Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/269873567
CITATIONS READS
14 117
2 authors, including:
Andrew Whittaker
University at Buffalo, The State University of New York
447 PUBLICATIONS 5,079 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Development of modeling recommendations for squat reinforced concrete walls (ATC-114 /TASK ORDER
38) View project
Seismic isolation of large light water reactors using elastomeric and sliding bearings View project
All content following this page was uploaded by Andrew Whittaker on 30 December 2014.
Low aspect ratio reinforced concrete walls are important structural RESEARCH SIGNIFICANCE
components in many commercial buildings and nearly all safety- Peak shear strength is the key parameter in the design and
related nuclear structures. The accurate prediction of their shear performance assessment of buildings and structures
strength is important for both code-based design and seismic constructed with low aspect ratio reinforced concrete walls.
performance assessment. Existing empirical predictive equations for the nominal
Building codes, manuals of practice, standards and guidelines, shear strength of low aspect ratio walls are biased and cannot
and the literature provide a number of empirical equations for reproduce results from physical tests. A database of results
peak shear strength of reinforced concrete walls. These equations
vary substantially in functional form and do not account for all
from tests of 434 walls with aspect ratios of 2 and less was
variables that affect the response of low aspect ratio walls. There developed to improve predictive shear strength equations,
is substantial scatter in the predicted values of peak shear strength. finite element modeling of walls, and hysteretic models.
Improved empirical equations are developed using data from tests Empirical but unbiased predictive equations for peak shear
of 227 walls with the objective of predicting the peak shear strength are proposed for rectangular, barbell, and flanged
strength of low aspect ratio walls without bias and with a small walls, which are suitable for implementation in codes and
coefficient of variation (COV). standards and for use in the performance assessment of
buildings, safety-related nuclear facilities, and other structures.
Keywords: low aspect ratio; reinforced concrete; shear strength; shear walls.
DATABASE
INTRODUCTION Gulec and Whittaker2 reviewed and catalogued the results
Low aspect ratio (squat, short) reinforced concrete walls of tests of 434 low aspect ratio reinforced concrete walls
are widely used in conventional buildings and safety-related with three different cross sections, namely, rectangular,
nuclear structures. The resistance of such walls to lateral barbell, and flanged. Of the 434 walls, 150 had a rectangular
loadings is generally controlled by shear strength, with failure cross section and 284 had boundary elements in the form of
resulting from diagonal tension, diagonal compression, sliding columns (191) or flanges (93). The test specimens in the
shear, or a combination thereof.1,2 database have: 1) a minimum web thickness of 5 cm (1.97 in.);
Building codes,3 standards,4 and the literature5,6 provide a 2) a symmetric reinforcement layout; 3) no diagonal
number of empirical equations to predict the peak shear reinforcement or additional wall-to-foundation reinforcement
strength (that is, the maximum shearing resistance of a wall to control sliding shear; and 4) aspect ratios (hw/lw) less than or
to lateral force) of reinforced concrete walls. Prior studies6-8 equal to 2.0. Detailed information on each wall and its loading
have indicated that these equations do not provide unbiased protocol can be found in Gulec and Whittaker.2
estimates of peak shear strength and their predictions yield
significant scatter, which is problematic because shear PREDICTIVE EQUATIONS FOR PEAK
strength is the key variable for force-based design and SHEAR STRENGTH
performance assessment of walls with a low aspect ratio Gulec et al.7,8 investigated the performance of five
(ratio of height to length). Alternative empirical equations that procedures that are widely used to predict the peak shear
provide much-improved predictions of peak shear strength are strength of reinforced concrete walls: Chapters 11 and 21 of
developed by statistical analysis of data from tests of 227 ACI 318-08,3 Barda et al.,5 ASCE 43-05,4 and Wood.6 They
walls with an aspect ratio of 2 or less, considering those concluded that: a) the use of the equations significantly
variables that significantly affect the behavior of low aspect varied with respect to wall geometry (rectangular, flanged,
ratio walls. The walls in the database are rectangular, barbell and barbell); and b) the coefficients of variation (COVs)
(rectangular walls framed by boundary columns), or flanged associated with the distributions of the ratio of predicted to
(rectangular walls framed by boundary flanges). The use of experimental peak shear strength were generally large. The
these new equations is compared to those used in design Wood equation6 best predicted the peak shear strength of
practice in the U.S. at this time. rectangular walls and the ASCE 43-05 equation4 performed
best for walls with boundary columns or flanges.
It is important to note that the shear strength of low aspect
None of the five procedures explicitly considered the
ratio walls rapidly degrades with repeated cycling7,8 and that
effect of wall cross section type or boundary element
none of the procedures considered in this paper address this
important issue. Further, the predictive equations proposed
in the following are empirical, and their future use must not ACI Structural Journal, V. 108, No. 1, January-February 2011.
MS No. S-2008-327.R2 received January 13, 2010, and reviewed under Institute
preclude the development of physics-based models capable publication policies. Copyright 2011, American Concrete Institute. All rights reserved,
of fully characterizing the response of low aspect ratio including the making of copies unless permission is obtained from the copyright proprietors.
Pertinent discussion including authors closure, if any, will be published in the November-
reinforced concrete walls under random loading through failure. December 2011 ACI Structural Journal if the discussion is received by July 1, 2011.
Fig. 2Variation of ratio of predicted shear strength to Fig. 4Variation of ratio of predicted shear strength to
experimental peak shear strength with respect to aspect experimental peak shear strength with respect to vfyv for
ratio for Models Vn1 and Vm1a. Models Vn1 and Vm1a. (Note: 1000 psi = 6.89 MPa.)
Fig. 3Variation of ratio of predicted shear strength to Fig. 5Variation of ratio of predicted shear strength to
experimental peak shear strength with respect to hfyh for experimental peak shear strength with respect to befybe for
Vn1 and Vm1a. (Note: 1000 psi = 6.89 MPa.) Vn1 and Vm1a. (Note: 1000 psi = 6.89 MPa.)
reinforcement ratio h fyh, vertical web reinforcement ratio solid lines in each figure are linear fits to Model Vm1a
v fyv, vertical boundary element reinforcement ratio be fybe, predictions. These figures and the lines of best fit show that
normalized axial force P/(Aw fc), and fc. In a well-specified Model Vm1a accurately captures the peak shear strength for
model, the data points in Fig. 2 through 7 should be scattered all six design variables over their ranges. The majority of the
in a narrow band at approximately the value of 1.0 for the ratios for Model Vm1a are between 0.75 and 1.25, whereas
ratio of predicted to experimental peak shear strength. The the ratios for Model Vn1 are widely scattered and range
(barbells) and flanges. The barbell and flanged wall data sets
were merged for analysis because the two cross-section
types sustain a similar sequence of damage when subjected
to incremented cyclic lateral loading, namely: 1) cracking of
concrete due to shearing force; 2) spalling and crushing of
concrete; and 3) failure of diagonal compression struts.
Gulec and Whittaker2 showed by finite element analysis
that although barbells in low aspect ratio walls are fully
effective in resisting shearing force, flanges may not be.
Accordingly, the coefficients of Eq. (3) were computed for
four assumptions of effective flange width, beff = bf (flanges
fully effective), and fractions of the height of the wall (hw,
hw/2, and hw/4). (Note: a similar approach is used in ACI 318-083
to calculate the effective flange widths for flexure and axial
load design. In ACI 318-08,3 unless a more detailed analysis
is performed, the effective flange widths are required to be
extended from the face of the web to the smaller of 1/2 the
distance to an adjacent wall web and 25% of the total wall
Fig. 7Variation of ratio of predicted shear strength to
height.) Similar to rectangular walls, two exponents on the
experimental peak shear strength with respect to fc for
concrete contribution were studied, namely, 0.5 and 1.0. The
Models Vn1 and Vm1a. (Note: 1000 psi = 6.89 MPa.)
eight models are summarized in Table 3. These models are
valid for the range of data from which they were developed,
between 0.36 and 3.52. In Fig. 2 through 7, the data points namely, for an aspect ratio between 0.20 and 1.0; 0 and
corresponding to Model Vn1 indicate signs of model 1650 psi (11.4 MPa) for h fyh; 0 and 1950 psi (13.4 MPa)
underspecification and/or heterogeneous variance. for v fyv; 155 and 1890 psi (1.1 and 13.0 MPa) for be fybe; 0
Figure 8 presents the variation of experimental peak shear and 32% for P/At fc, where At is the total wall area; and 1450
strength and peak shear strength calculated using Model Vm1a and 8460 psi (10.0 and 58.3 MPa) for fc. The values of each
(normalized by total wall area and fc) with respect to aspect variable in this data set also distribute somewhat uniformly
ratio. Figure 8 shows that normalized experimental and across its range (refer to Gulec and Whittaker2 for
predicted peak shear strengths for the walls in the data set are detailed information).
less than the upper limit for shear stress in ACI 318-083 Table 4 presents the statistics for the ratio of predicted to
of 10fc. experimentally measured peak shear strength using the five
The need for an upper shear stress limit is unknown at this widely used predictive equations and the eight models of
time because the procedure does not overpredict the peak Table 3. Table 5 presents similar statistics for the eight
shear strength of walls that developed relatively high models of Table 3 but results are presented separately for
shear stresses. barbell and flanged walls to assess the performance of each
model for each cross section. The data presented in Table 3
Peak shear strength equations for walls with through Table 5 for Models Vm1a through Vm1h were
boundary columns or flanges obtained by setting an upper limit of 0.15 on P/At fc (axial
Eight models (equations) based on Eq. (3) were created force normalized by the product of total wall area and
using a data set involving walls with boundary columns concrete compressive strength) because a preliminary
Table 4Statistics for ratio of predicted to experimental peak shear strength obtained using five widely
used procedures and eight models based on Eq. (3)
Mean Median Standard deviation COV Minimum Maximum Overpredictions, % Error sum of squares
Vn1/Vpeak 0.86 0.82 0.370 0.429 0.182 2.984 28.8 3,971,144
Vn2/Vpeak 0.79 0.76 0.283 0.359 0.207 2.432 21.6 4,953,880
Vn3/Vpeak 0.80 0.79 0.195 0.243 0.404 1.389 13.1 3,980,676
Vn4/Vpeak 0.89 0.86 0.223 0.251 0.467 1.462 30.1 2,736,754
Vn5/Vpeak 0.52 0.48 0.165 0.317 0.269 1.026 0.7 9,638,110
Vm1a/Vpeak 1.00 0.98 0.151 0.151 0.725 1.688 46.4 548,395
Vm1b/Vpeak 1.00 1.00 0.173 0.173 0.702 1.860 48.4 649,504
Vm1c/Vpeak 1.00 0.99 0.130 0.130 0.723 1.445 47.7 482,817
Vm1d/Vpeak 1.00 1.00 0.151 0.151 0.704 1.625 49.0 575,108
Vm1e/Vpeak 1.00 0.99 0.110 0.110 0.705 1.264 44.4 432,125
Vm1f /Vpeak 1.00 0.99 0.134 0.134 0.680 1.383 43.8 545,335
Vm1g/Vpeak 1.00 0.99 0.129 0.129 0.672 1.344 45.8 597,209
Vm1h/Vpeak 1.00 0.98 0.151 0.151 0.649 1.428 43.1 742,765
Table 5Statistics for ratio of predicted to experimental peak shear strength obtained using eight models
based on Eq. (3)
Cross section Mean Median Standard deviation COV Minimum Maximum Overpredictions, %
Barbell 0.95 0.94 0.105 0.111 0.728 1.191 27.8
Vm1a/Vpeak
Flanged 1.05 1.05 0.172 0.163 0.725 1.688 66.2
Barbell 0.96 0.95 0.142 0.148 0.702 1.284 35.4
Vm1b/Vpeak
Flanged 1.04 1.03 0.193 0.185 0.712 1.860 62.2
Barbell 0.95 0.94 0.102 0.106 0.738 1.183 27.8
Vm1c/Vpeak
Flanged 1.05 1.05 0.139 0.132 0.723 1.445 68.9
Barbell 0.97 0.95 0.136 0.141 0.713 1.268 34.2
Vm1d/Vpeak
Flanged 1.04 1.04 0.157 0.152 0.704 1.625 64.9
Barbell 0.99 0.98 0.106 0.106 0.780 1.261 40.5
Vm1e/Vpeak
Flanged 1.01 1.00 0.115 0.114 0.705 1.264 48.6
Barbell 1.00 0.98 0.141 0.141 0.750 1.342 43.0
Vm1f /Vpeak
Flanged 1.00 0.99 0.126 0.126 0.680 1.383 44.6
Barbell 1.05 1.04 0.109 0.104 0.819 1.344 63.3
Vm1g/Vpeak
Flanged 0.95 0.95 0.130 0.137 0.672 1.246 27.0
Barbell 1.05 1.04 0.145 0.138 0.790 1.428 57.0
Vm1h/Vpeak
Flanged 0.95 0.95 0.138 0.146 0.649 1.339 28.4
analysis showed that these models overestimated the peak predicted to experimental peak shear strength of 0.99 and a
shear strength of walls with P/At fc greater than 0.15. COV of 0.11, which was the smallest among the models
As seen in Table 4, Model Vm1e , which uses an assumed investigated. The model also yielded the smallest error (or
effective flange width of hw /2 and 2 = 1.0, provided the best residual) sum of squares statistic of the eight models. The
estimate of peak shear strength with a median ratio of model that used an assumed effective flange width of hw/2
Fig. 13Variation of ratio of predicted shear strength to Fig. 15Variation of ratio of predicted shear strength to
experimental peak shear strength with vfyv for Model Vm1e. experimental peak shear strength with Abe/At for Model Vm1e.
(Note: 1000 psi = 6.89 MPa.)
Lines of best fit are shown in each figure. The data points are
generally scattered in a narrow band at approximately 1.0 for
both cross-section types. The majority of the ratios
associated with Model Vm1e are between 0.75 and 1.25. The
observed trends in the figures are generally weak, indicating
that the model can successfully predict the peak shear
strength of low aspect ratio walls with boundary elements
within the ranges of the design variables considered.
Figures 15 and 16 present the variation of the ratio of
predicted to experimental shear strength with Abe/At and
bf /hw, respectively, where Abe/At is the ratio of the total
boundary element area to total wall area and bf /hw is the ratio
of flange width to wall height. The weak trends seen in Fig. 15
indicate that the model can successfully account for the
effect of the area of the barbells or flanges on peak shear
strength. The data of Fig. 16, which is only relevant to low
aspect ratio, flanged walls, indicate that the assumption of an
effective flange width of hw/2 is reasonable. As seen in the
figure, the ratios of predicted to experimental peak shear Fig. 16Variation of ratio of predicted shear strength to
strengths do not exhibit a trend with respect to the ratio of experimental peak shear strength with bf/hw for Model Vm1e
flange width to wall height. and flanged walls.
where Aw (in in.2) is the wall area, fc (in psi) is the compressive
strength of concrete, Fvw (in lb) is the force attributed to
vertical web reinforcement (calculated as the product of area
of vertical web reinforcement and the reinforcement yield
stress), Fvbe (in lb) is the force attributed to boundary
Fig. 17Variation of ratio of predicted Model Vm1e and element reinforcement (calculated as the product of total
experimental shear strength (normalized using total wall area area of vertical boundary element reinforcement at each end
and fc ) with respect to aspect ratio. (Note: 1000 psi = of the wall and the reinforcement yield stress), P (in lb) is the
6.89 MPa.) axial force, hw is the wall height, and lw is the wall length.
(Note: 1 in. = 25.4 mm; 1000 lb = 4.4482 kN; 1000 psi =
6.895 MPa.)
Figure 17 presents the variation of experimentally measured The performance of Models Vm1a and Vrec is presented in
peak shear strength and peak shear strength calculated using Table 6; the COVs for the two models are similar. An upper
Model Vm1e (normalized with total wall area and fc) with limit on shear force of 10 f c A w is imposed in Eq. (4)
respect to aspect ratio. As seen in the figure and similar to because the experimentally measured peak shear strength for
rectangular walls, the need for an upper shear stress limit for rectangular walls was always less than 10 f c A w .
Model Vm1e is unclear at this time because the procedure did not The proposed peak shear strength equation for symmetric
yield unconservative estimations of the peak shear strength for shear-critical walls with boundary elements and an aspect
walls that developed relatively high shear stresses. ratio of 1.0 or less (VBE) is presented in Eq. (5).
Simplification of empirical equations for peak ( 0.04f c )A eff + 0.40F vw + 0.15F vbe + 0.35P
---------------------------------------------------------------------------------------------------------
shear strength V BE = hw lw (5)
The empirical equations presented herein provide
substantially better estimations of the peak shear strength of 15 f c A t
low aspect ratio reinforced concrete walls than current code
procedures,3 standards,4 and the literature.5,6 The optimized where Aeff (in in.2) is the total wall area for barbell walls and
peak shear strength equations can be further simplified with the effective area for flanged walls (equal to the sum of the
a little sacrifice in performance into a form suitable for areas of the web plus the effective flanges area, where the area
application in a design office. Model Vm1a performed best of an effective flange is the product of the flange thickness and
for rectangular walls and Model Vm1e performed best for one-half the wall height minus the web thickness). Equation (5)
walls with barbells or flanges. Accordingly, these models can be applied to all barbell walls and flanged walls for which
were used as the basis for the code-oriented design equations At /Aw 1.25. For the design of flanged walls with 1.0 At/Aw
presented in the following. To simplify these equations, two 1.25, the peak shear strength should be taken as the smaller
additional changes were made to Model Vm1a for rectangular of the values calculated using Eq. (4) and (5). Equation (5) is
walls and Model Vm1e for walls with boundary elements: a simplification of Model Vm1e, but the loss of performance
1. The numerical analyses performed using VecTor210 in is only modest (refer to Table 7).
Gulec and Whittaker2 and the evaluation of available data Figure 17 shows that the experimentally measured peak
show that the effect of the horizontal web reinforcement shear strength of walls with barbells and flanges is generally
ratio on peak shear strength is modest in comparison with the less than 15 f c A t . Accordingly, 15 f c A t is proposed as an
effects of the other variables. Accordingly, a term associated upper limit on peak shear strength per Eq. (5) until additional
with horizontal web reinforcement ratio was not included in data are available to support an alternative limit.
the equations for walls with either rectangular cross sections The experimental data used to develop the peak shear
or with boundary columns or flanges. strength equations for low aspect ratio walls were based on
2. The coefficient 7, which is associated with aspect ratio tests of walls loaded in the plane of the web. For walls with
in Eq. (3) (that is, [hw/lw]7) was 0.58 for Model Vm1a for flanges or orthogonal walls sharing common boundary
rectangular walls and 0.48 for Model Vm1e for walls with elements, where the flanges or boundary elements are
boundary elements. This coefficient was set equal to 0.5. subjected to loading perpendicular to the web, the equations
The proposed peak shear strength equation for shear- must be used with care because orthogonal loading will
critical rectangular walls with an aspect ratio of 1.0 or less generally degrade the integrity of the boundary elements and
(Vrec) is presented in Eq. (4) may effectively reduce the cross section to a rectangular