Performance of RC buildings after KahramanmaraŞ Earthquakes:lessons toward performance based design

Baris Binici, Ahmet Yakut, Koray Kadas, Ozan Demirel, Ugur Akpinar, Afsin Canbolat, Firat Yurtseven,Orkun Oztaskin, Selin Aktas and Erdem Canbay

Department of Civil Engineering, Middle East Technical University, Ankara, Turkey

Abstract: Two simultaneous earthquakes occurred in the Kahramanmaraş-Pazarcık and Kahramanmaraş-Elbistan districts of Turkey on February 6, 2023, and with magnitudes of 7.7 and 7.6, respectively. These events caused the highest estimated loss recorded in Turkey within the last century from natural disasters. The key reason for the extensive loss was the proximity of eleven cities to the earthquake epicenters. Middle East Technical University teams investigated the building sites in Gaziantep, Kahramanmaras, Hatay, Adiyaman and Adana. The ground motion recordings revealed that in certain locations of Gaziantep, Kahramanmaraş and Hatay, the ground motion levels exceeded the maximum credible earthquake level defined for a return period of 2,475 years in the Turkish Earthquake Code. Residential building performance was investigated with respect to the construction year, which is a good indicator of compliance with modern seismic codes and inspection procedures. About 97% of the collapsed buildings were constructed prior to 2000, whereas over 5,000 buildings,which were built after 2000, collapsed or required urgent demolition. Most of the buildings with minor or greater structural damage sustained heavy infill wall damage rendering occupancy impossible. Aside from damage in older construction with significant structural deficiencies, the damage in some of the more recent and better constructed buildings was observed to be surprisingly poor. This can be attributed to the level of ground motion, significant ductility demands, poor material and workmanship and damage to non-structural elements. With the estimated total loss of above 100 billion dollars and over 50,000 casualties, the current seismic design criterion based on ductility and acceptance of structural damage should be reevaluated to ensure a more resilient urban environment in high seismic regions.

Keywords: earthquake damage; Kahramanmaras earthquake; RC Building; seismic codes; poor performance

1 Introduction

Two earthquakes ofMw=7.7 and 7.6 hit Kahramanmaraş-Pazarcık and Kahramanmaraş-Elbistan,respectively, on February 6th of 2023. Subsequently, a significant aftershock ofMw=6.6 occurred in Nurdağı,followed by another earthquake ofMw=6.4 in Hatay-Defne. This sequence of events caused more than 100 billion dollars in damage and over 50,000 casualties in Gaziantep, Kahramanmaraş, Hatay, Malatya and Adıyaman with less but significant damage in Adana,Şanlıurfa, Diyarbakır, Elazig, Osmaniye and Kilis.Survey teams from the Earthquake Engineering Research Center (EERC) of the Middle East Technical University(METU) were deployed to the field to carry out a detailed damage assessment focused on the Adana, Gaziantep,Hatay, Kahramanmaraş and Adıyaman regions (METU,2023). In the meantime, damage assessors pre-trained by the Ministry of Environment and Urbanization(MEU) carried out global building damage assessments according to the methodologies developed by Ilkiet al.(2020) and Bodurogluet al. (2013). It is important to note here that the buildings for which inquiries are made were assessed by the government appointed assessors,including those from government organizations and universities.

The Turkish Earthquake Code evolved in parallel to other modern codes (mostly following North American Guidelines) in the last five decades. The TEC (1975),which was followed by all buildings designed pre 2000,suggested the use of a simple inelastic design spectra for four earthquake zones. The highest inelastic design force was 10% of the building weight for ductile reinforced concrete frames with infill walls. An equivalent lateral force procedure was recommended in TEC (1975),whereas the detailing of column and beam ends had confining end zones but there was no requirement for the unsupported length of column longitudinal rebars. The key shortcomings of TEC (1975) are as follows:

(1) Low design seismic forces (compared to more recent versions in 1998, 2007 and 2018 as discussed later)

(2) Use of only the equivalent lateral force procedure and absence of considering higher mode effects and irregularities with modal analysis

(3) Allowing the use of concrete compressive strength as low as 18 MPa (in seismic zones 1 and 2)and plain rebars

(4) Lack of explicit capacity design principles for shear design of columns

(5) Absence of the strong column-weak beam concept

(6) Lack of any axial force limit on columns for ductility requirements

TEC (1998) borrowed its fundamentals from UBC(1994) and made up for almost all the shortcomings of TEC (1975). A simplified version of the seismic hazard study by Gülkanet al. (1993) was employed to define the design response spectra in TEC (1998). Capacity design concepts, detailing requirements for confinement,and modal analysis requirements for irregular buildings became the standard application in Turkey, in addition to the 0.5fcAgrequirement for RC column axial load limit after 2000. TEC (2007) continued the same rules of TEC-1998 for building design with the addition of a new chapter on seismic assessment. The most recent code (TBEC, 2018) that came into action in 2019 employs a site-specific hazard map for the design spectra and preserves the approach of TEC (1998) with some modifications in line with ACI 318-17 for member design and detailing (i.e., the column axial load limit was reduced to 0.4fcAg). This brief discussion of the past codes is perhaps a prerequisite to understanding the sustained damage-building age relationship. A comparison of the elastic response spectra given in different codes for soft and stiff soils are given in Fig. 1.

The quality of construction and inspection in Turkey in relation to the seismic damage as reported after past earthquakes should be noted (Yakutet al., 2022;Bayraktaret al., 2015; Spenceet al., 2003). However,emphasizing the deficiencies of the 1975 code alone does not adequately explain the building damage that is discussed later in detail. The use of plain bars, insufficient splice lengths, lack of proper confining reinforcement at column ends in addition to low concrete compressive strength (about 8–10 MPa in older construction) due to construction errors and lack of inspection are the wellknown key issues that resulted in non-ductile response of pre-2000 buildings in Turkey.

The objective of this study is to present the observed damage in reinforced concrete buildings after the Kahramanmaraş Earthquakes. First, a brief evaluation of the recorded strong ground motions is presented by comparing the response spectra with the code given design response spectra at different locations. Next, the observed building damage is presented by focusing on the construction year, ground motion level and lateral force resisting system (frame vs. shear walls) and infill walls.Finally, the reasons for expected damage, especially in newer construction, are explored by presenting the ductility and deformation demands in relation to the performance limits given in TBEC (2018).

2 Evaluation of recorded strong ground motions

The first earthquake with its epicenter in Pazarcik occurred with a rupture of the North-Anatolian fault well above 300 km between Hatay and Malatya. The second earthquake occurred due to the rupture of the Çardak-Sürgü fault zone in the east-west directions spanning between Kahramanmaraş and Malatya. Strong motions from the earthquake recorded by several stations operated by AFAD (2020), are shown on the topographical map of the earthquake area in Fig. 2. The figure also shows thePGAof the ground motion stations in the most severely stricken region.

Response spectra of the nine selected stations with different ground motion intensities are presented in Fig. 3. Note that the larger of the ground motions from subsequent earthquakes are used to obtain the response spectra. It can be seen that significant intensities were observed in Göksun, Fevzipaşa and Hatay regions,which were well above the design response spectra with a return period of 475 years used in building design (i.e.,DD-2) exceeding the maximum credible earthquake levels estimated for a return period of 2,475 years (i.e.,DD-1) in certain periods. In Kahramanmaras, Kirikhan and Turkoglu stations, the response spectra of recorded ground motions were above the DD-2 levels of seismic hazard not exceeding DD-1. The ground motions recorded in Afşin, Gölbaşı and Erzin were below the DD-2 levels at low- to medium period ranges. The ratio of the spectral accelerations to the corresponding DD-2 levels are presented in Fig. 4. It can be observed that spectral acceleration values exceeded the TBEC (2018)values by a factor of 1.75 and 4.65 in the 0–0.5 s and 0.5–1.5 ranges, respectively, for the high intensity regions.This demonstrates the significant underestimation of the seismic design demands at certain locations.

Fig. 1 Comparison of code design spectra

3 Damage count

Fig. 2 Fault locations, strong motion stations and highest PGA values (upper values for Pazarcık and lower values for Elbistan)

Fig. 3 Comparison of design spectra according to TEC (1975), TEC (1998), TBEC (2018) for different locations (TEC (1998) and TEC(2007) provide the same design response spectra)

The number of buildings that sustained various levels of damage as of March, 2023 are given in Table 1(Ministry of Environment, Urbanism and Climate Change). Buildings are classified as Urgent Demolition if they are identified to be on the verge of collapse. It can be seen that significant damage was observed in five cities (Gaziantep, Kahramanmaraş, Adiyaman, Hatay,Malatya) whereas the remaining six cities were affected to a lesser degree due to the remoteness to the fault zones. According to the data disseminated by the Turkish Statistical Institute (TSI), the population of buildings in eleven affected cities is around 1,800,000 as of the end of 2021. In the affected region, almost 87 percent of the buildings are made of reinforced concrete, and other structural systems (masonry, steel etc.) constitute about 13 percent of the entire inventory.

The reported number of total/partially collapsed buildings was about 50,000, which did not satisfy the seismic design target of no collapse. These buildings were the key cause of the fatalities. It was stated by the Minister of Environment, Urbanisation and Climate Change (Kurum, 2023) that about 97 percent of the collapsed buildings had a construction year before 2000.As discussed above, construction year 2000 is usually regarded as the onset of the significant change in the practice. More than 5,000 buildings with a construction year of after 2000 collapsed or sustained severe damage and/or required urgent demolition. Despite possible construction deficiencies observed in some of those buildings, the lack of seismic resistance and observed severe damage in a significant number of buildings resulted in an estimated loss of about 100 billion dollars.

4 Observed performance of RC buildings

4.1 Performance of RC buildings constructed before 2000

The deficiencies in the building stock of Turkey constructed before 2000 was observed in the Kocaeli,Duzce, Bingöl, Van, Elazig and İzmir Earthquakes(METU, 2011a, 2011b, 2011c; Bayraktaret al., 2015;Scawthorn, 2000; Spenceet al., 2003; METU, 2020a,2020b; Yakut, 2021). Similar buildings in the earthquake zones were constructed by using smooth reinforcing bars, insufficient steel reinforcement detailing and low concrete strength resulting in heavy damage and collapse (Fig. 5). The presence of soft stories due to the use of the first story as a commercial store in the ground level was one of the key structural reasons of collapse in many buildings. In those floors, infill walls are removed to create open floor space resulting in plastic hinging in columns and pancake type collapse as shown in Fig. 6.Several buildings have experienced beam-column joint failures due to insufficient transverse reinforcement of the joints. The distinction between the seismic performance of buildings constructed before and after 2000 was observed at several locations, where adjacent buildings behaved differently. Figure 7 depicts such an occurrence in Antakya, Hatay, in which the building on the left was constructed after 2000 and showed a satisfactory performance, whereas the brown building on the right was built before 2000 and collapsed. Another interesting type of failure was the overturning of the building from its base due to the inability of transferring lateral forces to the foundation (Fig. 8). It was observed that in regions of strong ground shaking with soft soil conditions, buildings with one or two basements survived in Adıyaman-Golbasi, while structures built on surface foundations or with small embedment depths failed by loss of bearing capacity and overturning (Fig. 8).

Fig. 4 Ratio of measured spectral accelerations to code given values

Table 1 Damage assessment results of MEU as of March 2023 (number of units)

Fig. 5 Building collapse in Antakya and Kahramanmaraş(inadequate detailing, plain bars)

The building collapses for the older construction is observed to stem from the following sources:

(1) Lack of lateral rigidity

(2) Insufficient lateral design forces given in the older codes

(3) Lack of reinforcement detailing for the column transverse reinforcement and beam-column joints

The drift demands in relation to deformation capacities for these buildings is discussed later.

4.2 Performance of RC buildings constructed after 2000

Fig. 6 Building collapses in Gaziantep, Malatya and Kahramanmaras (pancake collapse, soft story collapse)

Fig. 7 Difference in seismic performance of two adjacent buildings (gray building on the left: post-2000, brown building on the right: pre-2000)

These buildings presumed to be designed and constructed according to the Turkish Earthquake Code(1998) usually performed better than the older buildings.However, more than 5,000 buildings constructed after 2000 collapsed, violating the code given performance objective. This appears to be an important observation demanding further investigation on the design and construction quality of those buildings. Some examples of these heavily damaged buildings are shown in Fig. 9.The possible reasons for this damage can be attributed to:

(1) Use of flexible joist slabs as diaphragms

(2) Insufficient engineering design to distribute lateral forces to vertical load bearing elements, perhaps due to blind use of building design software

(3) Possible detailing errors on building construction site

(4) Underestimation of seismic demands

(5) Insufficient investigations of local soil conditions prior to building construction especially in the Hatay and Gölbaşı region.

Such heavy damage observed in new buildings raises concerns about the target seismic performance of residential buildings nearly in compliance with the current seismic code. The significant disruption of city life,heavy monetary loss and long recovery times may require reassessment of the performance targets of buildings.

4.3 Buildings with shear walls

Fig. 8 Overturning of buildings and ground settlement in Gölbaşı

Fig. 9 Damage to Post-2000 buildings

Tunnel form building construction is a popular structural system for low-cost mid-rise buildings in Turkey. Typically, they are constructed with shear walls having 20 cm thick walls, 0.2% web reinforcement and 1% boundary region reinforcement. The area of the walls to the floor area is around 2.5% in these buildings as no columns are used. There were about 200 buildings with 4 to 11 stories that were affected by the strong motions.It can be stated that all the tunnel form buildings satisfied the collapse prevention performance target without any record of collapse. Typical damage observed in these buildings is shown in Fig. 10. In tunnel form buildings with 4 to 6 stories, almost no damage was observed in the shear walls. The infill walls typically constructed with autoclaved aerated concrete or BIMS sustained diagonal cracks and separation from the wall interfaces. The most important structural damage was the shear cracks observed in the link beams and cracking in the slabs.As the lateral strength of the walls is utilized, the load transfers from the diaphragm and link beams to the walls were not sufficient. For the taller tunnel form buildings with 11 to 13 stories, shear wall damage in the form of corner crushing rebar buckling accompanied by shear cracking at the first floor was seen as evidence of flexure shear interaction. It can be stated thatPI(PI=ratio of cross-sectional area of shear walls to the building total floor area as stated by Hassan and Sozen (1997)) value of about 0.002 eliminated collapse, whereasPIvalues around 0.005 were sufficient to limit the damage in the shear walls.

4.4 Precast buildings

Many industrial precast buildings in Kahramanmaras and Gaziantep with one or two stories sustained damage(Fig. 11). Typically, the main direction of these industrial buildings have a span of about 20 m while the spans in the other direction are 7.5 m. The story heights vary between 7 to 10 m. The building frame columns are fixed at the base with a socket connection, whereas the prestressed (for long spans) roof girders are pinned to the column corbels usually with two grouted anchors embedded into the corbels. Two buildings that were under construction collapsed due to the overturning of the girders as the pin anchors failed due to in and out of plane deformation demands. Interestingly no indication of column base hinging was observed in most buildings,indicating sufficient sizing and detailing of the columns.In Gaziantep, most of the prefabricated buildings showed satisfactory performance. In a few lightly damaged buildings, the typical damage observed was in column corbels due to the girder rotations causing local crushing of concrete, which is repairable (Fig. 11). In short,the observed damage on low rise precast factory type buildings demonstrates the importance of the seismic resistance in the beam-column connections, which determined the final damage state.

4.5 Performance of non-structural elements

Infill wall damage continues to be an important observation after the Kahramanmaraş earthquakes.Several levels of damage were observed depending on the strong motion levels (Fig. 12). In regions with even low-level recorded accelerations (PGA<0.1 g), the infill wall-column/beam interface cracks were visible. In regions with moderate levels of strong motion, the infill walls sustained inclined cracks with varying widths (0.5–2 mm). Under such damage, despite the absence of any structural damage, the occupants left the building and were reluctant to occupy it after the earthquake. Similar to past observations in Van (METU, 2011b; 2011c),the infill walls are observed to be the key components to establish the damage state of a building affecting the psychology of the occupants. For regions that have experienced significantPGAvalues, the collapse of the infill walls was observed due to combined in and out of plane demands. In the latest seismic code starting from 2019, the separation of the infill walls is deemed necessary while preserving the out of plane strength of the infill walls. Otherwise, more strict drift limits should be followed. No such detailing was observed in the newer construction, which appears to point to the continuation of inappropriate infill construction techniques. Unlike the structural damage, the infill wall damage was similar in buildings constructed both before and after 2000, exhibiting no significant difference in their performance. Furthermore, the damage was similar in all the infill walls made of hollow clay brick,autoclaved aerated concrete or BIMS blocks, indicating that none of the block materials showed superior seismic performance. Employing appropriate solutions for infill wall construction appears to be one of the most important issues for seismic risk reduction based on our observations.

Fig. 11 Precast buildings in Kahramanmaraş and Gaziantep

Fig. 12 Infill wall damage examples

5 Seismic performance in relation to drift demand

To investigate the reasons for the extreme damage observed in relation to the recorded ground motions,drift demand map of the area was estimated by using the methodology of Akkaret al. (2005) along with Ay and Akkar (2008), who proposed the maximum ground story drift demand equations from a simplified shear-flexure beam model with statistical corrections. Accordingly,the maximum inelastic inter-story drift ratio (MIDR) can be approximated by the following equation:

where1γand2γare the correction factors as a function of estimated sum of beam to column stiffness,his the story height,His the total building height andSdis the spectral displacement demand for a prescribed fundamental period (T) . The following assumptions were made in using these equations:

(1) Story height is assumed as 3 m.

(3)Sdvalues were determined from the spectral accelerations of each recorded ground motion at the assumed fundamental period.

(4) Drift demands were determined for constant fundamental period values of 0.25, 0.50, 1.0, and 1.5 s.These values represent typical low- (2–4 stories) to mid-rise (5–10 stories) buildings as well as ranges of amplified spectral accelerations in the recorded ground motions.

(5) The drift ratios are spatially interpolated by using the calculations at recorded motion locations.

The results of drift demands are presented in Fig. 13.It can be observed that in regions of significant damage(Nurdagi, Islahiye and Hatay), drift demands of around 2%–4.5% were calculated. The drift demands increase with the period and reach up to 4.5% in Islahiye forT=1.5 s.Even if the buildings are code compliant in terms of seismic design, significant damage is expected due to such as excessive drift limits. The response modification factor-drift demand relations are plotted in Fig. 14. It can be observed that forRvalues of 6 to 8, the drift demands can grow as high as 6%. These plots show the significant deformation demands on the low period (lowrise) buildings designed as high ductility systems. The drift demand increases with the period of the building regardless of the response modification factor. For low period buildings, drift demands of as high as 8–9 percent are obtained whereas for high period buildings, lower drifts on the order of 5 percent were computed.

Fig. 13 Estimated drift demand maps for T=0.25, 0.50, 1.0 and 1.5 s

Fig. 14 Response modification factor (R)-MIDR relations for periods T=0.25, 0.50, 1.00 and 1.50 s

The extensive damage observed after the Kahramanmaras Earthquakes, whose spectra at some places reached as high as DD-1 level (maximum credible earthquakes in Turkish Earthquake Code), if not larger,appears to require the need to question whether assuming high ductility (i.e., response modification factors up to 8)and limiting drift limits computed based on a supposed design spectra may ensure safety and be economical in structural designs. The higher drift demands at larger response modification factors due to unexpectedly larger ground motions seem to be among the most important factors affecting the extent of damage. There is perhaps a need to rethink the high ductility design strategy if one cannot accurately estimate drift demands. In countries such as Turkey with limited ability for building inspection, multiple design errors and inadequate attention to details, it could be more economical to design for higher seismic forces with more strict drift control. The evidence on the success of drift control can be seen from the tunnel form buildings. It is observed that with sufficient shear wall area in building design,effective interstory drift ratio control (less than 2%)can be provided, even under MCEs with corresponding response modification factors of 2 to 4.

6 Conclusions

The Kahramanmaraş Earthquake reconnaissance studies conducted by the Middle East Technical University team is presented in this study. The building damage investigations are divided into two categories based on building construction period. For buildings constructed before 2000, attributes and patterns of damage are observed to be similar to the past earthquakes in Turkey. However, it was perhaps overly optimistic to expect better response as thousands of recently constructed buildings collapsed. The reasons for these collapses can be attributed to the intensity of the strong ground motion, design and construction detailing errors and inadequate inclusion of soft soil effects. The high drift demands estimated in the region are alarming from the point of view of performance-based earthquake engineering, which relies on deformation demand and capacity calculations. When the ground motions reach such intensities and the inability to estimate building response are added to the possible design and construction problems, the application of classical design approaches need to be revisited to minimize the potential seismic risks. Even though collapses could have been prevented with more strict building inspections, the monetary loss and the psychology of the earthquake victims could be helped by using simpler structural engineering rules such as drift controls and shear wall area requirements rather than complicated and detailed response analysis.