Nonlinear Pushover Analysis of RC Structures

R. Martino’, E. Spacone and G. Kingsley 3

Abstract
With the inclusion of the Non - Linear Static Procedure (NSP) or pushover analysis
Downloaded from ascelibrary.org by Istanbul Universitesi on 08/11/13. Copyright ASCE. For personal use only; all rights reserved.

into the Federal Emergency Management Agency Document 273 (Fema 273), the
need for non - linear pushover analysis tools for structural design in seismic zones is
apparent. Focusing on reinforced concrete structures, this entails the ability to
perform the NSP on lateral force resisting systems consisting of moment frames,
structural walls, or any combination of these. The authors are currently developing a
pushover analysis tool, using an existing finite element program, which can
accomplish these tasks. This paper details the steps in performing the NSP, compares
preliminary results of the NSP with a full non - linear dynamic analysis, and
discusses a simplified procedure to model shear deformations in reinforced concrete
members.
Introduction
As the United States, Japan, and Europe move towards implementing Performance
Based Design concepts, new seismic design provisions based on this philosophy will
require engineers to perform non-linear analyses of structures. These analyses can
take the form of a full, non-linear dynamic analysis, or a static non-linear pushover
analysis. Availability of an easy to use and accurate non-linear pushover analysis
tool for reinforced concrete structures, which can easily be implemented in a design
office, becomes evident. Even though recent years have seen a great amount of
research in the development of non-linear models, there is still a lot of work to do for
reinforced concrete structures, including modeling of bond slip, structural walls,
structural joints and non-structural members.
The focus of this paper is the non-linear pushover analysis procedure based on the
provisions for seismic rehabilitation of buildings contained in FEMA 273. Results
from a non-linear dynamic analysis will be compared with those from a non-linear
pushover analysis performed on reinforced concrete moment resisting frames of six,
twelve, and twenty stories. In comparing the pushover analysis to the full dynamic
analysis on reinforced concrete frames of increasing numbers of stories, the
limitations of the pushover analysis may be quantified, and its benefits over the
dynamic analysis can be understood.
Finally, the paper discusses how nonlinear shear deformations can be modeled in
nonlinear frame analyses. With the inclusion of a simple rule for modeling shear
deformations, the pushover analysis may be extended to structural walls. The results

’ Structural Engineer, KL&A of Colorado, Golden CO 80401.

2 Assistant Prof., Dept. of CEAE, University Colorado, Boulder, CO 80309-0428.
3 Principal, KL&A of Colorado, Golden CO 80401.
1

Copyright ASCE 2004 Structures 2000

Advanced Technology in Structural Engineering
 of a numerical analysis of a column subject to large shear deformations are compared
to the actual experimental test of the column.
Pushover Analysis According to FEMA 273
The pushover analysis, as described by FEMA 273 [l], is supposed to represent a
static approximation of the response a structure will undergo under dynamic
earthquake loading. The static approximation consists of applying a vertical
distribution of monotonically increasing lateral loads to a model which captures the
material non-linearities of the structure. Based on the desired building performance
Downloaded from ascelibrary.org by Istanbul Universitesi on 08/11/13. Copyright ASCE. For personal use only; all rights reserved.

level, the response spectrum for the design earthquake, and the base shear vs. roof
displacement plot, the maximum expected displacement, or target displacement, can
be found. The procedure is briefly outlined in figure 1.

Lateral Loads Building Performance Response Spectrum Target Displacement

- ---.. _“-~““““““---_.“““.~“-“”
Lateral
Loads Roof Displacement

Performance Level Period. T Roof

Figure 1: Pushover Analysis Procedure as Defined by FEMA 273

In addition to the parameters shown in figure 1, the target displacement depends on

the initial stiffness of the structure, Ki, the calculated effective stiffness of the
structure, K,, and the post - yield stiffness of the structure, UK,, which are shown in
figure 2.
Base .

Roof Displacement 6,
Figure 2: Calculation of the Target Displacement, 6 t
Non - Linear Pushover Analysis vs. Non - Linear Dynamic Analysis
In order to understand the advantages and limitations of the pushover analysis,
pushover and full non - linear dynamic analysis were performed on moment resisting
frames of six, twelve, and twenty stories designed according to the UBC code [6].
The resulting target displacement obtained from the pushover analyses can then be
compared with the maximum and minimum roof displacements induced by the

Copyright ASCE 2004 Structures 2000

Advanced Technology in Structural Engineering
 dynamic analyses. The dimensions of each frame are shown in figure 3. Also shown
are the fundamental, second, and third periods of each frame.

~ 361 36;

I
Downloaded from ascelibrary.org by Istanbul Universitesi on 08/11/13. Copyright ASCE. For personal use only; all rights reserved.

20 @
14'
36' 36'

I I I
-4 #JmJL m+
Six Story Frame Twelve Story Frame Twenty Story Frame
Figure 3: Design Frame Dimensions and Periods
The buildings are located in the Los Angeles area, on stiff soil, with 5% structural
damping, and were designed for an earthquake with a 10% probability of exceedance
in 50 years (Life Safety). The response spectrum representing these conditions was
formulated from FEMA 273. For the dynamic loading, the 1940 El Centro ground
motion was considered. Since a comparison is to be made between dynamic and
pushover responses, the ground motion should represent the same conditions as the
response spectrum. The SAC [3] joint venture provides a number of ground motion
records, which are scaled so that the mean response spectrum matches that given by
FEMA 273. The 1940 El Centro ground motion was selected for the analyses.
For the six-story frame, the pushover analysis gives a target displacement of 18.8 in.,
while the dynamic analysis yields a maximum dynamic displacement at the roof of 15
in., as shown in figure 4. The target displacement is a maximum expected
displacement for any earthquake with a 10% exceedance probability in 50 years.
Figure 4 also shows the maximum inter-story drifts for the dynamic analysis and the
displaced shapes at significant response times. The displaced shapes show how the
ground motion excites the second mode too, while the load distribution in the
pushover analysis is enforcing a first mode response. However, the analysis is limited
here to one earthquake only, and different ground motions would most likely have a
different impact on the frame. A complete comparison (which is the object of future
investigations) should include a suite of earthquakes.
For the twelve-story frame, the pushover analysis gives a target displacement of 30.8
in., while the dynamic analysis yields a maximum dynamic displacement at the roof
of 15 in., as shown in figure 5. There is a much larger discrepancy than what was
obtained in the six story frame. Figure 5 also shows the dynamic inter-story drifts
and displaced shapes. It is seen that while the pushover analysis is enforcing a first
mode response in the structure, the dynamic loading is exciting the second and higher
modes too.

Copyright ASCE 2004 Structures 2000

Advanced Technology in Structural Engineering
 Finally, applying the same procedure to the twenty story frame, the pushover analysis
returns a target displacement of 43.3 in., while the dynamic analysis yields a
maximum displacement at the roof of 12 in., as shown in figure 6. Again, the
discrepancy is very large. Figure 6 also shows the dynamic inter-story drifts and
displaced shapes. In this case too, the pushover analysis enforces a first mode
response in the structure while the dynamic loading excites the higher modes too.

Base Shear Pushover Analysis Roof Dynamic Am

(kiP4 v y = 140
Downloaded from ascelibrary.org by Istanbul Universitesi on 08/11/13. Copyright ASCE. For personal use only; all rights reserved.

16(h /

0 5 10 1.5 20 Time
6)
Dynamic Inter-story Drifts Dynamic Displaced Shapes J tl = 4,,4 s
7-

Floor
Level
t6= 6.34 s

-0.5 0 0.5 -20 -10 0 10 20

% of Total Height Displacement(in)
Figure 4: Pushover Analysis and Dynamic Analysis - Six Story Frame
The previous figures seem to indicate that the response of the three buildings is
mostly elastic, since the tails of the responses vibrate around the undeformed
configuration. A more careful look at the results reveals that most of the inelasticity
concentrates in the beams, while no plastic hinges form in the columns. This is due to
a) the fact that the fundamental periods of the three buildings fall into the descending
branch of the 1940 El Centro earthquake response spectrum b) the buildings were
designed according to the UBC code provisions, following the strong column-weak
beam philosophy. Older buildings might have shown a different response pattern.
Finally, only one ground motion was used.
The preliminary conclusion that can be drawn from these results is that for shorter,
stiffer structures, such as the six story frame, the pushover analysis is accurate and
includes expected displacements. However, as higher buildings are considered, the
fundamental period of the structure increases and higher modes are more likely to be
excited. This was illustrated in the twelve and twenty story frame results.
Generalization of the results requires however more analyses. In particular, additional
nonlinear dynamic analyses must be run using different input ground motions.

Copyright ASCE 2004 Structures 2000

Advanced Technology in Structural Engineering
 Roof
Base Shear Pushover Analysis Displacement (in) Dynamic Analysis
o&Is) _ vy=17okips -. 20 Max Disp = 14.2 in

~~~~~~~~~~~ . . 1<

0 5 10 15 20 25 30 35 Roof Disp 5 10 15 20 Time

Downloaded from ascelibrary.org by Istanbul Universitesi on 08/11/13. Copyright ASCE. For personal use only; all rights reserved.

(in) (s)
Dynamic Inter-story Drift Dynamic Displaced Shapes
l5 t1 = 3.10 8
i min t3 t’ t6 max t2 = 3.34s

.
.:
. t6 = 5.56 s
..
. j . .* i t7 = 5.85 s
r ” 1 I
-0.1 -0.05 0 0.05 0.1 -20 -10 0 10 20
% of Total Height Displacement (in)

Figure 5: Pushover Analysis and Dynamic Analysis - Twelve Story Frame

Base Shear Roof Dynamic Analysis

Pushover Analysis
(kiDS) Displacement (in)

0 20 40 60 80 Roof Disp(in) 0 5 10 15 20 Time (S)

Dvnamic Inter-storv Drift Dvnamic DisDlaced ShaDes

tl = 1.68 s
t2 = 2.25 s
Flmr:: [ \N” “’ t3 = 3.06 s
t4 = 3.44 s
Level Level 10 I t5 = 3.74 s
5 I t6 = 4.46 s
t7 = 4.80 s
OC I
,
-0.05 0 0.05 -20 -10 0 10 20
% of Total Height Displacement (in)

Figure 6: Pushover Analysis and Dynamic Analysis - Twenty Story Frame

Copyright ASCE 2004 Structures 2000

Advanced Technology in Structural Engineering
 Benefits of Performing the Pushover Analysis
A question that surfaces when considering the pushover analysis is why should one
perform it when one can do a full dynamic analysis or just design the structure by an
appropriate code such as the UBC. The pushover analysis may be more convenient
than the full dynamic analysis because of computational time. The results shown
above took minutes with the pushover analysis and several hours with the full
dynamic analysis. As long as the structure is stiff and the first mode effects dominate
the response, the pushover analysis is more practical for use in a design office.
Downloaded from ascelibrary.org by Istanbul Universitesi on 08/11/13. Copyright ASCE. For personal use only; all rights reserved.

Another reason for performing a pushover analysis after the structure has been
designed or retrofitted using appropriate codes or design guidelines, is that it yields
additional information on the limit states, the plastic hinge sequence and the force
redistribution caused by a seismic event. The designer can change the design
configuration to obtain a desired plastic hinge sequence under the applied lateral
loads. The pushover analysis also yields detailed member information such as
maximum inter-story drift demands and plastic hinge rotations, thereby increasing the
effectiveness and efficiency of the design. This in turn allows the owners to make a
more informed decision on what kind of design they wish to have for their building.
Modeling Shear Deformations in RC Members
The frame analysis capabilities added to FEAP [5] by Spacone et al [4] include
displacement-based and force-based beam elements with a fiber cross section. The
fibers contribute to the coupled axial and flexural response of the cross section. The
resulting section stiffness is thus full, as indicated in Figure 7a. The simplest way to
add shear deformations to the section model is to add a nonlinear shear force - shear
deformation relationship that is uncoupled from the axial and bending deformations.
Such a constitutive law results in the section stiffness of Figure 7b, where the
diagonal shape of the last two terms indicates decoupling of the shear deformations
from the other deformations. Furthermore, while bending and shear forces are not
related at the section level, implementation of this constitutive law in a force-based
element couples bending and shear forces at the element level through equilibrium.
This implies that if shear failure happens before bending failure at the section level,
because element equilibrium, the element bending moments will be bound by the
element shear forces.
, ‘a b coo
bdeO0
= c e f 0 0
0 0 0 g 0
OOOOh

b)
Figure 7: Section Stiffness with and without Shear Deformation
It can be seen from figure 7, that in order to achieve a relation between shear
force, V, and shear strain, 3: the values g and h must be quantified. This was

Copyright ASCE 2004 Structures 2000

Advanced Technology in Structural Engineering
 accomplished through a shear hysteretic law. The shear law implemented here is a
modified version of Fillipou et al’s [2] model. The law, shown in figure 8, exhibits
the following characteristics. Upon first loading in either direction, the law is bi-
linear with elastic stiffness and strain hardening (A-B-C, A-F-G). Unloading always
takes place parallel to the initial stiffness (C-D, G-H). Upon reloading for the first
time in the opposite direction, there is a stiffness reduction due to the crack closing
(D-E). Upon second loading in the original direction it experiences further reduction
in stiffness as larger cracks must now close (H-I). From that point it loads back to the
last maximum shear strain excursion, at which time it rejoins the strain hardening
Downloaded from ascelibrary.org by Istanbul Universitesi on 08/11/13. Copyright ASCE. For personal use only; all rights reserved.

curve (I-J-K).

t
V, Shear Force 1

Ymax y ,Shear Strain

Figure 8: Shear Hysteretic Law

Accuracy of Shear Element Formulation

To determine the accuracy of the shear element formulation described above, an
analysis was performed on column R-3, which was tested at the University of
California at San Diego [7]. A single force-based element was used for the analyses.
It can be seen from figure 8 that the numerical result compares well with the test data
in predicting the initial yield strength of the column and in predicting its initial
200

150

100
Lateral
Force
(tips)

Top Displacement(in)
Figure 9: Numerical and Test Results of Column R-3

Copyright ASCE 2004 Structures 2000

Advanced Technology in Structural Engineering
 elastic stiffness. However, as the load cycles increase, the numerical result is no
longer accurate. Mainly, the numerical result cannot capture the strength drop due to
severe shear cracking of the column. A more accurate shear constitutive law is
needed in order to capture such a behavior. However, the constitutive law may be
used for pushover analysis of shear walls, and the force-based element accurately
traces yielding of columns when its nominal shear strength is reached.
Conclusions and Future Work
The nonlinear pushover analysis appears to be accurate for stiffer, reinforced concrete
Downloaded from ascelibrary.org by Istanbul Universitesi on 08/11/13. Copyright ASCE. For personal use only; all rights reserved.

structures whose higher modes do not significantly contribute to the overall dynamic
response. Structural walls in short buildings fall into this category and can be
analyzed using this technique rather than more complex nonlinear dynamic analyses.
On the other hand, the response of taller, more flexible frames includes strong
participation from higher models, and this preliminary study seems to indicate that a
nonlinear pushover analysis, which follows mostly the fundamental mode shape,
would not be appropriate. To complete the development of a software tool for
nonlinear pushover analysis of building structures, a simple force-based beam
element with shear deformations was developed and is briefly described in the paper.
A more accurate shear constitutive law would allow extension of the element to
cyclic loads and would yield a better assessment of a member loss of carrying
capacity due to shear failure.
Acknowledgements
This work was sponsored by the Colorado Advanced Software Institute and by
KL&A of Colorado. Their financial support is greatly appreciated.
References
FEMA. (1997). NEHRP guidelines for the seismic rehabilitation of buildings. Report
No. FEMA-273. Washington, D.C.
Filippou, F.C., D’Ambrisi, A., and Issa, A. (1992). “Nonlinear Static and Dynamic
Analysis of Reinforced Concrete Subassemblages” Earthquake Engineering
Research Center, Report No. UCB/EERC-92/08
SAC, Structural Engineers Association of California, Applied Technology Council,
and California Universities for Reasearch in Earthquake Engineering,
http://quiver.eerc.berkeley.edu:8080/index.html
Spacone, E., and Monti, G. “FEAP Users Guide for Frame Analysis - Theory”
University of Colorado at Boulder, Department of Civil, Environmental and
Architectural Engineering, Boulder Co 80309-0428
Taylor, R.L. (1998). FEAP - - A Finite Element Analysis Program. Version 7.1.
Department of Civil and Environmental Engineering, University of California at
Berkeley, Berkeley CA 94720- 17 10
UBC (1997) U nzy orm Building Code, International Conference of Building Officials,
Volume 2
Xiao,Y., Priestley, M.J.N., and Seible, F. (1993). “Steel Jacket Retrofit for Enhancing
Shear Strength of Short Rectangular Reinforced Concrete Columns.” University
of California, San Diego, Structural Systems Research Project, Report No. SSRP-
92107.

Copyright ASCE 2004 Structures 2000

Advanced Technology in Structural Engineering