Content uploaded by Ahmed Ghobarah
Author content
All content in this area was uploaded by Ahmed Ghobarah on Feb 04, 2015
Content may be subject to copyright.
The August 17, 1999, Kocaeli (Turkey)
earthquake — damage to structures
Murat Saatcioglu, Denis Mitchell, René Tinawi, N. John Gardner,
Anthony G. Gillies, Ahmed Ghobarah, Donald L. Anderson, and David Lau
Abstract: The 1975 Turkish code provisions are first reviewed to provide the background for design and detailing of
structures prior to the earthquake. The performance of reinforced concrete and masonry structures is described indicat
-
ing many of the deficiencies in design, detailing, and construction execution. The behaviour of precast concrete struc
-
tures, steel structures, and industrial facilities is also presented. The provisions of the 1997 Turkish building code are
summarized and a description of new construction provides evidence of both excellent and poor construction practice.
Some examples of retrofitting of damaged structures soon after the earthquake are also presented.
Key words: seismic design, earthquake, Kocaeli, structures, codes, concrete, precast concrete.
Résumé : Les clauses du Code turc de 1975 sont, en premier lieu, évaluées afin de fournir les bases pour la concep
-
tion et la documentation de structures construites avant le tremblement de terre. La performance de structures en béton
armé et en maçonnerie est décrite, indiquant plusieurs des déficiences dans la conception, la documentation et
l’exécution de la construction. Le comportement de structures en béton préfabriquées, de structures en acier et
d’installations industrielles est aussi présenté. Les clauses du Code du bâtiment turc de 1997 sont résumées, et un des
-
cription de nouvelles constructions fournit les preuves de bonnes et mauvaises pratiques de construction. Quelques
exemples de remise en état de structures endommagées, suivant le tremblement de terre, sont aussi présentés.
Mots clés : conception paraséismique, tremblement de terre, Kocaeli, structures, code, béton, béton préfabriqué.
[Traduit par la Rédaction] Saatcioglu et al. 737
Introduction
The magnitude, M
b
, 7.4 earthquake that occurred along
the North Anatolian Fault at 3:01 a.m. on August 17, 1999,
resulted in over 20 000 deaths, 50 000 injured, and over $30
billion in damage. Information on lifelines and earthquake
preparedness can be found in Gillies et al. (2001). Figure 1
illustrates the concentration of damage in regions along the
fault line, particularly in Izmit, Adapazari, Sapanca, Golcuk,
and Yalova. Although Avcilar, a western suburb of Istanbul,
is at a considerable distance from the fault, it also experi-
enced damage due to the presence of soft soil conditions. A
total of 140 000 structures collapsed, which represent 7.7%
of the building stock in the epicentral region, while 28.6%
of buildings suffered light to moderate damage.
This paper describes different types of structural damage
that were observed during several separate site visits by the
authors. The reasons for the poor performance of some
structural systems are discussed and placed into context with
the Turkish codes and construction practice.
Turkish seismic code requirements
Many reinforced concrete buildings that suffered damage
during the earthquake were designed in a period when the
1975 edition of the Specifications for Structures to be Built
in Disaster Areas, which had been issued by the Ministry of
Reconstruction and Resettlement of the Government of Tur
-
key (Ministry of Public Works 1975), was in effect. In this
code, emphasis is placed on reinforced concrete frame build
-
ings with masonry infills, since this type of structural system
dominated the building inventory in the earthquake-stricken
areas.
1975 building code
The formula for the design base shear in the Turkish code
resembles those in the North American building codes of the
era. The design base shear, F,isgivenas
Can. J. Civ. Eng. 28: 715–737 (2001) © 2001 NRC Canada
715
DOI: 10.1139/cjce-28-4-715
Received September 21, 2000. Revised manuscript accepted
May 29, 2001. Published on the NRC Research Press Web
site at http://cjce@nrc.ca on July 31, 2001.
M. Saatcioglu
1
and N.J. Gardner. Department of Civil
Engineering, University of Ottawa, Ottawa, ON K1N 6N5,
Canada.
D. Mitchell. Department of Civil Engineering, McGill
University, Montreal, QC H1A 2K6, Canada.
R. Tinawi. Département de génie civil, École Polytechnique,
Montréal, QC H3C 3A7, Canada.
A.G. Gillies. Department of Civil Engineering, Lakehead
University, Thunder Bay, ON P7B 5E1, Canada.
A. Ghobarah. Department of Civil Engineering, McMaster
University, Hamilton, ON L8S 4L7, Canada.
D.L. Anderson. Department of Civil Engineering, The
University of British Columbia, Vancouver, BC V6T 1Z4,
Canada.
D. Lau. Department of Civil and Environmental Engineering,
Carleton University, Ottawa, ON K1S 5B6, Canada.
Written discussion of this article is welcomed and will be
received by the Editor until December 31, 2001.
1
Author to whom all correspondence should be addressed
(e-mail: murat@genie.uottawa.ca).
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Thursday, July 26, 2001 4:58:41 PM
Color profile: Generic CMYK printer profile
Composite Default screen
[1] F = CW
where W is the total structural weight and C is the seismic
coefficient defined as
[2] C = C
o
KSI
where C
o
is the seismic zone coefficient (C
o
= 0.1 for seis
-
mic zone 1, which is the zone of the area affected by the
earthquake); K is the coefficient related to structural type
(K = 0.8 for ductile concrete frames with unreinforced ma
-
sonry infills and K = 1.5 for non-ductile frames with
unreinforced masonry infills); and S is the spectral coeffi
-
cient, a function of the fundamental period of the structure
and soil type, and is equal to the inverse of the period (T) for
soft and alluvial soils with a high water table, similar to the
soil conditions encountered in the disaster area. The maxi
-
mum value of coefficient S is limited to 1.0; I is the impor
-
tance factor (I = 1.0 for ordinary residential, office, and
industrial buildings).
Therefore, for the majority of building structures with
fundamental periods of less than 1.0 s, the design base shear
varies between 8% and 12% of the building weight, depend
-
ing on whether the concrete frame is designed to be ductile
or non-ductile, respectively.
The specifications provide provisions to incorporate the
effects of non-structural components on structural elements
to prevent the creation of “short columns,” which have
behaved poorly in past earthquakes. The non-structural ele
-
ments are required to be separated from the frame if the
storey drift exceeds 0.25%.
1975 design and detailing requirements
Concrete frame members are required to be confined at
the ends, in the potential plastic hinge regions. Contrary to
observations of construction practice, the code required col
-
umns to be confined by closely spaced transverse reinforce
-
ment. The minimum volumetric ratio of rectilinear and spiral
transverse reinforcement is specified to be the greater of
1.0% or 0.12 times the ratio of the concrete compressive
strength to the yield of the transverse reinforcement. The
spiral pitch is limited to one-fifth of the column core or
80 mm, whichever is less. The hoops are required to be
placed with a spacing not to be less than 50 mm and not to
exceed 100 mm. The minimum tie diameter is specified to
be 8 mm. The hoops are expected to have 135° bends with
10 bar diameter extensions into the confined core. The mid
-
dle portions of the columns between the confined ends are
permitted to have reduced amounts of transverse reinforce
-
ment, with maximum spacing limits of one-half the column
dimension, 200 mm, or 20 times the smallest longitudinal
bar diameter, whichever is smaller. Hoops are required at the
ends of beams, within a distance of at least twice the beam
© 2001 NRC Canada
716 Can. J. Civ. Eng. Vol. 28, 2001
Fig. 1. Fault locations and regions of structural damage.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Thursday, July 26, 2001 4:58:45 PM
Color profile: Generic CMYK printer profile
Composite Default screen
depth, with a maximum spacing limit of one quarter of the
beam depth. Beam–column joints are to be designed for
joint shear, requiring transverse reinforcement in the joints.
Under no circumstances is the joint shear reinforcement al
-
lowed to be less than that needed in the column middle re
-
gion.
The beams are designed for shear under the effects of
gravity loads and end moments due to the earthquake. The
spacing of transverse shear reinforcement is limited to the
beam width or one half of the beam depth. The longitudinal
beam reinforcement is required to be continuous at the top
such that it is not reduced below one-quarter of the larger
amount required at either end near the support. Bottom bars
are required at the ends, near the supports, such that the area
of bottom reinforcement is not less than one-third of the area
of the top reinforcement. Additional requirements are speci
-
fied pertaining to reinforcement splicing and development.
Performance of reinforced concrete
structures
The predominant structural system used in Turkey con
-
sists of reinforced concrete frames with masonry infills.
Concrete, which is locally available, is generally preferred
over other construction materials for economic reasons. The
majority of concrete is used for cast-in-place construction,
with an increasingly larger percentage being ready-mix con
-
crete. Precast concrete construction is popular for industrial
buildings. Concrete shear walls have gained greater popular
-
ity only in recent years.
Reinforced concrete frame buildings
The majority of collapses during the earthquake were at
-
tributed to the poor performance of reinforced concrete
frames and masonry infill walls. Buildings with 4–6 storeys
suffered the heaviest damage, inflicting most of the casual
-
ties. Structures close to the region of faulting were subjected
to very high accelerations and velocities, resulting in very
high seismic demands.
Inspection of collapsed and damaged buildings revealed
that very little or no aseismic design had been implemented
during the design and construction of reinforced concrete
frame systems. It has been generally acknowledged that
there has been very poor regulatory control over both struc
-
tural design and construction. It was clear that the structural
layouts used were susceptible to very high drift demands due
to lack of proper lateral load resisting systems and extensive
presence of soft storeys. The high seismic demands became
increasingly critical due to the amplification of ground mo
-
tion by soft soil. The only mechanism of defence for such
structures with inadequate lateral load resisting systems is
the ability of the structural members to undergo inelastic de
-
© 2001 NRC Canada
Saatcioglu et al. 717
Fig. 2. Different levels of contribution of masonry infills to reinforced concrete frame responses: (a) limited masonry damage; (b)ex
-
tensive damage to masonry, no apparent distress in frames; (c) total masonry damage with some distress in first-storey columns; (d)to
-
tal damage to masonry and structural collapse.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Monday, July 30, 2001 10:44:34 AM
Color profile: Generic CMYK printer profile
Composite Default screen
© 2001 NRC Canada
718 Can. J. Civ. Eng. Vol. 28, 2001
Fig. 3. Examples of soft storey reinforced concrete buildings.
Fig. 4. Use of “Asmolen” one-way slab system and resulting strong beam and weak columns: (a) “Asmolen” concrete slab system and
(b) strong-beam weak-column connection.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Thursday, July 26, 2001 4:59:18 PM
Color profile: Generic CMYK printer profile
Composite Default screen
formations without experiencing brittle failures. Unfortu
-
nately, all of the frame buildings inspected lacked
appropriate seismic design and detailing practices, which
could have provided the required ductility and energy ab
-
sorption. Proper design practices were missing in spite of the
seismic design requirements of the 1975 Turkish code.
Causes of damage can be viewed under two categories:
(i) factors contributing to increased seismic demands and
(ii) factors contributing to reduced ductility and energy ab
-
sorption.
Factors contributing to increased seismic demands
Lateral bracing for reinforced concrete frame structures
was provided by unreinforced brick and (or) concrete ma
-
sonry walls. The brick masonry used was often in the form
of unreinforced hollow architectural blocks. During the
earthquake, these walls were able to participate in lateral
load resistance to varying degrees and were often damaged
prematurely, developing diagonal tension and compression
failures or out-of-plane failures. The degree of lateral load
resistance depended on the amount of masonry used and the
framing system provided. In contrast to modern moment re
-
sisting frames of North American practice, the use of light
partitions, such as dry walls, was not common in the earth
-
quake-stricken areas. Instead, masonry was used extensively
for interior partitioning, as well as exterior enclosure of
buildings, increasing wall-to-floor area ratios. Therefore, in
spite of lower strength and expected brittleness of this type
of masonry walls, the frames did benefit somewhat from
such extensive use of masonry until the threshold of elastic
behaviour was exceeded. Beyond the failure of brittle ma-
sonry, there was no lateral load resisting system with suffi-
cient stiffness to control lateral drift, thereby resulting in
high drift demands on the frame members. Figure 2 illus-
trates different degrees of masonry failure, resulting in par-
tial damage, severe damage, and collapse of the frame
structure.
Most buildings in Turkey are designed to have commer
-
cial space at the first-storey level, generally used for stores,
as illustrated in Fig. 3. Furthermore, many buildings have a
larger floor area above the ground-storey level. Both of these
features result in soft storeys at the street level and added
mass above the ground-storey level, placing excessive defor
-
mation demands on the highly critical first-storey columns.
Another reason for increased seismic demands on struc
-
tures was the common use of a structural slab system called
Asmolen. This is a one-way slab system, consisting of con
-
crete joists with masonry units placed in between the joists,
forming a deep structural slab system, as illustrated in
Fig. 4a. Columns used are usually smaller in size, resulting
in flexible and weak vertical elements relative to the adjoin
-
ing horizontal members at beam–column joints. This system,
totally in violation of the strong-column weak-beam design
philosophy, places heavy deformation demands on the col
-
umns, especially at the first-storey level, increasing storey
drifts and forcing hinging to occur in the columns. Figure 4b
shows column damage resulting from a strong-beam weak-
column connection.
Factors contributing to reduced strength and deformability
During the seismic response, the failure of brittle masonry
walls placed a heavy demand on the first-storey columns of
multistorey buildings. The columns sustained heavy damage
mostly because of lack of sufficient transverse reinforce-
ment. The transverse reinforcement consisted of 8 mm diam-
eter smooth reinforcement, generally placed at 300 mm or
wider spacing. In some buildings some of the ties were left
out as illustrated in Fig. 5. The ties did not appear to be suf
-
ficient either in terms of amount or detailing. This resulted
in widespread column shear failures as illustrated in Fig. 6.
In the majority of cases, the transverse reinforcement was
limited to perimeter ties with 90° hooks. Columns that were
subjected to heavy axial compression and flexural compres
-
sion resulted in the crushing of concrete due to lack of con
-
finement. Figure 7 shows additional examples of column
damage caused by lack of sufficient transverse reinforce
-
ment, this time resulting in compression crushing rather than
diagonal tension failures. First-storey column failures ac
-
counted for the majority of building collapses and hence a
significant portion of the overall casualties.
The lack of transverse reinforcement was also observed in
monolithic beam–column connections. Beam–column con
-
nections in the majority of buildings did not contain any
transverse reinforcement, suggesting that joint shear design
was never a consideration in these buildings. Figure 8 illus
-
trates damage to beam–column connections lacking joint re
-
inforcement.
Deformation capacities of some structural elements were
impaired because of unintended interference of non-
structural elements with the structure. As masonry walls par
-
ticipated in lateral load resistance of the framing system,
short-column effects were created around window and other
openings. Columns not designed for the increased shear as
-
© 2001 NRC Canada
Saatcioglu et al. 719
Fig. 5. Lack of transverse reinforcement in concrete columns.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Monday, July 30, 2001 10:44:42 AM
Color profile: Generic CMYK printer profile
Composite Default screen
© 2001 NRC Canada
720 Can. J. Civ. Eng. Vol. 28, 2001
Fig. 6. Lack of transverse shear reinforcement and resulting diagonal tension failures.
Fig. 7. Lack of column confinement reinforcement and resulting crushing.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Thursday, July 26, 2001 4:59:49 PM
Color profile: Generic CMYK printer profile
Composite Default screen
© 2001 NRC Canada
Saatcioglu et al. 721
Fig. 8. Lack of joint reinforcement and resulting failures in beam–column connections.
Fig. 9. Interference of non-structural elements with lateral load resisting system: (a) short column effect and (b) interference of stair-
way landing slabs.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Thursday, July 26, 2001 5:00:19 PM
Color profile: Generic CMYK printer profile
Composite Default screen
© 2001 NRC Canada
722 Can. J. Civ. Eng. Vol. 28, 2001
Fig. 10. Frame shear wall building with lightweight concrete masonry blocks: (a) apartment building with narrow shear walls and
(b) diagonal tension cracks in shear wall.
Fig. 11. Excessive damage caused by shear in older shear walls.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Thursday, July 26, 2001 5:00:50 PM
Color profile: Generic CMYK printer profile
Composite Default screen
sociated with reduced unsupported height suffered brittle
shear failures as depicted in Fig. 9a. In some buildings, the
landing slabs of staircases were connected to columns, and
either applied unexpected lateral forces or caused short col-
umn effects as shown in Fig. 9b.
Additional problems were observed, associated with irreg
-
ularities in structural elements, reducing deformability of el
-
ements. Although most floor plans had symmetric layouts,
there were cases where torsional effects created by asymme
-
try had adverse effects. Columns with plan offsets suffered
damage. Two cases of cranked columns were found to sur
-
vive the earthquake without significant damage, mainly be
-
cause they were overdesigned, but with some signs of
distress, requiring retrofitting.
Reinforced concrete shear wall structures
Use of reinforced concrete shear walls is limited in Tur
-
key, especially in older buildings. A number of buildings
were found with walls having cross sections with relatively
small aspect ratios, which in some cases resembled elon
-
gated rectangular columns. These buildings performed rea
-
sonably well. Figure 10 illustrates an apartment complex
under construction with lightweight concrete masonry units
and this type of walls. These buildings survived the earth
-
quake with minor damage to the structural framing system,
but major damage to the masonry. Figure 10b shows a shear
wall in this complex that developed diagonal shear cracks
wide enough to suggest some yielding of the reinforcement,
survived the earthquake, and saved the structure. There were
other shear wall buildings with older and significantly lower
quality concrete. Although the concrete in these walls was
damaged extensively, the shear walls did save the structures
from collapsing. This is illustrated in Fig. 11.
Properly designed and detailed shear wall structures per-
formed exceptionally well even in regions close to the fault.
Figure 12a shows the overall view of an apartment building
within the Tupras Oil Refinery in Izmit, which performed
well except for minor damage at a cold joint in one of these
walls due to poor concrete placement (see Fig. 12b).
Performance of precast concrete structures
A significant number of precast concrete industrial struc
-
tures are located in the epicentral area. In general, the pre
-
cast structures did not fair well during this earthquake.
Figure 13 provides an interesting contrast between two
structures that totally collapsed and an adjacent structure
that remained standing. These three structures are located in
the industrial park region of Yeni Sanayi about 2 km from
the fault. The two structures that collapsed were still under
construction with only bare frames, without any roof dia
-
phragms and without the side panels. These unfinished struc
-
tures had inadequate connections between the columns and
the beams and many of the columns failed at their bases as
shown in Fig. 13c. The third structure was identical to the
ones that collapsed, but the structural wall panels and roof
diaphragms were completed. This structure underwent only
slight hinging at some columns near their bases, with some
evidence of connection distress at one beam–column joint
(see Fig. 13d).
© 2001 NRC Canada
Saatcioglu et al. 723
Fig. 12. Frame wall building in Izmit without significant damage: (a) exterior view and (b) structural wall with concrete placement
problem.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Thursday, July 26, 2001 5:01:02 PM
Color profile: Generic CMYK printer profile
Composite Default screen
Figure 14a shows one form of construction for precast in
-
dustrial buildings that was very common in Turkey. This
construction consists of precast columns with short cantile
-
vers, which support cranked beams and, in turn, support pre
-
cast stringers. The precast columns are fixed at their bases,
since they are grouted into sockets in the foundation perime
-
ter wall footing as shown in Fig. 14b. The damage at the
base of the central columns is partly due to the lack of dia
-
phragm action in this unfinished structure. Figure 15a illus
-
trates the complete collapse of a similar precast structure
that was completed before the earthquake. The main reasons
for the collapse are the following:
(i) The presence of partial height masonry infills led to fail
-
ure in some columns at the level of the top of the dam
-
aged infills where the columns contained splices of the
vertical bars.
(ii) The columns contained inadequate confinement (see
Fig. 15b).
© 2001 NRC Canada
724 Can. J. Civ. Eng. Vol. 28, 2001
Fig. 13. Two precast concrete industrial buildings under different stages of construction: (a) incomplete building with frame elements
only; (b) building with bracing walls in place; (c) shear distress at base of column; (d) distress at beam–column joint.
Fig. 14. Typical precast industrial building: (a) frame with short cantilevers supporting cranked beam and (b) column socket at base.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Thursday, July 26, 2001 5:01:28 PM
Color profile: Generic CMYK printer profile
Composite Default screen
© 2001 NRC Canada
Saatcioglu et al. 725
Fig. 15. Collapse of completed precast industrial building: (a) influence of partial height masonry infills; (b) inadequate confinement in
columns; (c) thin corrugated fibre concrete diaphragm.
Fig. 16. Severe damage to frames of precast warehouse structure: (a) large deformation of columns and (b) flexural hinging at column
base.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Thursday, July 26, 2001 5:01:54 PM
Color profile: Generic CMYK printer profile
Composite Default screen
(iii) The diaphragm consisted of very thin corrugated fibre-
concrete panels with very flexible clip-on fasteners as
shown in Fig. 15c.
(iv) The following connection failures took place: cantile-
ver–column, cantilever–beam, and beam–stringers con-
nections.
Figure 16a shows the severe damage to another industrial
structure. Figure 16b shows the significant flexural hinging
that occurred at the base of one of the interior 400 ×
450 mm columns in an industrial precast building. The verti
-
cal reinforcement consisted of 4–20 mm diameter bars. The
column ties consisted of 8 mm diameter deformed bars,
spaced at 150 mm. The ties were anchored with 135° bends
and had free end extensions of 60 mm. Other columns along
the centre line were 260 × 510 mm, which suffered flexural
hinging about their weak axis. Another deficiency of this
type of construction is the connection between the double
cantilever heads to the interior columns. Figure 17a shows a
double cantilever head, in a building in the same industrial
park, which has collapsed as a result of failure of the bolted
connection to the column. Figure 17b shows the connection
bolts for the double cantilever head after the failure. A dif
-
ferent structural system in the same industrial park utilized
pretensioned long-span tapered beams which fitted into slots
formed at the top of the interior and exterior columns. Fig
-
ure 18a shows the failure of a precast industrial building
which experienced large lateral drift at the roof level. Fig
-
ure 18b shows an interior column where loss of support for
the beam has occurred due to the extremely large lateral drift
at the top. Figure 18c shows the significant flexural hinging
over a height of about 800 mm from the base of one of the
interior columns. The 450 × 450 mm column had 2–20 mm
diameter vertical bundled bars in each corner. The column
ties consisted of 6 mm diameter plane bars at 100 mm spac
-
ing. The ties were anchored with 90° bends with 120 mm
free end extensions. Although the column behaved in a very
ductile manner, the lack of proper diaphragm action and in-
adequate connections between the beams and the columns
contributed to the collapse.
Figures 19a and 19b illustrate another type of failure that
occurred in precast columns within the industrial area be-
tween Izmit and Adapazari. The complete failure of this un-
finished structure was due to the absence of diaphragm
action, failure of the double cantilever to column connec
-
tions (see Fig. 19a) and mid-height column failures (see
Fig. 19b). Figure 19c shows a mid-height failure. The verti
-
cal reinforcement of the 260 × 510 mm rectangular columns
consisted of 12–15 mm diameter continuous bars and 4–
20 mm diameter bars that were curtailed at the top of the
failure zone. The transverse reinforcement consisted of
6 mm diameter plain ties at a spacing of 175 mm. The trans
-
verse hoops were anchored with 135° bends into the core.
Although there were many examples of precast buildings
with poor design, detailing, and construction, there were also
large numbers of precast industrial buildings in the
epicentral area that performed extremely well, as shown in
Figs. 20a and 20b.
Steel structures
Apart from the special facilities such as oil refineries,
there were very few steel industrial structures. Figure 21
shows a one-storey steel structure in an industrial park in the
epicentral region. The I-shaped beams and columns were
formed with galvanized light-gauge, cold-formed steel. The
steel structure had tension-only bracing in the roof and ten
-
sion-only bracing every fourth bay along the side walls. This
structure with its partially completed roof diaphragm did not
suffer any damage.
© 2001 NRC Canada
726 Can. J. Civ. Eng. Vol. 28, 2001
Fig. 17. Collapse of industrial building: (a) failure of double cantilever and (b) failure of connection.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Thursday, July 26, 2001 5:02:06 PM
Color profile: Generic CMYK printer profile
Composite Default screen
Industrial facilities
Many industrial facilities suffered damage and interrup
-
tion of operations as a result of the earthquake. The most
significant damage occurred at the Tüpras refinery (Izmit),
which supplies 96% of the country’s requirement of petro
-
leum products. Three separate fires started due to the earth
-
quake, one of which took several days to bring under control
with substantial fire damage. Figure 22 illustrates the extent
of tank damage in one of the four tanks, inflicted by the fire.
The main fire was caused by a pipe rupture during the earth
-
quake, exposing oil products at extreme temperatures to oxy
-
gen. One of the fires started because of the collapse of a
115 m high reinforced concrete chimney that ruptured pipes
carrying hot H
2
S gas which only needed oxygen to ignite, as
illustrated in Figs. 23 and 24. The failure occurred at about
one-third of the height from the base due to the additional
support provided by connecting pipes at this level. The
refinery suffered ruptures of many pipes due to large dis
-
placements. Over 60 main pipeline systems were broken.
Damage to the liquid storage tanks in the refinery was exten
-
sive. About 80% of the tanks suffered from various types of
failure such as buckling of the supports of elevated tanks,
buckling of the tank walls, and elephant foot buckling of the
tank wall near the base, as depicted in Figs. 25–28. Many
tanks suffered damage to the floating roof system with the
consequence of liquid spillage with associated increased fire
hazard. Several tanks were destroyed by the fire. An inter
-
esting mode of failure of several tanks is the bulging in
-
wards and outwards of the roof causing internal suction
resulting in inward deformation of the walls. One of the
tanks that suffered roof damage had a diameter of 27.4 m
and a height of 12.8 m and was filled to the 12 m level at
the time of the earthquake. Other civil works in the refinery
© 2001 NRC Canada
Saatcioglu et al. 727
Fig. 18. Failure of industrial precast building: (a) evidence of large displacements at tops of columns; (b) loss of beam support in col
-
umn slotted connection; (c) severe hinging at base of column.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Thursday, July 26, 2001 5:02:24 PM
Color profile: Generic CMYK printer profile
Composite Default screen
© 2001 NRC Canada
728 Can. J. Civ. Eng. Vol. 28, 2001
Fig. 19. Complete collapse of unfinished precast structure:
(a) failure of double cantilevers; (b) mid-height column failures;
(c) close-up of column failure.
Fig. 20. Examples of precast structures that performed well.
Fig. 21. Cold-formed steel structure.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Thursday, July 26, 2001 5:02:52 PM
Color profile: Generic CMYK printer profile
Composite Default screen
facility suffered severe damage, including the port facilities
and pipe supports, as shown in Figs. 29 and 30, and the col
-
lapse of one of the administration buildings and the heavy
damage of others. Other less spectacular damage to indus-
trial facilities includes collapse of structures, partial roof
collapse, failure of tanks, and large displacement of equip-
ment. Another industrial facility that suffered minor damage
was a fertilizer factory just west of Izmit. The factory sur-
vived the earthquake without significant damage, other than
minor hairline cracks in some of the concrete walls and mi-
nor soil failure at its harbor. The operation had to be
stopped, however, because a large kiln, which is shown in
Fig. 31, was forced off its support.
New seismic code requirements and
construction practice
1997 building code
The 1997 Turkish code (Ministry of Public Works 1997),
with subsequent amendments in 1998, distinguishes between
four seismic zones, with zone 1 being the highest with an ef
-
fective ground acceleration coefficient, A
0
, of 0.40. All of
the damaged areas reported in this paper are located in zone
1. There are two methods for obtaining design force levels,
the equivalent seismic load method and the mode superposi
-
tion method. The design base shear, V
t
,isgivenas
[3]
V
AISTW
RT
AIW
t
a
0.1
01
1
0
()
()
where A
0
is the effective ground acceleration coefficient; I is
the importance factor (e.g., 1.5 for post-disaster buildings;
1.4 for schools; 1.2 for sports facilities, cinemas, and thea
-
tres; 1.0 for other buildings); R(T
1
) is the seismic load reduc
-
tion factor; S(T
1
) is the spectrum coefficient; and W is the
total building weight including contributing live loads (e.g.,
0.3 times the live load for residential and office buildings).
The spectrum coefficient, S(T
1
), depends on the natural
period of the building, T
1
, as well as the local site class as
given.
The first natural period of the building is given by
[4]
TCH
1 tN
0.75
where C
t
is the coefficient used in calculating the period
(0.07 for reinforced concrete frames or eccentrically braced
steel frames; 0.08 for steel frames; 0.05 for all other frame
buildings).
For structures containing reinforced concrete structural
walls, the coefficient, C
t
, is calculated from
© 2001 NRC Canada
Saatcioglu et al. 729
Fig. 22. Tank damage caused by fire.
Fig. 23. Chimney failure at Tupras Oil Refinery.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Monday, July 30, 2001 10:44:58 AM
Color profile: Generic CMYK printer profile
Composite Default screen
[5]
C
AlH
jj
t
wwN
0.075
0.2([ (/)])
/212
where A
wj
is the gross section area of the jth structural wall
in the first storey; H
N
is the total height of the building, m;
l
wj
is the effective length of the jth structural wall in the first
storey, in the direction of the earthquake.
The seismic load reduction factor, R(T
1
), is given as
[6]
RT
RT
T
TT
RT R T T
a
A
A
aA
1.5
1.5
()
()
()
() ( )
0
where R is the structural behaviour factor.
Table 1 gives the values of the structural behaviour fac
-
tors, R, for different types of structural systems. The Turkish
code distinguishes between nominal and highly ductile lev
-
els depending on the reinforcing details.
The 1997 code contains design and detailing requirements
for concrete structures that are comparable to those in the
North American codes.
Shear wall structures under construction in 1999
One very large complex of apartment buildings in Izmit
having 10–20 storeys and located about 5 km from the fault
sustained no visible damage. Figure 32a shows the overall
view of the apartment complex and Fig. 32b shows a 15-
storey structure under construction. Figure 32c shows the
plan view of the structural walls and Fig. 32d shows the
foundation mat. The reinforcement detailing in a typical wall
at the second-storey level is shown in Fig. 32e. The aspects
that contributed to the excellent performance include the fol
-
lowing:
(i) The 700 mm thick, heavily reinforced foundation mat
was founded on firm ground (see Fig. 32d).
(ii) The multiplicity of the symmetrically located shear
walls (see Fig. 32c) in both principal directions pro
-
vided excellent drift control, limited torsional effects,
and provided redundancy in resisting lateral loads.
(iii) The 200 mm thick walls were reinforced with concen
-
trated reinforcement (14–20 mm diameter bars at the
base and 11–12 mm diameter bars at the second-storey
level) at the ends of the walls confined with 8 mm di
-
ameter ties spaced at 200 mm. Two layers of welded
wire fabric (6 mm diameter wires with a vertical and a
horizontal spacing of 150 mm) provided the uniformly
distributed vertical and horizontal reinforcement in the
web regions of the walls. One small variation from
North American practice was the use of welded wire
fabric, the ends of which simply overlapped the con
-
fined cage of vertical bars, rather than having the hori
-
zontal bars project inside of the confined region.
© 2001 NRC Canada
730 Can. J. Civ. Eng. Vol. 28, 2001
Fig. 24. Pipe ruptures caused by chimney failure.
Fig. 25. Wrinkling of tank wall.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Thursday, July 26, 2001 5:03:22 PM
Color profile: Generic CMYK printer profile
Composite Default screen
© 2001 NRC Canada
Saatcioglu et al. 731
Fig. 26. Local buckling of steel tank wall.
Fig. 27. Elephant foot buckling of a tank wall.
Fig. 28. Close-up view of the elephant foot type buckling.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Thursday, July 26, 2001 5:03:46 PM
Color profile: Generic CMYK printer profile
Composite Default screen
(iv) The buildings were not only well designed, but it was
evident that construction and material quality control
practices were strictly enforced.
Frame wall structures under construction in 1999
Figure 33a illustrates the details of a frame wall structure
under construction, which had the following deficiencies:
(i) The columns had insufficient confinement, ties with 90°
bend anchorages, and had lap splices located at the floor
slab levels (see Fig. 33b).
(ii) The shear walls had insufficient confinement at the ends
of the walls, 90° bend anchorages for the 8 mm diame
-
ter ties (see Fig. 33c).
(iii) The poor construction control and very small concrete
cover provided over the reinforcement is evident in
Fig. 33d.
The poor detailing in this structure represents numerous
violations of the Turkish seismic code requirements.
Frame structures under construction in 1999
A six-storey reinforced concrete cast-in-place frame struc-
ture under construction in Yuvacik is shown in Fig. 34a.
This 3 bay by 4 bay structure has the very common
“asmolen” floor slab system. This structure, which is south
of Izmit and about 5 km from the fault, suffered structural
damage to the columns and joints. The traditional masonry
infill had not been placed before the earthquake. The 250 by
600 mm corner column contained 10–15 mm diameter
smooth vertical bars. These vertical bars were hooked at the
ground floor slab level and the dowel bars coming up from
the basement overlapped the vertical column bars. Hinging
took place just below the hooks. The column ties consisted
of 8 mm diameter smooth bars with 90° bend anchorages
and 60 mm free end extensions. The spacing of the ties was
200 mm. Shear distress occurred in the joints of a corner and
an edge column. There was no joint shear reinforcement in
any of the joints. In addition, insufficient concrete cover had
been provided. These details in this structure provide strong
evidence that there was insufficient inspection and enforce
-
ment of building regulations, despite the well-established
seismic code requirements of 1975 and the more stringent
requirements of 1997.
Precast concrete structures under construction in 1999
Although there were many examples of precast concrete
structures with poor lateral load resisting systems and details
under construction in 1999, there were many examples of
excellent systems under construction. Figure 35 shows the
beam–column connections in a three-storey precast building
under construction near the epicentral region. The hollow
© 2001 NRC Canada
732 Can. J. Civ. Eng. Vol. 28, 2001
Fig. 29. Dislocated pipes along the seashore.
Fig. 30. Failure of pipe supports.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Thursday, July 26, 2001 5:04:01 PM
Color profile: Generic CMYK printer profile
Composite Default screen
core slabs are connected to the edge beams by a pour strip,
which has perimeter diaphragm reinforcement as well as ad
-
equate connection to the beams through protruding stirrups
in the pour strip. This design clearly addresses the need for
diaphragm reinforcement and adequate connection between
the diaphragm and the exterior frame. This form of construc-
tion, which is similar to North American practice, has stiffer,
stronger diaphragms due to the load bearing requirements on
the floors. In contrast, the single-storey precast industrial
structures that suffered damage had very flexible, weakly
connected roof diaphragms, which were designed for very
light gravity loads.
Examples of retrofitting
One month after the earthquake there were several exam
-
ples of retrofitting of damaged structures. Figure 36 shows
the retrofit of a corner column of an eight-storey reinforced
concrete frame structure in Avcilar, west of the airport in Is
-
tanbul. Due to lack of symmetry in the wall system and the
torsional eccentricities, the corner column was subjected to
large displacements. This column was retrofitted at the
ground-storey level with steel angles or channel sections at
the corners of the column and with welded steel batten
plates. This retrofit enhances the strength and stiffness of the
column over its clear height but may lack proper continuity
with the column above and the basement level.
Figure 37 shows severe damage to the ground-storey col
-
umn of a five-storey reinforced concrete frame structure on
Darica Road in Gebze, between Istanbul and Izmit. Shear
failure occurred at the base of the column just above the lap
splice in the vertical reinforcing bars. The 400 by 520 mm
column contained 16–25 mm diameter bars with 8 mm di
-
ameter ties at 200 mm spacing. The ties had 90° bend an
-
chorages. A similar five-storey structure immediately across
the street had similar damage to the ground-storey columns.
Figure 38 shows the retrofitted ground-storey columns. The
columns were retrofitted by encasing the existing columns
with at least 200 mm of reinforced concrete on all sides. The
resulting dimensions were 650 by 850 mm for the edge col
-
umns and 650 by 1080 mm for the corner columns. The
added vertical bars were 14 mm in diameter, while 8 mm di
-
ameter ties were added around the perimeter. The added ties
were spaced at 100 mm near the top and bottom of the col
-
umn, and at 150 mm in the central region. Although this ret
-
rofit strengthened and stiffened the ground-storey columns,
there was no attempt to provide continuity with the structure
above and below the ground storey.
© 2001 NRC Canada
Saatcioglu et al. 733
Fig. 31. Kiln of a fertilizer factory in Izmit, dislocated during the earthquake.
Structural system
Nominal
ductility
High
ductility
1. Cast-in-place reinforced concrete
Frames 4 8
Coupled walls 4 7
Wall structures 4 6
Frames and walls 4 7
2. Prefabricated reinforced concrete
Frames with moment connections 3 6
Single-storey hinged frames fixed
at base
—5
Precast structural walls — 4
Frames with moment connections
and cast-in- place walls
35
3. Structural steel
Frames 5 8
Single-storey frames fixed at base 4 6
Braced frames or cast-in-place
concrete walls
Concentrically braced frames 3 —
Eccentrically braced frames — 7
Reinforced concrete walls 4 6
Frames with braced frames or
reinforced concrete walls
Concentrically braced frames 4 —
Eccentrically braced frames — 8
Reinforced concrete walls 4 7
Table 1. Structural behaviour factors, R.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Thursday, July 26, 2001 5:04:09 PM
Color profile: Generic CMYK printer profile
Composite Default screen
© 2001 NRC Canada
734 Can. J. Civ. Eng. Vol. 28, 2001
Fig. 32. Shear wall structures under construction: (a) overall view of apartment buildings; (b) 15-storey structure; (c) plan view of
structural walls; (d) foundation mat; (e) typical wall reinforcement details.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Thursday, July 26, 2001 5:04:29 PM
Color profile: Generic CMYK printer profile
Composite Default screen
There were examples where epoxy filling of cracks was
observed even for cases of severe shear cracking in columns.
It is noted that the epoxy filling of cracks should be viewed
as a temporary measure, since the main structural deficien
-
cies had not been corrected.
Conclusions
The 1999 Kocaeli earthquake resulted in the collapses and
severe damage to several types of concrete structural sys
-
tems.
Structural deficiencies
The following deficiencies in the different types of struc
-
tures were observed:
Concrete frame structures
•
Unreinforced masonry infill walls suffered brittle failures
and increased the base shear level, demonstrating that
non-ductile frames with brittle infill walls are poor lateral
load resisting systems for earthquakes.
•
Many collapses at the ground-storey level were due to soft
storeys resulting from the commercial usage and the re
-
duced dimensions of ground-storey residential buildings.
© 2001 NRC Canada
Saatcioglu et al. 735
Fig. 33. Frame wall structure under construction: (a) overall view; (b) column reinforcement details; (c) wall reinforcement details;
(d) insufficient concrete cover.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Monday, July 30, 2001 10:45:19 AM
Color profile: Generic CMYK printer profile
Composite Default screen
•
Use of thick slabs resulted in “strong column” – “weak
beam” structures that resulted in column failures.
•
Presence of partial infills created “short columns” that
failed in shear.
•
Lack of beam–column joint shear reinforcement resulted
in many shear failures in joints.
•
Inadequate detailing of structural members.
•
Although unreinforced masonry is a poor structural sys
-
tem, the extensive use of masonry infills probably limited
the drift sufficiently to limit the response of the frame
structures to the elastic range and hence helped to prevent
collapse of some buildings.
Precast concrete structures
•
Inadequate roof diaphragms and inadequate diaphragm
connection in single-storey industrial buildings permitted
large relative lateral displacements of frames.
•
Complete structural collapse was caused by inadequate
beam-to-beam, beam-to-column, and purlin-to-beam con
-
nections.
•
Full-height and partial-height unreinforced masonry infills
resulted in excessive seismic demands on precast concrete
columns.
© 2001 NRC Canada
736 Can. J. Civ. Eng. Vol. 28, 2001
Fig. 34. Concrete frame structure under construction.
Fig. 35. Precast concrete structure under construction.
Fig. 36. Retrofit of corner column.
Fig. 37. Shear failure of ground-storey column.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Thursday, July 26, 2001 5:05:14 PM
Color profile: Generic CMYK printer profile
Composite Default screen
© 2001 NRC Canada
Saatcioglu et al. 737
•
Poor detailing of columns, including inappropriate place
-
ment of splices, inadequate confinement reinforcement,
and poor details of column ties, resulted in many different
types of brittle failures.
Inadequate enforcement of design and construction
It is generally acknowledged by site-visit teams from
many countries, including experts in Turkey, that the prime
factor that led to poor structural performance was the inade-
quate and sometimes nonexistent regulatory enforcement of
both design and construction.
Acknowledgements
The authors gratefully acknowledge all of the assistance
provided by Turkish officials during the three site visits fol
-
lowing the August 17, 1999, earthquake. The support re
-
ceived from Professors Mustafa Erdik and Ozay Yuzugullu
of the Kandilli Observatory and Earthquake Research Insti
-
tute in Bogazici University is gratefully acknowledged. Spe
-
cial thanks are extended to Guy Leduc for his expert
guidance during one of these visits. The research grant fund
-
ing, provided by the Natural Sciences and Engineering Re
-
search Council of Canada (NSERC), made these site visits
possible.
References
Gillies, A., Anderson, D.A., Saatcioglu, M., Tinawi, R., and Mitchell,
D. 2001. The August 17, 1999, Kocaeli (Turkey) earthquake —
lifelines and earthquake preparedness. Canadian Journal of Civil
Engineering, 28. In press.
Ministry of Public Works and Settlement. 1975. Specifications for
structures to be built in disaster areas. Part III — earthquake di
-
saster prevention. Government of the Republic of Turkey.
Ministry of Public Works and Settlement. 1997 (Revised 1998).
Specifications for structures to be built in disaster areas.
Part III — earthquake disaster prevention. Government of the
Republic of Turkey.
Fig. 38. Concrete columns after retrofit.
I:\cjce\Cjce_28\cjce-04\L01-043.vp
Thursday, July 26, 2001 5:05:19 PM
Color profile: Generic CMYK printer profile
Composite Default screen
