Professional Documents
Culture Documents
3
3
It is the roof
Procedures displacement, termed the Target
Displacement in FEMA 356. In particular,
3.1 Introduction this study was aimed at identifying and
This chapter summarizes the results of quantifying the errors in these procedures
studies to assess the ability of current when applied to SDOF systems. For this
approximate nonlinear static procedures to purpose, approximate total displacements
estimate the maximum displacement of computed with ATC-40 and with FEMA
inelastic structural models. Initial studies 356 were compared with the results of
evaluated both the Coefficient Method of nonlinear response history analyses of
FEMA 356 and the Capacity Spectrum SDOF oscillators. The nonlinear
Method of ATC-40. The use of NSPs responsehistory analyses are “exact” for
(nonlinear static procedures) has the assumptions made for the properties of
accelerated in the United States since the the oscillator (damping ratio and type of
publication of ATC-40, FEMA 273/274 hysteretic behavior) and for the particular
and FEMA 356 documents. As a ground motion record. Thus these results
consequence there is valuable information are a useful benchmark to evaluate the
available on the practical application of approximate procedures. Of particular
these inelastic analysis procedures (see interest is the extent to which the
Appendix B, “Summary of Practice using approximate methods might tend to
Inelastic Analysis Procedures”). Various overestimate or underestimate
researchers and practicing engineers have displacement demands (introduce bias)
found that, in some cases, different and the spectral regions or strength levels
inelastic analysis methods give for whichythese biases are likely to occur.
substantially different estimates for Errors were quantified through statistical
displacement demand for the same ground analyses. A large number of SDOF
motion and same SDOF oscillator systems (with a wide range of periods of
(Aschheim et al., 1998; Chopra and Goel vibration, lateral strengths, and hysteretic
1999a,b, 2000; Albanessi et al., 2000; behavior) were subjected to a relatively
Kunnath and Gupta, 2000; Lew and large number of recorded earthquake
Kunnath, 2000; Yu et al, 2001; Zamfirescu ground motions. Ground motions included
and Fajfar, 2001; MacRae and Tagawa, near fault and far fault records
2002). representative of site conditions ranging
from rock to very soft soil. However, it is
The disparities in displacement predictions recognized that there may be some
highlight the need for comparison and situations that deviate from those used in
further study of these different approaches this investigation. Caution should be used
(see Appendix A, “Summary of Research when extrapolating the results presented in
on Inelastic Analysis Procedures”). The this evaluation for ground motions and site
objective of this evaluation was to study conditions that differ substantially. Section
the 3.2 describes the period of vibration,
accuracy of the approximate methods damping ratio, lateral strength, and
described in ATC-40 and FEMA 356 for hysteretic behavior of the SDOF systems
estimating the maximum displacement that were considered in this investigation.
demand of inelastic This section also describes the types and
single-degree-offreedom (SDOF) systems. characteristics of the recorded ground
This global displacement is a spectral motion records that were used as well as
displacement, termed the Performance the error measures computed in this study.
Section 3.3 describes the evaluation of the is discussed on page 105 of FEMA 450-2,
simplified inelastic analysis procedure in NEHRP Recommended Provisions for
ATC-40 to estimate the maximum Seismic Regulations for New Buildings
displacement of inelastic systems using and Other Structures, Part 2: Commentary
equivalent linearization. Section 3.4 (BSSC, 2003). Nine levels of normalized
provides a corresponding evaluation of the lateral strength were considered,
simplified analysis procedure in FEMA corresponding to R = 1, 1.5, 2, 3, 4, 5, 6, 7
356. In particular, this chapter provides an and 8. Four different hysteretic behaviors
evaluation of coefficients C1, C2 and C3 were used in this
in this method. Finally, Section 3.5 study (see Figure 3-1):
summarizes the dynamic response of • The elastic perfectly plastic (EPP) model
nonlinear elastic, or rocking, oscillators. A is used as a reference model. This model
complete compilation of the evaluation has been used widely in previous
study data is provided in Appendix C, investigations and therefore it represents a
“Supplemental Data on the Evaluation of benchmark to study the effect of hysteretic
Current Procedures.” behavior. Furthermore, recent studies have
shown that this is a reasonable hysteretic
3.2 Evaluation Procedures model for steel beams that do not
3.2.1 Hysteretic Characteristics experience lateral or local buckling or
SDOF systems with initial periods of connection failure (Foutch and Shi, 1998).
vibration between 0.05 s and 3.0 s were • The stiffness-degrading (SD) model
used in this investigation. A total of 50 corresponds to the modified-Clough
periods of vibration were considered (40 model, as originally proposed by Clough
periods between 0.05 s and 2.0 s, equally (1966) and as modified by Mahin and Lin
spaced at 0.05 s, and 10 periods between (1983). This model was originally
2.0 s and 3.0 s, equally spaced at 0.1 s proposed as representative of well detailed
intervals). The initial damping ratio, β, was and flexurally controlled reinforced
assumed to be equal to 5% for all systems. concrete structures in which the lateral
In this study the lateral strength is stiffness decreases as the level of lateral
normalized by the strength ratio R, which displacement increases.
is defined as • The strength and stiffness-degrading
(3-1) (SSD) model is aimed at approximately
where m is the mass of the SDOF reproducing the hysteretic behavior of
oscillator, Sa is the spectral acceleration structures in which lateral stiffness and
ordinate corresponding to the initial period lateral strength decrease when subjected to
of the system, and Fy is the lateral yield cyclic reversals. In this model, the amount
strength of the system. The numerator in of strength and stiffness degradation is a
Equation 3-1 represents the lateral strength function of the maximum displacement in
required to maintain the system elasticity, previous cycles as well as a function of the
which sometimes is also referred to as the hysteretic energy dissipated. This model is
elastic strength demand. Note that this similar to the three-parameter model
R-factor is implemented in IDARC (Kunnath et al.,
not the same as the response-modification 1992). When properly calibrated, this
coefficient conventionally used for design model can reproduce the response of
purposes. This R-factor poorly detailed reinforced concrete
is the design R-factor divided by the structures relatively well. An example is
overstrength factor, omega sub-zero. This shown in Figure 3-2, in which the
load-deformation relationship of a poorly
detailed beam-column joint tested at the response-history analyses were run as part
University of Washington (Lehman et al., of this investigation. In this study, the
2000) is compared with the response results computed with nonlinear response
computed with the SSD model. A single history analyses are the benchmark
set of parameters representing severe maximum displacements,( Δ i)ex. The
strength and stiffness degradation was maximum displacements estimated with
used for this model. The type of simplified inelastic procedures of ATC-40
degradation that is captured by this model and FEMA 356 are the approximate
only includes cyclic degradation. Note that maximum displacements, ( Δ i)app of the
the post-elastic stiffness in any cycle is inelastic system. It should be noted that the
always equal to zero or greater. Thus, nonlinear response-history analyses are
These are representative of site class B, as “exact” only for the SDOF oscillator with
defined by the NEHRP Recommended the assumed properties and for the
Provisions for Seismic Regulations for particular ground motion. The uncertainty
New Buildings and Other Structures, Part of the modeling assumptions with respect
I, Provisions (BSSC, 2000)1. The second to the actual building is not included in
group consisted of records obtained on either the nonlinear response history
stations on very dense soil or soft rock analyses or the approximate analyses. The
with average shear wave velocities nonlinear response-history results are a
between 360 m/s (1,200 ft/s) and 760 m/s, convenient benchmark. In order to
while the third group consisted of ground evaluate the accuracy of these approximate
motions recorded on stations on stiff soil procedures, an error measure was defined
with average shear wave velocities as the ratio of
between 180 m/s (600 ft/s) and 360 m/s. approximate, ( Δ i)app, to benchmark, ( Δ
These are consistent with site class C and i)ex, maximum displacement as follows:
D respectively. The fourth group (3-2)
corresponds to ground motions recorded
on very soft soil conditions with shear This error measure was computed for each
wave velocities smaller than 180 m/s, period of vibration T and each level of
which can be classified as site class E. normalized lateral strength R. Values of
Finally, the fifth group corresponds to 20 ET,R larger than one indicate that the
ground motions influenced by near field approximate method overestimates the
forwarddirectivity maximum displacement of the SDOF
effects. Detailed listings of the ground system and values smaller than one
motions are presented in Appendix C. indicate underestimation. A total of
320,000 individual errors were computed
3.2.3 Error Measures and Statistical Study in this study. In order to identify whether
The maximum displacement of each the approximate methods, on average, tend
inelastic SDOF system was estimated with to overestimate or underestimate
the simplified inelastic procedures in maximum displacements of inelastic
ATC-40 and FEMA 356 when subjected to systems, mean errors were computed as
each of the ground motions. The maximum follows:
displacement of each inelastic SDOF (3-3)
system was then computed using nonlinear
response history analyses. The maximum where n is the number of records in each
displacement is defined as the maximum group of ground motions. Mean errors
of the absolute value of the displacement were computed for each hysteretic
response. A total of 180,000 nonlinear behavior type, each period of vibration (or
for each normalized period of vibration as elastic SDOF system that has a period and
will be explained later) and each level of a damping ratio that are larger than those
normalized lateral strength. Therefore, of the initial values for the nonlinear
mean errors computed with Equation 3-3 system. The elastic SDOF system that is
do used to estimate the maximum inelastic
not allow for underestimations in a displacement of the nonlinear system is
spectral region to be compensated by usually referred to as the equivalent or
overestimations in another spectral region. substitute system. Similarly, the period of
Information on the bias for each period, vibration and damping ratio of the elastic
for each type of hysteretic behavior, for system are commonly referred to as
each level of normalized lateral strength, equivalent period and equivalent damping
and for each site class is retained. The ratio, respectively. The concept of
sample mean error computed with equivalent viscous damping was first
Equation 3-3 is an unbiased estimator of proposed by Jacobsen (1930) to obtain
the mean error of the approximate solutions for the steady
population. Therefore, it provides an forced vibration of damped SDOF systems
estimate of the average error produced by with linear force displacement
the approximate methods. However, it relationships but with damping forces
provides no information on the dispersion proportional to the nth power of the
of the error. In order to obtain a measure velocity of motion when subjected to
of the dispersion of the errors produced by sinusoidal forces. In this pioneering study,
the approximate the stiffness of the equivalent system was
methods, the standard deviation of the set equal to the stiffness of the real system
error was computed as and the equivalent viscous damping ratio
(3-4) was based on equating the dissipated
energy per cycle of the real damping force
The square of the sample standard to that of the equivalent damping force.
deviation of the error computed with Years later, the same author extended the
Equation 3-4 is an unbiased estimator of concept of equivalent viscous damping to
the variance of the error in the population. yielding SDOF systems (Jacobsen, 1960).
The standard deviation of the error was Since then, there have been many methods
computed for each type of hysteretic proposed in the literature. Review of the
behavior, for each level of normalized earlier equivalent linear methods can be
lateral strength, and for each site class. found in Jennings (1968), Iwan and Gates
(1979), Hadjian (1982), Fardis and
3.3 Evaluation of Capacity-Spectrum Panagiatakos (1996), while a review of
Method of ATC-40 some recent methods can be found in
3.3.1 Summary of the Approximate Method Miranda and Ruiz-García (2003). The
The simplified inelastic analysis procedure Capacity Spectrum Method as documented
in ATC-40, a version of the in ATC-40 is based primarily on the work
Capacity-Spectrum Method (CSM), is of Freeman et al. (1975). In equivalent
based on equivalent linearization. The linear methods, the equivalent period is
basic assumption in equivalent linear computed from the initial period of
methods is that the maximum vibration of the nonlinear system and from
displacement of a nonlinear SDOF system the maximum displacement ductility ratio,
can be estimated from the maximum μ. Similarly, the equivalent damping ratio
displacement of a linear is computed as a function of damping ratio
in the nonlinear system and the
displacement ductility ratio. The main idealized bilinear system designated as β0
differences among the many equivalent in ATC-40 documentation. Table 3-1
linear methods that are available in the shows the variation of κ with respect to β0
literature stem primarily from the for different hysteretic behaviors types.
functions used to compute the equivalent The equivalent period in Equation 3-5 is
period and equivalent damping ratio. As based on a lateral stiffness of the
discussed in Section 2.4.2, the Capacity equivalent system that is equal to the
Spectrum Method according to ATC-40 secant stiffness at the maximum
uses the secant stiffness at maximum displacement. It only depends on the
displacement to compute the effective displacement ductility ratio and the Figure
period and relates effective damping to the 3-3 shows the variation of equivalent
area under the hysteresis curve (see Figure periods for different post-yield stiffness
2-13).These assumptions ratios for a wide range of displacement
result in an equivalent period, Teq, and ductility ratios. The equivalent period
equivalent damping ratio (referred to as becomes longer as the displacement
effective viscous damping,β eq, in ATC ductility ratio increases and as the
40) given by post-yield stiffness ratio decreases. Figures
(3-5) 3-4 and 3-5 present the variation of κ and
(3-6) effective damping value, βeq, with
where T0 is the initial period of vibration changes in the ductility ratio, respectively.
of the nonlinear system, α is the post-yield The calculations were done assuming
stiffness ratio and κ is an adjustment factor elastic perfectly plastic (EPP) behavior to
to approximately account for changes in represent a system that has full hysteretic
hysteretic behavior in reinforced concrete loops (i.e., a non-degrading system). It can
structures. ATC-40 proposes three be seen that for structures with type A
equivalent damping levels that change behavior (systems having full hysteretic
according to the hysteretic behavior loops), the κ value is 1.0 for displacement
of the system. Type A hysteretic behavior ductility ratios less than 1.3. For ductility
denotes structures with reasonably full ratios larger than 1.3, κ decreases up to a
hysteretic loops, similar to the EPP value of 0.77 at a displacement ductility
oscillator in Figure 3-1. The corresponding ratio of 3.4 and remains constant at 0.77
equivalent damping ratios take the for
maximum values. Type C hysteretic larger ductilities. Similarly, for structures
behavior represents severely degraded with type B hysteretic behavior, the value
hysteretic loops (e.g., SSD), resulting in of κ is constant and equal to 0.67 for
the displacement ductility ratios less than 1.6,
smallest equivalent damping ratios. Type decreases to 0.53 for ductility ratio of 3.4,
B hysteretic behavior is an intermediate and remains constant for larger ductilities.
hysteretic behavior between types A and C For structures with type C hysteretic
(e.g., SD). The value of κ decreases for behavior, the κ factor is equal to 0.33
degrading systems (hysteretic behavior regardless of the level of ductility demand.
types B and C). ATC-40 suggests an initial The equivalent damping ratio in the
elastic viscous damping ratio (first term on equivalent linear spectrum method
the right hand side of Equation 3-6) of documented in ATC-40 rapidly increases
0.05 (5%) for reinforced concrete once the structures yields and remains
buildings. The terms to the right of κ in constant for ductility ratios higher than
Equation 3-6 represent the equivalent 3.4. The maximum equivalent damping
hysteretic viscous damping for an ratios for hysteretic behavior types A, B,
and C are 0.40, 0.29 and 0.20, ground motion. Specifically, in Equations
respectively. According to Equations 3-5 3-5 and 3-6, μ must be known in order to
and 3-6, structures with hysteretic compute βeff and Teq. However, when
behaviors type B and C will reduced evaluating a structure, the maximum
hysteretic energy dissipation capacity displacement ductility ratio is not known.
produced Consequently, iteration is required in order
by narrower hysteretic loops. to estimate the maximum displacement.