FINITE ELEMENT MODELING OF CONCRETE MASONRY-INFILLED

STEEL FRAMES

W. El-Dakhakhni1 , M. Elgaaly 2 , and A. Hamid 3

ABSTRACT

Masonry infilled panels in framed structures have been long known to affect strength,
stiffness and ductility of the composite structure. In seismic areas, however, ignoring the
composite action is not always on the safe side, since the interaction between the panel and
the frame under lateral loads dramatically changes the stiffness and the dynamic
characteristics of the composite structure and hence its response to seismic loads. This study
presents a simple practical method of estimating the stiffness and the lateral load capacity of
concrete masonry-infilled steel frames (CMISF) failing in corner crushing (CC) mode, as well
as the internal forces in the steel frame members using finite element method (FEM) analysis.
In this method, each masonry panel is replaced by three struts with force-deformation
characteristics based on the orthotropic behavior of the masonry infill. A simplified steel
frame model is also presented based on the documented modes of failure of CMISF. The
method can be easily computerized and included in the analysis and design of three-
dimensional CMISF structures. The proposed technique accounts for the non-linear behavior
that occur in the steel frame due to formation of plastic hinges, and in the masonry panel due
to crushing. It has been demonstrated that the proposed model can predict both the stiffness
and the ultimate lateral load capacity of such systems.

Key Words: Concrete Masonry, FEM Modeling, Infilled Steel Frames, Orthotropic
Masonry, Three Strut Model.

1
Research Assistant, Department of Civil and Architectural Engineering, Drexel University,
Philadelphia, PA, 19104, USA. E-mail : [email protected]
2
Professor, Department of Civil and Architectural Engineering, Drexel University,
Philadelphia, PA, 19104, USA. E-mail : [email protected]
3
Professor, Department of Civil and Architectural Engineering, Drexel University,
Philadelphia, PA, 19104, USA. E-mail : [email protected]
INTRODUCTION

Masonry infilled panels can be found as interior and exterior walls in reinforced
concrete and steel framed structures. Since they are normally considered as architectural
elements, their presence is often ignored by structural engineers. However, they tend to
interact with the surrounding frame when the structure is subjected to strong earthquake
loads; the resulting system is referred to as an infilled frame. Ignoring the effect of the infill in
stiffening and strengthening the surrounding frame is not always a conservative approach,
since the stiffer the building, usually, the higher seismic loads it attracts. If the panel is
overstressed and hence failed partially or wholly, the high forces previously attracted and
carried by the stiff infilled frame, will be suddenly transferred to the more flexible frame after
the infill is partially or fully damaged. In addition, change in stiffness distribution can result
in higher seismic forces due to torsional effects. Because of the complexity of the problem
and the absence of a realistic, yet simple analytical model, the effect of masonry infill panels
is often neglected in the nonlinear analysis of building structures. Such an assumption may
lead to substantial inaccuracy in predicting the lateral stiffness, strength, and ductility of the
structure. It will also lead to uneconomical design of the frame since the strength and stiffness
demand on the frame could be largely reduced. The current study aims to present a simple
method of predicting the stiffness as well as the ultimate load capacity of concrete masonry-
infilled steel frames (CMISF). The method can be used to produce design aids and to develop
a conceptual approach of analysis and design of such systems.

FAILURE MODES OF INFILLED FRAMES

Based on the knowledge gained from the research work during the last five decades,
failure modes of masonry infilled frames can be categorized into five distinct modes, namely:
1. Corner crushing mode (CC mode), represents crushing of the infill in at least one of its
loaded corners, as shown in Fig. 1-a. This mode is usually associated with infill of weak
masonry blocks surrounded by a frame with weak joints and strong members.
2. Sliding shear mode (SS mode), represents horizontal sliding shear failure through bed joints
of a masonry infill, as shown in Fig. 1-b. This mode is associated with infill of weak mortar
joints and strong frame.
3. Diagonal compression mode (DC mode), represents crushing of the infill within its central
region, as shown in Fig. 1-c. This mode is associated with a relatively slender infill, where
failure results from out-of-plane buckling instability of the infill.
4. Diagonal cracking mode (DK mode), in the form of a crack connecting the two loaded
corners, as shown in Fig. 1-d. This mode is associated with weak frame or frame with weak
joints and strong members infilled with a rather strong infill.
5. Frame failure mode (FF mode), in the form of plastic hinges in the columns or the beam-
column connection, as shown in Fig. 1-e. This mode is also associated with weak frame or
frame with weak joints and strong members infilled with a rather strong infill.

It is worth mentioning that only the first two modes are of practical importance, since the
third mode is very rare to occur and requires a high slenderness ratio of the infill to result in
out-of-plane buckling of the infill under in-plane loading, this is hardly the case when
practical panel dimensions are used. The fourth mode should not be considered a failure mode
because the infill can still carry more load after it cracks. The fifth mode should be considered
in the case of RC frames, but for steel frames infilled with unreinforced hollow concrete
masonry blocks, this mode hardly occurs. The study conducted herein models the CC mode
only, which is the most common mode of failure for CMISF.

DEVELOPMENT OF CMISF MODEL

Subjecting a bare masonry panel to a diagonal loading results in a sudden failure

because of unconfinement. This failure is usually initiated by a stepped crack along the loaded
diagonal, dividing the panel into two separate parts. This behavior was investigated by
ElDakhakhni (2000) using ANSYS5.3 FE program. The ASTM E-519 diagonal tension test
specimen, representing this loading case, and the FE model used to duplicate the experimental
failure mechanism are both shown in Fig. 2. Unlike the unconfined panel, as soon as a
diagonal crack develops within an infilled panel (usually at a much lower load and deflection
levels than ultimate) the panel finds itself confined within the surrounding frame and bearing
against it over contact lengths, as shown in Fig. 3. The contact lengths provide enough
confinement to prevent failure and allows the panel to carry more load until existing diagonal
crack continues to widen and new cracks appears leading, eventually, to ultimate failure. This
behavior was repeatedly reported in the literature by many researchers [Polyakov (1956),
Stafford-Smith and Carter (1969), Flanagan et al. (1992), Saneinejad and Hobbs (1995) and
Seah (1998)].

From the above discussion it is more rational to consider the panel to be composed of two
diagonal regions as shown in Fig. 3, one connecting the top beam to the leeward column and
the other connecting the windward column to the lower beam. From a different aspect, as
reported by many researchers, [Reflak and Fajfar (1991), Saneinejad and Hobbs (1995) ,
Mosalam et al. (1997a,b,c), Buonopane and White (1999), and Furutani et al. (2000)], the
bending moments and shearing forces in the frame members cannot be replicated using a
single diagonal strut connecting the two loaded corners. Based on the above, it is suggested
that, at least two additional off-diagonal struts located at the points of maximum field
moments in the beams and the columns are required to reproduce these moments as shown in
Fig.3. Furthermore, since the interaction between frame members and the infill depends on
the contact length which, in turn, is affected by the stiffness and the deflected shape of the
frame members, the use of a multi-strut model will allow for the interaction between
different panels in multi-story buildings. This is due to the fact that some beams (and/or
columns) will be loaded from the upper and lower panels (or left and right panels) at different
locations within the span (or height), which will affect their deflected shape and hence the
panels strains, and consequently changing the failure load. Using the ANSYS5.3 FE
program, ElDakhakhni (2000) modeled a single panel infilled frame using PLANE42 plane
stress elements connected to the frame BEAM3 elements with CONTACT12 contact
elements. The stressed part of the panel, as shown in Fig. 4, is in the form of a diagonal area.
The use of multi-strut model rather than a single strut will better represent the actual stressed
area within the infill, and will also facilitate the modeling of the progressive failure occurring
at the corners contact region, not just at the corner points.
The development of the infilled frame analytical model is divided into two parts. The first
part deals with the development of a FE model for the geometrical representation of the
structural systems components, namely, the steel frame and the infill panel. The second part
deals with the material model suggested for these two components.

STEEL FRAME GEOMETRICAL MODEL

The steel frame members were modeled with ANSYS5.3 FE program using
BEAM3, an elastic beam elements connected by non-linear rotational spring elements,
COMBIN39, at the beam-column joints. The concentration of non-linearity in the frame
joints only, is based on the fact that due to the limited infill ductility and thus limited frame
deformation at the peak load except at the loaded corners, the maximum field moments as
well as the bending moments at the unloaded joints are lower than that at the loaded joints
and has been found to be, at most, 20% of the plastic moment capacity of the section
[Saneinejad and Hobbs (1995)]. Unlike the model suggested by Seah (1998), which allows
for the interaction between the axial and shear forces and the bending moment at the
connection, no translational springs were used at the joint of the suggested model, instead,
the DOF coupling option provided in ANSYS5.3 was used to couple both the beam and the
column nodes at the beam-column connection in the two planer translational DOF and forced
them to undergo the same displacement.

INFILL PANEL GEOMETRICAL MODEL

Saneinejad and Hobbs (1995) showed that for steel frame members infilled with plane
concrete panel, the points of maximum field moment developed within the frame members lie
approximately at the end of the contact lengths, and are located at distances from the beam-
column connection given by
2 (M pj + c M pc) 2 ( M pj + b M pb)
ch = 0.4h b l = 0.4l (1-a,b)
ct bt
where, c is the ratio of the column contact length to the height of the column and b is the
ratio of the beam contact length to the span of the beam; h is the column height and l is the
beam span. Mpj is the minimum of the columns, the beams or the connections plastic
moment capacity, referred to as the plastic moment capacity of the joint; Mpc and Mpb are the
column and the beam plastic moment capacities, respectively; c and b are the normal
contact stresses on the face of the column and the beam, respectively; c and b are the ratios
between the maximum elastic field moment developed within the height of the column to Mpc
and that developed within the span of the beam to Mpb respectively; and finally t is the
thickness of the panel.
It is worth mentioning that these contact lengths are not constant and they vary throughout the
loading history. Saneinejad and Hobbs (1995) suggested that, near failure, either c and c
or b and b will reach their respective upper bound values of co and o or bo and o,
depending on whether the infill failure is initiated on the columns or the beams face
respectively. They suggested a method to determine c and b , and demonstrated, based on
FEM analysis, that 0 =0.2. For simplicity, and assuming full crushing of the infill in the
loaded corners region near failure, it is suggested that c, b, c , and b have reached their
respective upper bound values. The upper bound values of c and b , namely co and bo,
suggested by Saneinejad and Hobbs (1995) are given by
f c f c
co = bo = (2-a,b)
1 +3 r 1 +3
2 4 2


where, f c is the compressive strength of the plane concrete panel; and is the coefficient of


friction between the steel frame and the concrete panel; and r is the panels aspect ratio

  
defined as r = h/l <1. In order to modify the above equations to suit the concrete masonry, f c


should be replaced by f m-0 and f m-90 in equations 2-a and 2-b, respectively, where f m-0 and
f m-90 are the compressive strength of the masonry panel parallel and normal to the bed joint
respectively; since, unlike the isotropic plane concrete, concrete masonry is anisotropic or, at
best, orthotropic. Furthermore, since between steel and masonry is small (usually about
0.3), and the shrinkage of the concrete infill may result in a separation between the frame and
the panel; a rational assumption will be to neglect the friction between the steel frame and the
masonry. A similar assumption was also suggested by Liauw and Kwan (1982) assuming
friction to be a strength reserve. It is also worth mentioning that a recent study conducted by
Flanagan et al. (1999) concluded that the method suggested by Saneinejad and Hobbs (1995)
estimated twice the capacity obtained from experimental work that the best results were
obtained with = 0. Also since r < 1, then r4 is very small and can be neglected.

Summarizing the above assumptions and simplifications, it is suggested that the distances
from the beam column connection to the points of maximum field moments in the frame
columns and beams (which are also approximately the contact lengths) are to be given by
equations 5 and 6, respectively,
2 (M pj + 0.2 M pc ) 2 (M pj + 0.2 M pb )
c h = 0.4h bl = 0.4l (3-a,b)
f
t m 0 t f m 90
Referring to Fig. 3 it is more practical to use strut instead of plate elements to represent the
two regions of the panel. Assuming that the equivalent uniformly loaded diagonal region of
the panel to be of area equal to A, where A is to be determined later, hence each region of the
panel shown in Fig. 3 will be of area =A/2. Furthermore, assuming uniform contact stress


distribution along the contact areas, each region will be replaced by two struts, each of area
A1=1/2 (A/2)=A/4, located at the beginning and the end of the contact length. Combining the


two struts connecting the loaded corners, from the two regions, into one strut of area
A2 =2 A1= A/2 results in representing the whole panel by three struts, an upper strut
connecting the upper beam with the leeward column with area A1=A/4, a middle strut
connecting the two loaded corners with area A2=A/2, and finally a lower strut of area A1=A/4
connecting the windward column with the lower beam, where A=2A1+A2. A diagram showing
the proposed geometrical model for a typical CMISF is shown in Fig. 5.
Saneinejad and Hobbs (1995) replaced the panel by a single strut with an area Ad given by
b
(1 c ) c h t c + b l t
fc fc
Ad = (4)
cos
where, fc is a reduced strength for the concrete to account for the ultimate design limit state;
is defined as tan-1 = (h/l); and b is an equivalent uniform shear stress developed on the
beam-infill interface and is defined as b = b. Again based on the previous discussion, it is
suggested to neglected the b term in equation 4. Furthermore, since this study deals with the
behavior of infilled frames up to failure, no material reduction factor will be employed. Then
fc will be assumed to reach its upper bound value for masonry parallel to bed joint of f m-0. At
failure, referring to the discussion following equations 2-a and 2-b, c will also reach its
upper bound value of co which is, for concrete masonry, approximately equal to f m-0. This
will result in reducing the fc /co factor to unity. In this assumption, the effect of the biaxial
state of stress in the infill corners vicinity was neglected for simplicity. Based on the above, it
is suggested that the total diagonal struts area, A, is to be calculated by
(1 c ) c h t
A = (5)
cos

STEEL FRAME MATERIAL MODEL

The ultimate moment capacity of the non-linear rotational spring, representing the
beam-column joint, is defined as the minimum of the columns, the beams or the
connections ultimate capacity, Mpj , which will be referred to as the plastic moment capacity
of the joint, as defined earlier in equations 1 and 2. The rotational stiffness of the spring can
be calibrated so that the lateral stiffness of the frame model matches that of the actual bare
frame, which can be obtained experimentally or using simple elastic analysis or, in case of
semi-rigidly connected members, using available data on modeling semi-rigid connections
[Chen and Lui (1991)]. The joint behavior is shown in Fig. 5, where, el is the maximum
elastic rotation that the joint can undergo without yielding; pl is the maximum plastic
rotation before the joint undergoes moment reduction below Mpj; and ult is the maximum
plastic rotation beyond which the joint cannot sustain any moment.

INFILL PANEL MATERIAL MODEL

Based on the available literature, it is evident that the stressed part of the panel is a
diagonal region connecting the two loaded corners. It is therefore justifiable to assume that
the panel properties in the diagonal direction are the properties governing the behavior of the
infill panel. Masonry panels have been known to be anisotropic, [Hamid and Drysdale (1980),
Khattab and Drysdale (1992), Mosalam et al. (1997c) and Seah (1998)]. A close
approximation is to consider the anisotropic masonry panel to be orthotropic. Due to the fact
that the panel behaves as if it was diagonally loaded, constitutive relations, of orthotropic
plates [(Shames and Cozzarelli (1992)] and axes transformation matrix, are used to obtain the
Youngs modulus, E, of the panel in the diagonal direction using the following equation
1
E = (6)
1 2 0 90 1 1
cos + + cos sin + sin
4 2 2 4

E0 E0 G E 90
where, E0 and E90 are Youngs moduli in the direction parallel and normal to the bed joints
respectively; 0-90 is Poissons Ratio defined as the ratio of the strain in the direction normal
to the bed joints due to the strain in the direction parallel to the bed joints; and G is the shear
modulus.
It is common to relate the initial Youngs modulus of quasi-brittle materials such as concrete
and masonry to their ultimate compressive strength. Therefore it seems rational to assume
that, not only Youngs modulus will change, but also the ultimate strength of the masonry
panel in the direction, f m- . A simple way to account for this direction variation is to
relate E to f m- using the same factor relating E90 to f m-90 and neglect the shear stress effect,
since the infill is failing in a CC mode, as well as the effect of the biaxial state of stress in the
infill corners vicinity. The reason for choosing E90 and f m-90 is that it is a common practice
as well as a standard test (ASTM E-447), to obtain the strength of masonry prisms in a
direction perpendicular to bed joints i.e. the vertical direction, which is usually the loading
direction in load-bearing walls. The assumption that the masonry compressive strength varies
according to the angle of loading was investigated by Hamid and Drysdale (1980) and a value
of f m-0 =0.7 f m-90 was suggested by Seah (1998). Fig. 6-a shows the orthotropic model for
the masonry panel.
Based on non-linear FE analyses, Saneinejad and Hobbs (1995) suggested that the secant
stiffness of the infilled frames at the peak load to be half the initial stiffness. This might be
directly interpreted into the stress strain relation for the masonry panel by assuming that the
secant Youngs modulus at peak load Ep is equal to half the initial Youngs modulus, E , i.e.
Ep= 0.5 E .This assumption is justified since the stiffness of the infilled frame is affected
primarily by the stiffness of the infill [Dhanasekar and Page (1986)].
As shown in Fig. 6-b knowing Ep and f , it is now an easy task to determine the strain
corresponding to the peak load p . Instead of using the parabolic stress-strain relation shown
in Fig. 6-b, it is suggested to approximate it into a tri-linear relation which is simpler and
more practical for analysis as shown by the thick lines in the same figure. Unless more
accurate data are available, the parameters in Fig. 6-b will be assumed according to the
following
1 = p - 0.001 2 =p + 0.001 u = 0.01 (7-a,b,c)

Knowing the stress strain relation along with the area (from equation 5) and the length of each
of the three struts (which can be easily calculated knowing the panel dimensions and the
contact lengths given by equations 3-a and 3-b) makes it possible to obtain a force-
deformation relation for each strut. As shown in Fig. 6-c, by simply multiplying the strains 1,
2 and u by the length of each strut resulting in obtaining 1, 2 and u respectively. Also
multiplying the stress, f m- , by the area of each strut results in obtaining Fu for each strut. In
fact assuming that E and f m- are the same for all struts and neglecting the minor difference
in the inclination angle between the middle strut and both the upper and the lower strut, will
result in finding only two distinct force-deformation relations, one for the upper and lower
struts and another for the middle strut.
It is worth mentioning that, the use of a macro model, that is neither a single strut, nor a plate
to represent the panel was previously suggested by some researchers. Chrysotomou (1991)
and Mander and Nair (1994) suggested multiple strut models. Mosalam et al. (1997c) used a
truss with contact elements. Due to the complexity of the problem, most of the properties of
these macro models suggested by different researchers were not justified based on the
material level, unlike the suggested model. In fact, the areas and the stiffnesses of the
members representing the panel were generally selected to match either some experimental
findings like the stiffness and/or the ultimate load or the natural frequency under seismic
loading. In other cases the properties of the macro model were selected merely to match some
properties of a more sophisticated micro model.

MODELING OF TEST SPECIMENS USING THE PROPOSED METHOD

The suggested method was used to model three CMISF specimens. Two of the
specimens were tested at the University of New Brunswick under monotonic racking load by
Yong (1984) and Richardson (1986) and the third was tested in Cornell University by
Mosalam et al. (1997a) under quasi-static displacement. The first two specimens are identical
single panel CMISF with different masonry strength. The reasons for choosing these
specimens are primarily for the experimental results consistency as well as to verify the effect
of changing the masonry strength on the CMISF model behavior, and because these
experimental results were duplicated by Seah (1998) using a very sophisticated micro FE
model consisting of series of plane stress elements connected by ten springs at each node. The
third specimen is a one fourth scale, two bay-single story CMISF with semi-rigid connection,
this specimen was chosen to demonstrate the effectiveness of the method to model semi-rigid
connections and the effect of using three struts on changing the bending moment diagram of
CMISF. The details of the calculations involved in the FE modeling of these specimens can
be found elsewhere [ElDakhakhni (2000)].

The ANSYS5.3 FE program was used to generate the load-deflection relation of specimen
WB2 tested by Yong (1984), utilizing the proposed technique. The load-deflection relations
for the bare and the infilled frame model are shown in Fig. 7-a along with test results for
comparison. It is worth mentioning that the specimen was loaded until the lateral
displacement reached approximately 20 mm at which it was assumed that the specimen failed.
It might be noted that a sudden drop of the load-deflection curve occurred at approximately
312.0 kN. This was due to the development of the diagonal crack discussed earlier. The
developing of the diagonal crack affects neither the stiffness nor the ultimate load capacity of
the infilled frame and it will result only in a sudden drift, affecting the overall ductility of the
system, which is outside the scope of this study. The figure shows the capabilities of the
proposed method to predict both the stiffness and ultimate load capacity up to failure. The
model appears to overestimate the ultimate capacity by about 9% and acceptably estimates
the average stiffness up to failure.
Another specimen WD7, tested by Richardson (1986), was modeled using the same technique
utilizing the ANSYS5.3 FE program and the load-deflection relations for the bare and the
infilled frame model are shown in Fig. 7-b along with test results for comparison. Again the
proposed model is efficient in duplicating the test results up to failure. The model
underestimated the failure load by 10% and the experimental test data show that the infilled
frame gradually degrades and eventually at some point it will reach the ultimate capacity of
the bare frame.
It is worth mentioning that the proposed model can predict the stiffness and strength
efficiently up to failure, yet the post peak behavior and the ductility of CMISF systems are
highly uncertain and will require further research, currently being undertaken by the authors
at Drexel University.

Specimen Q21SSB, tested by Mosalam et al. (1997a) was modeled using the same technique;
Fig. 8-a show the load-deflection relation of the bare frame model and the infilled frame
model along with envelope of the cycling loading test after correcting it to exclude the effect
of the lack of fit between the frame and the infill. The model accurately represents the infilled
frame up to a deflection of 6 mm, at which the model underestimated the specimen capacity
by less than 2%. After this displacement, failure occurred in the specimen yet the model
continued to carry more load, but with a very low stiffness, then it gradually loses its strength
and fails. Fig. 8-b shows the bending moments in the model members at a load of 41.5 kN,
before failure. These moments have the same trend as those obtained by Mosalam et al.
(1997b) and suggested by Reflak and Fajfar (1991), Saneinejad and Hobbs (1995),
Buonopane and White (1999), and Furutani et al. (2000).

CONCLUSIONS

This paper presents an analytical method of predicting the stiffness and the ultimate
load capacity of CMISF failing in CC mode. Based on the present study, the following
conclusions can be inferred:

1. The proposed analytical technique predicts the lateral stiffness up to failure, and the
ultimate load capacity of concrete masonry-infilled steel frames (CMISF) to an acceptable
degree of accuracy. The technique accounts for the nonlinear behavior that occurs in both the
steel frame due to formation of plastic hinges, and in the masonry panel due to crushing. The
technique considers the diagonal tension cracking in the masonry joints merely as a
serviceability limit state.
2. The use of three struts instead of a single one is justified based on the observed bending
moments in the frame members, which cannot be generated using a single strut. Furthermore,
the three struts do not fail simultaneously, which is the case in actual infill panels, since the
crushing starts at the corners and keeps propagating in the corner region leading to failure of
the panel. The use of the three struts will also facilitate modeling the interaction between the
different panels in multi-story buildings.
3. The technique presents a macro-model that is more easy and practical to apply and require
much less time than techniques based on treating the panel as a plate, meso-models, or
descretizing the panel as a series of plane stress elements interconnected by a series of springs
or contact elements, micro-models.
4. In order to use this technique in actual multi-story, multi-bay frame structures, three
diagonal struts should replace each infill panel following the steps of the proposed method.
This process can be easily computerized and included into the FE programs used in structural
analysis in order to automatically generate the diagonal struts and place them in their proper
locations with their respective properties.
5. Instead of using the actual nonlinear stress-strain relation, an option, which might not be
available in many structural analysis software, a simplified tri-linear stress-strain relation is
employed for the masonry. A similar relation is also used in modeling the steel frame load-
deformation relation. It is worth mentioning that this simplification results in a less solution
time, specially, in multi-story 3-D structures with large number of DOF.

ACKNOWLEDGEMENTS

The work presented herein is part of a study performed at Drexel University under Grant No.
CMS-9730646 from the National Science Foundation (NSF); Dr. Shih-Chi Liu is the NSF
program director for this research. The results, opinions, and conclusions expressed in this
paper are solely those of the authors and do not necessarily reflect those of the NSF.

REFERENCES

American Society for Testing and Materials, ASTM E 447-92b, Standard Test Methods for
Compressive Strength of Masonry Prisms, Annual Book of ASTM Standards, Vol.
04.05, West Conshohocken, PA.
American Society for Testing and Materials, ASTM E 519-81, Standard Test Method for
Diagonal Tension (Shear) in Masonry Assemblages, Annual Book of ASTM
Standards, Vol. 04.05, West Conshohocken, PA.
ANSYS5.3, ANSYS Inc. 201 Johnson Road, Houston, PA 15342-1300.
Buonopane, S. G., and White, R. N., 1999. Pseudodynamic testing of masonry infilled
reinforced concrete frame, ASCE Journal of Structural Engineering, Vol. 125, No. 6,
pp. 578-589
Chrysostomou, C. Z., 1991. Effect of degrading infill walls on the nonlinear seismic
response of two-dimensional steel frames, Ph.D. thesis , Cornell University, Ithaca,
NY, 1991.
Chen, W. F., and Lui, E. M., 1991. Stability Design of Steel Frames, CRC Press.
Dhanasekar, M., and Page, A. W., 1986. The influence of brick masonry infill properties on
the behavior of infilled frames, Proceedings of the Institution of Civil Engineers, Part
2, Vol. 81, pp. 593-605.
Drysdale, R. G., and Hamid, A. A., 1979. Behavior of concrete block masonry under axial
compression, ACI Journal. Vol. 76, pp. 707-721.
ElDakhakhni, W. W. (2000) Non-Linear Finite Element Modeling of Concrete Masonry-
Infilled Steel Frame, M.Sc. thesis, Drexel University, Philadelphia, PA, 19104.
Flanagan, R. D., Bennett, R. M., and Barclay, G. A., 1992. Experimental testing of hollow
clay tile infilled frames, Proceedings, 6th. Canadian Masonry Symposium, University
of Saskatchewan, Saskatchewan, Canada. pp. 633-644.
Flanagan, R. D., Bennett, R. M., and Barclay, G. A., 1999. In-plane behavior of structural
clay tile infilled frames, ASCE Journal of Structural Engineering, Vol. 125, No. 6,
pp. 590-599.
Furutani, M., Imai, H., and Matsumoto, T., 2000. 12th World Conference on Earthquake
Engineering, Auckland, New Zealand, February 2000, on CD-ROM.
Hamid, A. A. and Drysdale, R. G., 1980. Concrete masonry under combined shear and
compression along the mortar joints, ACI Journal. Vol. 77, pp. 314-320.
Khattab, M. ,and Drysdale, R. G. 1992. Test of concrete block masonry under biaxial
tension-compression, Proceedings, 6th. Canadian Masonry Symposium, University of
Saskatchewan, Saskatchewan, Canada. pp. 645-656.
Liauw, T. C., and Kwan, K. H., 1982. Non-linear analysis of multi-storey infilled frames,
Proceedings of the Institution of Civil Engineers, Part 2, Vol. 73, pp. 441-454.
Mander, J.B and Nair, B. 1994. Seismic resistance of brick-infilled steel frames with and
without retrofit. TMS journal February 1994. pp. 24-37.
McBride, R. T., 1984. The behavior of masonry infilled steel frames subjected to racking,
M.Sc. thesis, Department of Civil Engineering, University of New Brunswick,
Fredericton, NB, Canada.
Mosalam, K., White, R. N., and Gergely, P., 1997a. Seismic Evaluation of Frames with Infill
Walls Using Quasi-static Experiments. Technical report NCEER-97-0019.
Mosalam, K., White, R. N., and Gergely, P., 1997b. Seismic Evaluation of Frames with
Infill Walls Using Pseudo-dynamic Experiments. Technical report NCEER-97-0020.
Mosalam, K., White, R. N., and Gergely, P., 1997c. Computational Strategies for Frames
with Infill Walls: Discrete and Smeared Crack Analyses and Seismic Fragility.
Technical report NCEER-97-0021.
Polyakov, S. V., 1956. Masonry in framed buildings (Godsudarstvenoe Isdatel' stvo
Literatury Po Stroidal stvui Architecture. Moscow, 1956). Translated by G. L. Cairns in
1963.
Richardson, J., 1986. The behavior of masonry infilled steel frames. M.Sc. thesis,
University of New Brunswick, Fredericton, NB, Canada.
Saneinejad A., and Hobbs, B., 1995. Inelastic design of infilled frames. ASCE Journal of
Structural Division, Vol. 121, No. ST4, pp. 634-650.
Seah, C. K. 1998. A universal approach for the analysis and design of masonry infilled
frame structures Ph.D. Thesis, University of New Brunswick, Fredericton, NB.,
Canada.
Shames, I. H., and Cozzarelli, F. A., 1992. Elastic and Inelastic Stress Analysis, Prentice
Hall , Englewood Cliffs, New Jersey 07632.
Stafford Smith, B., and Carter, C., 1969. A method of analysis for infilled frames.
Proceedings, the Institution of Civil Engineers, Vol. 44, pp. 31-48.
Yong, T. C., 1984. Shear strength of masonry panels in steel frames. M.Sc. thesis,
University of New Brunswick, Fredericton, NB, Canada.
 P P

(a ) C C m o d e (b ) S S m o d e

P P

(c ) D C m o d e (d ) D K m o d e

(e ) F F m o d e

Figure 1. Different Failure Modes of Masonry Infilled Frames:

(a) Corner Crushing Mode; (b) Sliding Shear Mode; (c) Diagonal Compression Mode
(d) Diagonal Cracking Mode; (e) Frame Failure Mode

(a) (b)

Figure 2. The Diagonal Tension Specimen : (a) ASTM E-519Test Setup; (b) Shear Stress
Contours and Failure Mode Obtained Using The ANSYS5.3 FE Model
 E F A B

G H C D
E
F

G Plastic hinges
H

A Diagonal crack
B
P
C
D Region 1
Contact lengths Region 2

Contact lengths

Figure 3. The Infill Panel Separation into Two Diagonal Regions, and The Resulting Bending
Moment Diagrams for a Different Bays in Multi-Story Infilled Frame Building

CONTACT12 Element

BEAM3 Element
P

PLANE42
Element

(a) (b)

Figure 4. ANSYS5.3 FE Model of a Single panel Infilled Frame: (a) Schematic Diagram;
(b) Principal Stress Contours
 b l beam-column joint
A1=A/4

c h A2=A/2

h
A1=A/4
column beam
Moment
M pj
l

el pl ult Rotation

Figure 5. The Proposed CMISF Model and The Beam-Column Connection Behavior

fm -90
' , E9 0

fm -' , E

fm -0' , E0

(a)
Stress F orce
E Ep
fm -' Fu

1 p 2
u
Strain
1 2 D efor m ation
u

(b) (c)

Figure 6. Concrete Masonry Behavior : (a) Orthotropic Model

(b) Tri-Linear Stress-Strain Relation; (c) Typical Force-Deformation Relation for Struts
 700 600
Infilled
Infilled Frame
600 Frame
500 Model
Model
Test
Test Results
500 Results
Bare
Bare 400 Frame
Frame
Load (kN)

Load (kN)
400
300
P
300
P
200
200

100 100

0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60 70
Deflection (mm) Deflection (mm)

(a) (b)

Figure 7. Load-Deflection Relations for Specimen : (a) WB2, (b) WD7

50
45 Infilled
Fram e
40 M odel
C yclic
Test
35 Env.
Bare
Load (kN)

30
4 1 .5 k N
25
20

15 P

10
5
M om en t S c ale 0 0 .5 1 .0 2 .0
0
0 4 8 12 16 20 24
Deflection (mm)

(a) (b)

Figure 8. Specimen Q21SSB : (a) Load-Deflection Relations, (b) Bending Moment Diagram
obtained using ANSYS5.3 (drawn on the tension side)