Académique Documents
Professionnel Documents
Culture Documents
Abstract: Backbone curves are used in seismic design standards as the basis for developing component models for nonlinear static and
Downloaded from ascelibrary.org by Iowa State University on 01/24/19. Copyright ASCE. For personal use only; all rights reserved.
dynamic analysis. Herein, data from 240 tests of rectangular, barbell, and flanged shear-critical walls are assembled and analyzed to develop
an improved cyclic backbone curve for use in nonlinear response-history analysis. Values of drift ratio and normalized lateral force at the
control points on the cyclic backbone curve are identified. The effects of design parameters, including wall aspect ratio, axial load, day-of-test
concrete compressive strength, vertical and horizontal web reinforcement ratio, reinforcement ratio in the boundary elements, and yield
strength of reinforcement, on the coordinates of the control points are addressed. The accuracy of equations for peak shear strength is assessed
using the collected test data. DOI: 10.1061/(ASCE)ST.1943-541X.0002277. © 2019 American Society of Civil Engineers.
Introduction strength of the component, and an abscissa value equal to the de-
formation at which significant strength degradation begins (line
Nonlinear static and nonlinear dynamic analyses are used to assess CD). Beyond point D, the component responds with substantially
the seismic behavior of existing and new buildings and other reduced strength to point E. At deformations greater than point E,
structures. In the United States, the first actionable documents to the component strength is essentially zero : : : ” (FEMA 1997b).
use nonlinear analysis were FEMA 273 (FEMA 1997b) and FEMA In this figure, QCE is mean expected strength. For evaluation of
274 (FEMA 1997a): the first generation of tools for performance- force-controlled components, QCE was replaced by QCL : a mean-
based earthquake engineering. The emphasis in these documents minus-one standard-deviation estimate of strength.
was on nonlinear static (pushover) analysis, which was deemed The coordinates of the control points on the idealized normal-
by its authors as an interim step between (1) elastic analysis of ized force-displacement (deformation) relationship of Fig. 1 were
models of building frames for lateral (seismic) forces calculated defined by analysis of data available through the mid-1990s by
using a response reduction factor and (2) nonlinear dynamic analy- domain experts. Generally, results of cyclic inelastic tests of com-
sis, which was functionally challenging in the mid-1990s due to a ponents were used to generate such relationships, and Fig. 1 was
lack of commercially available software and the computational ex- subsequently described as a cyclic backbone curve. [This curve is
pense. Fig. 1 reproduces Figs. 2–5(a and b) of FEMA 273: an ideal- generally different from one obtained under unidirectional mono-
ized component load versus deformation curve for depicting tonic loading (e.g., Deierlein et al. 2010).] Fig. 2 illustrates the con-
component modeling and acceptability, now widely described struction of a cyclic backbone curve from the reversed cyclic,
as a backbone curve. These figures were used “ : : : throughout inelastic testing of a shear critical, reinforced concrete (RC) wall
the Guidelines to specify acceptance criteria for deformation- (Luna et al. 2015, 2018).
controlled component and element actions : : : Linear response Table 6-18 (modeling parameters and numerical acceptance
is depicted between point A (unloaded component) and an effective criteria for nonlinear procedures-members controlled by shear)
yield point B. The slope from B to C is typically a small percentage of FEMA 273 assigned values to the variables d, e, and c [as iden-
(0–10%) of the elastic slope, and is included to represent phenom- tified in Fig. 1(b)] for shear-critical walls carrying low axial gravity
ena such as strain hardening. C has an ordinate that represents the load: 0.75%, 2.0%, and 0.40%, respectively. The nominal shear
strength of a shear-critical wall was to be calculated using
1
Assistant Professor, Dept. of Civil Engineering, Amirkabir Univ. of Section 21.6 of American Concrete Institute (ACI) 318-95 using
Technology, 15875-4413 Tehran, Iran (corresponding author). Email: specified reinforcement yield strength, and there “ : : : should be
epackachis@aut.ac.ir
2 no difference between the yield and nominal strengths : : : .”
Undergraduate Student, Dept. of Civil Engineering, Indian Institute of
Technology Gandhinagar, Gujarat 382355, India. Email: nikhil.sharma@
FEMA 273 and 274 were updated and merged in 2000 as FEMA
iitgn.ac.in 356 (FEMA 2000) and subsequently reissued as an ASCE standard
3
SUNY Distinguished Professor, Dept. of Civil, Structural and in 2006 (ASCE 2006) and 2013 (ASCE 2013). The control points
Environmental Engineering, Univ. at Buffalo, Buffalo, NY 14260. Email: for the cyclic backbone curve for shear-critical walls in ASCE
awhittak@buffalo.edu 41 were essentially identical to those described in FEMA 273.
4
Senior Principal, Simpson Gumpertz and Heger, San Francisco, ASCE 41 updated the cyclic backbone curve of FEMA 273,
CA 94111. Email: rohamburger@sgh.com FEMA 356, and ASCE 41. A trilinear lateral force–lateral defor-
5
Director of Projects, Applied Technology Council, Redwood City, mation relationship to peak strength (Fig. 3) was adopted for
CA, 94065. Email: ayse@atcouncil.org
Note. This manuscript was submitted on January 17, 2018; approved
shear-critical RC walls based on the work of Wallace (Elwood et al.
on August 31, 2018; published online on January 24, 2019. Discussion 2007). A point on the curve (F in Fig. 3) was associated with the
period open until June 24, 2019; separate discussions must be submitted onset of cracking.
for individual papers. This paper is part of the Journal of Structural In Fig. 3, shear resistance is normalized by peak shear
Engineering, © ASCE, ISSN 0733-9445. strength, and lateral deformation is normalized by story height h.
Database Development
Gulec et al. (2008) and Gulec and Whittaker (2009, 2011) cata-
logued and processed data from tests of 434 low-aspect-ratio
RC walls to propose an empirical equation for peak shear strength.
Luna et al. (2015, 2018) assembled data published between 2010
and 2015. Only 240 of the datasets assembled by Gulec and Luna
included information useful for the evaluation of the coordinates of
Fig. 3. Load-deformation relationship for members controlled by control points on the cyclic backbone curve. Appendix B of NIST
shear. (Reprinted from ASCE 2013, © ASCE.) GCR 17-917-45 identifies the original sources of the data, which
are not repeated here. All of these data were processed.
The influence of wall aspect ratio, vertical web reinforcement
ratio, horizontal web reinforcement ratio, vertical reinforcement ra-
Parameters g, d, and e in Fig. 3 represent yield drift ratio, drift ratio tio in the boundary elements (i.e., portion along wall edge strength-
corresponding to the onset of loss of shear strength, and ultimate ened by longitudinal and transverse reinforcement), normalized
drift ratio, respectively, where drift ratio is defined as a relative axial compressive load, and concrete uniaxial compressive strength
lateral displacement between the top and bottom of a story, divided on the coordinates of the control points were investigated.
by story height. Parameters f and c are associated with the strength A drawing of a typical RC shear wall, including dimensions and
at the onset of cracking and the residual strength, respectively. details of reinforcement, is presented in Fig. 4.
B B Hor. Bars
Ver. Bars hw
(b)
l be
Ver. Bars
tw
Downloaded from ascelibrary.org by Iowa State University on 01/24/19. Copyright ASCE. For personal use only; all rights reserved.
l ba
Ver. Bars
tw t ba tw
(e)
Fig. 4. Schematic drawings of low-aspect-ratio shear walls: (a) elevation and cross-section; (b) section B-B for a rectangular wall without boundary
elements; (c) section B-B for a rectangular wall with boundary elements; (d) section B-B for a barbell wall; and (e) section B-B for a flanged wall.
Fig. 5. Measured and idealized cyclic backbone curves. of the hysteresis loop) of the cyclic loops were collected and
processed.
data analysis. The figures include the test data, a trend line based on ment, ρv f yv , has a significant effect on V A =V C . An increase in
a least-squares fit, and lower and upper bounds for a confidence ρv fyv leads to a decrease in V A =V C in both rectangular and flanged
level of 95%. walls because (1) the lateral strength of a low-aspect-ratio wall is
dependent on the vertical reinforcement ratio, and (2) the force at
cracking should be independent of ρv f yv . Increases in ρh f yh
Drift at Cracking, δA and ρvb fyvb reduce V A =V C for rectangular walls, with a smaller
The effects of design variables on the drift ratio at cracking are pre- effect observed for flanged walls. Axial compression affects the
sented in Figs. 6 and 7. Of the six design variables considered here, normalized force at cracking: as the axial force increases, V A =V C
f c0 has the greatest effect on rectangular (planar) walls: the greater decreases. A reduction in the normalized force at cracking, from
the value of fc0 , the greater the drift ratio at cracking. The effect f ¼ 0.6 to f ¼ 0.5, is supported by the data.
of f c0 on the response of walls with barbells and flanges is small.
Per the data. The drift ratio at cracking is also affected by aspect
(moment-to-shear) ratio, axial compressive force, and boundary Drift Ratio at Yield, δB
element reinforcement ratio. As expected, the drift ratio at cracking Figs. 10 and 11 present data for the drift ratio at yield δ B . The
increases with aspect ratio. dashed lines in Figs. 10 and 11 are the ASCE 41 value of 0.4%
The trend lines in both Figs. 6 and 7 indicate that an increase in (Table 1), which underestimates most of the recorded data for
axial compressive force leads to a decrease in the drift at the onset rectangular walls. Wall aspect ratio affects the yield drift ratio,
of cracking, which is counterintuitive, calling into question (1) the although the effect on rectangular walls is small. Concrete uniaxial
use of trend lines for widely scattered data, (2) the accuracy of the compressive strength fc0 has a significant effect on the yield drift
reported data, and (3) the association here of cracking with the first ratio for both types of wall.
Drift Ratio at Peak Strength, δC to a drift ratio of 1% (ASCE 41 value if the axial force ratio is
0.05 or smaller). The other corresponds to a drift ratio of 0.75%
Figs. 12 and 13 present data for the drift ratio at peak strength, δ C . (ASCE 41 value if the axial force ratio is greater than 0.05). Fig. 13
Two dashed lines are shown in the panels. One corresponds clearly shows that of the six design variables considered in this
study, the drift ratio at peak strength is most affected by f c0 and Drift Ratio at Zero Residual Strength
M=Vlw : the greatest slopes of the trend lines in the corresponding Table 10-20 of ASCE 41 (Table 1) sets the residual strength ratio
panels. The drift ratio at peak strength decreases with increas- and ultimate drift ratio (i.e., point E on the backbone curve of
ing axial force ratio: the values proposed in ASCE 41 are not Fig. 1) equal to 0 and 1.0%, respectively, if the axial force ratio
unreasonable. ½ðAs − As0 Þf y þ P=ðtw lw fc0 Þ is greater than 0.05, and 0.2 and 2.0%
Fig. 11. Effect of design variables on δB for barbell and flanged walls.
otherwise. The effects of design variables on the drift ratio at strength. The projected drift ratio at zero residual strength was
zero residual strength δ 0D are presented in Figs. 14 and 15, respec- calculated by extending the slope of the postpeak branch to zero
tively. The panels of Figs. 14 and 15 present the experimentally force. The dashed line corresponds to the ASCE 41 ultimate drift
measured or projected drift ratio at point D for zero residual ratio of 1.0% for values of the axial force ratio greater than 0.05,
Fig. 13. Effect of design variables on δ C for barbell and flanged walls.
for which the residual strength ratio is set equal to 0. For values of Normalized Peak Lateral Strength
the axial force ratio in the range from 0.05 to 0.10, 1.0% is a low
estimate of δ 0D . The value of δ 0D decreases as the axial force ratio Figs. 16 and 17 present the effects of design variables on normal-
increases. ized peak lateral strength, V C =lw tw , where the normalized peak
Fig. 15. Effect of design variables on δ 0D for barbell and flanged walls.
strength is calculated as the average of the peak lateral strengths in the flanged walls because only the web area is used to normalize the
the first and third quadrants of loading, divided by the area of the peak lateral strength.) The normalized peak lateral strengths of the
web lw tw. (The flanges contribute to shear resistance, as is clear rectangular and flanged walls vary with f c0 , ρv f yv , and ρh f yh : as
from the y-axis range on the reported data in Figs. 16 and 17. Sub- these values increase, V C =lw tw increases. Aspect ratio also influen-
stantially greater values of normalized peak strength are achieved in ces the normalized peak lateral strength of flanged walls, with an
Fig. 17. Effect of design variables on V C =lw tw for barbell and flanged walls.
increase in aspect ratio leading to a decrease in normalized strength. Normalized Initial Stiffness
Interestingly, this trend is not seen in rectangular walls. Axial com-
pressive force in the range considered here increases normalized Figs. 18 and 19 present the effects of some of the design var-
peak lateral strength, which is an expected result. iables on normalized initial stiffness, K e =K t . Here, the onset of
Fig. 19. Effect of design variables on K e =K t for barbell and flanged walls.
5 5 5
0 0 0
0 5 10 15 0 5 10 15 0 5 10 15
Downloaded from ascelibrary.org by Iowa State University on 01/24/19. Copyright ASCE. For personal use only; all rights reserved.
(a) Vpt /lw tw [MPa] (b) Vpt /lw tw [MPa] (c) Vpt /lw tw [MPa]
15 15
With stress limit Without stress limit Mean=1.05
St. Dev.=0.49
Mean=0.72 Mean=0.92 COV=0.46
St. Dev.=0.56 St. Dev.=0.66
COV=0.77 COV=0.72
Vpp /lw tw [MPa]
5 5
0 0
0 5 10 15 0 5 10 15
(d) Vpt /lw tw [MPa] (e) Vpt /lw tw [MPa]
Fig. 20. Predicted and measured normalized lateral strengths of RC shear walls: (a) ACI 2014 (Chapter 11); rectangular walls; (b) ACI 2014
(Chapter 11); barbell and flanged walls; (c) ACI 2014 (Chapter 18); rectangular walls; (d) ACI 2014 (Chapter 18); barbell and flanged walls;
and (e) Barda et al. (1977); flanged walls.
The ACI equations are widely known and not repeated here. Analysis of the data of Fig. 20(a) indicates the equation proposed
Fig. 20 presents predicted normalized peak strength V pp =lw tw in Chapter 11 of ACI 318, which includes the upper shear stress
versus measured normalized peak strength V pt =lw tw , where V pp limit, reasonably predicts the peak lateral strength of rectangular
and V pt are the predicted and measured peak lateral strengths, re- RC walls and is the more accurate of the two ACI equations. The
spectively. The statistically estimated values of the average, stan- statistics of Fig. 20(c) indicate the best estimate of the peak lateral
dard deviation (SD), and coefficient of variation (COV) for each set strength of flanged walls is provided by the Barda equation,
of empirical predictive equations are presented in the figures. although the scatter in the data is large.
Each of the ACI equations imposes an upper limit on the pre-
pffiffiffiffiffi pffiffiffiffiffi pffiffiffiffiffi
dicted shear strength: 10 fc0 dtw ð¼ 8 fc0 lw tw ¼ 8Aw f c0 Þ for
pffiffiffiffiffi
the Chapter 11 equation, assuming d ¼ 0.8lw and 8Aw f c0 for the Summary and Recommended Changes to the ASCE
Chapter 18 equation, all in US units These limits on shear stresses 41 Cyclic Backbone Curve
are intended to avoid, albeit indirectly, failure in compression of
Backbone curves are used in seismic design standards as the basis
diagonal concrete struts. Two sets of results are presented in
for developing component models for nonlinear static and dynamic
Figs. 20(a and b): measured versus predicted shear strengths with
analysis. Herein, data from 240 tests of rectangular, barbell, and
the limit on shear stress (solid black circle) imposed and removed
flanged shear-critical walls, taken from catalogues assembled by
(open black square). Mean values greater than 1.00 (above the
Gulec and Luna, are assembled and analyzed to develop an im-
dashed line) indicate that the predictive equation for nominal
proved cyclic backbone curve (or idealized force–displacement
strength overestimates the measured strength (average of values
relationship) for use in nonlinear response-history analysis. The
in the first and third quadrant). For the rectangular walls, the pre-
pffiffiffiffiffi idealized force–displacement relationship is a piecewise linear
dicted maximum shear strength is limited by 8Aw fc0 in 67 of 136 curve defined by control points, where the coordinates of force
walls. For the barbell and flanged walls, the predicted maximum
pffiffiffiffiffi are normalized by peak shear strength.
shear strength is limited by 8Aw f c0 in 57 of 104 walls. The variability in the coordinates of the control points is great.
The mean values in the right-hand panels of Figs. 20(a and b) Ideally, the coordinates of the control points should be expressed in
are greater than their counterparts in the left-hand panels because a functional form, which accommodates all design variables that
flanges have been shown to resist significant shear in the plane of impact calculations. However, in keeping with the format of ASCE
the web (e.g., Gulec et al. 2008; Gulec and Whittaker 2009, 2011). 41, for shear walls and other framing systems, the recommended
Barda, F., J. M. Hanson, and W. G. Corley. 1977. “Shear strength of Luna, B. N., J. P. Rivera, S. Epackachi, and A. S. Whittaker. 2018. Seismic
low-rise walls with boundary elements.” ACI Spec. Publ. 53: 149–202. response of low aspect ratio rein forced concrete shear walls for build-
Deierlein, G. G., A. M. Reinhorn, and M. R. Willford. 2010. Nonlinear ings and safety-related nuclear structures. MCEER-18-0002. Buffalo,
structural analysis for seismic design. NEHRP Seismic Design NY: Univ. at Buffalo.
Technical Brief No. 4. Gaithersburg, MD: NIST. Luna, B. N., J. P. Rivera, and A. S. Whittaker. 2015. “Seismic behavior of
Elwood, K. J., A. B. Matamoros, J. W. Wallace, D. E. Lehman, J. A. Heintz, low aspect ratio reinforced concrete shear walls.” ACI Struct. J. 112 (5):
A. D. Mitchell, M. A. Moore, M. T. Valley, L. N. Lowes, and 593–603. https://doi.org/10.14359/51687709.
C. D. Comartin. 2007. “Update to ASCE/SEI 41 concrete provisions.” Mahin, S. A., and V. V. Bertero. 1976. “Problems in establishing and pre-
Earthquake Spectra 23 (3): 493–523. https://doi.org/10.1193/1 dicting ductility in a seismic design.” In Proc., Int. Symp. on Earth-
.2757714. quake Structural Engineering, 613–628. Rolla, MO: Univ. of Missouri.
Epackachi, S., N. Sharma, A. S. Whittaker, and A. Hortacsu. 2018. NIST. 2017. Recommended modeling parameters and acceptance criteria
“A cyclic backbone curve for squat reinforced concrete shear for nonlinear analysis in support of seismic evaluation, retrofit,
walls.” In Proc., 11th US National Conf. on Earthquake Engineering. and design. NIST GCR 17-917-45. Gaithersburg, MD: Applied
Oakland, CA: Earthquake Engineering Research Institute. Technology Council.