Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/245285741

Modeling Masonry Shear-Compression: Role of Dilatancy Highlighted

Article  in  Journal of Engineering Mechanics · November 2004


DOI: 10.1061/(ASCE)0733-9399(2004)130:11(1289)

CITATIONS READS
73 813

1 author:

Gideon van Zijl


Stellenbosch University
212 PUBLICATIONS   1,251 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Stephan Zeranka BEng 2010 View project

Durability of lightweight foamed concrete View project

All content following this page was uploaded by Gideon van Zijl on 02 April 2016.

The user has requested enhancement of the downloaded file.


Modeling Masonry Shear-Compression:
Role of Dilatancy Highlighted
Gideon P. A. G. van Zijl1

Abstract: In-plane shear and compression are important modes in masonry walls, both in load bearing structures and in framed
structures. By these mechanical actions shear forces caused by wind and earthquakes are resisted. Upon shear-slipping along bed joints,
brick units in masonry also undergo upward translation, or dilatancy, causing global volume increase. If this dimensional change is
prevented, large compressive stresses may build up, increasing the resistance to slip by the Coulomb-friction nature. If this shear-
compression interaction is not modeled correctly, large errors may be made. In the extreme case, unlimited shear resistance may be
predicted by computational models. A discrete crack modeling approach for masonry, which captures the shear-compression dilatational
behavior realistically, is elaborated. Shear-compression experiments on small masonry specimens as well as 1 m⫻1 m masonry walls are
analyzed as validation.
DOI: 10.1061/(ASCE)0733-9399(2004)130:11(1289)
CE Database subject headings: Masonry; Walls; Constitutive models; Dilatancy; Compression; Shear forces.

Introduction to propose, develop, and implement computational strategies and


finite element models for structural masonry. Computational mod-
Shear failure is an important, often governing mode of failure in els based in continuum plasticity (Lourenço et al. 1998) and con-
buildings, which are laterally loaded by wind and earthquake ac- tinuum damage (Papa et al. 2000; Addessi et al. 2002) have been
tion. The shear acts in combination with compression, which is formulated and implemented for in-plane behavior of masonry. In
caused by self-weight and floor loads. Confinement by, for in- all these cases plane stress was assumed. In many cases, however,
stance, structural frames to in-fill walls may also lead to shear- the size of the masonry units cannot be considered small in com-
compression. In frictional, cementitious materials like concrete, parison with the structural size, rendering a continuum approach
soils, or masonry, dilatational behavior upon shearing, as opposed invalid. In these cases a discrete modeling strategy is appropriate
to isochoric plasticity of metals and plastics, is a well-known (Page 1978; Stankowski et al. 1993; Lourenço and Rots 1997;
Giambanco et al. 2001). The bricks are modeled with standard
phenomenon. It has been reported as early as in the 19th century
continuum finite elements, while interface elements capture the
by Reynolds (1885), discussed comprehensively more recently by
joint behavior. Usually, the nonlinear behavior is restricted to the
Vermeer and de Borst (1984), and has been observed in soils by
joints, which represent the weak links in masonry. The interface
amongst others, Roscoe et al. (1963) and Rowe (1971), in con-
constitutive models capture tensile bond failure, as well as shear-
crete (Green and Swanson 1973), rock (Michelis 1981), as well as
slipping along the joints, while a compressive limit function was
in masonry experiments (Van der Pluijm 1992, 1998). Volume
included by Lourenço and Rots (1997). Plane stress is assumed
increase accompanies inelastic shearing deformation. If such vol-
for the interface models employed for two-dimensional analysis
ume increase is prevented or resisted by confining structural ele-
of masonry in-plane behavior. Constitutive laws for plane inter-
ments, wedging occurs, causing pressure buildup. In the case of faces, applicable in three-dimensional analysis of structural ma-
pressure-dependent strength, which is a well-known characteristic sonry, were formulated by Van Zijl (2000) and Giambanco et al.
of the mentioned dilatational materials, a significant strength in- (2001).
crease may accompany such confined shearing. However, none of the interface models capture the shearing
In this paper, a computational model to capture this phenom- dilatation along masonry interfaces appropriately. The models by
enon accurately in structural masonry is described. Since the late Lourenço and Rots (1997) and Giambanco et al. (2001) address
1970s masonry researchers have made a significant research effort this phenomenon. Both models are cast in a nonassociative plas-
ticity framework, which allows the prescription of a dilatancy
1
Professor of Civil Engineering, Faculty of Engineering, Univ. of coefficient different from the friction coefficient. Their models
Stellenbosch, Private Bag X1, Matieland 7602, South Africa. E-mail: provide for a variable dilatancy coefficient, governed by shear-
gvanzijl@sun.ac.za. Faculty of Architecture, Delft Univ. of Technology, slipping displacement along the interface. In the case of Lourenço
The Netherlands. and Rots (1997) a pressure dependence is also considered,
Note. Associate Editor: Franz-Josef Ulm.. Discussion open until April whereby the dilatancy coefficient is reduced with increased nor-
1, 2005. Separate discussions must be submitted for individual papers. To
mal pressure. However, neither of these dilatancy formulations
extend the closing date by one month, a written request must be filed with
the ASCE Managing Editor. The manuscript for this paper was submitted reproduce the volume increase, or uplift upon shearing along a
for review and possible publication on June 20, 2002; approved on masonry interface, measured in shear experiments (Van der Plu-
February 12, 2004. This paper is part of the Journal of Engineering ijm 1992, 1998; Vermeltfoort et al. 1993). Inappropriate dilatancy
Mechanics, Vol. 130, No. 11, November 1, 2004. ©ASCE, ISSN 0733- modeling has the potential of predicting unconservatively high
9399/2004/11-1289–1296/$18.00. resistance to confined shear. This has inspired the research re-

JOURNAL OF ENGINEERING MECHANICS © ASCE / NOVEMBER 2004 / 1289


Fig. 3. Top: Discrete modeling strategy for masonry. Bottom:
Interface constitutive model [based on data from Lourenço and Rots
Fig. 1. Shear test for characterizing masonry joint shear behavior (1997)].
[based on data from Van der Pluijm (1992, 1998)]

same order as the shear-slipping deformation. This highlights di-


ported here. The formulation and implementation of an interface
latancy as a potential source of significant inelastic volume in-
material law based in computational plasticity is discussed. A
crease in masonry shear walls. However, it can be seen that the
variable dilatancy coefficient is incorporated to reproduce experi-
dilatancy depends on the confining pressure, with significantly
mental measurements of brick normal uplift during shearing
reduced uplift with increased confining pressure. Also, at large
along a brick–mortar interface. The formulation is verified by
shear-slip deformation, the uplift is arrested and in some cases
analyzing small masonry shear experiments where the dilatant
even reversed, indicating a “smoothening” effect on the bed joint–
volume increase was prevented. This was done to test the ability
brick interface.
of the model to capture the pressure buildup and shearing resis-
tance increase accurately. Finally, large shear wall experiments
are analyzed to validate the model.
Discrete Model

Knowing that the joints in masonry form the weak link, a discrete
Dilatant Behavior of Masonry
modeling strategy can be employed by including predefined, po-
tential cracks at the joints in the form of interface elements. These
Small masonry shear experiments were performed by Van der
interface elements then capture cracking behavior, while the
Pluijm (1992, 1998) to characterize the Coulomb-friction charac-
bricks behave linear elastically, or linear viscoelastically if creep
ter of masonry shearing resistance along bed joints—Fig. 1.
is considered. This strategy can be extended to include vertical
In these shear experiments normal force control was applied to
(potential) cracks through the bricks as well. The discrete model-
maintain constant normal pressure 共␴兲 during the (displacement-
ing strategy is shown schematically in Fig. 3.
controlled) shearing. Thereby, the specimen edges could translate
A plane stress interface model was formulated by Lourenço
in the direction normal to the shearing deformation. This uplift, or
and Rots (1997). It is based on multisurface plasticity, comprising
displacement normal to the shear-slipping bed joint, was also
a Coulomb friction model combined with a tension cutoff and an
measured. The ranges in uplift are shown in Fig. 2 to be of the
elliptical compression cap, Fig. 3. Softening acts in all three
modes, governed by tensile fracture energy GIf , shearing fracture
energy GIIf , and compressive fracture energy Gc, as elaborated by
Lourenço and Rots (1997). The softening degradation is preceded
by hardening in the case of the cap mode.
This model is enhanced here to capture the observed dilatant
behavior of masonry joints, Fig. 2. The interface model is derived
in terms of the traction and interface opening displacement vec-
tors

T = 兵␴␶其T
共1兲
U = 兵u␯其T
with ␴ and u⫽stress and relative opening displacement, respec-
tively, in the interface normal direction and ␶ and ␯⫽shear stress
Fig. 2. Dilatant displacement normal to the joint upon inelastic shear and relative opening displacement, respectively.
of masonry [based on data from Van der Pluijm (1992)] The inelastic constitutive behavior is given by

1290 / JOURNAL OF ENGINEERING MECHANICS © ASCE / NOVEMBER 2004


with
Ṫ = D共U̇ − U̇ p兲 共2兲 up =
⌿0

冓 冔
1−

␴u
共1 − e−␦␯p兲 共11兲

It has three parameters, namely, the dilatancy at zero normal con-


D = diag关knks兴 共3兲
fining stress and shear-slip 共⌿0兲, confining (compressive) stress at
the elastic stiffness matrix containing the interface normal stiff- which the dilatancy becomes zero 共␴u兲, and the dilatancy shear-
ness kn and shear stiffness ks. U̇ p denotes the plastic opening slip degradation coefficient 共␦兲. These parameters can be obtained
velocity vector. Here, the plastic opening in the interface normal by, for instance, least-squares fitting to experimental data. The
direction may be interpreted as the crack width and the plastic second step to derive the dilatancy equation is to differentiate Eq.
(11) and rearrange the result to yield

冓 冔
opening displacement in the tangent direction as the inelastic
shear-slipping. The plastic opening velocity is given by the flow ␴ −␦␯
rule in standard plasticity fashion ⌿ = ⌿0 1 − e p 共12兲
␴u
⳵g
U̇ p = ␭˙ 共4兲 This formulation captures the smoothening effect of confining
⳵T pressure, as well as large shear-slipping deformation along an
where ␭⫽amount of cracking. The traction derivative of the plas- interface, see Fig. 2.
tic potential function g defines the cracking direction. Inelastic
shearing, or shear-slipping, is initialized if the Coulomb-friction
limit function Thermodynamic Consistency
f = 兩␶兩 + ␴⌽ − c 艋 0 共5兲 To study limitations of the model, the thermodynamics approach
is breached, where c⫽adhesion of the brick–mortar interface; for a surface of discontinuity proposed by Coussy (1995) is fol-
⌽⫽friction coefficient; and ␬˙ = equivalent plastic opening rate, lowed. The surface energy dissipation rate for this model can be
defined by written as (Coussy 1995; Désir et al. 1999):

␬˙ = ␭˙ 共6兲 冉
␾ = ␭˙ TT
⳵g
⳵T
+h
⳵f
⳵h
冊艌0 共13兲
Note that both the coefficient of friction and the adhesion de-
crease exponentially with increased plastic strain ␬, with the rate where h⫽hardening defined by
of decrease governed by fracture energy (Lourenço and Rots
1997). As an example, the adhesion softening is defined as 冋 冉 冊册
h = c0 1 − exp −
c0
GIIF
␬ 共14兲

c = c0 exp − 冉 冊 c0
GIIF
␬ 共7兲 which follows from rewriting Eq. (5) as
f = 兩␶兩 + ␴⌽共␴,␬兲 − c0 + h共␬兲 艋 0 共15兲
where c0⫽initial, virgin adhesion. The plastic potential function
governs the cracking direction. A suitable function is given by Eq. (13) can be elaborated, cf. Eqs. (8) and (15), to read

g = 兩␶兩 + ␴⌿ − c 共8兲 ␾ = ␭˙ 共兩␶兩 + ⌿␴ + h兲 艌 0 共16兲


where ⌿⫽dilatancy coefficient. From Eqs. (4) and (8) it follows Dissipation occurs during active inelasticity, i.e., for f = 0. By con-
that sidering this equality in Eq. (15) and substituting it in Eq. (16),
the expression for the surface energy dissipation rate becomes
u̇ p
⌿= sgn共␶兲 共9兲
␯˙ p ␾ = ␭˙ 关c0 + 共⌿ − ⌽兲␴兴 艌 0 共17兲
which provides a way of formulating a dilatancy coefficient from Restrictions to the dilatancy Eq. (12) are implied by Eq. (17) in
experimental data as shown in Fig. 2. By integration of Eq. (9) the order to satisfy the second law of thermodynamics. Since ␭˙ 艌 0 it
shear-slip induced normal uplift is found to be follows that

up = 冕 ⌿d兩⌬␯ p兩 共10兲 共⌽0 − ⌿兲␴ 艋 c0


Notice in Eq. (12) that ⌿ is non-negative and varies between ⌿0
共18兲

which is the link to derive a suitable dilatancy equation. First, an and zero. For masonry, the maximum mobilized dilatancy coeffi-
equation is chosen to describe the inelastic normal displacement cient in this range, ⌿0, is less than the friction coefficient ⌽,
upon shearing. In Fig. 2 it is seen that the normal displacement rendering the factor between brackets on the left-hand side of Eq.
component depends on the confining stress and the shear-slip. A (18) positive. Therefore to enforce the inequality in Eq. (18) the
linear stress-dependence is adopted, which gives a reasonable re- normal stress must be restricted to ␴ 艋 c0 / ⌽. This is done by
flection of the decrease in dilatancy with increased normal pres- restricting the tensile strength to f t 艋 c0 / ⌽, which is usually the
sure. Also, a more sophisticated relation is not justified, as only case for masonry (Van der Pluijm 1998).
three levels of confining pressure were employed in the shear
experiments. In Fig. 2 it can be seen that, for a given, constant
confining pressure, the rate of inelastic normal displacement de- Verification of Dilatancy Model
creases exponentially with inelastic shearing displacement. From
these observations, a description of the normal uplift upon shear- To confirm that the dilatancy formulation embedded in the dis-
slipping has been formulated as crete modeling approach captures the observed normal uplift

JOURNAL OF ENGINEERING MECHANICS © ASCE / NOVEMBER 2004 / 1291


Fig. 5. Comparison of the measured (Van der Pluijm 1992, 1998) and
numerical dilatant normal displacement upon shearing displacement
along the joint of (a) clay brick masonry; and (b) calcium silicate
masonry
Fig. 4. Masonry shear tests (Van der Pluijm 1992, 1998), showing
finite element mesh for reanalysis
are calculated from the measurements at the gauge points and the
elastic properties of the bricks (Van der Pluijm 1992). The agree-
upon shearing along a masonry joint, the shear experiments on ment is not surprising because the model parameters have been
small masonry specimens by Van der Pluijm (1992, 1998) have calculated from these experimental responses. Nevertheless, it
been reanalyzed—Fig. 4. The parameters shown in Table 1 have shows that the chosen form of the dilatancy equation is appropri-
been employed and were obtained from reported results on small ate.
masonry experiments by Van der Pluijm. Also, reported average
Young’s moduli of E = 17.4 and 13.4 kN mm−2 for the clay bricks
Confined Shearing Verification
and calcium silicate units, respectively, as well as Poisson’s ratio
␯ = 0.2 were employed. The interface stiffnesses kn, ks have been A more severe test of the model is presented by the confinement
calculated to match the average observed elastic responses of the of the dilatancy and the associated normal stress buildup. Van der
two types of masonry. Fig. 4 also shows the finite element mesh Pluijm (1998) modified his experimental setup to control the dis-
used for the analyses. placement normal to the joint. He performed shear tests on two
The experimental and numerical dilatancy responses are com- specimens of the clay brick masonry type, Table 1. In the first test
pared in Fig. 5 for the two types of masonry investigated. Note he applied an initial normal displacement to cause an average
that the joint normal and shear displacements are shown, which (compressive) stress of −0.1 N mm−2. The normal boundaries

Table 1. Parameters Employed for the Interfaces in Clay Brick and Calcium Silicate Masonry
Symbol Clay brick Calcium silicate Definition
kn 共N mm 兲
−3
825 438 Normal stiffness
ks 共N mm−3兲 345 182 Shear stiffness
f t 共N mm−2兲 0.4 0.1 Tensile strength
GIf 共N mm−1兲 0.012 0.005 Tensile fracture energy
c0 共N mm−2兲 0.87 0.28 Adhesion
GIIf 共N mm−1兲 0.006– 0.09␴ 0.02– 0.03␴ Shear fracture energy
⌽0 1.1 0.97 Initial friction coefficient
⌽r 0.70 0.75 Residual friction coefficient
⌿0 0.74 0.67 Initial dilatancy coefficient
␴u 共N mm−2兲 −1.57 −1.22 Stress at which dilatancy is zero
␦ 5.6 17 Dilatancy softening gradient

1292 / JOURNAL OF ENGINEERING MECHANICS © ASCE / NOVEMBER 2004


Fig. 7. Response to inappropriate dilatancy modeling

racy with the discrete strategy, employing the variable dilatancy


formulation. Furthermore, a limit to the pressure buildup is found
Fig. 6. (a) Schematical setup of the normal confined masonry
well below the crushing stress of the masonry.
shear tests (Van der Pluijm 1998). Comparison of experimental and
Inappropriate shearing dilatancy modeling with the discrete
numerical (b) shear force versus normal reaction force and (c) shear
approach can lead to large errors. This is illustrated in Fig. 7 by
force versus joint shear displacement.
employing various sets of the dilatancy model parameters. For
even a small, constant dilatancy coefficient an unlimited strength
is predicted for the shear specimen of Fig. 6(a). A dilatancy co-
efficient of zero reproduces the unconfined shearing response.
were then fixed in this position, before the displacement- This forms a lower limit to the confined shearing resistance, but is
controlled shearing was commenced, as shown schematically in overly conservative in this particular case.
Fig. 6(a). The same procedure was followed for the second test,
except that an initial average tensile stress, ␴0 = + 0.1 N mm−2 was
applied instead of the compression in the first test. Validation: Shear Wall Analyses
In Fig. 6(b) the normal force buildup upon shearing is shown.
A significant difference can be seen in the measured shear force To validate the dilatancy model, masonry shear wall experiments
evolution for the two tests. The vertical shift is not captured by have been analyzed. Full details of the experiments can be found
the computational model, as shown by the coinciding solid lines in Vermeltfoort et al. (1993). Three identical walls, 100 mm thick
depicting the numerical results. A possible explanation is the large and 1 m square, with an off-center opening of one brick wide and
spread in shearing dilatancy of masonry materials, indicated by six layers high, were tested in shear. The wall and its experimen-
the hashed areas in Fig. 5. Also, shear test data under normal tal setup are shown schematically in Fig. 8. Each wall was rigidly
compressive stress were employed to derive the dilatancy formu- connected to a steel beam along the lower edge, as well as along
lation in the previous section. To characterize the dilatancy in the the upper edge. These horizontal beams were mounted in a large
tensile-shear regime, shear experiments under constant normal steel frame. The horizontal force was applied to each wall by a
tension should be performed. Currently, Eq. (12) assumes a horizontally mounted actuator, Jack 1 in the figure, pressing
stress-independent dilatancy in the tensile regime. This is only the against the top beam. Instead of a constant vertical load, three
case while mode I separation has not yet occurred. Through vertical actuators, Jacks 2–4, were used to control the upper beam
tensile-shear coupling (Lourenço and Rots 1997), the interface position. After an initial vertical precompression with the three
model prescribes zero shearing resistance and, therefore, zero vertical jacks, calculated to cause an average compressive stress
shearing dilatancy once mode I resistance has been lost. of 0.3 N mm−2 in the masonry wall, they were controlled to allow
Van der Pluijm limited the normal force to 27.5 kN to protect
the test apparatus, at which point he switched back to force con-
trol of the normal boundary to sustain the limit force. This has
been simulated numerically, Fig. 6(b). However, the case of un-
restricted normal force buildup has also been analyzed. These
numerical responses are shown with dotted lines. Due to the
smoothing of the interface, a point is reached where no further
dilatancy occurs and the normal force is arrested. This point co-
incides with the pressure at which the dilatancy becomes zero, ␴u.
The measured response for the case of initial compression 共␴0 =
−0.1 N mm−2兲 indicates that a limit point was approached just
before the switch to force control was made, which confirms the
numerical result.
In Fig. 6(c) the numerical and experimental shear force-
deformation responses are compared. Reasonable agreement is
found.
These experiments confirm the confined dilatant shear strength Fig. 8. Shear wall experimental setup, showing horizontal measure-
enhancement in masonry. This is captured with reasonable accu- ment locations Mits 1 and 7

JOURNAL OF ENGINEERING MECHANICS © ASCE / NOVEMBER 2004 / 1293


Table 2. Discrete Model Parameters Employed in Shear Wall Analysis
Symbol Definition
−3
kn 83 N mm Normal stiffness
ks 36 N mm−3 Shear stiffness
ft 2.25 N mm−2 Tensile strength
GIf 0.018 N mm−1 Tensile fracture energy
c0 0.35 N mm−2 Adhesion
⌽ 0.75 Friction coefficient
⌿0 0.6 Initial dilatancy coefficient
␴u −1.3 N mm−2 Stress at which dilatancy is zero
␦ 5 Dilatancy softening gradient
GIIf 0.05 N mm−1 Shear fracture energy
fc 8.5 N mm−2 Compressive strength
Gc 5.0 N mm−1 Compressive fracture energy
␬p 0.093 Compressive plastic strain at f c

no further vertical translation. Once the vertical confinement was


in place, the horizontal jack was activated to displace the upper
beam at 10 mm h−1.
The intention was to install a fully confined condition. How-
ever, the large dilatational wedging forces during shearing caused
flexure of the beams, which was unfortunately not measured. To
simulate these experimental boundary conditions, the upper and
lower steel beams have been included in the model. At the con-
nection points with the vertical jacks, as well as the ground bear-
Fig. 9. Shear wall numerical response. Top: Frame stiffness influence
ings, the displacement has been controlled computationally. Also,
on global response. Bottom: Influence of creep behavior.
the horizontal displacement rate of the upper beam has been con-
trolled. The steel beams have been considered to act elastically,
with Young’s modulus 210 kN mm−2 and Poisson’s ratio 0.3. comparison with the computational results. The measured hori-
The required parameters for analyzing the experiments have zontal load-drift responses for the three walls are shown in Fig.
been obtained from separate characterizing experiments (Vermelt- 10. The less ductile behavior of Wall J1G was due to separation of
foort et al. 1993). Compressive tests on masonry specimens were some of the bricks in the top layer from the upper beam. After the
performed to obtain the masonry compressive strength and experiment, closer investigation during breaking down the speci-
Young’s modulus. Tensile tests were performed to measure the men confirmed poor filling of the bonding material in those re-
joint bond as well as fracture energy. Shear-slip parameters have gions (Vermeltfoort et al. 1993). The difference in the responses
been obtained from shear test results on masonry of the same clay of Walls J2G and J3G is attributed to the different mobilized
brick type and mortar joints, performed by Van der Pluijm (1992). failure mechanisms, as the crack patterns show in Fig. 11. Nev-
The model parameters for the discrete modeling strategy, found in ertheless, the coefficient of variation in peak horizontal resistance
this way, are tabulated in Table 2. Note that the indicated interface of 15% is within the expected range for masonry. The sliding
normal and shear stiffness have been calculated to give the mea- failure along a horizontal crack in Wall J3G caused lower, but
sured, average masonry Young’s modulus of 8 kN mm−2 when tougher resistance to shear loading than the stepped failure ob-
acting together with the bricks, for which the average measured
Young’s modulus of 16.7 kN mm−2 and Poisson’s ratio of 0.15
have been prescribed.
The need to include the steel beams in the model is confirmed
in Fig. 9(a). A significant influence of the finite frame stiffness on
the horizontal force-drift response has been computed. Further-
more, the influence of time dependence on the wall response has
been studied by activating viscoelastic behavior in the brick units
along the lines described by Van Zijl et al. (2001a). Typical ma-
sonry creep data, with an effective creep coefficient of 0.85 after
1 day has been employed. In Fig. 9(b) the wall response is shown
for three different durations of the experiment, ranging from 0 to
10 h. Note that the experimental shearing displacement rate of
10 mm h−1 resulted in a duration of about 2 h per test. Within this
range, no significant influence is computed. Therefore creep has
been omitted from the subsequent analyses.
Only two of several deformation measurements are indicated
in Fig. 8, namely, the horizontal drift of the wall (Mit 1) and that
of the upper beam (Mit 7). The drift measured by Mits 1 and 7, Fig. 10. Experimental horizontal force-drift (Mit 1) shear wall re-
the horizontal force, and the crack pattern are employed here for sponses

1294 / JOURNAL OF ENGINEERING MECHANICS © ASCE / NOVEMBER 2004


Fig. 11. Shear wall experimental failure modes

served in Wall J2G. In the analyses, no attempt has been made to


seed a particular mode of failure by, for example, modeling ma-
terial or geometrical imperfections. By considering “perfect” ma-
terial, the stepped crack pattern shown in Fig. 12, which agrees
with the observed crack pattern of Wall J2G, has been computed.
All three failure modes have been activated in the analysis. In
the stepped crack, tensile failure/cracking occurs along the verti-
cal joints, while shear-slipping occurs along the horizontal joints.
In four regions inelastic compressive deformation evolves, indi-
cated by the contours of algebraic minimum principal stress su-
perposed on the deformed wall in Fig. 12. In Fig. 13(a) the nu-
merical and experimental force-drift responses for Wall J2G are
shown. Note that both the Mits 1 and 7 measurements are shown
to explain the discrepancy in postpeak experimental and compu-
tational responses. A possible explanation for the sudden stop in
horizontal displacement measured by Mit 1, as opposed to the
continued drift by the frame (Mit 7), is the disturbing influence by Fig. 13. Shear Wall J2G experimental and numerical responses. Left:
the cracked zone on the local deformation of the half brick in the Horizontal load-drift response measured on frame (Mit 7) and wall
right upper corner, i.e., where this drift measurement was made. (Mit 1). Right: Adjusted load-drift response on wall [based on data
An adjustment to the Mit 1 measurement, by forcing it to follow from Van Zijl et al. (2001b)].
the initial relation with the Mit 7 measurement (Van Zijl et al.
2001b), is shown in Fig. 13(b). Reasonable agreement is seen
between the adjusted experimental and the numerical response. shearing, and pressure buildup if it is confined, has been cited. In
The role of dilatancy in the shear wall response has been stud- the confined shearing case, a large increase in peak resistance is
ied by repeating the analysis for two constant dilatancy coeffi- observed, driven by the dilatational wedging and subsequent
cients of ⌿ = 0 and 0.4, respectively. It can be seen that a conser- pressure-dependent strength increase. However, the dilatancy is
vative global response may be predicted by employing zero arrested at high pressure and at large shearing deformation,
dilatancy, Fig. 14. However, with the latter, relatively low, but whereby the strength enhancement is limited.
constant dilatancy coefficient, a 50% overestimation of the peak A computational model to simulate masonry shearing dila-
resistance is computed. Nevertheless, the computed crack pattern tancy has been elaborated. The dilatancy formulation is incorpo-
remains as shown in Fig. 12. rated in the Coulomb-friction part of an interface constitutive law,
which is set in a multisurface plasticity framework. Observed
dilatancy dependence on the confining pressure, as well as shear-
Conclusions slipping along masonry joints is considered. It has been shown
that the model captures the phenomenon of shearing dilatancy
The phenomenon of shearing dilatancy has been discussed. Ex-
perimental evidence of masonry inelastic volume increase upon

Fig. 12. Computed shear wall deformations at 3 mm (left) and 5 mm Fig. 14. Computed sensitivity to discrete dilatancy modeling of shear
(right) drift wall force-drift response

JOURNAL OF ENGINEERING MECHANICS © ASCE / NOVEMBER 2004 / 1295


accurately. In unconfined shear the volume increase measured in model for masonry: Parameter estimation and validation.” J. Eng.
masonry shear experiments is reproduced. In the case of confined Mech., 124(6), 642–652.
shear, where the normal uplift during shearing is prevented by the Michelis, P. N. (1981). “Work-softening and hardening behaviour of
boundary condition, the experimentally measured normal force granular rocks.” Rock Mech., 14, 187–200.
Page, A. W. (1978). “Finite element model for masonry.” J. Struct. Div.
buildup and accompanied strength and toughness increase of ma-
ASCE, 104(8), 1267–1285.
sonry are captured by the model. The experimental and computa-
Papa, E., Taliercio, A., and Mirabella-Roberti, G. (2000). “A damage
tional characterization of masonry dilatancy in the tension-shear model to predict the behaviour of masonry under sustained load.”
region remain to be addressed. Proc., 12th Int. Brick/Block Masonry Conf., Madrid, Spain, 1777–
The effects of inappropriate dilatancy modeling have been 1790.
demonstrated. A constant, nonzero dilatancy coefficient has been Reynolds, O. (1885). “On the dilatancy of media composed of rigid par-
shown to lead to the computation of unlimited resistance to con- ticles in contact.” Philos. Mag., 20, 469.
fined shear. This is the case for even a small dilatancy coefficient. Roscoe, K. N., Schofield, A. N., and Thurairajah, A. (1963). “Yielding of
However, a zero dilatancy coefficient, adopted as a pragmatic clays in states wetter than critical.” Geotechnique, 13(3), 211–240.
solution to avoid such unconservative prediction of shearing re- Rowe, P. W. (1971). “Theoretical meaning and observed values of defor-
sistance, may be extremely conservative. mation parameters for soil.” Proc., Roscoe Memorial Symposium on
The model has been validated by the analysis of shear wall Stress-Strain Behavior of Soils, Cambridge, 143–194.
experiments. Also for this case study, the role of dilatancy has Stankowski, T., Runesson, K., and Sture, S. (1993). “Fracture and slip of
been highlighted, confirming the danger of overestimation of ma- interfaces in cementitious composites. I: Characteristics.” J. Eng.
Mech., 119(2), 292–314.
sonry shear wall resistance by employing a constant dilatancy
Van der Pluijm, R. (1992). “Deformation controlled shear tests on ma-
coefficient. sonry.” Rep. BI-92-104, TNO Building and Construction, Delft, The
Netherlands (in Dutch).
Van der Pluijm, R. (1998). “Overview of deformation controlled com-
References bined tensile and shear tests.” Rep. TUE/CCO/98.20, Eindhoven Univ.
of Technology, Eindhoven, The Netherlands.
Addessi, D., Marfia, S., and Sacco, E. (2002). “A plastic nonlocal damage Van Zijl, G. P. A. G. (2000). “Computational modelling of masonry creep
model.” Comput. Methods Appl. Mech. Eng., 191, 1292–1310. and shrinkage.” Dissertation, Delft Univ. of Technology, The Nether-
Coussy, O. (1995). Mechanics of porous continua, Wiley, Chichester. lands.
Désir, J. M., Ben Romdhane, M. R., Ulm, F.-J., and Fairbairn, E. M. R. Van Zijl, G. P. A. G., De Borst, R., and Rots, J. G. (2001a). “A numerical
(1999). “Steel-concrete interface: Revisiting constitutive and numeri- model for the time-dependent cracking of cementitious materials.” Int.
cal modeling.” Comput. Struct., 71(5), 489–503. J. Numer. Methods Eng., 52(7), 637–654.
Giambanco, G., Rizzo, S., and Spallino, R. (2001). “Numerical analysis Van Zijl, G. P. A. G., Rots, J. G., and Vermeltfoort, A. T. (2001b). “Mod-
of masonry structures via interface models.” Comput. Methods Appl. elling of shear-compression in Masonry.” CDROM-Proc. 9th Cana-
Mech. Eng., 190, 6493–6511. dian Masonry Symposium, P. H. Bischoff, J. L. Dawe, A. B. Schriver,
Green, S. J., and Swanson, S. R. (1973). “Static constitutive relations for and A. J. Valsangkar, eds., Fredericton, Canada.
concrete.” AFWL-TR-72-244 Air Force Weapons Laboratory, Kirtland Vermeer, P. A., and De Borst, R. (1984). “Non-associated plasticity for
Air Force Base, N.M. soils, concrete and rock.” Heron, 29(3), 1–64.
Lourenço, P. B., and Rots, J. G. (1997). “Multisurface interface model for Vermeltfoort, A. Th., Raijmakers, Th. M. J., and Janssen, H. J. M. (1993).
analysis of masonry structures.” J. Eng. Mech., 123(7), 660–668. “Shear tests on masonry walls.” Proc., 6th North American Masonry
Lourenço, P. B., Rots, J. G., and Blaauwendraad, J. (1998). “Continuum Conf., 1183–1193.

1296 / JOURNAL OF ENGINEERING MECHANICS © ASCE / NOVEMBER 2004

View publication stats

You might also like