You are on page 1of 22

Research Paper

Earthquake Spectra
Quantitative assessment of 2021, Vol. 37(2) 1013–1034
Ó The Author(s) 2020
seismic design provisions for Article reuse guidelines:
sagepub.com/journals-permissions

buildings with torsional DOI: 10.1177/8755293020970967


journals.sagepub.com/home/eqs

irregularities

D Jared DeBock, M.EERI1, Michael Valley, M.EERI2,


and Conrad (Sandy) Hohener3

Abstract
ASCE/SEI 7 design requirements for seismic-induced torsion in buildings are evalu-
ated to determine their effectiveness for resisting seismic-induced collapse of tor-
sionally irregular buildings. The ASCE/SEI 7-16 provisions are found to be generally
conservative for most torsionally irregular building configurations—exceptions are
some buildings that rely heavily on lines of lateral resistance orthogonal to the design
earthquake force to resist torsional moments, and also some torsionally flexible
buildings designed using modal response spectrum analysis. Modifications to provide
better consistency in collapse resistance over a large range of building configurations
and degrees of torsional irregularity are recommended. The study also demonstrates
that buildings classified as extremely torsionally irregular may not need to be prohib-
ited from Seismic Design Categories E and F, as long as the lateral system is propor-
tioned properly.

Keywords
Torsion, irregular, ASCE 7, collapse, FEMA P695, FEMA P2012, seismic design,
earthquake
Date received: 27 January 2020; accepted: 2 October 2020

1
Department of Civil Engineering, California State University Chico, Chico, CA, USA
2
Magnusson Klemencic Associates, Seattle, WA, USA
3
Degenkolb Engineers, Los Angeles, CA, USA

Corresponding author:
D Jared DeBock, Department of Civil Engineering, California State University Chico, 400 West First Street, Chico, CA
95929, USA.
Email: ddebock@csuchico.edu
1014 Earthquake Spectra 37(2)

Introduction
ASCE/SEI 7-16 (ASCE, 2016) requires that all buildings with nonflexible diaphragms (in
plane) consider the effects of torsion due to earthquake shaking. Furthermore, buildings
classified as torsionally irregular or extremely torsionally irregular are required to meet
additional, more stringent, strength and deflection criteria to ensure adequate collapse
safety. Additional requirements imposed on buildings with torsional irregularity include
applying an accidental torsional moment (which is also amplified for static analysis), satis-
fying drift and stability requirements at the building’s most critical edge rather than at the
center of mass (CM), and, for extremely irregular structures, increasing the forces 30% by
requiring a redundancy factor r = 1.3. In addition to the added requirements, buildings
classified as extremely torsionally irregular are prohibited in Seismic Design Categories
(SDC) E and F.
ASCE/SEI 7 quantifies torsional irregularity by the ratio of a building’s maximum edge
drift to the average drift of its two ends, given a static lateral force applied with 5% eccen-
tricity (in the most critical direction) from the CM (illustrated in Figure 1)—hereafter
referred to as the torsional irregularity ratio (TIR). For multi-story buildings, the static
force is applied with the same vertical distribution that is used for the other aspects of
design (e.g. the Equivalent Lateral Force Procedure force distribution). Thresholds for
classifying a building as torsionally irregular and extremely torsionally irregular are
TIR . 1.2 and TIR . 1.4, respectively.
Sources of torsional irregularity (as defined by TIR in ASCE/SEI 7) are torsional flexi-
bility, relative to translational flexibility, and eccentricity between a building’s center of
rigidity (CR) and CM. Therefore, TIR is a function of stiffness, but not strength. Figure 2
shows the effect of eccentricity on TIR for three different plan aspect ratios; while TIR
generally increases with eccentricity, it is seen that it saturates and even declines at
extremely large levels of eccentricity. An additional shortcoming of TIR is that larger plan
aspect ratios (b/a in Figure 2) generally result in larger TIR, causing long narrow buildings
to be easily classified as irregular. Despite its shortcomings, however, TIR is a satisfactory
means of classifying torsional irregularity in many situations. TIR has always been the

Figure 1. Illustration of torsional irregularity ratio (TIR) calculation.


DeBock et al. 1015

Figure 2. Illustration of how the torsional irregularity ratio (TIR) computed in ASCE/SEI 7-16 Table
12.3-1 varies with plan aspect ratio and eccentricity for a single-story structure (after Korolyk and
Wagner, 2016). The horizontal lines show the ASCE/SEI 7-16 thresholds for Type 1a torsional irregularity
and Type 1b extreme torsional irregularity. The thickened lines in the building plan (right) represent lines
of lateral resistance. All lines of lateral resistance are equally stiff for computing TIR in this example.

sole measure of torsional irregularity in ASCE/SEI 7, and therefore, it is also used to


quantify torsional irregularity in this study.

Objectives
ASCE/SEI 7-16 design requirements for seismic-induced torsion in buildings are evaluated
with two primary objectives: (1) gauge their effectiveness for providing resistance to
seismic-induced collapse for torsionally irregular buildings and (2) propose modifications
to ASCE/SEI 7 torsion design requirements that attempt to reduce deficiencies and/or
unnecessary sources of conservatism.

Method
Method overview
Torsion design provisions of ASCE/SEI 7 are assessed by comparing the collapse resis-
tance of torsionally irregular archetype buildings to the collapse resistance of a baseline
archetype that is regular in plan. Ideally, the design provisions should result in collapse
resistance that is relatively constant, independent of torsional irregularity. Avoiding
declines in collapse resistance as torsional irregularity increases is of primary importance,
but avoiding excessive conservatism is also emphasized (e.g. we try to avoid recommend-
ing requirements that cause large increases in collapse resistance in torsionally irregular
buildings relative to regular buildings).
Archetype buildings are represented by single-story nonlinear three-dimensional (3D)
models that are calibrated to represent the aggregate behavior of multi-story buildings. To
use single-story representative models effectively, the following steps are taken: (1) the
models are calibrated, so that their design properties (e.g. drift and stability coefficient)
are consistent with that of a multi-story design; (2) the nonlinear backbones of the single-
story models approximate the aggregate behavior of a multi-story building backbone; (3)
1016 Earthquake Spectra 37(2)

Figure 3. Plan view of the baseline archetype (left) and a generic archetype (right). Thickened lines
represent lines of lateral resistance.

nonlinear response history results for a set of single-story archetypes are validated by com-
parison to nonlinear response history results of corresponding multi-story 3D models. It is
important to note that single-story models are known to give widely varying results that
often do not agree with more sophisticated multi-story models, and prior studies (e.g.
Anagnostopoulos et al., 2010, 2015; De Stefano and Pintucchi, 2008) have shown that tak-
ing the three aforementioned steps is essential for gleaning meaningful results from single-
story models.

Development of archetype designs


Plan configurations. More than 2000 single-story archetype models are analyzed in this study.
The baseline configuration has a 1:1 plan aspect ratio with lines of lateral resistance along
the perimeter (as in Figure 3, left). Torsionally irregular archetypes are created by adjusting
the aspect ratio of the building plan and varying the locations of the lines of lateral resis-
tance, as illustrated by the generic plan shown on the right side of Figure 3.
Plan configurations of the majority of the archetypes have three categories, illustrated
in Figure 4; these are as follows:

1. Symmetric, ‘‘Sym’’—torsional irregularity is due to torsional flexibility only;


2. Double asymmetric, ‘‘Double Asym’’—inherent eccentricity in both orthogonal
directions;
3. Single asymmetric, ‘‘Single Asym’’—inherent eccentricity in one direction.

Additional ‘‘mixed’’ and ‘‘three-sided’’ archetypes with 7:3 aspect ratio are also
included. The mixed archetypes have three reinforced concrete (RC) wall lines and one line
of moment frames, as shown in Figure 5. The three-sided archetypes have plans identical
to the mixed archetypes, except that the line of moment frames is removed. The mixed
and three-sided archetypes are categorized as ‘‘Sym,’’ ‘‘Asym A,’’ and ‘‘Asym B,’’ as
shown in Figure 5.

Baseline archetypes. The baseline archetypes define three categories of buildings from which
all of the other archetypes are derived: ‘‘short,’’ ‘‘mid-rise,’’ and ‘‘mixed.’’ The short base-
line archetype defines a class of short-period buildings (upper bound design period,
DeBock et al. 1017

Figure 4. Schematic plan configurations of the archetype designs.

CuTa = 0.3 s) whose designs are predominately force-controlled, with little sensitivity to
P-Delta effects. Its properties are similar to a 2-story wood shear wall building from a
prior study (DeBock et al., 2016). The mid-rise baseline defines a class of buildings with
longer periods (CuTa = 2.0 s) whose designs are often controlled by drift limits and/or
stability requirements as they become more torsionally irregular. At a macrolevel, its prop-
erties are similar to the 12-story baseline RC special moment frame (SMF) building devel-
oped for the vertical irregularity studies in Chapter 6 of FEMA P2012 ‘‘Assessing Seismic
Performance of Buildings with Configuration Irregularities’’ (FEMA, 2018). The mixed
system archetypes represent a combination of special reinforced concrete shear walls
(RCSWs) and RC SMFs, either as a dual system or as a combined system in accordance
with section 12.2.2 or 12.2.3 of ASCE/SEI 7. The baseline archetype for the mixed system
1018 Earthquake Spectra 37(2)

Figure 5. Illustration of ‘‘mixed’’ archetype configurations. Three-sided archetypes have identical wall
layouts, but with the moment frame removed.

represents a building whose lateral force resisting system (LFRS) is comprised solely of
RCSWs. The properties of this baseline archetype are calibrated to represent the behavior
of a mid-rise RCSW building (CuTa = 0.93 s). The RCSW parallel to the moment frame
in the irregular archetypes is designed stronger than the walls in the wall-only direction
because it is stiffer than the moment frame and attracts more force. The lines of resistance
representing RC SMFs are similar to the 8-story baseline RC SMF developed for the ver-
tical irregularity studies in Chapter 6 of FEMA P2012, except that they are weaker,
because the frame is softer than the wall and attracts less force. Key design properties of
the baseline variants are summarized in Table 1. For comparison, the design properties of
a 12-story RC SMF building and 8-story RC wall building from FEMA P2012 are also
included in Table 1.
The purpose of the archetypes is to capture a range of global system behaviors that
could affect the impacts of torsion design requirements (e.g. buildings with periods in the
constant acceleration part of the spectrum and with low P-Delta sensitivity vs buildings
with periods beyond the constant acceleration part of the spectrum with significant P-
Delta effects).
DeBock et al. 1019

Table 1. Baseline archetype design properties


Baseline archetype Design Analytical Rc Cdd Cs (g)e Max drift Stability
period period ratio (%) coefficient
(CuTa, s)a,b (T1, s) (u)f

Short 0.3 0.41 6.5 4 0.154 0.7 0.01


Mid-rise 2.0 2.35 8 5.5 0.044 1.2 0.07
Comparison:12-story 2.13 2.58 8 5.5 0.044 1.2 0.07
RC SMF
Mixed, baseline has 0.93 1.15 6 5 0.108 0.7 0.02
only RC walls
Comparison: 8-story 0.93 1.14 6 5 0.108 Not checked Not checked
RCSW
RC: reinforced concrete; SMF: special moment frame; RCSW: reinforced concrete shear walls.
a
Cu is the coefficient for the upper limit on the calculated building period.
b
Ta is the approximate (empirical) fundamental building period in ASCE/SEI 7.
c
Response modification factor used to reduce seismic demands to account for system ductility.
d
Deflection amplification factor used by ASCE/SEI 7 to approximately convert elastic displacements to inelastic
displacements.
e
Seismic base shear coefficient.
f
A measure of P-Delta sensitivity in ASCE/SEI 7.

Proportioning the lateral system for seismic design. For each archetype, the properties of the four
lines of lateral resistance are adjusted to meet strength and stiffness requirements, consider-
ing the effects of inherent torsion, accidental torsion, and any additional code requirements
that apply (e.g. accidental torsion amplification). Two methods are used to proportion the
lateral system to meet code requirements: Method 1 assumes that the designer can vary
strength and stiffness of the system independently (decoupled)—for example, adjustments
to reinforcing ratio and/or section sizes in a RC moment frame; Method 2 assumes that
strength and stiffness remain proportional (coupled)—for example, adding an additional
wall to a shear wall building or adding an additional bay to a moment frame. The design
processes for proportioning lines of lateral resistance for each method are as follows:
Method 1: decoupled strength and stiffness:
1. Start with the ‘‘Baseline’’ lateral resistance;
2. Adjust stiffness to meet drift requirements, if necessary;
3. Adjust strength to exactly meet strength requirements. Sometimes this results in a
decrease in strength for certain lines of lateral resistance;
4. Adjust stiffness and/or strength to meet stability requirements, if necessary.

Method 2: coupled strength and stiffness:


1. Start with the ‘‘Baseline’’ lateral resistance;
2. Scale strength and stiffness by exactly the same amount until strength, drift, and
stability requirements are satisfied. Some lines of lateral resistance may have their
strength and stiffness decreased.

Decoupling strength and stiffness, as in Method 1, has been shown by prior studies (e.g.
Anagnostopoulos et al., 2010) to produce unrealistic behavior and questionable results.
Therefore, it is important that the proportioning of the archetypes by Method 1 enforces
strength, drift, and stability requirements, so that the resulting buildings are code-
1020 Earthquake Spectra 37(2)

conforming. Nonetheless, the resulting properties still may not be realistic in many cases,
but they allow the archetype design space to test the limits of what is allowed by the code
(i.e. strength and stiffness both at the bare minimums). The second method, coupled
strength and stiffness, is probably more realistic and a better approximation of real build-
ing behavior in many cases.

Nonlinear modeling and quantification of collapse resistance


Determination of collapse resistance using the FEMA P695 approach. Incremental dynamic analy-
sis (IDA; Vamvatsikos and Cornell, 2002) is performed for the 22 FEMA far-field ground
motions from FEMA P695 (FEMA, 2009). Both components of the ground motion are
applied simultaneously to a 3D model. Two IDAs are performed for each pair of ground
motions—one with the ground motion in its original orientation and another with the
ground motion rotated 90°, so that a total of 44 IDAs are performed for each archetype.
The buildings are considered collapsed when the maximum interstory drift ratio at any
location is greater than 10%, even if the structural model is still stable. Since some gravity
systems may not be capable of accommodating drift ratios as large as 10%, additional
analyses were performed with a 6% threshold to verify that the trends reported later in
the results section (and the conclusions we draw from them) do not change.
The median intensity causing collapse is divided by the maximum considered earth-
quake (MCE) intensity to determine a collapse margin ratio (CMR). The CMR is used in
the FEMA P695 method to compute probability of collapse, given MCE ground shaking.
In this study, we compare CMR of each variant to the CMR of the baseline to determine
the impacts of design and configuration characteristics on collapse resistance.

3D modeling approach. The 3D single-story models of the archetypes are made in OpenSees
(PEER, 2016). The LFRS is modeled with nonlinear shear springs located at each line of
lateral resistance. The diaphragm is modeled as rigid with a lumped mass located at its cen-
ter. For symmetric archetypes, the mass is offset by 5% of the building dimension in each
orthogonal direction (simultaneously) to induce torsion; the effect of the mass offset on
the collapse resistance is minimal except for in configurations that are torsionally flexible.
The rotational inertia of the building mass is determined assuming an even distribution of
the mass across the building plan. Leaning columns, or P-Delta columns, are located in
four quadrants of the building, typically one radius of gyration away from the CM. Figure
6 shows a plan view of a 3D model.
Validation of the simplified modeling approach for representing more complex multi-
story structures was performed on the mid-rise archetypes. A detailed explanation of the
validation process is available in Appendix A of FEMA P2012, but is summarized here.
Since the mid-rise archetypes are designed to emulate the macro-response of a 12-story RC
moment frame, a set of 10 3D 12-story lumped plasticity moment frame models with vary-
ing degrees of torsional irregularity were developed and used to check the performance of
the single-story 3D models. The lumped plasticity modeling approach used for the multi-
story 3D models is identical to that of the models described in Chapters 3 and 6 of FEMA
P2012, except that they are modeled in 3D rather than 2D; they incorporate cyclic dete-
rioration and post-capping negative stiffness. P-Delta effects in the multi-story models are
captured by vertical loads on four leaning columns, each on radius of gyration from the
CM. Comparisons of the multi-story and simplified model results are shown in Figures 7
and 8. For the cases tested, the trends in collapse resistance (quantified with CMR) in the
DeBock et al. 1021

Figure 6. Plan view of a 3D archetype model.

Figure 7. Comparison of the effects of eccentricity on CMR for a 2:1 aspect ratio of 3D 12-story
lumped plasticity moment frame model versus a single-story model that is intended to represent ‘‘mid-
rise’’ building behavior.
From FEMA P2012.

Figure 8. Comparison of the effects of torsional flexibility on CMR for a 2:1 aspect ratio of 3D 12-
story lumped plasticity moment frame model versus a single-story model that is intended to represent
‘‘mid-rise’’ building behavior.
From FEMA P2012.
1022 Earthquake Spectra 37(2)

Figure 9. Backbone properties of the baseline archetypes representing short, mid-rise, and RC wall-
type buildings.

simplified models, as a building becomes more sensitive to torsion, are similar for the sim-
plified single-story models and the more sophisticated multi-story models.

Nonlinear backbone models for lines of lateral resistance. The nonlinear backbones of the simpli-
fied models are calibrated to emulate the global properties of more sophisticated multi-
story models and are shown in Figures 9 to 11. The backbones for the short archetypes
are similar to a 2-story wood shear wall building from DeBock et al. (2016). The back-
bones of the mid-rise archetypes have a global force–displacement relationship similar to
the 12-story RC SMF baseline building from Chapter 6 of FEMA P2012. Simplified RC
DeBock et al. 1023

Figure 10. Example of nonlinear backbone calibration: simplified mid-rise model versus the 12-story
RC SMF lumped plasticity model from FEMA P2012.

Figure 11. Backbone properties for lines of lateral resistance representing the ‘‘wall’’ and ‘‘frame,’’ that
make up the mixed system (in the long direction of the mixed system archetypes).

wall backbones are calibrated to emulate the global force–displacement characteristics of


the 8-story RC wall building in Chapter 5 of FEMA P2012.
The process for calibrating the nonlinear backbone of the simplified models involves
(1) setting the initial stiffness to hit the target period, (2) setting the strength to match the
strength observed in the corresponding multi-story model from FEMA P2012, (3) setting
the height and post-yield displacement capacity, so that drifts and P-Delta effects in the
simplified models align with drifts and P-Delta effects in the multi-story FEMA P2012
models. Once these steps were completed, it was verified that the collapse resistances, that
is, CMRs, of the single-story simplified models are comparable (less than 10% different)
than the multi-story models that are used to calibrate them.
Comparisons of the elastic design properties of simplified single-story models and com-
parative multi-story models are shown in Table 1. Figure 10 shows example pushovers
1024 Earthquake Spectra 37(2)

comparing force versus drift ratio for the simplified mid-rise model to force versus inters-
tory drift ratio at a critical (bottom) story of the 12-story lumped plasticity moment frame
model from FEMA P2012; the 12-story example is highlighted because it was the most dif-
ficult to calibrate, due to the way the drifts concentrate at lower stories of the multi-story
model when it is damaged. Note that the heights of the single-story models are determined,
so that P-Delta effects best approximate those of their multi-story counterparts at inters-
tory drifts in the range of 3%–6%, where P-Delta effects are critical for simulating the
onset and progression of collapse. Heights of the short, RC wall, and mid-rise, single-story
nonlinear models are 12, 69, and 56 ft, respectively. Note, the mid-rise version is shor-
tened, so that P-Delta effects at the onset of collapse (when drifts are concentrating in a
few stories of the multi-story building) can be captured; since these drift concentrations are
less prevelant in the linear design step, a larger height (66 ft) is used to compute design
drifts that are consistent with the multi-story design (see Table 1).
The pushover results are shown with and without P-Delta to illustrate its significance.
The mid-rise simplified models somewhat over-predict P-Delta effects at low levels of dam-
age (i.e. drifts below 3%) because they are calibrated to capture the level of P-Delta that is
expected when the building is damaged and drifts are concentrated in fewer stories in the
building. As expected, P-Delta is most influential for the mid-rise archetypes and least sig-
nificant in the short archetypes.
In the mixed system archetypes, lines of lateral resistance representing the RC wall in
the mixed direction use a backbone similar to the backbones computed for the baseline RC
walls in the non-mixed direction; except that they start out stronger than the walls that are
in the wall-only direction because the wall in the mixed direction is stiffer than the frame
and attracts more force. Figure 11 shows the backbones for lines of resistance in the mixed
direction of the mixed archetype variants. The RC wall backbones are noticeably ductile,
and this is because they mimic a flexural-controlled 8-story shear wall (the baseline 8-story
RC wall from FEMA P2012). Shear critical squat RC walls, which are common in low-rise
construction, have lower drift capacities with peak strength occurring approximately near
2% interstory drift ratio (e.g. Rivera and Whittaker, 2019), which is similar to the drift
capacity of the short simplified model.
The cyclic properties of the nonlinear springs are modeled with peak-oriented material
properties having relatively low cyclic (i.e. cycle-to-cycle) deterioration. Figure 6 shows
cyclic behavior of the short baseline model. Cyclic deterioration effects are minimal in the
simplified models because cyclic deterioration is a system/material-dependent property,
and this study is intended to capture torsion effects in a more generic sense, primarily
focused on non-brittle building systems. For ductile systems, collapse resistance has been
shown to be relatively insensitive to cyclic deterioration properties in comparison with
other parameters, such as peak strength and post-capping negative stiffness, that is, in-
cycle deterioration (Ibarra, 2003, 2005). However, where cyclic deterioration is an impor-
tant feature in the development of a collapse mechanism, the simplified models may over-
predict the collapse resistance. If it is assumed that cyclic deterioration properties of the
elements in a given LFRS do not worsen when torsional irregularity is introduced (which
is reasonable in many cases because detailing requirements stay the same), then a bias in
the absolute collapse resistance that is due to using a generic level of cyclic deterioration
should have little impact on relative comparisons between archetypes.
DeBock et al. 1025

Figure 12. Effects of scaling strength and stiffness of the nonlinear backbones, illustrated with the short
archetype, with zoomed-in views of the initial stiffness shown in the top right of each plot. P-Delta
effects are excluded for this illustration. The solid black line represents the baseline; blue-dotted and
red-dashed lines represent incremental increases in strength and stiffness. (a) Identical increases in
strength and stiffness (coupled). (b) Stiffness increase less than strength increase (decoupled). (c)
Strength increase less than stiffness increase (decoupled).

Scaling of nonlinear backbones. The strength and stiffness of the nonlinear backbones that repre-
sent lines of lateral resistance are scaled to match the changes in strength and stiffness that are
determined in the design step. Scaling of the strength and stiffness is applied to each branch of
the nonlinear backbone; therefore, ductility does not change. Effects of modifying strength
and/or stiffness of the backbones are illustrated in Figure 12; Figure 12(a) shows perfectly
coupled strength and stiffness; Figure 12(b) and (c) shows decoupled strength and stiffness.

Results
Collapse performance under current code requirements
Figure 13 shows trends in collapse resistance with TIR for short and mid-rise archetypes
that are proportioned according to ASCE/SEI 7-16. The vertical axis ordinate, collapse
resistance relative to the baseline, is computed by dividing the CMR of the archetype by
the CMR of the baseline. Relative collapse resistance greater than 1.0 means that an arche-
type has better collapse resistance than the baseline and vice versa for relative collapse
resistance less than 1.0. The horizontal axis ordinate is TIR, as computed in Table 12.3-1
of ASCE/SEI 7-16. At high levels of eccentricity, TIR saturates and even decreases as
inherent eccentricity increases (see Figure 2); this limitation of the TIR causes some of the
trends to turn around or ‘‘double back’’ at high levels of torsional irregularity (e.g. the
‘‘2:1 single asymmetric’’ archetypes) because TIR begins decreasing even as the structure
becomes increasingly irregular (in terms of its eccentricity).
With only a few exceptions, collapse resistance increases with increasing torsional irre-
gularity due to the additional design requirements imposed on torsionally irregular build-
ings. The increase in collapse resistance that is observed at TIR = 1.4 is due to triggering a
redundancy factor (r) of 1.3. Note that some systems with TIR \ 1.4 may not comply with
Table 12.3-3 of ASCE/SEI 7-16 and therefore would also be designed for a redundancy
factor of 1.3. For context, the redundancy factor is used by ASCE/SEI 7 to penalize sys-
tems that lack redundancy (see, for example, Wen and Song, 2003, for more background
on redundancy).
1026 Earthquake Spectra 37(2)

Figure 13. Collapse resistance relative to the baseline designed with ASCE/SEI 7-16 provisions. (a) Short
variants. (b) Mid-rise variants. Fitted median and logarithmic standard deviation are overlaid. Archetypes
with decoupled and coupled strength and stiffness are plotted with solid and dashed lines, respectively.

Observations
A number of sensitivity studies have been conducted to determine how effective the various
torsion design requirements are for avoiding declines in collapse resistance due to torsional
irregularity. Some of the primary observations from the code-conforming archetypes are
as follows:

1. Accidental torsion design requirements are relatively unimportant for buildings


that are not torsionally irregular, as shown by prior studies, for example,
Stathopoulos and Anagnostopoulos (2006, 2010) and DeBock et al. (2014). Note
that the building designs for Figure 13 excluded accidental torsion if the building
was not irregular, as per ASCE/SEI 7-16.
2. Buildings that rely heavily on lines of lateral resistance perpendicular to the design
earthquake force to resist torsional moments tend to be the lowest performers,
because those lines of resistance also resist seismic demands in the perpendicular
direction and consequently are ‘‘overworked.’’ This is most prevalent and observa-
ble in the single-asymmetric archetypes. It is well-recognized in the literature that
inherent eccentricity can lead to significant loading in lines of resistance that are
orthogonal to the ground motion direction; Goel and Chopra (1990) is one of the
earlier studies to have demonstrated this.

Additional observations, which are expanded in the following sections, are as follows:

1. The results of this study do not support the prohibition of most extremely torsion-
ally irregular buildings in SDC E and F;
2. Small modifications to torsion provisions may provide more consistent collapse
performance across all building types studied herein;
3. Measuring drifts at the building edge rather than the CM for checking drift limits
and stability requirements is important for providing adequate collapse resistance,
although this requirement may produce conservative results for buildings with
higher TIRs;
DeBock et al. 1027

4. Flawed as it may be, TIR appears to be a useful indicator of torsional irregularity


for most designs. However, triggering a torsional irregularity in buildings with
.75% of their strength on the same side of its CM may be a useful additional
requirement;
5. Accounting for accidental torsion by offsetting the mass 5% in each direction with
modal response spectrum analysis (MRSA), rather than adding static accidental
torsion moments to a concentric analysis, may not be a conservative enough design
approach for highly torsionally flexible buildings, and it could potentially lead to
unsafe designs.

Recommended minimum torsion design requirements. Through a number of iterations, a set of


minimum torsion design requirements is recommended. Results for the same archetype
design space, but designed with the recommended provisions, are shown in Figure 14. The
lowest performing irregular archetype in Figure 14 (mid-rise single asymmetric with maxi-
mum eccentricity) is 3% worse than the baseline, and the rest of the irregular archetypes
outperform the baseline. Figure 15 shows the performance of three-sided and mixed sys-
tem variants, which also generally improve, as torsion design requirements take effect. The
recommended revised torsion design requirements are arguably less restrictive than ASCE/
SEI 7-16 requirements and result in more consistent collapse resistance for the configura-
tions studied. The proposed changes to the current ASCE/SEI 7-16 torsion design provi-
sions, which are used to proportion the archetype models, are as follows.
Apply the 100%–30% orthogonal load combination rule for torsionally irregular build-
ings. In many circumstances, ASCE/SEI 7 allows the two orthogonal earthquake direc-
tions to be considered independently for design (i.e. not simultaneously applied). Using the
orthogonal load combination procedure for torsionally irregular buildings increases the
design force on lines of resistance that resist a significant portion of the torsional moment
from the lateral earthquake force in the orthogonal direction (in addition to lateral forces
from the direction in which they are oriented). Because these lines of lateral resistance are

Figure 14. Collapse resistance relative to the baseline designed with the suggested modifications to the
ASCE/SEI-7 torsion design provisions. (a) Short variants. (b) Mid-rise variants. Fitted median and
logarithmic standard deviation are also overlaid. Archetypes with decoupled and coupled strength and
stiffness are plotted with solid and dashed lines, respectively.
1028 Earthquake Spectra 37(2)

Figure 15. Collapse resistance relative to the baseline designed with the suggested modifications to the
ASCE/SEI-7 torsion design provisions. (a) Short three-sided variants. (b) Mixed variants. Fitted median
and logarithmic standard deviation are also overlaid. Archetypes with decoupled and coupled strength
and stiffness are plotted with solid and dashed lines, respectively.

‘‘double-worked’’ in an earthquake (i.e. resist the ground motion to which they are parallel as
well), it should be reflected in their design. The beneficial effects of adding this requirement
are most apparent in ‘‘1:1 single asymmetric’’ archetypes (e.g. compare Figures 13 and 14
results for 1:1 single asymmetric archetypes); as per the rules in ASCE/SEI 7-16, these types
of structures can have lower collapse resistance than regular buildings, whereas they have
higher collapse resistance than regular buildings if the orthogonal combination is considered.
Classify buildings as torsionally irregular if .75% of the story strength is on one side of
the CM. Rationale for this recommendation is given in the following section.
Limit the 5% mass offset method for ‘‘directly’’ simulating accidental torsion in MRSA
for buildings that are extremely torsionally irregular: This has already been added to
ASCE/SEI 7-16 in Supplement #2. However, this limitation is probably not necessary for
all extremely torsional irregular buildings and deserves further study. Rationale for this
recommendation is explained in its own section later in the article.
Relax the triggers for requiring a redundancy factor (r) of 1.3: ASCE/SEI 7-16 requires
that all buildings classified as extremely torsionally irregular have r = 1.3, regardless of
whether the system is redundant or not. However, the results indicate that only buildings
that are either extremely torsionally irregular in both orthogonal directions or have lines
of lateral resistance on only one side of the CM actually need to be designed using r = 1.3
to provide adequate collapse resistance.
Reconsider if prohibiting the equivalent lateral force (ELF) design procedure is needed
in ASCE/SEI 7: The archetypes in this study are proportioned with static procedures and
are shown to perform satisfactorily under the proposed design provisions. In addition, the
ELF design procedure has been shown in prior studies (e.g. FEMA, 2018) to consistently
produce more conservative designs for multi-story buildings than the modal response spec-
trum procedure; there may not be a need to prohibit the ELF procedure.
The prohibition of configurations classified as extremely torsionally irregular in SDC E
and F may not be necessary, according to the findings of this study: All of the archetype
DeBock et al. 1029

models in this study with TIR . 1.4 would be classified as extremely torsionally irregular
and banned in SDC E and F by ASCE/SEI 7. However, the majority of the archetypes with
TIR . 1.4 show collapse resistance greater than the baseline, after being proportioned with
the recommended provisions. This finding was further investigated by analyzing the arche-
types with the FEMA P695 near-field ground motions that have pulse characteristics; the
absolute collapse capacity reduced, but nearly identical trends in the collapse resistance rela-
tive to the baseline were observed (\5% difference). It should be noted that there is a wide
variety of pulse periods in the near-field record set, so similar trends are to be expected.

Rationale for triggering Type 1a torsional irregularity when .75% of strength is on one side of the
CM. In many cases, buildings with significant inherent eccentricity are already classified as
torsionally irregular due to having TIR . 1.2. However, buildings that are rectangular in
plan can have significant inherent eccentricity in the long direction and still have TIR \ 1.2.
For example, the three-sided and mixed system archetype buildings all have TIR less than 1.2
for the cases where lines of resistance in the short-direction are at or near the perimeter. To
illustrate this point, Figure 16 shows the performance of the mid-rise three-sided variants with
the proposed modifications, with and without the additional torsional irregularity classifica-
tion. The circled results in Figure 16(b) would not be classified as torsionally irregular based
solely on TIR (as it currently is in ASCE/SEI 7-16). As with the ‘‘1:1 single asymmetric’’
archetypes with TIR . 1.2, this structural configuration relies on the strength of the short
end walls for resistance to earthquakes in both directions. Consequently, the 100%–30%
orthogonal combination needs to be triggered to help these walls to be less ‘‘overworked.’’

Importance of checking drift and stability requirements at the building’s edge for torsionally irregular
buildings. Enforcing drift and stability limits at the building edge (rather than CM) is a cur-
rent ASCE/SEI requirement that is important for providing adequate collapse resistance
for torsionally sensitive buildings. Figure 17 shows steady declines in collapse resistance
for the mid-rise archetypes when drift and stability checks are performed at the CM rather
than the building edge. Note that all other design requirements are unchanged for creating

Figure 16. Trends in collapse resistance for mid-rise three-sided variants. (a) Designed with the
recommended torsion provisions. (b) Designed with the recommended torsion provisions, except that
torsional irregularity is determined only by TIR (as in ASCE/SEI 7-16).
1030 Earthquake Spectra 37(2)

Figure 17. Trends in collapse resistance for mid-rise archetypes proportioned according to ASCE/SEI
7-16 to resist seismic forces, but with a redundancy (r) of 1.0 and drift/stability checks performed at the
CM, rather than the edge of the building. Archetypes with decoupled strength and stiffness are
represented by solid lines; dashed lines represent archetypes with coupled strength and stiffness.

the archetype designs in Figure 17, except that the redundancy factor (r) is held at 1.0 to
remove the sudden jump in the trend that is observed when r changes to 1.3.

Application of 5% mass offsets to simulate accidental torsion with MRSA. ASCE/SEI 7 allows for
accidental torsion to be accounted for in one of the two ways in the MRSA design proce-
dure: (1) move the CM 5% of the plan dimension in each of the four directions and design
for an envelope of the demands, (2) add static accidental torsion moments (that result in
story moments equal to the story shear times 5% of the orthogonal plan dimension multi-
plied by the torsion amplification factor) to a concentric MRSA analysis, according to sec-
tion 12.8.4 of ASCE/SEI 7-16. Accounting for accidental torsion by applying 5% mass
offsets in lieu of the static accidental torsional moments set forth in section 12.8.4 of ASCE/
SEI 7-16 may not provide adequate collapse safety for some building configurations whose
fundamental mode of vibration is a torsional mode with relatively little modal mass in the
lateral direction. This is caused by the torsion mode(s) having too little effect on the analy-
sis, due to the fact that traditionally only the translation component of the mass is used to
determine modal participation in MRSA (e.g. as discussed in Basu et al. (2014)). The sym-
metric archetypes are examples of buildings with torsional modes that have little modal par-
ticipation in their lateral directions.
Table 2 shows the effects on the required strength and stiffness for plan-symmetric tor-
sionally irregular archetypes when the two design methods for accidental torsion are used.
At moderate levels of torsional irregularity, the 5% mass offset method significantly
increases force and drift demands. However, at high levels of torsional irregularity, offset-
ting the mass 5% makes very little difference in the design values. This finding is consis-
tent with a previous study by De la Llera and Chopra (1994) and is also illustrated in
DeBock et al. (2019); they showed that simulating accidental torsion with 5% mass offsets
in MRSA amplifies accidental torsion when the lateral and torsional modes have similar
periods, but the amplifying effect of the dynamic analysis diminishes as the torsional
period becomes significantly longer than the translational period. For reference, the tor-
sional periods for the buildings in Table 2 are .40% longer than the lateral periods.
DeBock et al. 1031

Table 2. Comparison of strength and stiffness requirements in symmetric mid-rise archetypes using a
concentric MRSA with static accidental torsional moments versus MRSA with 5% mass offsets to directly
simulate accidental torsion
Plan aspect TIRa Required strength relative Required stiffness relative
ratio to baseline to baselineb
Static 5% mass Difference Static 5% mass Difference
accidental offsets (%) accidental offsets (%)
torsion torsion

1 1.31 1.15 1.12 23 1.05 1.05 0


1.56 1.66 1.40 216 1.47 1.00 232
2.25 2.71 1.41 248 4.42 1.00 277
2 1.22 1.14 1.45 27 1.00 1.40 40
1.53 1.35 1.20 211 1.42 1.20 215
2.04 2.27 1.14 250 3.74 1.12 270
4 1.26 1.17 1.55 32 1.00 1.50 50
1.60 1.43 1.24 213 1.65 1.30 221
2.09 2.37 1.17 251 3.99 1.20 270
a
Due to differences in stiffness, TIR is slightly different between the versions with static accidental torsional moments
versus those designed with MRSA and 5% mass offsets. The reported TIR values are for the latter.
b
The stability coefficient and story drift of the baseline are 0.07% and 1.2%, respectively, so the required stiffness does
not increase immediately as torsional irregularity is introduced.

Figure 18 shows the performance of symmetric mid-rise archetypes when they are
designed using MRSA with mass offsets equal to 65% of the perpendicular plan dimen-
sion. The results show that directly accounting for accidental torsion in the MRSA proce-
dure using 5% mass offsets in the structural model can lead to poor performance in

Figure 18. Collapse performance of symmetric archetypes designed with MRSA with accidental torsion
applied directly by offsetting the mass 65% of the perpendicular building dimension in the structural
model.
1032 Earthquake Spectra 37(2)

extremely torsionally irregular buildings. Therefore, it is recommended to use statically


applied accidental torsional moments for buildings classified as extremely torsionally irre-
gular; this method is shown to be somewhat conservative in Figure 14.

Limitations
The results obtained by this study should be viewed in light of a number of limitations
inherent to the modeling and analysis methods employed. First, it is important to note
that the models are generic, single-story, and relatively simple, as necessitated by the large
number of building variants considered in this study. By employing the simplified model-
ing method, we expect to observe trends that are similar to what one would get from com-
plex detailed models, and this is verified for selected cases, but the overall results are more
crude than what more detailed multi-story models would produce. Notable behaviors that
are not captured by the simplified models are higher-mode effects and biaxial loading
effects (e.g. I- and H-shaped shear wall configurations in which the walls are not planar
and uncoupled, as is assumed in the simple models employed in this study). Failure of the
gravity system is also not explicitly modeled, and some structural systems will not be able
to achieve 10% (or 6%) drift without local gravity failures. The models also incorporate
limited cyclic (cycle-to-cycle) deterioration because they are generic in nature; therefore
any system-specific differences in the effects of torsion that are due to the effects of cycle-
to-cycle deterioration are not captured in this study. The modeling method also assumes
that increasing or decreasing the strength of the LFRS does not change its ductility, which
may not always be the case.
In addition, this study focuses on CMR, which is a measure of the median collapse resis-
tance. Differences in uncertainty between regular and irregular buildings are beyond the
scope of this study and are not considered. If the performance of irregular archetypes is less
certain than regular archetypes, then a larger CMR would be required to have the same
collapse probability.
The focus of the study is on high seismic category buildings, where torsion design pro-
visions are most restrictive; therefore, nonductile archetypes are not included in the assess-
ment. Although past studies (e.g. DeBock et al., 2014) have observed similar trends in
collapse resistance versus torsional irregularity for ductile and nonductile systems, this
study does not verify that conclusion for the archetype configurations that are studied.
For seismic design of torisonally irregular buildings, this study suggests solutions that are
strongly tied to the US Standard ASCE/SEI 7. The goal is to make suggestions that are prac-
tical to implement under the current code construct. Consequently, more elegant (and perhaps
more effective) solutions are not considered. Novel solutions to seismic torsion design have
been put forth by past research (see Anagnostopoulos et al. (2015) for an excellent summary
of the expansive literature on seismic torsion in buildings). Notably, De La Llera and Chopra
(1995) propose a method that considers the radius of gyration of the floor plate relative to the
maximum plan dimension along with the ratio of the torsional and lateral frequencies to pro-
portion buildings for torsion, and that method was later validated by Lin et al. (2001) through
observations of real building responses to earthquake shaking.

Conclusion
Assessments of a broad suite of archetype building configurations indicate that ASCE/SEI
7-16 seismic design provisions for torsion may be conservative in general. However, the
DeBock et al. 1033

provisions may be unconservative for buildings configured such that lines of lateral resis-
tance orthogonal to the design earthquake force are heavily utilized to resist torsional
moments.
Modifications to the ASCE/SEI 7 seismic design provisions for torsion are suggested,
based on the results of the study. For the archetype buildings studied, the proposed modifi-
cations somewhat decrease the overall conservatism of the seismic torsion provisions and
improve the performance of buildings with configurations that pose increased collapse risk
under current ASCE/SEI 7-16 requirements. These include applying the 100%–30% ortho-
gonal combination rule for torsionally irregular buildings, classifying buildings with high
strength eccentricity as torsionally irregular, and softening the triggers for requiring a
redundancy factor (r) of 1.3.
For the archetype buildings considered, the results indicate that the prohibition of
extremely torsionally irregular buildings in SDC E and F may be unnecessary, given that
the proposed modifications to the ASCE/SEI 7-16 seismic design procedures are imple-
mented and accidental torsion is applied statically (i.e. not by offsetting the mass 65% in
MRSA). In addition, the results suggest that prohibiting the ELF procedure on the basis
of torsional irregularity may not be necessary.

Acknowledgements
The study is part of a project managed by the Applied Technology Council (ATC) under FEMA
contract HSFE60-12-D-0242 task order HSFE60-16-J-0223, entitled ‘‘ATC-123: Improving Seismic
Design of Buildings with Configuration Irregularities.’’ Any opinions contained in this article repre-
sent those of the authors and do not necessarily represent the opinions of ATC or FEMA. The
authors thank the ATC-123 working group members, technical committee, and project review panel
for their valuable feedback and contributions.

Declaration of conflicting interests


The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/
or publication of this article.

Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or
publication of this article: FEMA contract HSFE60-12-D-0242 task order HSFE60-16-J-0223, enti-
tled ATC-123: Improving Seismic Design of Buildings with Configuration Irregularities.

References
Anagnostopoulos SA, Alexopoulou C and Stathopoulos KG (2010) An answer to an important
controversy and the need for caution when using simple models to predict inelastic earthquake
response of buildings with torsion. Earthquake Engineering & Structural Dynamics 39: 521–540.
Anagnostopoulos SA, Kyrkos MT and Stathopoulos KG (2015) Earthquake induced torsion in
buildings: Critical review and state of the art. Earthquakes and Structures 8(2): 305–377.
ASCE (2016) Minimum design loads and associated criteria for buildings and other structures. ASCE
Standard ASCE/SEI 7-16, 8 August 2017. Reston, VA: American Society of Civil Engineers.
Basu D, Constantinou MC and Whittaker AS (2014) An equivalent accidental eccentricity to
account for the effects of torsional ground motion on structures. Engineering Structures 69: 1–11.
De la Llera JC and Chopra AK (1994) Using accidental eccentricity in code-specified static and
dynamic analyses of buildings. Earthquake Engineering & Structural Dynamics 23(9): 947–967.
1034 Earthquake Spectra 37(2)

De la Llera JC and Chopra AK (1995) Estimation of accidental torsion effects for seismic design of
buildings. Journal of Structural Engineering 121(1): 102–114.
De Stefano M and Pintucchi B (2008) A review of research on seismic behaviour of irregular building
structures since 2002. Bulletin of Earthquake Engineering 6(2): 285–308.
DeBock DJ, Hohener C and Valley M (2019) Does accidental torsion prevent collapse? How collapse
potential for torsionally flexible structures is affected by the method of considering accidental
torsion. Structure Magazine, March, 2019, pp. 14–16.
DeBock DJ, Liel AB, Haselton CB, Hooper JD and Henige RA (2014) Importance of seismic design
accidental torsion requirements for building collapse capacity. Earthquake Engineering &
Structural Dynamics 43(6): 831–850.
DeBock DJ, Wade KF, Cook DT and Haselton CB (2016) New developments in FEMA P-58
seismic risk assessment of wood light-frame buildings. In: Proceedings of the 85th annual SEAOC
convention, Maui, HI, 12–15 October.
FEMA (2009) FEMA P695: Quantification of building seismic performance factors. Prepared by
ATC. Available at: https://www.fema.gov/media-library-data/20130726-1716-25045-9655/fema_
p695.pdf
FEMA (2018) FEMA P2012: Assessing seismic performance of buildings with configuration
irregularities. Prepared by ATC. Available at: https://www.fema.gov/sites/default/files/2020-08/
fema_assessing-seismic-performance-irregularities_p-2012.pdf
Goel RK and Chopra AK (1990) Inelastic seismic response of one-storey, asymmetric-plan systems:
Effects of stiffness and strength distribution. Earthquake Engineering & Structural Dynamics
19(7): 949–970.
Ibarra L (2003) Global Collapse of Frame Structures Under Seismic Excitations, PhD Dissertation,
Department of Civil and Environmental Engineering, Stanford University, Stanford, California.
Ibarra LF, Medina RA and Krawinkler H (2005) Hysteretic models that incorporate strength and
stiffness deterioration. Earthquake Engineering & Structural Dynamics 34(12): 1489–1511.
Korolyk MJ and Wagner AM (2016) Improving design for plan torsion. In: Proceedings of the 85th
annual SEAOC convention, Maui, HI, 12–15 October.
Lin WH, Chopra AK and Llera JCDL (2001) Accidental torsion in buildings: Analysis versus
earthquake motions. Journal of Structural Engineering 127(5): 475–481.
PEER (2016) OpenSEES (Open system for earthquake engineering simulation). Available at:
Opensees.berkeley.edu (accessed March 2016).
Rivera JP and Whittaker AS (2019) Damage and peak shear strength of low-aspect-ratio reinforced
concrete shear walls. Journal of Structural Engineering 145(11): 04019141.
Stathopoulos KG and Anagnostopoulos SA (2006) Importance of accidental design eccentricity for
the inelastic earthquake response of buildings. In: Proceedings of the 13th European conference,
Earthquake Engineering, Geneva, 3–8 September.
Stathopoulos KG and Anagnostopoulos SA (2010) Accidental design eccentricity: Is it important
for the inelastic response of buildings to strong earthquakes? Soil Dynamics and Earthquake
Engineering 30: 782–797.
Vamvatsikos D and Cornell CA (2002) Incremental dynamic analysis. Earthquake Engineering &
Structural Dynamics 31(3): 491–514.
Wen YK and Song SH (2003) Structural reliability/redundancy under earthquakes. Journal of
Structural Engineering 129(1): 56–67.

You might also like