You are on page 1of 12

Engineering Structures 25 (2003) 103–114

www.elsevier.com/locate/engstruct

Performance of reinforced concrete buildings during the August


17, 1999 Kocaeli, Turkey earthquake, and seismic design and
construction practise in Turkey
H. Sezen a,∗, A.S. Whittaker b, K.J. Elwood a, K.M. Mosalam c
a
Graduate Student Researcher, Pacific Earthquake Engineering Research Center, University of California, Berkeley, 1301 South 46th Street,
Richmond, CA 94804-9471,USA
b
Associate Professor, Department of Civil, Structural, and Environmental Engineering, 230 Ketter Hall, University at Buffalo, Buffalo, NY
14260, USA
c
Assistant Professor, Department of Civil and Environmental Engineering, University of California, Berkeley, 733 Davis Hall, Berkeley, CA
94720-1710, USA

Received 26 March 2002; received in revised form 17 July 2002; accepted 17 July 2002

Abstract

A large number of reinforced concrete buildings collapsed or were heavily damaged during the 7.4 magnitude earthquake that
struck northwestern Turkey on August 17, 1999. Recorded peak ground accelerations were relatively low (0.3 g–0.4 g) compared
to the magnitude of the structural damage, and the elastic acceleration response spectra from the recorded motions were comparable
with the elastic design spectra specified in the current Turkish seismic code. Seismic code requirements are discussed and compared
with observed details. Many structural deficiencies were highlighted by the earthquake damage, including: reinforced concrete
columns with insufficient confinement and transverse reinforcement, 90-degree hooks at the end of column ties, poor detailing in
beam-column joint regions, strong-beam and weak-columns, soft and weak stories, and poor quality construction. Buildings with
shear wall structural elements generally performed well.
 2002 Elsevier Science Ltd. All rights reserved.

Keywords: 1999 Turkey earthquake; Earthquake damage; Reinforced concrete; Seismic behavior; Codes; Design

1. Introduction

On August 17, 1999, a Mw 7.4 earthquake occurred


on the 1500-km-long North Anatolian fault in northwest-
ern Turkey. The epicenter of the earthquake was near
Izmit, 90 km east of Istanbul (Fig. 1). Following the
earthquake, the Pacific Earthquake Engineering
Research Center dispatched a reconnaissance team to the
epicentral region to learn first hand about the perform-
ance of the civil infrastructure. The geographic region
that was impacted by the earthquake was somewhat nar-
row banded and centered around the fault, and stretched
from Istanbul in the west to Gölyaka and Düzce in the Fig. 1. Map of affected region showing recorded peak ground accel-
erations in circles (as percentage of acceleration of gravity).
east. Damage to building construction was severe and
widespread (Sezen et al. [1], Aschheim [2], Scawthorn
[3]). Estimates for economic losses were around 20

Corresponding author. Tel.: (510) 231-9510; fax: (510) 231-9471. billion US dollars. The official death toll was over
E-mail address: sezen@ce.berkeley.edu (H. Sezen). 17,200, with some 44,000 people injured and thousands

0141-0296/03/$ - see front matter  2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 1 - 0 2 9 6 ( 0 2 ) 0 0 1 2 1 - 9
104 H. Sezen et al. / Engineering Structures 25 (2003) 103–114

left homeless. Some 77,300 homes and businesses were was the largest of these earthquakes. Following the Erz-
destroyed, and 244,500 were damaged. The majority of incan earthquake, the first seismic design code for build-
deaths and injuries were in the cities of Kocaeli, Sakarya, ings was published in 1940. The base shear, V, was cal-
and Yalova. culated as the product of a lateral force coefficient, C,
This paper describes briefly the state-of-practice for and the weight of the building, W, namely:
building seismic design and construction in Turkey, and
V⫽CW (2)
compares the US and Turkish codes. The performance
of the reinforced concrete frame and wall buildings and The value of C was set equal to 0.10 regardless of the
their components during the 1999 Kocaeli earthquake is location of the designed structure in Turkey. The base
presented, and evaluated considering the seismic design shear force was distributed over the height of the build-
and construction practice in the epicentral region. ing using a uniform load pattern.
In 1968, the earthquake code was substantially
changed. In addition to introduction of limitations on the
2. Turkish code requirements for reinforced cross-sectional dimensions of reinforced concrete
concrete construction components, modern concepts relating to spectral shape
and dynamic response were included. The uniform load
Two codes influence the design and construction of profile of earlier codes was replaced with a first mode
reinforced concrete buildings in Turkey: the “earthquake (inverted triangle) profile similar to that used in current
code” (Specification for Structures to be Built in Disaster US codes such as the 1997 Uniform Building Code [8].
Areas, [4,5]) and the ‘building code’ (TS-500, Building In 1975, important provisions were added to the seis-
Code Requirements for Reinforced Concrete [6]). The mic code, including new methods for calculating earth-
earthquake code includes procedures for calculating quake loads on buildings and ductile detailing require-
earthquake loads on buildings. The building code ments for reinforced concrete structures. The lateral
presents requirements for the design and detailing of force coefficient of the 1975 code was defined as
reinforced concrete components but does not include C ⫽ CoKIS (3)
ductile detailing requirements for use in seismic design.
Such requirements are found in the earthquake code. where Co was a seismic zone coefficient and varied
between 0.10 and 0.03, K was a coefficient related to
2.1. Building code requirements for reinforced the type of framing system, I was an importance factor,
concrete and S a spectral coefficient.
Geometry and detailing requirements for reinforced
The building code is similar to ACI 318-95 [7] but concrete components were modified in the 1975 code.
does not address design and detailing for earthquake With respect to transverse reinforcement spacing and
loads. The engineer is referred to the earthquake code detailing requirements, columns were divided into three
[4,5] for such information. Early editions (e.g., 1969) of regions as shown in Fig. 2a: confinement regions at each
the building code were based on allowable stress design. end of the column clear height, middle region, and beam-
Major changes were introduced into the code in 1981 column joint regions. A maximum tie spacing (s1 in Fig.
and 1985. The latest version of the code [6] permits both 2a) of 200 mm or 12 times the longitudinal bar diameter,
allowable stress and ultimate strength design. For the and 135-degree hooks were required on ties in confine-
latter, earthquake effects are considered using two load ment regions. In the middle region, the amount of trans-
combinations, verse reinforcement was based on satisfying gravity and
earthquake forces that were calculated using Eq. (3).
U ⫽ G ⫹ P ⫹ E and U ⫽ 0.9G ⫹ E (1) Response-spectrum and linear and nonlinear dynamic
where U is the required strength, G, P and E are the analysis procedures were introduced in the 1998 edition
dead, live, and earthquake load effects, respectively. of the seismic code. The lateral force coefficient C of
Earthquake loads are calculated following the pro- the 1975 code was replaced by
cedures defined in the seismic code of the time. A(T) AoIS(T)
C⫽ ⫽ (4)
Ra(T) Ra(T)
2.2. Earthquake codes
where A is the spectral acceleration coefficient; T is the
Major earthquakes in Turkey have led to substantial fundamental period; and Ao is the effective ground accel-
changes in the practice of seismic design and construc- eration coefficient. The importance factor varies between
tion. Fifty-seven destructive earthquakes struck Turkey 1 and 1.5, and is equal to 1.0 for ordinary structures.
in the 20th century, most occurring along the 1500-km- The spectrum coefficient, S, which defines the design
long North Anatolian fault. The M 7.9 Erzincan earth- acceleration spectrum, is given by three equations in the
quake of December 27, 1939, in northeastern Turkey, short-period, constant acceleration, and constant velocity
H. Sezen et al. / Engineering Structures 25 (2003) 103–114 105

Fig. 2. Column details from (a) 1975 and (b) 1998 Turkish seismic codes.

ranges. These ranges are delineated by spectrum charac- 3. Design and construction practice prior to the
teristic periods, which vary as a function of soil type. 1999 Kocaeli earthquake
The maximum spectral amplification is 2.5, which is
identical to that of US codes [8] for 5% damping. The From discussions with local engineers in Turkey and
seismic load reduction factor, Ra, is similar to the inspection of buildings, it is clear that the Turkish earth-
response modification factor in US codes, except that the quake code provisions were rarely enforced for the
seismic load reduction factor reduces linearly from the engineering of commercial and residential construction
maximum value of structural behavior factor, R, which in the epicentral region prior to the August 17, 1999
is tabulated in the code, to 1.5 at zero period. The earthquake [9]. The ductile reinforcement details
maximum value of R depends on the assumed ductility described in the 1975 earthquake code were rarely
(high or nominal) of the system and varies between 3 observed in buildings inspected by the authors after the
and 8. August 17, 1999, earthquake. Details of the seismic
In the 1998 edition of the earthquake code, reinforced design and construction practice prior to the 1999 Koca-
concrete buildings are classified as systems of either high eli earthquake are provided in [1].
or nominal ductility, based on the detailing of the
components. Detailing requirements are more stringent
for systems with high ductility. The detailing require- 4. Response spectra
ments for columns being designed for high or nominal
ductility levels are very similar. Information on the Fig. 3 presents 5% damped linear elastic acceleration
transverse reinforcement requirements along the height response spectra for rock and soft soil sites calculated
of a column is shown in Fig. 2b. All hoops must have using the provisions of the 1997 Uniform Building Code
135-degree seismic hooks at both ends. The shear [8] and the 1998 Turkish earthquake code [5] for the
strength of a column must exceed the shear force asso- regions of highest seismicity in each country. The Uni-
ciated with the plastic moment in the column. The only form Building Code (UBC) spectra shown in the figure
major provision that is not applicable for columns include near-field amplification factor of 1.0. The elastic
designed for a nominal ductility level is the spacing of spectral demands of the two current codes are very simi-
transverse reinforcement in the confinement zones, lar.
which is required to be half the spacing used in the col- The peak ground accelerations (PGA) recorded in the
umn middle region. For columns in frames of nominal region affected by the earthquake are shown in Fig. 1.
ductility, the maximum spacing of the transverse Recorded peak ground accelerations were relatively low
reinforcement between the confinement zones was (0.3 g–0.4 g) compared to the extent of the structural
increased by a factor of 2 over the spacing shown in damage. Fig. 3 includes the elastic response spectra at
Fig. 2b. 5% damping from one horizontal component of the
106 H. Sezen et al. / Engineering Structures 25 (2003) 103–114

reinforced concrete moment frames, respectively; Co was


set equal to 0.10. These allowable stress design spectra
were scaled to the strength level by multiplying the spec-
tral ordinates by 1.4. To construct the spectra for the
1998 Turkish code, Ao and the importance factor, I, were
set equal to 0.40 and 1.0, respectively; and R was set
equal to 4 and 8 for reinforced concrete moment-
resisting frames of nominal and high ductility, respect-
ively. To construct the spectra for the 1997 UBC, soil
type SE was assumed; near-field factors were not con-
sidered; the importance factor was set equal to 1.0, and
R was set equal to 3.5 and 8.5 for ordinary moment-
resisting frames (OMRF) and special moment-resisting
frames (SMRF), respectively. (The OMRF and SMRF
of the UBC correspond approximately to frames of nom-
inal and high ductility in the Turkish earthquake code,
respectively.)
Fig. 3. Elastic response spectra for 5% damping. For modern reinforced concrete moment-resisting
frames of high ductility (equivalent of SMRF in the US),
ground motion recorded at the following sites: YPT the ordinates of the 1998 Turkish lateral-force-coef-
station in Yarimca near Gölcük (PGA 0.23 g, soft soil), ficient spectra exceed those of the 1997 UBC. Recogniz-
SKR station in Adapazari (PGA 0.41 g, stiff soil), and ing that the prescriptive details of the 1997 UBC and the
the DZC station to the east of Gölyaka (PGA 0.32 g, 1998 Turkish code for frames of high ductility are very
soft soil). The elastic acceleration response spectra from similar, the performance of buildings designed to either
the recorded motions are not substantially different from code should be similar if the construction standards are
the elastic design spectra specified in the current US and similar and the fundamental period exceeds 0.2 sec. For
Turkish seismic codes. short-period buildings and identical construction stan-
Fig. 4 presents lateral-force-coefficient spectra (C in dards, buildings designed and detailed according to the
Eq. (1)) for the 1975 and 1998 Turkish codes and the 1998 Turkish earthquake code will be stronger than
1997 UBC for reinforced concrete moment-resisting buildings designed and detailed using the 1997 UBC [8].
frames on soft soil sites. Such frames were chosen for
the purpose of comparison because reinforced concrete
moment-resisting frames are the most common seismic 5. Typical moment-resisting frame construction
framing system in Turkey for building construction. To
construct the spectra for the 1975 Turkish code, K was The construction quality for residential and commer-
taken as 0.80 and 1.50 for ductile and nonductile cial buildings in the epicentral region varied widely.
Although the general construction quality for residential
building was poor according to US standards, many of
the engineered reinforced concrete commercial buildings
(e.g., office buildings) were observed to have satisfactory
construction quality. Many of the failures and collapses
of engineered commercial construction observed by the
reconnaissance team, and reported in Sezen et al. [1],
can be attributed to the use of nonductile details and not
poor quality construction.
Residential buildings in the epicentral region typically
range in height from two to seven storeys. Fig. 5a shows
a photograph of a three-storey moment resisting frame
building that was under construction at the time of the
earthquake. A plan of the second floor is shown in Fig.
5b. Note that ten of the columns are stronger in the x-
direction (parallel to the street), whereas five columns
are stronger in the y-direction. The column orientations
and locations are such that all of the moment-resisting
Fig. 4. Lateral-force-coefficient spectra for the 1975 and 1998 Turk- frames include one or more columns with their weak
ish seismic codes and the 1997 Uniform Building Code. axis perpendicular to the frame direction. These obser-
H. Sezen et al. / Engineering Structures 25 (2003) 103–114 107

Fig. 5. 3-storey building near Golcuk; (a) elevation, (b) floor plan.

vations, which were typical of most buildings in the buildings in Gölcük. One of the buildings collapsed
epicentral region, would suggest that the framing system completely, whereas the adjacent building exhibited
is much stiffer and stronger in the direction perpendicu- shear cracks in the first storey. Careful inspection of the
lar to the street assuming that similar rebar are used in first storeys in both buildings showed that the buildings
all beams and all columns. Such framing likely possesses had a similar plan footprint and construction details. It
limited strength and stiffness, which if coupled with non- is likely that the two buildings were nominally identical
ductile reinforcement details, results in a vulnerable and that both buildings were constructed by the same
building in the event of earthquake shaking. contractor. Both buildings were likely subjected to simi-
lar levels of earthquake shaking, yet one building perfor-
med well, while the other collapsed. This raises many
6. Behavior of moment-resisting frame construction questions regarding the limit state for nonductile
moment frames. Small differences in the strength of
Reinforced concrete moment-resisting frame buildings these nonductile buildings possibly caused by the vari-
behaved poorly during the Kocaeli earthquake. Accord- ation in material strength, construction practice, and
ing to official estimates, more than 20,000 moment- workmanship could account for the drastic difference
frame buildings collapsed, and many more suffered in performance.
moderate to severe damage. Three- to seven-storey
apartment buildings were hard hit, although many had
been constructed in the past 20 years. These apartment 7. Role of infill walls in response of moment-frame
buildings were likely designed and detailed to comply buildings
with the requirements of the building code [6] and 1975
earthquake code [4] for construction in a first-degree Hollow clay tile and gas–concrete masonry infill walls
seismic zone. Many of the collapses are attributed to the are widely used in the epicentral region. Many of the
formation of soft first storeys that formed as a result of buildings were constructed with hollow clay tile infill
differences in framing and infill wall geometry between walls in the frames perpendicular to the sidewalk.
the first and upper stories, the use of nonductile details, Frames parallel to the sidewalk were often infilled with
and poor quality construction in some cases. hollow clay tile only above the first storey to allow for
Fig. 6 shows two six-storey nonductile moment-frame commercial space on the ground level. Such an arrange-

Fig. 6. Variability of building response.


108 H. Sezen et al. / Engineering Structures 25 (2003) 103–114

ment of tile infill walls created stiffness discontinuities


in these buildings, which may have contributed to their
collapse by concentrating the drift demands in the first
storey. These walls are almost always unreinforced and
abut the frame columns without being tied to frame
members. The high in-plane stiffness of the masonry
infill walls, which is developed by diagonal strut action,
can dictate the response of the more flexible moment-
resisting frame. Damage to masonry infill walls was con-
centrated in the lower stories of buildings because of
higher story shear demands on the strength of the
moment frame-infill wall system.
Fig. 7 shows views of a collapsed apartment building
in Gölcük. The first two storeys of this building failed
completely, but damage in the upper four storeys with Fig. 8. Failure of two stories of a moment-frame building with infill
unbroken glass windows was limited. The long infill masonry walls.
walls in the upper four storeys have significant elastic
strength and stiffness—probably much greater stiffness in the back face of the building and in the two faces
and strength than the moment-resisting frame. If the perpendicular to the sidewalk. The front of the building
infill walls in the upper four storeys of the building are was open in the first storey. The lateral stiffness of the
indicative of the infill in the failed storeys, the first- and building was considerably larger in the direction perpen-
second-storey infill walls likely played an important role dicular to the sidewalk compared with parallel to the
in the collapse of the building. The brittle fracture of sidewalk. Deformations are concentrated in the first sto-
the first- and second-storey masonry infill walls prior to rey of this building parallel to the sidewalk, due to the
flexural yielding of the columns would have overloaded weakness and flexibility of the moment frame because
the nonductile first- and second-storey columns in shear, of lack of masonry infill walls in the front of the build-
likely resulting in the observed gravity load failure. ing. The first-storey columns in this building were sever-
The first two storeys of the building in Fig. 8 col- ely damaged and likely close to failure due to gravity
lapsed. The masonry infill walls and moment-frame con- load instability.
struction in the third and fourth storeys (first and second
storeys in the photograph of the collapsed building) suf-
fered major damage. Damage in this building reduced 8. Detailing and response of components of
with increased height above the ground. Failure of the moment-frames
masonry infill in the first and second storeys of the build-
ing likely precipitated the collapse of the building. Previous sections of this paper on moment-frame con-
Irregular placement of masonry infill walls can pro- struction have focused on the response of moment-frame
duce discontinuities of stiffness in moment- frame build- systems. This section addresses the response of the
ings. Consider the building in Fig. 9 in which the components of moment frames, namely beams, columns,
moment frame is both flexible and weak in the first sto- and beam-column joints. Based on the authors’ obser-
rey by comparison with the upper storeys. In the first vations, typical details for an existing column and beam
storey of this building, masonry infill walls are present reinforcing bar details are shown in Fig. 10. Bent-up bars

Fig. 7. Views from a collapsed apartment building in Golcuk.


H. Sezen et al. / Engineering Structures 25 (2003) 103–114 109

are shown in the typical beam section. Corner column


bar are spliced above the floor slab with lap lengths of
40–70 bar diameters. Side-face column rebar are either
spliced per corner rebar or terminated above and below
the joint with 180-degree hooks as shown Fig. 10. No
additional transverse reinforcement for the purpose of
confinement is provided in the hinge, joint, or splice
regions. In general, typical 80-mm–120-mm thick
reinforced concrete floor slabs performed well. Vulner-
abilities and description of the damage to the commonly
used joisted floor slab framing system are provided in
Sezen et al. [1].

8.1. Beams

Typical beam spans ranged between 3 and 5 m. Beam


widths and depths ranged between 200–250 mm and
500–600 mm, respectively. Transverse ties are smooth
reinforcing bars of 6–10 mm diameter with 90-degree
hooks. Bent-up longitudinal reinforcing bars, often used
for reasons of economy to provide shear resistance to
gravity loads and to increase negative moment-resistance

Fig. 9. Formation of a soft and weak storey.

Fig. 10. Typical beam and column rebar details observed by the
reconnaissance team. Fig. 11. Damage to nonductile reinforced concrete beam.
110 H. Sezen et al. / Engineering Structures 25 (2003) 103–114

for gravity loads at supports, however do not properly the column dimensions and details observed by the
resist cyclic shear forces induced by earthquake shaking. reconnaissance team for a typical building, and compares
Little damage to interior moment-frame beams was with those dimensions and details given in Chapter 21
observed by the reconnaissance team because columns of ACI 318-95 [7], the Turkish building code [6], and
were generally weaker than beams. One type of beam recent Turkish seismic codes, [4,5]. The data of Table
damage is shown in Fig. 11. The building in this figure 1 coupled with field observations clearly show that the
suffered a partial storey collapse because the fault rup- Turkish earthquake code provisions were not followed
tured beneath the building. The beams shown were for- in detailing reinforced concrete columns, especially in
ced to accommodate the partial collapse and were badly potential plastic hinge zones at the end of the columns.
damaged at the beam-column connection due to slip of In damaged columns, the authors did not observe a tie
the smooth longitudinal beam reinforcing bars. In many spacing of less than 100 mm or 135-degree hooks at the
cases, beam bottom reinforcing bars were inadequately end of column ties as required by all codes except for
anchored through the beam-column joint. the Turkish building code, TS-500 [6], which in general
does not include ductile detailing requirements for col-
8.2. Columns umns.
The majority of moment-frame component failures
Column plan dimensions range between 150 mm×500 was in columns and was due to (a) the use of nonductile
mm–250 mm×800 mm. Most columns in residential con- details and unconfined lap splices, (b) excessive beam
struction are blade columns with a depth-to-width ratio strength, and (c) interaction between the columns and
of approximately three. The longitudinal rebar ratio the masonry infill walls. Lap splices in moment-frame
ranges between 1% and 2%; 12–16 mm diameter smooth columns were typically made immediately above the
rebar are generally used. Transverse ties are smooth floor framing or the foundation. The exposed lap splice
rebar of 6–10 mm diameter with 90-degree hooks. The of Fig. 12a is from a moment-frame building in Adapaz-
spacing of transverse ties is typically 200–250 mm uni- ari. The lap splices in this column were approximately
form along the clear height of the column. Table 1 shows 35 bar diameters in length and were located in a plastic

Table 1
Observed and code specified column detailing and dimensions

ACI 318–95 Turkish 1975 Turkish 1998 Observed

Ductile Nonductile
Turkish 1975 TS500-1985

Min width (W) ⱖ300 mm ⱖ250 mm ⱖ250 mm ⱖ250 mm typical:


ⱖ0.4∗D ⱖH/20 250 mm
ⱖD/3 common:
200 mm

Min depth (D) ⱖ300 mm ⱖ250 mm ⱖ250 mm ⱖ300 mm typical:


400–600 mm

ρ (longitudinal) 0.01ⱕρⱕ0.06 0.01ⱕρⱕ0.04 0.008ⱕρⱕ0.04 0.01ⱕρⱕ0.04 typical:


0.01 to 0.02

Column middle ⱕ150 mm s2ⱕ200 mm ⱕ200 mm suⱕ200 mm typical:


zone: ⱕ6∗(db) long. s2ⱕ12∗(db) long. ⱕ12∗(db) long. suⱕW/2 150–250 mm
Tie spacing s2ⱕD/2
Hooks 135° 90° or 135° 135° all 90°

Column end ⱕ100 mm s1ⱕ100 mm scⱕ100 mm typical:


zones: ⱕW/4 s1ⱖ50 mm scⱖ50 mm 150–250 mm
Tie spacing scⱕW/3 100 mm or less
not observed
Hooks 135° 135° 135° 135° all 90° hooks
no 135° hooks

W: Minimum member dimension (width), D: longer member dimension (depth), H: column height, ρ: longitudinal reinforcement ratio, (db) long.:
longitudinal bar diameter, s1, s2, sc, su : shown in Fig. 2.
H. Sezen et al. / Engineering Structures 25 (2003) 103–114 111

Fig. 12. Column failures.

hinge zone. Transverse reinforcement was widely spaced The photograph of a building (Fig. 14) that was under
with 90-degree hooks, and no cross ties were present. construction in Adapazari at the time of the earthquake
The 90-degree hooks on the ties opened during the earth- shows severe damage in the beam-column joints, but the
quake and the limited strength and confinement afforded horizontal transverse ties maintained the integrity of
by the ties were lost. the joints.
The transverse tie details of Fig. 12b were common,
namely smooth rebar, widely and often unequally spaced
ties (200–250 mm), and 90-degree hooks. The wide 9. Behavior of shear-wall construction
spacing of the ties resulted in shear failures (Figs. 12b
and 12d), buckling of longitudinal rebar (Fig. 12c), and Buildings constructed using shear walls as the primary
poor confinement of the core concrete. lateral load-resisting system performed quite well in the
1999 Kocaeli earthquake. Some buildings with a dual
8.3. Beam-column joints wall-frame lateral load-resisting system were damaged
because the shear walls were not sufficiently stiff to keep
Damage to beam-column joints is shown in Fig. 13. the displacements of the nonductile framing system in
Much of the framing (Fig. 13a) is essentially intact but the elastic range. Storey collapses were not observed in
many of the beam-column joints are heavily damaged. buildings containing shear walls, but it should be noted
A view of one of the damaged joints is shown in Fig. that shear walls were not widely used in the epicentral
13b. Beam reinforcing bars anchorage in the joint is region. The reconnaissance team toured a number of
inadequate and no transverse ties are present in the buildings that would be classified as dual wall-frame sys-
joint region. tems in the United States. The most significant damage
112 H. Sezen et al. / Engineering Structures 25 (2003) 103–114

Fig. 13. Building collapse due to failure of beam-column joints.

umns shown (Fig. 15b) failed and shortened. These


components displaced out of the plane of the facade as
seen in Fig. 15b. Another example of damage to beams
and columns in a dual wall-frame building is shown in
Fig. 16. No cracks were observed in the shear wall, but
the right end settled approximately 500 mm due to bear-
ing failure of the supporting soils. Although the shear wall
was likely sufficiently stiff to protect the nonductile
frame, the rotation at the base of the shear wall and the
settlement of the footings beneath the moment-frame col-
umns contributed to the failure of the first-story columns.
Blade columns or narrow shear walls were often con-
structed near stairwells (Fig. 17). These walls or blade
columns were detailed similarly to regular moment-
Fig. 14. Damage to new moment-frame beam-column joint.
frame columns with light transverse reinforcement with
90-degree hooks and no cross ties. The damage shown
observed by the team in a dual wall-frame building is in Fig. 17 is similar to that observed in moment-frame
shown in Fig. 15. The wall and first-storey exterior col- columns.

Fig. 15. Collapsed dual wall-frame building in Adapazari, (a) view of collapsed 5-storey building, (b) failure of the shear wall and perimeter col-
umns.
H. Sezen et al. / Engineering Structures 25 (2003) 103–114 113

10. Conclusions

Much has been written in the aftermath of the August


17, 1999, earthquake about the poor quality of residential
and commercial construction in the epicentral region.
The detailing and quality of the residential construction,
perhaps most of it not rigorously engineered, was poor
by modern US and Turkish standards. The reconnais-
sance team also documented many collapses of commer-
cial reinforced concrete moment frame buildings that
were likely carefully engineered. Both poor construction
practices and the continued use of nonductile seismic
detailing were the primary reasons for most of the build-
ing collapses. Shear reinforcement was lacking in most
damaged columns observed by the authors. In contrast
with the code design provisions, common use of 90-
degree hooks for transverse reinforcement reduced the
lateral strength and confinement of columns. Short col-
Fig. 16. Damaged wall frame building due to ground failure and umns, poor detailing in beam-column joints, strong-
wall rotation. beam weak-columns, and use of inconsistent unre-
inforced masonry infill walls were among other reasons
for the widespread destruction in the region affected by
the earthquake. For the most part, buildings with shear
walls survived with limited or no damage.

Acknowledgements

The work described in this paper made use of the


Pacific Earthquake Engineering Research Center shared
facilities supported by the Earthquake Engineering Cen-
ters Program of the National Science Foundation under
Award Number EEC-9701568. This support is grate-
fully acknowledged.

References

[1] Sezen H, Elwood KJ, Whittaker AS, Mosalam KM, Wallace JW,
Stanton JF, Structural Engineering Reconnaissance of the August
17, 1999 Kocaeli (Izmit), Turkey Earthquake. PEER 2000/09.
Technical Report. Berkeley, CA.: Pacific Earthquake Engineering
Research Center, University of California, CA December 2000.
http://nisee.berkeley.edu/turkey.
[2] Aschheim M, (coordinator), Gulkan P, Sezen H, (major
contributors) Chapter 11: Performance of Buildings, in Kocaeli,
Turkey earthquake of August 17, 1999 Reconnaissance Report.
Earthquake Spectra. Supplement A to Volume 16. December 2000.
pp. 237–279
[3] Scawthorn CR, editor. The Marmara, Turkey earthquake of August
17, 1999: Reconnaissance Report, Technical Report MCEER-00-
0001. Buffalo, N.Y.: Multidisciplinary Center for Earthquake
Engineering Research, State University of New York, NY.
March 2000.
[4] Ministry of Public Works and Settlement. Specification for struc-
tures to be built in disaster areas. Government of Republic of Tur-
key. 1975.
Fig. 17. Shear failure of wall/blade column. [5] Ministry of Public Works and Settlement. Specification for struc-
114 H. Sezen et al. / Engineering Structures 25 (2003) 103–114

tures to be built in disaster areas; Part III—earthquake disaster building code, Volume 2, Structural Engineering Design Pro-
prevention. Government of Republic of Turkey. 1998. visions. Whittier, CA, 1997.
[6] Turkish Standards Institute. TS-500 Building Code Requirements [9] Gulkan P. Chapter 15: Building code enforcement prospects: the
for Reinforced Concrete. Ankara, Turkey, 1985 (in Turkish). failure of public policy, in Kocaeli, Turkey earthquake of August
[7] ACI American Concrete Institute. Building code requirements for 17, 1999 Reconnaissance Report. Earthquake Spectra. Supplement
structural concrete (ACI 318-95). Farmington Hills, MI, 1995. A to Volume 16. December 2000. p. 237–79.
[8] ICBO International conference of building officials. Uniform

You might also like