Académique Documents
Professionnel Documents
Culture Documents
ARTICLE
Experimental testing of slender load-bearing masonry walls
with realistic support conditions
Clayton Pettit, Erum Mohsin, Carlos Cruz-Noguez, and Alaa Elwi
Abstract: Slender, load-bearing masonry walls with slenderness ratios (h/t) greater than 30 are required to be designed as
Can. J. Civ. Eng. Downloaded from cdnsciencepub.com by UNIV CALGARY on 03/15/22
pinned-pinned elements according to North American provisions for masonry, CSA S304-14 and TMS 402-16. This provision
neglects the contribution of the reactive stiffness of the foundation to the strength of the wall and its effect on the redis-
tribution of bending moments along its height. Eight full-scale masonry walls were built with different degrees of base stiff-
ness and tested under an eccentric axial load. The results of the tests showed an increased load-bearing capacity and
decreased deflections with increased rotational base stiffness. Experimental data were used to determine key design param-
eters, including the effective flexural rigidity and moment distribution along the height of the walls. Comparing the values
of effective flexural rigidity determined from experimental results to code provisions, it was found that both codes tend to
underestimate the effective flexural rigidity of the walls.
Key words: masonry wall, slender, out-of-plane, full-scale testing, base stiffness.
Résumé : Les murs en maçonnerie minces et porteurs dont le rapport d’élancement (hauteur/épaisseur, h/t) est supérieur à
30 doivent être conçus comme des éléments articulés conformément aux dispositions nord-américaines relatives à la
maçonnerie, à la norme CSA S304-14 et à la norme TMS 402-16. Cette disposition ne tient pas compte de la contribution de
la rigidité réactive de la fondation à la résistance du mur et son effet sur la redistribution des moments de flexion le long de
For personal use only.
sa hauteur. Huit murs en maçonnerie en grandeur réelle ont été construits avec différents degrés de rigidité de la base et
mis à l’essai sous une charge axiale excentrée. Les résultats des essais ont montré une portance accrue et une diminution
des déformations avec une rigidité de base de rotation accrue. Des données expérimentales ont été utilisées pour déter-
miner les paramètres clés de conception, y compris la rigidité de flexion efficace et la répartition des moments le long de la
hauteur des murs. En comparant les valeurs de rigidité de flexion effective déterminées à partir des résultats expérimen-
taux aux dispositions du code, on a constaté que les deux codes tendent à sous-estimer la rigidité à la flexion effective des
murs. [Traduit par la Rédaction]
Mots-clés : mur en maçonnerie, mince, hors plan, essais en grandeur réelle, rigidité de la base.
0
1. Introduction of the masonry assemblage (fm ), a key component in many design
provision equations, remains ambiguous. Although the testing
Current North American standards for masonry design (CSA of masonry prisms is universally accepted as a viable means to
S304-14 (CSA 2019); TMS 402-16 (MSJC 2016)) define a slender ma- determine the compressive strength of the masonry assemblage,
sonry wall as having an effective height-to-thickness (kh/t) ratio designers often use prescriptive values of fm 0
available in design
(also referred to as the slenderness ratio) greater than 30. The provisions (Table 4 in CSA S304 and Table 2 in TMS 402), as prism
greater the slenderness ratio of a wall, the more susceptible it is testing can be rather difficult and time-consuming. Some studies
to slenderness effects. Slenderness effects are critical, as they have focused on calibrating such tables, and other studies have
may lead to premature material failure (crushing) or instability been conducted to determine empirical relationships between
(buckling). Experimental studies of slenderness effects in full- the component compressive strength and the assemblage com-
scale specimens have proven challenging because of the inherent pressive strength through experimental programs (Khalaf et al.
difficulties in testing walls with high slenderness ratios while 1994; Sarangapani et al. 2005; Ross 2013; Fortes et al. 2015, 2018).
using realistic block sizes. As a result, conventional code provi- Research on slenderness effects experienced by load-bearing
sions for slender masonry walls are based on a small set of testing masonry walls began in the 1970s when Yokel et al. (1970, 1971)
data (Yokel et al. 1970, 1971; Hatzinikolas et al. 1978a, 1978b; ACI-SEASC developed an experimental program to test reinforced masonry
Task Committee on Slender Walls 1982), and the resulting design walls with various slenderness ratios (20, 32, and 40) in a single
guidelines for slender masonry walls tend to be conservative. curvature. All wall specimens were tested under eccentric axial
The study of masonry walls has proven challenging owing to load and fixed-pinned boundary conditions. Comparing the test
the unique material nonlinearity of the assemblage components results to design provisions at the time (National Concrete Masonry
(block, mortar, and grout) as well as their interactions with one Association 1968), it was found that the flexural capacity was under-
another. Therefore, the estimation of the compressive strength estimated by up to a factor of 6 for walls with kh/t ratios of 20 and
Can. J. Civ. Eng. 49: 95–108 (2022) dx.doi.org/10.1139/cjce-2020-0297 Published at www.cdnsciencepub.com/cjce on 3 February 2021.
96 Can. J. Civ. Eng. Vol. 49, 2022
32 and overestimated the flexural capacity for walls with kh/t of Although rotational stiffness at the base of the wall would naturally
40. This overestimation was attributed to stability issues, as large be provided by the foundation on which the wall is constructed,
displacements were observed under small increases in load. It neglecting this rotational stiffness has been traditionally consid-
For personal use only.
was also found that even though the boundary condition at the ered realistic, because it is assumed that under repeated cyclic
base of the walls was designed to restrict rotation, rotation of the loading, a wall will experience sufficient degradation at the bot-
base was still observed. From this, it was concluded that any base tom courses, effectively becoming a hinge and releasing the base
stiffness was quickly degraded under relatively small amounts of moment. However, as experimental programs focusing on the
rotation. behavior of CMU masonry walls have focused on pinned-pinned
Walls in double curvature were later tested by Hatzinikolas boundary conditions (Liu and Dawe 2001; Liu and Hu 2007; Bean
et al. (1978a, 1978b). The program tested 68 masonry walls under Popehn et al. 2009), the validity of this assumption has not been
pinned-pinned boundary conditions. The slenderness ratios for verified.
the test specimens ranged from 12 to 22. Concentrated moments Recent studies (Isfeld et al. 2019; Pettit 2020) suggest that the
were applied at each end through an eccentric axial load at the stiffness at the base can significantly reduce the out-of-plane
top of the wall and an adjustable pin at the base of the wall to cre- deflections, as the moment profile along the height of the wall is
ate a bottom eccentricity. Walls in double curvature exhibited a redistributed by the reactive moment at the base. Figure 1 sche-
much higher load-carrying capacity (up to a factor of 4) than the matically shows the effect of moment redistribution on a wall
walls in a single curvature. However, the failure mode of the subjected to an eccentric axial load at the top of the wall and
walls in double curvature was found to be much more explosive lateral load located at the midspan, which is a typical loading
than that in a single curvature. This was due to snap-through scenario of walls. Accounting for the base rotational stiffness
buckling, as walls in double curvature would buckle in the first creates an inflection point, reducing the effective wall length,
mode at failure (i.e., shifting from double to single curvature). To which increases the critical buckling load of the wall. The total
prevent this sudden failure, design provisions began to limit the moment at the midspan of the wall at a given load also decreased.
axial stress that could be applied to a wall. In the study by Pettit (2020), four wall specimens (slenderness ra-
As demands for taller walls grew, concerns regarding the stabil- tio of 12) had varying degrees of base stiffness (0, 2300, 5000, and
ity and failure modes of slender masonry walls have increased. 9500 kNm/rad). The walls were tested under a non-reversed cyclic
To address these concerns, the American Concrete Institute (ACI) lateral load and a gravity load. It was observed that walls with a
and the Structural Engineers Association of Southern California base stiffness experienced up to an 84% increase in load-carrying
(SEASC) jointly tested 30 full-scale reinforced concrete and ma- capacity, 89% reduction in base rotation, and a 40% reduction in
sonry wall panels under combined axial and lateral loading in midspan deflection compared to the pinned-pinned control wall.
1982. Of the 30 panels tested, nine were constructed using con- Diminishing returns were also observed as the relative increase
crete masonry units (CMU) with slenderness ratios of 29, 36, and in lateral load capacity compared to the control wall was 67%,
48. All the walls were tested using pinned-pinned boundary con- 84%, and 83% for the 2300 , 5000 , and 9500 kNm/rad specimens,
ditions. The findings from this study allowed for the determina- respectively. Upon monitoring the base of the wall for possible
tion of allowable deflection limits, confirming that a ductile degradation under cyclic loading, no significant damage was
response was achieved with under-reinforced sections, and estab- observed, indicating that the pinned base assumption may be
lished limitations on the axial loads to be applied to the walls. conservative. Although promising, neither study tested any speci-
The findings of this study formed the basis of modern North mens with a slenderness ratio greater than 30.
American slender wall design provisions. This study aims to provide experimental evidence on the influ-
Both CSA S304 and TMS 402 assume a pinned base for very slen- ence of rotational base stiffness on the key structural parameters
der walls in CSA S304 and TMS 402. This provision was a result of of slender masonry walls. To accomplish this task, the rotational
the SEASC testing, which included only pin-ended specimens. stiffness provided by a foundation at the base of the wall was
evaluated as a function of the foundation, geometry, and soil Fig. 2. Wall specimen cross-section.
properties. From this, critical values of rotational stiffness were
selected and incorporated with specialized fixtures into eight
slender masonry walls. The walls were divided into two groups of
four. One group had a slenderness ratio of 28.6, and the other had
a slenderness ratio of 33.9. All walls were tested under an eccen-
tric axial load. No lateral load was applied to the walls. The
results were analyzed to study the effect of rotational stiffness on
the load-bearing capacity, deflections, moment redistribution,
and effective flexural rigidity.
All walls had Gr. 400 15M (nominal diameter and area of 15 mm 2.3.2. Vertical load application
and 200 mm2, respectively) steel rebar from the same batch. An MTS hydraulic ram with a capacity of 6000 kN was used to
Standard tensile tests following ASTM E8 (ASTM 2016) were con- apply an eccentric axial load to the top of the wall specimens
ducted to determine the material properties of the rebar and through a pin-type arrangement. The axial load was transferred
weldable rebar (dowels). The results indicated an average yield from the machine to the wall using a C-shaped steel cradle.
strength (fy) of 423 MPa and 403 MPa and an ultimate strength of
568 MPa and 567 MPa for the rebar and weldable rebar, respec- 2.3.3. Foundation stiffness
tively. The coefficient of variation in the yield strength of the The rotational support stiffness R at the base of the wall speci-
rebar was 1.4%. The measured values of the modulus of elasticity mens was simulated by attaching an HSS steel beam to the bot-
for the rebar and weldable rebar were 215000 MPa and 201000 MPa, tom assembly through a fully fixed connection, while the other
respectively. end of the beam was simply supported on the strong floor (Fig. 6).
The moment of inertia and length of the steel member were cho-
2.3. Test setup sen to provide rotational support stiffness levels typical of strip
The test setup used in this experimental program consisted of
foundations, using the analysis model described next.
a system for the application of an eccentric vertical load at the A finite-element (FE) analysis model developed using OpenSEES
top of the wall, a lateral bracing system, and a bottom fixture to (McKenna and Fenves 2001) was used to determine the stiffness of
simulate the base rotational stiffness (Figs. 4 and 5). the foundation, including the presence of soil. Concrete strip foot-
ings were selected in this study as the prototype foundation, as they
2.3.1. Lateral bracing system are commonly used for masonry walls in soils with moderate to
Lateral support was provided at the top loading point to con- high load-bearing capacity (Fig. 7). Pile foundations, found in areas
trol the lateral movement of the wall specimen through a steel with softer soils, are not part of the scope of this study, and their
column fixed to a strong floor. Two lateral braces were installed rotational mechanisms are likely to be significantly different.
between the stiff column and the top pin assembly using an The magnitude of the rotational restraint that strip founda-
eyebolt. tions provide at the base of the wall is a function of the soil prop-
The bottom support of the wall specimen was provided through erties and foundation geometry. Two wall types were considered
an assembly that consisted of a 150 mm diameter steel shaft sup- in the calculation of the rotation stiffness: slender, single-story
ported by a spherical bearing at each end. The spherical bearings masonry walls, and multi-story, non-slender masonry walls. Slen-
were pressed into large steel housings, which had provisions to der, single-story masonry walls are typically subjected to smaller
bolt into T-slots at the base of the testing machine. The assembly gravity loads than their multi-storey counterparts; however, they
was fitted with fixtures that allowed it to freely rotate to simulate are more susceptible to bending owing to their longer unsup-
a pinned connection and to simulate a reactive stiffness, as dis- ported spans. Therefore, the design of tall walls is often governed
cussed in Section 3.3.3. by the flexural moment, with relatively small shear forces at the
Fig. 6. Simulation of base rotational stiffness. [Colour online.] Fig. 7. Shallow strip footing foundations.
Fig. 8. (a) Finite element foundation model and (b) moment–rotation relationship. [Colour online.]
Can. J. Civ. Eng. Downloaded from cdnsciencepub.com by UNIV CALGARY on 03/15/22
stiffness of each spring was calculated to correspond to the modu- Table 2. Numerically determined support stiffness.
lus of the subgrade reaction of the soil multiplied by the tributary
Base Strip footing Modulus of
area of the footing corresponding to each spring.
stiffness dimensions subgrade
To determine the rotational stiffness of the foundation, a moment
(kNm/rad) (mm) Soil type reaction (kN/m3)
was specified at the top node of the foundation (Fig. 8a), and the
rotation at the top of the foundation was monitored. With the 1000 Clayey medium 56 000
soil properties and the concrete in the foundation being assumed dense sand
elastic while all nonlinearity occurs in the wall itself, the rela-
tionship between the applied moment and foundation rotation
is linear (Fig. 8b). The slope of the applied moment – foundation
For personal use only.
Fig. 9. Finite element model of HSS sections. [Colour online.] Table 4. Summary of test results.
Support Base Maximum Midspan
stiffness Peak load moment moment deflection
Specimen h/t (kNm/rad) (kN) (kNm) (kNm) (mm)
W1 28.6 0 527 – 43.0 27.3
W2 28.6 1000 758 –24.1 60.8 28.8
W3* 28.6 5000 968 –40.5 62.6 22.8
W4* 28.6 10 000 818 –45.6 53.2 14.9
W5 33.9 0 481 – 35.9 27.0
W6 33.9 1,000 607 –21.6 50.1 32.2
W7 33.9 5,000 746 –48.5 51.3 28.2
Table 3. Simulated base rotational stiffness. W8 33.9 10 000 804 –61.4 52.1 26.5
Rotational *Walls failed with crushing of the top course.
stiffness HSS HSS section
Can. J. Civ. Eng. Downloaded from cdnsciencepub.com by UNIV CALGARY on 03/15/22
Fig. 10. Load – midspan deflection response: (a) Group I and (b) Group II.
Can. J. Civ. Eng. Downloaded from cdnsciencepub.com by UNIV CALGARY on 03/15/22
Fig. 11. Total moment profile at 250 kN of axial load: (a) Group I and (b) Group II.
For personal use only.
to 5000 kNm/rad (W3 and W7) was 27.7% and 22.9% for Groups I x
ð1Þ Mfo ¼ PA e
and II respectively. When increasing the level of rotational base h
stiffness from 5000 kNm/rad (W3 and W7) to 10 000 kNm/rad (W4
and W8), the relative increase in load-carrying capacity was
ð2Þ Mso ¼ PAþSW D
–15.5% and 7.8% for Groups I and II respectively. The rationale for
specimen W3 having a higher load-carrying capacity than speci-
men W4 is that specimen W4 failed locally at the top of the wall x x
prior to peak load due to inadequate anchorage between the wall ð3Þ Mbs ¼ Ru 1 ¼ P0 l 1
h h
and the fixture.
It should be noted that this study did not include the lateral where PA is the applied axial load, e is the eccentricity of the axial
loads. This is significant because lateral loads often govern slen- load, x is the distance from the base of the wall to the point of in-
der wall design, as their resulting moments are a function of terest, h is the height of the wall, PA+SW is the combination of the
the wall height. Lateral loads would result in additional deflec- applied axial load and the self-weight of the masonry at the point
tions and second-order effects, resulting in a lower load-bearing of interest, D is the deflection at the point of interest, R is the
capacity. rotational base stiffness, u is the base rotation, P 0 is the load cell
reaction of the moment arm, and l is the length of the moment
3.3. Moment distribution arm.
The first-order moment was determined using statics (eq. 1), Figure 11 shows that the presence of a rotational base stiffness
while the second-order moment was determined by multiplying redistributes the moment along the wall height compared to a
the midspan deflection by the axial load (eq. 2). The reactive pinned-pinned case, with negative moments near the base. The
moment at the base of the wall was determined by multiplying level of moment redistribution increases with the level of rota-
the measured rotation by the rotational base stiffness (eq. 3). The tional base stiffness. Specimens with high base stiffness (W4
total moment profiles (the superposition of the primary, second- and W8) experience much larger negative moments at the base
ary, and base stiffness profiles) were plotted at an identical axial compared to specimens with smaller levels of base stiffness (W2
load of 250 kN (Fig. 11) for all the walls. This load was chosen as it and W6).
represents approximately 10% of the compressive capacity of the When a base stiffness is present, an inflection point is created
wall specimen cross-section, which is the maximum allowable in the moment diagram. The position of the inflection point
axial load for walls with h/t > 30 in CSA S304-14 (CSA 2019). moves higher along the wall with increased rotational base
Fig. 12. Deflection profile at 250 kN of axial load: (a) Group I and (b) Group II.
Can. J. Civ. Eng. Downloaded from cdnsciencepub.com by UNIV CALGARY on 03/15/22
stiffness. For example, specimen W2 (rotational base stiffness of because of the presence of a base stiffness, moment redistribu-
1000 kNm/rad) has an inflection point located approximately tion occurs, and the total moment at the midspan of the wall is
350 mm from the base of the wall, whereas specimen W4 (rota- equal to PAe/2 + PA+SWDms – Ru /2. Thus, the moment at the mid-
tional base stiffness of 10 000 kNm/rad) had an inflection point span was reduced by an amount of Ru /2. Table 5 shows the
located approximately 2100 mm from the base of the wall. moment reduction at the midspan for the walls determined
experimentally.
3.4. Deflection profiles Table 5 also shows that as the rotational base stiffness of the
Figures 12a and 12b show the deflection profiles of the Group I wall increases, the midspan moment reduction also increases.
and II walls, respectively, at an axial load of 250 kN. In general, This is expected because higher levels of rotational base stiffness
both Group I and II walls experience smaller deflections as the result in higher levels of moment redistribution. Comparing
base rotational stiffness is increased, with the exception of speci- specimens from Groups I and II, specimens from Group II exhib-
men W4. This was attributed to the experimental observation ited a larger second-order moment compared to Group I with the
that specimen W4 experienced a loss of contact between the top same level of rotational base stiffness. This is reasonable, as the
of the wall and the loading beam on the tension side owing to higher slenderness ratio of Group II resulted in larger deflections
inadequate anchorage. This resulted in increased eccentricity at and, in turn, larger second-order moments.
the top, which caused additional moments and deflections. The
3.6. Effective flexural rigidity
decrease in deflection with increased base stiffness is due to the
fact that the rotational stiffness at the base forces the wall into 3.6.1. Design provisions
double curvature, stiffening the walls. Specimens without rota- Effective flexural rigidity is a key parameter in determining
tional base stiffness experienced up to 4-times more deflection the second-order effects of slender masonry walls. Although simi-
than those with high levels of base stiffness. lar in that both design provisions utilize the elastic modulus of
Comparing the deflection profiles of the wall specimens of the masonry (Em) and both the gross and cracked moment of iner-
Groups I and II, it is seen that Group II experiences higher deflec- tia (Io/In and Icr, respectively) in the calculation of the effective
tions than Group I. This is expected because of the higher slen- flexural rigidity (EIeff), several distinctions between the two provi-
derness of the walls in Group II. sions can be found. Table 6 lists the equations used within each
code provision to determine the effective flexural rigidity.
3.5. Midspan moment In the formulas listed above, e is the virtual eccentricity, ek is
Table 5 compares the midspan moments in the walls at an axial 0
the kern eccentricity, fm is the compressive strength of the ma-
load of 250 kN. The total moment at the midspan for pinned- sonry assemblage, b is the total width of the cross-section, bg and
pinned with eccentric load at the top wall can be calculated as bug are the grouted and ungrouted widths of the cross-section
the sum of the first- and second-order moments, as PAe/2 + respectively, t is the thickness of the unit, tf is the thickness of
PA+SWDms. Without a base stiffness, there is no moment redistrib- the flange, taken as 36.2 mm as recommended by Drysdale and
ution, and this is the final design moment for the wall. However, Hamid (2005), n is the modular ratio between the steel and
12 12 12 12 12 12
Cracked moment of inertia (Icr)
bc3 2 bc3 Pf
(8a) Icr ¼ þ nAs ðd cÞ (8b) Icr ¼ þ n As þ ðd cÞ2
3 3 fy
Cracked neutral axis (c)
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
As fy þ Pf
ðnAs Þ2 þ 2nbAs d nAs (9b) c¼ 0 b
(9a) c¼ 0:64fm
b
Table 7. Moment of inertia comparison. TMS 402 calculation. The measured peak loads exceeded those
corresponding to the maximum allowable axial loads in both
Ig/In Icr
provisions.
( 107 mm4) ( 107 mm4)
The effective flexural rigidity at the midspan of the specimens
For personal use only.
Support Peak by each standard was determined using eq. 4a in Table 8, and
stiffness load eq. 4b in Table 6. The moments of inertia are listed in Table 7, and
Specimen (kNm/rad) h/t (kN) CSA TMS CSA TMS the kern eccentricity (ek) was determined by dividing the section
W1 0 28.6 527 57.2 57.2 3.7 18.0 modulus (Se) by the effective area (Ae) of the cross-section, while
W2 1000 28.6 758 57.2 57.2 3.7 39.7 the virtual eccentricity formulation (ev = Mfp/Pf) simplifies such
W3 5000 28.6 968 57.2 57.2 3.7 76.8 that the virtual eccentricity is equal to the load eccentricity (ev =
W4 10 000 28.6 818 57.2 57.2 3.7 48.4 e) because of the loading condition of the wall specimens (Mfp =
W5 0 33.9 481 57.2 57.2 3.7 15.4 Pfe). As per CSA S304 requirements, a stiffness limit state factor
W6 1000 33.9 607 57.2 57.2 3.7 23.7 ( f er ) of 0.75 was applied to each value of effective flexural rigidity
W7 5000 33.9 746 57.2 57.2 3.7 38.2 to account for the variation in stiffness along the height of the
W8 10 000 33.9 804 57.2 57.2 3.7 46.3 wall. In the TMS 402 provisions, additional parameters needed
within the effective flexural rigidity calculation include the
cracking moment Mcr (eq. 10) and the maximum first-order
moment Mu, which for this loading scenario can be taken as the
masonry materials (n = Es/Em), As is the area of steel reinforcement, applied axial load multiplied by the eccentricity of the load. It
and d is the distance from the extreme compressive fibre of the was found that in all instances, the applied moment exceeded
cross-section to the steel reinforcement. It is noted that in both that of the cracking moment, resulting in the TMS prediction of
design standards, the equations for the moment of inertia neglect EIeff to be limited to EmIcr.
the presence of webs. This is a conservative simplification, as the loss
in the moment of inertia (3.7% for the specimens tested) is negligible. Pf
ð10Þ Mcr ¼ þ fr Sn
Table 7 compares the values of both gross and cracked moments An
of inertia as calculated by each standard using the equations
defined in Table 6. For the gross moment of inertia, defined as Ig where Pf is the factored axial load, An is the net area of the cross-
and In in CSA S304 and TMS 402, respectively, the standards utilize section, fr is the modulus of rupture, and Sn is the section modulus
the same formula. As the section is uncracked at this instance, the of the cross-section. Table 8 summarizes the calculated effective
centroid of the cross-section coincides with the location of the rein- flexural rigidities for both the standards. TMS 402 consistently pre-
forcing steel, and thus the influence of the steel and axial load dicted larger values of EIeff when compared to those calculated
within the gross moment of inertia calculation is ignored. The two using the CSA S304 standard. This is a result of the inclusion of the
provisions diverge when calculating the cracked moment of iner- axial load within the cracked moment of the inertia term in TMS
tia. In the CSA S304 provisions, the cracked neutral axis (c) is calcu- 402. Thus, the increase in loadbearing capacity is directly included
lated based on a linear stress distribution (assuming service in the TMS calculations, resulting in an increased effective flexural
conditions), while the cracked neutral axis in the TMS 402 provi- rigidity with increased load-bearing capacity, unlike the CSA S304
sions is calculated based on a nonlinear stress distribution (assum- standard, in which the effective flexural rigidity remains independ-
ing ultimate limit state conditions). Another difference is the ent of the load-bearing capacity.
inclusion of an axial load term in the cracked moment of inertia
equation in the TMS 402 provisions, whereas this is not done 3.6.2. Analytical prediction of flexural rigidity
in S304. As this study focused on the effective flexural rigidity An estimate of the flexural rigidity of the specimens at peak
(flexural rigidity at peak load), the peak axial load was used in the load, or effective flexural rigidity, was calculated by solving
the elastic differential equation governing the beam-column 1 sinðkxÞ x
ð13Þ uðxÞ ¼ R u þ þ ðPAþSW Þe
behavior: EIk2 sinðkhÞ h
sinðkxÞcosðkhÞ x
@2 @ 2 uðxÞ @ 2 uðxÞ 1þ cosðkhÞ
ð11Þ EI þ PAþSW ¼ qðxÞ sinðkhÞ h
@h2 @h2 @h2
where h is the height of the wall; EI is the flexural rigidity; PA+SW where
is the compressive axial load due to a combination of the applied rffiffiffiffiffiffiffiffiffiffiffiffiffi
axial load and the self-weight of the masonry; u(x) is the deflec- PAþSW
ð14Þ k¼
tion at distance x from the base of the wall; and q(x) is the later- EI
ally distributed load function acting on the wall.
With the absence of a lateral load in this study, the q(x) term in Referring to eq. 13, it is seen that if the deflection and axial load
eq. 11 was taken as zero. The effects of the axial load eccentricity at a specific point, along with the wall height, base rotational
and rotational base stiffness were accounted for through the stiffness, and base rotation are known, the flexural rigidity (EI)
boundary conditions required to solve the differential equation. can be determined. As the flexural rigidity varies with the height
For a simply supported masonry wall subjected to an eccentric of the wall, the value of EI obtained from eq. 13 corresponds to
axial load at the top and a reactive boundary condition at the bot- the value of EI at a specific location in which the deflection and
tom, the following boundary conditions were defined: axial load were measured, allowing the capture of phenomena
such as tensile cracking to be effectively accounted for in the
ð12aÞ uð0Þ ¼ 0 evaluation of EI as the input load and deflection of the point are
dependent on such phenomena.
Using experimental values of midspan deflection, axial load,
ð12bÞ uðhÞ ¼ 0 wall height, base stiffness, and base rotation at peak load, the
effective flexural rigidity of the test specimens was estimated by
iterating the values of EI until both sides of eq. 13 were satisfied.
Ru
ð12cÞ u00 ð0Þ ¼ Such value of EI is termed “experimental.” Table 9 shows the ex-
EI perimental effective flexural rigidity at the midspan of the wall
obtained through the solution of the differential equation using
PA e test data and a comparison with the relevant code provisions.
ð12dÞ u00 ðhÞ ¼ It can be seen that the experimental EI increases with an increase
EI
in the rotational base stiffness. The presence of the base rotational
where R is the value of the rotational stiffness at the base of the stiffness decreases the moment at the midspan of the wall, allowing
wall; u is the slope at the base of the wall; and e is the eccentricity the wall to carry an additional axial load before failure occurs. This
of the compressive axial load. Solving eq. 11 results in the follow- additional axial load stiffens the wall at the critical section, thereby
ing expression: increasing the moment of inertia as cracking is delayed.
Table 10. CSA S304 design moment at the midspan of the wall (P = 250 kN).
Support stiffness MT,CSA MT,exp
Specimen h/t (kNm/rad) k Pcr (kN) Cm (kNm) (kNm) MT,CSA/MT,exp
W1 28.6 0 1.00 320 0.60 21.7 9.3 2.33
W2 28.6 1000 0.90 395 0.55 11.9 8.1 1.47
W3 28.6 5000 0.72 617 0.43 5.7 5.5 1.04
W4 28.6 10 000 0.61 860 0.40 4.5 3.1 1.45
W5 33.9 0 1.00 228 0.60 Instability 9.7 –
W6 33.9 1000 0.90 281 0.55 39.3 8.5 4.62
W7 33.9 5000 0.72 439 0.42 7.7 5.7 1.35
W8 33.9 10 000 0.56 726 0.40 4.8 1.9 2.53
Comparing the experimental results to the design provisions, the determination of the total moment of the wall (MT,CSA, eq. 15).
Can. J. Civ. Eng. Downloaded from cdnsciencepub.com by UNIV CALGARY on 03/15/22
it is seen that both codes underestimate the experimental effec- The effective length factor used in the determination of the criti-
tive flexural rigidity of the wall with CSA S304 being the more cal buckling load (Pcr, eq. 16) reduces the wall height based on the
conservative of the two provisions. This is a result of TMS 402, magnitude of rotational stiffness found at each end of the wall,
including the effects of the axial load in the calculation of the resulting in a larger buckling load and, in turn, a smaller total
cracked moment of inertia. CSA S304 neglects the effects of the design moment. Based on the ratio of moments at each end of
axial load, and the calculation of the cracked moment of inertia the wall, the Cm factor accounts for alternate loading conditions
is only a function of the cross-section properties. This resulted in such as an eccentric axial load, as was the case in this study. To
a constant effective flexural rigidity for all the eight wall speci- determine the effectiveness of these parameters in predicting
mens. Another key difference is the reduction factor ( f er ) of 0.75 the wall response, the total moment at the midspan of the
used in CSA S304 for effective flexural rigidity. This further wall calculated using CSA S304 (MT,CSA) was compared to that
decreases the already lower effective flexural rigidity predicted obtained during the experiment (MT,exp). It should be noted that
by CSA S304 compared to TMS 402. TMS 402 does not include these parameters as it is explicitly
Although TMS 402 accounts for the effects of axial load, the stated that the provided expressions in the provision are derived
standard is limited in that it does not account for the gradual deg- for pinned walls under a uniform distributed load; thus, total
For personal use only.
radation of EIeff from 0.75EmIn to EmIcr (eq. 4b) based on wall load- moment comparisons are made using CSA S304 only. Table 10
ing. Instead, designers are led to use EmIcr for all situations where summarizes the key parameters used to determine the total
the applied moment exceeds the cracking moment. This often moments at an axial load (P) of 250 kN. The k and Cm factors were
results in the cracked moment of inertia being used in situations estimated using the moment profile of the wall (Fig. 11). The ex-
where the actual moment of inertia is significantly larger. CSA perimental total moment (MT,exp) was determined by summing
S304 accounts for this transition through the virtual eccentricity the first, second, and base moments in Table 5, while the Mfo and
(ev) parameter in eq. 4a. EIeff parameters are listed in Tables 5 and 8, respectively. It should
Based on the observations above, the following recommenda- be noted that if the total moment at peak load was to be deter-
tions were made to improve the accuracy of the predicted effec- mined, CSA S304 would predict instability for all eight wall speci-
tive flexural rigidity. In the CSA standard, the determination of mens because the influence of the axial load on the effective
the cracked moment of inertia should be modified to reflect a flexural rigidity is not considered.
nonlinear stress distribution in the masonry at the ultimate load, 2 3
Cm
as opposed to the currently assumed serviceability stress distri- ð15Þ MT;CSA ¼ Mfo 6 7
bution. This is due to the definition of the effective flexural rigid- 4 P 5
1
ity being determined at the peak load, a point at which the stress Pcr
in the masonry is highly nonlinear. The cracked moment of iner-
tia expression should also be modified to include the effects
of axial loads, because the current equation for the cracked p 2 EIeff
ð16Þ Pcr ¼
moment of inertia is derived from masonry beams where axial ðkhÞ2
loads are rare. This inclusion can be done in a similar manner to
TMS, in which the effects of axial load are included by artificially
converting the axial load into an equivalent amount of steel rein- M1
ð17Þ Cm ¼ 0:6 þ 0:4 0:4
forcement. The limitations of the stiffness factor ( f er ) should M2
also be considered. This could be accomplished by neglecting the
use of this parameter if the effective flexural rigidity is taken as Several observations are presented in Table 10. Comparing
EmIcr, as this is the minimum value of stiffness a wall may exhibit; MT,CSA/MT,exp, it is seen that walls without a base stiffness (k = 1.0)
thus, a further reduction is unnecessary. In the TMS 402 provi- present the lowest correlation with the experimental results.
sions, an equation similar to eq. 4a should be implemented to This is interesting because the total moment of the wall is
account for the gradual decrease in flexural stiffness from Mu to directly proportional to the effective flexural rigidity of the wall,
Mcr. As seen in the CSA S304 section of Table 8, the effective flex- indicating that the effective flexural rigidity used was overly con-
ural rigidity (EIeff) of the midspan at peak load was significantly servative. This is reinforced by CSA S304, which predicts a critical
larger than the limit of EmIcr, concluding that the TMS 402 use of buckling load of less than 250 kN in specimen W5, indicating
EmIcr for all cases where Mu > Mcr may be overly conservative. that the wall experiences instability, which is inconsistent
with the experimental results. Comparing walls with base stiff-
3.7. Design moment ness, the correlation of MT,CSA, and MT,exp appears to improve.
To account for base stiffness and loading conditions other This is largely due to the significant increase in the critical buck-
than a uniformly distributed load, CSA S304 includes both an ling load due to the inclusion of the k factor. It is also seen that
effective length factor (k) and moment diagram factor (Cm) in the Cm factor improves the accuracy of the results, as the
ASTM. 2016. Standard test methods for tension testing of metallic materials.
4. Conclusion Standard ASTM E8. ASTM (American Society for Testing and Materials)
Based on the experimental testing of slender masonry walls International, West Conshohocken, Pa.
Bean Popehn, J.R., Schultz, A.E., and Tanner, J.E. 2009. Influence of second
with varying levels of base rotational stiffness, the following con-
order effects on slender, unreinforced masonry walls. In Proceedings of
clusions were drawn: the 11th Canadian Masonry Symposium, Toronto, Ont., 31 May–3 June
2009.
CSA. 2019. Design of masonry structures. Standard CSA – S304-14 (R2019).
1. The effect of the base stiffness on the loadbearing masonry Canadian Standards Association (CSA), Rexdale, Ont.
walls is significant. Walls with a base stiffness demonstrate a Drysdale, R.G., and Hamid, A.A. 2005. Masonry structures: behaviour and
higher load-bearing capacity and smaller midspan deflections design. Canadian Masonry Design Centre, Mississauga, Ont.
Fortes, E., Parsekian, G., and Fonseca, F. 2015. Relationship between the
compared to pinned-pinned walls. However, additional test-
compressive strength of concrete masonry and the compressive strength
ing should be performed using cyclic loads to investigate the of concrete masonry units. Journal of Materials in Civil Engineering,
degradation of the base under repeated rotational demands. ASCE, 27(9): 04014238. doi:10.1061/(ASCE)MT.1943-5533.0001204.
In addition, tests should incorporate the effects of lateral Fortes, E., Parsekian, G., Fonseca, F., and Camacho, J. 2018. High-strength
concrete masonry walls under concentric and eccentric loadings. Journal
loads.
For personal use only.
Sarangapani, G., Venkatarama Reddy, B., and Jagadish, K. 2005. Brick-mortar Io moment of inertia about the walls’ gross cross-sectional
bond and masonry compressive strength. Journal of Materials in Civil En- area (CSA S304)
gineering, ASCE, 17(2): 229–237. doi:10.1061/(ASCE)0899-1561(2005)17:2(229).
Yokel, F., Mathey, R., and Dikkers, R. 1970. Compressive strength of slender k effective length coefficient
concrete masonry walls. Building Science Series 33. Building Research Di- l length of the moment arm
vision, National Bureau of Standards, Washington, DC. LL applied lateral load
Yokel, F., Mathey, R., and Dikkers, R. 1971. Strength of masonry walls under Mbs rotational base stiffness moment
compressive and transverse loads. Building Science Series 34. Building
Research Division, National Bureau of Standards, Washington, DC. Mcr cracking moment
Mfo first-order moment
Mfp maximum factored primary moment
List of symbols Mso second-order moment
Mu applied moment
Ae effective net area (CSA S304) MB moment at the bottom of the wall
An effective net area (TMS 402) MT moment at the top of the wall
b total width of the masonry wall MT,CSA total moment as calculated by CSA S304
bg grouted width of the masonry wall MT,exp total moment as determined from the experiment
bug ungrouted width of the masonry wall
Can. J. Civ. Eng. Downloaded from cdnsciencepub.com by UNIV CALGARY on 03/15/22