremes being about 10 percent of
the value predicted by mesh 1F4. It is concluded that away from the critical region the
mesh has little influence on the failure load and the load versus deflection response.
However, there is some effect on the post cracking behaviour with the increase in the
3-21
300
250
-,-
z
~
200
0
0 0.2 0.4 0.6 0.8 1 1.2
Deflection (mm)
Table 3.3 - Ultimate load from the numerical analyses for the models with two or
four elements along the thickness.
number of elements through the thickness. It is to be noted here that the values of a.1
were based on equation 3.30 and a large variation occurs between analyses due to the
varying mesh sizes. Taking a.1 = 20 for mesh 1F4 reduces the ultimate load from
218 k:N to 188 kN and thus the fracture load is sensitive to the value of a.1 adopted.
3-22
important factor in the mesh sizing. The CPU times for each of the models is given in
Table 3.4. The analyses were performed on a DEC Alpha running UNIX with a single
300 MHz processor and 128 megabytes of RAM. Reviewing the analytical results
together with the CPU time, models 1C2, 1D2, and 1E2 are found to be the most
efficient. Of these three models, model1D2 is the cheapest with a minium of sacrifice
in the accuracy of the results. This model is chosen for further investigation.
Models similar to 1D2 were discretised with 1, 3 and 41ayers through the thickness of
the slab. The models are designated as 1D1, 1D3, and 1D4 and are shown in
Figure 3.8. Details of meshes 1D1 to 1D4 are given in Table 3.5 and the values of a1
The ultimate loads obtained for slabs 1D1 to 1D4 are tabulated in Table 3.7. Except
for mesh 1D 1 the failure loads are in the same range with the maximum difference of
less than two percent. A variation of 22 % was obtained between modellD1 and 1D2.
The load versus deflection behaviour is plotted for each case in Figure 3.9 where it is
shown that similar results are obtained for meshes 1D2, 1D3, and 1D4. It is concluded,
therefore, that two elements through the thickness of the slab are sufficient to obtain
accurate results.
To study the influence of the order of integration on the results of finite element
analyses models 1D2 and 1E2 were analysed using 2 x 2 x 2 Gaussian quadrature
Figure 3.8 - Finite element meshes used to study the effects of different division
through the thickness of the slab.
3-24
350
300
250
--,
~ 200
as 150
0 --Mesh 101 - 1 layer
...1
-+--Mesh 1D2 - 2 layers
100 _..,_Mesh 103 - 3 layers
~Mesh 104 - 4 layers
50
0
0 0.2 0.4 0.6 0.8 1
Deflection {mm)
Figure 3.9 - Comparison of the load versus deflection curves for the lD mesh
with one to four layers through the thickness.
Table 3.4 - CPU times for Models 1A2 to 1E2 and 1F4
(3 x 3 x 3 Gauss point integration).
Table 3.5 - Properties of models with different numbers of elements through the
thickness.
Maximum Minimum
Model Total Total Total
nodes elements DOF x/y Max(x, y)/z x/y Min(x, y)/z
Note: * x/y is the aspect ratio of an element in the plan of the slab.
Max(x ,y)/z and Min(x,y)/z are the maximum and the minimum aspect ratios
through the thickness for the , respectively
Table 3.7 - Ultimate load from the numerical analyses with different
numbers of elements through the thickness.
1D1 293
1D2 240
1D3 243
1D4 242
3-26
The load versus deflection curves for mesh 1D2-2 and 1E2-2 are compared in
Figures 3.10 and 3.11 with the corresponding curves obtained using the 3 x 3 x 3
Gaussian integration (that is meshes 1D2-3 and 1E2-3). The figures show that the
order of integration scheme has no significant effect on the numerical results. The
The CPU time for each of the analyses are shown in Table 3.9 and it is seen that
The model studied in Section 3.2.2 of this thesis is similar to that of punching shear in
plain concrete. In order to assess the validity of the model the results are compared
Nielsen et al. proposed a plasticity model for the pullout of circular discs in concrete.
The failure surface is formed by a solid of revolution, with the remainder of the slab
remaining rigid. The failure surface that gives the lowest failure load P, as shown in
(3.3lb)
3-27
300
250
--
z
~
"C
200
150
as
0
..J
100
~Mesh 102- 3 x 3 x 3 Gaussian integration
50 -+-Mesh 102- 2 x 2 x 2 Gaussian integration
0
0 0.2 0.4 0.6 0.8
Deflection (mm)
Figure 3.10 - Load versus deflection curves for modei1D2 with different
integration schemes.
300
250
--
~
"C
200
150
as
0
..J
100
~Mesh 1E2 - 3 x 3 x 3 Gaussian integration
- Q - Mesh 1E2 - 2 x 2 x 2 Gaussian integration
50
0
0 0.2 0.4 0.6 0.8 1
Deflection (mm)
Figure 3.11- Load versus deflection curves for slab 1E2 with different integration
schemes.
3-28
do
Table 3.8 - Ultimate loads for slabs 1D and 1E with different sampling points.
Table 3.9 - Solution time for slabs 1D and 1E for different integration
quadrature.
wherec = ~a 2 -b2 , do is the diameter of the rigid plate; ¢ is the angle of internal
friction for the concrete; and ho is the depth of conical part in the failure surface. The
a= do +hotan¢ (3.32a)
2
P HV
2
fcp [ ho\do+hotarup
( co~ +lc(h-ho )+l
)1-sincp ~d1
(d12Vl2) 2 J-m ((dl2)1-a2J~~
1-c--ab (3.33)
concrete; d1 is maximum diameter of the failure surface; and l and m are the material
l=1-(k-1).1L (3.34a)
fcp
m = 1- (k + 1) ft (3.34b)
fcp
where
k = 1+sin¢ (3.34c)
1-sin cp
3-30
The ratio J;jcp is notionally based on the ratio of the tensile strength to the
correlate the dimension d1 with experimental observations. Nielsen et al. (1978) found
that a value 400 :s;; C1(o.t :s;; 1000 gives the best fit with the experimental data. In
equation 3.33 the efficiency factor, v, is introduced to account for the imperfect
assumption that concrete is a perfectly plastic material and to calibrate the model with
v p;-.
= 4.22 < 10 (3.35)
In verifying the model with experimental data, Nielsen et al. showed a reasonable
mean strength correlation but with a coefficient of variation as high as 21 percent. The
coefficient of variation, however, was generally smaller when individual series were
considered.
average shear stress through the depth of the section taking (} = 26° to the soffit of the
p
-r =----;------ (3.36)
n(do+2h)h
If a lower bound to the failure surface is taken at(} =45° (as per ACI 318, 1999 and
p
't' = ---,-----.,.- (3.37)
tr(do+h)h
With an angle of internal friction of t/J =30°, taking do =}n de where de is the depth
of a square column (that is taking an equivalent circular area) and with -r =0.33..[1;,
h = 100 mm and de = 100 mm, by equations 3.36 and 3.37 the bounds of the ultimate
load are
110 kN ~ Pu ~ 162 kN
With the Nielsen et al. refilied model given by equations 3.31 to 3.35, and with the
bounds of {t
Jcp
= 400 and 1000, the failure load for the pullout of a 100 mm square
The fmite element results (given in Table 3. 7) are of the order of 25 percent higher
than the model proposed by Nielsen et al. (1978) and suggest that taking a 45° failure
3.3.1 Introduction
Dragosavic and van den Beukel (1974) and Park and Gamble (2000) established that
the punching shear strength of flat slabs is not sensitive to the reinforcement content.
In this section the influence of the reinforcement content on the punching shear
3-32
predicted by the finite element modelling is assessed. The effect of the integration
scheme on the results of the finite element models is also considered. For this purpose
the behaviour of a concentrically loaded reinforced concrete slab, resting on the central
square column, with varying amounts of steel is investigated. The results are compared
with the results obtained from the punching predictions of the Nielsen et al. (1978)
square concentric column (shown in Figure 3.13) is considered. The slab is uniformly
loaded and is reinforced with 21 bars at 100 mm spacing in each direction. The depth
to the centroid of all reinforcements (in both directions) was taken as 40 mm from the
top surface.
The slab, shown in Figure 3.13, was modelled using 363 by 20-node isoparametric
solid elements using the mesh shown in Figure 3.14. One quarter of the slab is
modelled due to symmetry. The reinforcing bars were modelled as embedded line
elements within the concrete elements. The CEB-FIP Model Code (1990) states that
dowel effects are not significant when only flexure reinforcement is used due to the
thin layer of concrete above the steel. The finite element formulation used here does
not allow for any dowel shear. The finite element model, Model 2, consisted of 2064
nodes for the concrete with 44 additional nodes used to model the 22 embedded
reinforcing bars in the concrete and include the two ~ area bars at the lines of
symmetry. The bottom face of the central element was restrained against vertical
1r'--1050---+---1050---,r{
l
I
~ ~ ....
21 Bars in each directions
~~~~~~~~
'-------..,.-t-.,r------40-~ X
1
*
150T
.j'-100
COLUMN REACTION
z
y X
eo =0.002
An elastic plastic stress-strain model was used to model the reinforcing steel with
perfect bond assumed between the reinforcing steel and the concrete. The yield
strength of the bars was taken as 400 MPa and the elastic modulus as 200 GPa.
0.0182, 0.0235, 0.0285 and 0.0318 with the ratio being calculated using the effective
depth of 110 mm. The governing yield line pattern is as shown in Figure 3.15. The
m =0.9 p fsy d 2 ; p =Ast I bd ; and fsy, Ast, b, and d are the yield stress of the
reinforcing bars, the bar area, the width of the panel and the effective depth from the
determining the punching shear capacity of the slab, the dowel effect was not
considered and, hence, the punching shear capacity is not a function of amount of
flexural reinforcement.
3-35
/
I
/
2100
~-----210D---------~
The ultimate loads determined for different percentages of longitudinal steel are
presented in Table 3.10. With the increase in the reinforcement ratio the ultimate load
is also increased but at a decreasing rate with the increasing reinforcement content.
The load deflection curve for each of the analyses (with numerical integration over 27
Gauss points) is presented in Figure 3.16. In the elastic region the load versus
deflection responses of the slab are similar, but significant variation is observed in the
reinforcement content.
At the punching shear limit as predicted by the Nielsen et al. (1978) model, the slab
fails in a non-ductile mode and the volume of reinforcement has only a small influence
3-36
400 -0.45%
--+- 1.01 o/o
350 ___._ 1.40 o/o
300 ------ 1.82 o/o
2.35 o/o
z 250 --8-
-
~
,
ca
200
---+-- 2.85 o/o
--&-- 3.18 o/o
.9 150
100
50
0
0 5 10 15 20 25
Deflection of symmetrical edge node (mm)
on the failure load. The load versus strain in the reinforcement near the column for the
different analyses is shown in Figure 3.17. The figure shows that ductile failures, as
The shear capacity of the slab can be obtained using the model of Nielsen et al. (1978),
as discussed in Section 3.2.3 of this thesis. The flexural capacity is obtained using
yield line theory for the critical yield line pattern as shown in Figure 3.15. The failure
load for the slab versus reinforcement content is plotted in Figure 3.18 and it is shown
For the slab with reinforcement ratios of 0.0101, 0.0182 and 0.0318 further analyses
deflection results and load versus maximum steel strain for these analyses are shown
in Figures 3.19 and 3.20. Ultimate loads are presented in Table 3.11. The figures show
that the order of integration has little influence on the results of the finite element
analyses. The CPU time required for the analyses is given Table 3.12 and it is seen
that the integration on the reduced 8-point integration is 30 percent more time efficient
Table 3.11 - Ultimate load for the analyses with different Gauss sampling points.
-11-0.45%
400
-+- 1.01%
350 Yield Strain .........,_ 1.40%
~ 1.82%
300 ~2.35%
--+- 2.85%
-250
z
-
~
, 200
as
0
~ 3.18%
...J 150
100
50
0
0 5 10 15 20 25
Maximum strain in reinforcement over the column (E-03)
Figure 3.17 - Load- strain developed in steel for various reinforcement contents.
400
350
• • •
300
•
z 250 Punching shear capacity
-
,ca 200
~
0
(Nielsen et al., 1978)
...J 150
100
Flexural capacity (Yield line theory)
50 • FE results
0
0 0.5 1 1.5 2 2.5 3 3.5
Reinforcement content in one direction (%)
400
350
300
-250
-
~
"0
C\1
0
200
-s- 1.01 o/o(2 2 2)
---- 1.01 o/o (3 3 3)
...I 150 ~ 1.82 o/o (2 2 2)
400
350
300
z 250
-
~
"0
C\1
0
200 -s- 1.01 o/o (2 2 2)
---- 1.01 o/o (3 3 3)
...I 150
~ 1.82 o/o (2 2 2)
0
0 1 2 3 4 5 6
Maximum strain in reinforcement over the column (E-03)
Figure 3.20 - Load versus steel strain curves found using 3 x 3 x 3 and 2 x 2 x 2
integration rules.
3-40
Table 3.12 - CPU time used for nonlinear analyses with 2 x 2 x 2 and 3 x 3 x 3
Gaussian integration.
3.4.1 Introduction
The behaviour of cracked reinforced concrete structures is complex. While micro and
fracture models are becoming more widely available to study the mechanics of
models have limited applications to large scale concrete structures. Other less
numerically demanding, smeared and discrete crack models are available. The 3D
parameters. The element response is modelled by using factors for tension stiffening,
compression softening and shear retention. In this study, a linear tension softening
model was used, as discussed in Section 3.1.4. The tension softening model depends
on three factors, the tensile strength of the concrete ft, the initial elastic modulus, Ec,
and ultimate cracking strain of the concrete for zero stress, eu . Of these parameters eu
is selected using van Mier's (1987) equation (equation 3.30). In the analyses
3-41
performed in this study a shear retaining factor of f3 = 0.2 was used unless stated
otherwise.
With the development of cracks, the stiffness of the concrete reduces with the rate of
stiffness degradation given by the slope of the tension softening curve. In reinforced
concrete structures, an increase in the slope of the concrete constitutive model over
that of concrete in uniaxial tension is needed to account for the tension stiffening
effects.
(3.39)
where ecr is the cracking strain and is taken as e = %c .The uniaxial tension strength
is taken as ft =0.33..[1; wherefcp is the in-situ concrete strength (ACI-318, 1999).
To study the sensitivity of the finite element modelling to eu the reinforced concrete
plate modelled in Section 3.3.2 with p = 2.85 percent was analysed for different values
of a1. All other parameters were unchanged and the shear retention factor was taken as
f3 = 0.2.
The results of the analyses are shown in Figures 3.21 and 3.22 where the load versus
deflections and load versus maximum steel strain, respectively, are shown. The
parameter a1 is found to have significant effect in the deflection response of the slab
3-42
450
400
350
-300
z ---Aipha1 = 31
~ 250 ~Aipha1 =24
"D
as 200 ~Aipha1 =15
0
-+-Aipha1 =8
...I
150
_._Aipha1 =2
100
-ir-Aipha1 =0
50
0
0 2 4 6 8 10
Deflection (mm)
Figure 3.21 - Load versus deflection curves for model 2 with p = 2.85 % for
different values of a1.
450
400
350
-300
z
~ 250 ---Aipha1 = 31
"D
as ~Aipha1 = 24
0 200
...I
150
~Aipha1 =15
-.-Aipha1 =8
100 _._Aipha1 =2
50 -ir-Aipha1 =o
0
0 0.2 0.4 0.6 0.8 1 1.2
Strain for central steel near the column (E-3)
Figure 3.22 - Load versus strain in steel near the midspan for model 2 with
p = 2.85 % for different values of a1.
3-43
and on the failure load. The development of steel strains are also affected by the
selected values for a1 in the range from cracking of concrete to the point where the
tension stress in the concrete is zero. The load versus strain curve converge to a
With the development of cracks, the shear stiffness of the concrete is reduced but not
. to zero due to the effects of dowel shear and aggregate interlock. The shear retention
factor f3 is used to calibrate the shear capacity for cracked concrete, however, its exact
calibration is a function of the aggregate type, the concrete strength, crack widths and
spacing and the amount of reinforcement crossing the cracks. As such, a realistic
Although the importance of this factor is recognised in the nonlinear finite element
mathematical model Some studies have taken a constant value, while others use a
variable factor based on the tension strain. In studies by van Mier (1987),
Channakeshava and Iyengar (1988), Barzegar (1989) constant values for f3 were used
varying between f3 = 0.1 and f3 = 0.5. Other studies such as those by Al-Manaseer and
Phillips (1987), and Foster (1992) used a variable shear retention factor model In this
section the sensitivity of the finite element results is examined for different values of
The reinforced plate subjected to punching shear shown in Figures 3.13 and 3.14 were
used for the study with a reinforcement content of p =2.85 percent. The results of the
3-44
analyses for load versus deflection and load versus maximum strain in the
reinforcement are plotted in Figures 3.23 and 3.24, respectively. The results of the
analyses show that the trend of the load deflection response is not affected by {3,
however, the ultimate load is influenced by the value of f3 selected. The observations
from the load versus strain response in the reinforcement, shown in Figure 3.24, are
similar to those for the load versus deflection response. Thus the value of f3 adopted,
while not influencing the overall stiffness behaviour of the slab, has significant effect
(Nielsen et al., 1978) have been adopted. In this section sensitivity analyses are
undertaken to determine the influence of¢ on the finite element results. Values of
As for the previous sensitivity studies the reinforced concrete plate shown in
Figures 3.13 and 3.14 with p = 2.85 percent were used for the analyses. The other
reinforcement are presented in Figures 3.25 and 3.26, respectively. The analyses show
3-45
500
450
400
--
~
"D
350
300
250
---Beta = 0.99
~Beta= 0.5
-+-Beta = 0.2
"'
0
..J 200 -Beta=0.1
150 __.__Beta = 0.05
-.tr- Beta = 0
100
50
0
0 2 4 6 8 10 12
Dflection (mm)
Figure 3.23 - Load versus deflection curves for model 2 with p = 2.85 % for
different values of {J.
500
450
400
-,- 350
~ 300
250
-.Beta = 0.99
~Beta =0.5
"'
0 200
..J
-+-Beta = 0.2
150 -+-Beta = 0.1
~Beta= 0.05
100
-A-Beta= 0
50
0
0 0.5 1 1.5 2
Figure 3.24 - Load versus strain in the steel near the midspan for model 2 with
p = 2.85 % for different values of f3.
3-46
400
350
300
--
z.:.= 250
"D 200
m --Phi = 35 Degrees
0
..J 150 -a- Phi = 30 Degrees
........,_Phi = 25 Degrees
100
50
0
0 2 4 6 8
Deflection (mm)
Figure 3.25 - Load versus deflection curves for model 2 with p = 2.85 % for
different values of ;.
400
350
300
z.:.= 250
-
"D
m
0
200
--Phi = 35 Degrees
..J 150
-a- Phi = 30 Degrees
100 ........,_Phi = 25 Degrees
50
o~~~~~~~~~~~~~~~~~~~~~~
Figure 3.26 - Load versus strain in steel near the midspan for model 2 with
p =2.85 % for different values of tfl.
3-47
that for the range of ifJ examined the general response of the slab is unaffected. There is
In the non-associated plasticity model the dilatancy angle lfl is always less than the
friction angle ifJ and reduces to zero for concrete subject to high confming pressure.
Vermeer and de Borst (1984) adopted a value of 0° ~ lfl ~ 20°. A value of 12.6° was
used in this study as recommended by Vermeer and de Borst. For the sensitivity test
the analyses was carried out for values of lfl = 1°, 12.6° and 18°. The other parameters
were fixed at
0
az = 15.3 /3= 0.2 ifJ = 30.
a= 15 ° J1 =0.2 eo=0.002
The reinforced concrete plate given in Figures 3.13 and 3.14 was again used for the
analyses with p = 2.85 percent. The results of the analyses for load versus deflection
and load versus strain in the reinforcement are presented in Figures 3.27 and 3.28,
respectively.
The analytical results show that the trend of load deflection and the load versus strain
response in the reinforcement plot are not affected by lfl, however, the ultimate loads
are influenced to some extent, for lflless than 12.6°. Thus the value of lfl adopted,
while not influencing the pattern of general response of the slab, has some marginal
350
300
250
--,i 200
_._Psi= 18.0 Degrees
ca -a-Psi= 12.6 Degrees
0 150
..J --+-Psi = 7.0 Degrees
100
50
0
0 1 2 3 4 5 6 7
Dflection (mm)
Figure 3.27 - Load versus deflection curves for model 2 with p = 2.85 % for
different values of Vf.
350
300
--,i 250
200
ca 150 _._Psi= 18.0 Degrees
0
..J -e- Psi = 12.6 Degrees
100 --+-Psi = 7.0 Degrees
50
0
0 0.2 0.4 0.6 0.8 1 1.2
Strain in central steel near column (E-3)
Figure 3.28 - Load versus strain in steel near the midspan for model 2 with
p = 2.85 % for different values of Vf.
3-49
The slope of the softening branch of the concrete stress-strain curve, Ec1, is given by
Ecf =~c. In Hognestad (1951) model Ec" =- fcm and for /em= 25 MPa this gives
':1 :1 0.012
The sensitivity test of the compression softening slope Ecf in the results of the
nonlinear fmite element analyses was conducted using Model 2 with a reinforcement
The study was undertaken for ( =-24, -12, and -6. The results of the analyses for load
versus deflection and load versus strain in the reinforcement are presented in
Figures 3.29 and 3.30, respectively. No effect is observed in the results for varying Ecf·
3.5 Conclusions
In this chapter, mesh grading is shown to have significant influence on the results of
the fmite element analyses. The integration scheme adopted is important in terms of
the solution time required for the analysis but has little effect on the overall results.
The finite element model predictions are in reasonable agreement with the closed form
plasticity model of Nielsen et a1. (1978) and in close agreement with the
3-50
350
300
--,
~ 200
250
-a-Zeta= 24
co 150 -Zeta=12
0 -tr-- Zeta = 6
...1
100
50
0
0 2 4 6 8
Deflection (mm)
Figure 3.29 - Load versus deflection curves for model 2 with p = 2.85 % for
different values of '·
350
300
-,-
~ 200
250
co 150 ~Zeta=24
0
...1
-Zeta=12
100 -tr--- Zeta = 6
50
0
0 0.2 0.4 0.6 0.8 1 1.2
Strain in central steel near column (E-3)
Figure 3.30 - Load versus strain in steel near the midspan for model 2 with
p = 2.85 % for different values of'·
3-51
Nielsen et al. 's simplified model with (} =26° . The numerical model also suggests
that the ACI 318 and AS 3600 codes are conservative in taking the failure plane at an
The results of the finite element models studied in Section 3.2.2 of this chapter reflect
the variability in the results of finite element analyses. Considering all the aspects
discussed above, including the results of the plasticity model of Nielsen et al. (1978),
it is concluded that the model 1D2-2 is the best compromise for efficiency and
accuracy for the study of punching type failures. That is, a fme mesh within the critical
region with a maximum plan aspect ratio of two and two elements ·through the
thickness of the slab is sufficiently accurate and efficient for solving punching
reduced integration saves storage and considerably reduces the solution time without
significantly sacrificing accuracy. Away from the punching zone the aspect ratio of the
Two modes of failure are clearly seen from the model, that of flexure and that of shear.
Once the punching limit is reached increasing the quantity of steel leads to only a
small increase in the capacity of the slab. This observation is made, however, in the
The results of the finite element modelling for a reinforced concrete flat slab in
punching shear show that using a reduced 8 point integration scheme gives a good
solution without sacrificing accuracy. Thus, as for the plain concrete slab analysed in
3-52
Section 3.2.2 of this chapter the use of a reduced integration scheme is appropriate in
The results of the fmite element model are consistent with the observation of
Dragosavic and van den Beuk.el (1974) in that the negative moment reinforcement
does not significantly influence the punching shear strength. The finite element results
are also consistent with the theoretical model for strength. It is concluded, therefore,
understanding the finite element results. The shear retention factor f3 and the
descending slope of the tension stress-strain law (as given by a1) were observed to
have a significant influence on the results of the finite element analyses. The other
parameters have only minor effect on the results. The commonly used values may be
adopted for all these parameters in the fmite element modelling of punching shear type
problems. The value of f3 = 0.2 for the shear retention factor is found to be suitable.
The value of a1 must be carefully chosen to represent the tension stiffening effect,
4.1 Introduction
To verify the generality of the finite element model, described in Chapter 3 and
studies have been numerically simulated. The problems selected comprised of a beam
failing in flexure, shear and torsion to more complex systems like slab-column
connections. The specimens selected were based on the level of details provided in the
experimental study reports. The aim was to test the model against a range of
Mansur and Rangan (1978) tested a series of floor beams framing into spandrel beams.
The beams were subjected to combined flexure, shear, and torsion. Beam SB-2 is
selected for the analyses. The specimen consisted of a floor beam spanning 3.0 metres
and attached to the mid-span of the spandrel beam of 1.5 metres span. Both the floor
beam and the spandrel beam had a depth of 300 mm and a width of 180 mm. Details
of the specimen and the loading arrangements are shown in Figure 4.1 and the details
of the steel reinforcement arrangements are shown in Figure 4.2. The properties of
The finite element mesh used for the analysis is shown in Figure 4.3 with one half of
the specimen modelled on one side of the line of symmetry. The model consisted of
4-2
I.P.=Inflexion point
Spartdrel Bearn
(a)
(b)
Figure 4.1 - (a) Idealisation of the spandrel and floor beam system within the
frame; (b) outline of test assembly and symmetrical half portion
used for FE modelling.
4-3
-
90
I• 9 @ 85 c(_c • 1
I ~ D
2-10 mm dia.
~
-~
6.3 mm dia. stirrups
2-16 mm and 1-12 mm dia.
r~~ ~
,...- 750
SPANDREL BEAM SB - 2
Lines of symmetry
0
1 0 3@100 J
I
21 @ 130 c/c J
2-12 mm dia.
V:l
~~
3000
~~
L.
~
J
I 150
[D
g
--1
6.3 mm dia. stirrups
2-20 mm dia.
J. 180
FLOOR BEAM
Figure 4.2- Specimen detail of the T -beam (Mansur and Rangan, 1978).
Table 4.1 - Material properties of reinforcing bars (Mansur and Rangan, 1978).
Plane of symmetry
1
Surface fully restrained
Figure 4.3 - Finite element model of specimen SB-2 (Mansur and Rangan, 1978).
992 nodes and 148 elements to model the concrete with additional 401 nodes for the
reinforcing steel modelled as embedded bar elements. The ends of the spandrel beam
were fully restrained against vertical translation and twist. The material properties
eo=0.002
In Figures 4.4 to 4.7 the results of the finite element analyses are compared with the
experimental results for deflections, twists and torque. The load versus deflection
curves of the analytical results generally compare well with the experimental results
(Figure 4.4) but give a slightly stiffer response than that of the experiment.
4-5
180
160
140
--
z
~
120
100 --Floor beam (Experimental)
D.
,; 80 ---Floor beam
ca
0 ------- Spandrel beam (Experimental)
...1 60 ~Spandrel beam
40
20
0
0 10 20 30 40
Deflection (mm)
Figure 4.4 - Load versus deflection curves of floor and spandrel beams.
180
160
140
z 120
.lll:
-Q. 100 --Experimental
ca 80
"C
-+-FE
0
...1 60
40
20
0
0 5 10 15 20 25 30
Average angle of twist of spandrel beam (E-3 Radian/m)
180
160
140
--
z
~
a. 100
120
,ca 80
0
..J 60 -Experimental
40 -+-FE
20
0
0 5 10 15 20
Torque, T (kN m)
18
16
-E
z~
14
12
-
1-
o)
10
8
..
::s
C"
0
1-
6 -Experimental
-+-FE result
4
2
0
0 5 10 15 20 25 30
Average angle of twist of beam (E-3 Radian/m)
The load versus average angle of twist in the spandrel beam, shown in Figure 4.5, and
the torque developed in the spandrel beam, shown in Figure 4.6, both compare well
with the experimental data. In Figure 4.7 the torque versus the angle of twist at the
centre of spandrel beam are compared for the finite element and the experimental data.
While the cracking torque was slightly higher for the numerical model, the general
Strains in the top and the bottom reinforcing bars in the floor beam were monitored at
the locations shown in Figure 4.8. The results of the load versus strain are shown in
Figure 4.9 and a good correlation is observed between the finite element and
experimental data. Strains near the middle of the bottom longitudinal bars and in the
stirrups near the mid-span of the spandrel beam were measured at the locations shown
in Figure 4.8. The finite element results are compared with the experimental data in
Figures 4.10 and 4.11 for the longitudinal bars and the stirrups, respectively. The finite
element results for the longitudinal bars compare well with the experimental data. The
results for the stirrups show a reasonable overall trend but are significantly different
The relatively high strains in the gauged stirrup at low loads in the spandrel beam
(shown in Figure 4.11) and the stiffer response of the sections than that evidenced in
the load displacement curves, indicate that the beam may have been subject to some
pre-cracking before testing. On the whole, however, it is concluded that the finite
element model gives a good prediction to the overall response of the floor-spandrel
beam system.
4-8
-Gauge points
180
160
140
--
z 120
~
a. 100
'ti'
as 80
0 -Experimental
..J 60 -+-FE
40
20
0
0 5 10 15 20
Strain {E-3)
180
160
140
--
z~ 120
a. 100
'ti'
as 80 -Experimental
0 -+-FE
..J 60
40
20
0
0 2 4 6 8
Strain {E-3)
Figure 4.9- Load versus steel strain curves for floor beam (a) Bottom bar
b) Top bar.
4-10
180
160
140
z 120
.llli::
-100
a.
-c
ca 80 -Experimental
0 -+-FE
..J 60
40
20
0
0 1 2 3 4 5 6 7
Strain (E-3)
Figure 4.10 - Load versus steel strain curves for the bottom longitudinal bars of
the Spandrel beam.
180
160
140
z 120
.llli::
-100
a.
-c
ca 80
0 -Experimental
..J 60
-+-FE
40
20
0
0 2 4 6 8 10 12
Strain (E-3)
Figure 4.11 - Load versus steel strain curves for the stirrup of the spandrel beam.
4-11
Lim and Rangan (1992) tested three slab-column connections for punching shear with
two of the three specimens containing shear studs at the connections to increase the
punching shear strength of the slab. Slab 1 contained no shear connectors and was
Slab 1 consisted of a 4950 by 3170 mm slab with a column of 250 mm square located
at the middle of its free edge. The specimen details and the arrangement of the steel
reinforcement are shown in Figure 4.12. The base of the column was pin supported.
The slab edge adjacent to the column was free while the opposite edge and the side
edges rested on 1-beams. The slab was restrained against lifting at the comers by
placing a beam across all top comers of the slab. However, no direct connection was
made from the supports to the slab and, thus, the slab was free to slip horizontally
against the boundaries. The equilibrating horizontal forces for the horizontal reaction
measured at the base of column could then only have come from friction of the slab
against the 1-beams, or from restraint developed at the comers where the slab was
prevented from lifting, or both. The slab was loaded via 24 equal point loads evenly
The finite element mesh used for the analyses is shown in Figure 4.13 and the
boundary conditions are shown in Figure 4.14. One half of the slab-column connection
was modelled due to symmetry (Figure 4.13). The slab-column connection was
modelled using 504 by 20-node isoparametric brick elements. The reinforcing steel
4-12
~---3170-~.
1B
I
A
i 1 I 10 W8 @ 220 mm
h-1--1-- ·- --- -- foo-l--
9 W8 @1 ?5 mm
I I
I ~ I
I I
I
~ --- I Top bars
---
~
4950 I I
I @ IX) I
I I
xl I IX)
S:
S:
I
I
I ~
I
N
~ iI
I
I - I
1-1--'---- --- '"-- f--'-- 13 W8 @ 220 mm
IJ. I I 3 W8 @ 125 mm
D C 16 W8 @ 220 mm
3170
, .. .. , 110
~___1
W8 @ 140 mm
r
800
w 175 mm
nnn>b"I
Simple support with
frictional restraint only
L j IE
SUPPORT
I--
R6 @ 00 mm
4 Y20
250
COVER=15 mm
SECTION X - X
~ X
(a) PLAN
y
i
I"""' ~ X
(b) ELEVATION
Figure 4.13 - Finite element model of Slab 1 (Rangan and Lim, 1992).
(a) Plan, (b) Elevation and (c) Perspective view.
4-14
EDGE RESTRAINED IN Z
EDGE RESTRAINED IN Z
1 1
L 125
C. --. ---------~-------------------------------
COLUMN
J
---X
~
z
I ·...... . ..
Spring with -~
stiffness, K S1mple support
-------11- X
Note: K=O, 800, 2400, Infinity N/mm
SECTION 1 -1
was modelled using 78 embedded bar elements. To model the boundary friction the
comers of the slab (marked A and Bin Figure 4.14) were restrained in the x, y, and z
directions and the fixity of the column in x was replaced with a spring of variable
stiffness. These boundary conditions being statically equivalent to fixing the column
base in the x-direction and applying horizontal springs at the comers. The stiffness of
the spring at the base of the column was then tuned to simulate different boundary
conditions.
The recorded mean cylinder strength at the time of testing was 25 MPa. In the finite
element modelling, the in-situ strength of the concrete was taken as fcp = 0.9 fcm . The
eo=0.002
The properties of the reinforcing bars used are given in Table 4.2.
The spring at the base of the column (in the x-direction) was modelled using
restrained). The stiffness K =800 N/mm was selected to match the measured horizontal
The calculated failure loads from each of the analyses are presented in Table 4.3. The
load versus mid-panel deflection and load versus vertical column reaction are given in
Figures 4.15 and 4.16, respectively. The vertical reaction versus the horizontal
0 39 Flexure
00 19 Punching Shear
45
40
C\1 35
..e
z 30
e
"C
2s --+-Experimental
"'
.2 20 --o- K = 0.0 Nlmm
-e- Experimental
160
---Q-K =0.0 N/mm
140 --e- K =800 N/mm
~K =2400 Nlmm
120 --tr- K =INFINITY
_100
z
-
.lll::
a:
N
80
60
40
20
0
0 10 20 30 40
Rx (kN)
Figure 4.17 - Vertical reaction {Rz) versus the horizontal reaction (Rx) at the base
of the column.
With the increase of stiffness, the load carrying capacity of the system is found to
decrease. The transfer of load through the connection depends on the combined effect
of the shear and the moment interaction. With the increase of spring stiffness the
The mid-span deflections for the finite element analyses and the experimental slab are
shown in Figure 4.15. A reasonable correlation is seen between the experimental data
and the finite element model for the slab. Figure 4.15 also shows that the slab
deflections are not sensitive to the spring stiffness. The main variance between the
4-19
numerical deflections and those recorded in the laboratory is in the pre-cracked elastic
The vertical reaction at the base of the column is plotted against the total slab load in
Figure 4.16. The figure shows that the vertical reaction in the column and hence, the
The greatest sensitivity to the boundary conditions is seen in the development of base
shear at the column, shown in Figure 4.17. With K =0, that is with no lateral restraint
at the base of the column, no unbalanced moment occurs at the column-slab joint.
With infinite spring stiffness the horizontal reaction and consequently the unbalanced
moment at the connection is significantly greater than that measured in the laboratory.
The best match is with K = 800 N/mm. In the load versus the horizontal reaction of the
In addition to influencing the column base shear and, hence, the unbalanced moment
in the column, the spring stiffness affects the strains in the negative moment
reinforcing steel. Figure 4.18 shows the strain in the short span central top steel at the
edge of the column. The strain is small when the lateral restraint is zero and increases
with increasing spring stiffness. The maximum strain in the short span bottom
reinforcement, near the mid-span, is shown in Figure 4.19. The reinforcement for the
spring stiffnesses of K = 0.0 N/mm and K = 800 N/mm were found to have yielded
before the advent of punching failure. This indicates some load redistribution occurred
in the slab prior to failure. No strain measurements on the steel were recorded in the
experimental study.
4-20
~K=o.o
YIELD STRAIN -EI- 1<::800 Nlmm
~ 1<::2400 Nlmm
....-tr- K:: INFINITY
~ 1<::0.0 Nlmm
-e- K=SOO Nlmm
--+- K=2400 Nlmm YIELD STRAIN
....-tr- K=INFINITY
This study shows the influence of boundary conditions in the laboratory test results.
The boundary conditions have a significant effect on the moment transferred from the
slab to the column in flat plate and flat slab structures. This in tum affects the capacity
of the joint and puts into question how well the experimental model simulates a real
structure. This study also showed that the slab deflections and the transfer of vertical
load to the column are not significantly influenced by the boundary conditions.
In modelling the punching type problems the behaviour of the cracked concrete is
important in obtaining accurate and reliable results for the failure load and failure
mechanism. In this respect, the sensitivity analyses presented in Chapter 3 show that
4-22
the parameters a1 and f3 are of fundamental importance. Model 2 (K = 800 Nlmm) was
chosen for further sensitivity analyses for the concrete tension parameters a1 and the
Sensitivity to a1
With the development of cracks, the stiffness of the concrete reduces. This reduction
and its rate is dependent on Eu (refer Figure 3.1). The ultimate uniformly distributed
load on the slab predicted from the finite element analyses with a constant shear
retention factor of f3 = 0.2 and for different values of the parameter a 1 are given in
Table 4.4 and shown in Figures 4.20 and 4.21. The failure load, the displacement of
the slab and the shear to moment ratio at the junction of the column and the slab are
significantly influenced by the slope of the descending branch of the tension stress-
strain curve for the concrete. The development of strain history in one of the critical
bars (central top bar in the short span) is presented in Figure 4.22. The parameter a1 is
seen to be influential in the strain history of the steel and the failure mode.
Sensitivity to J3
For fully cracked plain concrete the shear stiffness becomes small after the onset of
crossing the cracks on the influence of aggregate interlock is significant. Although the
satisfactory mathematical model. The model with K=800 N/mm is analysed with
40
35
N- 30
.e~
- 25
' C 20
as
----Experimental
_._Aipha1 =25
.2
15 -+-Aipha1 =15
~
1- 10 ---tr-Aipha1 =5
5 ......e-Aipha1 =0
0 --~~~~~~~~;-~~~~~~~~,_~~~~
0 10 20 30 40 50
Deflection (mm)
Figure 4.20 - Load versus deflection curves for the Rangan and Lim (1992) slab
with different values of a1•
160
140
120
-100
z
-
~
N
a:
80
60
----Experimental
_._Aipha1 25=
-+-Aipha1 =15
40 ---tr-Aipha1=5
20 ---+-Aipha1 =0
0
0 5 10 15 20 25 30
M(kNm)
Figure 4.21 - Shear versus moment at the column slab connection for the Rangan
and Lim (1992) slab with different values of a1o
4-24
40
-e- Alpha1 = 25
__._Aipha1 = 15
=5
"""'1!1-- Alpha1
-e- Alpha1 =0
Figure 4.22 - Load versus strain for the central top bar for the short span for the
Rangan and Lim (1992) slabl fordifferent values of a1o
2-A 0 21.7
2-B 5 23.5
2 15 32.4
2-C 25 34.4
5 Experimental 22.0
4-25
The effect of the shear retention factor on the slab deflections, the distribution of
vertical to horizontal loads at the base of the column, the development of strain in the
top reinforcement at the centre of the column and on the failure loads are given in
Figures 4.23 to 4.25 and Table 4.5, respectively. The analyses show that the shear
retention factor does not influence the results of the load versus displacement; the
distribution of horizontal to vertical reaction and, hence, the shear to moment ratio at
the slab-column intersection; or the development of strain with increasing applied load
in the negative moment reinforcement in the vicinity of the connection. However, the
value of f3 chosen has a significant influence on the failure load. The failure load
2 0.20 32.4
5 Experimental 22.0
4-26
50
45
.e-
40
C\1
35
-,~ 30
ca 25
.S! 20
-+-Experimental
=0.9
---e...- Beta
--b:- Beta =0.5
s 0 15 -+-Beta =0.2
1-
10 --e- Beta = 0.01
5
0
0 20 40 60 80
Deflection (mm)
Figure 4.23 - Load versus deflection curves for the Rangan and Lim (1992)
slab-column connection with different values of {3.
200
180
160
140
-120
z
e1oo -+-Experimental
N
a: 80 - - 0 - Beta = 0.9
--b:- Beta =
0.5
60 -+-Beta 0.2 =
40 --e- Beta 0.01 =
20
0
0 10 20 30 40 50 60
Rx(kNm)
Figure 4.24 - Vertical versus horizontal reactions at the column for the Rangan
and Lim (1992) slab-column connection with different values of {3.
4-27
50
45 Yield limit
-40
N
.e
35
-
~30
1125
0
~Beta=0.9
5
0
0 2 4 6 8 10
Strain {E-3)
Figure 4.25 - Load versus strain for the central top bar for short span for the
Rangan and Lim (1992) slab1 for different values of f3.
Foutch et al. (1990) tested four post-tensioned slab-edge column connections. The
specimens are identical in the external dimensions and had similar boundary
conditions. The outlines of the specimens with the loading points are shown in
specimens is shown in Figure 4.27. The locations of the tendons for specimens S 1 and
S2 are shown in Figure 4.28a. The location of the tendons for specimens S3 and S4 are
Note:
D: Distance of point
of application P
A: Central edge point
VALUE OF D (mm)
Specimen D
PLAN
S1 1070
points
S2 610
for whiffle tree
S3 610
/////// S4 305
/......_
Steel Plate 300X3 OOX25
300X300 Column
A_L
2030 T1 00
D
p
Load applied th rough
whiffle tree
'V"
LABOR ATORY FLOOR /\.
ELEVATION
Figure 4.26 - Outline with key points for the Foutch et al. (1990) specimens.
4-29
1520
1 ...
2 13
L
1 128 '1
• 47 77
L77
128 ~~ !1-
~I-~------1~5~2~0------~-~~~
'
13 SECTION 2-2
52 100
1245
II I I
100
Bors used for slob: #3 (9.5mm) deformed bos.
0
0
N
0
0 Stirrups #3 Deformed Bors
N
2030
0
0
N
1
0
0
N
0
0
N
II)
Ol
8 #6 Deformed Bars
D
SECTION 3-3
LABOR ATORY FLOOR
SECTION 1-1
1520 1520
=i-1
.3 45
.3 @75
1 r-h 1 1 15 1 h _J_ 1
]'j
i-1-J 11 5 f-J
.3 @75
.3 45 260
_j_
150 405 405 405 10.30
75 75 115
2_J 2_j
z 1520
1066
_l_
!--:---:=====~======== :;:zss~s1 ~ X
SECTION 1-1
z
1219 1520
102
l
T
-rn I
~
51
25.5
25.5 .. y
183* 180* 245 .245 180* 183*
610
-
~~ ~
610
SECTION 2-2
The properties of the concrete, reinforcement and tendons used in the test are
presented in Tables 4.6 to 4.8. The tests were carried out using whiffle tree to apply
four vertical loads arranged symmetrically at a distance of 1.07, 0.61, 0.61 and
0.305 metres from the front face of the column for specimens S 1, S2, S3 and S4
Table 4.6 - Concrete properties used for the FE model (Foutch et al., 1990).
a(Degrees) 15 15 15 15
</>(Degrees) 30 30 30 30
Description Values
The specimens were modelled using the mesh shown in Figure 4.29 with one half of
the specimen modelled due to symmetry. The finite element model had 276 20-node
solid isoparametric concrete elements with 1755 nodes. The bonded reinforcement and
the prestressing tendons were modelled using embedded elements defined by 120
nodes for specimens Sl and S2 and 114 nodes for specimens S3 and S4. The
reinforcement was placed in the same locations for all specimens. Perfect bond was
assumed between the bonded non-prestressed reinforcement and the concrete. The
pinned supports were applied at the top and at the bottom of the column, matching that
4-33
-- - ~- --x
1 ..
1524 .. 1
Plane of symmetry
Column
PLAN
z z
~~ ~ IT-
- -
r- r-
2 030 -x 20 30 y
- -
- -
(b) 3D view
Figure 4.29 - Finite element model of the slab-column (a) Plan and elevations
showing boundary conditions; and (b) 3D view.
4-34
of the experimental test set up. The Hognestad (1951) model was used to model the
concrete compressive stress strain diagram with the concrete strength taken as
fcp = 0.90 !em. where !em is the mean cylinder strength of the concrete at the time of test.
Figures 4.30 to 4.33 show the load versus displacement of the slab edge (shown as
point A in Figure 4.26) for slabs S 1 to S4, respectively. In Figures 4.34 to 4.37, the
load versus strain in the steel reinforcement is compared for different bonded
reinforcement at the locations shown in Figure 4.27. In all cases the numerical model
is in good agreement with the measured response of the slab-column connection. The
60
50
-40
z~
-
"C
as
0
30
...1
20 - Experimantal
--FE
10
0
0 10 20 30 40 50 60
Edge deflection at A (mm)
100
-Experimental
90
--FE
80
--
z
~
70
60
50
"C
as
0 40
...1
30
20
10
0
-1 9 19 29 39 49 59
Edge deflection (mm)
80
70
60
z 50
-Jill::
"C
ca
0
40
..J 30 -Experimental
-4-FE
20
10
0 --~~~~~~~~~~~~~~~~~~~~~~
0 10 20 30 40 50 60
Edge deflection (mm)
160
140
120
--
zJill:: 100
"C 80
ca
0
..J 60
40 -Experimental
-4-FE
20
0
0 10 20 30 40
Edge deflection (mm)
60
50
--z
Jill::
40
30
"'0
as
0
...1
20 ~Experimental: Gage 6,7
-+--Experimental: Gage 8
-11- FE: Gage 6
10 -+-FE: Gage 7
-.-FE: Gage 8
0
0 5 10 15 20
Strain (E-3)
Figure 4.34 - Load versus reinforcement strain history for Specimen Sl.
100
90
80
-c
z
70
60
50
"'0
as
0 ~Experiment: Gage 4
...1 40
-+--Experiment: Gage 5
30 --11- Experiment: Gage 6
-11- FE: Gage 4
-+--FE: Gage 5
-.-FE: Gage 6
-1 4 9 14
Strain (E-3)
80
70
60
-,-
z
~
50
40 -a- Experiment: Gage 6
as __..._Experiment: Gage 7
0
...1 30 --t1- Experiment: Gage 8
-----FE: Gage 6
20
~FE:Gage7
10 ......._FE: Gage 8
0
0 5 10 15 20 25 30
Strain (E-3)
Figure 4.36 - Load versus reinforcement strain history for Specimen S3.
160
140
120
-z 100
-
,as
~
0
80
-a- Experiment: Gage 5
__..._Experiment: Gage 6
--t1- Experiment: Gage 8
...1 60
---FE: Gage 5
40 ~FE:Gage6
......._FE:Gage8
20
0
0 5 10 15 20
Reinforcement strain (E-3)
Figure 4.37 - Load versus reinforcement strain history for Specimen S4.
4-39
Bums and Hemakom (1977) conducted tests on a 1/3 scale specimen of prestressed
Figure 4.38). Bonded reinforcing bars were used only in the negative moment regions
(see Figure 4.39) with all positive moment reinforcement being unbonded prestressed
tendons. The arrangements of the tendons were as shown in Figure 4.40. The slab was
uniformly loaded and unloaded a number of times, with 16 different loading patterns,
before being tested to destruction. In this study only the uniform load wad modelled.
For the purpose of the analysis the overhang portion of the slab was neglected and one
quarter of slab was modelled with symmetry. The cable profile is also taken as
symmetrical. The finite element mesh used is shown in Figure 4.41. The slab was
modelled using 1282 twenty-node solid isoparametric elements with 7520 nodes. The
bonded reinforcement was modelled as two-node embedded bars and the straight and
curved tendons were modelled as embedded two-node and three-node unbonded bars,
respectively. An additional 216 nodes were used to define the bonded bars and 310
eo= 0.002
4-40
. :L
. G~ . .. .. c2 Portion of the specimen
u u
. ... .. . . .
4
4
4
• 4
for FE modeling
.
,4
.
.. .. .·• . . . . Span/h = 44
4.
0
co 4 •
co
0 0
v
. c4'l!J .. .c3. Prototype - 9 .144 m span
4
I
4
: 4
,[
PLAN
I. 10,000 .I
---17621----3048
v
·I· 2048
·I· 3048----11--102
r--
i
1"1{ K· K 1 ~25 mm
bearing
ball
v
r'1
ELEVATION
Figure 4.38 - Plan and elevation of slab-column connection with the portion used
for f"mite element analyses for the Bums and Hemakom (1977)
specimen.
4-41
0
0
0
o~m
each way
/,
*/•
~Line
5 Bars each
of
Column used in ';nalyses
D
8-9.5 mm dia.
bars
Ties: 6 mm dia.
SECTION X-X
(assumed)
0
+
762
NOTE:
Length of each bar if not mentioned = 1.524 m
Reinforcement layout symmetrical about diagonal
@ 254 mm c/c
@ 108 mm c/c
u u L.:J
"'
tO
a if ~
1-
NOTE:
~
~ Areo of tendons 23.22 mm2
a= ~ 0 = 188 mm
v
0 ;! 0 ~ c
a= f=
0 X 0
,.... X
0
t J
0 ~
0 a= ~
v
0
0 0 0 [
;!
f=
1404 ~~~ ~
II
ll
a
ll127oll
14D4J 1
a
111 ?oil
1L 1401411
a a
IL75
a
@
@
254 mm c/c
108 mm c/c
v
0 0 0 0 [
;!
10,000
Bonded bars
SECTION X-X
"
c
£Q)
E
E
4E 0~
0
Q)
c
0
0::
u ---x
Plane of symmetry
4670
:!===F ~~~~~~4::e:=E670~~i~
FRONT AND SIDE ELEVATIONS
Figure 4.41- Finite element model of Burns and Hemakom (1977) slab-column
(Plan, elevations and 3D view with boundary from below).
4-44
with the in situ concrete strength was taken as 0.9fcm· The reported cylinder strength
was 34.0 MPa. The properties of the non-prestressed reinforcement and tendons used
The locations of main points to measure the deflection (point C) and strain in steel
(gauge 1 and 2) are shown in Figure 4.42. Figure 4.43 compares the load versus
mid-panel deflection response at C (see Figure 4.42) from the finite element model
with the experimental data. A good correlation is observed. The failure load obtained
in the finite element model was 9.8 kPa and compares well with the experimental
result of 9.9 kPa. Calculated from the strains measured in the steel of the columns, the
reaction in column C4 at failure was reported to be 150 kN. In the finite element
model the load in column C4 at failure was 159 kN, within 6 percent of the
experimental result.
Table 4.10 -Material properties of reinforcing steel (Burns and Hemakom 1977).
Bars
Material Property #2 #3
Area As (mm2) 32 71
0
co
<.0
..q-
L---f----------------------------~~----x
01 .. 4680 ----~
Figure 4.42 - Location of strain points for finite element analysis of Burns and
Hemakom (1977) slab.
12
-
"'e
.....
z
10
-
~
"C
ca
.2
8
6
"C
.S! 4 --Experimental
c.
c. --FEM
<C
2
0
0 50 100 150
Deflection at C (mm)
Figure 4.43 - Load deflection curves for the Burns and Hemakom (1977) slab.
4-46
Property Value
No strain histories were gauged in the test. The maximum strain developed at key
reinforcement locations in the finite element analyses are shown in Figure 4.42. The
load versus strain plots in the bonded reinforcement for the finite element model at
these locations are shown in Figure 4.44. The plots show that these bars had yielded
before failure. In the test the slab was reported to have failed by punching at the
interior column connection C4. The finite element results predict that the slab failed in
4.6 Conclusions
In this chapter finite element models were developed for beams and slab-column
connections reported to have failed in shear-torsion or punching shear mode from four
incorporated in developing these models. The overall observation is that finite element
modelling is able to compute the failure loads and failure modes of the experiments.
Measured data such as load versus deflections for example, were shown to correlate
16
14
-.e
C'll
z 10
12
~Strain location 1
-~
"D
ca
.2
8
.........._Strain location 2
6
s0
1- 4 Yield limit
2
-1 4 9 19
Steel strain (E-3)
Figure 4.44 - Load versus strain curves for bonded bars in the negative moment
region of the slab.
From this study it is concluded that the finite element models used in DIANA (1997)
give reasonable results within a tolerable range. However, the studies must be
the stress-strain curve for the concrete in tension was shown to dramatically effect the
flexural response of the slabs and the shear retention factor was shown to highly
influence the failure load, particularly for slabs failing in punching shear. In
conclusion the model can be used for studying the behaviour of slabs failing in
5.1 Introduction
As discussed in Chapter 3, the shear retention factor f3 and the tension stiffening
parameter a1 are two important parameters required for calibration of the finite
element modelling of the concrete structures. These parameters are investigated below
A number of experimental and analytical studies have been undertaken to study the
Stone and Carino (1983) and Hellier et al. (1987), pullout tests were originally
(Malhotra, 1975, Stone and Carino, 1983, Stone and Carino, 1984, Yener and Li,
1991, Yener 1994) have been used to calibrate the results of pullout tests with the
strength of the concrete. Klinger and Mendonca (1982), Ichihashi et al. (1985) and
DeVries et al. (1999) are just some of the many experimental studies undertaken on
Figure 5.1 shows a schematic diagram of a typical pullout test of an embedded plate.
Stone and Carino (1983) carried out two similar pullout tests using large-scale
specimens having apex angles (2a, refer Figure 5.1) of 54 and 70 degrees. Figure 5.2
shows the test set-up used in these studies. Based on discontinuities in the load-strain
histories observed in the embedded gauges, Stone and Carino concluded that pullout
~--------d~------~
d2 =Disk size p
d, =Int. dimension of
reaction ring
h =Embedment depth
2cx =Apex angle
t
ring
\4
\
Idealized failur~\ <l
4
surface \ <l
<l
<l
4
4
<l
<l
d
<l
4
4
<l 4
<l
II) Completion of circumferential cracking from the disk edge to the reaction ring at
approximately 2/3 of p u•
III) Shear failure of the concrete matrix and degradation of aggregate interlock
beginning at 0.8Pu·
Stone and Carino showed that the increase in apex angle changes the failure surface
laboratory tests, which showed a good agreement with the observed data up to the
5-3
\.l,.__j__jj_~~~>portir>g
bars @ 150 c/ c
AROUND CIRCUMFERENCE OF HOOP
1370 Diameter
2030 Square
initiation of cracking. Using an axisymmetric finite element model Hellier et al. (1987)
studied the behaviour of pullout specimens similar to that shown in Figure 5.1 for an
apex angle of 70 degrees. They concluded that the ultimate load carrying mechanism
Yener (1994), Yener (1994) state that in the post micro-cracking stage stresses are
continuously redistributed with crack propagation and the formation of new cracks
until failure. The residual strength in the pullout test is a consequence of the crushing
In this study a large apex angle is used in order that the failure mechanism is not
influenced by the boundary conditions. Under these conditions failure of the anchor by
pullout is similar to that of pure punching shear where no moment is introduced into
the connections.
A study of the literature of punching failure of slab shows that the failure cone forms
at a slope of less than 45 degrees with the horizontal (Regan, 1981, Hammill and
Ghali, 1994, Walker and Regan, 1987, Alexander and Simmonds, 1986). In this study
the apex angles were kept large to ensure a minimal influence on the punching cone
from the remote boundary. Two test series were undertaken in this study with concrete
compressive strengths of nominally 20 MPa and 40 MPa. Each test series consisted of
three specimens of different sizes and with different anchorage embedment depths (h)
(see Figure 5.3). The apex angle was between 130 and 144 degrees. The specimens are
5-5
t
1 ..
D2
I
..I
~ 1DOX100X20
1
steel plate
.v
at the centre
I
1 1
1t
D2 --
J
---x
11:<~
L so--il ..
~
01
I
I
.. If--so
Rig
Reactive farce p
!~ ~
<3
h <3~
----r
(h+250)
Steel
;1 ~_)
Plate
250
;1 ___i
.-I•t---- D2 -------i·~l
SECTION 1-1
designated as PXX-YYY where XX is the nominal concrete strength and YYY is the
embedment depth of the plate. For example P20-1 00 was cast with a nominal concrete
The pullout load was applied via a 20 mm thick by 100 mm square plate, embedded at
rod, as shown in Figure 5.4. Three embedment depths were tested, 50, 100 and
150 mm. For the specimens with 100 mm and 150 mm embedment depths a net of
wire mesh of 4 mm diameter was placed at a depth of 30 mm from the top surface
(see Figure 5.4). The function of the wire mesh was to avoid any accidental damage to
the specimen during curing. No damage was observed to the surface of any specimen
at the location of Macalloy rod entering into the specimen. The size of wire mesh in
plan was small as not to have any influence on the failure. A typical detail of the wire
mesh is shown in Plate 5.1. In all specimens, the pulling rod rested on the base of the
formwork. The rod was enclosed in a greased 40 mm diameter PVC pipe below the
plate to ensure no bond occurred between the concrete and the Macalloy bar below the
anchor plate. The formwork for all specimens was made of 18 mm thick plywood.
Once all parts were in place for the required position the assembly was braced. The
wire mesh was held in position using plastic bar chairs or 6 mm diameter steel wires
projected from the bracing. The assembly is shown in Figure 5.4 and Plate 5.2.
The concrete used in the study was provided by a local ready-mix concrete supplier.
The specifications for the concrete were a maximum aggregate size of 10 mm; a slump
..........._
~ '100
L~
f---100~
OOX100X20 mm
29 mm Macalloy bar
concentric to the plate
SECTION 1-1
t <1
L
(h+250) <!
Wire<1 mesh
~
'-----;;----t-1 OOX1 OOX20 mm steel plate
<!
---~..._-.:r---+-PVC pipe coated with
lubricant inside and
..
I~ - - - D 2 - - - . . -..l l outside both
Note: D3 = 200 mm and 280 mm for specimens PX:X-100 and PX:X-150, respectively.
No mesh was provided for specimens PXX-50.
The actual strengths at the time of testing are given in Table 5.1 together with the
concrete tension strength obtained from split cylinder tests (tested in accordance with
AS 1012.10-1985) and the measured elastic modulus of the concrete. The concrete
compressive strength was measured using 150 mm diameter by 300 mm high cylinders
The concrete was poured carefully so not to disturb the assembly and an electric
needle vibrator was used to consolidate the concrete. A steel trowel was used for
finishing the top surface of each specimen. For each pour ten large cylinders (150 mm
diameter x 300 mm height) and fifteen small cylinders (100 mm diameter x 200 mm
height) were cast to measure the properties of the concrete. One day after the concrete
was placed, the specimens were covered by hessian and plastic with the hessian
5-10
watered twice daily for a minimum of 7 days. The cylinders were striped from their
The reaction rig was fabricated of rectangular tubular section having external size of
100 mm by 50 mm and was placed on top of the concrete blocks concentric with the
central pullout rod. The reaction rig was connected to a 1000 kN capacity hydraulic
jack as shown in Figure 5.5. The load was measured using a 500 kN capacity load cell.
In test series P20 displacements were measured at four points along a central line. In
the second test series (P40) displacements were measured at three symmetrical points
on top of the concrete blocks. These points and their positions are presented in
Figure 5.5. Linear variable differential transducers (LVDT's) were used to measure
displacement with the transducers connected to a data acquisition control unit. The
reading accuracy of the transducers was 0.01 mm and their maximum travel ranges
were 100 mm. A typical arrangement for the test set-up is shown in Plate 5.3. The
output of all electronic measurements was channelled through a HBM signal amplifier
-Reaction rig-
1 1
L _j
2_j
"" - - 1 girders - - - - - - - M
!!
"
!!
r
(h+250)
L_
~
Concrete b)ock
'---Lf--- Reaction rig ----+-'---;;---'
C;ncr~te
17
..
block
•
(h+250)
_j_
SECTION 1-1 SECTION 2-2
Note:
a=85mm for h=100mm For clarity stiffeners
a=120mm for h=100mm are not shown.
a=150mm for h=150mm 01 is as shown in Figure 4.3.
b=100mm for all h = embedment depth
Plate 5.3 - Test set up showing L VDTs and steel girders resting on reaction rig.
5-13
The jack was slowly pressurised to apply the load at the rate of approximately
0.3 kN/minute. Thus depending on the specimen the testing times varied from Y2 hours
to 2 hours. The failure took place suddenly in all tests as expected and the dimensions
of the failure cones were measured and are given in Figures 5.6 and 5.7. Some of the
typical failure cones are presented in Plate 5.4. In all cases the failure occurred well
away from the boundaries and, thus, the boundary did not influence the results.
The peak load and the corresponding displacement at point A (see Figure 5.5) for all
tests are given in Table 5.2. The angle of the failure cone was significantly less than
45 degrees in all specimens. The plan dimensions of the top of failure surface are
presented in Figures 5.6 and 5.7 for series P20 and P40, respectively. The load
deflection curves at different points for series P20 are plotted in Figures 5.8 to 5.10
The experimental results are compared with the Nielsen et al. (1978) and
ACI 318-1999 models as shown in Table 5.3. The ACI model (ACI 349-1990) used to
characteristic cylinder compressive strength in MPa and Ac is the area defined by the
projected area of a 45 degrees cone radiating from the edge of the anchor plate. The
in-situ compressive strength was taken as 0.9 fcm . For the Nielsen et al. model the
lower bound was taken for ft = fcp/1000 and the upper bound ft = fcp/400 (see
Section 3.2.3). The angle of friction was taken as l/J = 30° and the efficiency factor
was taken as 0.91 for the P20 specimens and 0.69 for the P40 specimens, as obtained
from equation 3.35. Table 5.3 shows that the ACI model gives a safe prediction of the
Failure line
Figure 5.6- Plan dimensions of failure cone for test series P20-YYY.
5-15
c
L
o Central
Central plate
plate
Figure 5.7- Plan dimensions of failure cone for test series P40-YYY.
5-16
80
70
-
~50
:; 40
60 LVDTB LVDTA
~ 30
20
10
0
0 0.2 0.4 0.6 0.8 1
Displacement (mm)
140 LVDTA
120
-,-
z
~
100
80
ca 60
0
..J
40
20
0
0 0.1 0.2 0.3 0.4 0.5 0.6
Displacement (mm)
250 ~LVDTC
LVDT LVDTA
LVDTB
200
--
~ 150
i0 100
...1
50
0
0 0.1 0.2 0.3 0.4 0.5
Displacement (mm)
Figure 5.10- Load versus displacement curves for series P 20-150 specimens.
120
LVDTC
100 LVDTB
--
z
~
'0
80
60
tiS
0
...1 40
20
0
0 0.1 0.2 0.3 0.4 0.5
Displacement (mm)
Figure 5.11 - Load versus displacement curves for series P 40-50 specimens.
5-20
180
160
140 LVDTB
-120
z
,
c1oo
ca 80
LVDTA
0
..J 60
40
20
0
0 0.1 0.2 0.3 0.4 0.5 0.6
Displaceemnt {mm)
350
300
-~ 250
-
,ca
0
200
150
..J
100
50
0
0 0.1 0.2 0.3 0.4
Displacement {mm)
Table 5.2 - Failure load, corresponding displacements and the average slope of
the cone.
P20-50 73 0.337 10
Table 5.3- Comparison of peak load with ACI 318-1999 and Nielsen et al. (1978)
models.
The primary aim of the pullout tests was to calibrate the shear retention factor f3 and
the tension stiffening factor a1 for the problems failing in punching shear. All
specimens were modelled using 20 node solid isoparametric elements with one quarter
of the specimen modelled with symmetry. The effect of the rigid plate action was
modelled using constant displacement for the nodes representing the plate and it was
assumed that the plate was not bonded to the concrete. The planes of symmetry were
modelled using the appropriate boundary conditions and the top nodes of the elements,
where the reaction rig was placed, were restrained against the movements in the
vertical direction (see Figure 5.14). DIANA (1997) was used to carry out the finite
element analyses.
The finite element meshes used to model the specimens are shown in Figure 5.15. The
meshes are designated FE-50, FE-100 and FE-150 and correspond to a plate
embedment depth of 50, 100 and 150 mm, respectively. The number of nodes and the
elements contained in the meshes are given in Table 5.4. The in-situ strength of the
pullout concrete block was taken as fep = 0.9 fern and the tensile strength was taken as
ft = 0.33~ fern . Material properties used for the concrete elements are
The tension stiffening parameters a1 for the different models of P20 and P40 are given
in Table 5.5.
Table 5.4 - Nodes and element numbers for finite element meshes.
Specimen Parameter a1
P20-50 7.0
P20-100 8.0
P20-150 9.3
P40-50 12.3
P40-100 12.3
P40-150 10.6
5-24
y y
t
1· 12po~
~~~=zz:zz;JJ-IPortion used for I
I / FE modelling I
I
I
;
12 00 f- - --f<eelll~l--·
- - - -- X I
L50
'--1 OOX 1OOX20 steel
~~~g~~~U plate at the centre
I
~'-- ..._ ___ ..___ X
PLAN
Note:
Plane with nodes restrained
X-direction
Plane with nodes restrained
1
350
in Y -direction
- - - - - Plane with nodes restrained
in Z -direction
L y X
600---1
ELEVATION
y X y X
Model FE-150
Figure 5.15 - Three dimensional views of meshes used for different models.
5-26
The peak loads of the fmite element analyses are compared with the experimental
results in Table 5.6. It is seen that the finite element predictions are higher than the
experimental results in all cases but are mostly within 10 % of the experimental
results. The load versus deflection curves are plotted in Figures 5.16 to 5.21 for
predicted by the finite element analyses are seen in these figures to be within a
Figure 5.22. It is observed that the values of the parameters {3 = 0.2 and az as given
Table 5.5 are generally suitable for the modelling of these problems.
5.6 Conclusions
For large apex angle the pullout in the plain concrete takes place with the average
angle of the failure cone much less than 45 degrees to the horizontal. This agrees with
the findings of Stone and Carino (1983). The ACI model with the effective stress area
taken at 45 degrees from the outside of the anchorage plate gave conservative
predictions of the failure load while the Nielsen et al. (1978) model was
non-conservative.
The finite element model gave reasonable prediction given the limitation of the model
in analysing fracture type problems. The finite element analyses verify that the values
of {3 = 0.2 and a 1 as given by van Mier (1987) are suitable for modelling of pullout
90
80
70
-60
~50
"C
40
_.! 30
-Experimental
20 -FE
10
0
0 0.5 1 1.5
Displacement (mm)
Figure 5.16 - Load versus deflection curves at point A for specimen P20-50.
160
140
120
z~ 100
:;- 80
~ 60 -Experimental
40 -FE
20
o~~~~~~~~~~~~~~~~~~~~
Figure 5.17- Load versus deflection curves at point A for specimen P20-100.
5-28
250
200
--
~ 150
'C -Experimental
.9 100 -FE
50
o~~~~~~~~~~~~~~~~~;-~~~
Figure 5.18 -Load versus deflection curves at point A for specimen P20-150.
140
120
--
z
~
'C
100
80
ca 60 -Experimental
0
..J
-FE
40
20
0
0 0.1 0.2 0.3 0.4 0.5 0.6
Displacement (mm)
Figure 5.19 - Load versus deflection curves at point A for specimen P40-50.
5-29
250
200
--
~
"C
150
ca
.9 100 -Experimental
-FE
50
o~~~~~~~~~~~~~~~~~~~~~
Figure 5.20- Load versus deflection curves at point A for specimen P40-100.
400
350
300
z~ 250
:;;- 200
ca
.9 150
100 -Experimental
50 -FE
0
0 0.2 0.4 0.6 0.8
Displacement (mm)
Figure 5.21 - Load versus deflection curves at point A for specimen P40-150.
5-30
1.937E-3
l.S41E-3
.744E-3
.647E-3
.55E-3
'.453E-3
L.x .356E-3
l.259E-3
1.162E-3
1.649E-4
I
(a) P20-50
1.74E-3
t.G63E-3
'.597E-3
.511E-3
.434E-3
f .35BE-3
1.291E-3
$,285E-3
1.12BE-3
1.516E-4
I
(b) P20-150
1.612E-3
1.549E-3
.4B6E-3
.423E-3
.359E-3
.2S6E-3
~.233E-3
%.17E-3
1.187E-3
1.43BE-4
I
(c) P40-100
i.!BSE-2
1.S47E-3
.834E-3
.721E-3
.SBBE-3
'.4S6E-3
~ .383E-3
l.27E-3
i.157E-3
1.442E-4
I
(d) P40-150
Table 5.6 - Comparison of peak loads between experimental and f"mite element
results.
Specimen FE!Expt
Experimental Finite element
6.1 Introduction
The test specimens were designed as a one-third-scale model of the prestressed flat
plate building shown in Figure 6.1. The portions of the slab at the edge and at the
Four specimens were constructed and tested to failure with the primary variable being
the level of prestress. Specimens S 1, S2 and S3 were designed as edge panels while
Specimen S4 was design as a comer panel. Specimen S 1 was a control specimen and
had no prestressing. Specimens S2, S3 and S4 had unbonded prestressing in both the
The design of the slab of the prototype and the models was based on the analyses of
the frames using the equivalent frame method and using the simplified method of AS-
3600 (1994). The worst case of the results from these analyses was considered to
design the slab. Using the simplified method of AS-3600 it is assumed that the column
strip and the middle strip take 70 and 30 percent of the load, respectively. The
specimens were developed as a 113 scale model of the prototype. The design of the
prototype and the model were carried out for a live load of 3.0 kPa. All the designs
were done with the concrete having the characteristic strength of 25 MPa and with
yield strengths of the conventional reinforcement and prestressing steel of 450 MPa
and 1250 MPa, respectively. A 10 mm clear cover was used in the model slabs. To
give bias towards the punching shear failure the flexural reinforcement was increased
by 50 percent in all slabs over that required for the design moments.
6-2
0 0 0 0
-77~
Region undbr
stud~!
0 0 0 0
0 0 0 0
1
J
L,2ooo •I• 12000 I I I
12000 I I8 12ooo----J
o---- p;~u~n-i:ind~rl
stuay l
-----------"
SECTION 1-1
The design of the prototype and the model slabs are given in Appendix A. For slabs
S 1, S2 and S3 the slabs were designed for an average prestress of 0.0, 1.0 and 2.0 MPa
respectively. For the corner panel slab the average prestress was 2.1 MPa.
The columns of all specimens were overdesigned so as not to fail before the slab. A
200 mm x 200 mm section was used for the test slabs and contained eight 16 mm
6-3
diameter 400 grade hot rolled deformed bars. A 6 mm diameter diamond arrangement
was used for the column tie reinforcement and the ties were spaced at 80 mm centres.
Details of the reinforcement arrangements for the test specimens are presented in
Section 6.3.
The boundary of the slab was taken at the second point of the contraflexure away from
the design connection as recommended by Rangan and Hall ( 1983) and Rankin and
Long (1988). The normalised bending moment and shear force distributions for the
model and the specimen slabs with edge and comer column (based on the frame
analyses) are shown in Figures 6.2 and 6.3, respectively. The size of the column was
determined to simulate the stiffness of the prototype column framing into the
Figure 6.4. The slabs were 3565 x 4900 mm and 100 mm thick. The 553 mm long
200 mm square column was cast monolithically with the slab at the middle of the long
edge. The three sides of the slab away from the column support were supported on the
!-sections of the testing frame with a bearing width of 100 mm. To carry the
connection, the slab along the long edge opposite the connection was attached to the
supporting frame via shear connectors (refer Section 6.5). The column base rested on a
load cell mounted within a steel box connected to the load frame.
The specimen for the comer column connections (Specimen S4) consisted of a
3565 mm x 2550 mm by 100 mm thick slab, as shown in Figure 6.5. The 553 mm
6-4
3565 435
NORMALISED S.F.
Key:
--Specimen
----------Model
NORMAliSED B.M
End with column
I in the specimen
NORMALISED S.F.
Figure 6.2 - Normalised bending moment and shear force diagrams comparing
the prototype slab design with the specimen design for edge
column specimens Sl, S2 and S3.
6-5
'
,//' Note:
Specimen
~ 0.7
-~ Model
End with column
in the specimen
00
NORMALISED B.M.
3565 435
~--------~0010-----------~
NORMALISED S.F.
2550 450
Key:
--Specimen
-------- Model
NORMALISED B.M. End with column
in the specimen
2550 450
'-------.JOOD-------'
NORMALISED S.F.
Figure 6.3 - Normalised bending moment and shear force diagrams comparing
the prototype slab design with the specimen design for the corner
column specimen 84.
6-6
~--------------3565--------------~____t
200X200 mm
COLUMN
SECTION 1-1
Hinge~ sup:~ L 100
200
ylll
A 35 65-----•-'1 B
-r----- '=-=~=--=~=--=~=--=~=--=~=--=~=--=~-=-=~-=-=~1=,
,,
,,II
II,,
200X200 mm
-COLJJK4N_________ XI
4900
,,_j
:I,,
II
100 Ji'
~ 100
------------=-=:-.=..=:-.=..=:-J
D
L-,, c
49eo
" "
. . "
~
6
Simpl e support Simple su
~ 200mmX200mm
1 .. 2350 -- ~
Column
'--200
2350 -I
SECTION 11-11
Figure 6.4 - Dimension of slab and column of specimens Sl, S2 and S3.
6-7
Ill
y z
~----3565----~1
-, }---------2550--------~
----l!r----1 00
:I
!i
0
~
• •...
Hinged support
.
H1nged support
~ x::t
100
I
l ~200 X 200
!!JI
?.!"" COLUMN
mm :I
I'
I X
200 -----t -ll
2350 - - - - - - - - - - - 1..
0 SECTION 11-11
II _-:j
z
~-------------3565------------~~ I100
E
6
O • H;ogod '"PPOrt
1--200 SECTION 1-1
• Hinged support Xl
long by 200 mm square column was cast monolithically with the slab at the comer of
the free edges. The two sides of the slab opposite the column connection were
connected to the testing frame via shear studs (refer Section 6.5). The bearing widths
Details of the bonded reinforcement are given in Figure 6.6 and Table 6.1 for
specimens S1, S2 and S3 and in Figure 6.7 for specimen S4. The corresponding bar
bending schedules for ~ese specimens are shown in Figures 6.8 and 6.9. In all slabs
the top reinforcement normal to the hinge supported edges were cut off to 80 mm from
y
3565
I
A B
~~~to~
4900 r--X
IL
0
I
_j
A
D
I I
c
D c
8
y
3565
I
A 8
I ~
H
r--"'- •·f--------1----
I I
G
I I
I I
I I
I
I
~ I
I
I I{) I
200x200 I N I
COLUMN I I
@ 1--X
4900
0~ :
I
IL --r- ~~
I >- ------r--- _j
I
I
I
I I
~
I I
I E
I I
I I
I I
I I
I I
D c
F
For slob:
CLEAR COVER = 10 mm (Top and bottom)
CLEAR COVER = 25 mm (Sides)
Figure 6.6 - Arrangement of bonded reinforcement for specimens Sl, S2 and S3.
6-9
z
. 3515
I I 80 mm Studs
@ 2 00 mm c/c
1 _l
II
r
L653
II
jll
___L:::,_
I X
~
L A
R6 @ 80 mm (2 Sets)
8 Y16
For sl ob:
CLEAR COVER = 10 mm (Top and bottom)
CLEAR COVER 25 mm (Sides)
For column:
CLEAR COVER 20 mm (Side)
SECTION 1-1
200
r
l
l--2oo__j
SECTION 11-11
(c) Sections
J. 3565
I
! 9 Y10 @ 275 mm
- I
~- ~-- ~----- .... -~-l 7 Y10 @ 160 mm
I I 10 Y10 @ 110 mm
2 550 : l : 9 Y10 @ 275 mm
IL
I
I
I
:
I
_jl
200x200
COLUMN
r'J
- ~
- - ----:----
~ c
T
X
0 10 Y10 @ 80 mm
6 Y10 @ 285 mm
--Top bars 200 Y16
7 Y10 @ 125 mm
Bottom bars
PLAN
6 Y10 @ 280 mm
1I--
'-----------'
___j
200
SECTION 11-11
G
X @ 200 mm c/c
!QQf
II L653 Y10
L 2 Sets)
CLEAR COVER
CLEAR COVER
10 mm (Top and bottom)
25 mm (Sides)
SECTION 1-1
60 60
4840
3. Top bars along X-direction (a) Bar away from column (b) Bar within column region
3455
(a) 60
300
3455
(b)
400
4800
155
Figure 6.8 -Bar bending schedule for specimen Sl, S2 and S3.
6-12
soc== 3455
100
soc== 2440
100
2440/3455
Table 6.1 -Bonded reinforcement detail for specimens Sl, S2 and S3.
A 26Y10@60 26W8@60
c 6Y10@200 6Y10@200
D 7Y10@340 10WB@240
E 13Y10@125 13Y10@125
G 12Y10@90 13Y10@90
H 10Y10@240 11W8@220
6-14
The prestressing ducts and the tendons were arranged as per the design given in
Appendix A with small variations to avoid conflicts with the bonded reinforcement.
Figures 6.10, 6.11 and 6.13 and are summarised in Table 6.2. The details of the cable
profiles for tendons in specimens S2 and S3 are shown in Figure 6.12. The
Appendix A.
The formwork for the slabs consisted of 18 mm thick timber plywood to form the base
of the slab and 100 mm x 50 mm timber for the edges. The plywood was rested on the
adjustable rectangular hollow section steel beams that were propped at regular
intervals (refer Figure 6.14). All specimens were constructed in place on the testing
frame, shown in Plate 6.1, which was fabricated as a stiff I-beam frame tied to the
First the conventional bonded reinforcement was cut and bent to the required shapes
and sizes. The column reinforcement cage was prepared and placed in position with
the load cell assembly. The bottom layer of bonded reinforcement was laid in position
followed by the top layer of the bonded reinforcement. All the crossing points of
bottom x- and y- direction bars were tied. Next the ducts and the prestressing wire
were placed and tied in position. The test assembly and reinforcement prior to the
concrete pour are shown in Plates 6.2 to 6.5 for slabs S 1 to S4, respectively.
6-15
85516851
420
640
200x200
column
640
5
362~ /
/
v- Central cable
3mm off
--x
388
375
640
640
420
~1-~----3565-------~1
Effective Prestresses:
y
! 1150
2365 - - - j
7@320 mm I
120~ ~
rE
ro
0
E
0
CX)
0
1'- N
!-~¢ 0
""'""
h
I-'
X
~§ 4900
ro E
0
1'- 0
..- N
l~
.. I
~1--- 3565 --.....-~
Effective Prestresses:
1 0
y
Lgso--~1~.,__-15oo--~,__-15oo--~~-gso_J
L~~---------------------4900---------------------_J~
z
CABLE PROFILE ALONG LONG SPAN
·~
J ------
-vi
18 r
'.J
1.
----- 1(0
-
~ X
CABLE
2000
PRO~LE
3565
ALONG SHORT SPAN
1565
:I
Note: Not to the scale
y =~
l~-oo~------3565------- _[205
J
0 L, 8@125
I = _____j 23BOe------'•l
7@340
t_,, '
_j L 120
TENDON LAYOUT PLAN
356~
I
r-18
_j X __j
0 -;QJ
200n 156
~ SECTION 1-1
t 0
y
150n 105o--
~ SECTION II - II
Effective Prestresses:
Form work
(18mm thick plywood)
I Lc:~,~ti1 ~on--be_a_m--~~------~------~~~------~~------~n
2400 connected to frame
Laboratory floor
All the specimens were cast with commercial ready mix concrete with an electric
needle vibrator used to compact the concrete. The surface of the slab was levelled
using wooden screeds and the surface was finished using steel trowels. Concrete
cylinders and prisms were cast with the slabs as control specimens.
The control specimens were stripped the day followed the concrete pour, wrapped in
hessian and kept with the slab specimens. One day after the concrete was placed the
surface of the slab was covered with the wet hessian and plastic and, with the
exception of specimen S4, the hessian was watered twice daily for a minimum of 7
days. Slab S 1 was stripped after 30 days and slab S2 and S3 after 7 days. Slab S4 was
The boundary supports were chosen to simulate that of the prototype structure. For
hinged edges the movement of the slab in horizontal direction was restrained by using
the shear connectors welded to the top flange of the !-beams of the testing frame, as
shown in Figure 6.15. The shear connectors were 78 mm long, 13 mm diameter with
centres and an 8 mm diameter reinforcing bar was placed behind the connectors
towards the edge of the slab. In specimens S 1, S2, and S3 the connectors were welded
along the long edge opposite to the connection. In specimen S4 the shear connectors
were welded along both of the edge supports. The shear connectors provided an
equilibrating horizontal reactive force to the horizontal reaction developed at the base
of the column.
The base of the column was connected to a 20 mm thick steel plate with the
longitudinal column reinforcement welded to the plate. The column base plate was
fitted to a 115 mm long, 70 mm diameter steel boss which was connected to the load
cell assembly. In specimens S1, S2 and S3 the edge column specimens were
symmetrical about the x-axis and no horizontal support was provided to the short
edges of the slab. Thus, no horizontal y-direction reaction (refer Figure 6.4) could be
maintained by the column support. Pins were provided along y-direction to maintain
the column in position while the load cells were assembled on both sides of the boss to
measure the horizontal x-direction reactions. The base of the boss sat on an assembly
with the load cell that measured the vertical reaction in the column. The load cells and
pins were co_nnected to a box support via adjusting screws which were tightened to
f ~Rx
--l
200mm X 200mm
Column
l----200
Rz
DETAIL A
t
Studs @ 200 Welded
to 1-Section
1-Beam of
test frame
The column support assembly for specimen S 1 was designed as a hinged support
capable of measuring the vertical and the horizontal (x-direction) reactions. The steel
section attached with the load cell measuring the vertical reaction was fitted to a
circular groove at the base of the box, as shown in Figure 6.16. Post analysis of the
specimen S 1 results, however, showed that while the vertical reaction measurement
testing frame via the base connection of the vertical load cell and, thus, was lost to the
horizontal load cell measuring system. The problem was that since the steel section
holding the load cell to measure the vertical section was fitted in the groove, the lower
6-25
..
l=!~;;;f"'o---oY16
Long. bars
t--it---oW6 (2Sets)
base plate
X
115mm long 70mm dia. boss
fitted to 105x 105x25mm plate
portion of the assembly could not move freely in horizontal direction and a part of the
horizontal reaction developed was transmitted to the base. The column support
assembly was modified for specimens S2 and S3. For specimen S2 and S3 the vertical
load cell was connected to a roller system, as shown in Figure 6.17 and Plate 6.6.
The column support assembly used for specimen S4 is shown in Figure 6.18 and
Plates 6.7 and 6.8. The assembly was modified to that used for the edge column
connection tests to account for the lack of reaction symmetry in the corner panel test.
Load cells were added to the assembly to measure the support reactions in both the x
and y directions (refer Figure 6.5). The roller support at the base of the vertical load
cell (used in specimens S2 and S3) was replaced with 35 by 25.4 mm diameter high
strength steel ball bearings. The ball bearings were positioned uniformly around the
base of the assembly and located using an 8 mm thick timber plate with the ball
All the load cells placed to measure the horizontal and vertical column reactions of all
specimens except a load cell used to measure the vertical reaction in specimen S 1 were
of 200 kN capacity. The load cell used to measure the vertical reaction in specimen S 1
6.6 Prestressing
Specimens S2, S3 and S4 were prestressed to different degrees in both the x and the y
directions. In all cases the slabs were post-tensioned the day before the test to
minimise the loss of prestress due to relaxation. Twenty millimetre thick steel bearing
plates of 80 mm x 90 mm were placed at each end of the tendons with jacking from
..
Figure 6.17 - Support system at the column base for the specimen S2 and S3.
Plate 6.6 - Load cell and pin arrangement for specimens S2 and S3.
6-28
..
5 0
~ ~
reactions
8:Steel sections
9.Steel pins
225
-r-i-6:····•
• ••••••••
...._- 25 4mm diameter
boll bear,ngs
•••••
245 • • • 27mm hales
X
••••••
••••• L
8mm thick l1mber
plate
1451-
Figure 6.18- Support system at the column base for the specimen S4.
6-29
The required load for prestressing was determined with allowances for the loss due to
draw-in and relaxation. The extents of these losses were determined by carrying out
relaxation tests on the tendons in a similar environment to that of the slabs. Details of
the relaxation test and the calculation of losses are given in Appendix B. A 200 kN
capacity hydraulic jack was used to prestress in the tendons. The jack was placed
between the prestressing yoke and an anchorage plate, as shown in Figure 6.19, and a
load cell placed between the plates and connected to a HBM amplifier. The oil was
pumped to the jack until the required load was reached. During the process the wedges
of the prestressing grip were tapped gently, regularly and evenly into the barrel. Once
the desired prestressing load was applied the wedges were tapped into position.
The tendons were prestressed individually starting with the wire farthest from the
column. The jacking and nonjacking ends were alternated along each side of the slab
to distribute the prestress uniformly through the slab. The prestressing sequence is
shown in Figure 6.20 with the numbering in Figure 6.20 shown at the jacking end.
The use of a whiffle-tree to apply a distributed loading has been successfully used in
similar tests (Bums and Hemakom,1977, Lim and Rangan, 1992). Specimens S1, S2
and S3 were tested using twenty four concentrated loads applied via three identical
hydraulic jacks placed in parallel through the three sets of whiffle-trees. Each of the
jacks had a capacity of 270 kN and a maximum stroke of 150 mm. Three outlets from
6-31
Anchorage Plate
(80x90mm)
\_Barrel and
Dead end Jacking end wedge
the hydraulic pump regulated the three jacks with identical pressure in each of the
lines. The jacks were calibrated for pressure versus load in a Shimadzu Universal
Testing Machine with the results shown in Figure 6.21. The maximum variation
between the jacks is less than 3.0 percent. The arrangement of the loading points on
the edge panel slabs S 1, S2 and S3 are shown in Figure 6.22. The whiffle-tree
The cables from the whiffle-tree were attached to the loading points of the specimen
slab through the 20 mm holes using 20 mm diameter PVC pipes cast into the slab. The
load was applied to the top surface of the slab via 20 mm thick by 100 mm square
bearing plates. The arrangement of the cable system to apply load on the slab via the
loading tree is shown in Figure 6.24. A load cell was placed in one branch of the
whiffle-tree (see Figure 6.23) at the middle tier to measure of load applied to the slab.
y y
0 G)
®®:5® @ ® @)
0 CD
6
@
@ 10
Q
Column
@
'-.../
X ~2
16
14
18
20
@ - - - -- -- --- x(z2)_
~
0 -
-c0
® 5
1
@ 3
G) 9
® 5
~@® @@
H+H++H~~~~~-+-++~
~+H+H+H~-+-+-+~~~
H+H++H~~~~~~~+~
(c) Specimen S4
Note: The number inside the circle indicates the sequence of prestressing and are
250
200
-z
-
~
"C
ca
0
150
100
--Jack#173
-----Jack#174
..J ········ Jack#178
50
0
0 10 20 30 40 50
Pressure (M Pa)
---="'--
J+ck~
11)
+ + """
200X200 mm column
+ + + +
+ + + +
0 4700
+ + + +
1 00 mm wide bearing on
steel 1-section frame all + + + +
round the edges except 0
the one with the column
+ + + + 0
11)
NOTE: ----'-'r
Refer other drawings for details
of studs in 1-section and other
details.
565~ 800 ~ 800 1 800 1600
3565
Figure 6.22 - Arrangement of loading point in the slab (Specimens Sl, S2 and 83).
6-34
~--Cables attached
to slab loading
points
sections
Rod attached to
hydraulic jack
(a) Schematic drawing of a set of the whiftle-tree for specimens Sl, S2 and S3.
..
200x200mm
column
930x 130x25mm
ms plates welded
to RHS ot top
and bottom
RHS 203x 102x6.3
(b) Front elevation of whiftle-tree system for specimen Sl, S2 and S3.
1OOmm squa
20mm thick plate
<1
4
<1
<1
Figure 6.24 - Section through the slab showing connection of slab to the
whiftle-tree.
6-36
6.7.2 Specimens S4
Specimens S4 was tested using the sixteen concentrated loads applied via two identical
hydraulic jacks placed in parallel through the two sets of whiffle-trees. The
arrangement of loading points for the slab S4 is as shown in Figure 6.25. As for the
edge panel specimens, holes were cast in the slab to fit the loading cables. The loading
system for specimen S4 was similar to that used for the other specimens and is shown
in Figure 6.26. A load cell was located in the middle of one of the whiffle-trees
measuring IA of the total load. The arrangement of the whiffle-tree to apply the load in
Specimen S 1 contained only conventional reinforcing bars while all other specimens
contained both bonded deformed bars and unbonded prestressing wires. The bonded
diameter. The longitudinal bars of the column consisted of 400 grade hot rolled
deformed bars of 16 mm nominal diameter while the ties in the column were
fabricated from 6 mm diameter hard drawn wire. Tensile tests on the bars were
undertaken using an Instron Universal Testing Machine with the average results of
three specimens of each sized steel bars given in Figure 6.27 and Table 6.3.
6-37
3565
800
Rod attached to
hydraulic jack
~-········1..~-·:t>
l.ft• ~-
; ........ * ........ .
800
700
'iS 600
~ 500
;; 400 -------- Y16
-e
U)
3oo
en 200
100
0 ~~~~~~~~~~~~~~~~~~~~~~
carbon steel spring wire, conforming to AS 1472-1991 and marked as S8 was used for
the prestressing tendons. The stress-strain curve for the steel was obtained using an
Instron Universal Testing Machine and the average of the three results are given in
1400
-ca
1200
c. 1000
-.
==0 800
0
Cl)
600
u; 400
200
0
0 0.005 0.01 0.015 0.02 0.025
Strain
Corrugated plastic conduit, normally used for the electric wiring, was used as ducts to
house the prestressing wire in the specimens. The outside diameter of the duct was
16 mm and the clear inside diameter was 13 mm. A section of the duct is shown in
Figure 6.11.
Two types of barrels and wedges (yv2185M and W999L manufactured by CCL
System Ltd., United Kingdom) were used to grip the 8 mm diameter prestressing wire
and for applying the load to the slab via 12.7 mm diameter tendons connected to the
whiffle-tree systems and pulled down from the reaction floor. All the reinforcement
including the ducts for the prestressing were kept in position by using PVC and steel
bar chairs.
6.8.4 Concrete
Ready mix concrete delivered by a local supplier was used for all test specimens. The
specifications for the concrete were a maximum aggregate size of 10 mm, a slump of
80 mm and a target 28-days concrete strength of 25 MPa for specimen S 1 and 32 MPa
for specimens S2, S3 and S4. The target mean cylinder strength at time of testing was
25 MPa. The actual strengths at the time of testing are given in Table 6.5.
Seventeen cylinders of 150 mm diameter and 300 mm high and twenty cylinders of
size 100 mm diameter and 150 mm height were cast along with each specimen. The
cylinders were used to monitor the concrete strength gain with time for prestressing
and testing. The prestressing was done the day before the test when the concrete
strength was approximately 25 MPa. Specimen S4, however, showed a rapid strength
gain and the slab was prestressed at a concrete strength of approximately 37 MPa with
On the day of testing of the slab a minimum of six cylinders were capped with a
sulphur-ash capping compound and tested for the compressive strength. The tests were
corresponding to 20 MPa per minute. Two or three 150 mm x 300 mm cylinders were
1.5 MPa per minute. For specimens S 1, S3 and S4 three cylinders were tested in
uniaxial compression to obtain the stress-strain properties of the concrete. The load
was applied sat a slower rate of strain than for the standard tests, with the testing time
being about 7 minutes and the specimens gauged over a 150 mm interval. The stress-
strain plots of the concrete are presented in Figure 6.29. Three flexural prism tests
were conducted for the specimen S4 with the average modulus of rupture being 4.5
MPa. A summary of the material properties for the concrete is presented in Table 6.5.
6-43
40 4
35
-30
ca
a. 25
:E
,
';" 20
! 15
ti)
10
5
0
0 0.0005 0.001 0.0015 0.002 0.0025 0.003
Strain
Figure 6.29 - Stress versus strain relations for concretes in uniaxial compression.
6.9 Instrumentation
6.9.1 General
The applied load, reactions, deflections and steel strains were measured in all tests. In
this section details of the instrumentation used in the test are given. The output from
the electronic measuring devices was filtered through a signal conditioning amplifier
Section 6.7 and used to measure the applied loads. The horizontal and vertical
reactions at the base of the column, as shown in Figure 6.30, were measured using 200
6.9.3 Displacements
at five points for slabs S 1, S2 and S3 and three points for slab S4. The locations of the
LVDT's are given in Figure 6.31. The displacement transducers were linked to the
main data acquisition control unit and recorded using Labview (1999). The reading
accuracy of all transducers was 0.01 mm and their maximum travel range was
100mm.
For specimen S 1 mechanical dial gauges were installed at the top comers of the slab to
measure uplift and to the column base assembly unit to measure any horizontal
movement of the testing frame. Negligible movement (less than one division) was
recorded in any of the gauges. The dial gauges used were sensitive to 0.00254 mm
(0.000 1").
6-45
,___,
II
3565
~
---------- ----- --":..---:..-,
_L.-1•e------3565------'1
I
R~flum:'ll;:::.==R=x=z====,O=Oq~
I•
I
653] fRz
SECTION 1-1
J
200X200mm
It
~OLUMN
48 0 Rx1 r--' RJQ 47 00
00
~2550---1[
1
!i
I•
H
I•
:I
ii
·· colu~n T
J -so
I
653I Rz
: ---------- --=---=-..J
I SECTION 11-11
11--
Ill 1-4---
X
column
Rx1 Rx 2 100
200X200 mm 653 SECTION 1-1
COLUMN
xz X z
2550---l yr100
PLAN column
Ryz Ry,
653 Rz SECTION 11-11
(b) Specimen S4
Figure 6.30 - Reactions measured at the column support (a) specimens Sl, S2 and
S3; and (b) specimen S4.
6-46
Coordinates(mm):
B:A:
!1758,0,0)
1758, 1200,0)
c. 1758,-1200,0)
D: 50, 1200,0)
E: 50,-1200,0)
~~OL~M~-----tA-------- X
z
(o.o.o3~rr: ======::J....___,..
_ x
Coordinates(mm):
A: ~1700,1300,0)
,._c B: 1700,50,0)
C: 50, 1300,0)
~
~LC~OL~U_MN__~~B______-L---X
z
( O,O,O~~f-;======:::::J._-- .. X
(b) Specimen S4
Strains were measured using electronic strain gauges at key locations on the bonded
reinforcement. The locations of the strain gauges for the slab are shown in Figure 6.32.
Approximately 20 mm length of the top of the bars was grinded to make a plane
surface for fixing the gauge. Care was taken to grind the projected rib of the bar
without causing any appreciable loss of cross sectional area. The gauges were glued to
the surfaces of the reinforcing bars and were covered with a silicon compound. The
gauges were connected to a HBM signal conditioning amplifier which was connected
to the data acquisition system. In some cases the strain gauges were damaged during
concrete pour. Table 6.6 lists gauges that were damaged (X) and operational (./) at
A small load (less than 0.5 kPa) was applied to each slab to set the loading system and·
test the electronic measuring equipment. The load was then released, the gauges
zeroed, and the test started. The load was applied slowly by applying pressure in the
jack in the increments of approximately 0.7 kPa on the slab. The test was paused for
approximately two to five minutes between each load step. Measurements were
recorded electronically via the data acquisition unit. In specimen S 1 the top surface of
the slab was checked between load steps for surface cracks and marked as appropriate.
No surface cracking was measured for the prestress slabs due to safety considerations.
The test was continued beyond the punching shear failure for specimens Sl. However,
for specimens S2, S3 and S4 the tests were stopped as soon as the punching shear
failures occurred. The time taken to test each specimen was approximately 2 to 3
hours.
6-48
y y
Coordinates(mm):
1: 221, o. 75)
2: 221, 125, 75)
3: 221, -125, 75) 8
4: 115, 121, 85)
5: 115, -121, 85)
2 COLUMN 6
o~5eE==========~-x
(0,0,0,)3
0
(0,0,0,) 7
X
Coordinates(mm):
6: !1637, 30, 25)
7: 1637,. -30, 25) 9
8: 225, 1563, 15)
9: 225, -1563, 15)
(o,o,o3~ • X
(a) Specimen Sl
y y
Coordinates(mm}:
1: ~221, -10, 75~
2. 179, -10, 75
3 r21. 125. 75) 9
4 221, -125, 75)
COLUMN 5 115, 121. 85)
6 115, -121, 85)
2-<:: 3 COLUMN 7
06 X 0 X
(0,0,0,) 4 (0,0,0,) 8
Coordinates(mm}:
7, r637. 30. 24)
8: 1637, -30, 24) 10
15~
9: 230, -1563, 15
10: 230, -1563,
(o,o,o3~ • X
(b) Specimen S2
y y
Coordinates(mm):
1: (230, 0, 75)
2: (179, 0, 75)
3: !230.0, 125.0, 75.0) 9
4: 230, -125, 75)
COLUMN 5: 120, 120, 85)
6: 115, -120, 85)
2-...<:; 3 COLUMN 7
06 X 0 X
(0,0,0,) 4 (0,0,0,) 8
Coordinates(mm):
7: t1637, 30, 24)
8: 1637, -30, 24) 10
9: 230, -1563,15)
10: 230, -1563, 15)
TOP BARS BOTTOM BARS
z
(o,o,o~r.r:======~•--'X
(c) Specimen S3
Coordjnqtes(m m) · y
y
1: ~221, 30, 75)
h----------, 2: 221, 155, 75)
3: 30, 221, 85)
4: (190, 221, 85)
Coordingtes(mm)· 8
5: ~1637, 30, 25) 7
6: 1637, 110, 25)
7: 1463, 750, 25) COLUMN 6
~~~~5~~~~~x
8: (140, 1258, 15)
corner with TOP BARS BOTTOM BARS
column
(0,0,0~~1":;•=======---· X
(d) Specimen S4
Gauge Specimen
No.
Sl S2 S3 S4
1 X ./ ./ ./
2 X ./ ./ X
3 ./ ./ ./ ./
4 ./ ./ ./ ./
5 X ./ ./ ./
6 ./ ./ ./ ./
7 ./ ./ ./ ./
8 ./ ./ ./ X
9 X X ./ --
CHAPTER 7- TEST RESULTS AND COMPARISON WITH CODE MODELS
7.1 Introduction
The test results of the slab-column connections are given in this chapter. The measured
failure load and the reactions developed at the base of the column are discussed in
Section 7.2 together with plots showing the development of the reactions at the
column support with the increase of load. The load versus displacements are given in
Section 7.3. Individual test results are discussed in Section 7.4 with observed crack
In the figures presented in this chapter, the total load is taken as the sum of the applied
load, the self weight load and the load due to the whiffle-tree arrangement. For slabs
Sl, S2 and S3 the weight of each of the whiffle-trees was 3.68 kN and corresponds to
pressure of 0.63 kPa (3 whiffle trees per slab). In slab S4 the weight of each whiffle-
tree was 7.35 kN which is equivalent to a pressure of 0.81 kPa (2 wiffle trees for the
slab). The self weight component of the slab deflection, reactions and steel strain was
determined from the measured density of the concrete and the initial linear-elastic
The edge column specimens rested on I-beams on three sides and the slabs were
capable of carrying loads higher than the punching shear failure load when considered
as a slab supported on three sides. For specimen S I the test was continued beyond the
punching shear failure. For specimens S2, S3 and S4 the tests were stopped once the
7-2
punching failure had occurred. The discussions in this section are limited to
observation up to the punching shear failure. However, for specimen S 1 the plots for
displacements and reactions are presented covering the whole of the test including
As discussed in Section 6.5, for specimen S1 the load cells placed to measure the
horizontal reaction at the column base recorded only a part of the total reaction. To
correct the horizontal reaction a linear finite element analysis (given in Appendix C)
was undertaken. The finite element model showed the ratio of the horizontal to vertical
reaction to be 0.5. In the graphs that follow both the recorded and corrected horizontal
The punching shear failure loads and the corresponding reactions in the vertical and
horizontal directions at the base of the column are presented in Table 7.1. The
horizontal reaction at the time of failure was lower than the maximum values, which
were developed a few load steps before the punching failure. The values of the
maximum horizontal reaction Rx-Max are tabulated in Table 7.2 along with the vertical
reaction Rz-Max and the total load at the corresponding load step. The reaction histories
for specimen S 1 are given in Figure 7.1 with the measured reaction in horizontal
direction designated as the R~ and the corrected reaction as Rx. The load versus
vertical reaction is also shown in Figure 7.1 with the vertical reaction shown as Rz.
In Figures 7.2 and 7.3 the load versus vertical and horizontal reactions up to the
punching shear failure are presented. The corrected horizontal reaction (see Figure 7.3)
and S3 which were of similar configurations and concrete strengths. Figure 7.3 shows
20
18
ca 16
~ 14
-12
i
.2 10
"i 8
0 6
1-
4
2
o~~~~~~~~~~~~~~~~~~~~~~
0 20 40 60 80 100 120
Reaction (kN)
Figure 7.1- Load versus vertical reactions (Rz), corrected horizontal reaction
(RJ and measured horizontal reaction ( R~) for specimen Sl.
30
S3
-
ca
25
-
~ 20
"D
ca 15
.2
"i 10
0
1-
5
0
0 50 100 150
Vertical reaction (kN)
30
S3
25
-ca
-
c.
~
'C
20
ca 15
.2
"i
0 10
.....
'Corrected reaction
5
0
0 10 20 30 40 50 60
Horizontal reaction (kN)
Figure 7.3- Load versus horizontal reactions for specimens 81,82 and 83 up to
the punching failure.
Tables 7.1 and 7.2 show that for slabs S1, S2 and S3 the failure load increases with an
Specimen S4 failed in a punching mode when the total applied load (including self
weight and weight of the loading system) was 353.6 kN corresponding to a pressure of
39.0 kPa. The maximum total reactions measured in x- andy-horizontal directions for
the corner specimen S4 were 37.3 kN and 24.2 kN, respectively. The vertical reaction
at failure was 94.0 kN. However, corresponding to the maximum horizontal reaction
in x-direction of 37.3 kN, the horizontal reaction in y-direction, vertical reaction and
the total load on the slab were 23.9 kN, 86.3 kN and 33.4 kPa respectively. The load
50
45
-
~
as
D.
40
35
30
"D
as 25
.2
a; 20
15 15
.....
10
5
0
0 20 40 60 80 100
Reaction (kN)
Figure 7.4 - Load versus vertical reaction (Rz) and the horizontal reactions in
the x- andy-directions (Rx and Rv) for specimen S4.
7.3 Deflections
For slabs S 1, S2 and S3 deflections were measured at the five locations shown in
Figure 6.31 a. For slab S4 the deflection was measured at three points on the slab,
shown in Figure 6.31 b. The load versus deflection for the complete test of the
specimen S1 is shown in Figure 7.5. For specimen S1 the test was carried on beyond
the punching failure. However, the discussions are limited only to the behaviour up to
The load versus deflection curves up to punching shear failure for specimen S 1 are
shown in Figure 7.6. In Figures 7.7 and 7.8, the load versus deflection are plotted for
specimens S2 and S3, respectively. For slab S4 the load versus deflections are given
7-7
20
-ca
-8
0.
.lll::
"C
15 LVDTA
~ 10
1-
0 20 40 60 80 100
Deflection (mm)
20 LVDTB
-
: . 15
LVDTA
-
.¥
' tJ
~ 10
'ii
....0 5
0+-~~~~,_~~~~~~~~~~~~-L-L~
0 10 20 30 40
Deflection (mm)
-16
20
18 -
:.
.¥
14
-12
i
- 8
10
'ii
0 6 -LVDTE
.... 4 -LVDTD
2
OT-~~~~~~,_~~~~~~~~~~~~~~
0 2 4 6 8 10 12
Deflection (mm)
25
LVDTC LVDTA
-
ca
D.
20
c"C 15
ca
.2
'i 10
0
1-
5
0
0 5 10 15 20 25 30 35
Deflection (mm)
25
-20
ca
D.
,cca 15
0
'i 10
0
1- -LVDTD
5 -LVDTE
0
0 1 2 3 4 5 6
Deflection (mm)
30
25
-ca 20
-
D..
.lll:
' tJ
ca
.2
15
"i 10
0 -LVDTA
1- -LVDTB
5
0
0 5 10 15 20 25
Deflection (mm)
30
25
-ca
-
D..
.lll:
' tJ
ca
.2
20
15
"i
0 10 -LVDTD
1-
-LVDTE
5
0
0 1 2 3 4 5
Deflection (mm)
in Figure 7.9. In Figures 7.5 to 7.9 the deflections are plotted against the total load,
which includes an allowance for the deflections due to the slab self weight and weight
of the whiffle-tree. The self weight was calculated using the measured concrete
densities and deflections were determined from the initial linear response of the slab
In Figure 7.10 the load versus deflections at point A are plotted for slabs S 1, S2 and
7.4.1 Specimen Sl
and corresponds to a pressure of 15.5 kPa (excluding self weight). However, since the
slab was resting on 1-beams on three edges it could carry further load. The specimen
was loaded further until an applied load of 338.4 kN was reached with the pressure of
The development of cracks was monitored during the test. Flexural cracks first
appeared at the top of the slab, in the vicinity of the column when the applied load was
183 kN (30.5 kN in the load cell). Cracking on the slab soffit was more difficult to
to Yz of the failure load. The crack pattern for the top and the bottom surfaces of the
slab at the end of the test are shown in Figure 7 .11. Plate 7.1 shows the crack pattern
in the negative moment region at an applied load of 261 kN (43.5 kN on load cell).
7-12
50
45
LVDTB
_40
l. 35
c3o
"D
8 25
=as 20
'5 15
1- 10
5
o~~~~~~~~~~~~~~~~~~~~~
0 10 20 30 40 50 60
Deflection (mm)
30
-
as
25
-
~ 20
"D
as 15
.2
Sl
'ii 10
'5
1-
5
0
0 10 20 30 40
Deflection (mm)
Figure 7.10 - Load versus deflections at point A for specimens Sl, S2 and S3.
7-13
Distributed flexural
cracks in the negative
moment region
(a) Top surface of the slab (b) Bottom surface of the slab
Plate 7.1- Cracks in the vicinity of the column top for specimen Sl.
7-14
The punching shear failure surface is plotted in Figure 7.12. At the soffit of the slab
the failure started from the junction of the column and the slab. The measurements of
the punching cone as observed at the top and at the free edge of the slab are drawn in
Figure 7.12. The average angle of failure cone with the horizontal plane varied from
13 to 18 degrees.
The strain histories of different main tensile reinforcement at critical points (refer
Figure 6.32a) are given in Figures 7.13 and 7.14 for the working gauges. The strain
plots show some yielding of the central top bars at the punching failure. Figure 7.14
shows that yielding did not occur in the positive moment reinforcement and, thus,
sufficient reinforcement was provided to the slab to prevent a flexural failure before
7.4.2 Specimen S2
Slab S2 failed suddenly (with a loud bang) when the applied load was 324.6 kN
(excluding self weight). The corresponding total pressure applied to slab was 18.6 kPa
(excluding self weight). The slab failed in punching shear near the column connection.
The failure surface started from the junction of column and the soffit of the slab.
Fewer flexural cracks were observed in specimen S2 than for specimen S 1. This is
attributed to the influence of prestress in controlling the crack development. For safety
reasons cracks were not marked during the test. The crack pattern, at failure on the top
and the bottom surfaces of the slab are presented in Figure 7.15 and shown in
Plate 7 .2. The failure surface was measured and is presented in Figure 7 .16. The
punching failure cone formed an average angle of between 13 and 24 degrees with the
horizontal plane.
7-15
line
•
I
I 0
y
(mm) (mm)
X
507
-Y X
(mm) (mm)
0 507
206 507 .300 507
t .350
415
415
427
.300
0
525
525
.300
0
"
0
N
0 0
0 0
I"") I"")
Figure 7.12- Cracks near the column; (a) crack detail near the column at slab
top and; (b) punching cone as seen at the free edge of the slab of
specimen Sl.
7-16
20
18
-16
:. 14 Yield limit
c"C 12
8 10
1i 8
0 6
1- -Gauge2
4 --Gauge4
2
0
0 1 2 3 4 5
Strain (E-3}
Figure 7.13- Load versus strain histories for x- andy-direction top bars of
specimen Sl.
7-17
-Gauge&
-Gauge7
20
18
- 16
:.
~
14
- 12
i0 10
- 8
1ii
0 6 -GaugeS
..... 4
-Gauge9
2
o~~~~~~-L-L-L~-+~~~~~~~L-L-~
0 0.5 1 1.5 2
Strain (E-3)
Figure 7.14- Load versus strain histories for bottom bars of specimen Sl
(a)X-direction reinforcement and; (b) ¥-direction
reinforcement.
7-18
Punching failure
surface
(See Plate 5.2)
(a) Top surface of the slab (b) Bottom surface of the slab
Plate 7.2 - Failure crack at the top surface of the slab for specimen S2.
7-19
ck line
Y'---L---J\Lm~/'7,.-~
- "
Detail A
y X -Y X
(mm) (mm) (mm) (mm)
0 440 0 440
70 385 120 440
160 450 210 410
240 330 335 405
375 160 375 300
525 20 455 200
525 0 500 100
520 20
520 0
Figure 7.16- Cracks near the column; (a) crack detail near the column at slab
top and; (b) punching cone as seen at the free edge of the slab of
specimen S2.
7-20
The strain histories for the working strain gauges attached to the top and bottom
reinforcement (refer Figure 6.32b) are shown in Figures 7.17 and 7.18. None of the
bars had yielded at the point of failure indicating that the punching event occurred
7.4.3 Specimen S3
Specimen S3 failed in punching shear with a sudden bang. The applied load at failure
(excluding self weight) was 391.2 kN with the corresponding total pressure on the slab
The crack patterns developed at the top and the bottom surfaces of the slab are
presented in Figure 7.19. Because of the prestress effect these cracks could not be seen
easily. For reasons of safety no crack patterns were recorded until after failure of the
specimen and release of the applied load. Cracks developed at the top of the slab near
the column and are shown in Plate 7.3. The failure surface of the slab is shown in
Figure 7.20. At the soffit of the slab the failure surface started from the junction of
column and the slab, similar to that for specimens S 1 and S2. Due to the influence of
the prestress anchorage plates, the crack that formed on the free edge created a stepped
cone rather than a continuous cone. The average angle of the failure surface to the
horizontal was 26° and is steeper than that for specimens S 1 and S2.
The strain histories for the tension reinforcement (refer Figure 6.32c) are presented in
Figures 7.21 and 7.22 for the top and bottom reinforcement, respectively. None of the
25
Gauge3
Gaugel
-c
ca
0..
15
20
"D
ca
.2
1i 10
0
1-
5
0
0 0.5 1 1.5 2 2.5
Strain (E-3)
25
i 20
~
-"D 15
8
1i 10
0 -Gauges
1- 5
-Gauge&
0 0.5 1 1.5 2
Strain (E-3)
Figure 7.17- Load versus strain histories for top bars of specimen 82.
7-22
25
l20
~
- 15
i
: 10
0 -Gauge7
1- 5 -GaugeS
0 0.5 1 1.5 2
Strain (E-3)
25
-20
ca
D..
c"C 15
ca
.2
"i 10
0
1- -Gauge9
5
0
0 0.2 0.4 0.6 0.8 1
Strain (E-3)
Figure 7.18 - Load versus strain histories for bottom bars of specimen S2.
7-23
Punching failure
urface (See Plate 5.3)
.
(a) Top surface of the slab (b) Bottom surface of the slab
Plate 7.3 - Failure surface at the top of specimen S3 in the region of the column.
7-24
..... - "
Detail A
crack line
•
I
327 322 y X -Y X
(mm) (mm) (mm) (mm)
0 403 0 403
327 420 322 412
0 297 0 342 226
N
'<t 304 0
y
0
\ . 297 .. \ • 304 .. \
z
Punching failure surface projected on the
edge surface:
y z -Y z
(mm) (mm) (mm) (mm)
100 0 100 0
255 12 255 13
304 100 297 100
(b) DETAIL A - FRONT VIEW
Figure 7.20- Punching cone for specimen S3 (a) at the top surface above the
column; and (b) at the free edge of the slab.
7-25
30
-
ca
~ 20
25
,-ca
15
.2
'iii 10
0
1-
5
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Strain (E-3)
30
-
ca
25
,-
~ 20
ca 15
.2
-
'iii 10
0
1-
5
-GaugeS
-Gauge&
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Strain (E-3)
Figure 7.21 -Load versus strain for the top reinforcement of specimen S3.
7-26
30
-as
25
-
c.
~
"'C
20
as 15
.2
'iii
0 10 -Gauge7
1-
5 -GaugeS
0
0 0.2 0.4 0.6 0.8 1 1.2
Strain (E-3)
30
-
as
25
-
~
"'C
as 15
.2
20
'iii
0 10
1- -Gauge9
5 -Gauge10
0
0 0.2 0.4 0.6 0.8
Strain (E-3)
Figure 7.22 - Load versus strain for the bottom reinforcement in specimen S3.
7-27
7.4.4 Specimen S4
Specimen S4 failed at an applied load (excluding self weight) of 322 kN. The
corresponding total pressure on the slab (excluding self weight) was 35.4 kPa. The
slab failed in a brittle punching shear mode and was accompanied with a loud bang.
The punching shear surface is shown in Plate 7.4 and is mapped in Figure 7.23. The
failure surface started at the junction of the support column and the slab soffit. The
measurements of the punching cone along the top and along the free edge of the slab
The strain versus total applied load (including self weight) for the working strain
gauges (located as shown in Figure 6.32d) are presented in Figures 7.24 and 7.25 for
(Figures 7.24 and 7.25) show that the top reinforcement along the long edge had
yielded before failure. The negative moment reinforcement along the short edge and
the positive moment reinforcement had not yielded before failure indicating a reserve
of flexural capacity.
7.5.1 Introduction
The test results of specimens S 1, S2 and S3 show that increasing the prestressing force
gives an increase in the punching load carrying capacity of the connections. In the
same way the results of specimen S4 show that prestress has a significant influence on
the behaviour of comer column connections. In the following sections direct and
indirect comparisons are presented to show the influence of prestress on the punching
I
IZ
----
Detail A
fi
>----
X
crack line
z z
520
y
[!] 0 0
.[!] Ee
Figure 7.23- Crack details near the column of specimen S4 (a) at the top surface;
(b) on the face of the YZ edge; and (c) on the face of the XZ edge.
7-30
45
40 Gaugel
-35
ca
,-ca 25
c.
~
30
..2 20
1ii
0 15
1- Yield limit
10
5
0
0 1 2 3 4 5 6
Strain (E-3)
Figure 7.24- Load versus strain histories for X- andY-direction top bars of
specimen S4.
45
40
35
Tot
al 30
loa
d 25
(kP 20
a)
15
10
5
0
0 0.5 1 1.5 2 2.5 3
Strain (E-3)
Figure 7.25- Load versus strain histories for X-direction bottom bars of
specimen S4.
7-31
The theoretical moments and shear forces obtained from the equivalent frame model
and the simplified method of AS 3600 (1994), for the model structure shown in
Figure A.2, are presented in Tables 7.3 and 7.4 together with the measured values. In
Table 7.5 the shear force to bending moment ratios are given for these models together
with the measured values. The bending moments and shear forces calculated in
Tables 7.3 and 7.4 are for a design load equivalent to the measured failure load. While
the specimens were sized to attract a similar moment to shear ratio to that of a multi-
bay flat slab systems, this clearly was not achieved. Table 7.4 shows that the measured
shear forces were generally 15 percent lower than the design values while Table 7.3
shows that the moment in columns was approximately half of the code values.
Table 7.1 and Figure 7.26 show that the load carrying capacity for specimens S2 and
S3 increased by 15 % and 35 % with the prestress of 0.8 MPa and 2.2 MPa,
respectively, compared to the non-prestressed slab Sl. Thus, the prestress has
connections.
The tests show that the number and the spread of cracks were better controlled with
increasing amounts of applied prestress. The damage of the top surface of the slab at
failure was observed to be over a smaller conical region with the increase of prestress.
That is, the average slope of the failure cone increased with increasing prestress. The
slope of the failure surface measured from the horizontal varied from 13 to 26 degrees
for specimens S 1 to S3. The increase in average slope can be attributed to the
7-32
30
S3
25
20
-"'
,-
a.
..llil:
15
"'
.S!
"ii
0
1-
10
5
Deflection at Point A
(refer Figure 6.31a)
0 10 20 30 40
Deflection (mm)
Figure 7.26- Load versus deflection at the mid panel for specimens Sl, 82 and
S3.
7-33
Table 7.3- Comparison between the theoretical and measured bending moments
at the column connections for different specimens.
Table 7.4 - Comparison between the theoretical and measured load transferred to
the column for different specimens.
£_ ~-t)
M
confinement effects of the prestress. Regan (1981 ), Hammill and Ghali ( 1994), Walker
and Regan ( 1987) and Alexander and Simmonds ( 1986) all reported failure cones of
lower than 45 degrees. For the comer column specimen S4 the angle of failure cone
was approximately 18 degrees. The slopes of the punched cones observed in different
tests in this and the previous studies indicate that the ACI 318-1999 and AS 3600-
1994 models underestimate the size of the critical perimeter. The load transferred to
the edge column increased with the increase of prestress although the size of the slab
Figure 7.27 compares the crack pattern formed in the non-prestressed comer
Chapman (1974) and Lim and Rangan (1994) with that of the column of comer-
column specimen S4. The failure cone of specimen S4 is similar to that of the edge
column connections and to that used in the CEB-FIP Model Code 1990 (1993) rather
··..
·····... ·······. . ~Crack line
··...
------------, '• ..
I ··..
I
~--~---------····~·----~----x
...... / C r a c k lines
............ .
'•..
··... ·····
Figure 7.27- Comparison of the crack pattern developed at the top of the slab for
comer columns (a) reinforced concrete specimen reported by
Zaghlool and de Paiva (1973), Stamenkovic and Chapman (1974)
and Lim and Rangan (1994); and (b) prestressed specimen S4.
7-36
Different design codes have different models for predicting the punching shear
strength of flat plate slab-column connections. For the purpose of the comparison the
strength is normalised against the shear strength for zero moment (Vuo) and the
moment for zero shear (Mu 0 ). The normalised equations for different code models are
presented in Appendix D. All strength reduction factors and partial safety factors for
the material are taken as unity. Where the cube strength lfcu) is used in the code model
the cube strength is taken as feu = 1.25 fern , where !em is the compressive cylinder
The results of the tests are compared against various code models (see Appendix D) in
the following sections. Also included are the test data from other investigators with
1) all specimens included in the data pool with any mode of failure and; 2) only those
specimens included where the capacity of the slab was governed by punching shear
failure. That is where the load on the slab could not be increased for increasing
studies, however, specimens failing in a punching mode before reaching their flexural
capacity are not always clearly reported. For the purpose of the following comparisons
flexural mechanism and specimens with average reinforcement content of greater than
one percent are included in the second data set. A total of 40 specimens including 7
prestressed concrete connection specimens are included for the comparison of the test
results of edge column connections. Details of the data are given in Tables 7.6 and 7. 7.
Table 7.6- Test data for reinforced concrete edge column connections.
Note:# reported mode of failure. Mode of failure for some tests are not known from
the available literature and mentioned as "not reported" in these tables.
7-38
Table 7.6 (Continued) - Test data for reinforced concrete edge column
connections.
Note:# reported mode of failure. Mode of failure for some tests are not known from
the available literature and mentioned as "not reported" in the table.
7-39
Table 7.7- Test data for prestressed concrete edge column connections.
used for the comparison of test results of corner column connection specimens. The
details of these corner column connections are given in Tables 7.8 and 7.9. Only test
results of flat plate edge and corner column specimens without any shear
reinforcement (including ties in the spandrel) and without holes in the slab adjacent to
The normaslised moment-shear load path for specimens S1, S2 and S3 of the present
tests are compared with the ACI 318 (1999) model in Figure 7 .28. The figure shows
that the edge column specimens have a similar load path. In Figure 7.29 the
normalised load path for the corner column connection, specimen S4 is presented.
Compared to the edge column connections the slope of the interaction line for the
corner column connection is seen to be flatter and this is attributed to the effect of the
biaxial moment.
The test data given in Tables 7.6 to 7.9 for the edge and corner column connections are
Figures 7.30 to 7.33. Figure 7.30 shows the test data for the edge column connections
for the full data set. Figure 7.31 compares the data for the limited data set. Both
Figures 7.30 and 7.31 show a similar scatter in the data compared to the design model.
Similarly, Figures 7.32 and 7.33 show the scatter of test data for corner column
specimens for the full and limited data sets, respectively. Again the scatter in the data
is similar for the two sets. The scatter plots show that a large variation exists between
Table 7.8 - Test data for reinforced concrete corner column connections.
Table 7.8 (Continued) - Test data for reinforced concrete comer column
connections.
Table 7.9 - Test data for prestressed concrete comer column connections.
1.5
1
S3
ACI 318-1999
0.5
0 0.5 1 1.5
MufMuo
Figure 7.28 - Comparison of load path for edge column connection specimens Sl,
S2 and S3 with ACI 318-1999.
1.5
1
ACI 318-1999
0 0.5 1 1.5 2
MJMuo
Figure 7.29 - Comparison of load path for comer column connection specimens
S4 with ACI 318-1999.
7-44
1.5
0
:I
0
2;,1 A
> 0
X >C
0.5 •
- ~0 *
X
0
0 0.5 1 1.5 2
MJMuo
Figure 7.30- Comparison of test results with ACI 318-1999 model for edge
column connections (all data).
-ACI318·1999 • Speelman S1
+ Speelman S2 A Specimen S3
e EI-Salakawy et al. (1998) + Rangan et aL (1992)
2 - Mortin et aL (1991) - Rangan (1990)
0 Sunldja et al. (1982) ll Regan et aL (1979)
0 Stamenkovic et aL (1974) X Zaghlool (1971)
)I( Hanaon et aL (1968)
1.5
g
0
2;,1 A
> 0
llll X
~
>C
0.5 •
- _o X
0
Ql:
•
)I( 0
0 0.5 1 1.5 2
MJMuo
Figure 7.31- Comparison of test results with ACI 318-1999 model for edge
column connections (limited data set).
7-45
1.5
0 1 2 3
MJMuo
Figure 7.32- Comparison of test results with ACI 318-1999 model for corner
column connections (all data).
1.5
0
~ 1
0
0.5 X
~ ~ o"'r:JI x
¢ d«>x ¢
0 1 2 3
Figure 7.33- Comparison of test results with ACI 318-1999 model for corner
column connections (limited data set).
7-46
The normalised load path for specimens S 1 to S4 are compared with the AS 3600
model in Figures 7.34 and 7.35. The normalised load paths for the edge columns are
similar to those of the ACI 318 model (see Figure 7.28). However, the normalised path
for comer column connection specimen S4 (see Figure 7 .35) is steeper than that
The test data are plotted against the AS 3600 model in Figures 7.36 to 7.39.
Figure 7.36 shows the distribution of test data for edge column connections for all
data. Figure 7.37 shows the selected data set for edge connections. Similarly, the test
data for the comer column connections are plotted for all data against the code model
in Figure 7.38 and the selected data for the comer connection specimens failing in
punching shear are shown in Figure 7.39. There is a significant difference between the
code model and the experimental data for both data sets for both the edge and comer
column connections. Figures 7.36 to 7.39 show that the code model errs on the
conservative for the edge connections and generally for the comer connections,
although some data falls on the non-conservative side of the interaction curve.
Unlike the American and Australian code models, in BS 8110 the reinforcement
content is a factor in determining the strength of the connection. The shear strength is
The British code BS 8110 (1997) refers the designer to specialist literature for the
design of prestressed slab-column connections. In Figures 7.40 and 7.42 the test
7-47
g
.~
>
0 0.5 1 1.5
Figure 7.34- Comparison of load path for edge column connection specimens S1,
S2 and S3 with AS 3600-1994.
1.5
1
g
.; S4
0.5
o~~~~~+-~~~~~~~~~
0 0.5 1 1.5
M,*/Muo
Figure 7.35 - Load path for corner column connection specimens S4 with
AS 3600-1994.
7-48
1.5 0
0 0
:I
.~
>
0.5
1
6
6
+
~--
6
6
6
6
6
X • •
0 X
l •
0
0
•
0
::KcP
0
0 0.5 1 1.5 2
Mv"!Muo
Figure 7.36 - Comparison of test results with AS 3600-1994 model for edge
column connections (all data).
1.5 0
0
g
.~
~~·
1
>
•
0.5 ox 0
•
0
::Ko
0 0.5 1 1.5 2
Mv.,Muo
Figure 7.37 - Comparison of test results with AS 3600-1994 model for edge
column connections Oimited data set).
7-49
2
-AS 3600-1994
o Hammill et al. (1994)
•
¢
SpecimenS4
Walker et al. (1985)
o Stamenkovic et al. (1974) X laghlool (1971)
::K laghlool et al. (1970)
1.5
::1( X
0 ::1( X
..~
> 1
>tr.)Q 0
¢ 0
¢ ¢
~ ¢
0.5 ¢¢0 Xo
¢ ¢
0 0.5 1.5 2
Figure 7.38- Comparison of test results with AS 3600-1994 model for corner
column connections (all data).
1.5
)K X
0 )K X
:I
.c:: 1
~XJ
> 0
<> 0
X<> <>
<>
X
0.5 <>
0 0.5 1.5 2
Figure 7.39- Comparison of test results with AS 3600-1994 model for corner
column connections (limited data set).
7-50
data for the edge and corner column connections are plotted. The test data for the
prestressed connections in this study are compared to the code model together with the
data from other test series. Although BS 8110 (1997) does not strictly apply to the
design of prestressed specimens the prediction of strength for the test specimen is
consistent with the general data scatter. Figures 7.40 to 7.43 show that the British code
predicts well the mean of test data but considerable scatter exists.
For columns subjected to flexure but without axial force, equation D.l3
(see Appendix D) suggests that shear transfer via the slab is not a design
consideration. Thus there is poor correlation with experiments where no axial force
exists in the column. For example in specimen M.JE/2 by Stamenkovic and Chapman
(1974) (refer Figure 2.12) the slab-column connection is under the action of moment
without any axial force in the columns. By the code the capacity of this specimen is
not limited by shear capacity of the section. Other examples include specimen Z-V(4)
of Zaghlool (1971) and the corner column connection specimens, slab M/C/1 of
Stamenkovic and Chapman (1974), Z-ll(4) of Zaghlool (1971) and specimen NH4 of
The relative load path for the edge column specimens S 1, S2 and S3 and corner
column specimen S4 are plotted in Figures 7.44 and 7.45, respectively. The relative
load paths are steeper than those for the ACI 318 and the AS 3600 models.
The non-dimensionalised ratios of shear and moment for different test series are
plotted against the CEB-FIP model in Figures 7.46 to 7.49 for edge and corner column
7-51
-BS8110·1997 • Speelman S1
+
Speelman S2 A Specimen S3
e EI-Salakakwy et aL (1998) + Rangan at aL (1992)
- Martinet aL (1981) - Rangan (1990)
400 D Sunldja at aL (1982) <> Hawklna (1981)
A Regan at al. (1979) o Stamankavic at al. (1974)
X 2'aghlool (1971) :1: Hanaon at aL (1968)
350
300 X
z X X
c
.c
250 X X X
'&
e200
..
u;
: 150
.c
0
100
:1:
50
0
0 50 100 150 200 250 300 350 400
v.,(kN)
300
z
c.c 250
a,
c
2! 200
.
u;
ftl
Gl 150
.c
0
100
50
0
0 50 100 150 200 250 300 350 400
V811 (kN)
300 -898110-1997
• Specimen S4
0 Hanvnill et al. (1994)
250 <> Walker et al. (1987)
-
z
~200
0 Stamenkovic et al. (1974)
X Zaghlool (1971)
)I( Zaghlool et al. (1970)
-e
.s::.
0)
150 X
X
-;;
...
: 100
.s::.
CJ)
50
0
0 50 100 150 200 250 300
V8 ,,(kN)
-
z
~200
0 Stamenkovic et al. (1974)
X Zaghlool (1971)
)I( Zaghlool et al. (1970)
.s::.
a, X
-e...
0
: 100
150 X
.s::.
CJ)
50
0
0 50 100 150 200 250 300
Vett (kN)
S3
1.5
g
~ 1
0.5
0 0.5 1 1.5 2
MJMuo
Figure 7.44- Comparison of load path for edge column connection specimens S1,
S2 and S3 with CEB-FIP Model Code 1990 (1993).
1.5
~ 1
0.5
0 0.5 1 1.5 2
MJMuo
Figure 7.45 - Comparison of load path for comer column connection specimens
S4 with CEB-FIP Model code 1990 (1993).
7-54
1.5
[!,.
•
0
.(l
[!,.
X •
~
0
1 aa_, •o
~X'
0
0.5
0 0.5 1 1.5
MJMuo
Figure 7.46- Comparison of results with CEB-FIP model Code 1990 for edge
column (all data).
1.5 •
0
+
[!,.
X •
s
0
:I [!,.
1 -~ .0
[!,.
0.5 x•
0 0.5 1 1.5
MJMuo
Figure 7.47- Comparison of test results with CEB-FIP Model Code 1990 for edge
column (limited data set).
7-55
1.8
X
1.6
1.4
0
X "'
0
~ Xo
1.2
0
0
:I 1 oo8 0
~ 0 0
0 0 0
>0.8 X
X
-CEB-FIP 1990
0.6 • Specimen S4
0 Hammill et al. (1994)
0.4 0 Walker et al. (1987)
0 Stamenkovic et al. (1974)
X Zaghlool (1971)
0.2
0 lK Zaghlool et al. (1970)
0
0 0.5 1 1.5
MJMuo
Figure 7.48- Comparison of test results for comer column with CEB-FIP Model
Code 1990 (all data).
1.8
X
1.6
1.4 X "'
0
~ Xo
1.2
0
0 0 0 0
:I 1
~
>0.8
- CEB-FIP 1990
0.6 • Specimen S4
0 Hammill et al. (1994)
0.4 0 Walker et al. (1987)
0 Stamenkovlc et al. (1974)
0.2 X Zaghlool (1971)
lK Zaghlool et al. (1970)
0
0 0.5 1 1.5
MJMuo
Figure 7.49- Comparison of test results for corner column with CEB-FIP Model
Code 1990 (limited data set).
7-56
specimens. A wide scatter in the data is seen for both the edge and the comer column
specimens both for the full and limited data sets. For a number of test specimens from
different test series the model errs on the non-conservative and the general trend is
poor.
Figures 7.50 and 7.51 compare the Eurocode 2 model for edge column specimens with
the full and selected data sets, respectively. Figures 7.52 and 7.53 compare the
Eurocode 2 model for comer column specimens with the full and selected data sets,
respectively.
As for the British Standard BS 8110, Eurocode 2 does not make any allowance for
specimens where flexure is passed from the column to the slab in the absence of axial
force (see Figure 2.12). Therefore, for specimens without an axial force component the
comparison between the test data and the Eurocode 2 model is poor. With this
In all ~ases the comparison of the experimental data with the design codes show a
significant amount of scatter. In Tables 7.10 and 7.11 the means and the standard
deviations are given for the test data compared with each of the code models for the
edge and comer columns, respectively. In Table 7.12 the means and standard
deviations are given for all data for both the comer and edge connections.
7-57
400
350
300
z25o
.Ill:
"';:'
'Q
200 xx
X
..: 150
100
50
o~~~~~~~~~~~~
v.d (kN)
Figure 7.50 - Comparison of test results for edge column connection with
Eurocode 2 (all data).
400
350
300
z25o
.Ill:
"';:'
'Q
200 xx
X X
..: 150 &<
100
50
o~~~~~~~~~~~~
Vad (kN)
Figure 7.51 - Comparison of test results for edge column connection with
Eurocode 2 Oimited data set).
7-58
- Eurocode 2 (1992)
400 • Specimen S4
0 Hammill et al. (1994)
350 <> Walker et al. (1987)
o Stamenkovlcet al. (1974)
300 X zaghlool (1971)
:::K zaghlool (1970)
-250
-;!
~
.... 200
~~
DO
:::K X
150 X
• )I(
100
50
o~~~~~~~~~~~~~~~LWY
Figure 7.52 - Comparison of test results for corner column connection with
Eurocode 2 (all data).
-Eurocode 2 (1992)
400 • Specimen S4
0 Hammill et al. (1994)
350 <> Walker et al. (1987)
0 Stamenkovlc et al. (1974)
--
~
250
200 Do
J 150
3E~ :::K X
X
.)I(
100
50
0
0 50 100 150 200 250 300 350 400
vsd (kN)
Figure 7.53 - Comparison of test results for corner column connection with
Eurocode 2 (limited data set).
7-59
Table 7.10- Summary of analyses for the experimental data compared to design
codes for edge column connections.
Table 7.11- Summary of analyses for the experimental data compared to design
codes for corner column connection.
Table 7.12 - Summary of analyses for the experimental data compared to design -
codes for all data of edge and corner column connections.
DataSet I DataSet II
For the British Standard BS 8110, the mean measured shear strength to the capacity
calculated from the code (Ve.u) is 1.03 and 1.06 for the edge and comer columns,
respectively. When all the data for the edge and comer column connections are
considered it is 1.04. While the best fit is with BS 8110, the standard deviation is high
at 0.37. This implies that a significant amount of data lies on the non-conservative side
of the design model. The high standard deviation is of concern and requires a high
partial safety factor if the overall safety of the structure is to be maintained. For the
test data used in this study, BS 8110 gives a significantly improved prediction than
The Australian code of practice, AS 3600, has one of the largest mean experimental to.
theory ratios at 1.63 and 1.49 for edge and comer column connections, respectively.
7-61
Figures 7.36 and 7.38 show that no points fall below the theoretical strength for edge
connections and few data points fall below the theoretical strength for the comer
column connections. The mean and standard deviation of the combined edge and
comer connections for all data are 1.59 and 0.45, respectively. Assuming a normal
distribution of data there is a 91 percent level of confidence that the design model
The scatter of data for the ACI model is similar to that for the AS 3600 model but
inspection of Figures 7.30 and 7.36 show that more data points fall below the
theoretical limit than for the Australian code model. The mean of all data for the ACI
model is 1.45 with a standard deviation of 0.40. The confidence level of any one point
falling on the theoretically safe side of the interaction line (again assuming a normal
distribution) is 87 percent.
The Eurocode 2 model is of a similar form to the British model but with considerably
fewer data points falling on the non-conservative side. The mean experimental to
theoretical ratio for the combined data of edge and comer connections is 1.47; higher
than both the British and American codes but lower than the Australian code. The
corresponding standard deviation for the Eurocode 2 model of 0.55 gives a confidence
level of 80 percent. Comparing Tables 7.10 and 7.11 shows that there is a significant
variation in the mean and standard deviation between the edge and comer data sets.
The CEB-FIP (1993) model has one of the lower means for edge column specimens
(Table 7.10) but has a high standard of deviation of 0.37. The mean and standard
deviation for comer column connections are 1.72 and 0.63, respectively, and are the
7-62
highest of all models. An inspection of Figures 7.46 and 7.48 also shows that a number
The results of the statistical analysis carried out for the connections reinforced with
Tables 7.13 and 7.14, respectively. Except for the British code the means of the
experimental data to code ratio increased with a greater scatter when prestress is added
to the slab. It is also seen (Tables 7.10 to 7.15) that the means are generally higher
when the specimens not identified as punching failure are excluded from the data pool.
The generally poor predictions, however, of all code models together with the high
degree of scatter in the data suggests that the mechanics of punching shear are not well
represented by the code models. Further research is required into the load transfer
The analysis of results discussed above do not consider the effect of material partial
Table 7.12 is reproduced as Table 7.15 with the effects of the strength reduction
(or partial safety) factors included in the calculations. It is seen that BS 8110 yields the
most economical design while the CEB-FIP (1993) model the most conservative. The
confidence levels for the data set IT with and without the effect of the safety factors are
The calculations on the probability of a single data point falling below the theoretical
strength for data set IT with the material factors excluded is as high as 4 7 percent for
the British code (BS 8110) and down to 9 percent for the Australian code (AS 3600).
7-63
Table 7.13 - Summary of analyses for the experimental data compared to design
codes for edge column connections considering only non-prestressed
concrete connections.
Table 7.14- Summary of analyses for the experimental data compared to design
codes for edge column connections considering only prestressed
concrete connections.
Table 7.15 - Summary of analyses for the experimental data compared to design
codes for all data of edge and corner column connections
considering the effect of t/1 and Ym as appropriate.
Note: ifJ strength reduction factor and Ym. ~. and Yc are partial safety factors for the
materials depending on the code.
Table 7.16- Confidence level for data Set II with q, or Ym set equal to unity and
with the values set by corresponding code.
Note: ifJ strength reduction factor and Ym. ~. and Yc are partial safetys for the materials
depending on the code.
7-65
When material strength reduction factors (or partial safety factors) are included in the
calculations only the Australian code model gives a less than five percent probability
of failure. The ACI, Eurocode and CEB-FIP models give probabilities of a failure
between six and eight percent. The probability that an edge or comer column
7.6 Conclusions
All specimens tested in this study failed in a punching shear mode and the failure
surface started from the intersection of the column and slab soffit. The slopes of the
failure surface with reference to the horizontal plane were between 17 and 26 degrees.
The primary conclusion from the tests is that prestressing has an influence on the
carrying capacity of the connections and relatively larger unbalanced moments can be
transferred to columns. The formation of cracks and the extent of damage are also
controlled by the addition of prestress into the concrete slab. The failure mode,
slab-column specimens, using various code models, shows a large scatter in the data
and generally poor predictions. However, for the most part the codes give conservative
predictions with the exception of BS 8110 which shows a good prediction of the mean
of the test data although, as for the other models, the scatter is high.
different code models fails before the code requirement show that only the Australian
7-66
code gives a lower than five percent probability of failure, although the ACI,
Eurocode 2 and the CEB-FIP model code are not significantly greater than five
8.1 Introduction
In Chapter 3 of this thesis different material models, mesh patterns and aspect ratios
were investigated to obtain optimum results for the analysis of the punching type
problems. The model and the finite element program DIANA were tested and verified
range of problems such as simple flexure to the more complex punching shear in flat
and 7, the testing of four flat plate-column connections was discussed. These
specimens were developed from a one-third-scale model of the prototype flat plate
structure. Three of the specimens represented connections with edge columns while
the fourth specimen represented a comer column connection. Except for specimen S 1
the specimens were prestressed in two directions with the amount of prestress being
varied. A number of evenly distributed point loads were applied to the top surface of
the slabs simulating a uniformly distributed load. All the specimens failed in a
In this chapter, the experimental slabs reported in Chapter 6 are modelled using the
finite element program DIANA as discussed in Chapter 3. The concrete was modelled
using 20-node 3D isoparametric brick elements with the three degrees of freedom at
each node. The steel was modelled using embedded bar elements.
In addition to the numerical modelling of the test specimens, further analyses of the
as a part of the experimental testing program such as bonding of the prestressing steel.
8-2
The finite element meshes developed for slab-column connections S 1, S2 and S3 are
shown in Figures 8.1 and 8.2. One half of the specimens were modelled using
symmetry. The finite element models consisted of a total of 2137 nodes and 344
The reinforcement, including the prestressing wires, were modelled as bar elements
embedded in the concrete with an additional nodes used to define the embedded bars.
Straight reinforcing bars were modelled with two additional nodes per element while
the prestressing wires with parabolic prof:tles were defmed using three nodes per
element. The position of the reinforcement in the finite element model was as for the
test specimens. The specimens S2 and S3 differ only in the content of prestressing
wires and the intensity the applied prestress. Specimen S 1 had a different arrangement
The basic difference between the models is the reinforcement arrangement and
content. Details of the nodes and elements used for the reinforcement for the different
The fmite element mesh used to model specimen S4 is shown in Figures 8.3 and 8.4.
The mesh consisted of 2799 nodes and 460 brick elements to model the concrete and
110 embedded bar elements to model the bonded reinforcement and prestressing
wires. As for slabs S2 and S3, in slab S4 2-node bar elements were used to represent
the straight bars and 3-node element for the parabolic cables. Details of the finite
0 -x
Column - 3515
z
3515
Hinged slab
Hinged column edge
centre
FRONT ELEVATION
z
2400
Simply supported
Hinged edge
SIDE ELEVATION
Figure 8.1 - Plan and elevation showing the finite element mesh for edge column
models Sl, S2 and S3.
8-4
X
y
Plane of symmetry
0
Figure 8.2 - 3D view of f"mite element mesh for models Sl, S2 and S3.
25
[_ -x
Column
.Jo 3515
NOTE:
ml Load points
z
3515
Hinged slab
base
Hinged column
centre
FRONT ELEVATION
z
2500
Hinged slab
base
Hinged column
centre
SIDE ELEVATION
Figure 8.3 - Plan and elevation showing the finite element mesh for model S4.
8-6
X
y
implemented to solve the problem. In the early part of the analyses (the linear range)
load control was used while load-displacement control via an updated normal arc-
length procedure was used in the latter part of the analyses. Convergence was set at 1
percent of energy norm with a maximum of 50 iterations for any one load step.
The concrete was modelled using a cohesion-plastic strain curve based on the concrete
taken as 0.9fcm and tensile was taken as 0.33~ fcm , where !em is the mean cylinder
compressive strength of the concrete. The modulus of elasticity used for the concrete
(Ec) was taken to match the measured linear response of the experimental tests. The
concrete properties used for the finite element modelling are given in Table 8.2.
8-7
A bi-linear or tri-linear elasto-plastic model (refer Figures 3.4a and 3.4b) was adopted
for the conventional and prestressing reinforcement. The values of material parameters
used are given in Tables 8.3 and 8.4 for the conventional steel and prestressing wires,
respectively.
The ultimate load carrying capacities of the specimens obtained from the tests and the
numerical analyses are given in Table 8.5 and are in close agreement with that of the
experimental results. The exception is specimen S3 in which the finite element model
predicted a 20 percent higher failure load than that measured. The mean of the
numerical failure loads to the test failure loads was 1.07 with a standard deviation of
0.08.
The results of the individual tests are described below. The load versus deflections,
steel strains and reactions are compared with those of the experimental study and
The finite element model for the specimen S 1 predicted a punching failure load close
to that measured in the test (see Table 8.5). The deformed shape of the slab and the
contour plot for the vertical deflections at failure are shown in Figure 8.5. The load
versus deflection plot for the measured slab deflection (see Figure 6.31a for details) is
given in Figure 8.6. A reasonable correlation between the numerical model and test is
seen in the load-displacement history although the deflections show the finite element
Ec (GPa) 20 24 20 25
a(degrees) 15 15 15 15
l/J (degrees) 30 30 30 30
#
Note: 0.2 % Proof stress
8-9
#·
Note: 0.2 % Proof stress
(a) 3D view
l-.1~1E-2
l-.32~E-2
-.5B7E-2
-.69E-2
-.873E-2
~ -.1B6E-1
l-.12~E-1
l-.142E-1
B-.16E-1
1-.179£-1
(b) Contour plot (View from top with the column at bottom left corner)
0 10 20 30 40
Deflection (mm)
20
18
_16
l14
~
-12
"tJ
8 10
- 8 --Experimental E
"iii
0 6 · · ·······Experimental D
1- 4 --FEEandD
2
0 ~~~~~~~~~~~~~~~~~~~
0 2 4 6 8 10 12
Deflection (mm)
In Figure 8. 7 the load is plotted against horizontal and vertical reactions at the base of
the column. The analytical model gave a similar result to that measured up to the
initiation of the punching failure. As the punch took place, however, the reactions on
the test slab started to decrease and this was not seen in the finite element model.
The load versus steel bar strains at the points measured in the slab (see Figure 6.32a)
are plotted in Figures 8.8 and 8.9 for the top and bottom bars, respectively. These plots
show a good correlation between the finite element model and the experimental test.
As measured in the test a top bar in the x-direction located at the centre of the column
yielded before the punching failure (see Figure 8.8a). The finite element model
indicates, however, that the adjacent bars had not yielded before the punching failure.
Therefore the slab retained residual flexural capacity and failed in a true punching
failure mode.
18 degrees to the horizontal .. In the finite element modelling for Slab 1, the major
principal strains across the experimentally observed failure plane were obtained and
are plotted in Figure 8.10 for increasing load. The figure shows that first cracking
load cracking had occurred across a significant portion of the failure surface and it can
be said that the failure crack was initiated at between 40 and 60 percent of the failure
load. This observation is consistent with the experimental observations of Moe ( 1961)
whose experimental tests showed that the failure crack was initiated at about 50
percent of the failure load. Thus the results of cracking strains output in the finite
element model are consistent with those observed in the laboratory. The compressive
8-13
strain along the junction of the slab soffit and the column are shown in Figure 8.11.
The magfiitude of the principal tensile strain along this failure line is as high as 7600
Jl£, almost 100 times the concrete cracking strain. The magnitude of the maximum
compressive strain was 4890 J1£ occurring at the internal corner of the column but
reduces quickly away from the corner of the column. This indicates only local
crushing of the concrete near the slab-column connection. The principal tensile strain
contours are shown in Figure 8.12 and show the high tensile strains along the failure
surface.
20
18
-ftS
16
14
-
a..
~
"C
ftS
.2
12
10
"i 8
0 6
1-
4
2
0
0 20 40 60 80 100 120
Reaction (kN)
Figure 8.7 -Load versus horizontal (Rx) and vertical (Rz) reactions at column
support for specimen Sl.
8-14
20
18
_16
£f14
~ Yield limit
~2
~10
- 8
"i -Experimental Gauge 2
15 6 --FEGauge1
..... 4
~FEGauge2
2
o--~~~~~~~~~~~~4-~~~-+~~~~
0 1 2 3 4 5
Strain (E-3)
20
18
-
: . 14
c"D 12
16
CIS 10
.2
-"i 8
0
t-
6
4
-Experimental Gauge 4
--FEGauge4
2
0
0 0.5 1 1.5 2
Strain (E-3)
Figure 8.8 - Load versus strain histories for x- and y-direction top bars of
specimen Sl (a) x-direction reinforcement; and (b) y-direction
reinforcement.
8-15
20
18
-ftl
16
-,
c.
~
14
12
ftl 10
.2
'i 8 --Experimental Gauge 6
0 6 · · ·······Experimental Gauge 7
1-
4
• FE Gauges 6 and 7
2
0
0 0.5 1 1.5
Strain {E-3)
20
18
_16
.f 14
c12
i0 10
- 8
'i --Experimental Gauge 8
0 6
1- 4 - - - - -Experimental Gauge 9
2 - - F E Gauges 8 and 9
o--~~~~~~~~-+~~~~4-~~~~
0 0.5 1 1.5 2
Strain {E-3)
Figure 8.9 - Load versus strain histories for bottom bars of specimen Sl
(a)x-direction reinforcement; and (b) y-direction reinforcement.
8-16
At failure load
Maximum strain = 7656 J.l.l:.
A
B At 70 % of the failure load
Maximum strain = 901 j.J.f:.
A~
A~
At about 60 % of the failure load
Maximum strain = 462 JLE
A~
B SCALE:
t-----1 : 1491 j.J.f:.
SLAB
AB: Failure plane
_.4
.11 . .4 ;
COLU_MN.
:.1 •
4 .
Figure 8.10- Principal tensile strain along failure line for specimen Sl on the
vertical surface near the central symmetrical vertical plane.
8-17
Plane of SCALE:
symmetry f--------1 :4890 Ji-E
Figure 8.11 - Principal compressive strain at the slab and column junction for
specimen 81.
1.121E-l
2.189E-l
.964E-2
.839E-2
.714E-2
1 .589E-2
1.~65E-2
1.J4E-2
i.215E-2
1.985E-J
I
Figure 8.12 - Contour of principal tensile strain at the column region of the finite
element model for specimen 81.
8-18
The results of the finite element modelling of slab S2 gave a punching failure at a load
of 6 percent higher than that of the experiment. The deflected shape of the model at
the point of failure and a contour plot of the vertical displacements are shown in
Figure 8.13. The load versus deflection histories for different key points in the slab are
plotted in Figure 8.14. The deflections obtained from the finite element model
compare reasonably well with the test results although the model slab was stiffer than
The load versus horizontal and vertical reactions are given in Figure 8.15. The figure
shows a very good correlation between the numerical and the test results.
8-19
(a) 3D view
L.t32E-2
1-.32E-2
t-.SBBE-2
-.696E-2
-.883E-2
-"-.187E-l
l-.126E-1
•-.HSE-1
1-.163E-1
I-.182E-1
(b) Contour plot (View from top with the column at the bottom left corner)
25
a;- 20 Experhnental A
c.
~
-"C 15
~
"i 10
0
..... 5
0 10 20 30 40
Deflection (mm)
25
-20
ftl
c.
=- 15
"C
ftl
.S!
"i 10 -Experimental D
0 -Experimental E
..... 5
--FEDandE
0
0 1 2 3 4 5 6
Deflection (mm)
25 FERx FERz
-
ca 20
,-
a.
~
15
ca
.2
"i
10
0
.... 5
0
0 50 100 150
Reaction (kN)
Figure 8.15 -Load versus horizontal (Rx) and vertical (Rz) reactions at column
support for specimen S2.
Load versus strain histories for the reinforcing bars gauged in the experiment (see
Figure 6.32b) are shown in Figures 8.16 and 8.17 for the top and bottom bars,
respectively. These plots show a reasonable correlation between the finite element
model and the experimental test although the bottom x-direction reinforcement
indicates the finite element slab is stiffer than that of the experiment. This is consistent
with the deflection observations. The strain histories show the maximum strains in the
The distribution of principal tensile strain along the experimentally measured line of
failure is plotted in Figure 8.18. The maximum principal tensile strain along this
failure line was 8000 JlE, 120 times larger than the cracking strain. The compressive
strains at the slab soffit close to the column junction are plotted in Figure 8.19. For
8-22
FE Gauge 1
25
-c
ca
a.
20
15
"CJ
ca
.2
"i 10
0
.....
5 Yield line
0
0 0.5 1 1.5 2 2.5 3 3.5
Strain (E-3)
25
fti' 20
~
- 15
~ 10 --Experimental Gauge 5
~
..... 5
- - Expermental Gauge 6
- F E Gauges 5 and 6
0 0.5 1 1.5 2
Strain (E-3)
Figure 8.16- Load versus strain histories for x- andy- direction top bars of
specimen S2 (a) x-direction reinforcement; and (b) y-direction
reinforcement.
8-23
25
-20
ca
,-ca 15
~
~
.S!
"i 10 - Experiemntal Gauge 7
0 -Experimental Gauge 8
..... 5 --FE Gauges 7 and 8
0
0 0.5 1 1.5 2
Strain (E-3)
25
-20
:.
c"'C 15
!
"i 10
0 -Experimental Gauge 9
..... 5 --FEGauge9
0 ~~~~~~~~~~~~~~~~~~~~~
Figure 8.17 -Load versus strain histories for bottom bars of specimen S2
(a) x-direction reinforcement; and (b) y-direction
reinforcement.
8-24
SCALE:
1----------l : 1856f.U; ~COLUMN
Crocking strain = 70 f.-LC 4
Figure 8.18 - Principal tensile strain along failure line for specimen S2 on the
vertical surface near the central symmetrical vertical plane.
5240 J.t.e(Maximum)
Plane of Scale:
symmetry • '• -: •• : ~ 1-------l :5240 j.tt:
......
Figure 8.19 - Principal compressive strain at the slab soffit and column junction
of specimen 82.
8-25
simplicity the directions of compressive strains are not considered in the plot. The
maximum compressive strain was 5240 J.le at the comer of the column but reduces
quickly away from the comer of the column indicating only local crushing of the
concrete. The contours of the principal tensile strain in the region of the connection are
plotted in Figure 8.20. The distribution of the principal tensile strains obtained from
the finite element model show that the failure of the specimen is due to punching.
The results of the finite element model of slab S3 gave a failure load of 30.7 kPa, 20
percent higher than the experimental failure load. The deformed shape of the finite
element model and the vertical displacement contours are shown in Figure 8.21. The
load versus deflection graphs comparing the results of finite element model to the
experimental results are presented in Figure 8.22. Although the trends of the curves
obtained in the finite element model are similar to the experimental results, the finite
element model is stiffer and cracked later than was observed in the test.
Plots of the applied load versus horizontal and vertical reactions are given m
Figure 8.23. The reactions obtained from the numerical model are in good agreement
The numerical and experimental load versus strain histories at the locations shown in
Figure 6.32c, are compared in Figures 8.24 and 8.25 for the top and the bottom
reinforcing bars, respectively. As for the deflections the strains from the finite element
model indicate that the numerical model was stiffer and cracked later than the test
slab.
8-26
i.171E-1
n.lS<E-1
.137E-1
.119E-1
.192E-1
l.a«-z
!.677E-2
1.504E-2
i.331E-2
:.159E-2
Figure 8.20 - Contour of principal tensile strain at the column region of the finite
element model S2.
8-27
(a) 3D view
1-.126E-2
1-.295E-2
-.465E-2
-.635E-2
-.8B4E-2
f-.974E-2
l . ~. )( t-.114E-l
1-.131E-1
1-.HSE-1
~-.165E-1
(b) Contour plot (View from top with the Column at bottom left corner)
35
30
-
~
.:.:::
25
:; 20
as
.2 15 -Experimental A
"i -Experimental 8
0 10 ---FE A
1-
-+-FE 8
5
0
0 5 10 15 20 25
Deflection {mm)
35
30
-
a.as 25
.:.:::
:; 20
as
.2 15
"i -Experimental D
0
1- 10 -Experimental E
-----FE D and E
5
0
0 1 2 3 4 5 6
Deflection {mm)
35
-
: . 25
30
-,
.:..::
ca
.2 15
20
"i
0 10
1-
5
0
0 50 100 150 200
Reaction (kN)
Figure 8.23 - Load versus horizontal (Rx) and vertical (Rz) reactions at column
support for specimen S3.
8-30
35
30
-t.
~
25
:; 20
ca
.2 15 -Experimental Gauge 1
"i --Experimental Gauge 2
15 10
t- -FEGauge1
5 -+-FE Gauge 2
0
0 0.5 1 1.5 2 2.5 3
Strain (E-3)
35
-t.
~
25
30
:; 20
ca
.2 15
"i -Experimental Gauge 3
15 10 -Experimental Gauge 4
t-
-FE Gauges 3 and 4
5
0
0 0.5 1 1.5 2 2.5
Strain (E-3)
Figure 8.24- Load versus strain histories for x- andy- direction top bars of
specimen S3 (a) x-direction reinforcement; and (b) y-direction
reinforcement.
8-31
35
-
:. 25
~
30
:;- 20
8 15
"iii -Experimental Gauge 5
0 10
1- --Experimental Gauge 6
5 -FE Gauges 5 and 6
o--~~~~-r~~~~~~~L-~~+-~~~~~
0 0.5 1 1.5 2
Strain (E-3)
35
-
: . 25
~
30
:;- 20
ftS
,g 15
-Experimental- Gauge 7
"iii
0 10 --Experimental- Gauge 8
1-
5 -FE Gauges 7 and 8
0
0 0.2 0.4 0.6 0.8 1 1.2
Strain (E-3)
Figure 8.25 - Load versus strain histories for bottom bars of specimen S3
(a)x-direction reinforcement; and (b) y-direction
reinforcement.
8-32
The principal tensile strains along the measured failure line are plotted in Figure 8.26.
The maximum principal tensile strain was 8400 p,e, over 100 times the cracking strain
of the concrete. The principal compressive strains near the junction of slab soffit and
the column are plotted in Figure 8.27. The maximum compression strain was 4770 J.le
at the corners of the slab but, as for slabs S 1 and S2, reduces quickly away from the
corner. The principal tensile strain contours in the critical region of the connection are
The finite element model for slab S4 failed in punching shear at the load of 40 kPa and
compares favourably with the experimental result of 39 kPa. Figure 8.29 shows the
displaced shape and displacement contours obtained from the finite element analysis.
The load versus deflection plots for the points shown in Figure 6.31 b are given in
Figure 8.30. The numerical model predicted a stiffer response than measured in the
test. The earlier cracking observed in the experimental tests combined with the earlier
degradation of stiffness, compared to the finite element results, could be due to the
effect of the disturbance caused by the prestressing ducts. The duct may act as crack
inducers and with the unbonded tendons there is some loss of cross-sectional area.
In Figure 8.31 the load versus horizontal and vertical reactions are plotted. Some
differences between the finite element results and the experimental test are seen near
the failure load, otherwise the results show a reasonable correlation. The load versus
strain histories for the bar locations, shown in Figure 6.32d, are presented in
Figures 8.32 and 8.33 for the top and bottom reinforcement, respectively. The strain
histories determined from the finite element modelling are generally stiffer than those
measured.
8-33
LJ"¥
SCALE:
1----------l : 2080 P,l:
.ll ~
Figure 8.26 - Principal tensile strain along failure line for specimen S3 on the
vertical plane near the central symmetrical vertical surface.
4770 p.t:(Maximum)
#.--:-1--....,....":-"~~-J
Plane of SCALE:
symmetry 1--------l : 4 77 0 P,l:
Figure 8.27 - Principal compressive strain at the slab soffit and column junction
for specimen S3.
8-34
1.14SE-1
l.133E-1
.117E-1
.1B2E-1
t:~~=~
l.559E-2
L4B5E-2
I.251E-2
1.977E-3
I
Figure 8.28 - Contour of principal tensile strain at the column region of the finite
element model for specimen S3.
8-35
(a) 3D view
•-.137£-2
i-.32E-2
-.593£-2
-.SBSE-2
-.BSBE-2
~- .185E-l
l-.123E-l
1-.H2E-l
i-.!SE-1
1-.178E-l
I
(b) View from top (column located at the bottom left corner)
Experimental C
50
45
-c
ca
a.
40
35
30
"C
ca 25
.2
-
'ii
0
1-
20
15
10
5
0
0 10 20 30 40 50 60
Deflections {mm)
Experimental Ry
50
Experimental Rx
45
-c
:.
40
35
30
FERy
i0 25
-
'ii
20
15 15
1- 10
5
o~~~~-r~~~-+-L~~~~~-L~+-~~~~
0 20 40 60 80 100
Reaction {kN)
Figure 8.31 - Load versus horizontal (Rx) and vertical (Rz) reactions at column
support for specimen S4.
8-37
45
40
m 35
-i
~ 30
.2 20
25
- Experiemntal Gauge 1
'ii 15
0 -FEGauge1
..... 10 Yield limit
5 -+-FE Gauge 2
0
0 1 2 3 4 5 6
Strain (E-3)
45
40
m 35
-
~ 30
"C
~ 20
25
-Experimental Gauge 3
1i 15 -Experimental Gauge 4
0
..... 10 --.--FE Gauge 3
5 -FEGauge4
o--~~~-r~~~-+~~~~~~~~+-~~~~
Figure 8.32 - Load versus strain histories for x- and y- direction top bars of
specimen S4 (a) x-direction reinforcement; (b) y-direction
reinforcement.
8-38
45
40
Ci 35
-"'
~
"tJ
.2
30
25
20 -Experimental Gauge 5
"i 15 -Experimental Gauge 6
0
..... 10 -+--FE Gauge 5
5 ----FE Gauge 6
0
0 0.5 1 1.5 2 2.5 3
Strain (E-3)
45
40
Ci 35
-"'
~
"tJ
.2
30
25
20 -Experimental Gauge 7
"i 15
0
..... ----FE Gauge 7
10
5
0
0 0.5 1 1.5
Strain (E-3)
Figure 8.33 - Load versus strain histories for bottom bars of specimen S4
(a)x-direction reinforcement; and (b) y-direction
reinforcement.
8-39
The distributions of principal tensile strains along the experimentally measured lines
of failure near the free edges of the :XZ and YZ-planes are shown in
Figures 8.34 and 8.35. The maximum principal tensile strain was 11970 Jl£, 145 times
larger than the concrete cracking strain. The principal compressive strains developed
near the junction of the column and the slab soffit are shown in Figure 8.36. The local
maximum compression strain near the internal comer of the column was 5045 Jl£.
Away from the column comer the compression strain reduced rapidly indicating only
local crushing of the slab near the comer of the column. The principal tensile strain
contours are shown in Figure 8.37 further highlighting the high strain region in the
To cover the full range of prestress used in practice, a model with an average prestress
of 3 MPa was developed with a separate design being undertaken. As for the other
specimens the slab was designed for a factored design load of 14.75 kPa. All
parameters including the concrete properties and bonded reinforcement were kept as
for model S3. The new model with the prestress of 3.0 MPa is designated as
Model SE-3.0. The failure load obtained from finite element analysis is compared with
the finite element analyses of the other edge column connection models in Table 8.6.
The deflections at point A (see Figure 6.31a) on the slab are compared in Figure 8.38.
The failure loads against the average prestress applied to the slab are plotted in
Figure 8.39. The figures show an almost linear trend of increase in the load carrying
SCALE:
1------1 1009 """ COLUMN
Crocking strain = 80 JJ.£
Figure 8.34 - Principal tensile strain along failure line for specimen 84 near the
free XZ-surface.
SCALE:
1------1 2335 p,e COLUMN
Figure 8.35 - Principal tensile strain along failure line for specimen 84 near the
free YZ-surface.
8-41
COLUMN
SCALE:
:5045 f..U;
Figure 8.36 • Principal compressive strain along the slab soffit and column
junction for specimen S4.
I.HJE-1
f.128E-1
.ll~E-1
.SS2E-2
.MGE-2
~. 781E-2
t555E-2
l.~1E-2
1.26~E-2
1.119E-2
I
Figure 8.37 - Contour of principal tensile strain at the column region of the finite
element model S4.
8-42
40
--s1-o.o
35 -+-S2-0.7
-+-S3-2.2
-ca
a.
30
25
__._SE-3.0
,-
~
ca
20
0
...I
15
10
5
0
0 5 10 15 20 25
Deflection at A (mm)
Figure 8.38 - Load versus deflection plots for deflection at point A for finite
element results of edge column connections with varying
amounts of prestress.
Table 8.6 - Comparison of the failure loads for edge column connection models
with different prestress.
Sl-0.0 0 18.5
40
-35
cu
-
fl. 30
~
; : 25
,;; 20
cu
.2 15
1a
0 10
1-
5
0
0 1 2 3 4
Average prestress, (P/A)Avg (MPa)
Figure 8.39 - Load versus average prestress applied to slabs of edge column
connections (results off"mite element models).
To study the influence of prestress on the behaviour of corner column connections the
developed by designing tendons for a factored design load of 14.75 kPa (as for the
other models). The models are designated as S4-X.X, where X.X is the average
prestress in MPa applied to the slab. For example, Model S4-2.1 refers to model S4
with the average applied prestress in slab of 2.1 MPa. With the exception of the
arrangement and the intensity of the applied prestress, the amount and arrangement of
concrete material properties for all S4-X.X models (see Figures 8.3 and 8.4) were as
The failure loads for slab S4 with the different amount of prestress in slab are
compared in Table 8.7. The load deflection curves are compared in Figure 8.40 and
the failure load versus the applied prestress are plotted in Figure 8.41. As for the edge
column connections the load carrying capacity increases approximately linearly with
increasing prestress.
The horizontal reactions in x- and y-directions and the vertical reaction developed at
the base of the column are compared in Figures 8.42, 8.43 and 8.44, respectively. The
figures show that the prestress has only minor effect on the distribution of the
reactions. Strains in the x- and y-direction top bars at the locations shown in
Figure 6.32d are plotted in Figures 8.45 and 8.46, respectively. The strain histories of
x-direction bottom bars are shown in Figure 8.47. In no case had the steel yielded
indicating that the failure load was not limited by the flexural strength of the slab.
concrete structures. Finite element modelling was used to investigate the influence of
bonding of the prestressing steel on the behaviour of the edge and comer column
specimens.
The numerical models studied in the previous sections were modified with the
prestressing wires bonded to the concrete. The models are designated as S2-B, S3-B
and S4-B. All properties other than the bonding of the prestressing wires, including the
geometry and boundary conditions, were as for the previous models. The finite
8-45
50
45
-~ 35
40
c"D 30
ca 25 -s4-o.o
.2
"iii 20 -+-S4-1.0
0
..... 15 ......_84-2.1
_._S4-3.0
10
5
0
0 5 10 15 20
Deflections at A (mm)
Figure 8.40 - Load versus deflections at point A for slab model S4 with different
intensity of average prestress.
Table 8.7 - Comparison of the failure loads for corner column connection models
with different levels of prestress.
S4-0 0 18.6
60
"'5o
a.
.!II:
-40
;::
"ti
ca 30 W =10.0 {P/A)Avg + 17.9
.2
as 20
0
1- 10
0+-~~~~,_~~~~-r~~~~-+~~~~~
0 1 2 3 4
Average prestress, (P/A)Avg (MPa)
Figure 8.41 - Finite element results of failure load versus average prestress for
comer column connection model S4.
50
45
-ca
40
-
a.
.!II:
i
.2
35
30
25 --84-0.0 MPa
as 20
-+-84-1.0 MPa
0 15
1- 10 -lr- 84-2.1 MPa
5 --e-- 84-3.0 MPa
0
0 10 20 30 40 50 60
Reaction (Rx) (kN)
Figure 8.42 - Load versus x-direction horizontal reaction (Rx) at the column
support for different comer column models.
8-47
50
45
-as
a.
40
,-
~
35
30
as 25 --54-0.0 MPa
.2
"i 20 -+--54-1.0 MPa
0 15
..... 10 _.....54-2.1 MPa
Figure 8.43- Load versus y-direction horizontal reaction (Rv) at the column
support for different corner column models.
50
45
-a.
as
40
-,
~
as
.2
35
30
25 - - 54-0.0 MPa
Figure 8.44 -Load versus vertical reaction (Rz) at the column support for
different comer column models.
8-48
60
-
ca
50
-
~ 40
' tJ
ca 30
.2 -84-0.0 Gauge 1
"i 20 -+--84-1.0 Gauge 1
0 __..._ 84-2.1 Gauge 1
1- Yield limit
10 ---- 84-3.0 Gauge 1
0
0 1 2 3 4 5
Strain (E-3)
60 ~ 84-0.0 Gauge 2
-+--84-1.0 Gauge 2
-ca
50 -&- 84-2.1 Gauge 2
-e- 84-3.0 Gauge 2
-
~ 40
' tJ
ca 30
.2
"i 20
0
f- Yield limit
10
0
0 1 2 3 4
Strain (E-3)
Figure 8.45 - Load versus strain histories for x- direction top bars for
comer column connection with different amount applied
prestress.
8-49
60
-ca
50
-
~ 40
"C
ca 30
.2
-
ca
0
.....
20
10
--- 54-0.0 Gauge 3
~54-1.0 Gauge 3
.....,.__ 54-2.1 Gauge 3
--- 54-3.0 Gauge 3
0
0 0.5 1 1.5 2 2.5 3
Strain (E-3)
60
-
ca
50
-
~ 40
"C
ca 30
.2 -s- 54-0.0 Gauge 4
ca 20 ~54-1.0 Gauge 4
0 --&- 54-2.1 Gauge 4
.....
10 -e- 54-3.0 Gauge 4
0
0 0.5 1 1.5 2 2.5
Strain (E-3)
Figure 8.46 -Load versus strain histories for y- direction top bars for
corner column connection with different amount applied
prestress.
8-50
60
-
ca
50
-
~ 40
"C
ca 30
.2
'iii 20 - - 84-0.0 Gauge 5
0 ~84-1.0 Gauge 5
1-
10 .....,._ 84-2.1 Gauge 5
---*-- 84-3.0 Gauge 5
0
0 0.2 0.4 0.6 0.8 1 1.2
Strain (E-3)
60
-
ca
50
-
~ 40
"C
ca 30
.2
'iii 20 -e- 84-0.0 Gauge 6
0 --.-84-1.0 Gauge 6
1-
10 -tr- 84-2.1 Gauge 6
--e- 84-3.0 Gauge 6
0
0 0.2 0.4 0.6 0.8 1 1.2
Strain (E-3)
Figure 8.47 - Load versus strain histories for x-direction bottom bars of corner
column connections with different applied prestress.
8-51
element mesh is shown in Figures 8.1 and 8.2 for models S2-B and 83-B. Similarly the
finite element mesh for the model S4-B is as shown in Figures 8.3 and 8.4. With the
exception of a1 the material parameters used were as for the unbonded tendon
analyses. The slope of the softening branch of the concrete tension stress-strain law
(represented by a1) was modified to allow for additional tension stiffening effect due
The failure loads for the finite element analyses are compared in Table 8.9. The effect
of bonding the prestressing wires was to increase the failure load with the largest
relative increase for the specimen with higher levels of prestress. For specimens S3-B
and S4-B with 2.2 and 2.1 MPa of prestresses, respectively, the increase in the failure
load for the bonded compared to the unbonded prestress was 20 and 14 percent,
respectively.
The load versus deflection at point A on the slab (refer Figure 6.31a) is plotted in
Figure 8.48 for the bonded and unbonded analyses. The figure shows that overall
responses of the specimens are similar for both the bonded and unbonded models. The
load versus reactions are plotted in Figure 8.49. The distributions of the reactions are
Finally, the strain histories are compared for the bonded and unbonded models in
Figures 8.50 and 8.51 for the top and bottom reinforcement, respectively. As for the
deflections and reactions there is little difference between the bonded and unbonded
prestressing systems.
8-52
Table 8.9 - Comparison of failure loads for the unbonded and bonded finite
element models.
30
25
-
ca
-
~20
"C
~15
0
0 5 10 15 20 25
Deflection at A (mm)
30
25
-
-
:~. 20
"C
ca 15
.2
- - Rx (Unbonded)
'i
0 10 -e- Rx (Bonded)
1- -+- Rz (Unbonded)
5 -+- Rz (Bonded)
0
0 20 40 60 80 100 120 140
Reaction (kN)
Figure 8.49 - Load versus horizontal (Rx) and vertical (Rz) reactions at column
support for specimen S2.
8-54
30
25
(i
-
~ 20
"0
ftS
.2
15
Yield limit
a; --Gauge 1 (Unbonded)
'5 10 -a- Gauge 1 (Bonded)
1- -+-Gauge 2 (Unbonded)
~Gauge 2 (Bonded)
5 -+--Gauges 3 and 4 (Unbonded)
---&---Gauges 3 nad 4 (Bonded)
0
0 1 2 3 4
Strain (E-3)
30
_25
ftS
-
~20
"0
!15
-Gauges 5 and 6 (Unbonded)
"i 10
'5 -a- Gauges 5 and 6 (Bonded)
1-
5
0
0 0.5 1 1.5 2
Strain (E-3)
Figure 8.50- Load versus strain histories for top bars of specimen 82
(a)'x-direction reinforcement; and (b) y-direction
reinforcement.
8-55
30
25
-CIS
-
c.
,
~
20
CIS
.2
15
'iii
0 10
1- ----Gauges 7 and 8 (Unbonded}
5 -s- Gauges 7 and 8 (Bonded}
0
0 0.2 0.4 0.6 0.8 1
Strain (E-3)
30
-
CIS
25
-,
c.
~
CIS
.2
20
15
'iii
0 10 ----Gauges 9 and 10 (Unbonded}
1-
5 -s- Gauges 9 and 1o (Bonded}
0
0 0.2 0.4 0.6 0.8 1 1.2
Strain (E-3)
Figure 8.51 - Load versus strain histories for bottom bars of specimen S2
(a) x-direction reinforcement; and (b) y-direction
reinforcement.
8-56
The load versus deflection plot for the bonded model S3-B is compared with the
unbonded model S3 in Figure 8.52. The reactions at the column supports are shown in
Figure 8.53 and the strain histories for the top and the bottom reinforcement are
compared in Figures 8.54 and 8.55, respectively. The bonded and unbonded systems
have a similar overall response but with an increase in failure load for the bonded
The load versus deflection plots for points A and D of the slab (refer Figure 6.31 b) are
shown in Figure 8.56. The reactions are presented in Figure 8.57 and the strain
histories for the top and bottom main bars are compared in Figures 8.58 and 8.59,
respectively. As for slabs S2 and S3, the response of slab S4-B is similar to that of the
In this chapter the finite element model DIANA was used to analyse the experimental
test slab-column connections. The model was calibrated and verified in Chapters 3 to 5
of this thesis. The finite element model generally simulated the overall response of the
test specimens well although some variance occurred in some of the measured versus
specimens S 1 and S4 that occurred towards the end of the tests were not seen in the
finite element modelling. The fact that this redistribution was not also observed in the
experimental results of specimens S2 and S3 may indicate that at high loads some
locking of the pin base occurred. Alternatively, some load redistribution may have
40
35
Ci' 30
a.
c 25
, 20
"'
.2
-
'ii 15
0
1- 10
------ Unbonded
-a-Bonded
5
0
0 5 10 15 20 25 30 35
Deflection at A (mm)
40
35
Ci' 30
a.
c 25
,
"' 20 - Rx {Unbonded)
.2
-e- Rx (Bonded)
'ii 15
0 --+-- Rz (Unbonded)
1- 10 ~ Rz (Bonded)
5
0
0 50 100 150 200 250
Reaction (kN)
Figure 8.53 - Load versus horizontal (Rx) and vertical (Rz) reactions at column
support for specimen S3.
8-58
40
35
Ci' 30
c.
,c
as 20
25 Yield limit
.2 -Gauge 1 (Unbonded)
-
"i 15
0
1- 10
-e- Gauge 1 (Bonded)
-+-Gauge 2 (Unbonded)
~Gauge 2 (Bonded)
.......,.._Gauges 3 and 4 (Unbonded)
5 --b- Gauges 3 and 4 (Bonded)
0
0 1 2 3 4
Strain (E-3)
40
35
-c
c.as 30
,as 25
20
.2
"i 15 -Gauges 5 and 6 (Unbonded)
0
1- 10 -e- Gauges 5 and 6 (Bonded)
5
0
0 0.5 1 1.5 2 2.5
Strain (E-3)
Figure 8.54- Load versus strain histories for x- andy- direction top bars of
specimen S3 (a) x-direction reinforcement;.(b) y-direction
reinforcement.
8-59
40
35
"i' 30
c.
~ 25
'§ 20
'ii 15
'5
1- 10 ---Gauges 7 and 8 (Unbonded)
5 -e- Gauges 7 and 8 (Bonded)
o--~--~-L--~~~--~~--~~--~~~--~~
0 0.5 1 1.5
Strain {E-3}
40
35
"i' 30
c.
~ 25
"D
ca 20
.2
'ii 15 --Gauges 9 and 10 (Unbonded)
'5
1- 10 -e- Gauges 9 and 1o (Bonded)
5
0
0 0.5 1 1.5
Strain {E-3}
Figure 8.55 - Load versus strain histories for bottom bars of specimen S3
(a) x- direction reinforcement; and (b) y-direction
reinforcement.
8-60
50
45
40
-
"' 35
a.
~ 30
'a
25
"'
.S! - - A (Unbonded)
'ii 20 -e- A (Bonded)
0 15 -+- B (Unbonded)
1-
10 ~B(Bonded)
5
0
0 5 10 15 20 25
Deflections (mm)
50
45
-40
:.
~
35
-30 - - Rx (Unbonded)
i0 25 -e- Rx (Bonded)
- 20 -+- Ry (Unbonded)
'ii ~ Ry (Bonded)
0 15
1- 10 __.__ Rz {Unbonded)
--&- Rz (Bonded)
5
o~~~~~~~~~~-r~~~~~~,_~~~
0 20 40 60 80 100 120
Reaction (kN)
Figure 8.57- Load versus horizontal (Rx) and vertical (Rz) reactions at column
support for specimen S4.
8-61
50
45
-m
40
-
ll.
~
"D
m
.S!
35
30
25 ---Gauge 1 (Unbonded)
"i
20 -e- Gauge 1 (Bonded)
0 15 Yield limit --+-Gauge 2 ( unbonded)
1- 10 -+--Gauge 2 (Bonded)
5
0
0 1 2 3 4 5 6
Strain {E-3)
50
45
-m
40
35
-
ll.
~
"D
m
.S!
30
25 ---Gauge 3 (Unbonded)
20 -e- Gauge 3 (Bonded)
"i
0 15 --+-Gauge 4 (Unbonded)
1- -+--Gauge 4 (Bonded)
10
5
0
0 0.5 1 1.5 2 2.5
Strain {E-3)
Figure 8.58 - Load versus strain histories for top bars of specimen S4.
(a) x-direction reinforcement; and (b) y-direction reinforcement.
8-62
50
45
-40
:.
.lll::
35
-30
"tS -+-Gauge 5 (Unbonded)
~ 25
- 20 -e- Gauge 5 (Bonded)
'ii --Gauge 6 (Unbonded)
0 15
1- 10 --+-Gauge 6 (Bonded)
5
o.-~--~~~~,_~--~-L--~~~--~~--~~
0 0.5 1 1.5
Strain {E-3)
50
45
-40
:. 35
c"tS 30
~ 25
--Gauge 7 (Unbonded)
'ii 20
0 15 -e- Gauge 7 (Bonded)
1- -+-Gauge 8 (Unbonded)
10
--+-Gauge 8 (Bonded)
5
0
0 0.2 0.4 0.6 0.8 1
Strain {E-3)
Figure 8.59 - Load versus strain histories for bottom bars of specimen S4
(a) x-direction reinforcement; and (b) y-direction
reinforcement.
8-63
The difference between the measured and the predicted failure load was higher for the
specimen S3 compared to that of the other specimens and the stiffness degradation of
specimen S4 was greater than that produced by the finite element model. In these
specimens the closeness of ducting may have formed weakness in the slabs leading to
earlier cracking of the specimens than predicted by the finite element modelling.
The numerical models were used to study the effect of the level of prestress on the
comer column connections. Slabs were analysed for zero to 3.0 MPa of average
prestress. As for the experimentally observed edge connections, the level of prestress
was found to influence the load carrying capacity of the comer connections. For
3.0 MPa of prestress the increase in capacity of the connection was 85 percent above
Finally, the finite element models were used to access the differences between bonded
and unbonded prestressed slab systems. It was found that bonding the prestressing
tendons to the concrete did not influence the distribution of reactions, the overall
deflection response or the development of steel strains with load. Higher punching
failure loads, however, were obtained for the bonded prestressed slab system over that
of the unbonded system with the greatest load increase being for the slabs with the
9.1 Introduction
In this study the ultimate behaviour of flat plate slab-column connections has been
investigated with the focus being on prestressed concrete edge and comer slab-column
connections. The load path in these structures is three-dimensional and with the
non-linear material behaviour the response of the slabs under load is complex.
The research described in this thesis comprises of three parts; part one was an
investigation into the suitability of using finite elements to model punching shear
failure in flat plate slab-column connections using 3D solid brick elements. In part two
behaviour of flat plate-column connections. Part three dealt with the numerical
modelling of the test specimens with unbonded and bonded tendons. The effect of
The work undertaken in this thesis has identified a number of areas where the research
remains deficient. Recommendations for further study are given in this chapter.
In this study an investigation into the development of efficient finite element models
to analyses punching type failures was undertaken. Analyses of a plain concrete slab
panel under a pullout load were conducted using DIANA (1997) to study the effects of
mesh grading and the Gaussian quadrature used for numerical integration. The failure
loads obtained from the finite element model were compared with the ACI code model
9-2
and with the plasticity model of Nielsen et al. (1978) for anchorage pullout out
strengths. The ACI 318 was found to be conservative while the failure loads obtained
from the Nielsen et al. model compared well with the numerical results.
Numerical tests on the finite element mesh grading showed that the mesh grading has
a significant effect on the results of the analyses. It was observed that the mesh in the
critical region requires an aspect ratio not greater than 2 while away from the critical
zone the mesh aspect ratio may be increased. Two elements through the thickness of
the slab were found to be sufficient for modelling up to the peak load. Numerical
computational time and the disk space consumed for the analysis with 2 x 2 x 2
subject to a punching loading. Two distinct modes of failure were observed from the
model, that of flexure and that of punching shear. The primary failure mode is one of
flexure until a critical quantity of flexural reinforcement is used. Beyond this point
punching shear governs the failure. For increasing reinforcement content beyond the
critical point the failure load was independent of the quantity of flexural
reinforcement. While the effect of dowel action is not included in the numerical model
the results are consistent with the experimental observations of Dragosvic and
Sensitivity tests of the different parameters needed to model slabs using the 3D brick
element available in DIANA were carried out to provide an insight into the effect of
the parameters on the interpretation of the finite element results. The shear retention
factor f3 and the slope of the descending branch of the tension stress-strain law of
concrete (given by a1) were observed to have a significant influence on the results of
the finite element analyses. Other parameters such as friction angle, dilatancy angles,
slope of the descending branch of concrete stress-strain law were observed to have
only minor effects in the numerical results. The value of f3 = 0.2 for the shear retention
factor was found to give the most consistent numerical results for a wide range of
must be carefully chosen to represent the tension stiffening effect. The values of
different parameters recommended by Vermeer and de Borst (1984) and by van Mier
( 1987) were found to be suitable for the finite element modelling of punching shear
type problems.
tests are reported. The pullout tests were undertaken to further verify the shear
retention factor (/3) used in the numerical modelling. Six large square rectangular
prism blocks were tested with two different concrete strengths and with an anchor
plate embedded at three different depths. The test set-up was such that the boundaries
of the slabs were located far away from the anchor plates to elin_rinate boundary
behaviour generally well but with some exceptions. The shear retention factor f3 =0.2
was confirmed to be suitable for the numerical modelling of these problems.
9-4
tested, three specimens represented edge flat plate-column connections and one a
corner flat plate-column connection. The primary variable was the level of prestress in
the slab.
All four specimens tested failed in a punching shear mode with the failure surface
emanating from the intersection of the column and slab soffit. The prestress was found
to have an influence on the load carrying capacity of the connections. The load
carrying capacity increased with the increase in the intensity of the prestress applied to
In the edge column connections tested, the angle of the failure surface to the slab soffit
increased with increasing levels of prestress. This indicates that confinement of the
concrete across the failure surface has an effect on the angle of the failure surface. The
average failure angle was in the range of 17 to 26 degrees depending on the type of
connection and the intensity of the applied prestress. This is significantly less than the
45 degree failure angle assumed in the Australian and ACI code design models.
The test results were compared with the analytical shear to moment ratios calculated
from the,.equivalent frame and simplified code methods given in AS 3600 (1994) for
the load equal to the failure load measured in the tests. The shear force calculated from
these methods were close to the measured values but significant differences occurred
for the moments. Thus, the dimensional similitude model used to size the specimens
did not give the desired moment/shear ratio. This is possibly due to the zero vertical
9-5
displacement boundary condition used in the test at the second line of contraflexure.
Perhaps a better way of sizing of the laboratory specimens would be to use finite
element modelling to match the moment/shear ratios between the laboratory model
Shear force and moment data obtained from a range of experimental tests were
compared with predictions from five international codes of practice. There was a
significant degree of scatter between the test data and the predicted strengths.
Statistical analyses of the experimental failure load to the code model predictions were
undertaken for a wide range of test data. The British code model (BS 811 0) gave the
best fit for the mean of the test data but with a large scatter of results and low
confidence level as to the probability of failure being greater than the design strength.
The Australian code model (AS 3600) showed a poor correlation with the measured
data but with a similar standard of deviation to that of other codes of practice, thus,
giving the highest confidence level that the slab capacity will exceed the design
strength.
In Chapter 8, the results of numerical analyses on the specimens were presented. The
finite element model was able to simulate the behaviour of the specimens reasonably
carried out using the numerical modelling. The finite element studies further showed
On verification of the finite element models, the models were modified to investigate
the effect of bonding of the prestress and to further study the influence of the prestress
9-6
found to be marginally stronger than the unbonded connections. The finite element
model for the prestressed comer-column connections showed that prestressing in two
In summary, in this study the effect of prestress on the load carrying capacity of flat
plate-column connections was investigated. For both edge and comer column
linearly with increasing prestress. The influence of bonding the prestressing wires to
the slab was investigated using numerical modelling. The results showed that bonding
of the prestressing tendon gives a slight increase in the strength of the connection. The
relative increase in the capacity of the connection for the case of bonded tendons
From the research undertaken in this thesis a number of areas for further studies have
• numerical modelling
Details of further research needed in each of these areas are given below.
• The tension stiffening model needs rationalisation and the tension stiffening
and tension softening effects need to be separated. Modelling of tension
stiffening is essential in order to obtain the correct displacements in the slab but
it is not rational that elements along the punching failure surface have added
tension strength beyond cracking.
• Two 20-node brick elements through the thickness gave the satisfactory results
for studying punching shear strength in thin slabs. However, refinement of the
mesh through the thickness is needed if post failure behaviour is to be
investigated or for modelling thick slabs.
• fu this study one moment shear ratio in the connection was tested. Further tests
are needed varying the stiffness of the column and slab to obtain data over the
full range of moment shear interaction possible. Linear finite element analyses
may be used to develop these specimens.
9-8
theory dominates the literature. The fact that design models vary from country to
country is perhaps the best evidence that the fundamentals of load transfer between
slabs and their supporting columns are yet to be properly understood. For the models
investigated in this study the scatter in the test data to predicted results is significant.
transfer mechanism in slab-column connections. Only then can a more realistic design
model be developed.
REFERENCES
Aalami, B., (1972). "Moment-rotation relation between column and slab", ACI
Journal, Vol. 69, No.5, May, pp. 263-269.
Abdel Rahman, H. H., (1982). "Computational models for the nonlinear analysis of
reinforced concrete flexural concrete slab systems", Doctoral Thesis, University
College of Swansea, Swansea, United Kingdom.
ACI 318 (1963). "Building code requirements for reinforced concrete (ACI 318-63)",
American Concrete Institute, Farmington Hills, 144 pp.
ACI 318 (1971). "Building code requirements for reinforced concrete (ACI 318-71)",
American Concrete Institute, Detroit.
ACI 318 (1977). "Building code requirements for reinforced concrete (ACI 318-77)",
American Concrete Institute, Farmington Hills, 103 pp.
ACI 318 (1983). "Building code requirements for reinforced concrete (ACI 318-83)",
American Concrete Institute, Detroit.
ACI 318 (1989). "Building code requirements for reinforced concrete and commentary
(ACI 318-89/318R-89)", American Concrete Institute, Detroit, 353 pp.
ACI 318 (1995). "Building code requirements for structural concrete (ACI 318M-95)
and commentary (ACI 318RM-95)", American Concrete Institute, Farmington Hills,
371 pp.
ACI 318 (1999). "Building code requirements for structural concrete (ACI 318M-99)
and commentary (ACI 318RM-99)", American Concrete Institute, Michigan, 391 pp.
ACI 349 (1990). "Code requirements for nuclear safety related concrete structures",
American Concrete Institute, Michigan, 130 pp.
ACI 408.2R (1992). "State-of-the-art report on bond under cyclic loads", American
Concrete Institute, Detroit, 32 pp.
R-2
ACI Committee 421, (1992). "Abstract of: shear reinforcement for slabs", ACI
Structural Journal, Vol. 89, S56, September-October, pp. 587-589.
ACI Committee 435, (1991). "State of art report on control of two-way slab
deflections", ACI Structural Journal, Vol. 88, S53, July-August, pp. 501-514.
ACI-ASCE Committee 326, (1962). "Shear and diagonal tension: Part 3 -slabs and
footings", Journal of American Concrete Institute, Vol. 59, No.3, March, pp. 352-396.
Alexander, S. B., (1990). "Bond model for strength of slab-column joints", Doctoral
Thesis, University of Alberta, Canada, 228 pp.
Alexander, S. D. B., and Simmonds, S. H., (1992). "Bond model for concentric
punching shear", ACI Structural Journal, Vol. 89, No.3, May-June, pp. 325-334.
Alexander, S., (1999). "Strip design for punching shear", The Design of Two-Way
Slabs, ACI SP-183, pp. 161-179.
R-3
AS 1012.10 (1985). "Method for the determination of the indirect tensile strength of
concrete cylinders ('Brazil' or Splitting test)'', Standards Australia, Sydney, Australia.
AS 1472 (1991). "Carbon steel spring wire for mechanical springs", Standards
Australia, Sydney, Australia, 9 pp.
Barzegar, F., Behle, R., and Foroozesh, M., (1991). "Moment transfer and slab
effective widths in laterally loaded edge connections." ACI Structural Journal,
Vol. 88, No.5, September-October, pp. 615-623.
Bazant, Z. P., and Oh, B. P., (1983). "Spacing of cracks in reinforced concrete",
Journal of Structural Engineering, ASCE, Vol.109, No.9, September, pp. 2066-2085.
Bazant, Z. P., and Oh, B. P., (1984). "Numerical simulation of progressive fracture in
concrete structures: Recent developments", Proceedings, International Conference on
Computer Aided Analysis and Design of Concrete Structures, Pineridge Press,
Swansea, 1984, pp. 1-18.
Broms, C. E., (1990), "Shear reinforcement for deflection ductility of flat plates," ACI
Structural Journal, V. 87, No.6, November-December, pp. 696-705.
Brown, S., and Dilger, W. H., (1994). "Seismic response of flat plate column
connection", Annual Conference of the Canadian Society for Civil Engineering,
Winnipeg, June, pp. 388-397.
BS 8110 (1997). "Code of practice for design and construction (BS 8110: Part D",
British Standard Institution, London, 126 pp.
Bums, N. H., and Hemakom, R., (1977). "Test of a scale model post-tensioned flat
plate", Journal of Structural Division, ASCE, Vol. 103, No. ST6, June, pp. 1237-1255.
CEB-FIP Model Code for Concrete Structures 1990 (1993). Comite Euro-Intemational
du Beton/Federation International de la Precontrainte, Lausanne, Switzerland.
R-5
Chana, P., and Desai. S., (1993). "Design fro provision of resistance against punching
shear", Proceedings: Concrete 2000, Dundee University, September, Vol. 1,
pp. 815-825.
Chen, W.F., and Han, D. J., (1988). "Plasticity for structural engineers",
Springer-Verlag, New York, 606 pp.
Chiang, Y. J., and Lee, Y. L., (1994). "Evaluation of modeling accuracy of 20-node
solid elements by statistical factorial design", Computers & Structures, Vol. 52, No. 6,
pp. 1309- 1314.
Chiang, Y. J., and Tang, C., (1995). "Accuracy assessment to applying 20-node solid
elements to pressurized composite shells", Finite Elements in Analysis and Design,
Vol. 20, pp. 219-231.
Cooke, N., Park, R., and Young, P., (1981). "Flexural strength of prestressed concrete
member with unbonded tendons," Journal of the Prestressed Concrete Institute,
Vol. 26, No.6, November-December, pp. 52-80.
de Borst, R., and Nauta, P. (1985). "Non-orthogonal cracks in a smeared finite element
model", Engineering Computations, March, Vol. 2, pp. 35-46.
DeVries, R. A., Jirsa, J. 0., and Bashandy, T., (1999). "Anchorage capacity in
concrete of headed reinforcement with shallow embedments", ACI Structural Journal,
Vol. 96, No.5, September-October, pp. 728-736.
Dechka, D. C., and Dilger, W. H., (2000). "Full scale continuous slab-column frame
subjected to reverse cyclic loading", American Concrete Institute, 2000 Spring
Convention, San Diego, USA, March.
Dechka, D. C., Loov, R. E., and Dilger, W.H., (2000). "Prediction of punching shear
capacity by shear friction", American Concrete Institute, 2000 Spring Convention, San
Diego, USA, March.
DiStasio, J., and Van Buren, M.P., (1960). "Transfer of bending moment between flat
plate floor and column", Journal of the American Concrete Institute, Vol. 57, No. 3,
September, pp. 299-314.
DIANA-Release 6.2 (1997). TNO Building and Construction Research, Delft, The
Netherlands.
Dilger, W., and Cao, H., (1991). "Behaviour of slab-column connections under
reversed cyclic loading", Proceedings of the 2-nd International Conference of
High-Rise Buildings, China.
Dragosvic, AM. and van den Beukel, A., (1974), "Punching shear", HERON, vol. 20,
No. 2,48 pp.
Durrani, A. J., Du, Y., and Luo, Y. H., (1995). "Seismic resistance of nonductile
slab-column connections in existing flat slab buildings", ACI Structural Journal,
Vol. 92, No.4, July-August, pp. 479-487.
R-7
Elgabry, A. A., (1991). "Shear and moment transfer of concrete flat plates", Doctoral
Thesis, Department of Civil Engineering, The University of Calgary, Alberta, Canada,
February, 226 pp.
Elgabry, A. A., and Ghali, A., (1987). ''Tests on concrete slab-column connections
with stud-shear reinforcement subjected to shear-moment transfer", ACI Structural
Journal, Vol. 84, S46, September-October, pp.433-442.
Elgabry, A. A., and Ghali, A., (1990). "Design of stud-shear reinforcement for slabs",
ACI Structural Journal, Vol. 87, S35, May-June, pp. 350-361.
Elgabry, A. A., and Ghali, A., (1996). "Moment transfer by shear in slab-column
connections", ACI Structural Journal, Vol. 93, ST 2, March-April, pp. 187-196.
Elgabry, A. A., and Ghali, A., (1996). "Transfer of moments between columns and
slabs: Proposed code revisions", ACI Structural Journal, Vol. 93, No. 1,
January-February, pp. 56-61.
El-Salakawy, E. F., Polak, M. A., and Soliman, M. H., (1998). "Slab-column edge
connections subjected to high moments", Canadian Journal of Civil Engineering,
Vol. 25, pp. 526-538.
El-Salakawy, E. F., Polak, M. A., and Soliman, M. H., (1999). "Reinforced concrete
Slab-column edge connections with openings", ACI Structural Journal,
January-February, Vol. 96, No. 1, pp. 79-87.
Elstner, R. C., and Hognestad, E., (1956). "Shearing strength of reinforced concrete
slabs", Journal of American Concrete Institute, Vol. 28, No. 1, July, pp. 29-58.
Eurocode 2 (1992). "Design of concrete structures- Part 1: General rules and rules for
buildings", European Committee for Standardization, Brussels, 274 pp.
Falamaki, M., and Loo, Y. C., (1992). "Punching shear tests of half-scale reinforced
concrete flat-plate models with spandrel beams", ACI Structural Journal, Vol. 89, No.
3, May-June, pp. 263-271.
R-8
Foster, S. J., (1992). "The structural behaviour of reinforced concrete deep beams",
Doctoral Thesis, School of Civil Engineering, The University of New South Wales,
Sydney, Australia.
Foutch D. A., Gamble, W. L., and Sunidja, H., (1990). "Tests of post-tensioned
concrete slab-edge column connections", ACI Structural Journal, Vol. 87, No. 2,
March-April, pp. 167-179.
Furst, A., and Marti, P., (1997). "Robert Maillart's design approach for flat slabs",
ASCE Journal of Structural Engineering, Vol. 123, No.8, August, pp. 1102-1110.
Gardner, N. J., and Shao, X., (1996). "Punching shear of continuous flat reinforced
concrete slabs", ACI Structural Journal, Vol. 93, No.2, March-April, pp. 218-228.
Gerber, L. L., and Burns N. D., (1971). "Ultimate strength tests of post-tensioned flat
plates", PCI Journal, Vol. 16, No.6, November-December, pp. 40-58.
Gesund, H., and Kaushik, Y. P., (1970). "Yield line analysis of punching failures in
slabs" International Association for Bridge and Structural Engineering, Zurich,
Vol. 30-1, pp. 41-60.
Gilbert, R. 1., and Warner, R. F., (1978). ''Tension stiffening in reinforced concrete
slabs", ASCE Journal of Structural Division, Vol. 104, ST12, December,
pp. 1885-1900.
R-9
Gomes, R. B., and Regan, P.E., (1999). "Punching resistance of RC flat slabs with
shear reinforcement", ASCE Journal of Structural Engineering, June, Vol. 125, No. 6,
pp. 684-692.
Gomes, R., and Regan, P., (1999). "Punching strength of slab reinforced for shear with
offcuts of rolled steel !-section beams", Magazine of Concrete Research, April,
Vol. 51, No.2, pp. 121-129.
Guan, H., and Loo, Y. C., (1994). "Layered finite element method of analysis of
reinforced concrete flat plates", International Conference on Computational Methods
in Structural and Geotechnical Engineering, Hong Kong, December 12-14,
pp. 984-989.
Guan, H., and Loo, Y. C., (1995). "On the improvement of layer-type finite element
model", Proceeding Fifth East Asia-Pacific Conference on Structural Engineering and
Construction, Gold Coast, Australia, pp. 73-78.
Guan, H., and Loo, Y. C., (1996). "Numerical analysis and computer visualization of
punching shear failure of reinforced concrete flat plates: Part I Methodology", Third
Asian-Pacific Conference on Computational Mechanics, Seoul Korea, pp. 259-264.
Guan, H., and Loo, Y. C., (1996). "Numerical analysis and computer visualization of
punching shear failure of reinforced concrete flat plates: Part II Applications", Third
Asian-Pacific Conference on Computational Mechanics, Seoul Korea, pp. 265-270.
Hall, A. S., and Rangan, B. V., (1983). "Forces in the vicinity of edge columns in flat
slab floors", Magazine of Concrete Research (London), Vol. 35, No. 122, March,
pp. 19-26.
R-10
Hammill, N., and Ghali, A., (1994). "Punching shear resistance of comer slab-column
connections", ACI Structural Journal, Vol. 91, No. 6, November-December,
pp 697-707.
Hanson, N. W., and Hanson, J. M., (1968) "Shear and moment transfer between
concrete slabs and columns", Journal of the Portland Cement Association, Research
and Development Laboratories, Vol. 10, No. 1, January, pp. 2-16.
Hawkins, N. M., (1981). "Lateral load resistance of unbonded post-tensioned flat plate
construction", PCIJoumal, Vol. 26, No.1, January-February, pp. 94-116.
Hawkins, N. M., and Coreley, W. G., (1971). "Transfer of unbalanced moment and
shear from flat plate to column", Cracking, Deflection, and Ultimate Load of Concrete
Slab Systems, ACI Special Publication, SP- 30, American Concrete Institute, Detroit,
1971, pp. 147-176.
Hawkins, N. M., Bao, A. and Yamazaki, J., (1989). "Moment transfer from concrete
slab to columns", ACI Structural Journal, November-December, Vol. 86, No. 6,
pp. 705-716.
Hawkins, N. M., Bao, A., and Yamazaki, J., (1989). "Moment transfer from concrete
slabs to columns", ACI Structural Journal, Vol. 86, S70, November-December,
pp. 705-716.
Hellier, A.K., Sansalone, M., Carino, N.J., Stone, W. C., and Ingraffea, A. R., (1987).
"Finite-Element analysis of the pullout test using a nonlinear discrete cracking
approach", Cement, Concrete and Aggregates, American Society of Testing and
Materials, Vol. 9, No. 1, pp. 20-29.
Hemakom, R., (1975). "Strength and behavior of post-tensioned flat plates with
unbonded tendons", Doctoral Thesis, The University of Texas, Austin, December,
pp. 272.
R-11
Hognestad, E., (1951). "A study of combined bending and axial load in reinforced
concrete members", University of lllinois Engineering Experimental Station, Bulletin
Series No. 399, November, 128 pp.
Hu, H. H., and Schnobrich, W. C., (1988). "Nonlinear analyses of plane stress state
reinforced under short time monotonic loading", Civil Engineering Studies,
SRS No. 539, University of lllinois, Urbana, April.
Hueste, M. B. D., and Wight, J. K., (1999). "Modelling punching shear failures in
static and dynamic nonlinear analyses using a macro model", American Concrete
Institute, 2000 Spring Convention, San Diego, USA, March.
Hueste, M. B. D., and Wight, J. K., (1999). "Nonlinear punching shear failure model
for interior slab-column connections", ASCE Journal of Structural Engineering,
September, Vol. 125, No.9, pp. 997-1008.
Ichihashi, 1., Shibata, H., and Nomura, T., (1985). "Some test results of the strength of
embedded plates and or bolts in concrete", Transaction of the International Conference
on Structural Mechanics in Reactor Technology, Amsterdam, pp. 453-458.
Islam, S., and Park, R., (1976). "Tests on slab-column connections with shear and
unbalanced flexure", Journal of Structural Division, ASCE, Vol. 102, No. ST3, March,
pp. 549-568.
Jiang, D. H., and Zhou, K. R, (1991). "Imitation of punching failure for reinforced
connect slabs in finite elements method", Proceeding of Asian Pacific Computational
Conference on Mechanics, December, 221-226 pp.
Khwaounjoo, Y. R., Foster, S. J., and Gilbert, R. I., (1999). "Influence of boundary
conditions on punching shear behaviour of flat plate-column connections", 16th
Australian Conference on the Mechanics of Structures and Material, ACMSM16,
Sydney, December, pp. 145-150.
Kinnunen, S., and Nylander, H., (1960). "Punching of concrete slabs without shear
reinforcement", Nr. 158, Institutionen for Byggnadsstatik, Kungliga Tekniska
Hogskolans, Stockholm, 112 pp.
Klinger, R. E., and Mendonca, J. A., (1982). "Tensile capacity of short anchor bolts
and welded studs: a literature review", ACI Journal, Vol. 79, No. 27, July-August,
pp. 270-279.
Kuang, J. S., and Morley, C. T., (1992). "Punching shear behavior of restrained
reinforced concrete slabs", ACI Structural Journal, Vol. 89, No. 1, January-February,
pp. 13-19.
Lamb, J. W., (1984). "Moment transfer and joint stiffness in reinforced concrete flat
plate-column connections", Department of Civil Engineering, Royal Military College
of Canada, Canada, 184 pp.
Lew, H. S., Carino, N. H., Fattal, S. G., and Batts, M. E., (1982). "Investigation of
construction failure of Harbour Cay Condominium in Cocoa Beach, Florida", Building
Science Series No. 145, National Bureau of Standards, Washington, D. C., August,
135 pp.
Lim, F. K., and Rangan, B. V., (1994). "Strength of concrete slabs with stud shear
reinforcement in the vicinity of edge corner columns", Research Report No. 1/94,
School of Civil Engineering, Curtin University of Technology, April, 180 pp.
R-13
Lim, F. K., and Rangan, B. V., (1995). "Studies on concrete slabs with stud shear
reinforcement in vicinity of edge and comer columns", ACI Structural Journal,
Vol. 92, No.5, September-October, pp. 515-525.
Lin, C. S., (1973). "Nonlinear analysis of reinforced concrete slabs and shells",
Doctoral Thesis, UC_SESM 73-7, University of California at Berkeley.
Lin, C., and Scordelis, A. C., (1975). "Nonlinear analysis of RC shells of general
form", Proceedings ASCE, Vol. 101, ST3, March, pp. 523-538.
Lin, T. Y., and Bums, N.H., (1981). "Design of prestressed concrete structures", John
Wiley & Sons, 646 pp.
Long, A. E., "Punching failure of reinforced concrete slabs", Doctoral Thesis, Faculty
of Applied Science and Technology, Queen's University of Belfast, May 1967,
210pp.
Loo, Y. C., and Falamaki, M., (1992). "Punching shear strength of reinforced concrete
flat plates with spandrel beams", ACI Structural Journal, Vol. 89, S36, July-August,
pp. 375-383.
Loo, Y. C., and Guan, H., (1997). "Cracking and punching shear failure analysis of
RC flat plates", Journal of Structural Engineering, Vol. 123, No. 10, October,
pp. 1321-1330.
Loo, Y. C., Chiang, C. L., and Guan, H., (1995). "Performance of prediction method
for punching shear strength of post-tensioned concrete flat plates", 14th Australian
Conference on the Mechanics of Structures and Materials, Hobart, Australia,
pp. 130-135.
Luo, Y. H., and Durrani, A. J., (1995). "Equivalent beam model for flat-slab buildings-
part ii: exterior connections", ACI Structural Journal, Vol. 92, No. 2, March-April,
pp. 250-257.
R-14
Luo, Y. H., Durrani, A. J., Bai, S., and Yuan, J., (1994). "Study of reinforced detail of
tension bars in frame comer connections," ACI Structural Journal, Vol. 91, No. 48,
pp. 486-496.
Luo, Y. H., Durrani, A., and Conte, J., (1995). "Seismic reliability assessment of
existing r/c flat-slab buildings", Journal of Structural Engineering, Vol. 121, No. 10,
October, pp. 1522-1530.
Mansur, M. A., and Rangan, B. V., (1978). "Torsion in spandrel beams", Journal of
Structural Division, ASCE, Vol. 104, ST7, pp. 1061-1075.
Mast, P. E., (1970). "Plate stresses at columns near the free edge", ACI Journal,
Vol. 67, No. 11, November, pp. 898-902.
Materson, D. M., and Long, A. E., (1974). "The punching strength of slabs, a flexural
approach using finite elements", Shear in Reinforced Concrete, ACI Special
Publication, SP-42, American Concrete Institute, Detroit, pp. 747-768.
Megally, S., and Ghali, A., (1994). "Design considerations for slab-column
connections in seismic zones", ACI Structural Journal, Vol. 91, No. 3, May-June
1994, pp. 303-314.
Megally, S., and Ghali, A., (1996). "Nonlinear analysis of moment transfer between
columns and slabs", 1st Structural Specialty Conference, May 29-June 1, Edmonton,
Alberta, Canada, pp. 321-332.
R-15
Megally, S. and Ghali, A., (1999). "Design for punching shear in concrete", The
Design of Two-Way Slabs, ACI SP-183, pp.37-66.
Megally, S., and Ghali, A., (2000). "Punching of concrete slabs due to column
moment transfer", ASCE Journal of Structural Engineering, February, Vol. 126, No.2,
pp. 180-189.
Megally, S., and Ghali, A., (2000). "Avoiding punching failure of slabs in
earthquakes", American Concrete Institute, 2000 Spring Convention, San Diego, USA,
March.
Mehrain, M., and Aalami, B., (1974). "Rotational stiffness of concrete slabs", ACI
Journal, Vol. 71, September, pp. 429-435.
Moe, J., (1961). "Shearing strength of reinforced concrete slabs and footings under
concentrated loads", Bulletin No. D47, Portland Cement Association, Research and
Development Laboratories, Skokie, lllinois, April, pp. 1-130.
Menetrey P., (1998). "Relationship between flexural and punching failure", ACI
Structural Journal, July-August, Vol. 95, No.4, pp. 413-419.
Mortin, J.D., and Ghali, A., (1991) "Connection of flat plates to edge columns," ACI
Structural Journal, Vol. 88, No.2, March-April, pp. 191-198.
Nielsen, M. P., Braestrup, M. W., Jensen, B. C., and Bach, F., (1978). "Concrete
plasticity", Specialpublikation udgivet af Dansk Selskab, October, 129 pp.
Noor, A. K., and Babu.ska, 1., (1987). "Quality assessment and control of finite
element solutions", Finite Elements in Analysis and Design 3, pp. 1-26.
Ockleston, A. J., (1955). "Load test on a three storey reinforced concrete building in
Johannesburg", Structural Engineer, London, Vol. 33, No. 10, October, pp. 304-322.
R-16
Ottosen, N. S., (1977). "A failure criterion for concrete," Journal of Engineering
Mechanics Division, ASCE, Vol. 103, No.4, pp. 527-535.
Ozbolt, J., Vocke, H., and Eligehausen, R., (2000). ''Three-dimensional numerical
analysis of punching failure", American Concrete Institute, 2000 Spring Convention,
San Diego, USA, March.
Park, H., (1999). "Numerical study on RC flat plates subjected to combined axial and
transverse load", Structural Engineering and Mechanics, Vol. 8, No.2, pp.137-150.
Park, R., and Gamble, W. L., (2000). "Reinforced concrete slabs", John Wiley &
Sons, Inc., New York, 716 pp.
Park, R., and Islam, S., (1976). "Strength of slab-column connections with shear and
unbalanced flexure", Journal of Structural Division, ASCE, Vol. 102, No. ST9,
September, pp. 1879-1991.
Park, R., and Islam, S., (1976). "Strength of slab-column connections with shear and
unbalanced flexure", Journal of Structural Diyision, ASCE, Vol. 102, No. ST9,
September, pp. 1879-1901.
R-17
Pavlovic, M. N., and Poulton, S. M., (1985). "On the computation of slab effective
widths", ASCE, Journal of Structure Division, Vol. 111, No.2, February, pp. 363-377.
Pecknold, D. V., (1975). "Slab effective width for equivalent frame analysis", ACI
Journal, Vol. 72, No.4, April, pp. 135-137.
Polak, M. A., (1998). "Modelling punching shear of reinforced concrete slabs using
layered finite elements", ACI Structural Journal, January-February, Vol. 95, No. 1,
pp. 71-80.
Prak:hya, G. K. V., and Morley, C. T., (1990). "Tension stiffening and moment-
curvature relation of reinforced concrete elements", ACI Structural Journal, Vol. 87,
No.5, September-October, pp. 597-605.
Rangan, B. V., (1987). "Punching shear strength of reinforced concrete slabs", Civil
Engineering Transaction Institution of Engineers, Australia, Civil Engineering,
Vol. CE29, No.2, April, pp. 71-78.
Rangan, B. V., (1990). "Punching shear design in the new Australian standard for
concrete structures", ACI Structural Journal, Vol. 87, S15, March-April, pp. 140-166.
Rangan, B. V., (1990). "Tests on slabs in the vicinity of edge columns", ACI
Structural Journal, Vol. 87, No.6, November- December, pp. 623-629.
Rangan, B. V., and Hall, A. S., (1983). "Forces in the vicinity of edge columns in flat
plate floors (VI-tests on R. C. models)", UNCIV Report No. R-203, The University of
New South Wales, Sydney, Australia, January, 240 pp.
Rangan, B. V., and Hall, A. S., (1983). "Moment and shear transfer between slab and
edge column", ACI Journal, May-June, Vol. 80, No.3, pp. 183-191.
Rangan, B. V., and Lim, F. K., (1992). "Test on concrete slabs with stud shear
reinforcement in the vicinity of edge columns", Research Report No. 1192, School of
R-18
Rankin, G. I. B., and Long, A. E., (1988). "A rational approach to punching at interior
slab-Column Connections", Proceedings of American Concrete Institute Annual
Convention, Florida, March, pp. 45-55.
Regan, P. E., (1981). "Behaviour of reinforced concrete flat slabs", CIRIA Report 89,
February, London, 89 pp.
Robertson, I. N., and Durrani, A. J., (1991). "Gravity load effect on seismic behavior
of exterior slab-column connections", ACI Structural Journal, Vol. 88, S28, May-June,
pp. 255-267.
Robertson, I. N., and Durrani, A. J., (1992). "Gravity load effect on seismic behavior
of interior slab-column connections", ACI Structural Journal, Vol. 89, No. 1,
January-February, pp. 37-45.
Roddeman, D. G., and Jansen L. F., (1993). "An a priori geometry check for a single
isoparametric finite element", Computers & Structures, Vol. 47, No. 1, pp. 69-72.
Rots, J. G., (1985). "Bond-Slip simulation using smeared cracks and/or interface
element", Delft University of Technology, Netherlands, November, 56 pp.
R-19
Rots, J. G., Nauta, P., and Kusters, G. M.A., and Blaauwendraad, J. (1985) "Smeared
crack approach and fracture localization in concrete", HERON, Vol. 30, No. 1, 48 pp.
Scordelis, A. C., Lin, T. Y., and Itaya, R., (1959). "Behaviour of a continuous slab
prestressed in two directions", Journal of the American Concrete Institute, Vol. 31,
No.6, September, pp. 441-459.
Scordelis, A. C., Lin, T. Y., and May, H. R., (1958). "Shearing strength of prestressed
concrete lift slabs", Journal of the American Concrete Institute, Vol. 30, No. 4,
October, pp. 458-506.
Scordelis, A. C., Pister, K. S., and Lin, T. Y., (1956). "Strength of a concrete slab
prestressed in two directions", Journal of the American Concrete Institute, Vol. 28,
No.3, September, pp. 241-256.
Sih, G. C., and DiTommaso, A., Eds. (1985). "Fracture mechanics of concrete:
Structural application and numerical calculation", Martinus Nijhoff Publishers,
Dordrecht, 276 pp.
R-20
Simmonds, S. H., and Alexander, D. B.S., (1987). "Truss model for edge column-slab
connections", ACI Structural Journal, Vol. 84, July-August 1987, pp. 296-302.
Smith, S. W., and Bums N. D., (1974). "Post-tensioned flat plate to column
connection behavior", PCI Journal, Vol. 19, No.3, May-June, pp. 74-91.
Stamenkovic, A., and Chapman, J. C., (1974). "Local strength at column heads in flat
slabs subjected to a combined vertical and horizontal loadings", Proceedings,
Institution of Civil Engineers, London, Part 2, June, pp. 205-232.
Stone, W. C., and Carino, N. J., (1983). "Deformation and failure in large-scale
pullout tests", ACI Journal, Vol. 80, No. 46, November-December, pp. 501-513.
Suidan, M., and Schnobrich, W. C., (1973). "Finite element analysis of reinforced
concrete", ASCE Proceedings, Journal of Structural division, Vol. 99, STlO, October,
pp. 2109- 2121.
Sunidja, H., Foutch, D. A., and Gamble, W. L., (1982). "Response of prestressed
concrete plate-edge column connections", Civil Engineering Studies, Structural
Research Series No. 498, University of Dlinois, Urbana, and March, 232 pp.
Timoshenko, S. P., and Woinowsky-Krieger, S., (1959). "Theory of plates and shells",
McGraw-Hill International Book Company, New York, 580 pp.
van den Beukel, A., (1976). "Punching shear at inner, edge, and comer columns",
HERON, vol. 21, No.3, 30 pp.
Vecchio, F. J., and Collins, M.P., (1990). "Investigating the collapse of a warehouse",
Concrete International: Design and Construction, Vol. 12, No. 3, March 1990,
pp. 72-78.
Vermeer, P. A., and de Borst, R., (1984). "Non-associated plasticity for soils, concrete
and rock", HERON, Vol. 29, No.3, 65 pp.
Walker, P.R., and Regan, P. E., (1987). "Comer column-slab connections in concrete
flat plates", ASCE Journal of Structural Engineering, April, Vol. 113, No. 4,
pp. 704-720.
Wong, Y. C. and Coull, A., (1980). "Effective slab stiffness in flat plate structures",
Proceedings of Institution of Civil Engineers, Part 2, 69, London, September,
pp. 721-735.
Yamada, T., Nanni, A., and Endo, K., (1992). "Punching shear resistance of flat slabs:
influence of reinforcement type and ratio", ACI Structural Journal, Vol. 88, No. 4,
September-October, pp. 555-563.
Yener, M., (1994). "Overview and progressive finite element analysis of pullout tests",
ACI Journal, Vol. 91, No. S6, January-February, pp. 49-58.
Yener, M., and Li, G. C., (1991). "Progressive finite element fracture analysis of
pullout concrete", Journal of Structural Engineering, Vol. 117, No. 8, Auglist,
pp. 2351-2371.
Zaghlool, E. R. F., (1971). "Strength and behaviour of comer and edge column-slab
connections in reinforced concrete flat plates", Doctoral Thesis, Department of Civil
Engineering. University of Calgary, Alberta, 366 pp.
R-22
Zaghlool, R. F., de Paiva, H. A. R., and Glockner, P. G., (1970). "Tests of reinforced
concrete flat plate floors", Proceedings of American Society of Civil Engineers,
Journal of Structural Division, March, pp. 487-507.
Zerbe, H. E., and Durrani, A. H., (1990). "Seismic response of connections in two-bay
reinforced concrete frame subassemblies with a floor slab", ACI Structural Journal,
Vol. 87, S40, July-August, pp. 406-415.
Zhou, K. R., and Jiang, D. H., (1991). "Imitation of punching failure for reinforced
connect slabs in finite elements method", Proceeding of Asian Pacific Computational
Conference on Mechanics, December, pp. 221-226.
APPENDIX A- DESIGN OF PROTOTYPE AND TEST SLABS
The flat plate-column structural system shown in Figure A.l was considered as a
prototype structure for the design of the experimental specimens. A one third scale
model (see Figure A2) was generated from the prototype structural system shown in
Figure A 1 to develop the test specimens. The columns were 600 mm square and the
slab 300 mm thick in the prototype structure and were reduced to 200 mm square
A live load of 3 kPa and a total dead load of 8.2 kPa were considered for the design of
the prototype structure. The dead load included the self weight of the slab plus a 1 kPa
superimposed dead load. In the design the load factors of 1.25 and 1.5 were used for
the dead load and the live load, respectively. The design load of 14.75 kPa was used
A concrete of strength of 25 MPa was used for both the prototype and model
structures and the yield strengths of the conventional reinforcing bars and prestressing
tendons were assumed to be 450 and 1250 MPa, respectively. The modulus of
elasticity of the concrete was taken as 20 GPa and for the conventional and
c c c
c c c
c c c
J
L12000 .I I 12000 I I I 12000 I I I 12ooo----J
-
4( po
-
f'U
l:: -
bo
:--
Portion
stuoy
unde~
__j
4( po
- / \v ~
SECTION 1-1
D D D D
Region un+r
s~
D D D
o
L_~ All columns 200x200
D D D D
1
L! 0 77-T7
Region und r
study"
J
~000 4000 I. 1000
1~ ~0
?U
[
I~
Portion unde~
jStUOy __j
1 :JO
1:' ;>0
v ~
SECTION 1-1
The design stress resultants for the prototype and the model structures were
determined using the simplified method outlined in AS 3600-1994 and using the
equivalent frame method for the model frames. The results from both methods were
compared and the design was carried out considering the worst case from these
analyses. In the model slabs, the flexural moment reinforcement was increased by a
Two dimensional frame analyses were carried out for the internal and edge frames in
the x- and y-directions of the model structure. The width of the internal frame was
taken equal to the centre to centre distance of the adjacent panels, while for the edge
frame the effective width was taken from the edge of the slab to the centre of the
panel. The bending moment diagrams for the design regions of these frames are shown
in Figure A3. In Figure A4 the bending moment diagrams are given for the analyses
using the simplified design method of AS 3600. The bending moments shown in these
figures are the total bending moments and were distributed to column and middle
The specimens were developed with the slabs extending to second line of
contraflexure as discussed in Section 3.2 of this thesis. The length of the column for
the specimen slab-column connections (see Figure A5) was set to give the correct
stiffness to the model slab. The flexural stiffness of the column of the 1/3 scale
4
slab-column connection is 2 E4nim , where Em, Im and Lm are the modulus of
elasticity, moment of inertia and length of the column for the model structure. The
flexural stiffness of the column of the slab-column connection for the test specimen is
3
Es Is where E8 , Is and Ls are modulus of elasticity, moment of inertia and length of
Ls
the column for the test specimen, respectively. Equating these stiffnesses, the required
(Al)
A-5
48
Note:
10.5 Units:
BM (kN)
SF (leN)
"~: : : : : :"'
24
::-----=:::::::::: I
~a...--------------41000f-------------
BMD FOR X-PIRECTION EXTERNAL FRAME
Key:
End with column
I in the specimen
Y QIRfCTIQN INTERNAl SpAN EDGE FRAME Not to the scale
Units:
BM {KNm)
SF (kN)
Note: Total bending moments are shown. These are distributed to column and middle
strips.
Figure A.3 - Bending moment diagrams from the analyses of the model frame.
A-6
Note:
~----------~ooo-----------~
I End with column
in the specimen
Diagrams - Not to the scale
Units:
8MD FOR EDGE COLUMN STRIP
BM (kNm)
SF (kN)
~-----------400o-----------~
74.7kNm
24.
\
Key:
End with column
I In the specimen
Y-Q!RFC]ON INJFRNA! FQGE CO! liMN SJBIP Not to the scale
Units:
BM (kNm)
SF (kN)
Figure A.4 - Bending moment diagrams for the model frames from the simplified
design method (AS 3600 - 1994).
A-7
600mmx600mm
column
200mmx200mm
column OO
1
[
T
Prototype slab-column
Using Lm = 1.35 min the equationAl gives Ls = 0.51 m. Considering laboratory and
other constraints the length of the column for the laboratory specimens was taken as
Ls =0.60m.
The fmal geometry of the test specimens and the positions of the conventional and
reinforcing steel ratio in the design slabs came to 1.1 percent which was then increased
other investigators. The table shows that the quantity of steel used in this study is
typical of that used in studies on the punching shear strength of plate slabs.
A-8
Table A.l • Reinforcement contents of some typical flat plate slabs tested
Hammill and Ghali (1994) NH1. NH2 and NH4 Comer 1.47
#Note: Maintained this reinforcement content within a distance from the face of the
column equal to the effective depth of the slab.
APPENDIX 8 - RELAXATION TEST AND PRESTRESS LOSSES
Two types of prestressing losses were considered in the tests, these being the loss of
prestress caused by the anchorage slip and loss due to relaxation of the tendons. Tests
for evaluating the prestress losses were carried out for spans of 3.82 m for the edge
column specimens and 2.44 m for the comer column specimen. The set up used for
these tests is shown in Figure B.l. A load cell was placed between the anchorage plate
and the barrel-wedge system to regularly monitor load in the prestressing wire. The
output terminal of the load cell was connected to an HBM amplifier and a data
acquisition unit.
To test for the losses the load was applied gradually and the wedges were tapped in the
barrel gently and uniformly. Once the load reached the desired level, the wedges were
taped to fit tight in the barrel, the pump was released and the wire was cut. The loss
due to anchorage draw in was obtained from the change in the load measured at the
cutting of the wire from the jack. Three tests were undertaken for each span. Load
readings from the amplifier were taken regularly for two days to determine the
relaxation losses. The averages of the three tests for each series are shown in
Table B.l. The average room temperature during the test was recorded to be 20
degrees with the tests undertaken in the same environment as the test slabs.
The load to be applied to the prestressing wires in each test was determined by adding
the load given in Table B.2 to the required effective prestressing force. Further, a load
of 1 kN was added to the first two cables prestressed in each slab to compensate for
plate
Anchorage Plate
(100x100mm)
\_Barrel and
Dead end Jacking en d wedge
I 3.82 5 4 6
IT 2.44 8 2 3
Additioanlload (kN) 10 7 14
APPENDIX C- HORIZONTAL REACTION FOR SPECIMEN S1
In the post analysis of the test data of specimen S1 a problem with the experimental
set-up for the measuring of the horizontal reaction became evident. The support details
and support reactions for specimen S 1 are detailed in Figure C.1. In the test
arrangement the connection of the load cell measuring the vertical load allowed for an
accidental transfer of horizontal load into the testing frame and this component was
unmeasured.
The horizontal reaction Rx was split into two components Rh1 and Rh2 , as shown in
Figure C.1c. The arrangement of the load cells measured Rh1 only and Rh2 was not
measured. Using a linear finite element model the component of the horizontal
reaction Rh2 is determined. The mesh used in the finite element analyses consisted of
20-node 3D brick elements and is shown in Figure C.2. The finite element program
DIANA (1997) was used for the analysis with the elastic modulus taken as
Ec = 20.0 GPa.
From the finite element analysis the ratio of the vertical to horizontal reaction was
calculated and the missing component determined. The finite element analysis gave a
ratio of the total horizontal reaction to vertical reaction of 0.5. The ratio of the
measured component of the horizontal load Rht to the vertical reaction Rz is shown in
Figure C.3 up to the point of punching failure. The measured ratio is almost constant
at 0.28 and supports the calculated correction factor being a constant. Further, in
Figure 7.3 the corrected horizontal reaction for specimen S 1 is plotted with the
measured horizontal reactions for specimens S2 and S3, that were of similar geometry.
The figure shows that the corrections based on the finite element model appear to be
reasonable.
C-2
I")
0
<0
I")
0
"'
R,
Figure C.l - Support system showing the source of error and actual distribution
of horizontal reaction.
C-3
X
y
Plane of symmetry
0.35
0.3
0.25
-N
a:,.. 0.2
.c
a: 0.15
0.1
0.05
0
0 20 40 60 80 100 120
Rz(kN)
D.1 Introduction
Different codes use different models for the design of punching shear. In this appendix
a brief review of these models is presented. Altogether five codes were considered in
• ACI 318-1999
• AS 3600-1994
• BS 8110-1997
• Eurocode 2-1992.
Australian and CEB-FIP model codes and the test results were compared in Chapter 7
The eccentric shear model for the design of flat plate-column connections given in
ACI 318 (1999) is based on many studies including those by DiStasio and Van
Buren (1960), Moe (1961) and Hanson and Hanson (1968). The ACI code defines the
critical section as the section with its perimeter b0 to be minimum with the perimeter
taken as not closure than d/2 to the columns, where b0 is the perimeter of the critical
D-2
section and d is the effective depth of the slab. The critical perimeters for edge and
· comer column connections are as shown in Figures D.l and D.2, respectively.
For exterior rectangular columns in a two-way flat slabs, where any portion of the
column cross section is closer than 4 times the slab thickness from the free edge, the
(D.la)
(D.1c)
where f3c is the ratio of long side dimension to the short side dimension of the column;
as is 30 for edge columns and 20 for the comer columns; and J; is the characteristic
non-prestressed members provided that one or more of the following conditions apply:
• The minimum distance from the column to the discontinuous edge is less than
• J; s; 35.0MPa.
• 0.9 ~fpc~ 3.5 MPa, where /pc is the effective compressive stress in the
Column section
Mux
ex
(C1 +d/2)
1----Free edge
Figure D.l - Critical section and location of moments for edge column
connections.
section
uy
Y----x cc2 *-d/2)
Free edges
(C1 +d/2)
Figure D.2 - Critical section and location of moments for corner column
connections.
D-4
The ACI code takes a fraction of the unbalanced moment, 'W Mu. of the total factored
unbalanced moment, Mu, as being transferred to the slab by flexure. The remaining
shown in Figures D.3 and D.4 for edge column connections and in Figure D.5 for
corner column connections. The total shear stress developed at the critical section is
the sum of the shear stress developed due to part of the factored unbalanced moment,
Mu, and the uniform shear stress due to the factored shear force, Vu. The maximum
(D.2)
where A is the area of the critical section; ex and ey are the eccentricities of moment
Figures D.l and D.2 for edge and corner column connections; and lx and ly are
properties of the critical section analogous to polar moment of inertia about the y and
x directions, respectively.
The terms in equation D.2 are as defined in Figures D.l and D.2 and equations D.3
and D.4 for edge and corner connections, respectively. The moments Mux and Muy are
the moment Mu calculated at the centroid of the critical section in the x and y
directions, respectively.
(D.3a)
D-5
~y Shear stresses on
critical section due Vu
/ and Mua
Span
be in
~X
Note:
Mucx =Mua +Mub
Critical section
Free
Vu _ Vu + l'x Mux ex
- A Jx
Figure D.4 - Shear stress distribution due to Vu. and "{vy Muy for a typical column
(Shown for the edge column).
D-7
Shear stresses on
critical section due
Mux1 and Muy1
Free edge
"--x
Note:
Mucx =Mux1 +Mux2
Mucy =Muy1 +Muy2 Vu ,Mucx ,Mucy :Forces to be transmitted
by slab-column connection
A=du (D.3b)
(D.3c)
(D.3d)
[(CI+d /2)d3 +(Ct +d /2)3d] +(C +d)d 2 +2(C +d/2)d[(q +d/2) ex]2 (D.3e)
Jx 2 ex 1
6 2
(D.3t)
1
Yvx =1- 2 Ct+d/2 (D.3g)
1+3 C2+d
(D.3h)
(D.4a)
A=du (D.4b)
(D.4c)
(D.4d)
0-9
1
rvx = 1- 1+ -2 q +d t2
I----,,----::-
(0.4g)
3 Cz+d
1
Yvy =1- --=:2---;=== (0.4h)
1+- Cz+d
3 Ct+d/2
For rectangular column connections without shear reinforcement in the slab, the
governing equation is
(0.5)
where Vu is the maximum factored shear stress, ifJ is the strength reduction factor and
Av = ifJVc (0.6)
.,., n bod
Vu Mu
--+-..!:!:..- 1 (0.7a)
if' Vuo ifJ M uo
(0.7b)
In equations D.7a and D.7b, Vu and Mu are the strength limit state shear force and
moment at the centroid of the column and Vuo and Muo are, respectively, the shear
capacity for zero moment and moment capacity for zero shear. The suffixes X and Y
denote the directions of the moment being considered. The relationships for Vuo and
Muo are given in equation 0.8 below. Since in ACI 318 (1999) the moments are
D-10
calculated at the centroid of the critical section, the terms in the moment-shear
interaction equation D.7a are modified to reflect the eccentricities x andy about they-
and x- axes, respectively. In this study all the code comparisons are made by
considering the moment and shear at the centroid of the slab-column intersection.
f.J:/3
Vuo (D.8a)
(-1 _Yv ex X y)
Yry 'y
ly
Ac lx
f.J:/3
Muo (D.8b)
rvx exflx
[M uo ]y = f.J:/3jJ (D.8c)
rvx eX X
Finally, it is worth highlighting that the moments [Mu]x and [Mu]r are the moments at
the centroid of the column about y and x-axes, respectively, and Mux and Muy are the
moments calculated at the centroid of the critical section in the x and y directions,
respective! y.
D.3 AS 3600-1994
concrete slabs with rectangular columns and without shear reinforcement in the slab is
limited by
(0.9)
where Vuo is the ultimate shear strength for zero moment and is given by
(D. lOa)
D-11
with
and where v* is the factored design shear force; l/J is the strength reduction factor; u is
the length of the critical shear perimeter defined by a line parallel to the edge of the
column at a distance of d/2 from the column face; d is the effective depth of the slab
averaged around u; a is the linear dimension of the critical shear perimeter measured
parallel to the direction of M/; M/ and v* are, respectively, the unbalanced bending
moment in the direction being considered and shear force transferred at the centroid of
the column at the collapse limit state; f3h is the ratio of the largest dimension of the
support (Y) to the overall dimension (X) at right angles to Y ( f3h =Vx ,Y ~ X ;refer
Figure D.6); rJcp is the average intensity of effective prestress in concrete in MPa and
(D.lla)
Taking M uo =Vuo Safu and substituting into equation D.lla gives the interaction
equation
v* M*
--+ v 1 (D.llb)
l/J Vuo l/J M uo
Boundary of effective
area of column
r
Boundary of effective
area of column
r-----l~/2
[
r- ---- d/2
I ~--..,...-,...--t---' I
..
I
I
I I
I - - - Critical shear ----l . :
I
l \ \
:
I
perimeter :
I ..
<I
I
I
I
y
\ I I I
L --- J
X ~2
_v
M*_ _ [ _ v_
Muo Muo X
M*] +[M*] _v_
Muo y
(D.11c)
Substituting equation D.lOa in the equation forM uo we write the ultimate moment
(D.12)
The shear perimeter considered in BS 8110 code is different from those used in
American and Australian codes. The first shear perimeter recommended to be checked
is at a distance of l.Sd from the face of the column. Successive perimeters are
considered at 0.75d intervals for designing shear reinforcement. For corner columns
and edge columns with the bending about an axis parallel to the free edge (see
Figure D.7) the effective design shear force (Veff) is calculated to be 1.25 times the
y
4
Free edges Column ~ ~Cdtical section
Column
~1 ==--=-----r~
t~
1_ _
Mux
a ritical sections a
a=(C 1 + 1 .5d) a=(C 1 + 1.5d)
b=(C2 + 1.5d) b= (C2 +3d)
1---Free edge
(D.13)
For an edge column with a bending moment Mt equal to Muy. about an axis
perpendicular to the edge (as shown in Figure D.8), the design effective shear Veff is
given by
Veff
v=-- (0.15)
ud
where u is the perimeter of the critical section and d is the effective depth of the slab.
For a slab without any shear reinforcement v must be less than Vc where
D-14
y
4
Column : rCritical section
!c,i
! Muy ___ X Ib
a
a=(C 1 + 1.5d)
b= (C2 +3d)
~----Free edge
Figure D.S - Edge column with moment about the axis perpendicular to the free
edge.
and where Ym is the_ partial safety factor for strength of materials; Ast is the area of
tension reinforcement in the critical section; b is the width of the critical section; dis
the effective depth and feu is the characteristic cube strength of the concrete. For
feu< 25 MPa, the (ja/25) 113 term in equation D.l6a is taken as unity.
D-15
The approach taken by the CEB-FIP model code 1990 (1993) differs completely in
form from the American and Australian codes but has some similarity with BS 8110-
1997. For edge column connections with the moment normal to the free edge the
CEB-FIP Model code assumes a shear stress of 'isd on the perimeter u1 as shown in
(D.17)
where Psd and Msd are shear force and unbalanced moment calculated at the centre of
the column; Vp is the vertical component of the prestress force passing through the
column within the distance of D/2 from the column, D being the overall depth of the
square columns K=0.45); and W1 is a parameter of the control perimeter such that,
(D.l8)
where dl is an elementary length of the perimeter and e is the distance of dl from the
(D.19)
For edge connections with the loading at the interior of the slab and with loading
eccentricity in the direction perpendicular to the slab edge, the shear stress is
D-16
and O.SC1
assumed to be uniform on the perimeter ui as shown in Figure D.9b. The design shear
stress is given by
(D.20)
(D.21)
where YMis a partial safety factor;fck is the characteristic cylinder strength; and
written as
(0.24a)
(0.24b)
(0.24c)
YM Vu + YM Mu 1 (0.25)
Vuo Muo
For corner column connections the CEB-FIP Model Code assumes a shear stress of 'rsd
loading is towards the interior of the slab, the uniform shear distribution takes place
along the perimeter ui as shown in Figure O.lOb and the shear stress developed is
given by
(0.26)
The suffixes X and Y denote the moment in x and y directions, respectively. The
symbols are as defined as above for the case of edge column connections and the shear
(D.27)
(D.28a)
(D.28b)
For the comer column connections the interaction diagram is given by equation D.25
with
factors controlling the shear strength. The shear strength is zero for the slabs without
tensile reinforcement.
D-19
In Eurocode 2 ( 1992) the design model for slabs without shear reinforcement is
limited to columns having an aspect ratio of C1/C2 of less than 2. For edge and comer
columns the Eurocode 2 model for punching shear assumes the critical perimeter to be
at a distance of 1.5d from the faces of the column (shown in Figures 0.11 and D.12),
provided that the perimeter is less than 11d. For slabs where the perimeter exceeds
lld, the control perimeter is defined as shown in Figure D.13. For slabs without shear
(D.30)
where Vsd is the total design shear force, u is the critical perimeter, {3 is a coefficient to
take into account the effects of eccentricity of loading and is taken as 1, 1.5, 1.4 and
1.15 for concentric loading (no eccentricity), comer columns, edge columns and
(D.31)
where 'rRd is the basic resisting shear strength of the concrete and is taken as
0.0525 f% (D.32)
rc ck
d is the average of the effective depths of the slab in the x and y directions and rc is a
partial safety factor. The other parameters used in equation D.31 are
D-20
Column
perimeter
1-----Free edge
Figure D.ll - Control perimeter of Eurocode 2 for edge column.
edges
Column
~o~~rol
1
...... ..:..?.<J. perimeter
1.5d
c,, ~
l 2C2
;..--~-.,.
Figure D.13 - Control perimeter of Eurocode 2 for column with u > lld.
D-21
~ 0.9CTpc
P1 = Plx Ply+ < 0.15 (0.34)
Acfyd
where Pix, Ply are the ratios of tension steel in x and y directions, respectively, CTpc is
the average prestressing force resulting from the initial prestress without losses and /yd
is the design yield stress of the reinforcement. As for BS 8110 (1997), Eurocode 2
does not provide guidelines for the transfer of unbalanced moment into the column
without shear.