13th World Conference on Earthquake Engineering

Vancouver, B.C., Canada

August 1-6, 2004
Paper No. 2713

METHOD OF MODAL COMBINATIONS

FOR PUSHOVER ANALYSIS OF BUILDINGS

Erol KALKAN1 and Sashi K. KUNNATH2

SUMMARY

Nonlinear static procedures (NSP) are finding widespread use in performance based seismic design since
it provides practitioners a relatively simple approach to estimate inelastic structural response measures.
However, conventional NSPs using lateral load patterns recommended in FEMA-356 do not adequately
represent the effects of varying dynamic characteristics during the inelastic response or the influence of
higher modes. To overcome these drawbacks, some improved procedures have recently been proposed by
several researchers. A method of modal combinations (MMC) that implicitly accounts for higher mode
effects is investigated in this paper. MMC is based on invariant force distributions formed from the
factored combination of independent modal contributions. The validity of the procedure is validated by
comparing response quantities such as inter-story drift and member ductility demands using other
pushover methods and also the results of nonlinear time history analyses. The validation studies are based
on evaluation of three existing steel moment frame buildings: two of these structures were instrumented
during the Northridge earthquake thereby providing realistic support motions for the time-history
predictions. Findings from the investigation indicate that the method of modal combinations provides a
basis for estimating the potential contributions of higher modes when determining inter-story drift
demands and local component demands in multistory frame buildings subjected to seismic loads.

INTRODUCTION

Current seismic design practice in the United States is still governed by force-based design principles.
However, the emergence of performance-based seismic engineering has resulted in increasing use of
nonlinear methods to estimate expected seismic demands in a building structure. A widely used and
popular approach to establish these demands is a “pushover” analysis in which a mathematical model of
the building is subjected to an inverted triangular distribution of lateral forces. While such a load
distribution is based on the assumption that the response is primarily in its fundamental mode of vibration,
it can lead to incorrect estimates for structures with significant higher mode contributions. This

1
PhD. Student, University of California, Davis, CA, USA. Email: ekalkan@ucdavis.edu
2
Assoc. Prof., University of California, Davis, CA, USA. Email: skkunnath@ucdavis.edu
accentuates the need for improved procedures that addresses current drawbacks in the lateral load patterns
that are used in pushover analyses. Recently, several improved pushover procedures have been proposed
[1,2]. These procedures have been shown to provide more accurate estimates of interstory drift values than
conventional NSPs using inverted triangular, uniform or other lateral load patterns based on direct modal
combination rules suggested in FEMA-356 [3]. The effort required to implement these procedures in
routine analysis is significant and, therefore, not an attractive proposition in engineering practice. In order
to investigate alternative simple schemes to represent realistic lateral force demands, a new lateral load
configuration using factored modal combinations has been recently developed [4]. The accuracy of this
approach, to be referred to in this paper as the Method of Modal Combinations (MMC), has been
validated on two RC buildings [4]. Other pushover procedures, which are relevant to the work described
in this paper, are the methods described by Paret et al. [5] and Sasaki et al. [6].

The principal objective of this paper is to extend the validation study to additional buildings and to
compare predictions to other approaches such as the Modal Pushover Analysis (MPA) procedure proposed
by Chopra and Goel [2] and to the results of detailed time-history analyses. Three existing buildings with
varying story levels (4, 6 and 13) were selected for the evaluation. The six and thirteen story buildings
were instrumented during the Northridge earthquake, therefore, recorded base motions were utilized in
nonlinear time history (NTH) analyses. For the four-story building, which was not instrumented, three
different ground motions were used to predict expected seismic demands.

NONLINEAR STATIC PROCEDURES

The MMC procedure is evaluated by comparing computed interstory drift demands to nonlinear time-
history estimates and to other pushover procedures. One set of lateral load patterns was based on
recommendations in FEMA-356 while the second methodology considered in the comparative study is the
modal pushover analysis (MPA) of Chopra and Goel [2]. A brief overview is presented of the different
NSP methodologies used in the study.

Lateral Load Patterns Based on FEMA-356

In FEMA-356, several alternative invariant loading patterns are recommended for estimating equivalent
seismic demands. In this study, two loading patterns are considered. These two patterns are permitted
when more than 75 percent of the total mass participates in the fundamental vibration mode in the
direction under consideration. The following notations are used in this paper to describe these patterns:

NSP-1: The buildings are subjected to a lateral load distributed across the height of the building based on
the following formula specified in FEMA-356:

Wx hxk
Fx = V (1)
N
k
∑ Wi hi
i =1
where, Fx is the applied lateral force at level ‘x’, W is the story weight, h is the story height and V is the
design base shear, and N is the number of stories. The summation in the denominator is carried through all
story levels. This results in an inverted triangular distribution when k is set equal to unity.

NSP-2: A uniform lateral load distribution consisting of forces that are proportional to the story masses at
each story level.
Modal Pushover Analysis (MPA)

Modal Pushover Analysis (MPA), developed by Chopra and Goel [2], is essentially the extension of single
mode pushover analysis to multi-mode response, and use of the theory of response spectrum analysis to
combine the modal contributions. The basic steps of the procedures are as follows:

1. Compute the natural frequencies, ωn and mode shapes using an elastic model of the system.
2. Run pushover analyses with the loading patterns (sn= m φn) based on each mode independently.
3. Idealize each pushover curve as bilinear curves considering negative post-yield stiffness if
necessary.
4. Convert the idealized pushover curves into a set of capacity spectrum curves of the corresponding
SDOF system using the ADRS conversion from MDOF to SDOF (Note that guidelines provided
in ATC-40 [7] capacity spectrum procedure can be used for this purpose).
5. Compute the peak response corresponding to each SDOF system via a nonlinear response history
analysis (NRHA) based on an input ground motion for each SDOF system or via inelastic design
spectrum.
6. Convert the peak response of SDOF system to the target displacement of MDOF system for each
mode separately.
7. From the pushover database (Step 2), extract the peak inelastic response quantities of interest,
such as interstory drift and plastic hinge rotations independently for each mode.
8. By using SRSS, determine the combined peak response.

At the first glance, MPA procedure is an adaptation of NRHA for inelastic static analysis. However, this
process inherently requires considerable effort except if very few modes are considered in the evaluation.
In its original form, MPA is not a static method since it requires repetitive runs of SDOF response history
analyses for a given ground motion to identify the target displacement of each mode. Additionally, it
requires the use of one or more ground motions unless an inelastic design spectrum is used. Running the
pushover analyses independently in each mode and ignoring the contribution of other modes in the
formation of plastic hinges is an issue of concern for MPA since it may result in inaccurate estimates of
plastic hinge rotation, an important parameter for comparing acceptance criteria in performance-based
evaluation.

Method of Modal Combinations (MMC)

In this procedure (see Kunnath [4]) the spatial variation of applied forces is determined from:

Fj = Σ α n Γ n m Φ n Sa (ζ n , Tn ) (2)

where α n is a modification factor that can assume positive or negative values; Φ n is the mode shape
vector corresponding to mode n; S a is the spectral acceleration at the period corresponding to mode n;
and

Γ = ([Φ ] [m]{ι}) / M n in which M n = [Φ]T [m][Φ ]

T
(3)

If only the first two modes are used in the combination process, then Equation 2 would have the following
form:

Fj = α1Γ1mΦ1 Sa (ζ 1 , T1 ) ± α 2 Γ 2mΦ 2 Sa (ζ 2 , T2 ) (4)

Therefore, the procedure requires multiple pushover analyses wherein several combinations of modal load
patterns are applied. In order to arrive at estimates of deformation and force demands, it is necessary to
consider peak demands at each story level and then establish an envelope of demand values for use in
performance based-evaluation.

DESCRIPTION OF BUILDINGS

Three special moment resisting steel frame buildings were selected as representative case studies to
evaluate the MMC procedure.

4-Story Building

The building was designed in according to 1988 UBC specifications. It is 16.15m in elevation and has a
rectangular plan with plan dimensions of 33.27m x 19.2m. The structural system is composed of perimeter
MRFs to resist lateral loads and interior gravity frames. The floor plan and elevation view of the building
with beam and column sections are shown in Figure 1.

A B C D E
4

3
3@6.1 m

2 Moment resisting connection

Simple hinge connection

(a) Plan view of perimeter frames

1
9.5 m 9.14 m 12.2 m 2.43 m

A B C D E
W24x68 W24x68 W24x68 W16x26
Roof
W14x145 W14x145 W14x145 W14x145

W14x211 W14x211 W14x211 W14x211

W14x211 W14x211 W14x211 W14x211

W14x145 W14x145 W14x145 W14x145

4.26 m
W24x76 W24x76 W24x76 W21x50
3rdFloor

W24x94 W24x94 W24x94 W21x50

3@3.96 m

2ndFloor

W24x104 W24x104 W24x104 W21x50

1st Floor

(b) Elevation
9.5 m 9.14 m 12.2 m

Figure 1. Structural configuration of 4-story building

The columns of the MRFs are embedded into grade beams and anchored to the top of the pile cap, and the
foundation system is composed of drilled concrete piers with pile caps, grade beams and tie beams. All
columns are made of A-572 grade 50 steel. The girders and beams are made of A-36 steel. The floor
system is composed of 15.9 cm thick slab (8.3 cm light weight concrete and 7.6 cm composite metal deck)
at all floor levels and the roof. The outside walls are made of thin set brick veneer supported on a metal
stud wall. This building suffered significant flange fracture damage in beam-flange to column-flange
connections during 1994 Northridge earthquake [8]. All of the severely fractured beam-columns
connections were concentrated in the NS direction in the moment frame on Line-D (Fig. 1). No fracture
was observed in the NS direction moment frame on Line-A, and only one fracture was observed on Line-1
in the EW direction. This building was not instrumented. Further details of the building are given in
Krawinkler et al. [8].

6-Story Building

This moment frame steel structure was designed in accordance with 1973 UBC requirements. The
rectangular plan of the building measures 36.6m x 36.6m with a 8.2cm thick light weight concrete slab
over 7.5cm metal decking. The primary lateral load resisting system is a moment frame around the
perimeter of the building. Interior frames are designed to carry only gravity loads. The plan view and the
elevation of a typical frame are shown in Figure 2. The building was instrumented with a total of 13
strong motion sensors at the ground, 2nd, 3rd and roof levels. The building performed well in the
Northridge earthquake with no visible signs of damage. In constructing the building model, the columns
were assumed to be fixed at the base level (all columns are supported by base plates anchored on
foundation beams which in turn are supported on a pair of 9.75m - 0.75m diameter concrete piles). The
specified minimum concrete compressive strength at 28 days was 20.7 MPa. Section properties were
computed for A-36 steel with an assumed yield stress of 303 MPa. The total building weight (excluding
live loads) was estimated to be approximately 34,644kN. Additional details including calibration of the
building model is reported in Kunnath et al. [9].

A B C D E F G
A B C D E F G
1
W24x68
Roof
W14x90

2 W24x84
5th Floor
3 W24x68
4th Floor
5@4m

W14x132
6@6.1m

W24x68
4 3rdFloor
W27x102 2ndFloor
5
W14x176

W30x116
1st Floor
6
5.3m
7
6@6.1m 6@ 6.1m
Moment resisting connection
Simple hinge connection (b) Elevation

(a) Plan view of perimeter frames

Figure 2. Structural configuration of Burbank 6-story building

13-Story Building

This building is located in South San Fernando Valley about 5 km southwest of the Northridge epicenter
and is composed of one basement floor and 13 floors above ground. Built in accordance with the 1973
UBC code, this structure has been the subject of a previous investigation [9,10]. As shown in Figure 3, it
has a 48.8m square plan and 57.5m in elevation. The lateral load resisting system is composed of four
identical perimeter frames. The floor plan increases at the second floor to form a plaza level that
terminates on three sides into the hillside thereby making this level almost fixed against translation.
Recorded response of the building during the Northridge earthquake indicates a peak horizontal
acceleration of 0.44g, 0.32g and 0.33g at the ground, 6th floor and roof levels. Weld fracture damage was
observed primarily in the NS direction.

4 5 6 7 8 9 W27x84
Roof
B W14x500 W14x426 W14x398 W14x314 W14x287 W14x246 W14x167
W33x118
12th Floor
C W33x118
11th Floor
W33x130
10th Floor
D W33x130
48.8 m

9th Floor
12@4.013 = 48.2 m

W33x141
E 8th Floor
W33x141
7th Floor
F W33x152
6th Floor
W33x152
5th Floor
G W33x152
4th Floor
48.8 m
W33x152
3rd Floor
W33x152
Moment resisting connection 2nd Floor
W36x230
Simple hinge connection 1st Floor
4.88
W33x194 Plaza Level
(a) Plan view of perimeter frames
4.42

5@9.75 = 48.8 m

(b) Elevation

Figure 3. Structural configuration of the 13-story building

SUMMARY OF EVALUATION

The evaluation process consisted of comparing the computed demands using MMC with time-history
analysis and with other pushover methods. For the two instrumented buildings, the recorded base motions
served as the input accelerations for the time history analyses. Since the actual ground motions did not
produce significant inelastic behavior, the records were scaled so as to induce a peak interstory drift of
approximately 2 percent at any level. The target displacements for the pushover procedures were then
based on the peak time-history induced story drifts. This approach provides a rational basis for comparing
the demands obtained with different methods. Since instrumented information was not available for the
four-story structure, three ground motion records were selected from the recommended set in ATC-40 [7].
Comparison of interstory drift demands comprised the primary basis for the evaluation. Typical member
ductility demands (based on plastic rotations) were also evaluated for the MMC and FEMA procedures.
The MPA procedure does not lend itself to a direct evaluation of component ductility. Details of the
ground motions used for the time-history analyses of the three buildings are presented in Table 1.

Table 1. Details of ground motions used in nonlinear time history analyses

EQ. Scale Factor *

Year Earthquake Recording Station PGA (g) 4-Story 6-Story 13-Story

1989 Loma Prieta Hollister, South & Pine 0.370 1.0 - -
1989 Loma Prieta Gilroy #2 0.350 2.1 - -
1994 Northridge Moorpark 0.290 4.0 - -
1994 Northridge Burbank 0.299 - 2.3 -
1994 Northridge Woodland Hills 0.318 - - 1.9
* Scale factor used to achieve the target interstory drift ratio of two percent

Analytical Modeling

The nonlinear evaluations were carried out using the open source finite element framework, OpenSees
[11]. A nonlinear beam-column element that utilizes a layered ‘fiber’ section is used to model all
components in the frame models since the interaction of axial force and flexure is automatically
incorporated. The element is based on a force formulation that considers the spread of plasticity. Since the
objective of the evaluation is to evaluate various pushover procedures rather than simulate local
connection fracture, the modeling of the members and connections was based on the assumption of stable
hysteresis derived from a bilinear stress-strain model. Since the buildings are symmetric in plan, only two-
dimensional models of a single frame were developed for each building. In the case of the four-story
building, the exterior frame along EW direction (Line-1) was modeled. Similarly, frame models for the six
and thirteen story buildings were developed for the exterior frames in EW direction. The elastic models
were validated using available recorded data and typical simulations are displayed in Figure 4.

10 Recorded
Simulated
Disp. (cm)

0
th
6 Floor
-10
(a) 0 20 Time (sec) 40 60

40 Recorded
Simulated
Disp. (cm)

0
th
12 Floor
-40
0 10 20 30
(b) Time (sec)

Figure 4. Validation of analytical models: comparison of recorded and computed response (a) EW
response at roof level of 6-story building; (b) EW response at roof level of 13-story building
Evaluation of Method of Modal Combination (MMC)

The validation exercise presented in Figure 4 was obtained using elastic models indicating minimal or no
inelastic behavior. Hence, it was necessary to scale the recorded ground motions so that interstory drifts
reached a magnitude to cause yielding in elements and provide a more reasonable basis to evaluate the
adequacy of the pushover methods. The scale factors used to produce 2 percent peak story drifts are given
in Table 1 for each ground motion. The 5-percent damped elastic acceleration and displacement response
spectra computed from the amplified motions are presented in Figure 5. Also marked on these figures are
the first two modal periods for the four- and six-story models and three modal periods for the thirteen-
story model.

4.0 50
T2 4-Story Building
40
3.0 Mean T1
Max Sa at Tn
30

Sd (cm)
Sa (g)

2.0
T1 20
T2
1.0 Mean
10
Max Sd at Tn
0.0 0
0.0 1.0 2.0 3.0 0.0 1.0 2.0 3.0
(a) Period (sec) Period (sec)

3.0 20
6-Story Building
2.5 T1
15
2.0 T2 T2
Sd (cm)
Sa (g)

1.5 10

1.0
T1 5
0.5

0.0 0
0.0 1.0 2.0 3.0 0.0 1.0 2.0 3.0
(b) Period (sec) Period (sec)

3.0 100
13-Story Building
2.5
75
2.0 T3
Sd (cm)
Sa (g)

1.5 50
T2 T2
1.0 T3
25 T1
0.5 T1

0.0 0
0.0 1.0 2.0 3.0 4.0 0.0 1.0 2.0 3.0 4.0
(c) Period (sec) Period (sec)

Figure 5. 5-percent damped pseudo acceleration spectra and displacement spectra for (a) 4-story building
(based on three ground motions); (b) 6-story building; (c) 13-story building
Evaluation of Interstory Drift Demands

As indicated previously, three earthquake records were used for the nonlinear time history (NTH) analysis
of the four-story building. The MMC response is based on the envelope of demands resulting from Mode
1 ± Mode 2 (using peak Sa of the three ground motions). The resulting lateral forces using such a modal
combination is shown in Figure 6. Plots of the displacement and drift profile for both NTH runs and the
various pushover methods are shown in Figure 7. The peak displacement profiles are generally similar for
all methods. The variation of interstory drift indicates that both MPA and MMC capture the demands
with reasonable accuracy though the demand at the first story is slightly over-estimated. Of the two FEMA
methods, NSP-1 (first mode distribution) is a better indicator of seismic demands though the drifts at the
uppermost level are under-estimated. Note that NTH gives the highest demand at third story that implies
some contribution of the second mode in the response.

4 0.93 4 0.31 0.31 4 1.234 0.62

3 0.84 -0.09 3 3 0.75 3 0.93

2 0.56
± -0.34 2
= 2 0.21 2 0.90

1 0.23 -0.24 1 1 0.00 1 0.47

0 0.00 0.000 0 0.00 0 0.00

S1 (Mode 1) S2 (Mode 2) S1 +S2 S1 - S2

Figure 6. Spatial distribution of lateral forces for 4-story building, Sn = Γn m φn , for n = 1 and 2

4 4

3 3
Story Level

Story Level

2 2

NTH Mean
1 NSP-1 1
NTH Mean
NSP-2 NSP-1
MMC NSP-2
MPA MMC
MPA
0 0
0.00 0.01 0.02 0.00 0.01 0.02 0.03
Roof Drift Ratio Interstory Drift Ratio

Figure 7. Comparison of roof drift and interstory drift ratio using various methods for 4-story building
As in the case of the four-story building, the MMC procedure was developed from lateral forces using a
modal combination based on Mode 1 ± Mode 2. This combination led to the lateral force distribution as
indicated in Figure 8. As is evident from these distributions, the two combinations place increased
demands on either the upper or middle stories and is a function of both the mode shape and the spectral
demands at these modal periods. The resulting story demands are plotted in Figure 9 along with demand
estimates from the other methods. Significant higher mode effects are apparent in Figure 9b. MMC
captures the highest story drift as well as the other story drifts reasonably, however the drift demand is
overestimated at the first story. The MPA procedure generally under-predicts the demands at most levels.
For this particular building none of the FEMA procedures show good correlation with the time history
results in terms of the story drifts, they overestimate the demand at the lower levels and underestimate it at
upper levels.

6 1.73 6 0.64 0.64 6 2.37 6 1.09

5 2.03 5 0.38 0.38 5 2.41 5 1.65

4 1.70 -0.23 4 4 1.47 4 1.93

3 1.32
± -0.61 3
= 3 0.71 3 1.93

2 0.90 -0.71 2 2 0.19 2 1.61

1 0.64 -0.62 1 1 0.01 1 1.26

0 0.00 0.000 0 0.00 0 0.00

S1 (Mode 1) S2 (Mode 2) S1 +S2 S1 - S2

Figure 8. Spatial distribution of lateral forces for 6-story building, Sn = Γn m φn , for n = 1 and 2

6 6

5 5

4 4 NTH
NSP-1
Story Level
Story Level

NSP-2
3 MMC
3
MPA

2 2
NTH
NSP-1
1 NSP-2 1
MMC
MPA
0 0
0.00 0.01 0.02 0.00 0.01 0.02 0.03
Roof Drift Ratio Interstory Drift Ratio

Figure 9. Comparison of roof and interstory drift ratios using various methods for 6-story building
For the thirteen-story building, the MMC procedure was developed from lateral forces using a modal
combination based on Mode 1 ± Mode 2 + Mode 3. This combination led to the lateral force distribution
as indicated in Figure 10. Inclusion of third mode indicates a significantly altered load distribution. For
this building, the combinations based on the first two modes, as used for the other buildings, was also
evaluated in addition to the two combinations shown in Figure 10, and analyses were conducted for two
configurations separately. The resultant demands are given in Figure 11 for comparison. The predicted
demands from the other methods are also presented. In general demands predicted by MMC are in
agreement with the computed demand from NTH analysis. Incorporation of the third mode improved the
capability of MMC to capture the time-history demands. Except for the lower story levels, neither FEMA
procedures nor MPA show good correlation with the computed demand from NTH analysis.

14 1.97 14 0.82 0.83 14 0.54 14 3.34 14 1.69

13 3.87 13 1.43 1.43 13 0.73 13 6.0213 3.16
12 3.71 12 0.99 0.99 12 0.08 12 4.77 12 2.80
11 3.52 110.52 -0.43 -0.42
11 11 3.61 11 2.58
10 3.28 -0.01
10 -0.76 -0.76 10 10 2.51 10 2.52
9 3.00 -0.49 9 -0.82 -0.82 9 9 1.68 9 2.66
8 2.68 -0.90 8 -0.61 -0.618 8 1.17 8 2.97
7 2.34 ± -1.18 7 + 7-0.23 = 7 0.93 7 3.29
6 1.98 -1.31 6 6 0.21 6 0.88 6 3.50
5 1.64 -1.30 5 5 0.54 5 0.88 5 3.48
4 1.28 -1.17 4 4 0.73 4 0.85 4 3.18
3 0.92 -0.93 3 3 0.74 3 0.73 3 2.60
2 0.67 -0.72 2 2 0.66 2 0.60 2 2.05
1 0.26 -0.29 1 1 0.29 1 0.25 1 0.83
0 0.00 0.000 0 0.00 0 0.00 0 0.00
S1 (Mode 1) S2 (Mode 2) S3 (Mode 3) S1 +S2+S3 S1 -S2+S3

Figure 10. Spatial distribution of lateral forces for 13-story building, Sn = Γn m φn , for n = 1, 2 and 3

14 14 NTH 14
NTH
13 13 NSP-1 13 NSP-1
NSP-2 NSP-2
12 12 12
MMC MMC
11 11 MPA 11 MPA
10 10 10
9 9 9
Story Level

Story Level

Story Level

8 8 8
7 7 7
6 6 6
5 5 5
NTH
4 4 4
NSP-1
3 NSP-2 3 3
2 MMC 2 2
MPA
1 1 1
0 0 0
0.00 0.01 0.02 0.00 0.01 0.02 0.03 0.00 0.01 0.02 0.03
Roof Drift Ratio Interstory Drift Ratio Interstory Drift Ratio

Figure 11. Comparison of roof and interstory drift ratios using various methods for 13-story building
The displacement profile of the six and thirteen story buildings during the time history analysis is captured
in Figures 12 and 13. Also shown in these figures are the corresponding story drift histories. Though not
immediately evident from the figure, it was observed that the drift profile is initially representative of the
modal contributions to the response based on spectral demands and that these demands change as the
systems become inelastic and the modal periods shift along with the corresponding spectral demands.

Figure 12. Profile of response history for the six-story structure:

(a) Roof drift ratio history; (b) Interstory drift ratio history

Figure 13. Profile of response history for the thirteen-story structure;

(a) Roof drift ratio history; (b) Interstory drift ratio history
Evaluation of Component Ductility Demands

Another important parameter in seismic response analysis is the estimate of ductility demands in
individual components. FEMA-356, for example, uses component acceptance criteria using ductility
demands as the fundamental basis of its performance-based evaluation methodology. In this section, the
effectiveness of the MMC procedure to estimate component demands is investigated. Tables 2 and 3
show typical ductility demands for column elements at critical story levels in the six and thirteen story
buildings experiencing the highest deformation demand. Also given are the global system ductility
demands which are less than the observed local story and component ductility demands. Similar results
were obtained in a more comprehensive study by the authors examining ductility demands of RC and steel
buildings [12]. These results serve as evidence that designing a building to achieve a certain ductility
demand may result in much larger demands at the local level.

Comparisons of the ductility demand from pushover procedures with by nonlinear time-history analyses
show that the predicted demands are remarkably similar to those estimated. A more visual comparison is
provided in Figures 14 and 15 where the moment-rotation behavior of three typical column sections
undergoing inelastic deformation is displayed. While cumulative effects cannot directly be incorporated
into any static procedure, the ability of MMC to estimate component deformations is clearly demonstrated
in these figures.

Table 2. Typical ductility demands in 6-story building

Location NSP-1 * NSP-2 * NTH MMC
Global - 1.53 - 1.92
5th Story - 0.0 0.0 - 2.02
5th Story Column Interior 0.0 0.0 2.81 2.73
* NSP-1:Inverted triangle; NSP-2: Mass proportional

Table 3. Typical ductility demands in 13-story building

Location NSP-1 * NSP-2 * NTH MMC
Global - 2.08 2.24 - 2.05
7th Story - 2.19 1.32 - 2.59
7th Story Column Interior 3.28 1.67 3.69 3.74
9th Story - 1.30 0.0 - 1.90
9th Story Column Interior 1.61 0.0 2.60 2.70
* NSP-1:Inverted triangle; NSP-2: Mass proportional

CONCLUSIONS

The popularity of nonlinear static pushover analysis in engineering practice calls into question the validity
of conventional lateral load patterns used to estimate inelastic demands. The aim of the present work is to
develop alternative multi-mode pushover analysis procedures by indirectly accounting for higher mode
contributions but yet retaining the simplicity of invariant distributions in a theoretically consistent manner.
A new combination scheme is investigated in this paper and compared to both time-history procedures
and other pushover methods. The evaluation is based on a series of analyses of existing steel moment
frame buildings.
 1.5
NTH MMC

1.0

0.5

M/M y
0.0
-1 0 1 2 3
-0.5

-1.0

th
-1.5 5 Story Interior
θ/θ y Column

Figure 14. Comparison of ductility demands in typical interior column from MMC and NTH
for 6-story building (θ implies total rotation, θy implies yield rotation)

2.0 2.0
NTH MMC NTH MMC

1.0 1.0
M/M y

M/M y

0.0 0.0
-2 0 2 4 -2 -1 0 1 2 3

-1.0 -1.0

th th
-2.0 7 Story Interior -2.0 9 Story Interior
θ/θ y Column θ/θ y Column

Figure 15. Comparison of inelastic static demand using MMC and cyclic demand in typical interior
column of the 13-story building

This study indicates that pushover methods utilizing lateral force distributions based on a single mode are
not capable of predicting the story level at which the critical demands occur. On the other hand, the
results of modal combination procedures based on ongoing research appears to be promising in terms of
better estimating peak values of critical inelastic response quantities such as inter-story drifts and plastic
hinge rotations. It is shown that considering sufficient number of modes, interstory drifts estimated by
MMC is generally similar to trends noted from NTH analyses unless the building is deformed far into the
inelastic range with significant strength and stiffness deterioration.

Higher mode effects on seismic demand are dependent both on the frequency content of the ground
motion and the characteristics of the structural system even for regular low-rise buildings (based on
findings from the four-story building evaluation). In the present phase of the research, the force
distributions are based on modal contributions in the elastic state of the system. The influence of higher
modes in the inelastic phase of the response can be incorporated by introducing modification factors that
account for changes in spectral demands due to inelastic effects. Additional studies considering various
structural systems and ground motions are ongoing to further validate the methodology.

ACKNOWLEDGEMNT

Funding for this study provided by the National Science Foundation under Grant CMS-0296210, as part of
the US-Japan Cooperative Program on Urban Earthquake Disaster Mitigation, is gratefully acknowledged.
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the
author and do not necessarily reflect the views of the National Science Foundation.

REFERENCES

1. Gupta, B. and Kunnath, S.K. (2000). “Adaptive spectra-based pushover procedure for seismic
evaluation of structures.”, Earthquake Spectra, 16(2), 367-392.
2. Chopra, A.K. and Goel, R.K. (2002). “A modal pushover analysis procedure for estimating seismic
demands for buildings.” Earthquake Engineering and Structural Dynamics, 31,561-582.
3. FEMA-356. (2000). Prestandard and Commentary for the seismic rehabilitation of buildings,
American Society of Civil Engineers (ASCE), Reston, VA.
4. Kunnath, S. K. (2004). “Identification of modal combinations for nonlinear static analysis of
building structures.” Computer-Aided Civil and Infrastructure Engineering.19: 282-295.
5. Paret T.F., Sasaki K.K., Eilbeck, D.H. and Freeman, S.A. (1996). “Approximate inelastic procedures
to identify failure mechanisms from higher mode effects.” Paper No. 966, Proc. of the 11th WCEE,
Acapulco, Mexico.
6. Sasaki, K.K, Freeman, S.A. and Paret, T.F. (1998). “Multimode pushover procedure (MMP) – A
method to identify the effects of higher modes in a pushover analysis.” Proc. of the U.S. National
Conf. on Earthquake Engineering, Seattle, WA.
7. Applied Technology Council (ATC-40). (1996). Seismic evaluation and retrofit of concrete
buildings, Redwood, CA.
8. Krawinkler, H. and Al-Ali, A. (1996). “Seismic demand evaluation for a 4-story steel frame
structure damaged in the Northridge earthquake.” The Structural Design of Tall Buildings, 5(1), 1-
27.
9. Kunnath, S.K., Nghiem, Q. and El-Tawil, S. (2004). “Modeling and response prediction in
performance-based seismic evaluation: case studies of instrumented steel moment-frame buildings.”
Earthquake Spectra. 20 (3).
10. Uang, C.M., Yu, Q.S., Sadre, A., Youseff, N. and Vinkler, J. (1997). “Seismic response of an
instrumented 13-story steel frame building damaged in the 1994 Northridge Earthquake.”
Earthquake Spectra, 13(1), 131-149.
11. OpenSees. (2004). Open system for earthquake engineering simulation,
http://opensees.berkeley.edu.
12. Kunnath, S.K. and Kalkan, E. (2004). “Evaluation of seismic deformation demands using nonlinear
procedures in multistory steel and concrete moment frames.” ISET Journal of Earthquake
Technology. 41(1).