A site survey of damaged RC buildings in İzmir after the Aegean sea earthquake on October 30, 2020

An earthquake with a magnitude of M w = 6.6 and a depth of approximately 16.5 km occurred on


Introduction
Turkey lies in one of the most active earthquake zones worldwide. Based on statistical studies, in Turkey, a strong earthquake with a magnitude of M w between 6.0 and 6.9 occurs every two years, and a major earthquake with a magnitude M w of above 6.9 occurs every three years. Approximately 98 % of the Turkish population lives in earthquake-risk-designated areas. Since the establishment of the Turkish Republic in 1923, approximately 80,000 people have lost their lives owing to earthquakes. The total approximate direct and indirect costs of earthquake damage between 1980 and 2022 were estimated to be USD 40 billion [1,2]. Turkey has several fault lines that can generate strong earthquakes. These include the North Anatolian Fault Line (NAF) and East Anatolian Fault Line (EAF) (Figure 1, adapted from Turkish General Directorate of Mineral Research and Exploration, MTA [3]). The western part of the country has several fault line segments with the potential to generate strong earthquakes. These fault lines can all be categorised under the West Anatolian Fault Line (WAF) (Figure 1). The NAF in northern Turkey extends close to 1,500 km and spans from east to west. In eastern and southeastern parts of the country, the EAF extends from the Gulf of Iskenderun to the city of Hakkari in the shape of an arc. In the western part of the country, the WAF covers an area of approximately 45,000 km 2 (almost 6 % of Turkey's total land area).
Bayraklı, İzmir, located in the western part of Turkey, suffered a strong earthquake with a magnitude of M w = 6.6 on 30 October 2020 [4]. The hypocentre was located 16.5 km below the epicentre. The earthquake affected more than four million people, living primarily in the city of İzmir and surrounding towns and villages, as well as the residents of Samos Island, Greece. The focus of this article is on the reinforced concrete (RC) buildings located in and around İzmir; therefore, the structural damage observed in Samos are not included in this study. İzmir is the third most populated city in Turkey, with a total of population of nearly 4.4 million people. It also has the third highest gross domestic product (GDP) in the country, at 6.3 % of the total GDP [5]. The city centre is located approximately 67 km northeast of the epicentre of the earthquake, as shown in Figure 2. According to government data, 117 people lost their lives and approximately 32,000 people were injured owing to the earthquake. The earthquake caused the collapse of 11 RC buildings immediately after the main shock. According to detailed site investigations conducted by the government, close to 6,000 buildings experienced moderate-to-minor damage. Of these buildings, 511 had moderate damage, and 5,119 had minor damage. More than 500 buildings were structurally inadequate and were demolished [6]. Several scientific studies were conducted on the 2020 Aegean Sea earthquake. While some studies focused on the geotechnical and seismic aspects of earthquake [7][8][9][10], others focused on building damage [11][12][13][14]. The total direct cost of the earthquake damage was estimated to be approximately USD 400 million [15].
It is important to visually record earthquake-induced damage at a site immediately after an earthquake. This can provide valuable information to experts before structural demolitions and alterations occur. A reconnaissance team performed a technical visit to the earthquake-affected region within 24 h of the earthquake. This paper, which details the observations and outcomes of this site visit, adds to the current literature on the 2020 Aegean Sea earthquake by providing valuable onsite information on the seismic characteristics of the earthquake. These characteristics are based on data extracted from various strong ground motion stations, beginning from the closest Turkish station near the epicentre to the Bayraklı District where some RC buildings collapsed and others experienced moderate-to-major damage. The observations and findings of this site visit with respect to RC buildings are presented in this paper.

History
During the Neotectonic Age (approximately 12 million years BC), plate movement began in the north-south direction due to the collision of the larger Arabian and smaller Anatolian plates [16] (Figure 3). Because of the collision between the Arabian and Anatolian plates, the western part of Anatolia moves at a speed of 40 mm/year in a counter-clockwise direction [17]. There are two important largescale grabens (Gediz-GG and Büyükmenderes-BMG) in the regions with normal faulting mechanisms [18]. Both are young geological structures formed during the neotectonism of Western Anatolia and have the potential to generate destructive earthquakes. According to geological studies conducted in this region, strike-slip faulting mechanisms have been identified between these two important grabens in the N-S, NW-SE, and NE-SW directions [19][20][21]. The seismicity in and around İzmir is generally governed by both normal and strike-slip fault mechanisms located between, and parallel to, the Büyük Menderes and Gediz grabens ( Figure  4). These seismic activities are predominantly controlled by the İzmir, Tuzla, Karaburun, Yenifoça, Manisa, Kemalpaşa, Seferihisar, Menemen, Gülbahçe, and Dağkızılca fault lines as well as the Gediz Graben detachment fault [22]. The İzmir and Manisa fault lines, which exist to the south and east of the İzmir Bay area, are both governed by normal faults. This region also contains the Tuzla and Yenifoça faults, whose strike-slip mechanisms are located in the northeast-southwest direction of the bay area [23]. The formation of İzmir Bay began through normal faults that occurred during the Early Pliocene in the western part of Turkey [24,25].
In the Late Quaternary, early delta progradation of sediments occurred in the bay area [26]. Therefore, the local geology of the coastline of the İzmir Bay area consists of Quaternary alluvium surrounded by a Paleocene flysch zone (limestones) and Miocene sandstones/mudstones [27][28][29]. The depth to bedrock in the bay area varies from 900 to 1,200 m, and the groundwater level is between 1.0 and 10.0 m [30,31].

Earthquake hazard map
The Earthquake Hazard Map of Turkey, which was prepared based on earthquakes with a 10 % exceedance probability in 50 years, or the equivalent of a return period of 475 years, is shown in Figure 5 [32]. Based on the peak ground accelerations shown in the map, the most severe earthquake regions are the northern, eastern, and southeastern parts of Turkey, including the area near the Aegean Sea, the inner southwestern part of Turkey, and the area around Lake Van. Figure 6 shows a more detailed hazard map of western Turkey, with the expected peak ground motion acceleration values calculated based on the same return period of 475 years, along with active fault lines [32]. Based on the data in Figure 6, the largest peak ground acceleration values in the region are approximately 0.4 g. As this article focuses on the RC buildings located in the eastern part of İzmir Bay, it is important to emphasise the region's active fault lines and soil conditions. The map in Figure 7 shows the active fault lines and faulting mechanisms of the İzmir Bay area, including the dominant soil type [3]. Based on the map, the dominant type of soil is undifferentiated Quaternary deposits. An in-depth discussion of soil types is provided in the next section. Figure 8-a shows the aerial view of the İzmir Bay area and the region that experienced the most severe damage, which is shaded in blue. Figure 8-b provides information related to the detailed soil types in and around the bay area. According to the map, the dominant soil type of the bay area is undifferentiated Quaternary deposits, which is a type of soil that contains siliciclastic organics and freshwater carbonates. However, regions to the north, south, and west of the study area, each approximately 15 km away, have much better soil characteristics.

Aegean sea earthquake
The Aegean Sea earthquake occurred on 30 October 2020 at 14:51:23 local time [4]. The earthquake's epicentre was off the coast of Samos Island, Greece, which is located 35 km away from the southwestern part of the city of İzmir. The nearest inhabited Turkish town to the epicentre is the village of Payamlı. It is located in Seferihisar, a town in İzmir Province, which was 23.4 km away from the epicentre. The earthquake had a focal depth of 16.5 km and its magnitude was measured as M w = 6.6 [4]. Table 1 lists the time, location, magnitude, and depth of the earthquake as recorded by different agencies. The earthquake occurred due to normal faulting at a shallow crustal depth within the Eurasia tectonic plate of the eastern Aegean Sea. According to the USGS, the focal mechanism solution indicates that the earthquake occurred on a moderately dipping normal fault mechanism, indicating a north-south oriented extension, which is fairly common in the Aegean Sea [36]. Based on another study conducted by the MTA, an approximately 40 km long Samos fault line ruptured during the earthquake [39]. It was also stated that the strain energy probably shifted over to the western part of the Samos fault line, extending from the North-East to the South-West. The results from the moment tensor solution of the earthquake, as prepared by different agencies, are listed in Table 2.  Historically, the İzmir region has been very active and has experienced numerous large magnitude earthquakes. Since 1900, a total of 695 earthquakes with magnitudes greater than or equal to 4.0 have been recorded during the instrumental period. Out of these 695 earthquakes, the largest magnitude was 6.8, which occurred in 1955. A total of 332 earthquakes were also recorded during the non-instrumental period, pre 1900 ( Figure 10) [4,35]. After the mainshock of the earthquake, a tsunami occurred in the region with a maximum wave height of 3.82 m [39]. The waves in the northern direction hit the shore of the Bay of Sığacık, a district in Seferihisar, and in the southern direction, on the northern shore of Samos Island [40]. Based on the authors' site observations, the waves moved as much as 200 to 250 m inland, into the district of Sığacık. According to the earthquake data collected between October 30, 2020 and November 18, 2020 (20 days), a total of 4,320  Figure 11 shows the daily frequency distribution of these aftershock earthquakes. In order to determine the characteristics of the aftershocks, the total number of aftershocks is also categorized according to their magnitudes, as displayed in the figure. Table 3 lists the peak ground acceleration values recorded by AFAD at the nearest sixteen strong ground motion stations [32]. The locations of these stations are plotted on the map in Figure 12. The station closest to the earthquake was the Seferihisar station (station number 3536), which was 35 km from the epicentre. The most distant station was Bornova station (station number 3520), which was 76 km from the epicentre. The maximum peak ground acceleration recorded at the Kuşadası station (station number 0905) was in the East direction with a magnitude of 0.179 g.  , which were all associated with local site conditions. The acceleration peak at the epicentral distance of 70 km was much more pronounced than the other two peaks. This led to the immediate collapse of eleven buildings and severe structural damage in the Bayraklı district. Figure  13.b shows the variation in effective earthquake duration. Based on the data, the effective duration increased as the earthquake moved into the İzmir Bay area, leading to an increase from an average of 16 to 23 s in the horizontal directions. This increase was more prominent in the vertical component of the earthquake. The variations in the Arias and Housner Intensities of the Aegean Sea earthquake are plotted in Figure 14.a and b in the N-S, E-W, and vertical directions based on the ground motion data recorded at 16 stations ( Figure 12 for their locations). The data used in these figures clearly indicate that the intensity of the earthquake was greater in the regions around İzmir Bay, Bayraklı, and Karşıyaka.

Evaluation of strong ground motion data
Strong ground-motion data were evaluated at six stations to better understand the impact and characteristics of the earthquake, specifically in the bay area ( Figure 15). These stations are located along the earthquake route, starting from Kuşadası (station number 0905) to the station in the most affected area, Bayraklı (station number 3513), in the following order: 0905, 3536, 3518, 3513, 3519, and 3514. However, emphasis is given to the results extracted from the following three stations in and around the town of Bayraklı, which encircles the İzmir Bay area: 3513, 3518, and 3519. Figure 15. Locations of the six strong ground motion stations used to evaluate the Aegean Sea earthquake data Figure 16 shows the peak ground acceleration values of the Aegean Sea earthquake in the north-south and east-west directions recorded at six stations. According to the data, the largest peak ground acceleration values were recorded at station number 0905 (located in the town of Kuşadası). The values at this station in both directions were 0.18g and 0.15g, respectively. The acceleration values are also provided in the Bayraklı area, where building collapses and severe structural damage were observed. The nearest station to Bayraklı is number 3513; it recorded 0.11g and 0.10g peak ground accelerations in the north-south and east-west directions, respectively. The design spectrum curves of the current Turkish Building Earthquake Code (TBEC, 2018) [41] were compared with the measured spectral acceleration data, which were extracted from six stations in the north-south and east-west directions, as a function of varying damping ratios ( Figure 16). TBEC (2018) [41] defined a total of six local soil classes, identified using letters from ZA through ZF, where ZA defines a hard rock soil type and ZF defines an extremely loose soil type, which requires further soil testing and site evaluation. In this study, the design spectrum curves of the first five soil classes (excluding soil type ZF) were used to evaluate the north-south and east-west peak ground acceleration components of the Aegean Sea earthquake. Based on the spectrum values, the peak ground acceleration values at stations 3536 and 3514 were much lower than the demands associated with the five soil classes. Therefore, the structural damage at these locations was moderate to minor. However, as the seismic waves travelled along stations To further evaluate the impact of the Aegean Sea earthquake, the local soil classes of the selected stations were considered. Figure 17 shows the soil classes at these stations [42]. Based on this data, the soil properties of the İzmir Bay area appear to be the weakest because almost all the stations are located in soil classes ZD or ZE, which confirms the soil properties explained in Section 2.3. As expected, the weak soil amplified the peak ground acceleration and extended the effective duration by causing building collapse and severe structural damage. Although stations 3518 and 3519 both displayed acceleration patterns very similar to station 3513, the building damage near these two stations was not as severe as that at 3513. The severe building damage near station 3513 was largely attributed to the local soil type of the region (alluvial plain and delta deposits) ( Figure  8.b). Another reason for this problem is the lack of geotechnical studies, which were not mandated at the time of construction (for further information, see Section 4.5.5).

Earthquake performance of reinforced concrete buildings
Building an effective framework for an extensive damage assessment of structures resulting from an earthquake is a very important concept for evaluating existing earthquake design codes and regulations. Many methods have been proposed by academic scholars in this field. Some of these methods are extremely detailoriented and involve a resource-intensive process that requires full access to the structures in an earthquake-affected site, whereas  [43][44][45]. However, other methods are commonly used in Europe, such as the Greek Criteria [46], which were further elaborated based on extensive experience from past earthquakes [47]. This method has been used in other earthquakeprone countries such as Croatia [48].
In Turkey, the damage evaluation of RC buildings subjected to earthquakes is performed using the guidelines in [49] specified by the Turkish Chamber of Civil Engineers. Based on this document, the damage is classified into three groups: minor-, moderate-, and heavy. Minorly damaged buildings can be used immediately after an earthquake, moderately damaged buildings should be repaired and strengthened before occupation, and heavily damaged buildings should be demolished for economic reasons. The evaluation is performed in two stages (exterior and interior evaluation). If the building is partially or fully collapsed, the inter-story drift is greater than 0.02, or the building is tilted by more than 3°, the building is classified as heavily damaged. Interior evaluations are not conducted for these buildings. If the building is not classified as heavily damaged at the exterior evaluation stage, an interior evaluation is performed on the damage and condition of all the beams and columns of the critical story (often the ground story). Based on a numerical evaluation of the damage percentages of the vertical (columns) and horizontal (beams) elements, a building can be classified into any of the three damage levels. The buildings presented in this paper were classified as heavily damaged during the exterior evaluation process. As technology advances, there have been recent developments in post-earthquake evaluation methods. One such method requires extensive seismic instrumentation of buildings through rapid analyses of the outputs extracted from sensors [50]. Another one involves the use of machine learning techniques to quickly determine the earthquake-induced building damage using features such as spectral accelerations at a period of 0.3 s, fault distance, building's age, floor area, and the presence of irregularities [51]. In addition to these types of resource-intensive methods that provide in-depth technical information on the impact of earthquakes, an immediate damage assessment method might be more practical to fulfil the urgent requirement to organise earthquake aid and rescue operations. The most practical method is the on-site visual inspection of the earthquake-damaged site. This provides valuable technical information on earthquakes before human interference occurs. This approach was used by our technical team to assess the structural damage to RC buildings in Izmir.
The technical team visited the earthquake-affected locations 10 h after the earthquake to avoid missing any data related to the condition of the buildings during the search and rescue operations. Detailed observations and investigations were performed at collapsed, partially collapsed, and damaged building sites during the four-day stay of the technical team. The findings and evaluations explained in this section are the results of the data and evidence collected during numerous visits to related sites at various times. First, the overall impact of the earthquake is explained, and the findings of site visits are discussed to identify the reasons for the structural damage. With the exception of images from Google Maps, all photographs featured in this section were captured by (and are the properties of) the authors.

Building inventory and damage statistics in Izmir
In January 2019, the total number of buildings in Izmir was 830,447 [52]. Based on data provided by the Turkish Statistical Institute, almost 70 % of the buildings in İzmir are RC buildings; the remainder are masonry [14]. Residential buildings constitute approximately 90 % of all buildings. Figure 18 shows the distribution of RC buildings in İzmir.

Figure 18. Distribution of RC buildings in İzmir by percentage (adapted from [14]
As depicted in Figure 18, 73 % of the RC buildings have up to three floors, 25 % have four to eight floors, and the remaining (almost 2 %) have nine and more floors. As a group, 70 % of these RC buildings were built before 2000. Two main earthquake codes were in effect at that time: the 1975 and 1998 earthquake codes. Based on the number of years and buildings constructed, it is accurate to state that out of 70 % of the buildings, almost 60 % of them were built according to the rules and regulations of the 1975 Turkish Earthquake Code. Only a few government-approved institutions were permitted to investigate buildings (RC and masonry) at earthquake sites. Table 4 lists the findings of one of these institutions [14,53] total number of buildings without damage constituted 95 % of the total building inventory. The buildings with major damage and those that collapsed and demolished were all in the Bayraklı District (see Section 4.2). The distribution of the damage levels of the RC buildings in İzmir is also plotted in Figure 19 as a function of their construction years.
As the data indicates, most of the moderate-to-major damage was observed in buildings constructed between 1990 and 2000. Beyond 2010, the damage level shifted towards minor-level damage.

Collapsed buildings
As stated in Section 1, the earthquake caused the sudden collapse of 11 RC buildings. Out of these 11 RC buildings, four of them (Rıza Bey, Doğanlar, Emrah, and Block B of Yağcıoğlu Apartment Complex) completely collapsed, and seven of them (Yılmaz Erbek and Karagül Apartments, three blocks in the Barış Apartment Complex and two blocks in the Cumhuriyet Apartment Complex) partially collapsed in the Bayraklı District during the earthquake. All the collapsed buildings had structural framing systems consisting of RC columns and beams. These frames were infilled with brick masonry at all story levels except for the ground story. Most had shops on the ground story which may have produced a weak story at the ground-floor level. General information related to the collapsed buildings is presented in Table 5. The locations of these buildings are shown in Figure 20.
A nine-story building, the Rıza Bey Apartment, collapsed fully during the earthquake. The building construction was completed in 1994. The building consisted of 32 separate apartments and five shops at the ground-story level. Based on videos shot during the collapse, the collapse of this building started with the failure of the ground story columns. After the remainder of the building shifted one story down, all the other stories began to collapse progressively on top of each other without any translation in the horizontal direction. Photographs of the building before and after the earthquake are shown in Figure 21.
The Doğanlar Apartment Building, with its 21 apartments and four shops at the entrance level, collapsed approximately 1 min after the earthquake. Construction of this eight-story building was completed in 1992. One side of the building was adjacent to the other, with a small gap between them. During its collapse, the Doğanlar Apartment Building shifted away from the adjacent building. Photographs of the buildings before and after the earthquake are shown in Figure 22.  Another collapsed building was the eight-story Emrah Apartment Building, completed in 1993. The apartment housed six shops on the ground floor and 28 separate apartments. Photographs of the buildings before and after the earthquake are shown in Figure 23.
Block B of the Yağcıoğlu Apartment Complex collapsed fully after the earthquake, whereas Block A survived the earthquake with heavy damage. Each block of this apartment complex consisted of an 8 story building with 14 separate apartments and four shops located at ground level. Both the buildings were constructed in 1993. Photographs of the buildings before and after the earthquake are shown in Figure 24. The Yılmaz Erbek Apartment Complex consisted of two adjacent 10-story blocks.
The building was constructed in the late 1990s. The ground story level consisted of shops, and the remaining stories comprised separate apartments. The first two stories of one block collapsed during the earthquake. The partially collapsed building was separated from the other buildings, and the total height was reduced. Photographs of the building before and after the earthquake are shown in Figure 25.
The eight-story Karagül Apartment Building, with 28 separate apartments and shops at ground level, was heavily damaged during the earthquake. The building was constructed in the early 1990s. Each residential story consisted of four apartments, and only one-fourth of the building area (corresponding to one apartment) partially collapsed a few minutes after the earthquake. Photographs of the building before and after the earthquake are shown in Figure 26. The Barış Apartment Complex, which was comprised of four blocks with eight stories each, was constructed in 1992. Three of these buildings partially collapsed, and one experienced extensive damage without collapse. The first four stories of the two blocks and the first three stories of the other block collapsed on top of each other. The stories over the collapsed portions of the buildings survived the earthquake by leaning towards one side of the buildings. Photographs of the Barış Apartment Complex before and after the earthquake are shown in Figure 27.

Severely damaged buildings
There were numerous severely damaged buildings in the city of İzmir during the Aegean Sea earthquake. A typical damage type was diagonal cracking on structural (beams, columns, and shear walls) and non-structural (partition walls) elements, as shown in Figure 29.

Figure 29. Diagonal cracks on structural and non-structural elements
Structural damage to the columns was also significant in some of the buildings, as shown in Figure 30. However, some of the buildings constructed side-by-side were separated because of the collision (hammering effect) of these buildings moving in opposite directions during the earthquake. This type of motion damages the structural elements of buildings. An example of this behaviour is shown in Figure 31.

Evolution of Earthquake Codes in Turkey
The first Turkish Earthquake Code, TEC (1940), [54], was published in 1940 and was originally adapted from the Italian Earthquake Code, which was valid at that time. To date, the code has been modified nine times. The last four versions of this study were published in 1975 [55], 1998 [56], 2007 [57], and 2018. [41].
The partially and fully collapsed RC buildings in this earthquake were constructed in the 1990s (between 1990 and 1998) when the TEC (1975) [55] was in effect. TEC (1975) [55] was considered successful when it was prepared. According to TEC (1975) [55], earthquake loads on buildings were calculated by multiplying the earthquake load coefficient by the seismic weight (calculated using the self-weight and a portion of the live loads) of the building. Factors such as earthquake zones (four zones), building importance factors (two factors), soil types (12 types), fundamental period of the building, design acceleration spectra (four types), and ductility levels of the building (two types) were used to determine this coefficient. Based on this code, the calculated base shear for residential buildings was approximately 8 % to 15 % of the seismic weight of the building. An equivalent lateral static force procedure was used to distribute the total base shear to the story levels.
No minimum concrete compressive strength was defined for residential buildings in this code; however, a minimum concrete compressive strength of 22.5 MPa was specified for buildings located in first and second earthquake zones, with an importance factor of more than one. There was no provision related to the type (plain or deformed bars) and minimum yield strength of reinforcing steel bars. The largest sectional forces were known to develop at the beam and column ends. Therefore, provisions related to the confinement of the beam and column ends and the bending stirrup end into the concrete core (135° hooks) were also included in TEC (1975) [55]. Additionally, provisions related to short columns were incorporated into this specification. Beginning with TEC (1998) [56], a strong beam-stronger column methodology was introduced into the design of RC buildings in Turkey. This required the resisting moment capacities of the connecting columns of a joint to be greater (at least 20 %) than those of the connecting beams of the same joint in the same plane. When a building is subjected to earthquake forces, damage (plastic hinging) occurs first at the bottom ends of the first-story columns. Consequently, plastic hinges will always form at the beam ends if a building is constructed based on the strong beam-stronger column criterion. The collapse of a building requires the formation of hinges at both beam ends, resulting in the highest possible energy absorption capacity before failure. This criterion was also followed in later codes. Other provisions of TEC (1975) [55], such as the confinement of concrete at the beam and column ends and the bending of stirrups, were explained in more detail in TEC (1998) [56], TEC (2007) [57], and the Turkish Building Earthquake Code [41]. The calculated base shear for residential buildings was approximately 12.5 % to 25 % percent of the seismic weight of the building for these three later codes. A comparison of the last four Turkish earthquake codes for RC buildings is shown in Table 6.

Evaluation of collapsed and severely damaged buildings
All fully and partially collapsed buildings in this study were designed and constructed before TEC (1998) [56] was implemented. Therefore, the provisions of TEC (1975) [55] will be used to evaluate the design, construction, and behaviour of these buildings during this earthquake. Some collapsed or severely damaged buildings were structurally evaluated for their risk of collapse during earthquakes in recent years (between 2012 and 2018). Corresponding Structural Evaluation Reports (SER hereafter) [58,59] were prepared and submitted to the inhabitants of these buildings long before the 2020 Aegean Sea earthquake. The information in these SER [58,59] was also used to evaluate the behaviour of the buildings in this section. A summary of the problems resulting in the damage and collapse is shown in Figure 32. It is important to note that these problems are not only common to RC buildings in Turkey but also to other types of buildings, such as masonry and timber buildings, resulting from past earthquakes that occurred in the region and worldwide [60][61][62]. The details of these problems are discussed in the following sub-sections. * For all buildings in all earthquake regions: ** Buildings located in the first and second earthquake regions with an importance factor of more than one; *** Buildings located in the first and second earthquake regions with an importance factor of more than one or ductile behaviour

Problems related to materials
No minimum concrete compressive strength was specified in TEC (1975) [55] for residential buildings that collapsed during this earthquake. According to the SER, concrete compressive strengths ranging from 10 to 17 MPa was used in the construction of these buildings [58,59]. These values were determined using ultrasonic and surface-hardness tests performed on the buildings. However, these results should be interpreted with caution because nondestructive testing of concrete compressive strength is unreliable and should be used in combination with the traditional coring technique. The concrete compressive strengths obtained from these tests were lower than those specified in the structural drawings of the buildings. Based on site observations of the concrete members of the collapsed buildings, it was determined that the concrete used in these buildings consisted of round coarse aggregates of 30-50 mm in diameter. Round and large aggregates would result in poor interlocking behaviour and weak bond strength between the concrete and steel reinforcement. Photographs of some of the concrete samples are shown in Figure 33.  TEC (1975) [55] did not specify the type of steel reinforcement that should be used in RC buildings. Only plain longitudinal and transverse reinforcing bars were available in Turkey during the early 1990s. The steel-reinforcing bars exhibited a minimum yield strength of 220 MPa. Deformed reinforcing bars have been available since the mid-1990s. Visual observations of the collapsed buildings confirmed that combinations of these reinforcements were used during construction. For example, plain bars were used in the construction of the Emrah Apartment, both as longitudinal and transverse reinforcements. The longitudinal and transverse reinforcements used in the Rıza Bey Apartment were deformed and plain, respectively. In the Yılmaz Erbek Apartment, only deformed bars were used as reinforcement. Photographs of the reinforcements are shown in Figure 34. Most of the buildings in the area had moisture problems owing to the humid climate, high water table level, and insufficient water insulation, particularly at the ground level. Therefore, the reinforcements of the structural members of these buildings corroded extensively, resulting in a significant reduction in the cross-sectional area of the reinforcements. Examples of the corroded reinforcements are shown in Figure 35. It is important to note that these mistakes are repeated when building damage is investigated during most earthquakes. For instance, the field observations recorded after the Sivrice, Elazığ, Turkey earthquake on 24 January 2020 are an example of this [63].

Problems related to reinforcement configuration
As stated previously, TEC (1975) [55] required closely spaced stirrups (confinements) to be used at the end regions (confinement zones) of the beams and columns. The length of this region was a minimum of one-sixth of the column height, or 450 mm for the columns. TEC (1975) [55] also specified the maximum distance between the two stirrups in this region as 100 mm.
Site observations indicated that no confinement was used at the ends of the beams and columns. Examples of the reinforcement configurations of the columns and beams of collapsed buildings are shown in Figure 36. This finding was also supported by various SER [58,59] prepared for these collapsed buildings. Although TEC (1975) [55] required bending the ends of the stirrups into the concrete core (135° hooks), the ends of the stirrups were only bent by 90°, resulting in the opening of the stirrups after the spalling of the cover concrete in the beams and columns. Photographs of these stirrups in various collapsed buildings are shown in Figure 37.
As with the observations made in the materials section, problems related to the reinforcement configuration have been frequently observed in Turkey following an earthquake.  [55] did not define irregularities based on story stiffness. The effects of this type of irregularity were first introduced into the 1998 earthquake code [56]. The strong beam-stronger column requirement first appeared in TEC (1998) [56]. The first few stories of some of the buildings collapsed like a sandwich during the earthquake. These buildings had story stiffness irregularities at the ground-story level which likely played a major role in their collapse. Photographs of these buildings are shown in Figure 38. According to TEC (1975) [55], the transverse reinforcement was confined along the length of the short columns. An example of the damage caused to short columns after the earthquake is shown in Figure 39. Reinforcement confinement was not observed in the short columns.

Problems related to inspection of construction works
Before the beginning of the millennium, construction work in Turkey had not been properly inspected. Reports related to soil investigations and structural designs were not always reviewed by an institution. During construction, the strength of the materials (concrete and steel) used was not tested, and the correct application of the structural drawings in terms of the reinforcement configuration and dimensions of the structural members was not confirmed. In some collapsed and severely damaged buildings, the SER prepared before the earthquake indicated that the reinforcement configuration of the structural drawings differed from that of the constructed buildings. Furthermore, the sizes of the constructed structural members did not match those in the structural drawings of some buildings [58,59]. For example, a column of 250 × 800 mm in size in structural drawings was constructed in 1988 as 250 × 500 mm along the height of a 9-story building. The inspection mechanism before (the design stage) and during the construction process changed significantly during the 2000s. After the earthquake, the conditions of numerous damaged buildings in İzmir were investigated by experts from the Turkish Ministry of the Environment and Urbanization. Based on their findings, some buildings were determined to be unsafe for occupants and were either demolished immediately or designated for repair and/or strengthening. During this post-earthquake assessment process, we observed that the structural members of some buildings were damaged or destroyed for various reasons before the earthquake. Examples of this type of damage and destruction are shown in Figure 40. The bottom longitudinal reinforcement of the basement beam shown in this figure was cut at midspan, and the top longitudinal reinforcement was damaged at the support region to instal utility pipes inside the building. The stirrups were also cut from the damaged sections.

Problems related to soil investigation
In Turkey, early geotechnical site studies were initiated in 1986 as part of the requirements for zoning plans [63]. However, it was not until two major earthquakes occurred, the Kocaeli Earthquake in August 1999 and the Düzce Earthquake in November 1999, that the existing law was amended to require geological site reports as part of the building permit acquisition process [64]. Because of problems regarding the content of the geotechnical reports, the government intervened and furnished the details of a typical site report. As part of these efforts, in January 2000, the Ministry of Public Works issued a memorandum to make geotechnical site reports a structural design requirement. As the historical evolution of geotechnical site reports in Turkey suggests, soil-associated structural design parameters were not prioritised until the early 2000s. This underscores the problems in buildings constructed before 2000. Most likely, this problem played a significant role in the buildings that collapsed and experienced moderate-to-major damage in Bayraklı District [7,11].

Conclusions
The following conclusions can be drawn based on the technical team's site observations: -Local soil class (site effect) is an important parameter that should be incorporated into earthquake analysis and building design. The buildings in Bayraklı suffered mostly from this phenomenon. Therefore, thorough earthquake analyses should be performed for all existing buildings in the area. -TBEC (2018) [41] requires a site-specific elastic acceleration spectrum only for local soil class ZF. Although the code allows design engineers to request similar studies for other soil classes, there is no clear justification for this request. Thus, adding a statement to the code requiring a sitespecific elastic acceleration spectrum in areas like Bayraklı would help satisfy earthquake design requirements.
-To prevent any further loss of life and property during future earthquakes, a detailed earthquake hazard assessment should be prepared, specifically for buildings located in the İzmir Bay area. GRAĐEVINAR 75 (2023) 5, 451-470 Halit Cenan Mertol, Gokhan Tunc, Tolga Akis -The fully and partially collapsed buildings in İzmir during the Aegean Sea earthquake were constructed when TEC (1975) [55] was in effect. However, many requirements in the code were not followed during the design or construction stages.
-Inspections during the design and construction phases were not properly performed for buildings constructed before 2000. Therefore, all RC buildings constructed in Turkey before 2000 should be re-evaluated for their structural performance.
-Administrative precautions should be taken to sustain the structural integrity of RC buildings during their lifespan.