An Approximate Procedure to Compute the Drift Demands on Frame

Type Structures

S. Akkar, U. Yazgan & P. Gülkan

Department of Civil Engineering, Middle East Technical University,
06531, Ankara, Turkey

ABSTRACT: The accurate calculation of lateral deformation for structural systems is the prime objec-
tive in displacement-based design procedures. This paper describes a simple procedure to estimate the
maximum inter-story drift for regular frame type structures with different beam-to-column stiffness ra-
tios. Given the spectral displacement and beam-to-column stiffness ratio, the procedure calculates the
maximum ground story drift (GSDR) and maximum inter-story drift (MIDR) along the height of the
structure. A total of 148 near-fault ground motions recorded on dense-to-firm soil sites are used for the
statistical evaluation of the procedure. The procedure can be used with confidence for frames with
fundamental periods vary between 0.3s and 2.0s when they are subjected to near-fault records without
pulse. The approximations are in good agreement with the exact response history results of near-fault
records with pulse when the fundamental period to pulse period ratio is less than 2.2.
KEYWORDS: Inter-story drift demand, near-fault ground motions, approximate methods.

1 INTRODUCTION

The estimation of lateral displacements is the essence in displacement-based seismic design. The pre-
diction of local displacement demands may even become more challenging as they can play an impor-
tant role in design and performance evaluation of structures that may be subjected to intricate near-
fault ground motions with high-amplitude, pulse signals. Even if the ground motions do not exhibit
such complex features, inter-story drift values remain one of the primary concerns in design codes.
The local displacement demands induced by earthquake ground motions have been the subject of
research since 1930s. Westergard (1933) described how structural displacements could become critical
under pulse signals. Jennings and Newmark (1960) estimated the lateral displacements of buildings by
using uniform and continuous shear beams. Heidebrecht and Stafford-Smith (1973) derived the lateral
displacement and associated internal stresses by solving the fourth order partial differential equation.
The unexpected local structural failures due to the near-fault ground motions of the 1994 Northridge
and 1995 Kobe earthquakes urged the engineering community to re-examine the relationship between
the local displacement demand and structural capacity. Iwan (1997) proposed the drift spectrum as a
component to the response spectrum to describe the local displacement demand of pulse-like, near-
fault earthquake records. His drift spectrum is based on the solution of damped shear waves propagat-
ing along a shear beam and computes the drift by analogy with the shear strain of a shear beam. Hei-
debrecht and Naumoski (1997) described another way of defining ground story drift using one-
dimensional wave propagation along a shear beam and response spectrum concept. Using the fourth
order differential equation described by Heidebrecht and Stafford-Smith, Miranda (1999) derived an
expression for maximum inter-story drift ratio of general structural behavior (i.e., structures deforming
both in shear and flexure). This solution was limited to uniform lateral stiffness along the building
height but Miranda and Reyes (2002) improved the original proposal to non-uniform lateral stiffness.
Using the shear beam concept together with response history analysis, Chopra and Chintanapakdee
(2001) showed that the drift spectrum proposed by Iwan could be duplicated by using maximum 5
modes. Gülkan and Akkar (2002) found that the shear beam and response spectrum concepts could re-
produce the exact ground story drift within an error bound of ± 10 percent for shear frames with fun-
damental periods less than 2 seconds and when they are subjected to near-fault ground motions.
The objective of this paper is to describe a procedure to estimate the maximum ground story and
maximum inter-story drift for general structural behavior by using the modal analysis concept of frame
structures. The proposed procedure is based on a modification of the equation of Gülkan and Akkar
(2002) that uses the first mode continuous shear beam deformation pattern with spectral displacement.
The improved procedure uses beam-to-column stiffness ratio to account for the general structural be-
havior and modifies the local displacement demands computed by the shear beam behavior. Using a
total of 148 near-fault ground motions, the performance of the new procedure is evaluated for frame
structures with fundamental periods vary between 0.2 s to 2.0 s. In the ground motion data set, 56
near-fault records contain a significant pulse in their waveforms. In order to achieve a verification of
the procedure, the evaluations are conducted separately for ground motions with and without pulse
signals. Error statistics are presented to see how certain near-fault ground motion features influence
the performance of general procedure.

2 BASIC THEORY

2.1 Modal analysis

Equation 1 presents the decoupled equations of motion for a multi-degree-of-freedom (MDOF) system
with mass matrix m
φ iT mι
Y&&i ( t ) + 2ξ i ω i Y&i ( t ) + ω i2 Yi ( t ) = Γi u&&g ( t ); Γi = (1)
φ iT mφ i

In this expression, Yi is the modal coordinate, ξi is the viscous damping ratio, ωi is the radial fre-
quency, and Γi is the modal participation factor of the ith mode. The ground acceleration is designated
by u&&g ( t ) and φi is the ith mode shape vector. The vector ι describes the displacements of the masses
resulting from static application of a unit ground displacement. The elastic response of the system for
mode i is represented by the following expression
u i ( t ) = φi Γi Di ( t ) (2)

The response history Di(t) is computed by solving the equation of motion in (1) without modify-
ing the ground acceleration by Γi. The maximum response for the corresponding mode is computed by
using the spectral displacement value Sd(ωi,ξi). That is
u i ,max ( t ) = φi Γi Di ( t ) max = φi Γi S d ( ω i ,ξ i ) (3)

Equation 3 indicates that given two systems with the same frequency and viscous damping at
mode i, the lateral displacements are directly proportional to the product of φiΓi as Di(t) will be the
same for these two systems. If particular interest is focused on the first mode response, the contribu-
tion to the ground story displacement is represented by
C1 = Γ1φ11 (4)

Similarly, the contribution of the first mode to the top story displacement is
C top = Γ1φ1top (5)

In Equations 4 and 5, φ11 and φ1top are the first mode shape values at the first and top story, respec-
tively. This paper calls C1 and Ctop as ground story displacement and top story displacement contribu-
tion factors, respectively.
2.2 Effect of beam-to-column stiffness ratio on lateral displacement

Beam-to-column stiffness ratio, ρ, is a parameter to define the behavior for frame type building sys-
tems (Blume, 1968). It is the ratio of sum of the beam stiffnesses to column stiffnesses at the story that
is closest to the mid-height of the building, and it is constant for structures that have uniform lateral
stiffness along their height. The general form of ρ is given in Equation 6.

nbay ∑E (I / l)
b b
ρ= beams
(6)
nbay + 1 ∑E (I / l)
columns
c c

The variable nbay represents number of bays and Eb and Ec stand for the elasticity modulus of beam and
column members, respectively. The ratios (I/l)b and (I/l)c are beam and column rigidities, respectively.
The frame behavior is governed by flexure as ρ takes smaller values. The larger values of ρ imply
dominant shear behavior. The smaller ρ values imply columns stiffer than the beams.
1.0 1.0

ρ = 0.001
0.9 0.9
ρ = 0.005
0.8 0.8
ρ = 0.125
0.7 0.7
ρ=4 ρ = 0.001
0.6 0.6
ρ = 0.005
0.5 0.5
ρ = 0.125
0.4 0.4
ρ=4
0.3 0.3

0.2 0.2

0.1 0.1

0.0
0.0
0.0 0.5 1.0 0.0 0.2 0.4 0.6 0.8 1.0
( φ1,n- φ1,n-1)/max( φ1,n- φ1,n-1)
φ1 /φ1,roof

Figure 1. Sensitivity of deformation patterns on ρ for first mode behavior

The variation of ρ has a significant effect on the structural deformation. Figure 1 shows the fun-
damental mode shape and corresponding inter-story drift for frames with different ρ values. These
curves are drawn for non-dimensionless height. The mode shape for the smallest ρ value (ρ = 0.001)
represents cantilever behavior. As ρ grows, the behavior is controlled by both shear and flexural dis-
placements. When ρ goes to infinity, the structure becomes a shear beam. Figure 1 indicates that the
increase in ρ increases the lateral displacement for fundamental mode behavior. The largest difference
between the lateral displacements occurs as ρ takes values between 0.001 and 0.125 (transition from
flexural behavior to mixed mode behavior). The lateral displacement changes more gradually when ρ
takes values greater than 0.125. The inter-story drift for the first mode behavior also changes signifi-
cantly for ρ between 0.001 and 0.125. The maximum inter-story drift migrates from the upper half to
lower half as ρ takes values between 0.001 and 0.125. The migration of maximum inter-story drift to-
wards lower stories is more gradual for ρ > 0.125. In extreme cases, when structure acts as a cantilever
(ρ = 0) or in pure shear (ρ = ∞), the maximum inter-story drift occurs at the top or at the ground story,
respectively whenever the structural behavior is dominated by the first mode.
2.3 Effect of beam-to-column stiffness ratio on ground story and roof displacements

Figure 2 shows the period dependent variation of first mode ground story (C1ρ) and top story (Ctopρ)
displacement contributions for various ρ values. The contributions are normalized by the correspond-
ing shear frame contribution (i.e., C1shear or Ctopshear) to achieve a more representative view for the ef-
fect of ρ on structural displacement parameters of concern. The left frame in Figure 2 shows a strong
dependency of ground story displacement on beam-to-column stiffness ratio. As the structure shifts
from shear frame behavior to a mixed mode that is governed by both shear and flexure, the ground
story displacement becomes smaller (i.e., C1ρ/ C1shear< 1). The top story displacement seems to be less
sensitive to the changes in ρ as it is affected to a lesser extend by the variations in ρ. The top story
displacement tends to increase with respect to shear frame for very small ρ values (ρ ≤ 0.01) when
flexural behavior is very dominant. The highlighted observations suggest that the shear frame behavior
can accurately estimate the top story displacement for frames that have ρ > 0.01. This is not the case
for ground story drift and estimations based on shear frame behavior would probably overestimate the
local ground motion demands as the associated ground story displacements start to take smaller values
for decreasing ρ.

Ctopρ/Ctopshear C1ρ/C1shear
1.2 1.2
ρ = beam-to-column stiffness ratio
ρ = 0.01
ρ=∞
1.0 1.0
ρ=∞
0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

ρ =0.01
ρ = beam-to-column stiffness ratio
0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
T (s) T(s)

Figure 2. The variation of ground story and top story displacements with respect to shear frame for dif-
ferent ρ values

2.4 Effect of beam-to-column stiffness ratio on maximum inter-story displacement

The previous discussion on the first mode lateral displacement shapes indicated how the maximum in-
ter-story drift locations change along the structural height for different beam-to-column stiffness ra-
tios. Figure 3 displays the first mode variation of maximum modal inter-story drift amplifications in
terms of fundamental period and ρ values. The curves show the normalized maximum modal inter-
story drift with respect to the modal maximum ground story drift. This way of presenting the curves
enables one to see how the maximum inter-story drift is modified with respect to ground story drift for
a given ρ value. Inherent from the theoretical first mode behavior, the shear frames do not show a
modification in the maximum inter-story drift relative to maximum ground story drift as the maximum
inter-story drift always occurs at ground story level in this particular case.
 ( φ1,max- φ1,max-1)/φ1,1
6

ρ = 0.01
5

1
ρ= ∞

0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
T(s)

Figure 3. Period dependent amplification of maximum inter-story drift with respect to ground story
drift for different ρ values

2.5 Estimation of maximum ground story and maximum inter-story drift for general frame
behavior

The maximum ground story drift ratio (GSDRsh) of shear frames can be approximated by using Equa-
tion 3. Using the first mode shape of a uniform, continuous shear beam and inserting the correspond-
ing modal participation factor will yield the equation proposed by Gülkan and Akkar (2002).
sin( πh )
GSDRsh = 1.27 S d ( T ,ξ ) (7)
2H

The terms H and h in Equation 7 represent the building height and story height, respectively. Us-
ing the relationship between the building fundamental period and shear wave velocity, c, traveling
along the structure’s height, one can express Equation 7 in an alternative way (Gülkan and Akkar,
2002)
1.27 2π h
GSDRsh = sin S d ( T ,ξ ) (8)
h Tc

Equations 7 or 8 can form the basis for estimating the maximum ground story drift (GSDR) and
maximum inter-story drift (MIDR) ratios. The smooth variation of first mode ground story displace-
ment contribution with respect to shear frame contribution is used to modify Equations 7 and 8 to
compute ρ dependent GSDR. Fitting smooth curves on C1ρ/C1shear presented in Section 2.3 gives the
modification factor as a function of fundamental mode period T and ρ. The curve fitting is done by 2-
stage regression analysis that is based on minimizing the square root of the differences (error) between
the exact variation and the curve fits. Equation 9 gives the expressions for the modifying factor
b( ρ ) 1 1
γ 1 ( ρ ,T ) = a( ρ ) + ; a(ρ) = , b(ρ) = (9)
T 1 + 0.35 / ρ 0.65 8 + 25 ρ 0.4
The term γ1 is the modification factor that corrects the shear frame ground story drift to a more
general, ρ-based ground story drift demand expression that is presented in Equation (10).
GSDR( T ,ξ , ρ ) = γ 1 ( ρ ,T )GSDRsh ( T ,ξ ) (10)
 Note that the function γ1 becomes 1 as ρ tends to large values satisfying the boundary condition
for shear frame behavior.
The maximum inter-story drift ratio for general structural behavior is estimated through (10) and
the plots presented in Section 2.4 for the first mode variation of maximum inter-story drift with respect
to the maximum ground story drift. As in the case of Equation 9, 2-stage regression analysis is con-
ducted to fit curves on these plots to represent the second modification factor, γ2 that accounts for the
variation of maximum inter-story drift with respect to ground story drift. The function γ2 should be
taken as 1 whenever it gives values less than 1 for a particular T and ρ. Equation 11 shows the results
of the regression analysis
d( ρ )
( c ( ρ )− ) 1 0.07
γ 2 ( ρ ,T ) = e T ; c(ρ) = , d(ρ) = (11)
2ρ + 1 ρ 0.25

The modifying factor γ2 is equal to 1 as ρ goes to infinity. This satisfies the theoretical boundary
conditions between the maximum inter-story drift and ground story drift for shear frames. Equation 12
gives the expression of maximum inter-story drift ratio for general structural behavior
MIDR( T ,ξ , ρ ) = γ 1 ( ρ ,T )γ 2 ( ρ ,T )GSDR sh ( T ,ξ ) (12)

The expressions in (10) and (12) are capable of representing the exact variation of GSDR and
MIDR for ρ ≥ 0.125 that is the practical range of concern in most frame type structures (Blume, 1968)

3 EVALUATION OF THE PROPOSED PROCEDURE

A total of 148 near-fault ground motions were used in the evaluation of the equations proposed for es-
timating the GSDR and MIDR. The near-fault ground motions are records from soil site classes that
have shear wave velocities between 180 m/s and 760 m/s in the upper 30 m layer. The closest site-to-
fault distances of these records vary from 0.1 km to 20 km. The ground motion data set was divided
into two groups as records with and without pulse signals. The velocity waveforms were chosen to
identify the pulse signals, as it is easier to detect the pulse from the variation in velocity time series. A
total of 56 near-fault records contain pulse signals in the chosen data set. The remaining near-fault re-
cords do not exhibit a dominant pulse in their waveforms.
Response history analyses using these records were conducted for a group of single bay frame
models. A total of 19 generic frame models were used with fundamental mode periods ranging from
0.2 s to 2.0 s. The fundamental periods of the models satisfy T=0.1n relationship where n designates
the total number of stories. This fundamental period relationship corresponds to a shear wave velocity
of 120 m/s when the story height, h, is taken as 3 m. The beam-to-column stiffness ratio (ρ) for the
models were varied as ρ = 0.125, 0.25, 0.5, 0.75, 1.0, 1.5, 2.0, 3.0, 4.0, ∞. In this way, the evaluation
of the proposed procedure was established by a total number of 190 (19×10) frame models. The num-
ber of response history analyses conducted for the evaluation stage was 28120 (19×10×148).

3.1 Effect of near-fault ground motions on MDOF system response

The near-fault ground motions with forward directivity effect contain a high-amplitude, pulse signal in
the direction normal to the fault due to the alignment of fault rupture and wave propagation towards
the site. Along the fault parallel component, the forward directivity effect results fling pulse in the ac-
celeration trace due to the shear dislocation of the fault. As the fault rupture and wave propagation
move along opposite directions, the polarization of SH waves changes and the pulse effect is lost. This
is called backward directivity and near-fault ground motions that possess backward directivity have
broadband waveforms without any dominant pulse signal.
Figure 4 plots the maximum inter-story drift and ground story drift ratio scatters for the 56 near-
fault records with pulse signals in their velocity time histories. At each response history run, the fun-
damental period of the model was normalized with respect to the pulse period of the excitation in order
to display the pulse signal effect on the local displacement demand parameters (i.e., MIDR and
GSDR). The scatter data shows a significant amplification when the fundamental periods of the frame
systems are closer to the pulse signal period (i.e. T/Tp ≈ 1). As the structure’s fundamental period
moves away from the pulse period of the records, the amplifications both in the maximum inter-story
drift and ground story drift ratios fall down. Figure 5 shows similar scatter diagrams for the near-fault
data group that does not contain pulse signals. The scatter plots do not show a clear trend for this case
indicating that the MIDR and GSDR values vary randomly from record-to-record.
The near-fault records with pulse signals exhibit a strong amplification in the pseudo-velocity
spectrum in the vicinity of their pulse periods (Yazgan, 2003). The proposed expressions for MIDR
and GSDR will be sensitive to the pulse demand as the spectral displacement term in (10) and (12) is
directly related to pseudo-velocity spectrum. Therefore, the local displacement demand estimates of
the proposed procedure would be reasonable, as long as the first mode behavior governs.

GSDR MIDR
56 records with pulse 56 records with pulse
0.05 0.05

0.04 0.04

0.03 0.03

0.02 0.02

0.01 0.01

0.00 0.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
T/Tp T/Tp

Figure 4. Scatter diagrams for GSDR and MIDR computed from response history analyses when frame models
were subjected to pulse signals

GSDR MIDR
92 records without pulse
0.05 0.05 92 records without pulse

0.04 0.04

0.03 0.03

0.02 0.02

0.01 0.01

0.00 0.00
0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0
T (s) T (s)

Figure 5. Scatter diagrams for GSDR and MIDR computed from response history analyses when frame models
were subjected to excitations without pulses
3.2 Fundamental mode response in capturing the local displacement demand

Figure 6 shows the mean ratios of first mode response to exact response (considering all modes) for
MIDR and GSDR when the models were subjected to 56 pulse records. Mean ratios less than 1 indicate
that on average, the first mode response is unable to capture the exact response. The curves are drawn
for different ρ values and fundamental periods are normalized by the pulse signal period as in the case
of Figure 4. The plots in Figure 6 show that on average, the variation of ρ does not have an affect on
first mode response in estimating the MIDR or GSDR. The curves attain almost the same values re-
gardless of the changes in ρ. The curves also suggest that the first mode response captures the exact
response fairly well for fundamental periods that are slightly higher than twice of the pulse period. Af-
ter T/Tp > 2.2, the higher mode effects reduce the performance of first mode in representing the exact
response. The underestimations of exact response may reach to –20 percent for T/Tp > 2.6 due to the
employment of first mode response. Alavi and Krawinkler (2001) described this phenomenon for
MDOF behavior under pulse excitation. They stated that the first mode behavior could represent the
MDOF response when the building fundamental period is less than the pulse period. The present study
not only confirms this but also defines certain T/Tp limits for a reliable representation of local dis-
placement demands after examining a large number of near fault records. Figure 7 shows similar com-
parisons for the near-fault ground motion data that does not contain any significant pulse. The mean
variation of the curves for different ρ indicates that the first mode response can capture the local dis-
placement demands with an error of –5 percent for fundamental periods less than 1.5s. The higher
mode effects start to dominate displacements after T > 1.5s and the first mode can represent the exact
response within an error of -5 to -10 percent.
These observations point that the first mode contribution gives fairly good estimates for the exact
response for a relatively wide range of structural periods that are usually the concern in engineering
practice. Caution should be exercised as the unrealistic use of first mode behavior may cause signifi-
cant errors for long periods especially when buildings are subjected to near-fault records with pulse ef-
fects.
E[GSDR1/GSDRAll] 56 records with pulse E[MIDR1/MIDRAll] 56 records with pulse
1.4 1.4
ρ = 0.125 ... ∞ ρ = 0.125 ... ∞

1.2 1.2

ρ = 0.125 ρ = 0.125

1 1

0.8 0.8
ρ=∞ ρ=∞

0.6 0.6
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
T/Tp T/Tp

Figure 6. Adequacy of first mode in capturing the exact response of records that exhibit pulse signals
in their waveforms
E[GSDR1/GSDRAll] 92 Records without pulse 92 Records without pulse
E[MIDR1/MIDRAll]
1.4 1.4
ρ ρ==0.125
0.125......∞∞ ρ = 0.125 ... ∞

1.2 1.2

1 1

0.8 0.8

0.6 0.6
0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0
T (s) T (s)

Figure 7. Adequacy of first mode in capturing the exact response of records that have no significant
pulse signals

3.3 Verification of the proposed procedure

The estimations of GSDR and MIDR were evaluated by defining the following error term
∆ app
ETi ,ρ j = (13)
∆ exact

In Equation 13, ∆app designates either the approximate ground story or maximum inter-story drift
ratios computed from (10) or (12), respectively. The variable ∆exact is the corresponding value resulting
from response history analysis. Given a ground motion record, m, the error term ETi ,ρ j is computed for
the fundamental period Ti and beam-to-column stiffness ratio ρj. The mean error for pre-defined Ti and
ρj is calculated by

∑ (E ρ )
nm
1
E Ti ,ρ j = Ti , (14)
nm m =1
j m

A mean error value smaller than 1 will indicate that on average, the approximation yields values
less than the exact value computed from the response history analyses. A mean error that is larger than
1 will show that on average, the estimates are on the conservative side. The mean error equal to 1 will
point that the mean estimations duplicate the exact mean variation precisely.
Figure 8 shows the mean error trends in GSDR and MIDR estimations for ρ values ranging from
0.125 to ∞ and fundamental periods normalized with respect to the pulse period. On average the ap-
proximate method estimates safe GSDR and MIDR values for T/Tp ≤ 2.2. The errors for MIDR are
more sensitive to the changes in ρ. As beam-to-column stiffness ratios take larger values, the mean es-
timates tend towards the unsafe side. For T//Tp > 2.2, the unsafe errors can be as large as –20 percent.
On average, the approximate equations can predict the 80 percent of the mean exact response. The
mean error values of MIDR are more sensitive to the variations in ρ indicating that dispersion on the
mean estimations is larger with respect to GSDR.
Figure 9 presents a similar evaluation for the near-fault without a pulse in their velocity wave-
forms. The plots show that for fundamental periods between 0.3 s and 2.0 s, the approximate methods
estimate the mean variation of MIDR and GSDR within an error limit of 15 percent to –10 percent.
This indicates that on average, the procedure yields approximate GSDR and MIDR values that are 15
percent larger or 10 percent smaller than the mean exact response. The estimates are better for the
GSDR values. Similar to the pulse signal performance, the estimates of MIDR are more sensitive to the
variation in ρ and they present more dispersion when compared to GSDR estimates. The mean errors
tend to shift towards unsafe side as ρ and T increase. The dominance of higher mode effects in the
long periods is the primary cause of the unsafe predictions.
The trends in mean error statistics of Figures 8 and 9 are very similar to the plots of Figures 6 and
7, respectively that show the adequacy of first mode response in capturing the exact variation of local
displacements. This can be interpreted as the fairly good representation of the first mode response his-
tory results for GSDR and MSDR through Equations 10 and 12.
Error Error
Mean Error for GSDR Mean Error for MIDR
1.4 1.4
ρ = 0.125 ρ = 0.25 ρ = 0.125 ρ = 0.25
ρ = 0.50 ρ = 0.75 ρ = 0.50 ρ = 0.75
ρ = 1.0 ρ = 1.5 ρ = 1.0 ρ = 1.5
ρ = 2.0 ρ = 3.0 ρ = 2.0 ρ = 3.0
1.2 1.2 ρ = 4.0 ρ=∞
ρ = 4.0 ρ=∞

1.0 1.0

0.8 0.8

0.6 0.6
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
T/Tp T/Tp

Figure 8. Mean error statistics for the near-fault records that have pulse signals

Error Mean Error for GSDR Error Mean Error for MIDR
1.4 1.4
ρ = 0.125 ρ = 0.25
ρ=0.125 ρ=0.25
ρ = 0.50 ρ = 0.75
ρ=0.50 ρ=0.75
ρ = 1.0 ρ = 1.5
ρ=1.0 ρ=1.5
1.2 ρ = 2.0 ρ = 3.0
1.2 ρ=2.0 ρ=3.0
ρ = 4.0 ρ=∞
ρ=4 ρ=∞

1.0 1.0

0.8 0.8

0.6 0.6
0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0
T (s) T (s)

Figure 9. Mean error statistics for the near-fault records that do not exhibit pulse signals in their wave-
forms

4 SUMMARY AND CONCLUSIONS

Two simple expressions for estimating the ground story and maximum inter-story drift ratios are pre-
sented. These expressions modify the drift expression of Gülkan and Akkar (2002) by considering the
beam-to column stiffness ratio. The theoretical fundamental mode behavior of frame structures is used
in the derivation of expressions. A total of 148 near-fault ground motions recorded on firm to dense
soil sites are using for the evaluation of proposed expressions. Of the near-fault ground motion data set
56 records contain significant pulses in their waveforms. The evaluations are done separately for
ground motion records with and without pulse signals. The following observations and conclusion are
derived at the end of the study:

1. The ground story drift is sensitive to the variation in beam to column stiffness (ρ) ratio. Esti-
mations made by shear beam behavior (i.e. ρ = ∞) may produce very conservative results for
ground story displacement of frame structures that undergo lateral deformations both in shear
and flexure.
2. The pulse signal has a significant effect on the ground story (GSDR) and maximum inter-story
drift (MIDR) demands. These local displacement demand parameters tend to amplify when the
fundamental period (T) of the structure takes values closer to the pulse period (Tp) of the exci-
tation.
3. The first mode behavior can represent the exact GSDR and MIDR variation accurately for T/Tp
values less than 2.2. The higher mode effects are dominant for T/Tp > 2.2. In the case of near-
fault records without pulse, the first mode can represent the actual GSDR and MIDR demand
fairly good for T < 2.0s.
4. When only near-fault records with pulse signals are considered, the proposed expressions ap-
proximate the GSDR and MIDR demand within an error limit of ± 15 percent with respect to
the mean variation of exact response for T/Tp < 2.2.
5. Under the near-fault records that do not exhibit pulse signals, these approximations produce
errors of ± 15 percent with respect to the mean variation of exact response for a fundamental
period range of 0.3s to 2.0s.

REFERENCES

Alavi, B. and H. Krawinkler, (2001), “Effects of near-fault ground motions on frame structures,” The John A.
Blume Earthquake Enginering Center Report No. 138, Stanford University, Calfornia, February.
Blume, J.A., (1968) “Dynamic charactristics of multi-story buildings,” Journal of Structural Division, ASCE,
94, pp. 377-402.
Chopra, A.K. and C. Chintanapakdee, (2001) “Drift spectrum vs. modal analysis of structural response to near-
fault ground motions,” Earthquake Spectra, 17, pp. 221-234.
Gülkan, P. and S. Akkar, (2002) “A simple replacement for the drift spectrum,” Engineering Structures, 24, pp.
1477-1484.
Heidebrecht, A.C. and B. Stafford-Smith, (1973) “Approximate analysis of tall wall-frame structures,” Journal
of Structural Engineering, ASCE, 99, pp. 199-221.
Heidebrecht, A.C. and N.D. Naumoski, (1997) “Development and application of a displacement-based design
approach for moment-resisting frame structures,” Seismic Design Methodologies for the Next Generation
of Codes, Fajfar P. and Krawinkler H. eds., Balkema: Roterdam, pp. 217-228
Jennings, R.L. and N.M. Newmark, (1960) “Elastic response of multi-story shear beam type structures subjected
to strong ground motion,” Proc., Second World Conf. on Earthquake Engineering, Tokyo, Japan, Vol. II,
pp. 699-717.
Iwan, W.D. (1997) “The drift spectrum: a measure of demand for earthquake ground motions,” Journal of Struc-
tural Engineering, ASCE, 123, pp. 397–404.
Miranda, E., (1999) “Approximate seismic lateral deformation demands in multistory buildings,” Journal of
Structural Enginering, 125, ASCE, pp. 417-425.
Miranda, E. and C.J. Reyes, (2002) “Approximate lateral drift demands in multi-story buildings with non-
uniform stiffness,” Journal of Structural Enginering, ASCE, 128, pp. 840-849.
Yazgan, U., (2003) “A methodology for determination of performance baed design parameters,” Master Thesis,
Department of Civil Engineering, Middle East Technical University, Ankara Turkey, September.