Material Model For URM With A Application On The Seismic Assessment Of A Hospital In Tyrol, Austria
Suikai Lu ^{1} , Rudolf Heuer ^{2} , Marian Ralbovský ^{3} , Rainer Flesch ^{4}
^{1} Dipl.Ing., Scientist, Business Unit Transport Routes Engineering, arsenal research, 1030 Vienna, Austria, suikai.lu@arsenal.ac.at ^{2} Professor, CMSD – Center of Mechanics and Structural Dynamics, Vienna University of Technology, 1040 Vienna, Austria, rh@allmech.tuwien.ac.at ^{3} Dipl.Ing., Scientist, Business Unit Transport Routes Engineering, arsenal research, 1030 Vienna, Austria, marian.ralbovsky@arsenal.ac.at ^{4} Professor, Head of Business Unit, Business Unit Transport Routes Engineering, arsenal research, 1030 Vienna, Austria, rainer.flesch@arsenal.ac.at
1) Abstract The content of this paper presents a new model for unreinforced masonry (URM), based on the plastic material model by Ganz [1]. Here, the idea of Ganz [1] was followed, to use a combination of yielding surfaces in the stress space, where each surface describes one failure mode. Compared to the material model of Ganz [1], the new model was extended to cover even the tension strength aspects of masonry constructions in both directions (in plane, orthogonal and parallel to the horizontal joints) for each failure mode. The model still consists of only 5 convex yielding surfaces which describe the following failures, respectively:
) tension failure ) compression failure ) shear failure ) sliding along the horizontal joints ) tension failure in the horizontal joints
Additionally to the theoretical background an application is shown, where laboratory experiments are used to test and calibrate the material model and its parameters. Therefore, this model was implemented into the Finite Element Software ANSYS. The implementation involves all the failure modes and an automatic searching for the positions of the masonry structures in the global FEmodel. The result of this numerical implementation is the display of the cracked and yielded areas of the wall, respectively. The accuracy depends on the size of the finite elements chosen in the model. Furthermore, this model was applied to seismic assessment of a hospital located in Innsbruck, Tyrol, Austria. In a first step, insitu measurements have been carried out to detect the dynamic parameters (natural frequencies and mode shapes) for updating the finite element model with these properties. Afterwards, the FEmodel was analyzed by response spectra method using SRSS combination rule to simulate a code earthquake, as required in the Austrian national code for designing of buildings with seismic actions, ÖNORM B 4015 [9]. Finally, the new material model was introduced to display cracks in the structure occurring due to the earthquake excitation.
2) Introduction Based on the classical theory of plasticity, Ganz [1] formulated two material models in 1985, where he described yielding surfaces for each failure mode in masonry. First he formulated a model for URM, which covers only 5 yielding surfaces. Then he developed a model to include tension strength, where 12 surfaces were needed. The new model, presented in this paper extends the basic Ganzmodel (without tension strength) by including tension in a new effective approach capturing the main failure modes of unreinforced masonry structures.
Both the analytical derivation as well as the software implementation in form of a macro for the FE Software ANSYS, are presented.
To verify the model, experimental laboratory tests were analyzed numerically by using this macro, in order to confirm and prove the analytical work.
In the last part of this paper, a practical application on a lifeline structure, a hospital in Tyrol, Austria was analyzed by using this new approach.
3) Original Model by Ganz In 1985, Ganz [1] formulated a material model for URM, where the two components brick and joints were split.
For the component brick, he focused on the most generally form, perforated bricks (see Figure 1). Within a limiting approach this theory can be also applied to a solid brick.
Figure 1 – Brick Element
The considered forces and cross sections are defined as follows:
F
x
= F + F
x ,1
x , 2
F
x
F
y
F
y
stands for the normal force, and
= F
y ,1
+ F
y , 2
F
x
stands for the horizontal force, and
,1
,
F
x , 2
act upon
A ,
x
F
y ,1
,
F
y , 2
act upon
_{A}
xy
A ,
x
, respectively.
A
xy
, respectively.
F
xy
F
xy
= F
xy ,1
+ F
xy , 2
stands for the shear force, and
sectional area reads
F
xy ,1
,
F
xy , 2
act upon
A , A
x
xy
respectively. The total cross
A A
=
x
+
A
xy
+
A
0
(1)
(2)
(3)
(4)
By combining the uni and biaxial parts of the forces using principal forces, the following three equations for failure in brick result:
f
1
f
2
f
3
τ
2
= −
xy
τ
2
= −
xy
2
= +
τ
xy
σ σ
x
(
σ
x
y
+
σ
y
(
σ
≤ 0
f
cx
y
+
)(
σ
y
f
cy
)
+
≤
f
cy
0
)
≤
0
tension failure in brick
compression failure in brick
shear failure in brick
(5)
(6)
(7)
For the component mortar, supposing that the vertical joints are not filled, it is only necessary to focus on the horizontal joints. With this assumption the model equations are on the “conservative side”.
Sliding in the joints is modeled by means of the MohrCoulomb´s law,
f
4
= τ
2
xy
−
( c
−σ
x
tan( ))
ϕ
2
≤
0
sliding along the horizontal joints
(8)
Finally, a tension cutoff for the Mohr Coulomb´s friction law is formulated,
f
5
=
τ
2
xy
+
σ
x
⎡
⎢
⎣
σ
x
+
2 tan(
c
π
+
ϕ
4
2
)
⎤
⎥
⎦
≤
0
tension failure in the horizontal joints
(Eq 5 to Eq 9) describe the law for URM according to Ganz considering the components of
compressive strength
f
cx
,
f
cy
, respectively.
(9)
4) Modified Material Model The new model, developed within this research work, was expanded by the former model to consider also tension stresses. Taking the uniaxially exposed parts of the brick section, the governing equations can be written as:
β A ≥ F
t
x
x
,1
≥ −β A
c
x
F
y
,1
≡
where
F
xy
,1
≡
0
β
c
,
β
t
denote compression strength and tension strength of brick, respectively.
(10)
(11)
For the biaxially exposed parts, the inequation can be written by using the principal forces in terms of
≥ −β A
c
xy
(12)
(13)
the former derived material laws for brick (Eq 5 and Eq 7) can be replaced by
f
1
f
3
=
=
τ
2
xy
τ
2
xy
−
(
σ
x
−
+
(
σ f
y
cy
f
tx
)(
−
f
ty
σ
+
y
−
σ
y
f
ty
)
≤
0
)
−
f f
cy ty
≤ 0
(14)
(15)
The surface function
f
2
(Eq 6) remains unchanged.
To enclose tension in joints, criteria
tension cutoff
f
4
(Eq 8) can be remained unchanged too, but the equation for
f
5
(Eq 9) has to be modified (Figure 2).
Figure 2 – Tension cutoff
Finally,
f
5
=
τ
f
5
2
xy
+
can be written in terms of
σ
x
2
+ 2
c
⋅
cos
ϕ
−
sin
ϕ
⋅
f
tx
1 sin
−
ϕ
(
σ
x
−
f
tx
)
−
2
σ
x
f
tx
+
f
tx
2
≤
0
(16)
The new material model (Eq 14, 6, 15, 8, 16) can be displayed graphically as a combined yielding surface (Figure 3):
Figure 3 – Combined Yielding Surface of the extended model for URM
The failures of the corresponding five yielding surfaces can be interpreted in the same way as in the original model by Ganz (compare to the previous section).
5) Input URM parameters for analytical analysis
For numerical implementation of the URM model, some main input parameters are necessary. The following section presents an overview of the evaluation of the data used for the analysis.
) Compression strength orthogonal to the horizontal joints [2]:
f
cx
= K ⋅ β
c
0. 75
⋅ f
mc
0 .25
f
cx
(17)
with the module K = 1 . 0 ÷ 1 . 5 , and the compression strength of the considered mortar
f
mc
.
) Compression strength parallel to the horizontal joints [3]:
URM consisting of solid brick:
f
cy
= 0.75 ⋅ f
cx
f cy
URM consisting of perforated brick:
f
cy
= 0.5 ⋅ f
cx
(18)
(19)
) Tension strength orthogonal to the horizontal joints [4]:
The tension strength of URM depends only on the tension strength of the used mortar
written as:
f tx
f
tx
=
2
3
⋅ f
mt
f mt
and can be
(20)
) Tension strength parallel to the horizontal joints [3]:
In case of tension strength parallel to the horizontal joints, two different crack types should be treated separately.
f ty
Crack Type A (Figure 4) occurs, if bricks are made of low quality materials and if large portions of
normal stresses
σ
x
are exposed to masonry members.
Figure 4 – Crack Type A, taken from [5]
f
ty BT
,
_
where
A
h ⋅ f
b
,
bt horiz
^{=} 2 h
(
b
+
h
m
)
f
bt horiz
,
denotes the horizontal tension strength of used brick
(21)
Crack Type B (Figure 5) is typical for high strength bricks in combination with low quality mortar and/or
if the exposing normal stress σ
x is very small.
Figure 5 – Crack Type B, taken from [5]
f
ty BT B
,
_
where τ
l ^{=} 2 h
(
b
⋅
τ
u
b
+
h
m
)
u stands for shear strength.
(22)
) Young's modulus orthogonal to the horizontal joints [2]: E _{x}
E
x
=
1000 ⋅ f
cx
(23)
) Young's modulus parallel to the horizontal joints [6], [7]: E _{Y}
E
Y
E
X
^{=} 1
+
2 π
⋅
⋅
f
where
f
h
b
^{≈} 4 l
⋅
b
.
(24)
) Friction angle: ϕ ; and shear strength under no compressive stress: c The Friction angle ϕ varies normally between 20° and 40°, and the shear strength under no
compressive strength is between 0.2 – 2.5 MPa where some numbers for c are listed in [2] in table
3.4.
) Shear strength
τ
u
(
= c + tan(ϕ ) − σ
x
)
The shear strength is in accordance to a conventional MohrCoulomb friction law, where
the normal stress.
σ
x
(25)
denotes
6.) Implementation of the new model into the FESoftware ANSYS:
To implement the model into a FEsoftware, it should be taken into account, that
_{f} _{t}_{x} ≤
c
.
tan ϕ
If this condition is not maintained,
f
tx
^{c}
has to be set equal to tan ϕ
(26)
, before continuing the analysis.
The stress state of the analyzed object must be verified. Four different positions can be distinguished (Figure 6). Therefore, an extra condition (Eq 27), has to be considered.
Figure 6 – Positions of the analyzed stress state
f
5 / 2
T
=
c
⋅
cos
_{ϕ} −
sin
_{ϕ} ⋅
f
tx
−
f
tx
+ σ
^{x}
Together with condition
f ,
5
f
5
=
τ
2
xy
+
σ
x
2
+
2
c
⋅
cos
ϕ
−
sin
ϕ
⋅
f
tx
1 sin
−
ϕ
≥ 0 
then the result is 
1 
< 0 
then the result is 
0 
(
σ
x
−
f
tx
)
−
2
σ
x
f
tx
+
f
tx
2
≤ 0
> 0
then the result is
then the result is
(27)
0
1
,
(28)
the four positions P1  P4 can be separated numerically in the following way:
P1:
f
5
⋅
f
5 / 2
T
=
0 1
⋅
=
0
; P2:
f
5
⋅
f
5 / 2
T
=
0 0
⋅
=
0
; P3:
f
5
⋅
f
5 / 2
T
=
1 0
⋅
=
0
; P4
f
5
⋅
f
5 / 2
T
=
1
⋅
1
=
1
(29)
Only in case of P4, the analyzed stress point is outside the combined yielding surface, which means that cracks will occur. The combined yielding surface was implemented into ANSYS (Figure 7) including the conditions described previously.
Figure 7 – Screenshot of the implemented Program for URM
The implementation also involves an automatic detecting of the positions of masonry structures in the global coordinate system of the FEmodel.
7) Verification of the material model by numerical analysis of laboratory tests:
In 1982, Ganz et al [8] made experimental tests on URM. They exposed the test specimen stepwise until collapse. Those walls were exposed either uniaxially and/or biaxially. Also the angle of the horizontal joints of each specimen varied between 0; 22.5; 45; 67.5 and 90 degrees.
To verify the material model and to demonstrate the implemented macro, these experimental tests have been recalculated numerically and were compared to the test results.
Input data for the computer simulation are given in Table 1.
Table 1 – Input data for computer simulation
In Figure
implementation.
8,
K3
of
density ρ 
[kg/m³] 
905 
strengths f cx 
[N/mm²] 
7.6 
f cy 
[N/mm²] 
2.7 
f tx 
[N/mm²] 
0.03 
f ty 
[N/mm²] 
0.00 
Cohesion c 
[N/mm²] 
0.06 
Friction angle ϕ 
[°] 
39 
the test series 
is displayed graphically, to show the effectiveness of the
Figure 8 – left: numerical implementation; right: laboratory experiment on real test specimen
A summary of additional results is given in Table 2, where stresses were measured in the middle of the wall and were taken from the middle element of the FEModel.
Table 2 – Comparison between test results and numerically analyzed results
Test 
Angle of 
Ratio F H / F V 
Measured 
σ 
at laboratory 
Numerically analyzed σ 

horiz. joints 
crack (taken from [8]) 
crack 

σ 
X 
σ 
Y 
τ XY 
σ 
X 
σ 
Y 
τ 
XY 

[°] 
[ ] 
[N/m²] 
[N/m²] 
[N/m²] 
[N/m²] 
[N/m²] 
[N/m²] 

K1 
22.5 
1 / 10.9 
8.00 e4 
9.20 e5 
4.20 
e5 
7,89e4 
9,10e5 
4,14e5 

K3 
0.0 
0 
/ 1 
0 
7.63 e6 
0 
0 
7.61e6 
0 

K4 
90.0 
0 
/ 1 
1.83 e6 
0 
0 
2.70e6 
0 
0 

K6 
45.0 
0 
/ 1 
3.20 e5 
3.20 e5 
3.20 
e5 
3.19e5 
3.19e5 
3.19e5 

K7 
22.5 
0 
/ 1 
3.90 e5 
2.25 e6 
9.30 
e5 
3.99e5 
2.33e6 
9.64e5 

K8 
67.5 
0 
/ 1 
2.20 e5 
4.00 e4 
9.00 
e4 
2.28e5 
3.91e4 
9.43e4 

K10 
0.0 
1 / 3.2 
2.11 e6 
6.44 e6 
0 
2.40e6 
7.3e6 
0 

K11 
22.5 
1 / 3.1 
2.04 e6 
4.49 e6 
1.23 
e6 
2.07e6 
4.36e6 
1.13e6 

K12 
45.0 
1 / 3.2 
2.03 e6 
2.03 e6 
1.08 
e6 
2.05e6 
2.05e6 
1.05e6 
8) Seismic application of the material model to the Hospital LKH Innsbruck, Tyrol, Austria
This hospital (Figure 9) was built approximately in 1945 where the structure consists of unreinforced masonry (URM) with “AstMolin” (reinforced ribbed arch concrete) slabs and flat reinforced concrete slabs at the aisles. It exhibits a basement, a ground floor and five upper floors including one attic floor. Innsbruck´s geographically position is: 11.390 longitudes and 47.263 latitudes. The soil under this building is composed of middle dense to dense sandy and stony flint. The maximum PGA in accordance to Austrian national code for designing of buildings with seismic actions, ÖNORM B 4015 [9], is 1.54 m/s². However, the strongest earthquake was in 1572, and its PGA has been assessed to 1.48m/s² with a main duration of 3.4 sec and a dominant frequency of 2.7 Hz according to the Austrian earthquake catalogue of ZAMG (Zentralanstalt für Meteorologie und Geodynamik, the Austrian Central Institute of Meteorology and Geodynamics). I.e., structural designing according to the current Austrian code ÖNORM B 4015 leads to mechanically conservative results.
Figure 9 – Photograph of LKHInnsbruck
In a first step, the natural frequencies of the structure were measured by ambient vibration excitations as wind, humans in the structure, traffic and earth micro tremors. The advantage of this method is that no extra artificial exciter is needed. Therefore, the structure was instrumented by 25 sensor positions (partly triaxial) very sensitive seismic accelerometers (Wilcoxon Research 731AP31), positioned as shown in Figure 10 and Figure 11.
Figure 10 – Sensor position at the fourth upper floor
Figure 10 – Sensor position in the staircases
The results of the measured data were analyzed by the software MACEC, to get the natural frequencies and corresponding mode shapes. For comparison’s sake a finite element model (Figure 11) was built and modal analysis was performed by means of the software ANSYS, where mainly shell (shell 63) and line (line 188) elements were used. In this model, the total number of elements was approximately 80,000. In Figure 12, the numerical results are compared to the experimentally evaluated parameters.
Figure 11 – Finite Element model of LKH Innsbruck
Dynamic behavior: (mode shapes and natural frequencies)
Experimentally evaluated by In Situ Measurements 
Numerically analyzed by Finite Element Software, ANSYS 
f _{1} =2.423 Hz 
f _{1} =2.3443 Hz 


∆=3% 

f _{2} =2.580 Hz 
f _{2} =2.5874 Hz 


∆=0% 

f _{3} =2.945 Hz 
f _{3} =3.1026 Hz 


∆=+5% 

f _{4} =3.265 Hz 
f _{4} =4.1136 Hz 


∆=+20% 
Figure 12 – Comparison of Mode shapes and natural frequencies
The material parameters of the existing masonry were derived by a simple compression test on a test specimen taken from the original structure (see Figure 13 to Figure 14). The bricks are of the dimensions l/w/h = 250/100/60 [mm].
Figure 13 – Test Specimen of the real structure
Figure 14 – Axial compression test
The evaluated axial compression strength oriented orthogonal to the horizontal joints is
N/mm². The parameters according to Eq (17)  Eq (25) are summarized in Table 3.
f
cx
=1.68
Table 3 – URM Parameters
Parameter 
Floor 
MinValue 
MaxValue 

[N/mm²] 
[N/mm²] 
[N/mm²] 

Compression 

strength 

Orthogonal to the horizontal joints 
1.68 

Parallel to the horizontal joints 
0.92 
1.18 

Tension 
* 

strength 

Orthogonal to the horizontal joints ^{*} 

4th floor 
0.08 

3rd floor 
0.19 

2nd floor 
0.25 

1st floor 
0.35 

Ground floor 
0.38 

Parallel to the horizontal joints 

4th floor 
0.25 
0.30 

3rd floor 
0.35 
0.45 

2nd floor 
0.40 
0.54 

1st floor 
0.49 
0.68 

Ground floor 
0.52 
0.72 

Shear 

strength 

4th floor 
0.14 
0.17 

3rd floor 
0.20 
0.25 

2nd floor 
0.23 
0.30 

1st floor 
0.28 
0.38 

Ground floor 
0.29 
0.41 

Youngs 

Modulus 

Orthogonal to the horizontal joints ^{*} 
2000 
2500 

Parallel to the horizontal joints 
1500 
1850 

Shear 
850 
1000 

Modulus 
After determination of the mode shapes and natural frequencies, a response spectrum analysis was performed considering the code spectra of ÖNORM B4015 [9], where the assumed earthquake was acting in the weaker direction (eastwest), see Figure 10. Finally, the material model described above was used to analyze the cracks occurred due to the earthquake. Figure 15  Figure 21 show the effected cracks, and how many criteria are violated:
number 5 (red) stands for “no crack” and the number 0 (blue) indicates that all criteria are violated.
Figure 16 – Crack pattern of ground floor
1 ^{s}^{t} upper floor
2 ^{n}^{d} upper floor
3 ^{r}^{d} upper floor
Figure 17 – Crack pattern of 1 ^{s}^{t} upper floor
Figure 18 – Crack pattern of 2 ^{n}^{d} upper floor
Figure 19 – Crack pattern of 3 ^{r}^{d} upper floor
4 ^{t}^{h} upper floor
Attic floor
Figure 20 – Crack pattern of 4 ^{t}^{h} upper floor
Figure 21 – Crack pattern of attic floor
9) Conclusion
A 
powerful and precise tool for analyzing the bearing capacity of URM has been developed. The result 
of 
this research work is a macro implementation into the Finite Element software ANSYS. The user, 
e.g. a practical engineer, has only to enter the material parameters into the input mask, which are in most cases available (or, at least, can be estimated from the literature or specific building codes), and
so a precise analysis of URM becomes applicable. Although this procedure is a nonlinear approach, the implementation was elastic, which fulfils most of the practical demands of a practical engineer. As result of this application the computer shows the occurrence, localisation and even the failure mode of the cracks in specific URM members. The accuracy depends on the chosen element size of the finite elements.
10) Acknowledgements
The research work of this paper was funded by arsenal research (Fund No: 2.05.00187.4.0). The authors are grateful for this support and want to thank them. The research work on LKHInnsbruck hospital was supported in part by TILAK, Tiroler Landeskrankenanstalten Ges.m.b.H. The authors want to also thank them.
References
[1] 
Ganz, H. R., Mauerwerksscheiben unter Normalkraft und Schub. Institut für Baustatik und 2004. 
[2] 
Konstruktion, ETH Zurich, report no. 148, 1985. Code, EN 199611, Design of Masonry Structures, part 11 common rules for reinforced and 
[3] 
unreinforced masonry structures. Glitzka, H., Druckbeanspruchung parallel zur Lagerfuge. Mauerwerkskalender 1988, pp 489 – 
[4] 
496, Ernst & Sohn, 1988. Tassios Εργαστηριο Ωπλισµενον Σκνροδεµατοζ, Η Μηχανικη, τηζ, τοιχοποιιαζ. 
[5] 
Αθηνα, 1986. Vratsanou, V., Das nichtlineare Verhalten unbewehrter Mauerwerksscheiben unter 
[6] 
Erdbebenbeanspruchung. Ph. D. Thesis, Institut für Massivbau und Baustofftechnologie, Universität Fridericiana zu Karlsruhe TH, 1992. Graubner, C. A., Glock, C., Meyer, G., Abschätzung der Knicklänge mehrseitig gehaltener 
Wände aus großformatigen Mauersteinen. Bauingenieur, Juni 2004, pp 300305, Springer, 

[7] 
Gross, D., Seelig, T., Bruchmechanik mit einer Einführung in die Mikromechanik. 3. edt, 
[8] 
Springer, 2001. Ganz, H. R., Thürlimann, B., Versuche über die Festigkeit von zweiachsig beanspruchtem 
[9] 
Mauerwerk. Institut für Baustatik und Konstruktion, ETH Zurich, report no. 75023, 1982. Code, ÖNORM B4015 (2002), Belastungsannahmen im Bauwesen–Außergewöhnliche EinwirkungenErdbebeneinwirkungen, Grundlage und Berechnungsverfahren 
Paper should be fully refereed, please
Molto più che documenti.
Scopri tutto ciò che Scribd ha da offrire, inclusi libri e audiolibri dei maggiori editori.
Annulla in qualsiasi momento.