You are on page 1of 10

PRACTICAL MODELING FOR NONLINEAR SEISMIC RESPONSE OF RC WALL

STRUCTURES

A. Lepage 1 , S. L. Neuman 2 , and J. J. Dragovich 3

ABSTRACT

A simplified analytical model is proposed for modeling the nonlinear response of


flexural-yielding reinforced concrete walls using standard structural analysis
software. The program SAP2000 is used to implement the proposed model for
evaluating structural response by means of user-defined nonlinear response
history analysis. The model is useful for performing practical nonlinear static or
nonlinear dynamic procedures. The use of the model is illustrated by its
application to two structures previously tested in the laboratory.

The walls are modeled using a fine mesh of linear-response shell elements
coupled with uniaxial line elements. The use of line elements allows one to
invoke the typical nonlinear response parameters available for such elements.
The axial stiffness of the shell elements is gradually transferred to and from the
line elements using stiffness modifiers between 0 and 1 at the expected plastic
hinge region and its vicinity. The nonlinear model assigned to the line element
corresponds to a multilinear plasticity model. In this type of model the nonlinear
force-deformation relationship is given by a multilinear curve defined by a set of
points that need not be symmetrical with respect to the origin. The first slope of
the force-deformation curve on either side of the origin defines the range of linear
elastic response. The remaining segments define plastic deformations.

The experimental data used to validate the proposed analytical model show
agreement with the calculated response. The model is capable of capturing with
reasonable accuracy the main response parameters of the wall structures: initial
stiffness, onset of yielding, and yield strength. Additionally, the measured
displacement response waveforms as well as the amplitudes are reasonably
matched by the calculated values during the duration of strong base motion.

1
Assistant Professor, The Pennsylvania State University, University Park, lepage@psu.edu
2
Project Engineer, KPFF Consulting Engineers, Seattle, scott.neuman@kpff.com
3
Assistant Professor, Seattle University, jeffdrag@seattleu.edu
Introduction

In the seismic evaluation of existing buildings as well as in the design of new buildings
there are situations where special peer reviews of the seismic force resisting system require
advanced analysis using either the static nonlinear or the dynamic nonlinear procedure. The
main objective of this paper is to present a nonlinear modeling technique easily adaptable to
various wall configurations commonly found in practice.
A simplified model is proposed for representing the nonlinear response of flexural-
yielding reinforced concrete walls using standard structural analysis software. In particular, the
computer program SAP2000 (CSI, 2005) is used to implement the proposed model.
The use of the model is illustrated by its application to two structures previously tested in the
laboratory: Structure AI (Adebar and Ibrahim, 2002) and Structure M1 (Aristizabal and Sozen,
1976). The experimental data are used to validate the proposed model.

Test Structure AI

Description

A large scale test was conducted in the University of British Columbia (Adebar and
Ibrahim, 2002) on a reinforced concrete structural wall representing a prototype wall with a
height-to-length ratio of nearly 10. The test model is 1/4 scale and is 12.8 m (42 ft) high. The
lateral load in the test specimen is applied at a height 12.2 m (40 ft) from the wall base.
The wall is reinforced so that the critical section for flexural yielding is located at 457
mm (1.5 ft) away from the base. Typical details of the wall geometry and vertical reinforcement
are shown in Fig. 1. The vertical reinforcement had actual yield strength, fy, of 455 MPa (66 ksi)
and the concrete had cylinder compressive strength, f’c, of 49 MPa (7.1 ksi).

Figure 1. Cross section of wall (Structure AI)

During the test the wall was subjected to a constant uniform axial load of 1,500 kN (337
kip) applied at the top of the wall by means of external post-tensioning with negligible P-Delta
effects. The axial load represents gravity loads in the order of 0.1 f’c Ag. The lateral load is
applied using a displacement controlled hydraulic actuator with four full cycles of loading at
each target displacement level. The maximum lateral displacement, measured at 12.2 m (40 ft)
from the wall base, approached 200 mm (8 in) and corresponds to a mean drift ratio of nearly
1.6%. For a more detailed description of the experimental setup refer to Adebar and Ibrahim
(2002).
Proposed Analytical Model
A computer model representing the wall specimen is developed using SAP2000 (CSI, 2005).
The wall is modeled using six shell elements along the wall cross section with a single shell element
used for each of the 381x203 mm (15x8 in) end piers and four shell elements for the wall web, see
Fig. 2(a). The typical longitudinal dimension of the shell element is 305 mm (1 ft) except near the
base and near the point of load application where 152-mm (6-in) segments were used. At the base,
the shell elements were restrained with translational supports.
Special modeling assumptions were used near the critical section located at 457 mm (1.5 ft)
away from the wall base, where flexural yielding was expected to occur. In this region, see Fig. 2(b),
uniaxial line elements are introduced longitudinally in between the shell element nodes. The use of
line elements allows one to invoke the nonlinear analysis capabilities of SAP2000 using such
elements. To activate the line elements, the axial stiffness of the neighboring shell elements is
modified by a factor between 0 and 1. A transition region is used to gradually transfer the axial
stiffness between the shell and line elements. The line elements invoking nonlinear response extend
for a distance of 1.52 m (5 ft), approximately equal to the length of the wall.
The nonlinear model assigned to the uniaxial line element corresponds to a multilinear
kinematic plasticity model. In this type of model the nonlinear force-deformation relationship is
given by a multilinear curve defined by a set of points. The first slope on either side of the origin
defines the range of linear elastic response. The remaining segments define plastic deformations.
The slope derived from the last two points specified for a given direction of loading is extrapolated to
infinite deformation. The force-deformation curve need not be symmetrical with respect to the
origin.
To represent compression response, the force-deformation curve characterizing the line
element is derived using the tributary area of concrete. The compressive strength is defined by
multiplying area of concrete by f’c and the deformation is obtained from L f’c /Ec, where L is the
length of the element and Ec = 4700√f’c in MPa (57,000√f’c in psi). Post-yield compression stiffness
is set to 0.1% of the initial stiffness, nearly flat.
To represent tension response, the yield force is defined by multiplying fy by the tributary
area of steel. The deformation is derived by setting the slope equal to the initial slope used for
compression loading. Identical initial slope for both compression and tension loading is convenient
in order to characterize the initial overall stiffness of the wall as a function of gross section properties
or a desired fraction of it. Post-yield tension stiffness is set to 2% of the stiffness associated with the
area of steel only. Thus, a steel stress of approximately 1.1fy is attained for a strain ductility of 6.
The properties supporting the definitions of the line elements with nonlinear characterization
are included in Table 1. The structural model is preloaded with dead loads and the resulting state of
stress defines the starting conditions for the nonlinear static analysis.

Calculated Response
Results obtained from the nonlinear response analysis using the proposed analytical model
are compared with the measured response in Fig. 3. The experimental data corresponds to the peak
measured response at each target displacement level for both positive and negative direction of
loading. Force and deformation are measured at the point of load application. Satisfactory
agreement between the analytical and experimental data is observed.
(a) Complete model (b) Detail

Figure 2. Model description (Structure AI)

Table 1 – Properties of uniaxial nonlinear elements (Structure AI)


Compression Yield Tension Yield
Element Concrete Area Steel Area Force Displ. Force Displ.
mm2 mm2 kN mm kN mm
A 38,710 274 1,898 0.451 124.5 0.0296
B 58,060 276 2,846 0.451 125.7 0.0199
C 38,710 100 1,898 0.451 45.5 0.0108
200

160

Force (kN) 120

80

Calculated
40
Measured (- )
Measured (+ )
0
0 40 80 120 160 200
Displacem ent (m m )

Figure 3. Comparison of measured and calculated response (Structure AI)

Test Structure M1

Description

As part of a series of investigations that focused on the response of reinforced concrete


structures to earthquake motions, Aristizabal and Sozen (1976) performed tests using the
University of Illinois Earthquake Simulator, an experimental facility designed to subject small-
scale structures to laboratory base motions in one horizontal direction. Test Structure M1
comprised two planar systems of two walls with 25x178 mm (1x7 in) cross section, connected at
each level by beams spanning 101 mm (4 in), see Fig. 4. The size of the beam was the same at
all levels, with nominal dimensions of 25x38 mm for a span to depth ratio of 2.7.
The concrete used was small-aggregate concrete with measured compressive strength of
32.8 MPa (4.75 ksi) and a modulus of elasticity of 21.4 GPa (3,100 ksi). The steel used as
flexural reinforcement in walls and beams had a yield stress of 72 ksi and a Young’s modulus of
213 GPa (30,800 ksi). The amount of steel reinforcement in beams and walls is shown in Fig. 5
and 6. Transverse reinforcement was provided so that shear failures were precluded prior to the
development of flexural yielding. To improve the ductility of the first three level walls, the
longitudinal steel was confined by spiral reinforcement.
The ten-story coupled wall system had an effective weight at each floor level of 4.45 kN
(1,000 lbs) distributed equally to the walls without loading the beams. The test specimen had
heavy base girders bolted to the test platform of the earthquake simulator. The specimen-to-
simulator connection was designed to simulate a fixed-base condition for the test structure.
The planar structural units were mounted parallel to the direction of platform motion.
The base motion, patterned after the NS component of the 1940 Imperial Valley earthquake
(recorded at El Centro, California), used a time-scale compression factor of 2.5 in order to obtain
a relation comparable to conditions for a full scale building between the natural frequency of the
test structure and the frequency content of the earthquake record.
(a) Elevation (b) Section through beams
Figure 4. Dimensions of Structure M1

Figure 5. Cross section of wall, base to level 3 (Structure M1)

(a) Wall-beam connection (b) Cross section of typical beam


Figure 6. Beam reinforcement (Structure M1)
Proposed Analytical Model
The analytical model proposed for Structure AI is also adopted in the modeling of Structure
M1 with minor adjustments. Eight shell elements along each wall cross section are chosen for
Structure M1. Special modeling assumptions are used for the critical section near the wall base
where yielding is expected to occur. In this region, see Fig. 7, uniaxial line elements capable of
responding nonlinearly are introduced in between the shell element nodes. The stiffness of the shell
elements is modified in the vertical direction only and is set to zero in the expected hinge region
where the line elements represent all vertical stiffness in the wall. A transition region is used to
gradually transfer the vertical stiffness between the shell and line elements. The line elements
invoking nonlinear response extend above the fixed base a distance 1.25 times the wall length. Table
2 presents a summary of the uniaxial nonlinear element properties used to represent the wall hinge
region. Post-yield stiffness is taken as 0.5% of the initial stiffness, for tension and compression.
The beams are modeled with zero-length nonlinear links located at midspan between rigid
segments that extend to the wall centerline. Sources of linear and nonlinear actions in the beams are
condensed at the zero-length link into shear forces and vertical displacements accounting for flexural
and shear deformations. The nonlinear response of the beams is represented by the Wen (1976)
model with a yield force equal to 2M/L, where M is the flexural strength and L is the clear span. The
yield force of the beams was calculated to be 6.8 kN (1.5 kip). The beam post-yield stiffness is
defined as 0.5% of the initial stiffness.
The effects of axial loads induced by gravity loads are directly accounted for by preloading
the structural model before processing the nonlinear response history analysis. The analysis also
includes second order (P-Δ) effects. All masses are lumped at the beam level and constrained by
means of a rigid diaphragm. The nonlinear response is calculated using the method by Wilson
(1985) and includes participation of all ten translational modes of vibration.

Calculated Response
Using the proposed modeling technique, the top-displacement response is calculated and
compared to the experimental data. Figure 8 shows the results for various modeling assumptions.
Model 1 is a linear-response model with initial stiffness in beams and walls based on 0.5 times the
gross-section moment of inertia (Ig). This model follows the general recommendations found in
FEMA 356 (2000). The reduction of Ig in the walls and beams are conveniently approximated by
modifying only the vertical stiffness of all shell and line elements by a stiffness modifier equal to the
target fraction of Ig. Model 2 is the nonlinear version of Model 1 and uses multilinear kinematic
plasticity property for the nonlinear elements at the wall base and the Wen plasticity property for the
nonlinear zero-length link representing the beams. In Model 3, the beam initial stiffness is reduced
to 0.2Ig similar to the recommendations for coupling beams in Paulay and Priestley (1992) and NZS
3101 (1995). Model 4 uses 0.2Ig in beams and 0.3Ig in all wall segments similar to the
recommendations by Fenwick and Bull (2000). Model 5 is identical to Model 4 except for the use of
multilinear Takeda plasticity property (Takeda et al., 1970) for nonlinear elements in lieu of the
kinematic model.
Figure 8 reveals that Models 4 and 5 are capable of reproducing the measured displacement
waveform up to the instance of maximum displacement. After the peak displacement, the analytical
models cannot replicate the stiffness reduction of the actual test specimen caused by the observed
crushing and spalling at the exterior edges of the walls. The apparent period of vibration after 2.5
seconds is lower for the calculated waveform.
Figure 7. Model description (Structure M1)

Table 2 – Properties of uniaxial nonlinear elements (Structure M1)


Compression Yield Tension Yield
Element Concrete Area Steel Area Force Displ.1 Force Displ.1
mm2 mm2 kN mm kN mm

A 161 0 5.3 0.0876 -2 -2


B 323 26.6 10.6 0.0876 13.20 0.1094
C 484 26.6 15.8 0.0876 13.20 0.0729
D 806 0 26.4 0.0876 - -
E 968 0 31.7 0.0876 - -
1
Displacement values correspond to section property modifiers of 1.0.
2
For cases of zero steel area, tension response is linear elastic with negligible stiffness.
60
Model 1: LINEAR (0.5B,0.5W)
30

- 30

- 60
60
Model 2: KINEMATIC (0.5B,0.5W)
30

- 30

- 60
60
Model 3: KINEMATIC (0.2B,0.5W)
Di sp l acement , mm

30

- 30

- 60
60
Model 4: KINEMATIC (0.2B,0.3W)
30

- 30

- 60
60
Model 5: TAKEDA (0.2B,0.3W)
30

- 30 Measured
Calculat ed
- 60
0 1 2 3 4 5
Time, sec

Figure 8. Comparison of calculated and measured top-displacement response


Conclusions

A simple analytical model for calculating the nonlinear response of flexural-yielding


reinforced concrete walls was validated using experimental data. The model was satisfactory for
capturing the main response parameters: initial stiffness, onset of yielding, yield strength, and
displacement response when subjected to strong earthquake motions.
The proposed model may be easily implemented in many commercially available
structural analysis programs with basic nonlinear elements. Additionally, the model may be used
for a more accurate representation of various wall configurations in lieu of stick-model
oversimplifications.

References

Adebar, P., and A. M. M. Ibrahim, 2002. Simple Nonlinear Flexural Stiffness Model for Concrete
Structural Walls, Earthquake Spectra, 18(3), 407-426.

Aristizabal-Ochoa, J. D., and M. A. Sozen, 1976. Behavior of Ten-Story Reinforced Concrete Walls
Subjected to Earthquake Motions, Structural Research Series No. 431, Civil Engineering Studies,
University of Illinois, Urbana, Illinois.

CSI, 2005. SAP2000: Static and Dynamic Finite Element Analysis of Structures, Advanced 9.1.4,
Computers and Structures, Inc., Berkeley, California.

FEMA 356, 2000. Prestandard and Commentary for the Seismic Rehabilitation of Buildings, American
Society of Civil Engineers, Reston, Virginia.

Fenwick R., and D. Bull, 2000. What is the stiffness of reinforced concrete walls?, SESOC Journal, 13
(2), 23-32.

NZS 3101, 1995. Concrete Structures Standard, Part 1: Code, Part 2: Commentary, Standards New
Zealand, Wellington.

Paulay T., and M. J. N. Priestley, 1992. Seismic Design of Reinforced Concrete and Masonry Buildings,
John Wiley & Sons, 744 p.

Takeda, T., M. A. Sozen, and N. N. Nielsen, 1970. Reinforced Concrete Response to Simulated
Earthquakes, Journal of the Structural Division, ASCE, 96(ST12), 2557-2573.

Wen, Y. K., 1976. Method for Random Vibration of Hysteretic Systems, Journal of the Engineering
Mechanics Division, ASCE, 102(EM2), 249-263.

Wilson, E. L., 1985. A New Method of Dynamic Analysis for Linear and Nonlinear Systems, Finite
Elements in Analysis and Design, Elsevier Science, 1(1), 21-23.

You might also like