Вы находитесь на странице: 1из 15

Engineering Structures 30 (2008) 37933807

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

An alternative pushover analysis procedure to estimate seismic displacement demands


Sun-Pil Kim a , Yahya C. Kurama b,
a b

Hyundai Development Institute of Construction Technology, South Korea Department of Civil Engineering and Geological Sciences, University of Notre Dame, Notre Dame, IN, 46556, USA

article

info

a b s t r a c t
An alternative pushover analysis procedure is proposed to estimate the peak seismic lateral displacement demands for building structures responding in the nonlinear range. As compared with other pushover analysis procedures, the main advantage of the proposed procedure is that the effects of higher modes on the lateral displacement demands are lumped into a single invariant lateral force distribution that is proportional to the total seismic masses at the floor and roof levels. The applicability and validity of the proposed procedure, which is referred to as the Mass Proportional Pushover (MPP) procedure, are critically evaluated through comparisons with multi-degree-of-freedom nonlinear dynamic timehistory analysis results for a set of benchmarked three-story, nine-story, and twenty-story steel moment resisting building frame structures. The estimated demands are also compared with results from a Modal Pushover Analysis (MPA) procedure. The comparisons demonstrate that the proposed Mass Proportional Pushover procedure provides, on average, better roof and floor lateral displacement demand estimates than the Modal Pushover Analysis procedure. The improvement from the proposed procedure is larger for the nine-story and twenty-story structures than the improvement for the three-story structure and is also larger for the Design Basis Earthquake (DBE) ground motion set than the Maximum Considered Earthquake (MCE) set. 2008 Elsevier Ltd. All rights reserved.

Article history: Received 25 July 2006 Received in revised form 24 May 2008 Accepted 4 July 2008 Available online 8 August 2008 Keywords: Modal pushover analysis Seismic displacement demands Seismic design and analysis

1. Introduction As performance-based considerations become more common in the seismic design and evaluation of building structures, the estimation of peak lateral displacement demands under specified seismic intensity levels (e.g., Design Basis Earthquake, Maximum Considered Earthquake) has gained utmost importance [1]. The most accurate analytical procedure to estimate the seismic displacement demands of a structure responding in the nonlinear range is to conduct nonlinear multi-degree-of-freedom (MDOF) dynamic time-history analyses. However, since this procedure requires the consideration of a large number of earthquakes and is relatively complicated and time-consuming, the use of static nonlinear pushover analysis procedures [1] is generally considered to be more suitable for the seismic design and evaluation of regular building structures in standard practice. Among the static pushover analysis procedures that have been previously developed to estimate the seismic displacement demands of building structures, adaptive pushover methods [2,3]

Corresponding author. E-mail address: ykurama@nd.edu (Y.C. Kurama).

aim to capture the changes that occur in the vibration properties of a structure, and the associated variations in the inertia forces, as the structure is displaced into the nonlinear range during an earthquake. For each step of an adaptive pushover analysis, the lateral force distribution is evaluated and adjusted as necessary based on the nonlinear behavior of the structure. This approach can provide good estimates of the nonlinear seismic displacement demands; however, it is relatively complicated for use in common structural engineering design practice. A second static approach, the Modal Pushover Analysis procedure proposed by Chopra and Goel [47], uses invariant modal lateral force distributions in accordance with linear-elastic theory. This method is in general simpler, and thus, more practical than the adaptive pushover method for use in seismic design. For structures with increased higher mode effects, the Modal Pushover Analysis procedure allows the use of multiple modes in the estimation. The primary limitations of this approach are: (1) as many static pushover analyses of the structure need to be conducted as the number of modes considered; (2) reversal in the pushover load versus displacement relationships of the structure under higher-mode lateral force distributions is possible [5], resulting in an unstable solution; and (3) a modal combination procedure is needed to estimate the total demands, often resulting in an overestimation of the peak demands. The Mass Proportional

0141-0296/$ see front matter 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.engstruct.2008.07.008

3794

S.-P. Kim, Y.C. Kurama / Engineering Structures 30 (2008) 37933807

Pushover procedure proposed in this paper offers a possible improvement to address these limitations. The main simplification and difference of the Mass Proportional Pushover procedure as compared to modal pushover analysis is that the effects of higher modes on the lateral displacement demands are lumped into a single invariant lateral force distribution that is proportional to the total seismic masses (or weights) assigned to the structure at the floor and roof levels. Thus, only one pushover analysis of the structure is required with no need to conduct a modal analysis or modal combination, even when higher mode effects are considerable. The applicability and validity of the new procedure are critically evaluated through comparisons with MDOF nonlinear dynamic time-history analysis results and with modal pushover analysis results using a set of benchmarked three-story, nine-story, and twenty-story steel moment resisting building frame structures. It is shown that especially for the nine-story and twenty-story structures under the DBE ground motion set, the Mass Proportional Pushover procedure results in considerably better roof and floor lateral displacement demand estimates than the Modal Pushover Analysis procedure while also requiring a relatively simpler process. The method is currently limited to the estimation of peak lateral displacement demands, and the estimation of other seismic demand quantities is out of its scope. Note also that the proposed procedure should not be used for structures that have significant plan and/or height irregularities. Furthermore, an evaluation of the procedure for wall structures is not yet available; and thus, the results presented below are limited to frame structures only. 2. Proposed pushover procedure This section describes the Mass Proportional Pushover procedure proposed in this paper and its relationship with the Modal Pushover Analysis procedure in Chopra and Goel [47]. 2.1. Uncoupled modal response history analysis for nonlinear structures The governing differential equation of motion for a nonlinear MDOF structure under horizontal earthquake ground acceleration, ug (t ), is as follows:

where,

n =

Ln Mn

{n }T [m]{1} {n }T [m]{n }

(4)

and {n } is the nth mode shape. Then, the governing differential equation of motion for the structure can be written as:
N

[m]{ } + [c ]{ } + {fs }({u}, sign{ }) = u u u


n =1

{sn } g (t ). u

(5)

Also, because any set of N independent vectors can be used as a basis for representing any other vector of order N, the modal expansion of the lateral displacement vector, {u}, has the form
N N

{ u} =
n =1

{ un } =
n=1

{n }qn

(6)

where, qn are the modal amplitudes. Finally, substituting Eq. (6) and its derivatives into Eq. (5), pre-multiplying by {n }T , and using the mass and damping orthogonality of the mode shapes result in the following governing equation for a SDOF system:

Dn + 2n n Dn +

Fsn Ln

= g (t ) u

(7)

where, Fsn = {n }T {fs (Dn , sign Dn )} is the modal internal resisting force, Dn is the modal displacement (with Dn and Dn representing the modal velocity and acceleration, respectively), and n and n are the modal damping and frequency, respectively. Eq. (7) can be solved either by conducting a nonlinear SDOF dynamic time-history analysis or by using a nonlinear SDOF displacement response spectrum [8]. More details and assumptions related to the above formulation can be found in Chopra [9]. Note that the coupling that occurs between the modal responses of the structure in the nonlinear range is neglected in this procedure.
2.2. Pushover lateral force distribution In order to solve Eq. (7), equivalent nonlinear SDOF representations of the structure need to be determined using pushover analysis to define the relationship between Fsn and Dn . According to linear-elastic uncoupled modal response history analysis theory, the only lateral force distribution that can produce displacements proportional to {n } is {sn }. Therefore, the Modal Pushover Analysis procedure uses {sn } as the force distribution for the pushover analysis in each mode [47]. In contrast with this approach, the proposed Mass Proportional Pushover procedure uses the total seismic mass (or weight) distribution assigned to the structure at the floor and roof levels, [m]g {1} = [w]{1} as the lateral force distribution. Furthermore, the linear-elastic floor/roof lateral displacement vector resulting from the application of [w]{1} on the structure is used as the mode shape of the structure. There are several reasons and advantages for the above choice of the lateral force distribution rather than {sn }: (1) According to classical modal analysis theory, it is impossible to superpose modal responses of nonlinear structures not only because of the coupling that exists between the governing differential equations for the N modes but also because of the changes that occur in the vibration properties of the structure. In other words, the choice for the use of {sn } as the modal lateral force distribution vector in the Modal Pushover Analysis procedure is not based on nonlinear theory (i.e., is not unique). Thus, the only way to show the accuracy of an assumed force distribution is by

[m]{ } + [c ]{ } + {fs } = [m]{1} g (t ) u u u


with

(1)

{fs } = {fs }({u}, sign{ }) u

(2)

where, [m] and [c ] are the mass and classical damping matrices of the structure, {fs } is the internal resisting force vector, and {u} is the floor/roof relative lateral displacement vector with { } and { } u u representing the first and second time derivatives (i.e., floor/roof velocity and acceleration vectors), respectively. Eq. (2) indicates that the internal resisting force vector depends not only on the lateral displacements of the structure, but also on the displacement history. Even though linear modal analysis is not valid for nonlinear structures, it is used in the formulation below to transform Eq. (1) into a set of single-degree-of-freedom (SDOF) equations of motion. According to modal analysis theory, the vertical distribution, [m]{1}, of the lateral force vector on the right hand side of Eq. (1) can be expanded as the summation of the modal lateral force distributions, {sn } as:
N N

[m]{1} =
n =1

{sn } =
n =1

n [m]{n }

(3)

S.-P. Kim, Y.C. Kurama / Engineering Structures 30 (2008) 37933807

3795

(a) Idealized MDOF pushover base shear force versus roof displacement relationship.

(b) SDOF pseudo-acceleration versus displacement relationship.

Fig. 1. Equivalent single-degree-of-freedom representation.

comparing the results from SDOF analyses with results from MDOF nonlinear dynamic time-history analyses. In the proposed Mass Proportional Pushover procedure, a single pushover analysis is conducted with the lateral force distribution given by the total seismic mass (or weight) vector, [m]g {1} = [w]{1}. Based on Eq. (3), this implies that the effects of higher modes on the lateral displacement demands are lumped into the first mode. Note that this is a reasonable approximation only when the displacements of the structure are primarily governed by the first mode; and thus, structures that develop weak/soft story mechanisms are outside the scope of the proposed procedure. Note also that different lateral force distributions may result in different pushover yield/failure mechanisms in the structure and that the proposed lateral force distribution may tend to exacerbate weak story mechanisms relative to a first mode approach. (2) In order to define a relationship between Fsn and Dn in each mode, the Modal Pushover Analysis procedure requires as many static pushover analyses of the structure as the number of modes considered. (3) As a possible limitation of the Modal Pushover Analysis procedure, reversal in the pushover load versus displacement relationships of the structure under higher-mode lateral force distributions is possible [5], resulting in an unstable solution. (4) If nonlinear SDOF displacement response spectra (instead of dynamic time-history analyses) are used to solve Eq. (7), a different response spectrum may be needed for use with each mode since nonlinear response spectra depend, among other factors, on the post-yield stiffness ratio of the structure [10], which would be different for each mode. (5) The Modal Pushover Analysis procedure requires a modal analysis of the structure to determine the mode shapes and other modal properties such as {sn } and n based on Eqs. (3) and (4). In contrast, the Mass Proportional Pushover procedure does not require a modal analysis since the linear-elastic floor/roof lateral displacement vector resulting from the application of [w]{1} on the structure is used as the mode shape. (6) Finally, the Modal Pushover Analysis procedure uses approximate rules to combine the peak modal responses, such as the square-root-sum-of-squares (SRSS) rule and the complete quadratic combination (CQC) rule. The Mass Proportional Pushover procedure does not require the combination of peak responses since Eq. (7) is solved only once using a single equivalent SDOF representation of the structure as described below. 2.3. Equivalent single-degree-of-freedom representation In both the Modal Pushover Analysis procedure and the Mass Proportional Pushover procedure, the base shear force versus roof

displacement relationship obtained from the pushover analysis of the structure is converted into an equivalent SDOF pseudoacceleration versus displacement relationship as shown using the idealized bi-linear relationships in Fig. 1. In the case of the Modal Pushover Analysis procedure, the base shear force, Vbn from each modal pushover analysis is converted into a modal equivalent SDOF pseudo-acceleration, An using An = Vbn
Mn

(8)

where, Mn = n Ln is the effective modal mass. For the conversion from modal pushover roof displacement, urn to modal SDOF displacement, Dn , the Modal Pushover Analysis procedure uses

Dn =

urn

(9)

where, the modal participation factor n is calculated from Eq. (4) based on the linear-elastic mode shape {n } normalized with respect to the roof. In the Mass Proportional Pushover procedure, the conversion from the pushover base shear force, Vb to SDOF acceleration, A is accomplished by using the total mass, M of the structure as: A= Vb M

(10)

Note that the use of the total mass, M in Eq. (10) is done so as to include the representation of the entire seismic mass of the structure in the seismic demand estimates (i.e., it lumps all effective modal masses into the first mode). This selection is consistent with the use of the total seismic mass (or weight) distribution, [m]g {1} = [w]{1} as the lateral force distribution in the proposed procedure. The SDOF displacement, D is given by D= ur

(11)

with the participation factor calculated as:

{ue }T [m]{1} {ue }T [m]{ue }

(12)

where, {ue } is the floor/roof lateral displacement vector (normalized with respect to the roof displacement) obtained from the linear-elastic response range of the pushover analysis under the [m]g {1} = [w]{1} lateral force distribution. A summary of the Mass Proportional Pushover procedure proposed in this paper is as follows:

3796

S.-P. Kim, Y.C. Kurama / Engineering Structures 30 (2008) 37933807

1. Conduct a single nonlinear pushover analysis of the structure under the lateral force distribution given by [m]g {1} = [w]{1}. 2. Determine the MDOF base shear force versus roof displacement (Vb ur ) relationship from the pushover analysis of the structure. 3. Determine the equivalent SDOF pseudo-acceleration versus displacement (AD) relationship using Eqs. (10) and (11). 4. Find the maximum SDOF displacement, Dmax by solving Eq. (7) with Fs /L = A. 5. Calculate the maximum MDOF roof and floor lateral displacements of the structure as:

relationships obtained from the OpenSees models of the SAC frame structures under the [w]{1} lateral force distribution are shown in Fig. 3. For dynamic analysis, the building mass assigned to each frame was assumed to be lumped at the horizontal degrees of freedom of the beamcolumn joint nodes. The damping matrix was defined using Rayleigh damping [9] with a damping ratio of 3% for periods of T1 and 0.1T1 . The average acceleration method, which is one of two special cases of Newmarks method [9], was used to conduct the nonlinear time-history analyses. 3.3. Equivalent SDOF models for prototype structures

{umax } = Dmax {ue }.


3. Numerical approach

(13)

This section provides an overview of the numerical approach used to critically evaluate the proposed Mass Proportional Pushover procedure as follows: (1) prototype structures; (2) analytical modeling; (3) nonlinear SDOF models for prototype structures; and (4) ground motion records. 3.1. Prototype structures A set of benchmarked three-story, nine-story, and twenty-story prototype steel moment resisting building frame structures is used in this investigation. These buildings were designed by Brandow & Johnston Associates, Los Angeles, California as part of the SAC Phase II Steel Project [11]. Although the SAC buildings were not actually constructed, they were designed in compliance with the gravity, wind, and seismic building requirements of the 1994 Uniform Building Code [12] and represent typical low-rise, medium-rise, and high-rise buildings on stiff soil sites (Soil Type S2 in the 1994 Uniform Building Code) in Los Angeles, California. As shown in Fig. 2, the prototype structures have no significant plan or height irregularities. The moment resisting frames were designed only along the perimeter of the buildings, with the interior system consisting of gravity load resisting members. The design of the moment resisting frames in all three structures was governed by seismic loads. The moment resisting frames in the northsouth (NS) direction of the buildings are used in this paper. Table 1 provides the total heights, H for the SAC buildings as well as the building seismic mass assigned to each frame, M (assumed to be equally divided between the two moment resisting frames in each orthogonal direction of the building), the distribution of the seismic mass over the height of each frame, and the periods for the first three linear-elastic modes, T1 , T2 , and T3 (determined using the analytical model described below). More details on the SAC prototype buildings can be found elsewhere [11]. 3.2. Analytical modeling Two-dimensional nonlinear MDOF analytical models for the moment resisting frames in the NS direction of the prototype SAC building structures were developed using the OpenSees program [13]. To reflect flexural nonlinearity, the column and beam members were modeled using nonlinear beamcolumn elements that consider the spread of plasticity along the member. The beam and column cross sections were assumed to have bi-linear stiffness degrading [14] moment versus curvature relationships with a post-yield stiffness equal to 3% of the linearelastic stiffness. The columns of the three-story structure were assumed to be fixed at the base. For the nine-story and twentystory structures, basement levels were used with assumed pinned conditions for the column bases as shown in Fig. 2. Typical hysteretic base shear force versus roof displacement (Vb ur )

The bi-linear dashed-line plots in Fig. 4 show the equivalent SDOF pseudo-acceleration versus displacement (AD) relationships for the prototype SAC frame structures as obtained from the proposed Mass Proportional Pushover procedure. To obtain these relationships, first, smooth pushover base shear force versus roof displacement (Vb ur ) relationships were determined using the MDOF OpenSees models of the structures. Then, the MDOF Vb ur relationships were transformed into SDOF AD relationships as described previously. The smooth SDOF AD relationships of the SAC structures as obtained using the proposed procedure are shown with the smooth solid lines in Fig. 4. The idealized bi-linear AD relationships for the structures in Fig. 4 (i.e., the bi-linear slopes, 2 and 2 , and the yield point, Dy and Ay ) were determined by assuming: (1) the initial slope, 2 of the bi-linear relationship is the same as the initial slope of the smooth relationship; (2) the maximum displacement is approximately equal to five times the yield displacement, Dy ; and (3) the areas underneath the smooth and bi-linear relationships up to the maximum displacement are the same. The BISPEC program [15] was used to conduct nonlinear dynamic time-history analyses of the idealized bi-linear equivalent SDOF models for the SAC buildings. A stiffness degrading forcedisplacement relationship [14] was assumed for the hysteretic behavior of the SDOF models under cyclic loading. To facilitate comparisons with the nonlinear MDOF dynamic analyses described above, the SDOF system damping ratio in the Mass Proportional Pushover procedure was assumed to be equal to 3%. Similarly, the modal damping ratios used in the Modal Pushover Analysis procedure were determined from Rayleigh damping with a damping ratio of 3% for periods of T1 and 0.1T1 . While the damping ratios and the damping models used in the analyses of the structures can certainly be debatable, the use of the same damping ratios between the MDOF analyses, the Mass Proportional Pushover procedure, and the Modal Pushover Analysis procedure ensures that the differences in the lateral displacement demands from these three approaches are not caused by differences in damping. 3.4. Ground motion records A total of 40 far-fault ground motion records (LA01LA40) are used to evaluate the effectiveness of the Mass Proportional Pushover procedure in estimating the MDOF lateral displacement demands. The design-level ground motion ensemble (twenty recorded motions, LA01LA20, as shown in Table 2) corresponds to a 10% probability of exceedance in 50 years representing the Design Basis Earthquake (DBE) in IBC-2003 [16]. Similarly, the survival-level ensemble (ten recorded and ten generated motions, LA21LA40, as shown in Table 3) corresponds to a 2% probability of exceedance in 50 years, representing the Maximum Considered Earthquake (MCE) in IBC-2003. These ground motion records were compiled by the SAC Phase II Steel Project [17] for a site in

S.-P. Kim, Y.C. Kurama / Engineering Structures 30 (2008) 37933807

3797

(a) Three-story.

(b) Nine-story.

(c) Twenty-story. Fig. 2. SAC model frame structures.

3798 Table 1 Properties of SAC prototype frame structures Structure Total height, H (m)

S.-P. Kim, Y.C. Kurama / Engineering Structures 30 (2008) 37933807

Seismic mass, M per frame (kg) 1.48 10

Floor/roof seismic mass per frame (kg) Period (s) First mode, T1 Second mode, T2 0.33 0.84 Third mode, T3 0.17 0.49 0.48 10 0.52 106 0.51 106 0.49 106 0.54 106 0.28 106 0.29 106
6

Three-story Nine-story

11.88 37.17

4.50 106

Twenty-story

80.73

5.55 106

(floors 23) (roof) (floor 2) (floors 39) (roof) (floors 220) (roof)

1.01 2.26

3.96

1.37

0.79

(a) Three-story.

(b) Nine-story.

(c) Twenty-story. Fig. 3. Base shear force versus roof displacement behaviors of SAC structures.

Los Angeles with a stiff soil profile (Site Class D in IBC-2003), similar to the site conditions used in the design of the SAC prototype structures. For each ground motion, two horizontal components, rotated 45 degrees away from the fault-normal and fault-parallel orientations, were provided by the SAC steel project. The ground motions were scaled based on target linear-elastic smooth acceleration response spectra as described in Somerville et al. [17]. The probabilistic ground motion spectra published by the United States Geological Survey [18,19], modified to represent Site Class D, were used as the target spectra. To preserve the variability in the characteristics of the individual ground motions, the shapes of the acceleration response spectra of the records were not modified in the scaling procedure. Instead, for each ground motion, a single scaling factor was found that minimized the weighted sum of the squared error between the average 5%-damped linear-elastic acceleration

response spectra of the two horizontal components and the corresponding target response spectrum in the period range of 0.34 s. This scale factor was then applied to both components of the ground motion, thus retaining the ratios between the components at all periods. The weights used in the determination of the ground motion scaling factors were 0.1, 0.3, 0.3, and 0.3 for periods of 0.3, 1, 2, and 4 s, respectively. More details on the SAC ground motion records can be found in Somerville et al. [17]. Tables 2 and 3 provide the following information for the SAC LA01LA40 ground motion records: (1) record site location; (2) earthquake magnitude; (3) epicentral distance; (4) SAC scale factor; (5) peak ground acceleration, PGA; and (6) maximum incremental velocity, MIV. The MIV of a ground motion is equal to the maximum area under the acceleration time-history of the record between two successive zero-acceleration crossings. Previous research has shown that a strong correlation exists

S.-P. Kim, Y.C. Kurama / Engineering Structures 30 (2008) 37933807

3799

(a) Three-story.

(b) Nine-story.

(c) Twenty-story. Fig. 4. SDOF pseudo-acceleration versus displacement relationships for SAC structures. Table 2 SAC DBE-level ground motions LA01LA20 Record name LA01 LA02 LA03 LA04 LA05 LA06 LA07 LA08 LA09 LA10 LA11 LA12 LA13 LA14 LA15 LA16 LA17 LA18 LA19 LA20 Site location Imperial Valley, 1940, EI Centro Imperial Valley, 1940, EI Centro Imperial Valley, 1979, Array #05 Imperial Valley, 1979, Array #05 Imperial Valley, 1979, Array #06 Imperial Valley, 1979, Array #06 Landers, 1992, Barstow Landers, 1992, Barstow Landers, 1992, Yermo Landers, 1992, Yermo Loma Prieta, 1989, Gilroy Loma Prieta, 1989, Gilroy Northridge, 1994, Newhall Northridge, 1994, Newhall Northridge, 1994, Rinaldi RS Northridge, 1994, Rinaldi RS Northridge, 1994, Sylmar Northridge, 1994, Sylmar North Palm Springs, 1986 North Palm Springs, 1986 EQ magnitude 6.9 6.9 6.5 6.5 6.5 6.5 7.3 7.3 7.3 7.3 7.0 7.0 6.7 6.7 6.7 6.7 6.7 6.7 6.0 6.0 Distance (km) 10 10 4.1 4.1 1.2 1.2 36 36 25 25 12 12 6.7 6.7 7.5 7.5 6.4 6.4 6.7 6.7 Scale factor 2.01 2.01 1.01 1.01 0.84 0.84 3.20 3.20 2.17 2.17 1.79 1.79 1.03 1.03 0.79 0.79 0.99 0.99 2.97 2.97 PGA (g) 0.46 0.68 0.39 0.49 0.30 0.23 0.42 0.42 0.52 0.36 0.66 0.97 0.68 0.66 0.53 0.58 0.57 0.82 1.02 0.99 MIV (cm/s) 88.9 81.0 103 74.6 106 81.6 59.1 71.5 135 76.4 79.4 86.8 84.6 132 124 165 102 139 48.6 122

between the MIV and the peak lateral displacement demands resulting from a ground motion [20]. Note that reliable nonlinear dynamic time-history analysis results depend on the use of realistic ground motion records with

characteristics that are appropriate for the site soil conditions, site seismicities, and seismic demand levels considered. According to Somerville et al. [17], the SAC ground motion records provide a sample of the variability in earthquake characteristics through a

3800 Table 3 SAC MCE-level ground motions LA21LA40 Record name LA21 LA22 LA23 LA24 LA25 LA26 LA27 LA28 LA29 LA30 LA31 LA32 LA33 LA34 LA35 LA36 LA37 LA38 LA39 LA40 Site location Kobe, 1995 Kobe, 1995 Loma Prieta, 1989 Loma Prieta, 1989 Northridge, 1994 Northridge, 1994 Northridge, 1994 Northridge, 1994 Tabas, 1974 Tabas, 1974 Elysian Park (simulated) Elysian Park (simulated) Elysian Park (simulated) Elysian Park (simulated) Elysian Park (simulated) Elysian Park (simulated) Palos Verdes (simulated) Palos Verdes (simulated) Palos Verdes (simulated) Palos Verdes (simulated)

S.-P. Kim, Y.C. Kurama / Engineering Structures 30 (2008) 37933807

EQ magnitude 6.9 6.9 7.0 7.0 6.7 6.7 6.7 6.7 7.4 7.4 7.1 7.1 7.1 7.1 7.1 7.1 7.1 7.1 7.1 7.1

Distance (km) 3.4 3.4 3.5 3.5 7.5 7.5 6.4 6.4 1.2 1.2 17 17 11 11 11 11 1.5 1.5 1.5 1.5

Scale factor 1.15 1.15 0.82 0.82 1.29 1.29 1.61 1.61 1.08 1.08 1.43 1.43 0.97 0.97 1.10 1.10 0.90 0.90 0.88 0.88

PGA (g) 1.28 0.92 0.42 0.47 0.87 0.94 0.93 1.33 0.81 0.99 1.30 1.19 0.78 0.68 0.99 1.10 0.71 0.78 0.50 0.63

MIV (cm/s) 274 242 86.9 211 202 269 166 226 92.2 128 208 260 188 161 343 329 263 302 117 279

set of time histories that are realistic not only in their average properties but also in their individual properties. Nevertheless, it should be stated that the findings and conclusions presented in this paper may be limited to the site and seismic characteristics represented by the SAC ground motions. 4. Evaluation of demand estimates This section critically evaluates the proposed Mass Proportional Pushover procedure to estimate the seismic lateral displacement demands for nonlinear building structures. Demand estimates are obtained for the prototype SAC frame structures using the proposed procedure and are compared with estimates from MDOF nonlinear dynamic time-history analyses as well as from the Modal Pushover Analysis procedure described in Chopra and Goel [47]. The demand estimates from the Mass Proportional Pushover procedure were obtained by conducting nonlinear dynamic timehistory analyses of the idealized equivalent bi-linear SDOF models in Fig. 4 to solve Eq. (7) under the DBE-level and MCE-level ground motion records in Tables 2 and 3, respectively. MDOF models and equivalent bi-linear modal SDOF models (as obtained from the Modal Pushover Analysis procedure) for the SAC structures were also subjected to the same ground motion records. As described previously, the BISPEC program was used to conduct the SDOF dynamic analyses assuming bi-linear stiffness degrading hysteretic characteristics for the equivalent models. The MDOF analyses were conducted using the OpenSees analytical models (see Figs. 2 and 3). The following seismic demand estimates determined using the Mass Proportional Pushover procedure are evaluated below: (1) peak roof lateral displacements; (2) peak floor lateral displacements; and (3) peak inter-story drift angles. 4.1. Peak roof lateral displacement demand estimates The peak roof lateral displacement demands for the SAC frame structures were estimated using Eq. (13) with the peak SDOF displacement demands, Dmax , determined from the equivalent bilinear SDOF dynamic time-history analyses described above. The errors between the estimated peak roof displacement demands from the proposed Mass Proportional Pushover procedure and the exact demands determined from the MDOF nonlinear dynamic time-history analyses of the SAC structures are shown by the markers in Figs. 5 and 6. The error values were calculated as: error in roof displacement =

(ur ,max )estimated (ur ,max )exact (ur ,max )exact

(14)

where, (ur ,max )estimated is the displacement demand estimated from the pushover procedure and (ur ,max )exact is the displacement demand determined from the MDOF dynamic analysis. The mean and standard deviation values for the calculated errors in the peak roof displacement demands are provided in Tables 4 and 5. The positive mean errors indicate that the peak displacement demands are, on average, conservatively over-estimated by the approximate procedure. To provide a benchmark for the evaluation of the Mass Proportional Pushover procedure, the peak roof displacement demands for the SAC structures were also estimated using the Modal Pushover Analysis procedure. Peak demand estimates considering the first (i.e., fundamental) mode, first two modes, and first three modes (only for the twenty-story structure) were determined by combining the modal peak roof displacement demands using the SRSS rule. The resulting errors with respect to the exact displacement demands from the MDOF analyses of the structures are given in Tables 4 and 5 and shown using the markers in Figs. 5 and 6. As shown in Tables 4 and 5, on average, both approximate methods provide reasonable estimates for the peak roof displacement demands from the MDOF analyses. The Modal Pushover Analysis (MPA) results in Figs. 5 and 6 are for demand estimates obtained using the first mode only. The mean errors increase as more modes are included in the Modal Pushover Analysis procedure (see Tables 4 and 5). This is because, the first-mode estimates are, on average, larger than the exact solution and the consideration of higher modes increases this over-estimation based on the SRSS combination of the modal displacements. Note that this situation does not occur in the proposed procedure since the effects of higher modes are lumped into a single mode, and thus, there is no need for modal combination. Note also that while the proposed lateral force vector contains the higher modes, the degree to which higher modes may be important for the demand estimation varies with the number of stories and the excitation. This is discussed further below. It can be seen from the analysis results that the Mass Proportional Pushover procedure provides, on average, smaller peak roof displacement demand estimates than the estimates from the Modal Pushover Analysis procedure considering the first mode only. These reduced demand estimates are closer to the mean MDOF demands while still providing a conservative solution. One possible explanation for the reduced demand estimates from the proposed procedure is the use of the linear-elastic

S.-P. Kim, Y.C. Kurama / Engineering Structures 30 (2008) 37933807

3801

(a) Three-story.

(b) Nine-story.

(c) Twenty-story. Fig. 5. Errors in peak roof displacement demand estimates under LA01LA20.

Table 4 Percent errors in peak roof displacement demands under LA01LA20 Structure Proposed procedure Mean (%) Three-story Nine-story Twenty-story 4.5 3.1 8.5 St. dev. (%) 22.0 14.8 16.1 Modal Pushover Analysis (MPA) procedure No. of modes One mode Two modes One mode Two modes One mode Two modes Three modes Mean (%) 6.3 6.1 4.6 6.8 12.6 16.6 17.1 St. dev. (%) 23.0 22.0 16.2 15.4 18.6 17.5 17.4

Table 5 Percent errors in peak roof displacement demands under LA21LA40 Structure Proposed procedure Mean (%) Three-story Nine-story Twenty-story 15.9 6.6 10.6 St. dev. (%) 23.1 15.8 10.2 Modal Pushover Analysis (MPA) procedure No. of modes One mode Two modes One mode Two modes One mode Two modes Three modes Mean (%) 16.0 17.8 7.7 9.8 15.2 19.6 20.1 St. dev. (%) 23.1 23.1 16.5 16.6 12.4 12.2 12.5

3802

S.-P. Kim, Y.C. Kurama / Engineering Structures 30 (2008) 37933807

(a) Three-story.

(b) Nine-story.

(c) Twenty-story. Fig. 6. Errors in peak roof displacement demand estimates under LA21LA40.

lateral displacement vector from the application of [w]{1} on the structure as the mode shape of the structure. Tables 4 and 5 show that the Mass Proportional Pushover procedure provides better estimates for the mean peak roof displacement demands than the estimates from the Modal Pushover Analysis procedure for all of the cases considered in the investigation (i.e., considering the three-story, nine-story, and twenty-story structures; the DBE and MCE ground motion ensembles; and the MPA results including one, two, and three modes in the solution). It can be observed that the improvement provided by the proposed procedure is larger for the nine-story and twenty-story structures than the improvement for the three-story structure and is also larger for the DBE ground motion set than the MCE set. Based on these results, it is concluded that the Mass Proportional Pushover procedure provides better estimates for the MDOF peak roof displacement demands than the Modal Pushover Analysis procedure. These improved estimates, combined with relative ease of use, make the proposed procedure well-suited for the seismic design of regular building frame structures in practice. Note that a large amount of scatter is observed in Figs. 5 and 6 indicating a large variation in the errors for the peak roof displacement demand estimates from the different ground motion records used in the investigation (the error in the estimated peak roof displacement demand is more than 50% under some of the ground motions). For both the Mass Proportional Pushover procedure and the Modal Pushover Analysis procedure, the errors in the demand estimates vary significantly with the number of

stories (i.e., the three SAC structures) and the ground motion records. No significant correlation is observed between the errors and the maximum incremental velocity of the ground motion records; however, the mean errors under the MCE ground motion set are in general larger than the errors under the DBE ground motion set. The increased mean errors under the MCE set is possibly because of the increased nonlinearity in the structural response, and the corresponding increase in the mode shape variation and coupling between the modes, which are ignored in both approximate procedures. 4.2. Peak floor lateral displacement demand estimates The peak floor lateral displacement demands for a MDOF system can be estimated by multiplying the estimated peak roof displacement demand with the roof-normalized mode shape. In the case of the Mass Proportional Pushover procedure, this process results in Eq. (13), where ue (i.e., the floor/roof lateral displacement vector obtained from the linear-elastic response range of the pushover analysis under [m]g {1} = [w]{1}, normalized with respect to the roof displacement) is used as the normalized mode shape. Note that a different peak floor displacement vector would result from each ground motion record using this procedure since the peak roof displacement is different from each ground motion. For design purposes, a more reasonable approach would be to estimate a single peak floor displacement vector for the

S.-P. Kim, Y.C. Kurama / Engineering Structures 30 (2008) 37933807

3803

(a) Three-story.

(b) Nine-story.

(c) Twenty-story. Fig. 7. Mean peak floor displacement demand estimates under LA01LA20.

Table 6 Percent errors in mean peak floor displacement demands Structure Proposed procedure LA01LA20 (%) Three-story Nine-story Twenty-story 2.7 6.1 8.5 LA21LA40 (%) 22.9 4.6 8.3 Modal Pushover Analysis (MPA) procedure No. of modes One mode Two modes One mode Two modes One mode Two modes Three modes LA01LA20 (%) 4.2 4.4 15.5 17.1 17.3 20.8 21.1 LA21LA40 (%) 18.8 19.0 12.0 12.7 12.8 14.9 15.1

structure based on the estimated mean peak roof displacement demand from the entire ground motion ensemble. The resulting mean peak floor displacement demands for the SAC buildings, ui,max , estimated using the Mass Proportional Pushover procedure ( markers) are compared in Figs. 7 and 8 with the exact mean peak floor displacement demands determined from the MDOF nonlinear dynamic time-history analyses of the structures ( markers). The errors between the estimated mean peak floor displacement demands, ( i,max )estimated , and the exact mean demands, u ( i,max )exact , for the SAC buildings are given in Table 6. The error u values were calculated as:

error in floor displacements


K +1 i =2 K +1 i=2

ABS ( i,max )estimated ( i,max )exact u u (15)

( i,max )exact u

where, K = 3, 9, and 20 for the three-story, nine-story, and twenty-story structures, respectively, i = 2 represents the second floor level (see Fig. 2), and i = K + 1 represents the roof level. To provide a benchmark for the evaluation of the Mass Proportional Pushover procedure, the mean peak floor displacement demands for the SAC structures were also estimated using the

3804

S.-P. Kim, Y.C. Kurama / Engineering Structures 30 (2008) 37933807

(a) Three-story.

(b) Nine-story.

(c) Twenty-story. Fig. 8. Mean peak floor displacement demand estimates under LA21LA40.

Modal Pushover Analysis procedure. Mean peak demand estimates considering the first (i.e., fundamental) mode, first two modes, and first three modes (only for the twenty-story structure) were determined by combining modal mean peak floor displacement demands using the SRSS rule. The modal mean peak floor displacement demands were determined by multiplying each normalized mode shape with the corresponding estimated modal mean peak roof displacement demand. Following Chopra and Goel [6], each mode shape was determined as the floor/roof lateral displacement vector obtained from the pushover analysis of the structure under {sn } when the roof displacement reached the estimated modal mean peak roof displacement demand from the ground motion ensemble. This process was repeated for as many times as the number of modes considered. The resulting mean peak floor displacement demand estimates from the Modal Pushover Analysis procedure are shown using the markers in Figs. 7 and 8, with the errors between the estimated and exact mean demands [calculated using Eq. (15)] given in Table 6. Looking at the results, it is concluded that the Mass Proportional Pushover procedure provides, on average, better peak floor displacement demand estimates for the SAC buildings than the Modal Pushover Analysis procedure, with the exception of the three-story building under the MCE-level LA21LA40 ground motion set. The use of nonlinear mode shapes following Chopra and Goel [6] may have played a role in the increased errors in the mean peak floor displacement demand estimates from the Modal Pushover Analysis procedure; however, this was not

investigated. Note that the Modal Pushover Analysis (MPA) results in Figs. 7 and 8 are for demand estimates obtained using the first mode only. Similar to the peak roof displacement demands described previously, the errors in the estimated mean peak floor displacement demands from the Modal Pushover Analysis procedure tend to increase as more modes are included. 4.3. Peak story drift angle demand estimates The story (i.e., inter-story) drift angle is defined as the relative lateral displacement between two adjacent floor/roof levels divided by the story height. In the case of the proposed Mass Proportional Pushover procedure, the peak story drift angle demands are calculated from the peak floor/roof lateral displacement demands described previously. Once again, the estimated mean peak roof displacement demand from the entire ground motion ensemble is used to determine the mean peak story drift angle demands for each structure. The resulting mean peak story drift angle demands for the SAC buildings, ui,max , estimated using the Mass Proportional Pushover procedure ( markers) are compared in Figs. 9 and 10 with the exact mean peak story drift angle demands determined from the MDOF nonlinear dynamic time-history analyses of the structures ( markers). The errors between the estimated mean peak story drift angle demands, ( ui,max )estimated , and the exact mean demands,

S.-P. Kim, Y.C. Kurama / Engineering Structures 30 (2008) 37933807

3805

(a) Three-story.

(b) Nine-story.

(c) Twenty-story. Fig. 9. Mean peak story drift angle demand estimates under LA01LA20.

( ui,max )exact , for the SAC buildings are given in Tables 7 and 8.
The area errors investigate the total error in the story drift angle demands considering all of the stories in the structure, and are calculated as: area error in story drift angle
K +1 i=2 K +1 i =2

ABS ( ui,max )estimated ( ui,max )exact (16)

( ui,max )exact

where, K = 3, 9, and 20 for the three-story, nine-story, and twenty-story structures, respectively, i = 2 represents the second floor level, and i = K + 1 represents the roof level. The peak errors investigate the error in the largest mean peak story drift angle demand occurring in any story over the height of the structure, and are calculated as: peak error in story drift angle largest( ui,max )estimated largest( ui,max )exact largest( ui,max )exact

(17)

Note that while the peak error can be positive or negative (indicating an over-estimation or an under-estimation in the results, respectively), the area error is always positive. To provide a benchmark for the evaluation of the proposed Mass Proportional Pushover procedure, the mean peak story drift angle

demands for the SAC structures were also estimated using the Modal Pushover Analysis procedure. Mean peak demand estimates considering the first (i.e., fundamental) mode, first two modes, and first three modes (only for the twenty-story structure) were determined by combining modal mean peak story drift angle demands using the SRSS rule. The resulting mean peak story drift angle demand estimates from the Modal Pushover Analysis procedure are shown using the markers in Figs. 9 and 10, with the errors between the estimated and exact mean demands [calculated using Eqs. (16) and (17)] given in Tables 7 and 8. Note that, different from Figs. 58, the Modal Pushover Analysis (MPA) results in Figs. 9 and 10 are for demand estimates obtained using the first two modes for the three-story and nine-story structures and the first three modes for the twenty-story structure since the consideration of higher modes in the Modal Pushover Analysis procedure can result in better estimates for the peak story drift angle demands (see Tables 7 and 8). This is because, the story drift angle demands are sensitive to the derivative (i.e., slope) of the mode shape vector(s); and thus, the consideration of a larger number of mode shapes can improve the estimation. It is observed from the peak errors in Tables 7 and 8 that the Mass Proportional Pushover procedure underestimates the largest mean peak story drift angle demands, except for the threestory structure under the MCE-level ground motion ensemble. In general, the errors in the mean peak story drift angle demand estimates are quite large, ranging between 14.1% and 22.7% for

3806

S.-P. Kim, Y.C. Kurama / Engineering Structures 30 (2008) 37933807

(a) Three-story.

(b) Nine-story.

(c) Twenty-story. Fig. 10. Mean peak story drift angle demand estimates under LA21LA40.

Table 7 Percent errors in mean peak story drift angle demands under LA01LA20 Structure Proposed procedure Area error (%) Three-story Nine-story Twenty-story 19.4 16.1 22.7 Peak error (%) Modal Pushover Analysis (MPA) procedure No. of modes One mode Two modes One mode Two modes One mode Two modes Three modes Area error (%) 5.7 3.9 26.8 17.3 24.0 11.3 10.3 Peak error (%)

7.2 13.8 14.1

6.7 4.1
5.3 6.2 9.8 18.3 19.0

Table 8 Percent errors in mean peak story drift angle demands under LA21LA40 Structure Proposed procedure Area error (%) Three-story Nine-story Twenty-story 21.1 14.1 18.6 Peak error (%) 11.8 Modal Pushover Analysis (MPA) procedure No. of modes One mode Two modes One mode Two modes One mode Two modes Three modes Area error (%) 13.5 14.4 18.3 10.8 14.4 12.1 14.0 Peak error (%) 8.3 8.3 6.2 6.6

13.2 25.3

2.8
2.9 3.3

S.-P. Kim, Y.C. Kurama / Engineering Structures 30 (2008) 37933807

3807

the area errors and between 25.3% and +11.8% for the peak errors. Furthermore, the estimates from the proposed procedure are worse than those provided by the Modal Pushover Analysis procedure. The poor performance of the proposed procedure to estimate the peak story drift angle demands is because of the use of a single mode shape to represent the displaced shape of the structure during an earthquake and the inability of this mode shape to capture the changes in the lateral displacements between adjacent floor and roof levels. 5. Conclusions This paper proposes an alternative pushover analysis procedure, referred to as the Mass Proportional Pushover (MPP) procedure, to estimate the peak seismic lateral displacement demands for building structures responding in the nonlinear range. The main advantage of the Mass Proportional Pushover procedure over other approximate procedures (e.g., the Modal Pushover Analysis procedure) is the use of a single pushover analysis for the structure with no need to conduct a modal analysis to capture the effects of higher modes on the lateral displacement demands. Observations made through comparisons between the estimated displacement demands from the Mass Proportional Pushover procedure, the Modal Pushover Analysis procedure, and the exact demands from multi-degree-of-freedom nonlinear dynamic time-history analysis results are given below. Note that these observations are based on the mean seismic demands from a limited number of analyses considering three different steel building moment frame structures and two ground motion ensembles, and thus, generalizations drawn from the investigation should be used with caution. (1) Both the Mass Proportional Pushover procedure and the Modal Pushover Analysis procedure provide reasonable estimates for the mean peak roof and floor lateral displacement demands from the MDOF analyses. (2) The Mass Proportional Pushover procedure provides better estimates for the mean peak roof and floor displacement demands than the estimates from the Modal Pushover Analysis procedure. (3) The improvement from the Mass Proportional Pushover procedure is larger for the nine-story and twenty-story structures than the improvement for the three-story structure and is also larger for the Design Basis Earthquake (DBE) ground motion set than the Maximum Considered Earthquake (MCE) set. (4) The improved lateral displacement demand estimates, combined with relative ease of use, make the Mass Proportional Pushover procedure well-suited for the seismic design of regular building frame structures in practice. (5) As a result of modal combination, the mean errors in the peak roof and floor displacement estimates from the Modal Pushover Analysis procedure increase as more modes are included in the solution. This situation does not occur in the Mass Proportional Pushover procedure since the effects of higher modes are lumped into a single mode, and thus, there is no need for modal combination. (6) There is a large variation in the errors for the peak lateral displacement demand estimates from the different ground motion records used in the investigation. No significant correlation is observed between the errors and the intensity of the ground motions; however, the mean errors under the MCE ground motion set tend to be larger than the errors under the DBE ground motion set. The increased mean errors under the MCE set is possibly because of the increased nonlinearity in the structural response, and the corresponding increase in the mode shape variation and coupling between the modes, which are ignored in both approximate procedures.

(7) The Mass Proportional Pushover procedure provides relatively poor estimates for the peak inter-story drift angle demands, and the peak inter-story drift angle estimates from the Modal Pushover Analysis procedure are on average closer to the MDOF analysis results. The poor performance of the Mass Proportional Pushover procedure to estimate the peak inter-story drift angle demands is because of the inability of the selected mode shape to capture the changes in the lateral displacements of the structure between adjacent floor and roof levels. Acknowledgements This work was supported by the Korea Research Foundation (KRF-2005-214-D00168) with additional funding provided by the Department of Civil Engineering and Geological Sciences at the University of Notre Dame, Notre Dame, Indiana, USA. Their support is greatly appreciated. The opinions, findings, and conclusions presented in the paper are those of the authors and do not necessarily reflect the views of the organizations acknowledged above. References
[1] American Society of Civil Engineers (ASCE). Seismic rehabilitation of existing buildings. ASCE Standard No. ASCE/SEI 41-06. 2007. [2] Antoniou S, Pinho R. Development and verification of a displacement-based adaptive pushover procedure. J Earthq Eng 2004;8(5):64361. [3] Papanikolaou VK, Elnashai AS, Pareja JF. Evaluation of conventional and adaptive pushover analysis II: Comparative results. J Earthq Eng 2006;10(1): 12751. [4] Chopra AK, Goel R. A modal pushover analysis procedure to estimate seismic demands for buildings: Summary and evaluation. In: Fifth national conference on earthquake engineering; 2003. [5] Chopra AK, Goel R. Role of higher-mode pushover analyses in seismic analysis of buildings. Earthq Spectra 2005;21(4):102741. [6] Chopra AK, Goel R. A modal pushover analysis procedure for estimating seismic demands for buildings. Earthq Eng Struct Dyn 2002;31(3):56182. [7] Chopra AK, Goel R. A modal pushover analysis procedure to estimate seismic demands for unsymmetric-plan buildings. Earthq Eng Struct Dyn 2004;33(8): 90327. [8] Chopra AK, Goel R. Capacity-demand-diagram methods for estimating seismic deformation of inelastic structures: SDF systems. Report no. PEER-1999/02. Berkeley (CA): Pacific Earthquake Engineering Research Center, University of California; 1999. [9] Chopra AK. Dynamics of structures: Theory and application to earthquake engineering. Englewood Cliffs (NJ): Prentice-Hall; 2001. [10] Nassar A, Krawinkler H. Seismic demands for SDOF and MDOF systems. Report No. 95. Stanford (CA): The John A. Blume Earthquake Engineering Center, Stanford University; 1991. [11] Ohtori Y, Christenson RE, Spencer Jr BF, Dyke SJ. Benchmark control problems for seismically excited nonlinear buildings. J Eng Mech 2004;130(4):36685. [12] International conference of building officials. Structural engineering design provisions. Uniform building code, vol. 2. 1994. [13] OpenSees Development Team (Open Source Project), OpenSees: Open system for earthquake engineering simulation. Berkeley (CA): Pacific Earthquake Engineering Research Center, University of California; 19982002. http://opensees.berkeley.edu/. [14] Mahin S, Bertero V. An evaluation of some methods for predicting the seismic behavior of reinforced concrete buildings. UCB/EERC-75-5. Berkeley (CA): Earthquake Engineering Research Center, University of California; 1975. [15] Hachem M. BISPEC user manual. Berkeley (CA): University of California; 1999. www.ce.berkeley.edu/~hachem/bispec. [16] International Code Council (ICC). International building code. Falls Church (VA). 2003. [17] Somerville P, Smith N, Punyamurthula S, Sun J. Development of ground motion time histories for phase 2 of the FEMA/SAC steel project. report SAC/BD-97/04. Prepared for the SAC Joint Venture. Sacramento (CA): Structural Engineers Association of California; 1997. [18] Building Seismic Safety Council (BSSC). NEHRP recommended provisions for seismic regulations for new buildings and other structures, Parts 1 and 2, FEMA-302 and FEMA-303. Washington (DC): Federal Emergency Management Agency (FEMA); 1998. [19] Frankel A, Mueller C, Barnhard T, Perkins D, Leyendecker E, Dickman N. et al. National seismic hazard maps, June 1996: Documentation. US Geological Survey. Open-File report 96-532; 1996. [20] Kurama Y, Farrow K. Ground motion scaling methods for different site conditions and structure characteristics. Earthq Eng Struct Dyn 2003;32(15): 242550.

Вам также может понравиться