Large Structures and Infrastructures for Environmentally Constrained and Urbanised Areas 1

Comparative Assessment of Enhanced Multi-mode Pushover Analysis Methods for

Performance Based Design

Amelia KUSUMA Naveed ANWAR

Master Student Director
Asian Institute of ACECOMS
Technology AIT, Thailand
Thailand nanwar@ait.ac.th
kusuma.amelia@yahoo.com
Naveed Anwar, born 1958,
Amelia Kusuma, born 1986, received his civil engineering
received her civil engineering degree from the Univ. of Eng. &
degree from Asian Institute of Tech., Lahore and MS&Phd from
Technology, Thailand AIT, Thailand

Summary
Performance Based Design is a logical design process to determine the performance of such
structures where normal code-based prescriptive methods may not apply. Although Non-Linear
Time History Analysis has proven to be the most appropriate and reliable approach to follow, it
requires considerable effort, cost and skill, which may be either not practical or justified in many
cases. For such cases, Non-linear Static Pushover Analysis has become a popular “tool” for PBD.
However, NSPA has an inherent deficiency that its invariant load distribution cannot take the
higher-mode effect into consideration which may have significant effects in structure, especially
mid- to high-rise buildings. Several attempts had been made to develop NSPA so that the higher-
mode effects can be considered. This paper investigates the effectiveness of several enhanced
pushover methods in predicting the response of structures in comparison with NLTHA solution.

Keywords: Performance based design; Enhanced pushover analysis; Seismic evaluation.

1. Introduction
Performance Based Design (PBD) is a logical design process that will give a solution to achieve a
specified performance. Most Codes now incorporate a performance based design option as an
alternative to its prescriptive requirements. Linear analysis is far from accurate, while nonlinear
analysis is more difficult but can give rational result. For practical reasons, people may choose non-
linear pushover analysis than Non-Linear Time History Analysis (NLTHA). Despite of the “exact”
solution which NLTHA offered, higher computational cost compared to NSPA becomes a
disadvantage of NLTHA.

However, NSPA has several inherent deficiencies. One of them is that the NSPA invariant load
distribution cannot take the higher-mode effect into consideration which will take important role for
structures that their behavior will be affected by higher mode (e.g. tall buildings). This issue pushed
engineers to develop enhanced pushover procedures that can consider higher-mode effects.

2. VARIOUS METHOD
Pushover Analysis method can be categorized into three major groups based on how they analyze
the structure (lateral load vector). The groups are Single Mode Pushover Analysis Method
(SMPAM), Simple Pushover Analysis Method (SPAM), and Multi-mode Pushover Analysis
Method (MMPAM). Figure 1 shows the categorization of enhanced pushover methods.
2 34TH INTERNATIONAL SYMPOSIUM ON BRIDGE AND STRUCTURAL ENGINEERING, VENICE, 2010

2.1. Single Mode Pushover Analysis (SMPAM)

Pushover Analysis can be done using the mode shape of the structure as the pattern of its lateral
loading vector. The most common mode shape that is used as lateral loading vector for analysis is
the elastic first mode shape (fundamental mode shape) of the structure. However, these single mode
load distributions are inadequate when higher-mode (mode higher than the fundamental mode) of
the structure significantly contributes to the response of the structure.

2.2. Simple Pushover Analysis (SPAM)

Simple Pushover Analysis uses simple lateral load vector distribution. Two types of this loading
vector are uniform distribution and equivalent lateral force. FEMA-273 determined the uniform
distribution as s*i=mi , in which mi is the mass at i-th floor, and s*i is the lateral force at i-th floor.
The second type is the equivalent lateral force (ELF) distribution. Based on FEMA-273, ELF can be

obtained as , where the exponent k=1 for T1 < 0.5 sec (fundamental period), k=2 for
T1 > 2.5 sec and linear interpolation shall be used for values in between, hi and hx are heights from
the base to i-level floor and x, respectively.

2.3. Multi-mode Pushover Analysis (MMPAM)

In order to solve the problem of the structure which higher-mode effects are important, a number of
enhanced procedures which account for the higher-mode effects have been proposed. Most of the
current proposed methods have similarity with each other. They can be categorized into four groups
based on how they determine the lateral loading vector and also how they obtain the seismic
demands of the structures (results).

Figure 1: Category of Pushover Analysis Method

3. Methodology
This study will explore the ability of the current enhanced pushover procedure to simulate the
structure’s performance and seismic demands. The effectiveness of each method in the particular
type of structure will be investigated in this study. The methods that will be compared are Modal
Pushover Analysis (MPA), Consecutive Modal Pushover (CMP), Upper-bound Pushover Analysis
(UBPA) and also Simple Pushover Analysis (SPA) Method. All of those methods will be compared
with the “benchmark” solution from NLTHA.
 Large Structures and Infrastructures for Environmentally Constrained and Urbanised Areas 3

3.1 Modal Pushover Analysis (MPA)

Modal Pushover Analysis (MPA) that has been recently developed by Goel and Chopra, 2001 [1], is
an improved pushover procedure which take into account higher-mode effect in analyzing seismic
demand while retaining the simplicity of the loading pattern in pushover analysis. MPA utilizes the
concept of modal combination through several pushover analyses using invariant loading
distribution vector based on the elastic modes of the structure. The response of each pushover
analysis is being combined use SRSS or CQC combination rule to get the total response of the
structure.

3.2 Consecutive Modal Pushover (CMP)

Consecutive Modal Pushover (CMP) has been proposed by Poursha et.al., 2009 [2]. This method
uses a single-stage and multi-stage pushover analysis. The multi-stage pushover analysis is
conducted using an advantage of the consecutive implementation of MPA procedure. When one
stage of the modal pushover analysis has been performed, then the next stage (another modal
pushover analysis) begins with an initial state (stresses and deformations) from the end state
condition of the previous stage. CMP procedure is carried out with the various loading pattern
based on the modal properties of the linearly elastic structure.

3.3 Upper-bound Pushover Analysis (UBPA)

Upper-bound Pushover Analysis (UBPA) first proposed by Jan et.al, 2004 [3] determined the
distribution vector of the lateral loads over the height of the building by combining effects of the
first and second mode.

4. Structural System and Analytical Model

The structural system that will be used in this study is a 20- building. The floor plan and elevation
view of the structural systems in this study are shown in Figure 2. Nonlinearity in the frame section
will be carried out by defining hinge properties of each member. Fiber modeling will be used in
modeling nonlinearity of the shear wall section.
8400

19 x 3000
8400
8400

5000

6500 6200 6500

Figure 2: Floor plan and elevation view of the structures (all dimensions are in mm)

The computer program that will be used in this study is SAP2000 v.14.1.0 program [4]. This
computer program will be used for all the analysis that will be done in this study.

In order to obtain the “benchmark” result, Nonlinear Time History Analysis was conducted. Kobe
Earthquake ground motion was selected for the analysis. To facilitate a rational basis for comparing
different method’s procedure, the ground motion records were scaled so that it will be matched the
UBC’97 response spectrum (Z=2.00, Soil Profile D). Figure 3 shows the modified ground motion.
4 34TH INTERNATIONAL SYMPOSIUM ON BRIDGE AND STRUCTURAL ENGINEERING, VENICE, 2010

Earthquake Name Kobe, Japan

YEAR 1995
Earthquake Magnitude 6.90
EpiD (km) 24.20
Preferred NEHRP Based on D
Vs30
Preferred Vs30 (m/s) 312.0
PGA (g) 0.2668
PGV (cm/sec) 21.66
PGD (cm) 7.60

Figure 3: Modified Kobe Earthquake Ground Motion Detail

5. Evaluation of nonlinear static procedures

The performance of the structure can be investigated through some parameters which will be
obtained after the analysis. To validate the methods, they will be compared to NLTHA method as
the “benchmark” results. The results from NLTHA analysis will be presented in two values,
minimum (-) and maximum (+). The response evaluation parameters which were selected to
analyze the structural system are as follows:

(1) Story drift of the structure. Story drift is the lateral displacement that occurs in a single
story of multistory building. Story drift can be obtained as the relative drift between two
consecutive stories normalized by corresponding story height.

UBPA method can give quite similar response compared to NLTHA (+). CMP method seems
to over-estimate upper story and under-estimate bottom story drift profile. MPA method gives
good results for the bottom story drift profile compared with NLTHA (-) but overestimates it
at the upper storey. Figure 4 shows the inter story drift ratio obtained with each methods.

Figure 4 : Inter story drift profile

(2) Lateral displacement of the structure. To find the behavior of the structure when it is
subjected to the lateral loading can be observed from its lateral displacement.
 Large Structures and Infrastructures for Environmentally Constrained and Urbanised Areas 5

All methods can give reliable estimation for displacement profile. MPA, UBPA give close
results to the NLTHA. Only SPA underestimates displacement value of the structure. Figure 5
shows the peak displacement profile of the structure obtained with each procedure.

Figure 5 : Displacement profile

(3) Member hinge plastic rotation. Member hinge plastic rotation of the analyzed frame
members can determine the structure failure mechanism. Hinge properties of the members are
calculated based on the design of the structure.

From the results shows in Figure 7, UBPA underestimates column hinge plastic rotation, in
the other hand, CMP overestimates it. Beam hinge plastic rotation was estimated closely by
UBPA and CMP method at the bottom floor, but for the top floor, CMP overestimated it
while UBPA underestimated it. MPA can predict closest results among other methods for
both column and beam plastic hinge rotation value. Figure 6 shows the location of the
investigated beams and columns.

A4-C4 C4-E4
A4

A6

Figure 7 : Location of investigated beams and columns

6 34TH INTERNATIONAL SYMPOSIUM ON BRIDGE AND STRUCTURAL ENGINEERING, VENICE, 2010

Figure 7 : Beam and column plastic hinge rotation

(4) Base Shear. Total Base Shear of the structure is compared to verify the accuracy of each
method. Pushover methods cannot estimate total base shear of the structure well because
NLTHA includes all modes of the structure with its complexity while Pushover Analysis
method has limitation in including higher-mode effect because of many difficulties. CMP and
MPA give considerable estimation of the total base shear of the structure. Figure 8 shows
total base shear of the 20-storey building.

Figure 8 : Total Base Shear of 20-storey Building

(5) Story Shear Force and Shear Wall Shear Force. To find out how much shear wall takes
part in resisting the lateral forces when the structure is analyzed, shear force in shear wall
must be estimated and compared to the story shear force. In this study the results below was
taken for the 2nd level of the structure. Only CMP and MPA method seem can give reliable
 Large Structures and Infrastructures for Environmentally Constrained and Urbanised Areas 7

estimation while the others cannot predict well. Figure 9 shows comparison of Story shear
forces and Shear wall forces of 20-storey building.

Figure 9 : Story Shear and Shear Wall Shear Force

(6) Shear Wall Stress-Strain. In this study, shear wall in the structure was modeled using fiber-
modeling technique (Location of the nonlinear-links is shown at Figure 10). To find out
whether it already yielded or not, stress and strain of the fiber should be investigated. The
stress-strain yield point of concrete ((-) sign) is 49(MPa)-0.002, and the stress-strain yield
point of concrete ((+) sign) is 400(MPa)-0.0047. No shearwall has reach yield point for all
methods. Only CMP method can predict close value of stress and strain of shearwall
compared to NLTHA. MPA results cannot be obtained because the results for each mode
cannot be combined. Different loading direction (+ or -) will affect which fiber will take
compression/tension side, concrete will take main part in compression force while tension
force will be taken by steel. If nth-mode analysis has opposite loading direction, then stress
and strain result cannot be combined with the other mode result in MPA. Figure 11 shows
shearwall stress and strain value for different methods.

1U1 1U4 1R13

1B1 1B4 1R1

Figure 10 : Location of Nonlinear- links

Figure 11 : Shear Wall Stress and Strain

8 34TH INTERNATIONAL SYMPOSIUM ON BRIDGE AND STRUCTURAL ENGINEERING, VENICE, 2010

6. Conclusions
In the needs of Performance Based Design (PBD) “tools”, attempts have been made to include
higher-mode effect to the Pushover Analysis method so that it can obtain reliable results. Some
parameters have been investigated in this study, and compared to validate the methods. This study
compared all 3 methods (CMP, MPA and UBPA) that include higher-mode effects to the analysis
and compared to the Simple Pushover Analysis (SPA) and also “benchmark” result from NLTHA.
Research has been conducted and led to the following conclusions:

(1) Enhanced Pushover Analysis methods that have been studied in this research tend to be
conservative in giving displacement value and story drift ratio of the structure compared to the
NLTHA.
(2) To predict frame hinge plastic rotation value of the structure, the best overall methods is
MPA. CMP tends to overestimate it, while UBPA underestimate it.
(3) To obtain total base shear, story shear force, and shearwall shear force, only MPA and CMP
can give reasonable results compared with NLTHA. Other methods underestimate total base
shear value, story shear force and shearwall shear force.
(4) The Pushover methods tend to give reasonable results for overall structural response and
deformations, but are less reliable for determining the local response of members.

Based on the results and analysis of the investigated methods, it can be concluded that enhanced
pushover analysis methods is able to be used to evaluate particular response (displacement, drift,
base shear, story shear force, shearwall shear force) of the structure although some of the results
cannot give conservative results compared to NLTHA.

7. References
[1] CHOPRA A.K., GOEL R.K., “A Modal Pushover Analysis Procedure for Estimating
Seismic Demands for Buildings”, Pacific Earthquake Engineering Research Center, PEER
201/03 Jan.2001.
[2] POURSHA M., KHOSHNOUDIAN F., MOGHADAM A.S., “A Consecutive Modal
Pushover Procedure for Estimating The Seismic Demands of Tall Buildings”, Engineering
Structures 2009, 31:591-599.
[3] JAN T.S., LIU M.W., KAO Y.C., “An Upper-bound Pushover Analysis Procedure for
Estimating Seismic Demands of High-rise Buildings”, Engineering Structures2004, 26:117-
28.
[4] COMPUTER &STRUCTURES INCORPORATED (CSI), SAP 2000 v14.1.0, Berkeley
(USA, CA).