Anda di halaman 1dari 32

THE STRUCTURAL DESIGN OF TALL BUILDINGS, Vol.

5, 15 1 - 182 (1 996)

ANALYSIS OF THE NORTHRIDGE EARTHQUAKE


RESPONSE OF A DAMAGED NON-DUCTILE
CONCRETE FRAME BUILDING

M. SAIFUL ISLAM?
Nabih Yousszf & Associates, 800 Wilshire Blud, Suite 510, Los Angeles, CA 90017, 7J.S.A

SUMMARY
The response of a seven-story reinforced concrete building recorded during the Northridge earthquake is
analysed in this paper. The building was designed in 1965 to the lateral force requirements of 1964 Los
Angeles City Building Code. Non-ductile concrete moment frames and interior slab-column frames form
the lateral resisting system of the building. The building suffered severe damage during the Northridge
earthquake and was red tagged. This building was instrumented by as many as 16 sensors during the
earthquake. These accelerogram records were analysed using system identification techniques to obtain
important building response information. This also provided the opportunity to see if certain analytical
techniques that are commonly used by practising engineers and/or researchers and many of which are
currently being incorporated in various upcoming documents dealing with seismic evaluation of existing
buildings could have predicted the observed performance of the building.

1. INTRODUCTION
The 17 January 1994 Northridge earthquake provided an opportunity for structural engineers
to evaluate the performance of buildings, in general, and existing concrete buildings in particular.
Many of these concrete buildings were designed prior to 1976 and lack the strength and ductility
that we have come to expect in modern buildings. Although many of these sustained damage,
some even severe, very few actually collapsed. This paper analyses the response of one such
concrete building which sustained severe damage during the Northridge earthquake.
The building analysed herein is unique in that it was heavily instrumented during the
Northridge earthquake and even had limited instrumentation at the time of the San Fernando
Earthquake of 1971. As many as 16 sensors recorded the ground motion and the response of
the building during the Northridge earthquake. Analyses of these accelerogram records provided
important building response information including actual periods of vibration, story drifts, roof
displacement, torsional response, distribution of inertial forces, and story shear. This provided
a unique opportunity to test a number of analytical techniques/tools, both linear and nonlinear,
some of which are currently being incorporated in various upcoming documents dealing with
seismic rehabilitation of existing buildings (e.g. ATC 33.03, Proposition 122/ATC 401).This
paper will present these results and identify shortcomings and pitfalls of some of these analytical
tools.

t Project Manager.

CCC 1062-8002/96/030151-32 Received March 1992


0 1996 by John Wiley & Sons, Ltd. Revised M a y 1996
152 M. SAIFUL ISLAM

2. BUILDING DESCRIPTION
2.1. General
The Holiday Inn is a seven-story reinforced concrete structure located in Van Nuys, just east
of the San Diego freeway at Roscoe Boulevard. The building consists of roughly 63 OOO square
feet of floor area. The original construction drawings are dated 19 February 1965 and the
building is believed to have been constructed during 1965-66. Table I presents the building
summary. The building is essentially rectangular in plan with overall dimensions of
approximately 628 by 1500 in the North-South and East-West directions, respectively.
Figures 1 through 3 show the various elevations of the building taken after the Northridge
earthquake.

2.2. Structural system


Foundations supporting the Holiday Inn consist of 38 in deep pile caps, supported by groups
of two to four poured-in-place 24 in diameter reinforced concrete friction piles. All pile caps are

Table I. Building summary


Date of construction: 1965-1966
Date of drawings: 1965-1966
Design code: L. A. City Building Code 1964 (assumed)
Number of stories: 7
Plan dimensions: 628(N-S) x 1512(E-W)
Building height: 658p
Story heights:
Ground floor: 136
Second through sixth floor: 88
Seventh floor: 88%

Figure 1. North elevation of the damaged building, showing one bay shored up to the 5th floor
CONCRETE FRAME BUILDING NORTHRIDGE RESPONSE 153

Figure 2. South elevation, showing columns severely damaged below the 5th floor spandrel beams

Figure 3. Close-up view of the south perimeter frame, showing typical column damage

connected by a grid of tie beams and grade beams. Each pile is approximately 40 ft long and
has a design capacity of over 100 kips vertical load and up to 20 kips lateral load.
All structural weight and superimposed load on the building is carried by a system of reinforced
concrete flat slab and perimeter concrete beams supported by concrete columns. The concrete
slab is 10 in thick at the second floor, 8 i in thick at the third to seventh floors, and 8 in thick
154 M. SAIFUL ISLAM

North

Figure 4. Typical floor framing plan

at the roof. The typical framing consists of columns spaced at approximately 200centers in the
transverse (N-S) direction and 189 centers in the longitudinal direction. Figure 4 shows the
typical floor framing plan.
Lateral forces in each direction are resisted by perimeter spandrel beam-column frames as
well as interior slab-column frames. Typical interior columns are 2 0 x 2 0 between Ground
and 2nd and 18 x 18 above 2nd floor. The column reinforcing varies along the height of the
CONCRETE FRAME BUILDING NORTHRIDGE RESPONSE 155

column with 6- # 7 above 5th floor, 6- # 8 between 4th and 5th, 8- # 9 between 3rd and 4th,
and 10- # 9 below 3rd floor. The ties are # 3 @? 12 centers between Ground and 5th floor and
# 2 @ 12 centers above the 5th floor. Slab reinforcing in the column strip in both directions
typically consists of 16- # 6 at the top near the columns and 8- # 6 at the bottom.
Figures 5(a) and 5(b) show the elevation of the South and North perimeter frames, respectively.
The columns are 14 x 2 0 and have their weak axis oriented in the plane of the frames (see
Figure 6(a)). They are reinforced with 10-#9 between Ground and 2nd, 6-#9 between 2nd
and 4th, and 6- # 7 above the 4th floor. The reinforcements are spliced immediately above the
floor level and have a lap length of approximately 38 bar diameter (see Figure 6(d)). The ties
are # 3 @ 12 in on center between Ground and 4th and # 2 @ 12 in on center above the 4th
floor. At typical floor levels, the beams are 16 x 22.5 and are reinforced with 2-#6 at the
bottom and anywhere from 2-#8 to 3-#9 at the top.
The north perimeter frame has four bays of brick infill wall between the ground and second
floor (see Figure 5(b)). Nominal 1 in and + i n expansion joints separate these walls from the
exterior columns and the underside of the second floor spandrels, respectively (see Figure 6(c)).
Although these elements were not designed as part of the lateral force-resisting system, they
appear to have participated in resisting the imposed demand as is evident from the damage
sustained by these walls during both the Northridge (1994) and the San Fernando (1971)
earthquake.
In the transverse (N-S) direction, the perimeter columns have their strong axis in the direction of
the frames. These frames are hidden behind 1 in thick cement plaster supported by metal studs.
With the exception of some light framing members supporting the stairway and elevator
openings and non-structural brick filler walls, the structure is essentially symmetric. However,
as will be evident from the results presented in this paper, the participation of the non-structural
brick filler walls and exterior cement plaster appears to have contributed to the significant
torsional response experienced by the building during the Northridge earthquake.

2.3. Building material


Table I1 gives typical material properties. The structure is constructed of regular weight
reinforced concrete. The values in Table I1 are specified values which can significantly

Table 11. Properties of construction materials


~ ~ ~~

Specified f ; Assumed f ;
(psi) (Psi)

i
Columns, Ground-2nd floor 5 000 6 650
Columns, 2nd-3rd floor 4 000 5 320
Beams and slab, 2nd floor only 4000 5 320
All other concrete, 3rd-roof 3000 4 000
Minimum specified
yield strength, Assumed F,
Grade F, (ksi) (ksi)

i
Beams and slabs Intermediate-grade
Reinforcing steel (ASTM A-15 and A-305) 40 so
Columns bars Deformed billet bars
(ASTM A-432) 60 12
156 M. SAIFUL ISLAM

Roof
N+
7th flr.

,;
N R
5th f
l
rz
I
4th flr m
-@

-0
2nd flr.

1st flr
I I I I I I I I I

brick wall 1Partial brick wall


@I
Figure 5. (a) South perimeter frame elevation; (b) North perimeter frame elevation

underestimate the actual strength (referred to hereafter as the expected strength) of the in situ
material. Specified values of material strength are used in design and reflect the minimum value.
The expected values are always larger than the specified values because of the inherent
overstrengths in the original material and strength gained over time.
CONCRETE FRAME BUILDING NORTHRIDGE RESPONSE 157

Section B

Elevation D Section C

Figure 6. Beam sections and column detail

Furthermore, concrete exhibits a significant increase in both strength and stiffness and
reinforcing steel in strength when loaded at increased strain rates, e.g. at strain rates that are
expected during earthquakes. Compared to the normal rate of loading for a standard cylinder
test (35 psi s-I), the concrete in this building was strained at a rate estimated to be on the
order of 8 000 psi s - ' (building period assumed to be 1.5 s). This higher rate of loading alone
is estimated to increase the strength obtained from standard cylinder tests by as much as 15-20%.
In this study, therefore, the 'expected' values were used for all materials. The only exception
to this was in shear capacity calculation of beams and columns where the specified minimum
158 M. SAIFUL ISLAM

value of reinforcement was used. In the absence of in situ test results, the 'expected' values of
all materials were assumed to be approximately 20 to 33% higher than the specified minimum
values.

3. STRONG MOTION INSTRUMENTATION


The building response during the 17 January 1994 Northridge earthquake was recorded by
a total of sixteen CSMIP sensors. Figure 7(a) shows the location of these sensors. Ten of these
sensors recorded the North-South (transverse) response, five recorded the East-West

I I

Q
I- 150"O"

@)

Figure 7. (a) 1994 Northridge earthquake sensors; (b) 1971 San Fernando earthquake sensors
CONCRETE FRAME BUILDING NORTHRIDGE RESPONSE 159

(longitudinal) response and one recorded the vertical acceleration. All sensors were triggered
simultaneously at nominal 1% g vertical acceleration. Digitized response records for approximately
60 s are available for each of the 16 sensors. The time histories of all sensors are synchronized
within 0.02 s (Islam 1994).
The building also had limited instrumentation at the time of the 1971 San Fernando
earthquake. The building response during the San Fernando earthquake was recorded by Earth
Sciences AR-240 strong motion accelerographs located at the roof, fourth and ground floor
level. At each location, the response was recorded along all three principal directions:
North-South, East-West, and vertical. Figure 7(b) shows the location of the sensors during the
San Fernando earthquake.

4. EARTHQUAKE DAMAGE
The building experienced extensive structural damage during the 17 January 1994 Northridge
earthquake. The building was red tagged, and several bays along the perimeter of the building
were temporarily shored immediately after the earthquake. Shoring was provided up to fifth
ffoor level to those bays where the adjoining columns were damaged and their vertical load
carrying capacity was believed to have been compromised.
The structural damage was primarily confined to the longitudinal perimeter frames with
damage to the transverse direction frames limited to minor flexural cracks in the end bay beams.
Figure 8 shows the elevation of the longitudinal perimeter frames indicating the location of
major damage. The damage was most severe between the fourth and fifth floor levels of the south
perimeter frame (Line A). Figures 2, 3 and 9 show some of the damage sustained by this frame.
Damage primarily consisted of shear failure of the columns immediately below the 5th floor
spandrel beam. At several locations, extensive shear cracking may also have promoted the
buckling of vertical column reinforcement (see Figure 9).
In addition to damage to the columns, many beam-column joints below the 5th floor level
also sustained minor to moderate shear cracks. The damage was observed in beam-column
joints of both longitudinal frames although the south perimeter frame appeared to have slightly
more damage. Along with shear cracks, concrete spalling was also observed in one of the
joints. Besides these, structural distresses in the form of concrete spalling and hairline flexural
cracks were observed in several spandrel beams.
Figure 10 shows a close-up view of the infilled bay of the north perimeter frame located
towards the east end of the building. Cracks in the 2nd floor beam-column joint are clearly
evident. Cracks occurred at the same location during the 1971 San Fernando Earthquake. The
figure also shows a typical crack pattern in the non-structural brick infill walls along the north
face of the building. These cracks occurred at the 2nd floor beam soffit and near the corners of
each panel. The observed damages clearly suggests that these brick infill walls participated in
resisting the imposed seismic demand.
Nonstructural damage was not very extensive and was mostly confined to the 4th story. Doors,
windows and drywall partitions in the East-West direction suffered severe damage between the
4th and the 5th floor levels. This is attributed to the large deformation of this story during the
earthquake.

5. ANALYSIS METHODOLOGY
Four types of analysis were performed to: (1) establish the actual performance of the building
during the Northridge earthquake; and (2) to test the validity of analytical techniques/tools
160 M. SAIFWL ISLAM

I I I I

- I bays at lE't9" = 15D'-k I I

Damaged
removed following earthquake

@)

0 Indicates location of the most severe damage

Indicates temporary shoring

Figure 8. (a) Northridge earthquake damage-North perimeter frame; (b) Northridge earthquake damage-south
perimeter frame
CONCRETE FRAME BUILDING NORTHRIDGE RESPONSE 161

Figure 9. Close-up view of a damaged column at the 5th floor spandrel beam

commonly used by practising/research engineers and currently being incorporated in various


upcoming documents dealing with seismic evaluation .of existing buildings (e.g. ATC
33.03, LA City Division 95l, etc.).

5.1. Analysis of response records


The response of the building recorded by a total of 16 sensors during the Northridge
earthquake were analysed (Islam 1994) using basic system identification techniques to obtain
information about the dynamic response. Analyses of these accelerogram records provided
important building response information including actual periods of vibration, story drifts, roof
displacements, torsional response, distribution of inertial forces and story shear.

5.2. Three-dimensional elastic time history analysis


Three-dimensional elastic time history analysis of the building was performed using the
Northridge earthquake ground motion to establish elastic response parameters such as story
drifts and roof displacement. The ground motions were applied concurrently from the two
162 M. SAIFUL ISLAM

Figure 10. Close-up view of northeast end of the building, showing damage to the 2nd floor beam-column joint

orthogonal directions. For long period structures such as the Holiday Inn (i.e. structures whose
fundamental period is in the constant velocity and/or displacement region), research has shown
that global (roof,)displacements predicted by elastic analysis having element stiffness properties
equal to their stiffness at yield is either comparable to the inelastic displacement or overestimates
it (ATC 33.03 1995l, Seneviratna 19953, Newmark and Hall 19824). This generalized
rule is typically based on certain qualifications which are seldom complied with especially in
existing concrete structures. In spite of this shortcoming, this equal displacement assumption is
quite extensively used in the analysis of long-period structures. This study provides an
opportunity to test the validity of this assumption.
Two three-dimensional models were developed for the building, one with and one without
brick infill in the four bays of the north perimeter frame. The structural model included both
perimeter beam-column and interior slab-column frames. The building is assumed to be fixed
at the ground floor level. Building weight distribution used in the analysis consists of 1 341 kips
at roof level, 1 381 kips at typical Boor levels and 1751 kips at second floor, for a total of
9 999 kips. Figure 1 1 illustrates the three-dimensional computer model of the building, which
includes the infill wall.
Member stiffness properties were calculated from the moment-curvature curve at first yield
of the longitudinal reinforcement. For spandrel beams, the average of positive and negative
stiffnesses was used in the analysis. Four # 6 slab reinforcing were included in calculating the
negative flexural capacity (tension at the top) and stiffness of the beams. Table I11 shows
member stiffnesses used in the analysis. Five percent viscous damping was assumed for all
analyses. A panel zone rigidity of 50% was used to account for the fact that spandrel beams do
not cover the full width of the columns. For the interior slab-column frame, the sum of the
column dimension and three times the slab thickness was used to determine the gross moment
of inertia of the equivalent beam.
CONCRETE FRAME BUILDING NORTHRIDGE RESPONSE 163

I/
Figure 11. Three-dimensional model of the building

Table 111. Element stiffness properties used in the analysis


Member Effective moment of inertia (1,)t
Columns: 3rd-Roof 0.33
A. Perimeter frames Grd--3rd 0.42
Beams: 0.33
Columns: 3rd-Roof 0.42
B. Interior slab-column frames . Grd-3rd 0.70
Beams: 0.40

t Ig = gross moment of inertia.

Figure 12 shows the stiffness distribution between beam-column (exterior) and slab-column
(interior) frames as a function of story level. Clearly, the interior frames played a significant role
in providing the lateral resistance, especially in the transverse direction where there are a total
of seven interior frames. The results suggest that at typical floor levels the stiffness of each
interior slab-column frame is approximately one-half that of the perimeter beam-column frame.
However, at first story level, the interior and the perimeter frames have equal stiffness. This is
attributed to the increased story height at this level (136 as compared to 88.5 at typical floor
levels), which shifts a proportionally greater importance onto the flexural stiffness of the
columns. Given that the interior columns are 20 x 2 0 at this level compared to perimeter
frame columns which are 14 x 20, this in essence results in the interior frames resisting a larger
share of the seismic demand at the first story level.
164 M. SAIFUL ISLAM

(*) ROOF

- 7TH

6TH
5L. 5TH
,
I
I
,
,
,

v)
4TH

3RD

2ND
0 10 20 30 40 50 60 70 80 90 100
Stiffness (%)

Figure 12. Distribution of stiffness between interior and exterior frames: (a) longitudinal direction; (b) transverse
direction

5.3. Limit state analysis


Building structural components and sub-assemblies were analysed to establish respective
strength and deformation limit states.

5.4. Pushover analysis


Pushover analysis is simply a static nonlinear analysis where a predetermined lateral load
pattern that approximately represents the inertia forces is applied and the structure is pushed
under this load pattern until a mechanism is formed. There are several levels of sophistication
that may be used for pushover analysis, and the results are typically sensitive to the choice
of load pattern. However, for regular buildings of moderate height, it is common practice to
assume a triangular distribution for the load pattern. Accordingly, a triangular distributed load
pattern was used in this study.
A two-dimensional pushover analysis was performed on the south perimeter frame. The
objective of the pushover analysis was to determine if this analytical technique, which is currently
being incorporated in various seismic evaluation guidelines (e.g. ATC 33, ATC 4O/Proposition
122, etc.), could have predicted the observed damage. The program Drain-2DX, Version
1.03, developed at the University of California, Berkeley, was used to perform the pushover
CONCRETE FRAME BUILDING NORTHRIDGE RESPONSE 165

Table IV. Recorded peak acceleration and time of occurrence


E-W direction N-S direction
(longitudinal) (transverse)
Grd F1: S16: 0.459 @ 8.86 s S13: 0.429 @ 5.54 s
S1: 0,399 @ 5.58 s
2nd F1: S12: 0.339 @ 8.44 s S8: 0,409 @ 4.58 s
S7: 0.389 @ 4.12 s
3rd F1: S11: 0.369 @ 10.06 s S6: 0.459 @ 4.44 s
Northridge earthquake, 1994 S5: 0.419 @ 4.84 s
6th FI: S10 0.469 @ 9.12 s S4 0.339 @ 7,28 s
Roof S9: 0.589 @, 9.22 s S3: 0.579 @ 7.36 s
S2 0,569 @ 4.64s
Vertical Accel
Grd F1: S15: 0279
Grd FI: 0.139 0.259
4th F1: 0.2529 @ 7.9 s 0.2039 @ 9.1 s
San Fernando earthquake, Roof: 0,3279 @ 9.2 s 0.4069 @ 9.9 s
Vertical Accel
Grd F1: 0189

analysis. Drain-2DX employs a step-wise linear, iterative approach as an approximation of


nonlinear analysis.

6. ANALYSIS RESULTS
6.1. Response records
Table IV shows the peak acceleration and the time of occurrence during the Northridge
earthquake as recorded by each of the sixteen sensors in the building. Also included in the table
is the summary of building response recorded during the 1971 San Fernando earthquake. Figures
13(a) and 13(b) show relative floor displacement time histories. Relative displacement,
referred to hereafter simply as displacements, are obtained by subtracting the ground displace-
ment from the absolute displacement recorded at each floor. Figure 14 shows the roof
displacement time history in each of the two directions. Figures 15(a) and 15(b) show interstory
displacement time history traces for the bottom two floors. The peak values of displacements
and story drifts recorded during the Northridge earthquake are summarized in Table V. Figure
16 plots the difference of the drifts recorded in the N-S direction at each end of the floor. The
difference reflects the torsional response of the building.
The following conclusions can be drawn from the information presented in Tables IV and V
and Figures 13 through 16.
(1) In the E-W direction, the building vibration period appears to change significantly during
the strong motion period. The roof deflection trace in the E-W direction (Figure 14)
suggests the fundamental building period as being approximately 1.5 s during the early
part of the earthquake, 2.1 s between 10 and 20 s and 2.4s thereafter. It should be
pointed out that because of its location and orientation, the response recorded by this
sensor (sensor 9) is believed to have been purely translational and does not include any
torsional response of the building. This lengthening of building period is attributed to the
significant damage of the longitudinal frames during the earthquake.
166 M. SAIFUL ISLAM

0 4 8 12 16 20
Time (sec.)

0 4 8 12 16 2
Time (sec.)
Figure 13. Relative displacement: (a) East-West direction; (b) North-South direction

( 2 ) In the North-South direction, the fundamental building period appears to have been
approximately 2.2 s during the strong motion part of the earthquake and 2.0 s thereafter
(Figure 14). A lengthening of the building period, suggesting a loss of stiffness, does not
appear to occur in the transverse direction. This is consistent with the observed condition
where, apart from some flexural cracks in spandrel beams, no major damage could be
seen in the transverse frames.
CONCRETE FRAME BUILDING NORTHRIDGE RESPONSE 167

0 10 20 30 40 50 60
Time (sec .)

0 10 20 30 40 50 60
Time (sec .)

Figure 14. Roof displacement time history, Northridge earthquake

Table V. Recorded peak relative displacement and story drifts

Max displacement Interstory displacement Drift coefficient


(in) (in) (%I
[ Roof: S9: 9.20 in @ 9.34 s
{6th FI:
E-W direction 3rd F1:
S10:
S11:
8.23 in @ 9.34 s
3.62 in @ 8.52 s 1.98 in @ 8.52 s 1.9
2nd F1: S12: 1.64 in @ 8.52 s 1.64 in @ 8.52 s 1.o
Roof: S3: 6.94 in (4 7.40 s
S2: 9.05 in @ 10.64 s
6th FI: S 4 6.05 in @ 7.42 s
3rd FI: S6: 2.93 in @ 7.44 s 1.65 in @ 10.66 in 1.6
S5: 3.41 in @ 8.40 s 1.33 in @ 14.56 in 1.3
2nd F1: S8: 1.64 in @ 8.52 s 1.75 in @ 8.32 in 1.1
S7: 1.93 in @ 8.40 s 1.93 in @ 8.40 in 1.2
Note; Floor displ. = sensor displacement - ground displacement
168 M. SAIFUL ISLAM

0 4 8 12 16 2
T i (sec.)
(a)
3

io
h
1

B
m -1

-2

-3
3

i
n
1

A -1

-2

-3
0 5 10 15 2
T i m (sec.)
(b)
Figure 15. Inter-story displacement time history: (a) East-West direction; (b) North-South direction
CONCRETE FRAME BUILDING NORTHRIDGE RESPONSE 169

4 8 12 16 2
T i (sec.)

Figure 16. Time history of torsional response

(3) The torsional period of the building, at least during the initial period of the earthquake,
appears to have been on the order of 1.5 s (Figure 16).
(4) The second floor experienced as much as 1.64in displacement in the E-W direction
(Figure 15(a)). This suggests that the brick filler walls along the north face of the building,
which apparently had a 1 in caulking joint separating them from the adjoining columns,
may have acted as structural infill walls. The damage observed in the filler walls and the
significant torsional response recorded during the earthquake in an otherwise structural
symmetric building supports this hypothesis.
Table VI provides a summary of building periods observed during the Northridge earthquake
and those reported in earlier studies along with building periods calculated for the two models
analysed in this study. Also included in the table is the building period calculated using UBC
94 method A.
Noting that the building period is inversely proportional to the square-root of stiffness, the
information presented in Table VI suggests that the building lost approximately 80% of its
stiffness during the San Fernando earthquake (building period changed from 0.7 s to 1.5 s). This
is primarily attributed to the hairline/microcracks that are believed to have developed in
structural and non-structural elements (e.g. partition/cement plaster walls) of the building during
the San Fernando earthquake. An earlier study (Freeman 19765) indicated that non-structural
elements played a significant role in affecting the building response during the San Fernando
earthquake. The lengthening of period (and thus loss of stiffness) observed during the Northridge
earthquake in the longitudinal (East-West) direction is primarily attributed to the severe damage
sustained by the south perimeter frame.
Figure 17(a) shows the 5% damped response spectrum developed using the first 20 s of E-W
direction absolute acceleration recorded at the roof, 6th, 3rd and 2nd floor levels. Figure 17(b)
shows the Fourier spectrum of the displacement (relative to the base) response of the same. The
following important observations can be made from the results presented in these figures.
170 M. SAIFUL ISLAM

Table VI. Building periods

E-W direction N-S direction Torsional


(longitudinal) (transverse) 6)
(1) UBC 94 Method A 0.68 s 0.68 s
(2) Pre-1971 San Fernando
earthquake ambient vibration: 0.52 s 0.40 s
(3) San Fernando earthquake (1971)
early part of earthquake: 0.70 s 0.70 s
during peak response: 1.5 s 1.6 s
(4) Northridge earthquake (1994)
early part (0-10 s): 1.5 s 1.4 s
middle part (10-20 s): 2.1 s 2.2 s
towards the end (> 25 s): 24s 2.0 s
(5a) Three-dimensional model w/o infill 1.39 s 1.79 s 1.37 s
(effective stiffness properties) (0.46 s: 2nd mode) (0.58 s: 2nd mode)
(5b) Three-dimensional model w/infill 1.17 s 1.79 s 1.37 s
(effective stiffness properties) (0.37 s: 2nd mode) (0.58 s: 2nd mode)

(1) These response spectrum plots suggest that the acceleration response of lower floor levels
was dominated by response at periods on the order of 0.35 to 0.45 s while that of the
upper floors, although distributed over a broad range of periods (0.35 to 1.60 s), appeared
to have been concentrated more at periods 0.4 and 1.5 (2.1) s. These periods approximately
coincide with the second and first modal period, respectively, in the longitudinal direction
(see Table VI). The multiple peaks observed between 1.4 and 2.1 s reflect the shift in the
fundamental period of the building during the duration of the earthquake (see Figure 14
and Table VI).
(2) The displacement response of all floor levels, on the other hand, was dominated by the
response at 1.5 and 2.0 s which coincides with the first modal period in the longitidunal
direction of the building. As noted above, the two peaks at 1.5 and 230 s observed in the
Fourier spectrum reflect the lengthening of the fundamental building period in this
direction.
This suggests that higher modes significantly affected the acceleration response but not the
displacement response of the building. It is possible that the higher frequency (frequency is
inversely proportional to period) acceleration response, e.g. response at approximately 0.35 to
0.45s, may have become more pronounced once the south perimeter frame sustained the
severe damage. However, earlier work by the author (Islam 19942) has shown higher modes to
have contributed significantly even before the building is believed to have sustained severe
damage.
For buildings such as this where more than 80% of the total building reactive weight is effective
in the first mode, it is more common to assume contributions of higher modes to be negligible.
The results presented above clearly show that although that is true for the displacement
response, it is certainly not true for the acceleration response and thus, for the force
distribution on the building. This point is illustrated further later in this paper. These results
can be explained by noting that acceleration is obtained by twice differentiating the displacement
which in essence yields the following relation:
CONCRETE FRAME BUILDING NORTHRIDGE RESPONSE 171

0 1 2 3 4 0 1 2 3 4
S n d d wial(sa=) 8ndld M ( s a = )

Figure 17a. 5% damped response spectrum of E-W relative acceleration records

j m

0 1 2 3 4 5 0 1 2 3 4 5
Period (sec) Period (sec)

0 1 2 3 4 5 0 1 2 3 4 5
Period (sec) Period (sec)
Figure 17b. Fourier spectrum of displacement response
172 M. SAIFUL ISLAM

t
r

where S, and S, are acceleration and displacement, respectively and T can be viewed as the
building modal period. This in essence significantly amplifies the contribution of higher modes
to the acceleration response. For example, in this building where the first and second modal
period in the longitudinal direction is approximately 1.5 and 0.4 s, the above relationship
suggests that the ratio of second to first modal response is as much as 14 times more for
acceleration than it is for displacement.
In addition to the contribution of higher modes, the higher frequency content observed in
typical building acceleration response is also due to the fact that ground accelerations have
significantly more energy in this (higher) frequency range. For such a distribution of energy, the
acceleration response of even those buildings responding in a purely fundamental mode will
exhibit higher frequency content, while their displacement response will be predominantly
dictated by the resonant response at the fundamental period (Islam 19896). This suggests that
the distribution of forces on buildings will seldom, if ever, be triangular, as assumed in the model
codes.
Figure 18 presents a comparison of the response spectra developed using ground motions
recorded at the site between the UBC 94 and that proposed by LA City Division 95 default
spectrum. For periods greater than 1 s, the UBC design spectrum compares well with the
'recorded' spectrum while the Division 95 spectrum appears to underestimate it. At lower period
ranges (between 0 2 to 0.5 s), on the other hand, both UBC and LA City Division 95 spectrum
significantly underestimates the 'recorded' spectrum. Noting that the second translational modal
period of this building is in the 0.35 to 0.50 period range, use of the code and/or Division 95
spectra, therefore, would have significantly underestimated the contribution of the second
mode.

6.2, Elastic time history analysis


Figures 19 and 20 present story displacements and story drifts calculated using the two
three-dimensional models described earlier. Also shown on the same figures are the building
response actually recorded during the Northridge earthquake. Only the first and second story
drifts could be calculated from the recorded data. Table VII provides a summary of the same
results. The following conclusions can be drawn from these results.
CONCRETE FRAME BUILDING NORTHRIDGE RESPONSE 173
Roof

6
-0
a 5
3
$ 4
G
3

1
0 2 4 6 8 10 12 14 16

Displacement (inches)

Roof

6
-
$ 5
4

$ 4
3
3

1
0 2 4 6 8 10 12 14 16

Displacement (inches)
(b)
Figure 19. Story displacements-3D elastic time history analysis: (a) building East-West (longitudinal) direction;
(b) building North-South (transverse) direction (response @ C.O.M. same for with and without infill)

(1) The brick infill at the northeast end of the building can introduce significant torsion in
the building.
(2) Although neither of the models predict the observed roof displacement accurately, the
results appear to support the hypothesis that the brick infills probably interacted with the
confining frame.
(3) In general, roof displacements predicted by elastic analysis appear to overestimate the
actual displacement. Therefore, use of equal displacement rule would have yielded
conservative estimate of the inelastic displacement of the building. It should be noted
that the equal displacement rule although widely used to estimate global deformation,
e.g. roof displacement, is not typically valid for estimating individual story drifts.
174 M. SAIFUL ISLAM

-
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8

Drift Ratio (%)

(a)

4
-0
--
-0C.O.M
0S.W Corner (with infill)
ON.E Corner (with Infill)

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8

Drift Ratio (%)


(b)
Figure 20. Story drifts-3D elastic time history analysis: (a) building East-West (longitudinal) direction; (b) building
North-South (transverse) direction (response @ C.O.M. same for with and without infill)

Depending on structural component limit states, inelastic behavior of buildings typically


results in drift concentration in certain stories more than others. This is important
to bear in mind when reviewing the results presented in Figure 20.

6.3. Limit state analysis


6.3.1. Columns and beams. Figure 21(a) and 21(b) show moment-curvature diagrams
of typical longitudinal perimeter frame columns between the 4th and 5th floor levels and
spandrel beam at the 5th floor level. As noted earlier, above the 4th floor the columns
have # 2 ties @ 12 in on center. The member shear capacities were calculated using the
CONCRETE FRAME BUILDING NORTHRIDGE RESPONSE 175

Table VII. Comparison of maximum recorded and calculated roof displacement

Recorded Model 1 (w/Infill) Model 2 (w/o Infill)


(in) (in) (in)
Center of geometry 9.2 10.3 12.6
Longitudinal (E-W) direction - 11.0 12.6
North perimeter ~

9.8 12.6
Center of geometry - 8.7 8.7
Transverse (N-S) direction 6.94 11.8 8.7
West perimeter 9.04 14.3 8.7

200 -

fm = 4000. psi
2 160- - L. k s r
I
v) tw = 20. in.
._
a -
z 120-
Lw = 14. in.

s
W
Pu = 115. kips
Ductility = 2.1 4
I
0 80-
I

40-

0 ... . ( . I . . , . I . l , . I I . ) . I I I ( . l l l l . I I I I . I . . ( , , , . l , , , , (

0 120 240 360 480 600 720 840 9M) 1080 1200

fm = 4000. psi
2 80- - rn
V.
'

I
176 M. SAIFUL ISLAM

conventional method, which, in this case, suggests that approximately one set of ties is
effective in resisting the column shear. However, given the 14 in depth of the column parallel
to the direction of the longitudinal frame, the effectiveness of the ties in resisting the imposed
shear demand on the column is questionable at best.
At the component level, flexural limit state governs the behavior of the beams while shear
limit state governs the behavior of the columns. Stated another way, the shear capacity of the
column is typically less (although not by much) than the shear (2 MJh,,) associated with
flexural hinging of the column ends.
At the sub-assembly level, the expected performance is dictated by whether (i) the ties in the
column can be assumed to contribute to the shear capacity of the column, and (ii) the slab
reinforcing is included in determining the flexural capacity of the beams. If slab reinforcing is
not included and column ties are assumed to be effective, as was the case in the original design,
flexural yielding at beam ends dictates the limit state of typical sub-assemblage. For the scenario
in which the slab reinforcing is assumed to contribute to the beam flexural capacity, the limit
state of beam-column sub-assemblage above the 3rd floor appears to be dictated by flexural
yielding of beams at positive moment ends (tension at the bottom) followed by shear failure of the
columns. Below the 3rd floor, the limit state is flexural yielding of the beams at both ends. The
observed damage in the building appears to substantiate this later hypothesis.

6.3.2. Beam-column joints. Based on probable demands on the beams and columns, it is
estimated that the maximum shear demand on the beam-column joints at the lower floor levels
was between 8 and lo/&. It was mentioned earlier that although the original drawings indicate
the presence of several ties in the joint, the as-built condition appears to have been somewhat
different. At several joints where concrete had spalled off during the earthquake, no ties were
observed. Even without the ties, the probable joint shear stress demand of 8 to lo/& is not
considered to be excessive and cracking of beam-column joints at this stress level is typically not
expected. This appears to contradict the damage, although minor, observed in many of the
beam-column joints at the lower floor levels. One possible explanation of this inconsistency
may be that with beams not being centered with respect to the columns (see Figure 16(a)),
complex stress patterns may have developed at the joints which led to shear cracking and spalling
of concrete from the unconfined portion of the joint, i.e. the 4 in width of the column not confined
by the spandrel beams.

6.3.3. Story shear. The base and/or story shear actually experienced by the building
during the Northridge earthquake can be calculated in an approximate way by utilizing
DAlemberts principle, which states that the force at each level (at any given time) is the mass
at that level multiplied b y its maximum acceleration. As discussed earlier, absolute
acceleration was recorded at the ground, second, third and roof levels. By interpolating the
values for the remaining floors and multiplying the absolute acceleration (in gs) of each floor
by its weight cf = ma = w(a/g)),one can conveniently obtain the probable shear at any given
time during the earthquake. This procedure was used to calculate the probable seismic force
demand in the longitudinal (E-W) direction at select time intervals.
Figure 22 shows the approximate limiting shear capacity of each floor (indicated by the shaded
portion) along with probable recorded shear at 3.94 and 9.22s. The shear distribution at
3.94 s corresponds to the first major excitation in the longitudinal direction, and the distribution
at 9.22 s corresponds to the maximum roof acceleration. The limiting story shear capacity
indicated in the figure is calculated by summing up the shear capacity of all columns, both
interior and exterior.
CONCRETE FRAME BUILDING NORTHRIDGE RESPONSE 177

-9 6th t
ti
^I
4th -

2 n d [ . 1 . 1 I I I I
"':
\
#
0 500 loo0 1500 2000 2! 3
Story Shear (JSips)

Figure 22. Shear demand during Northridge earthquake versus calculated capacity (shaded portion indicates probable
story shear capacity)

The following important conclusions can be drawn from the results presented in Figure 22.
(1) The shear demand between the 4th and 5th floor levels is close to, if not the maximum,
experienced by the building at approximately 4 s into the earthquake. This suggests a load
distribution other than the triangular pattern commonly assumed for buildings such as
this which responds primarily in the first mode. This is consistent with the observations
made earlier and is further supported by the results presented in Figure 23. This is
discussed in more detail later in this section. This (the recorded story shear distribution),
along with the fact that the shear capacity of this story is significantly less than those of
the stories below, may have resulted in damage being concentrated between 4th and 5th
floor levels.
(2) The results suggest that the shear capacity of the entire story was exceeded at 3.94 s. This
would imply that the shear capacity of all the columns below the fifth floor level should
have been exceeded. This contradicts the damage observed where shear failure occurred
only in the south perimeter columns. Neither the interior columns nor most of the columns
of the north perimeter frame sustained any damage. This inconsistency might be explained
by the fact that (i) the calculated shear capacities probably underestimated the true
capacities of the members, and (ii) the torsional asymmetry introduced due to the presence
of brick infill walls along the north face of the building may have resulted in significantly
higher seismic demand on the south perimeter frame, thus resulting in their damage.
(3) The south perimeter frame columns probably reached their shear limit state at 3.94 s. The
shear distribution at 9.22 s is representative of the distribution that probably existed after
significant strength and stiffness degradation had occurred at the south perimeter columns.
Figure 23 shows floor acceleration and displacement recorded during the Northridge
earthquake as a function of story levels at select time steps. The acceleration response is shown
178 M. SAIFUL ISLAM

RF.
7TH

6TH

STH

4TH

3RD

ZND

GRD.
0.8 0.6 4.4 4.2 0 0.2 0.4 0.6 0.8
RELATIVE FLOOR ACCEL. (g)

RF.

7TH

6TH

STH

4TH

3RD

ZND

GRD.

Figure 23. Distribution of response recorded in the longitudinal (E-W) direction

at 3-94 and 9.22 s which coincides with the time of first major excitation and maximum recorded
acceleration at the roof, respectively. It is clear from the figure that the acceleration response
of the building was significantly influenced by higher translational modes of the building. The
displacement response of the building, on the other hand, was not found to have been influenced
by higher modes in any significant way. The influence of higher modes and especially the second
mode on the acceleration response was also evident in the results presented in Figure
17. Although it is conceivable that higher modes played a more significant role after the
columns of the south perimeter frame were severely damaged (e.g. at 9.34 s), at 3.94 s into the
earthquake the building is believed to have been responding primarily in the elastic mode (Islam
1994).
The results presented in Figure 23, therefore, clearly suggest that the distribution of seismic
force on the building during the Northridge earthquake was anything but triangular as assumed
in model codes. Also, it supports the results presented in Figure 22 showing that the
maximum shear force demand at 3.94s was not at the base but between the 4th and 5th
floor levels, i.e. the level at which the most severe damage was observed.

6.4. Pushover analysis


Figure 24 shows the results of the pushover analysis performed on the south perimeter frame
of the building. The results are presented in terms of base shear versus roof displacement.
Significant events in the progressive lateral response of the building such as first beam yielding
in flexure, first column yielding in shear, and formation of mechanism are shown on the
CONCRETE FRAME BUILDING NORTHRIDGE RESPONSE 179

1st Beam Yield (1)

0.00 4.00 8.00 12.00

Roof Displacement (in.)

Figure 24. Results of the pushover analysis. (Numbers in parentheses indicate sequence of hinge formations)

figure. For comparison, the maximum peak displacement (9-20 in) recorded in the East-West
direction and the displacement (6.3 in) recorded at approximately 4 s into the earthquake are
also shown on the same plot. Figure 25 shows the sequence of hinge formations.
Note that some of the requisite simplifications made in creation of the pushover model are
not likely to be present in the actual structure. Once such simplification relates to the assumption
that all structural components will behave in an elasto-plastic manner. Potential strength and
stiffness degradation, which are typically associated with concrete elements not detailed per
current ductile detailing requirements such as here, have not been taken into account. Further,
some of the component limit states such as shear failure of columns, which exhibit drastic strength
and stiffness degradation and are likely to lead to complete loss of that element have also been
assumed to perform in a ductile fashion. The pushover analysis curve presented in Figure 24,
therefore, should not be interpreted to indicate that buildings could actually achieve the roof
deflections indicated without premature brittle failure occurring in some of the elements, e.g.
columns.
The results of the pushover analysis suggests that the hinging begins with yielding of the
positive moment (tension at the bottom) end of the beam at the lower floors and progresses
upward. At roof displacement of approximately 5-7 in, a sway mechanism is formed with most
of the girders hinged and flexural hinges in the column formed at the base and immediately
below the 5th floor. At this stage, the shear demand on the columns between the 4th and 5th
floors is less than its shear capacity calculated assuming the column ties to be effective.
This behavior is not entirely consistent with the observed damage from the Northridge
earthquake because of the following.
(1) Based on the results of the pushover analysis, one would expect extensive flexural cracks
associated with both positive and negative moment at two ends of most of the beams. In
180 M. SAIFUL ISLAM

Assumed lateral
load distribubon

-1st column 1st b e a m n


y I e 1d1n g yielding

Indicates beam yield location.


@ Indicates column yield location.

Figure 25. Yielding sequence, pushover analysis

the actual building only minor positive flexural cracks at lower level beams were
observed-suggesting that the negative flexural capacity of the beams may not have been
exceeded.
(2) No flexural cracks were observed at the bases of the columns.
(3) Shear failures rather than flexural hinges occurred immediately below the 5th floor level.
This inconsistency is primarily attributed to the difference in the load distribution assumed in
the pushover analysis and that actually experienced by the building. It has been shown earlier
that at approximately 3.94s into the earthquake (first major excitation), the actual force
distribution on the building was such that the maximum shear demand was not at the base (as
would be expected for a triangular load pattern) but between the 4th and 5th floors. As noted
earlier, this, along with the fact that the columns below the 4th floor have significantly higher shear
capacity, is believed to be the primary reason that the shear capacity of the columns between
the 4th and 5th floors was exceeded before flexural hinges could form in most of the beams.

7. SUMMARY AND CONCLUSIONS


The subject building sustained severe structural damage during the 17 January 1994
Northridge earthquake. The south perimeter frame suffered the most severe damage and the
damage was primarily restricted to the columns between the 4th and 5th floor levels. The damage
to the south perimeter columns was so severe that it is believed that, had the earthquakes strong
motion duration been longer, the damaged columns would have lost their vertical load carrying
capacity completely leading to at least partial collapse of the building. The observed damage,
however, appears to be inconsistent with what one would typically expect in a building like this
CONCRETE FRAME BUILDING NORTHRIDGE RESPONSE 181

which is more or less structurally symmetric and has concrete frames as the primary lateral
bracing system.
Specifically, the questions of interest raised by this behavior are as follows:
(a) why was the damage primarily restricted to the longitudinal frame along the south side
of the building; and
(b) why was the most severe damage concentrated between the 4th and 5th floor levels and
not at the lower levels as one would normally expect.
In order to answer these questions as well as to see if the available analytical tools could have
predicted the observed damage, several types of analysis were performed as described in the
body of the paper. Based on the results of these analyses, the following conclusions may be
drawn.
(1) The brick filler walls between the ground and 2nd floors at the northeast end of the
structure interacted with the confining frame introducing asymmetry in the longitudinal
direction. This is believed to have led to higher displacement demand on the south
perimeter frame of the building.
(2) The building appears to have experienced significantly higher seismic shear (V/W =
-
0.20 0.25) than normally expected of such buildings and certainly significantly more
than it was originally designed to withstand. This is attributed partly to the significant
participation of the interior slab-column frames and partly to the inherent material and
system overstrength.
(3) It appears that many of the structural elements may have exceeded their elastic limit state
at approximately 4 s into the earthquake. However, the most severe damage, e.g.
breakdown of the entire load path in the south perimeter frame columns immediately
below the 5th floor level, may have actually occurred at approximately 9 s, which coincides
with the time of the peak ground acceleration in the longitudinal direction.
(4) The higher modes significantly influenced the acceleration response but not the
displacement response of the building (see Figures 17 and 23). Consequently, the
distribution of seismic forces on the building during the Northridge earthquake
was very different from the triangular distribution typically assumed in model codes for
buildings such as this whose displacement response is primarily dominated by fundamental
translational modes.
(5) Because of higher mode affects, the shear demand between the 4th and 5th floor levels
appears to have been close to, if not the maximum, experienced by the building. This
along with the fact that the building shear capacity of this story is less than those of the
stories below, may have resulted in damage concentrating between the 4th and 5th floor
levels.
( 6 ) The maximum shear demand on beam-column joints is estimated to have been between
&
8 to 10/ during the Northridge earthquake. This is not considered to be excessive and
cracking of beam-column joints at this level of stress is typically not expected. This is
inconsistent with the damage observed in many of the beam-column joints at lower floor
levels. One possible explanation of this inconsistency may be that, with beams not being
centered with respect to the columns, complex stress patterns may have developed at the
joints which lead to shear cracking and spalling of concrete from the unconfined (i.e. the
4 in width of the column not confined by spandrel beams) portion of the joints.
(7) In general, roof displacements predicted by elastic analysis appear to overestimate the
actual displacement. This suggests that for long period structures such as this the equal
182 M. SAIFUL ISLAM

displacement rule may be used to establish a conservative estimate of the inelastic


displacement of the building.
(8) Although not required by the code applicable at the time of its design, the original design
appears to have made an attempt to incorporate the weak-beam/strong-column
philosophy. However, by not including the probable contribution of the slab reinforcement,
the original design significantly underestimated the flexural capacity of the spandrel beams.
This forced the limit state from the desirable flexural yielding of beams to brittle shear
failure of the columns. The lack of adequate ties in the column exacerbated the shear
damage and allowed vertical column reinforcing to buckle where concrete had severely
cracked and/or spalled off.
(9) Pushover analysis results are sensitive to the assumed load distribution. A pushover
analysis performed on the longitudinal frame with a triangular load pattern was unable
to predict the damage observed in the building. This suggests that although the pushover
analysis is a useful tool, it certainly is not an infallible tool for assessing inelastic strength
and deformation demands or for exposing design weaknesses. Further, although pushover
analysis is a very useful tool for understanding theoretical failure sequences and building
responses, requisite simplifications in the creation of pushover models can result in failure
sequences that can be quite different from observed damage.

ACKNOWLEDGEMENT

Financial assistance provided by the California Seismic Safety Commission for part of this work
is gratefully acknowledged. Supplemental funding provided by Nabih Youssef & Associates is
also gratefully acknowledged.
The opinions expressed herein are those of the author and do not necessarily reflect the
views of the California Seismic Safety Commision or Nabih Youssef & Associates.

REFERENCES

1. ATC 33.03 (75% Draft Document), Guidelines for the seismic rehabilitation of buildings, Applied
Technology Council, October 1995.
2. M. S. Islam, Holiday Inn, Orion Avenue, Van Nuys, Proceedings of the 63rd Annual SEAOC Convention,
1994, pp. 117-163.
3. G . D. P. K. Seneviratna, Evaluation of inelastic MDOF effects for seismic design, Doctoral dissertation,
Stanford University, California, U.S.A., December 1995.
4. N. M. Newmark and W. J. Hall, Earthquake Spectra and Design, Earthquake Engineering Research
Institute, Oakland, 1982, 103 pp.
5. S. A. Freeman, Prediction of response of concrete buildings to severe earthquake motion, Douglas
McHenry International Symposium on Concrete and Concrete Structures, 1976, ACT SP 55-23,
pp. 589-605.
6. M. S. Islam, Modal coupling and wind-induced vibration of tall buildings, Doctoral dissertation, The
Johns Hopkins University, Baltimore, Maryland, U.S.A., 1988.

Anda mungkin juga menyukai