You are on page 1of 23

Engineering Failure Analysis 34 (2013) 606–628

Contents lists available at SciVerse ScienceDirect

Engineering Failure Analysis

journal homepage:

Failures of structures during the October 23, 2011 Tabanlı (Van)

and November 9, 2011 Edremit (Van) earthquakes in Turkey
Mucip Tapan a, Mustafa Comert b, Cem Demir b,⇑, Yusuf Sayan c, Kutay Orakcal c, Alper Ilki b
Van Yüzüncü Yıl University, Engineering Faculty, Van, Turkey
Istanbul Technical University, Civil Engineering Faculty, Istanbul, Turkey
Boğaziçi University, Civil Engineering Department, Istanbul, Turkey

a r t i c l e i n f o a b s t r a c t

Article history: The October 23, 2011 Tabanlı and November 9, 2011 earthquakes that hit Tabanlı and Edre-
Available online 27 February 2013 mit districts of Van province in Turkey and their impacts on different types of structures
are studied in this paper. According to United States Geological Survey (USGS), the magni-
Keywords: tudes of these earthquakes, which caused partial or total collapse of numerous buildings
Masonry buildings and more than 600 casualties, were 7.1 and 5.6, respectively. Other than negatively impact-
Reinforced concrete buildings ing all aspects of daily life, the earthquakes remarkably disrupted the economical activities
Structural damage
in the area. This paper summarizes the seismological characteristics of the affected region,
Structural failure
Van earthquakes
the general characteristics of the strong ground motion and the types of structural damage
observed during site investigations. Emphasis is given to the failures and seismic perfor-
mance of different types of structures, through detailed explanation of damage mecha-
nisms. The structural damage levels are observed to be directly related to the extent of
irregularities of the structural system, level of the insufficient quality in the workmanship,
and usage of inadequate building materials. It was also clearly observed that if a minimum
amount of engineering attention had been paid during the construction stages, and the
requirements of design and construction codes had been satisfied, most of the existing
buildings could have sustained the earthquakes without considerable damage.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Van Province is located in the eastern part of Turkey, quite close to the border between Turkey and Iran. The population of
Van city is slightly above one million, according to 2009 census. About half of this population lives in urban areas, while
other half lives in villages. Being one of the most seismically-active zones in Turkey, Van settles close to junction of East Ana-
tolian Fault (EAF) and North Anatolian Fault (NAF), Fig. 1 (left). In addition to these main faults, there are also numerous min-
or faults around Van. An extensive summary of the faults around Van city can be found elsewhere [2]. On October 23, 2011,
an earthquake of magnitude Mw = 7.1–7.2 hit the Tabanlı district of Van province at 13:41 (10:41 GMT) local time. Tabanlı is
approximately 30 km way from the center of Van city. Magnitude and source characteristics of this earthquake, as defined by
various institutions, are given in Table 1. In this table, ML is the local magnitude and Mw is the moment magnitude. This
earthquake affected particularly the Ercisß district, whereas it also caused structural damages, structural collapses and

⇑ Corresponding author. Tel.: +90 212 285 3838; fax: +90 212 285 6587.
E-mail addresses: (M. Tapan), (M. Comert), (C. Demir), (Y. Sayan), (K. Orakcal), (A. Ilki).

1350-6307/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628 607

Fig. 1. Seismotectonics of Turkey (modified from [1]) and region affected by the earthquakes [3].

Table 1
Characteristics of October 23, 2011 Tabanlı (Van) earthquake.

Institution KOERIa USGSb EMSCc DEMPd

Date 10/23/2011 10/23/2011 10/23/2011 10/23/2011
Time 13:41:21 (Local) 10:41:21 (GMT) 10:41:22 (GMT) 10:41:20 (GMT)
Latitude 38.757N 36.628N 38.86N 38.689N
Longitude 43.3602E 43.486E 43.48E 43.465E
Depth 5 km 20 km 10 km 19 km
Magnitude 6.6 (ML), 7.2 (Mw) 7.1 (Mw) 7.2 (Mw) 6.7 (ML), 7.0 (Mw)
Kandilli Observatory and Earthquake Research Institute.
United States Geological Survey.
European-Mediterranean Seismological Centre.
Turkish Prime Ministry – Disaster and Emergency Management Presidency.

Table 2
Death toll after Tabanlı (Van) earthquake.

Location Deaths
Van – City center 100
Ercisß 351
Other 153
Total 604

fatalities at the center of Van city. It should be noted that the Ercisß district is approximately 35 km way from the epicenter
(Tabanlı). The death toll distributed among settlement areas is summarized in Table 2.
Following the first earthquake, another earthquake of magnitude Mw = 5.6 hit the Edremit district of Van province on
November 9, 2011. While Tabanlı is to the north of Van city center, Edremit is located approximately 15 km southwest of
Van city center. Although the magnitude of the second earthquake was smaller than the first one, its impact on structural
damages was extensive, due to relatively high acceleration components in the center of Van city. The damages experienced
during the first earthquake were further increased during the second earthquake. Furthermore, many buildings, which expe-
rienced heavy damage during the first earthquake in Van city center, collapsed totally during the second earthquake. The
characteristics of this earthquake are presented in Table 3. It should be noted that this earthquake caused 40 casualties at
the center of Van city. A map showing the epicenters of two earthquakes, Van city center and the Ercisß district is presented
in Fig. 1 (right). It is important to note that the earthquakes were caused by two different fault ruptures. While the faulting
mechanism of the first earthquake was thrust type, the second one was caused by strike-slip faulting [2]. According to the
Governorship of Van, a total of 28,512 buildings were damaged heavily or collapsed during Tabanlı and Edremit earthquakes
[4]. The distribution of damage to buildings in Van province is presented in Table 4 [4].
On the day of the second earthquake, our reconnaissance team arrived in the region to carry out site investigation and
damage assessment. Subsequently, the affected area was visited five more times to examine the damages and the seismic
performance of different types of structures, such as school buildings, hospitals, hotels, prefabricated structures, historical
608 M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628

Table 3
Characteristics of November 9, 2011 Edremit (Van) earthquake.

Institution KOERIa USGSb EMSCc DEMPd

Date 11/09/2011 11/09/2011 11/09/2011 11/09/2011
Time 21:23:23 (Local) 19:23:33 (GMT) 19:23:33 (GMT) 19:23:33 (GMT)
Latitude 39.429N 38.429N 38.42N 38.447N
Longitude 43.234E 43.229E 43.29E 43.263E
Depth 5 km 5 km 6 km 6 km
Magnitude 5.6 (ML) 5.6 (Mw) 5.6 (Mw) 5.6 (ML)
Kandilli Observatory and Earthquake Research Institute.
United States Geological Survey.
European-Mediterranean Seismological Centre.
Turkish Prime Ministry – Disaster and Emergency Management Presidency.

Table 4
Number of damaged buildings after Tabanlı and Edremit earthquakes [4].

Area No damage Slight damage Moderate damage Heavy damage and collapse
Van (city center) 18,990 25,920 9380 13,500
Van (villages) 1296 3210 294 6090
Ercis (city center) 4958 8064 1428 5158
Ercis (villages) 1617 4750 229 3287
Edremit 1592 1463 58 477
Total 28,453 43,407 11,389 28,512

structures and non-engineered housing units. The visited regions included the centers of Van city, the Ercisß district, and a
large number of villages, which experienced severe damages, particularly after the first earthquake. The observed structural
damages and failures are evaluated and discussed considering the local site conditions and the strong motion data provided
by the national strong motion network of Turkish Prime Ministry Disaster and Emergency Management Presidency (DEMP)
[2], characteristics of structural systems, construction details, and characteristics of construction materials. Furthermore, a
summary of the building inspection system, construction practice, as well as the evolution of the seismic design code and the
obligatory seismic insurance system are also presented in the paper. Similar information can be found elsewhere [5–10] for
other recent earthquakes such as Athens 1999 earthquake in Greece, Izmit 1999 earthquake in Turkey, L’Aquila 2009 earth-
quake in Italy, Christchurch 2011 earthquake in New Zealand and Elazig 2011 earthquakes in Turkey.

2. Seismo-tectonic characteristics of the region

Van is located in the very intensely deformed and seismically active Eastern Anatolian plateau, which is the product of
collision between Arabian and Eurasian plates. The Arabian plate moves northward relative to Eurasia, resulting in lateral
motion of the Anatolian block to the west (Fig. 1 (left)). The motion is accommodated mainly along two active faults, the
North Anatolian Fault (NAF) and the East Anatolian Fault (EAF). These large, translational fault systems extend across much
of central and western Turkey and accommodate the western motion of the Anatolian block, as it is being squeezed by the
converging Arabian and Eurasian plates. In the area of Lake Van and further east, tectonics are dominated by the Bitlis suture
zone (in eastern Turkey) and the Zagros fold and thrust belt (toward Iran) [11], as well as number of small active faults run-
ning parallel to either the NAF or the EAF, which distribute the tectonic movements along these two main faults in this region
The October 23, 2011 Mw = 7.2 Tabanlı and November 9, 2011 Mw = 5.6 Edremit earthquakes occurred at east of the NAF–
EAF junction, on the Bitlis suture zone. The focal mechanism of Mw = 7.2 Tabanlı earthquake was found to be consistent with
oblique-thrust faulting, similar to mapped faults in the region [11]. Although surface fault rupture was not observed, this
earthquake was estimated to correlate well with the Everek fault [2]. The Mw = 5.6 Edremit earthquake was shown to follow
a dominantly strike-slip focal mechanism [3]. According to KOERI [3], this earthquake did not occur on a fault previously
indicated and discussed in the literature except by Kocyigit [12]. Studies on the historic seismicity of the region indicate that
Van and its surroundings have experienced other destructive earthquakes in the past [3].

3. Evaluation of strong ground motion

3.1. Characteristics of strong motion records

For the Mw = 7.2 Tabanlı earthquake with epicentral coordinates of 38.689N–43.465E [2], maximum PGA values were re-
corded at the Muradiye Directorate of Meteorology Building station, which has the smallest epicentral distance of
M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628 609

Repi = 46.6 km. The maximum PGA values recorded during the Tabanlı earthquake at the Muradiye station for NS, EW, and UD
directions are 0.182g, 0.173g, and 0.081g, respectively. For the Mw = 5.6 Edremit earthquake, with epicentral coordinates of
38.447N–43.263E [2], maximum PGA values were recorded at the Van Merkez Directorate of Public Works Building station
(Repi = 17.2 km). The maximum PGA values recorded during the Edremit earthquake at the Van Merkez station in the NS, EW,
and UD directions, are 0.151g, 0.251g, and 0.153g, respectively.
The strong motion records with the highest PGA values were processed in a simple manner using the software Seismo-
Signal [13], via first performing baseline correction on the three components of each acceleration time history record, and
then applying fourth-order Butterworth band filter for removing the undesirable frequency (noise) content in each record.
It should be mentioned that more detailed methods are available for processing of the records. For the Tabanlı earthquake,
the processed time histories yield peak ground velocities of 27.3 cm/s, 14.8 cm/s, and 5.9 cm/s for the NS, EW, and UD com-
ponents, respectively; whereas the maximum peak ground displacement is obtained as 5.5 cm for the NS component. For the
Edremit earthquake, peak ground velocities of 17.3 cm/s, 32.1 cm/s, and 6.3 cm/s are obtained for the NS, EW, and UD com-
ponents, respectively; whereas the maximum peak ground displacement is calculated as 6.8 cm for the EW component.

3.2. Spectral characteristics

The 5% damped acceleration response spectra obtained from the records of the Tabanlı and Edremit earthquakes are com-
pared in Fig. 2, with the 2007 Turkish Earthquake Resistant Design Code (TERDC) [14] spectrum, defined for seismic zone I
(PGA = 0.4 g) and for all soil classes, where Z1 represents the most stiff soil conditions and Z4 the softest. In the TERDC, the
shortest period range for a maximum spectral amplification value of 2.5 is defined between period values of TA = 0.10 s and
TB = 0.30 s for soil class Z1, and the longest period range is defined between period values of TA = 0.20 s and TB = 0.90 s for soil
class Z4. As seen in Fig. 2, none of the records exceed the design spectra. In order to evaluate spectral amplification charac-
teristics of the records, the 5% damped acceleration response spectra of the records are normalized with respect to their cor-
responding PGA, and are compared with the TERDC design (Table 5). Overall, normalized spectra of the records exceed the

900 Bitlis-NS

800 Malazgirt-NS
600 Muradiye-EW
Sa (cm/s2)

400 Ağrı-EW

300 Siirt-EW
100 Z3
0 0.5 1 1.5 2 2.5 3 3.5 4
Period (s)

Van Merkez-NS
900 Edremit-NS

800 Muradiye-NS
600 Van Merkez-EW
Sa (cm/s2)

400 Bitlis-EW

300 Malazgirt-EW
100 Z3
0 0.5 1 1.5 2 2.5 3 3.5 4
Period (s)

Fig. 2. Comparison of acceleration response spectra with design code [14] spectra, for the Mw = 7.2 Tabanlı (top) and Mw = 5.6 Edremit (bottom)
610 M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628

Table 5
Comparison of amplified spectral accelerations with TERDC [14] design spectrum.

Event Station Comp. Max. Sa/PGA Period range

TA (s) TB (s)
Mw = 7.2 Tabanlı earthquake Muradiye NS 3.44 0.27 0.57
EW 2.60 0.29 0.30
Bitlis NS 3.45 0.36 0.73
EW 4.81 0.25 0.76
Malazgirt NS 2.83 0.29 0.72
EW 2.80 0.44 0.60
Ağrı NS 3.00 0.81 1.02
EW 3.71 0.42 0.59
Siirt NS 3.56 0.36 1.00
EW 3.17 0.29 0.75
Mw = 5.6 Edremit earthquake Van Merkez NS 2.96 0.35 0.44
EW 3.13 0.30 0.45
Edremit NS 3.46 0.38 0.50
EW 4.15 0.39 0.66
Muradiye NS 3.99 0.25 0.49
EW 4.47 0.27 0.76
Bitlis NS 5.28 0.30 0.68
EW 3.19 0.28 0.51
Malazgirt NS 3.94 0.28 0.56
EW 3.87 0.44 0.60

amplification defined in the design spectrum, at various period ranges. In particular, records with significant PGA values (e.g.,
Muradiye station records of the Mw = 7.2 Tabanlı earthquake and Van Merkez station records of the Mw = 5.6 Edremit earth-
quake) reach amplification levels of 3–3.5, which exceeds the maximum amplification factor of 2.5 defined in the TERDC.

4. Inspection system, construction practice, seismic design code and obligatory insurance for seismic damage in Turkey

It is important to outline the building inspection system, general construction practice, seismic design code and seismic
insurance system in Turkey; since the seismic failures of existing structures are closely related with these issues in Van as
well as other parts of Turkey.

4.1. Inspection system

Like in many other parts of Turkey, most of the existing buildings in and around the affected region have not been prop-
erly subjected to serious inspection during their design and construction. The most recent building inspection law in Turkey
[15] authorizes independent engineering firms for inspection of design and construction of buildings. According to this law,
the independent inspection firms are responsible for any kind of malpractice that may occur during design or construction
phases. This law was enforced in 2001 for 19 pilot provinces of Turkey after the catastrophic Kocaeli (1999) and Duzce (1999)
earthquakes. However, Van province was not among these 19 pilot provinces, and the new building inspection law was en-
forced in Van province by the beginning of 2011. Nevertheless, prior to this date, officially, the buildings should have been
inspected during all phases of construction by an authorized civil engineer, an architect, a mechanical engineer, and an elec-
trical engineer, according to previous legal inspection practice [16]. However, as mentioned above, unfortunately, most of
the existing so-called engineered buildings were not actually subjected to this kind of inspection and a great majority of
the existing buildings do not have official permit for occupancy. It should also be noted that the new inspection law has some
important drawbacks since the so-called independent inspection firms are selected and paid by the constructor.

4.2. Construction practice

For the ordinary so-called engineered buildings, like in many other parts of Turkey, the owner or the shareholder of the
land is also the constructor of the building. The constructor, personally, is not required to have specific knowledge or expe-
rience on construction. While, the architectural and structural designs are conducted by architects and civil engineers, the
in situ application is generally executed by an uneducated foreman, without proper inspection. The non-engineered build-
ings, on contrary, are designed and constructed entirely by uneducated workers. While the so-called engineered buildings
are quite common in city centers, non-engineered housing units make up the great majority in districts and villages. Unfor-
tunately, properly engineered, inspected buildings with official occupancy permits make up only a small portion of existing
buildings in Van and in other cities of Turkey. In terms of structural systems; the buildings in Van city center are mostly 4–5
story reinforced concrete frame buildings, whereas there are also many unreinforced and confined masonry buildings of 2–3
stories. In Ercisß, while there are a lot of reinforced concrete medium rise buildings (4–5 stories), the ratio of masonry build-
ings is relatively higher with respect to Van city center. In villages, almost all buildings are unreinforced masonry built with
M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628 611

adobe, light-weight concrete blocks or stones, generally by using mud mortar. In terms of usage of construction materials,
unlike big cities like Istanbul, Ankara and Izmir, where the usage of ready-mixed concrete and deformed bars has been com-
mon since the beginning of 1990s, this was not the case until beginning of 2000s in and around Van. The infill walls of rein-
forced concrete frames are either constructed using hollow clay bricks or lightweight hollow concrete blocks, which have
also been used for construction of unreinforced and confined masonry buildings.

4.3. Seismic design code

In Turkey, the first seismic design code was issued in 1940 after the 1939 Erzincan earthquake, which has caused a death
toll of 39,968. After this preliminary seismic design code, the regulations on construction and seismic design have been up-
dated frequently. The last three versions of Turkish seismic design code [14,17,18] cover almost all aspects of the most en-
hanced international seismic design codes of the time for common types of buildings in Turkey. Therefore, it can clearly be
stated that if the buildings have been designed and constructed considering the seismic design code, the damages and fatal-
ities could have been minimized. Details on evolution of Turkish seismic design codes can be found elsewhere [19].

4.4. Obligatory seismic insurance

After the 1999 Kocaeli and Duzce earthquakes, an obligatory seismic insurance system has been established in Turkey.
According to this system, all occupants are required to pay a low amount of annual insurance premium for compensation
of potential structural damages after earthquakes. While this insurance is mandatory according to the law, only about
25% of the housing units in Turkey are insured. In recently developing cities like Van, the ratio of insured housing units is
even less. One important obstacle for more widespread insurance application is the financial support provided by the gov-
ernment to all citizens whose buildings are damaged, without considering whether they have insured their houses or not.
While this kind of approach of governments is understandable in terms of humanitarian purposes and political reasons, such
applications may cause further reduction in the number of insured housing units, and raise questions about the rationale and
need of an obligatory seismic insurance system.

5. Characteristics of the building materials

5.1. Concrete

Van city is quite far from the center of Turkey. Therefore, unlike many cities where utilization of ready-mixed concrete
was common at the beginning of 1990s, ready-mixed concrete has not been used during construction in Van until 2000s. In
addition, the utilization of hand-mixed concrete continued for the construction of some of the low-cost buildings after
2000s. To demonstrate the distribution of concrete quality according to the age of the buildings, a large number of concrete
cores are taken from the reinforced concrete buildings, which have experienced damages to different extents. Accordingly,
while the average concrete strengths are in the range of 10–15 MPa for the buildings constructed before 2005, the average
concrete strength exceeds 25 MPa after 2005, due to widespread usage of ready-mixed concrete. These results are based on
compression tests of 110 concrete cores taken from 22 different buildings.

5.2. Steel reinforcing bars

Like usage of ready-mixed concrete, deformed reinforcing bars have become common in the beginning of 2000s in Van.
Plain round bars have been used for construction of almost all reinforced concrete buildings before 2000s. For assessing the
mechanical characteristics of reinforcing bars used in construction, a large number of samples were taken from the damaged
buildings. The samples included main longitudinal bars and transverse bars of smaller diameter. The mechanical character-
istics of reinforcing bar samples are summarized in Table 6.

5.3. Masonry units

Different masonry blocks and mortars of different quality are used depending on the region and construction year. While
stone and adobe blocks are used more commonly in villages, lightweight hollow concrete blocks are more commonly used
for the construction of masonry buildings in the city center. Hollow core clay bricks, light weight hollow concrete blocks and
aerated concrete blocks are used for the construction of partitioning walls of reinforced concrete buildings. It should be
noted that the contribution of non-structural partitioning walls to the seismic capacity may be important when relatively
older so-called reinforced concrete frame buildings are extremely weak in terms of seismic resistance. Furthermore, it
was shown by several researchers that the presence of non-structural walls can play a role on improving the damping char-
acteristics [20] of the reinforced concrete frames, leading to a reduction in seismic demand. However, the contribution of
these non-structural walls is not dependable because of non-standard construction quality and their poor out-of-plane
612 M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628

Table 6
Characteristics of reinforcing steel bar samples.

Sample Diameter (mm) Yield stress (MPa) Maximum stress (MPa) Rupture strain (%) Date of construction Reinforcement type
1 14 365 432 46 1985 S220
2 14 353 430 37 1985 S220
3 14 274 388 40 1985 S220
4 14 308 396 45 1985 S220
5 8 423 441 39 1985 S220
6 10 355 436 35 1985 S220
7 12 293 452 38 1976 S220
8 12 296 459 41 1976 S220
9 10 264 402 48 1976 S220
10 10 265 414 45 1976 S220
11 8 385 469 38 1976 S220
12 16 366 548 26 1999 S220
13 16 344 541 28 1999 S220
14 14 355 538 35 1999 S220
15 14 413 585 24 1999 S220
16 12 323 477 35 1999 S220
17 12 393 439 32 1999 S220
18 9 381 531 38 1999 S220
19 8 419 550 41 1999 S220
20 16 268 403 29 1992 S220
21 16 342 466 30 1992 S220
22 8 304 457 32 1992 S220
23 8 301 439 32 1992 S220
24 12 351 464 24 1992 S220
25 12 325 460 29 1992 S220

performance. In addition, sometimes the non-structural walls integrated with the structural system can cause an increase in
seismic load demand because of increased overall stiffness of the structural system.

6. Damages to existing structures

6.1. Engineered buildings

A major part of the existing buildings in city center and district centers are medium-rise reinforced concrete frame build-
ings. Like in other parts of Turkey, most of the existing reinforced concrete buildings were not properly constructed in the
earthquake affected region, because of lack of an effective inspection mechanism. Consequently, most of the existing build-
ings have a range of deficiencies to varying extents. Therefore, it is not possible to define the large majority of existing build-
ings as properly engineered construction, even if they have been inspected in theory and approved legally. Therefore, these
buildings are defined as so-called engineered reinforced concrete buildings in this paper.

6.1.1. General characteristics of the so-called engineered reinforced concrete buildings in the region
Before the earthquakes, Van was a rapidly growing city attracting many people from the surrounding cities. Therefore, the
construction of medium-rise buildings (4–8 stories) has been increasing rapidly in the recent years. There are also few higher
buildings, reaching up to 14 stories. The buildings generally have a basement floor, and although ground conditions are gen-
erally poor, almost all buildings are constructed on shallow foundations (with only few exceptional new buildings con-
structed on piles or deep foundations). While generally the foundations are singular footings or continuous beams in case
of relatively old buildings, the newer buildings are generally built with mat foundations. Although the structural systems
of relatively older buildings are composed of reinforced concrete frames only, there are generally combinations of frames
and shear walls in relatively new and high buildings. A remarkable amount of low-rise buildings are constructed as confined
masonry with a weak reinforced concrete frame, and weak walls of lightweight hollow concrete blocks.
According to observations of the authors, collapsed and heavily damaged buildings constitute a minority of the existing
building stock. This observation is also consistent with the relatively low level of ground accelerations measured or esti-
mated for the Van city center, and the actual distribution of damage to approximately 200 school buildings outlined in
Section 6.5.1.
It should be noted here that as reported by the Governorship of Van [4], 111,000 residential units are examined officially,
and 28,000 of these units (25%) are heavily damaged or collapsed, 11,000 units are moderately damaged, 43,000 units are
slightly damaged, and 29,000 units did not experience any damage. Unfortunately, details of this examination are not cur-
rently available. It should be stated here that similar to the improper inspection of new buildings, damage assessment after
the earthquakes is also often carried out without a proper methodology. The tendency of the insufficiently trained and inex-
perienced damage assessment team members is to assign damage levels that are higher than what it should be, in order to
M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628 613

Fig. 3. Soft story damage and residual displacement (Van DSI building).

Fig. 4. Global system torsion and shear failures.

stay on the safe side and not have responsibility about the potential weak performance of the investigated building against a
future earthquake. As mentioned above, it is the authors’ observation that the percentage of the collapsed and heavily dam-
aged buildings should be significantly less. While the structural damage experienced by the buildings is not only a function
of quality of concrete, many people point out the concrete quality as the most important parameter relating to the seismic
performance of the existing buildings. However, the quality of concrete is only one of the many effective parameters on the
seismic performance of existing buildings, and not necessarily more effective than various other parameters. Consequently,
any reliable rapid seismic safety assessment procedure to be conducted before the earthquake should include the effects of
many important parameters; such as the dimensions of structural members, ground characteristics, seismicity of the region,
overall weight of the structure, structural irregularities, deteriorations with time, reinforcement details, as well as the quality
of construction materials.

6.1.2. Failures of so-called engineered reinforced concrete buildings

Many so-called engineered buildings have experienced structural damages to various extents. The major causes of dam-
ages were observed to be non-compliance with the known seismic design rules. It was clearly observed that the properly
constructed buildings withstood against both earthquakes, without considerable structural damage. The observed structural
damages can basically be attributed to:
614 M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628

Fig. 5. Typical short column failure (left), short 90-degree hooks and large stirrup spacing (middle) and buckling of column longitudinal bars (right).

Fig. 6. Slip of column longitudinal bars.

(i) Irregular structural systems causing unforeseen or improperly foreseen internal forces or displacements, such as
excessive lateral displacements concentrated at certain stories due to irregular distribution of stiffness through the
height (soft story) (Fig. 3), or additional shear effects due to unforeseen torsional actions formed because of irregular
distribution of stiffness in plan (Fig. 4). The unforeseen/undesired formation of short columns attracting high shear
forces can also be categorized among most commonly observed deficiencies (Fig. 5 (left)).
(ii) Insufficient ductility or absence of ductility due to improper reinforcement detailing or structural irregularities, such
as large spacing between stirrups, insufficient anchorage of stirrups without proper 135° hooks, and lack of sufficient
crossties to enhance the effectively confined column cross-sectional area (Fig. 5 (middle)).
(iii) Remarkably higher axial stresses on vertical structural members than foreseen in design, due to lower concrete quality
and/or smaller dimensions of vertical structural members compared to design drawings, causing not only limited ver-
tical load resistance against additional seismic actions but also causing a significant reduction in ductility of columns.
(iv) Premature failures, stemming from improper detailing, such as buckling of longitudinal bars due to large stirrup spac-
ing or small diameter of main bars (Fig. 5 (right)) and/or insufficient bond of main bars over the lap splices at the crit-
ical potential hinge regions of columns causing slip (Fig. 6), which in turn causes a remarkably reduced energy
M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628 615

Fig. 7. Shear-critical columns.

Fig. 8. Weak column–strong beam problem.

Fig. 9. Beam–column joint without proper design/detailing.

616 M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628

Fig. 10. Anchorage problem of beam longitudinal bars.

Fig. 11. Structural system problem (all members are strong in one direction).

dissipation capacity and ductility of columns. Arrangement of lap splices of plain round column main bars without a
proper hook at their ends is also among major causes of structural damage, due to prevention of attaining full flexural
capacities of the columns.
(v) Undesired shear damages due to insufficient application of capacity design principles, particularly through lack of suf-
ficient transverse reinforcing bars (Fig. 7).
(vi) Undesired column damages due to insufficient application of capacity design principles, particularly through forma-
tion of plastic hinges on the critical sections of columns rather than beams (weak column–strong beam) (Fig. 8).
(vii) Poor arrangement of beam–column joints, such as lack of transverse reinforcement (Fig. 9) and poor anchorage of
beam longitudinal bars into the joints (Fig. 10).
(viii) Poor workmanship during placing, compacting and curing of concrete leading to a non-homogenous and weak con-
crete, causing remarkable local damages.
(ix) Extended rectangular columns with a small dimension in one direction (such as 200–250 mm) preventing proper
installation of reinforcing bars and placing of concrete. In addition, many buildings in Van city and Ercisß district were
constructed with extended rectangular columns with their long directions aligned in the same direction (Fig. 11). This
kind of arrangement of extended rectangular columns caused many damages and residual displacements in the weak
direction of the structure.

As mentioned before, for both earthquakes, the ground accelerations and corresponding spectral accelerations were less
than that should have been considered during structural design according to Turkish seismic design code [14]. Then, consid-
ering the observed widespread structural damage, it can be extrapolated that an earthquake as defined as the design earth-
quake in the Turkish seismic design code [14] can cause much worse consequences in terms of structural damages and
resulting casualties, not only in Van but also in other cities in Turkey. It is also worthy to mention that the damaged struc-
M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628 617

Fig. 12. Non-engineered confined masonry buildings.

Fig. 13. Damages to unreinforced masonry buildings in Ercisß district and Van center.

tures were concentrated around the areas with weak soil properties (most parts of Van city center and Ercisß), whereas the
structural damage was marginal at around Edremit (epicenter of the second earthquake), where the ground characteristics
are significantly better.

6.2. Non-engineered buildings

6.2.1. Non-engineered buildings in Van city center and Ercisß district

A great majority of existing buildings at the center of Van and Ercisß are non-engineered. While there are many non-engi-
neered reinforced concrete buildings, there are also confined masonry and unreinforced masonry non-engineered buildings.
The main difference between non-engineered reinforced concrete buildings and confined masonry buildings is that the walls
are constructed before the construction of reinforced concrete members in case of confined masonry construction (Fig. 12).
While the non-engineered reinforced concrete buildings in Van city center were damaged in different extents after the first
and second earthquakes, in Ercisß, most of the damages occurred during the first earthquake. It is important to note that while
non-engineered reinforced concrete buildings and confined masonry buildings were constructed without any engineering
service, the seismic performance of regular confined masonry buildings was generally slightly better due to better contribu-
tion of walls to the seismic behavior. The contribution of walls was observed to be important against both vertical and hor-
izontal actions. However, it should be mentioned that the seismic performance of unreinforced masonry buildings were
remarkably worse than other buildings (Fig. 13). It should also be noted that the seismic performances of non-engineered
and so-called engineered reinforced concrete buildings were not significantly different due to the fact that the latter type
buildings were also constructed without proper engineering service and inspection.

6.2.2. Non-engineered buildings in villages

The structural damages were severe and widespread in villages, which were close to the epicenter of the first earthquake,
particularly when the ground conditions were unfavorable. The ground accelerations are estimated to be in the order of 0.5–
0.6g at and around the villages close to the epicenter of the first earthquake, where most of the destruction and total struc-
tural collapses were observed. In these villages, a great majority of the existing housing units are unreinforced masonry. The
walls of these structures are mostly constructed with adobe blocks and mud mortar for relatively older housing units
(Fig. 14), and with lightweight concrete blocks and cement mortar for newer ones (Fig. 15). There are also many housing
units built with a combination of these two different types of blocks, sometimes together with wooden structural members.
Furthermore, there are also many housing units built with lightweight concrete blocks bonded to each other by mud mortar
(Fig. 16). The thicknesses of adobe walls are generally about 50 cm, whereas the lightweight concrete block walls are
generally 25–30 cm wide. These thick adobe and concrete walls are preferred by the local people due to their good heat
618 M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628

Fig. 14. Adobe walls and mud mortar.

Fig. 15. Lightweight concrete block walls and cement mortar.

Fig. 16. Lightweight concrete blocks and mud mortar.

insulation characteristics. They provide a cool environment during hot seasons and warmer environment during cold sea-
sons. The roofs of these housing units, particularly for relatively older ones, are built with wood members and covered with
a thick layer of clay for heat insulation. It is clear that the clay layer causes increase in seismic demand through its effect on
the increase of mass. Many of this type of adobe structures collapsed totally during the earthquakes. Since the wooden
beams that carry the roof and the clay layer are aligned in one direction, all vertical loads are transferred to two walls in
the plan, and the remaining two walls are not subjected to vertical loads other than their own weights. Since the shear
strength and resistance of masonry walls to overturning moment are directly proportional to the axial load acting on the
member, when there is only limited axial load on two of the walls on which the wooden beams are not supported, these
walls can easily be toppled out-of-plane or can fail in shear either in-plane or out-or-plane. Other deficiencies observed
clearly are the lack of transverse ties, bond beams and a rigid slab to form a diaphragm action. Use of different construction
materials without a proper connection in between, and improper addition of walls to the housing unit for extension pur-
poses, are among other reasons of damages and partial collapses. The weak mechanical characteristics and durability of
M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628 619

the mud mortar are also among main reasons of poor seismic performance of non-engineered housing units in villages. Be-
sides these, it should be mentioned that the roofs of newer housing units are covered with relatively lighter materials.

6.3. Industrial structures

Van is a city that has been developing rapidly in recent years. Therefore, there are many industrial structures in the city.
Like in other parts of Turkey, most of the factories are constructed with reinforced concrete structural systems. A remarkable
portion of these structures are constructed with precast members. Nevertheless, there are also many industrial facilities con-
structed with steel structural systems.

6.3.1. Concrete structures

Generally, industrial structures have performed better than the residential buildings during the earthquakes; due to bet-
ter engineering design, construction, and inspection of industrial structures compared to that of residential buildings. The
relatively superior performance of industrial structures can also be attributed to the slightly better ground conditions of
the designated industrial areas. While cast-in-place industrial structures generally suffered only limited damages, some pre-
cast concrete industrial structures failed through total collapse (Fig. 17). The total collapse of few structures of the same type
is observed to be initiated because of failures of connections of the beams to the columns (Fig. 18). While the connections
should normally be fixed through bolts and nuts in addition to the grout to be filled in the hole of the beam at the support
section, no bolts and nuts were used in the failed precast structures examined. Interestingly, no bolts and nuts were shown
on the design drawings as well. Another major design error is the assumption of Z2 type local ground condition, whereas it
should have been Z3 or Z4, considering that an allowable soil stress of 120 kPa is taken into account during structural design.
According to Turkish Seismic Code, Z1 class soil corresponds to the stiff soil condition while the Z4 is the soft soil conditions
[14]. This erroneous assumption causes a remarkable reduction in seismic design forces. Furthermore, while the bottom lon-
gitudinal reinforcing bars are hooked at the end of purlins in the design drawings, such hooks are not made in practice. The
lack of hooks may have played an important role on the failure of the support sections of the purlins. The damages to purlins
may have had a significant effect on total collapse. An interesting observation on these totally collapsed columns, which may
shed a light on nonlinear design and analysis of such structures, is the observed plastic hinge length, which is measured to be
approximately twice the depth of the column (Fig. 19). Other deficiencies observed, which did not play a major role on the
failure of the facility, were locally observed poor concrete quality, insufficient cover thickness and insufficient number of
transverse crossties in the precast columns.

6.3.2. Steel structures

In the area, most of the existing industrial structures are reinforced concrete. There are also some silos and open sheds
constructed with steel. While some of these steel structures performed well, without experiencing significant damage, some
others totally collapsed due to poor design details (Fig. 20). The main weakness of the severely damaged steel tube sections is
observed to be lack of stiffeners at the base plate connection. It should be noted that according to the information provided
by the owner of the facility, the silos were filled during the first earthquake with cement to an approximate weight of 800 kN,
which has caused total collapse of the silos.

Fig. 17. Failure of a precast concrete industrial facility.

620 M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628

Fig. 18. Failures at connections of beams and columns.

Fig. 19. Plastic hinge length of the precast column.

6.4. Historical structures

6.4.1. Mosques
It was observed that Tabanlı earthquake did not cause significant damages on historical structures whereas Edremit
earthquake caused partial collapses and severe damages of several historical mosques. Fig. 21 shows Hüsrevpasßa mosque
after Tabanlı and Edremit earthquakes. It should be noted that the partial collapse of the mosque was due to an aftershock
after Edremit earthquake. The damages are generally concentrated around the critical regions, connecting stiff and flexible
regions of the structure (Fig. 22), or at the bases of stone columns due to excessive rotation of the rigid members (Fig. 23), or
at the domes of the mosques, which are relatively more flexible with respect to structural walls of the mosques.

6.4.2. Minarets
Minarets have been among the most vulnerable structures against earthquakes, due their slender structure. During these
earthquakes, many minarets, historical or contemporary, have experienced partial collapse or severe damage. While the
structural system of the historical minarets is composed of a masonry shell, the contemporary ones are constructed with
a reinforced concrete cylindrical shell. The collapsed minaret of Hazreti Ömer Mosque in Van city center is shown in
M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628 621

Fig. 20. Failure of tube columns of steel silos.

Fig. 21. Hüsrevpasßa Mosque after the first and the second earthquakes.

Fig. 22. Concentrated damages at intersections of stiff and flexible regions.

Fig. 24. The mosque and minarets were built in 1970. Concrete cores and reinforcing bar samples were taken from the col-
lapsed part of the minaret. It was observed that the average core strength of concrete and yield strength of the plain rein-
forcing bars are 40 MPa and 300 MPa, respectively. The main reason of damage is evaluated to be the abrupt change of
strength of the reinforced concrete shell section at the overlap zone of longitudinal reinforcing bars. While the thickness
of the minaret wall is about 25 cm below the failure zone, it is about 15 cm above. Other than this abrupt change, causing
a significant reduction in strength and stiffness, the double layer longitudinal and transverse bars below the failure zone is
changed into a single layer of longitudinal and transverse bars. Furthermore, the diameter of longitudinal bars is also reduced
at the same section.

6.5. Other structures

6.5.1. Schools
School buildings in city center and districts are generally low-rise reinforced concrete structures with shear walls. While,
owing to presence of sufficient amount of shear walls in two main orthogonal directions and the regular structural systems,
most of these school buildings have performed well, without experiencing considerable damage, few school buildings have
622 M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628

Fig. 23. Excessive base rotation of rigid members.

Fig. 24. Collapsed minaret of Hazreti Ömer Mosque.

Fig. 25. Damages to partition walls in reinforced concrete school buildings.

experienced widespread damage in partition walls (Fig. 25). Considering the possibility of school buildings to be used as
shelters after earthquakes, this type of non-structural damage should also be avoided through proper construction of the
partition walls. Few schools, particularly relatively older ones without shear walls, experienced slight to moderate structural
M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628 623

Fig. 26. Floor plan and view of the school building at Gedikbulak village.

damage as well. As an exceptional case, a reinforced concrete school building was collapsed totally in Gedikbulak village. The
design plan and as-built view of this school building, and its appearance after the first earthquake, are presented in Figs. 26
624 M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628

Fig. 27. Failure of the school building at Gedikbulak village.

Fig. 28. Inadequate lap splices in the shear walls.

M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628 625

Fig. 29. Masonry school buildings with multi-leaf stone walls (Dağönü village).

Table 7
Damage distribution for school buildings [21].

Damage No Slight Moderate Severe/Collapse Total

Number 44 105 32 8 189
Percentage 23% 56% 17% 4% 100%

and 27. As seen in Fig. 26, the structural system was changed during the construction stages. Nevertheless, the structural
system appears to be adequate, in terms of stiffness and strength, to withstand these moderate-level earthquakes without
total collapse. However, the deficiencies in reinforcement detailing such as insufficient anchorage of the beam longitudinal
bars into the shear walls (Fig. 27) and inadequate lap splices of the longitudinal bars of the shear walls seem to have im-
paired the efficiency of the shear walls, remarkably (Fig. 28).
While reinforced concrete school buildings have generally performed satisfactorily, the performance of masonry school
buildings in villages were generally not as good. An important reason of relatively higher damage in masonry school build-
ings in villages is the relatively high ground accelerations, compared to reinforced concrete school buildings which are gen-
erally located in city or district centers and at a larger distance from the epicenter. The masonry school buildings, which
experienced remarkable damage or partial collapse, were generally multi-leaf stone buildings, with a wooden slab which
is not strong enough to form a diaphragm action (Fig. 29). Other main reasons of poor performance of stone masonry school
buildings are irregular-shaped stones and low-quality binding mortar. The staff of Istanbul Technical University was invited
by the Ministry of Education to conduct a damage assessment survey for 189 school buildings in the center of Van city, the
Ercisß district, and villages. The distribution of damages is shown in Table 7 [21]. It is important to note that all severely dam-
aged or collapsed school buildings are stone masonry school buildings located in villages. The collapsed reinforced concrete
school building at Gedikbulak village, which is close to epicenter of the first earthquake, was not among the school buildings
assigned to the damage assessment team of Istanbul Technical University.

6.5.2. Hospitals
Several hospitals were examined at different regions of Van city center. Like most of other types of structures, all hospitals
examined in the region are constructed with a reinforced concrete structural system. While none of the hospital buildings
collapsed during earthquakes, all hospitals were out of service due to significant damage to non-structural components, ex-
cept a new hospital that was taken into service in 2011. Furthermore, relatively older hospital buildings were observed to
experience structural damages as well. Considering the low level of the ground acceleration caused by the earthquakes,
the hospital buildings should have exhibited a performance of immediate occupancy. Therefore, it is judged that certain de-
sign and construction deficiencies, particularly those causing insufficient lateral stiffness of the hospital buildings, are
responsible for the damages. These can be listed as inadequate selection of structural system, without proper rigid structural
members such as shear walls, and use of low quality concrete with low elastic modulus. The remarkable damages to non-
structural walls are attributed to the poor masonry workmanship, together with the high lateral drifts experienced. The
damages to structural and non-structural members observed at some hospital buildings are shown in Fig. 30. It is also impor-
tant to note that the non-fixed medical equipment and furniture may also cause injuries and deaths (Fig. 30).

6.6. Damages due to collapse of neighboring buildings

Some buildings in Van experienced damage resulting from collapse of neighboring or adjacent buildings. These types of
damage are quite unpredictable and it is difficult to mitigate risks of such damage, unless the owners of the neighboring
weak building do not take the necessary measures towards prevention of collapse or excessive lateral displacements. This
issue is quite important in Turkey, since many buildings are constructed adjacent to neighboring buildings, without any
spacing in between. It should be noted that there are rules on minimum spacing between adjacent buildings, since the first
Turkish seismic design code was published in 1940.
626 M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628

Fig. 30. Typical damages observed in hospital buildings.

6.7. Failures of non-structural components

Many non-structural components of the buildings experienced severe damage during the earthquakes. The most common
problem was observed to be the low quality construction of non-structural walls, which either failed out-of-plane, or expe-
rienced severe damage due to in-plane stresses. These types of damage, which could have been easily avoided through sim-
ple and improved connection details of non-structural walls to the structural system, are among the main reasons why many
buildings were not occupied for a long time after the earthquakes, even though the structural system itself had not suffered
significant damage. These damages clearly highlighted the importance of mitigation efforts on the non-structural compo-
nents, for rapid recovery and return to normal living conditions. Other common types of damage observed included damages
to stairs, generally due to improper reinforcement detailing, and damages due to pounding effects between adjacent build-
ings, due to lack of sufficient spacing in between.

7. Conclusions and recommendations

Conclusions and recommendations deducted from the observations in the earthquake region can be summarized as

a. With the exception of a single new hospital building, all hospitals in Van city (with a population of approximately one
million) were out of service after the earthquakes. Considering the low level of ground acceleration caused by these
earthquakes and the performance level required of hospital buildings, it is clear that hospital buildings in Van and
other cities should rapidly be assessed for seismic safety and performance, and the hospital buildings not possessing
adequate seismic performance characteristics should either be retrofitted or reconstructed. Otherwise, observations
indicate that in case of severe earthquakes, which may cause remarkably higher ground accelerations, most of the
existing hospital buildings may experience more extensive structural and non-structural damage, leading to long-
term hindrance of utilization of hospitals. At this point, special attention must be paid to prevent non-structural dam-
age (such as failure of or damage to non-structural walls), since some of the hospitals observed were out of service
because of non-structural damages only.
M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628 627

b. Reinforced concrete school buildings with shear walls generally performed well. Nevertheless, the slight structural
damages and the non-structural damages should have been better-avoided, for these buildings to be used as shelters
after earthquakes, and for putting the schools back in service without too much delay. The exceptional total collapse of
one reinforced concrete school building is attributed to weak connections between the foundation and the shear walls,
and the poor details of beam–column joints at floor levels. In addition, some single-story school buildings with multi-
leaf stone masonry walls in villages experienced damages to different extents, including total collapse. These failures
are attributed to the poor connection of wall units due to low-quality mortar and irregular-shaped stones used in the
construction, as well as poor diaphragm action.
c. Although the peak ground accelerations were in the order of 0.1–0.3g in densely-populated regions, many buildings in
these regions suffered heavy damage and collapse. However, it was clearly seen that all collapsed and damaged build-
ings were non-engineered (or so-called engineered) buildings, which were improperly constructed. The buildings,
which were constructed with a minimum amount of care, withstood the earthquakes without experiencing significant
damage. It was clearly observed that if a proper construction and inspection mechanism had been established, the col-
lapses and casualties could have been avoided. Therefore, to avoid similar losses and damages in future, proper con-
struction of new buildings must be enforced, and existing non-engineered buildings must be evaluated for seismic
performance, and retrofitted when necessary.
d. Unlike the common belief in Turkey that poor material quality is the major cause of poor seismic performance, mate-
rial quality was not observed to have major influence on the destructive effects of the earthquakes. Many buildings
with relatively good material quality (medium strength concrete and deformed reinforcing bars) collapsed or suffered
severe damage, while many buildings with remarkably lower-quality material characteristics (low-strength concrete
and plain reinforcing bars) survived without experiencing significant damage. Regularity of the structural system
including arrangement of non-structural walls, and detailing of the structural members, were observed to be impor-
tant characteristics that have governed the seismic performance of structures.
e. Some invaluable historical structures, such as mosques and minarets, were damaged to different extents. Immediate
action is required to protect and restore these monumental structures. The damages to such monumental structures
should be considered as a warning for many existing vulnerable historical structures spreading all around Turkey and
other parts of the world, and necessary measures should be taken to avoid failures of invaluable cultural heritages.
f. High or long unsupported non-structural (infill) walls should be confined by bond beams, in order to prevent out-of-
plane failure of non-structural walls, which can cause injuries or deaths.
g. It was observed that only a small portion of the buildings in the region were insured, although an obligatory seismic
insurance system had been established after 1999 earthquakes. The obligatory insurance agency must seriously con-
sider means to reach more people, and the government must consider adopting penalties for building owners who
resist having their properties insured.


The authors are grateful for the support of administrations of Istanbul Technical University (particularly Rector Prof. M.
Sahin), Boğaziçi University, and Van Yüzüncü Yıl University (particularly Rector Prof. P. Battal). The financial support of Istan-
bul Technical University and the logistic support of Van Yüzüncü Yıl University are gratefully acknowledged. The authors
also wish to thank to the technicians of Istanbul Technical University Mr. M. Uslu, H. Atesß and A. Sßahin, as well as the Building
Materials Laboratory of Istanbul Technical University. The concrete strengths for six buildings out of twenty three are pro-
vided by Turkish Ready-Mixed Concrete Association.


[1] United States Geological Survey. Implications for earthquake risk reduction in the United States from the Kocaeli Turkey earthquake of August 17,
1999. USGS circular report no. 1193; 1999.
[2] Turkish Prime Ministry Disaster and Emergency Management Presidency. <> [accessed at 15.02.12].
[3] Bogazici University Kandilli Observatory and Earthquake Research Institute. <> [accessed at 18.02.12].
[4] Governorship of Van. <> [accessed at 30.01.12].
[5] Decanini L, Liberatore L, Molliaioli F, De Sortis A. Estimation of near-source ground motion and seismic behaviour of RC framed structures damaged by
the 1999 Athens earthquake. J Earthquake Eng 2005;9:1–27.
[6] Bakir BS, Sucuoglu H, Yilmaz T. An overview of local site effects and associated building damage in Adapazari during the 17 August 1999 Izmit
earthquake. Bull Seismol Soc Am 2002;92:509–26.
[7] Decanini L, Liberatore L, Molliaioli F. Damage suffered by RC buildings during the 2009 L’Aquila earthquake. In: Proceedings of 14ECEE, a general
overview and a case study; 2010.
[8] Celebi M, Bazzurro P, Chiaraluce L, Clemente P. Recorded motions of the 6 April 2009 Mw 6.3 L’Aquila, Italy, earthquake and implications for building
structural damage: overview. Earthquake Spect 2010;26:651–84.
[9] Smyrou E, Tasiopoulou P, Bal IE, Gazetas G. Ground motion versus geotechnical and structural damage in the February 2011 Christchurch earthquake.
Seismol Res Lett 2011;82:882–92.
[10] Celep Z, Erken A, Taskin B, Ilki A. Failures of masonry and concrete buildings during the March 8, 2010 Kovancilar and Palu (Elazig) earthquakes in
Turkey. Eng Fail Anal 2011;18:868–89.
[11] United States Geological Survey. <> [accessed at 10.03.12].
[12] Kocyigit A. Neotectonics and seismicity of East Anatolia. In: Proceedings of the workshop on the geology of East Anatolia, Van; 2002.
628 M. Tapan et al. / Engineering Failure Analysis 34 (2013) 606–628

[13] SeismoSignal v.4.3.0. SeismoSoft Ltd.; 2011. <>.

[14] Turkish earthquake resistant design code. Ministry of Public Works and Settlement, Ankara; 2007.
[15] Construction inspection law. Ministry of Public Works and Settlement; 2001 [in Turkish].
[16] Building law [law no: 3194]. Official gazette no: 18749; 1985.
[17] Turkish earthquake resistant design code. Ministry of Public Works and Settlement, Ankara; 1998.
[18] Turkish earthquake resistant design code. Ministry of Public Works and Settlement, Ankara; 1975.
[19] Ilki A, Celep Z. Earthquakes, existing buildings, and seismic design codes in Turkey. Arab J Sci Eng 2012;37:65–380.
[20] Ozsayin B, Yilmaz E, Ispir M, Ozkaynak H, Yuksel E, Ilki A. Characteristics of CFRP retrofitted hollow brick walls of reinforced concrete frames. Constr
Build Mater 2011;25:4017–24.
[21] Istanbul Technical University. Van Ilinde Bulunan MEB Okul Binalarının Hasar Durumlarına ait Rapor. ITU technical, report; 2011 [in Turkish].