Skip to main content
European Commission logo print header

Istanbul Urban Earthquake Test Site

Final Report Summary - URBANQUAKE (Istanbul Urban Earthquake Test Site)

URBANQUAKE is designed to install and monitor seismic array networks in Istanbul where the probability of having a major earthquake in the next 30 years is very high. The real earthquake data that is collected using these arrays are highly valuable for the scientific community around the world since they constitute one of the most reliable source of information to develop, verify and modify the existing analysis and design procedures in earthquake engineering. In this final summary report, an executive summary of the work undertaken during the period of project is provided followed by a brief description of the project context and objectives. Main scientific achievements are described along with the dissemination activities that have been done so far to promote foregrounds of the project. Potential impact, plans for further development and exploitation of the URBANQUAKE also discussed.

Executive summary

Estimation of site specific ground motion characteristics has been a crucial issue in the assessment of the vulnerability of the existing structures, for retrofit and rehabilitation alternatives as well as in the design of new structures. The scientific aspects contain significant degree of uncertainties and reliable solution of this problem requires development of realistically comprehensive analytical models that rely on large amount of field evidence. Seismic arrays, which are deployed in seismically active areas to record ground motion during earthquakes, are one of the ways to collect field evidence for response of soil layers. The information and data provided by these arrays is considered very crucial for verification, calibration and development of predictive tools that are used in earthquake engineering.

In the URBANQUAKE project, a strong motion network of vertical arrays has been installed within the city of Istanbul, Turkey complementing already existing horizontal Rapid Response network. The existing high seismic activity of the region increases the scientific importance of these arrays because large number of earthquake records could be obtained within relatively short period. In order to achieve this objective, up to EUR 350 000 research grant is secured through various national institutions and agencies, and all this support is used to deploy 'vertical seismic arrays' which will complement and improve Istanbul Rapid Response Network composed of 100 strong motion stations and form a fully functional urban earthquake test site. Vertical seismic arrays are composed of several accelerometers installed in boreholes at various depths in the soil profile. The real earthquake data that is collected at urban earthquake test site is evaluated through numerical and analytical methods by utilising the detailed assessment of site conditions that are compiled based on the extensive microzonation project that was conducted by Istanbul Metropolitan Municipality. The data that has already started to accumulate from the urban earthquake test site enables progress in development and calibration of site response analysis and site-specific damage scenario methodologies.

The major impact of the project today and in the future is its unique potential to generate invaluable data about site effects in earthquake engineering which is today regarded as one of the most-needed type of information to develop, modify and verify the current analysis and design procedures in the profession. The impact potential of project and the achievements attained in the first project period has also been acknowledged by CORDIS; the project has been selected for special promotion on Technology Marketplace and is announced as an exploitable technology on official website.

URBANQUAKE has been presented in various international conferences and workshops as invited lectures and as articles in peer-reviewed publications. National and international funds that are obtained with the efforts of the researchers of this project are used to cover travel expenses of all dissemination activities. All these efforts for dissemination have brought an international recognition to Istanbul Strong Motion Network. An ongoing collaborative Seventh Framework Programme (FP7) project titled 'Seismic engineering research infrastructures for European synergies' (SERIES) has already proposed to use data that has been accumulated at URBANQUAKE to develop new techniques for experimental studies of seismic wave propagation. The institution (i.e. KOERI) and the researchers of this project are a part of the consortium formed for SERIES. In the meantime, collaborations with United States (US) and European based organisations such as Network for Earthquake Engineering Simulation (NEES) and the Euroseistest Site have been initiated to share and exploit foregrounds of the project.

Significant efforts have also been spent in order to continue to build the capacity of the test site. A comprehensive research project have been prepared which proposes to implement a large-scale research capacity development in geotechnical, geophysical and structural earthquake engineering area using resources offered by Turkish Central Planning Agency. The project proposal focused on acquiring state-of-the-art systems for laboratory and in-situ measurements of engineering properties of soil layers and soil-structure systems and extending the seismic urban network in Istanbul by additional arrays of both permanent and portable nature including an in-situ 'liquefaction' test site. The proposed development with a EUR 1.5 million budget will equip KOERI with the state-of-the-art research capacity in earthquake engineering that is compatible with those available in US and Japan and will provide a unique platform for collaborative research throughout European Union (EU) community. The researcher supported by URBANQUAKE has a leading role in preparation and execution of the proposed project which at this time has successfully passed the first evaluations and is in process of further review.

Strongly related to the achievements described above, academic performance of the researcher, which has accelerated during the period of this project, has enabled her to apply for a permanent position at KOERI. Her integration to the institution will contribute significantly to the research capacity and publication rate of the Institution and will pave the way for more research projects both at national and international scales.

Context and objectives:

The project had multiple objectives. The first objective was to deploy and ensure successful operation and monitoring of vertical seismic which would complement and improve Istanbul Rapid Response Network composed of 100 strong motion stations and form a fully functional urban earthquake test site. In order to achieve this objective, significant amount of financial support was needed, therefore one of the crucial tasks was to prepare and submit project proposals to various funding institutions and agencies and obtain the necessary research grant for instrumentation and field work.

The second objective involved analytical evaluation of real earthquake data collected at the URBANQUAKE by utilising the detailed assessment of site conditions at the strong motion array locations. Site conditions would be compiled based on the extensive microzonation project which was being conducted by Istanbul Metropolitan Municipality under the technical guidance of KOERI as well as on the site-specific investigations that would be conducted at the vertical seismic array sites by means of field and laboratory testings.

The third objective was to use the data that will accumulate at the urban earthquake test site to make progress in development and calibration of site response analysis and site-specific damage scenarios.

The fourth objective was to bring international recognition to urban earthquake test site and to disseminate scientific achievements related to the project through of international conferences and peer-reviewed articles; and by doing that to pave the way for collaborations with well-known research institutions to share and exploit the foregrounds of the project.

The fifth objective was to ensure integration of the researcher to the Institution by offering her a research environment where she could get a permanent position as a result of her academic performance.

All these objectives have been fulfilled during the course of the project. Moreover, significant efforts have been spent to obtain a much larger-scale financial support through national funding agencies for further building the capacity of URBANQUAKE and creating a center of excellence for earthquake engineering at the Institution.

Science and technology (S&T) results

In accordance with the objectives briefly described above, an urban earthquake test site has been formed by deploying vertical seismic arrays at selected locations in the western European side of Istanbul which are complementing the Istanbul Rapid Response Network composed of 100 surface strong motions. Each vertical array is composed of four or three borehole accelerometers and one surface accelerometer. Financial support for all instrumentation and field work is granted by national organisations such as The Scientific and Technological Research Council of Turkey (TÜBITAK) and Turkish Central Planning Agency. At this time, all instrumented sites are in operation.

Soil conditions at the instrumented sites are evaluated based on the geophysical and geotechnical data compiled from an extensive microzonation project for south western part of Istanbul which was conducted by the Metropolitan Municipality. The faculty members of KOERI was providing technical support to the project where approximately 2900 boreholes were drilled within an area of about 182 km2 to investigate local soil conditions. Soil conditions are further investigated at the vertical seismic array sites by means of site-specific field and laboratory testings. The sites are briefly described below.

Ataköy array

This array has been deployed in late 2006 before the start date of URBANQUAKE grant. It has been the example for other arrays that are installed within the context of this project. Brief information about the array is provided here since it constitutes a part of URBANQUAKE and its effective monitoring is one of the objectives of the project.

Ataköy vertical array (ATK) is located close to the Atatürk International Airport and represents a stiff-soil / rock site. Geologically, the site can be characterised with a Miocene age unit known as Güngören formation underlain by Eocene age Ceylan formation. Güngören formation is mostly associated with hard clay and sand layers but also includes limestone. Ceylan formation consists of sandstone and claystone and represents bedrock layer with an approximate depth of 100-150 m at this part of the city. Geotechnical site investigations carried out at the array location revealed that the soil profile is composed of alternating hard sandy clay and dense to very dense clayey sand layers down to 110 m while sandstone is encountered below this depth. Between the depths of 5 m-34 m there is a layer of highly weathered limestone with clay interlayers. These clay interlayers as well as the overlying sandy clay layer are classified as CL (PI=16-26, FC=52-56%). Below the depth of 35 m, clay layers are more plastic (PI= 51-56) and have higher fines content. Dense sand layers down to the depth of sandstone layer are classified as SM and compromised of 16-34 % fines with nonplastic nature. The equivalent shear wave velocity for the top 30 m of the soil profile (VS30) at the site is approximately 278 m/s.

ATK is composed of four triaxial accelerometers that are installed at depths of 25, 50, 70 and 140 m and one triaxial accelerometer on the ground surface. Sensors at depths of 25 m, 50 m and 70 m are Kinemetrics SBEPI, while the deepest sensor is Kinemetrics FBA ES-DH (with in-built compass). All four of the borehole sensors are connected to a 12-ch Kinemetrics K2 digital recorder. A second digital recorder, a 4-ch Kinemetrics K2 with internal three-component accelerometer, records the motion on the ground surface. Common triggering and GPS timing between the borehole sensors and the sensor at the ground surface allows for synchronised recording. Threshold-triggered data collected at a rate of 200 samples per second is periodically retrieved via ADSL connection. Among the borehole sensors only the deepest borehole sensor has a built-in compass for accurate orientation. The orientations of other three borehole sensors are determined through analysis of recorded acceleration-time histories (Kurtulus et al., 2008).

Zeytinburnu Array

Zeytinburnu vertical array (ZYT) is located close to the shoreline at south west of the city and represents a relatively soft-soil / rock site. According to recently conducted seismic damage scenario studies, Zeytinburnu district represents one of the most vulnerable areas in the city (Ansal et al., 2010). Geologically, the site can be characterised by Miocene age units known as Bakirköy and Güngören formations. Bakirköy formation overlies Güngören formation and is associated with a 5 to 10 m thick limestone layer containing thin layers of clay. Underlying these is the Paleozoic age Trakya formation composed of sandstone and shale (greywacke) which represents the bedrock layer. Trakya formation is encountered at a depth of about 200-250 m at this part of the city.

Geotechnical site investigations carried out at Zeytinburnu vertical array location revealed that the soil profile is composed of alternating clay and silt layers down to 230 m while greywacke is encountered below this depth. Between the depths of 5 m-32 m there is a layer of weathered limestone with clay interlayers. Clay layers alternate between CL (PI= 13-22) and CH (PI=37-67) while silt layers are classified as MH (PI=22-49). Fines content of CL, CH and ML materials are typically very high (FC=72-100). Thin layers of silty sand (SM) are encountered occasionally between the alternating clay and silt layers. VS30 at the site is equal to 263 m/s.

ZYT is composed of three triaxial accelerometers that are installed at depths of 30, 57, and 288 m and one triaxial accelerometer on the ground surface, but with more up-to-date instrumentation. All boreholes are instrumented with Kinemetrics FBA ES-DH (with in-built compass) while an external three-component accelerometer (Kinemetrics EpiSensor) is installed on the ground surface. All sensors are oriented to exact North. Sensors are connected to a 12-ch Kinemetrics rock digital recorder which allows for threshold-triggered data collected at 200 samples per second to be automatically sent to an ftp website of KOERI via ADSL connection.

Fatih array

Fatih vertical array (FTH) is located on a hill adjacent to a seismically monitored historical mosque within the old city and represents a stiff-soil / rock site. Historically, the site has always been an important location for religious structures (dating as early as 550 BC) most of which had been significantly damaged during the past major earthquakes. Ground accelerations as high as 0.2 g have been recorded at this site during the Mw= 7.4 1999 Kocaeli Earthquake which was located about 100 km away from Istanbul. Geologically, the site has similar features as ZYT site which can be characterised by the Miocene age units known as Bakirköy and Güngören formations underlain by the Paleozoic age Trakya formation representing the bedrock layer. Trakya formation is encountered at a depth of approximately 60-80 m at this part of the city.

Geotechnical site investigations carried out at this location revealed that the soil profile is composed of alternating clay, silty sand layers up to 40 m while greywacke is encountered below this depth. It was observed that the top 40 m of greywacke is considerably weathered and the intact material is located at a depth of about 80 m. Clay layers are mostly classified as CH (PI = 32-59), occasionally as CL (PI = 14-24) at shallower depths (< 20 m). Fines content of these materials are variable (FC = 63-93 %). Silty sand (SM) layers with FC = 14-43 % are encountered between the clay layers at the depths of 17 to 32 m. VS30 at this site is equal to 335 m/s. FTH is composed of four triaxial accelerometers that are installed at depths of 23 m, 60 m, 136 m and on the ground surface. Instrumentation and field set-up at this site is identical to that at ZYT array. Given the fact that Fatih Mosque is also instrumented and monitored by KOERI, the site is expected to provide valuable data not only for free-field soil response but also for soil-structure interaction related studies.

Istanbul Rapid Response Network (IRRN)

IRRN, operated by KOERI, is composed of 100 strong motion stations distributed within the city of Istanbul. The network consists of strong-motion instruments (Güralp CMG-5T) located at grade level in small- to medium-sized buildings. Full-recorded waveforms at each station can be retrieved using GSM and GPRS modems subsequent to an earthquake (Erdik et al., 2003). Out of 100 rapid response network strong motion stations, 55 stations are located in the European side of Istanbul. ATK, ZYT and FTH vertical arrays operates in synchronisation with rapid response network and provides reference bedrock motion for the ground motions recorded by this network.

As mentioned before a comprehensive site investigation study involving 2900 geotechnical borings was conducted for this area (OYO, 2007). The data included mostly standard penetration test (SPT), seismic reflection, seismic refraction measurements and some cone penetration test (CPT) and PS-logging measurements as well as a number of laboratory test results from disturbed soil samples. The information compiled is used to obtain a representative soil profile at each IRRN station. Measurements performed at borings conducted in the near vicinity (< 200 m) of the stations are taken into account in determining these profiles. Depths of engineering bedrock are typically estimated based on the 3D engineering bedrock model that was prepared during the Istanbul Microzonation project (OYO, 2007). All station sites have comparable VS30 values while depths of bedrocks are quiet variable.

Ground motions recorded

The arrays are relatively new, they have been in operation for the last few years, therefore only a number of small magnitude earthquakes have been recorded so far. Maximum ground accelerations recorded at the sites so far does not exceed 0.01 g. However, given the high seismic activity of the region, the arrays are bound to generate important scientific data.

Detailed assessment of site conditions at the strong motion array locations allowed for comprehensive numerical and analytical evaluation of the real earthquake data that has accumulated at the URBANQUAKE so far, enabling a number of research studies. The data recorded up to now represent low amplitude motions which allows for linear analysis of soil response only. More records of varying levels of intensity, source and path characteristics that will become available in future will provide opportunities to improve site response models.

One of the first utilisation of data generated by the arrays has been to investigate the predicted and modelled site response given the detailed information about in-situ soil conditions (Kurtulus, 2011). Among the earthquakes listed, 12 March 2008 Çinarcik, 3 October 2010 Marmara and 19 May 2011 Kütahya events have also been recorded by IRRN stations. The effect of distance can be readily seen by comparing response at Kütahya event with that observed during the two other events. Even if similar spectral accelerations are recorded and all at linear range, frequency response can be quiet different depending on the distance and source effects.

Amplification potential of FTH seems to be the greater than the others even though ZYT is a softer and deeper site. Observed spectral ratios show that the average amplification from bedrock to the surface of the soil reaches a factor of 4, 7 and 12 at periods of 0.9 s, 1.5 s and 0.5s at ATK, ZYT and FTH sites, respectively.

At ATK and FTH sites, the values of predominant periods observed from amplification spectra seem to be in agreement with theoretical 1D fundamental periods of 1.1 and 0.6 respectively, that are calculated from the measured wave velocity profiles using the well-known VS=4H/T0 relationship. However, calculated fundamental period suggests a longer period (2.7 s) for ZYT than that observed from the records obtained at this site. On the other hand, the observed higher amplification potential of FTH site is related to the similarity of the fundamental period of the site to that of the recorded event (approximately 0.6 s).

A significant variation is observed for all three events with average cov (coefficient of variation) in the range of 0.6 to 0.8. The observed high scatter of the ground motions recorded during each event suggests that a significant part of the observed variation can be related to site effects. The variations of ground motion parameters with respect to NEHRP (2001) site classes show some indication of site effects (i.e. stations sites identified as site C tend to have lower spectral response) but also demonstrate that VS30 alone is not a sufficient indicator for amplification potential, as observed from the significant scatter in the spectral response of stations located on site D soils. Almost all strong motion stations have comparable VS30 values.

Recorded and modelled site response

An attempt is made to estimate site-specific ground motion parameters at IRRN stations by performing 1D response analysis using Shake91 (Idriss and Sun, 1992) with the assumption that time histories recorded by the deepest accelerometers at ATK, ZTY and FTH vertical arrays can represent bedrock motions for these sites. Site response modelling is carried out for 10 of the IRRN stations that recorded all three events. In the analysis, the bedrock acceleration time histories recorded at ATK, ZYT and FTH are used as input motions. In general, there is a certain agreement between the recorded and the calculated parameters (mean squared error < 0.2). However, as seen in the figure, calculated ground motion parameters are different from each other depending on the input acceleration time history. ZYT seems to be providing the best-fitting bedrock motion. One reason could be the associated bedrock geology; another possible reason is the selection of ground motion parameter used in the comparisons. It may be necessary to investigate other ground motion characteristics to evaluate the most representative bedrock reference.

One other study involves developing a new methodology to characterise dynamic properties of soil layers by most-used geotechnical earthquake engineering parameters such as shear wave velocity and material damping. The data that has been collected at Ubanquake so far is utilised to verify the proposed technique (Kurtulus et al., 2011).

The technique is usually based on cross correlation of waves recorded at different receivers. Elgamal et al. (1995) applied cross-correlation to the Lotung array records and have been able to obtain soil response characteristics. Snieder and Safak (2006) used seismic interferometry by deconvolving waves instead of cross-correlating them and extracted the building response. Mehta et al. (2007) have applied the technique proposed by Snieder and Safak (2006) to Treasure Island array records to extract 1D velocity profile of soil layers. Here, a similar approach is used by first applying deconvolution to waves recorded at different depths to separate the response of soil layers from the incoherent waveforms excited by an earthquake. In contrast to customary, waveforms are deconvolved using the surface recording (instead of motion recorded at the bedrock level) in order to obtain a simple downgoing wave as proposed by Snieder and Safak (2006). Once response of layers are isolated from interacting up and downgoing waves, wave travel times between the receivers are calculated to obtain wave propagation velocities.

The two standard approach to calculate wave travel times between two recording points is to use the time differences between characteristics peaks or to determine the time lag where the cross-correlation of the waveforms has a maximum. However, these methods are acceptable for non-dispersive, non-attenuating media, where the waveforms do not change their shape as they travel. In fact, waves do change their shapes due to attenuation while travelling through soil layers. In other words, the phase shifts between the two records at different depths are caused by the combined effect of wave travel times plus the phase distortions due to damping. It is possible to eliminate the phase shifts introduced by damping on the calculated wave travel times by using the envelope functions of the waveforms. The travel times obtained from the envelope functions will be smaller; the difference representing the phase shift due to damping. The damping values corresponding to these phase shifts can be calculated and will constitute the intrinsic part of attenuation for the soil medium (Safak, 1995).

Acceleration-time histories recorded at ZYT site during the 19 May 2011 Kütahya event are used to demonstrate the applicability of this approach. Figure 13a shows the NS component of the raw records.

Cross-correlation is then used to calculate travel times for the deconvolved waves and their envelopes. As expected, the travel times calculated from the envelopes are smaller than those calculated from the signal itself. It can be shown that the envelope functions are not affected by the dispersive properties of the medium (Bendat and Piersol, 1985). This property of envelope functions allows for determination of phase shifts which can be related to intrinsic attenuation. The approximation comes from the use of the complex wave velocity to in order to incorporate damping to wave propagation. An important advantage of the technique described above is that it provides a way to separate intrinsic attenuation and radiation damping.

Another progress that was made during the course of the project has been the development of a microzonation methodology that incorporates site-specific response analysis to perform damage scenarios for urban areas (Ansal et al., 2009 and 2008). An application conducted for one of the most vulnerable districts in Istanbul has demonstrated the importance of site effects on seismic damage (Ansal et al., 2010).

The proposed methodology is composed of two main phases. The first phase involves generation of microzonation maps with respect to earthquake ground shaking parameters due to the selected regional earthquake hazard scenario. In the second phase, vulnerability of buildings and pipelines are estimated based on the calculated earthquake ground shaking parameters. Results are displayed in damage distribution maps for buildings and pipeline systems that are produced in GIS environment.

The first step is to adopt a grid system that divides the investigated urban area into cells (typically 250 m × 250 m) according to the availability of geological, geophysical and geotechnical data. Variations of earthquake shaking parameters for bedrock outcrop within the area are separately determined for a specified level of exceedance probability or using deterministic simulations. Site characterisation is performed based on available borings and other relevant information by defining one representative soil profile for each cell with shear wave velocities extending down to the engineering bedrock (shear wave velocity, Vs = 750 m/s).

Site specific earthquake characteristics on the ground surface for each representative soil profile are calculated using one dimensional site response analyses, Shake91. Hazard compatible acceleration time histories (in terms of expected fault type, fault distance, and earthquake magnitude) are selected and site response analyses are performed for a selected number of acceleration time histories. It was demonstrated by Ansal and Tonuk (2007) that if limited number of input acceleration time histories (e.g. three records as specified in some earthquake codes) are used, even with scaling to the same peak ground acceleration (PGA) amplitudes for site response analysis, the results in terms of PGA, peak ground velocity (PGV) and elastic acceleration response spectrum (SA) can be significantly different for different sets of input motion records. This would introduce an important uncertainty when estimating the damage distribution. Therefore to partially overcome this issue, one possible option is to use as many acceleration time histories as possible (e.g. 25 to 30) from the hazard compatible bin (in terms of fault type, earthquake magnitude and epicenter distance) as input motions for site response analyses. The selected time histories can be real earthquake acceleration records, or alternatively can be calculated using simulation models (Ansal et al., 2009). In the case of using real acceleration time histories, PGA scaling is adopted as suggested by Ansal et al. (2006a).

Site response analyses using Shake91 provide the variations of PGA and SA on the ground surface. Variation of PGV is determined through integration of acceleration time histories on the ground surface. Average of all spectral acceleration values between 0.1 and 1.0s periods of the elastic acceleration response spectrum (Saavg (0.1-1s)) is calculated as one parameter representing earthquake shaking intensity on the ground surface. Site-specific peak spectral accelerations corresponding to 0.2s (SABorcherdt) are also calculated through empirical relationship proposed by Borcherdt (1994) using equivalent (average) shear wave velocities for the top 30 m of soil profiles (Vs30). Superposition of empirically calculated values (i.e. SABorcherdt) with those calculated using Shake91 (e.g. Saavg (0.1-1s)) provides a general assessment of the variation of site effects and is used as a parameter for microzonation with respect to ground shaking intensity (Ansal et al., 2004a).

In order to assess seismic vulnerability for buildings, two parameters; site-specific short period (corresponding to 0.2 s) and long period (corresponding to 1 s) spectral accelerations are calculated. Site-specific acceleration response spectrum is used to determine spectral accelerations for the short period (Ss) and for the long period (S1). An approach is adopted to determine the best fit NEHRP (2003) envelope to the calculated average acceleration response spectra (Ansal et al., 2006b). All the requirements of the NEHRP design spectra are applied in obtaining the short (Ss) and long (S1) period spectral accelerations. The two independent variables in the developed optimisation algorithm are Ss and S1. The NEHRP design spectrum is preferred because of its flexibility in defining spectral accelerations (Erdik and Fahjan, 2005).

At this stage, microzonation maps for the investigated urban area may be prepared with respect to Vs30, NEHRP site classification, PGA, PGV, Saavg (0.1-1s) SABorcherdt, Ss and Sl. A map representing the ground-shaking intensity is prepared where the estimated relative shaking intensity levels are based on the superposition of two parameters: Saavg (0.1 - 1 s) and SABorcherdt. The approach adopted in the assessment of the calculated microzonation maps using Saavg (0.1 - 1 s) and SABorcherdt, involves the division of the area into three zones as A, B, and C (Ansal et al., 2004b). Since the site characterisations, as well as all the analysis performed, require various approximations and assumptions, it is preferred not to present the numerical values for the microzonation parameters. The variations of the parameters are considered separately and their frequency distributions are calculated to determine the boundaries between the three zones. The zone C shows the most unsuitable 33 percentile (e.g. high spectral accelerations), zone B the medium 34 percentile and zone A shows the most favorable 33 percentile (e.g. low spectral accelerations). The final microzonation map is a relative map defined in terms of three zones independent of the absolute values of the ground shaking intensity.

In the second phase of the procedure, vulnerability analyses for building and pipeline inventories are evaluated. Site-specific spectral accelerations Ss and Sl are used to assess the vulnerability of the building stock. The analytical estimation of structural damage is formulated based on Hazus (2003), where the vulnerability relationships (also called fragility curves) are developed in terms of spectral displacements, which in turn are calculated from the estimated mean inelastic drift capacities of buildings for various damage states. The mean drift demand of a typical building is estimated through nonlinear static procedures (NSPs), which are based on performance-based seismic evaluation (ATC 40, 1996; FEMA 273, 1997; FEMA 356, 2000). NSPs are based on the capacity (pushover) curve of the given building and the estimation of the inelastic spectral displacement demand consistent with the capacity curve. Vulnerability of pipeline inventory with respect to wave propagation is evaluated using site-specific PGV values as input parameters. Empirical correlations which relate damage rate to PGV are employed to predict damage in pipelines in terms of damage rate and number of pipe damages at each cell in the grid system. Results from vulnerability analyses are used to prepare damage distribution maps for the buildings and pipelines.

The methodology is developed into a software package (KoeriLossV2, 2007) to provide a practical tool for assessing the seismic vulnerability of an urban area. A pilot study is carried out using KoeriLossV2 to perform a damage scenario for Zeytinburnu district in Istanbul, Turkey, where building and gas pipeline inventories are available to some detail. The area of investigation is approximately 20 km2 occupied mostly with low- to mid-rise residential buildings. The available inventory indicated that natural gas pipeline system in the district consists of steel pipes with diameters changing between 102 mm and 762 mm.

A grid system with cells of 250 m × 250 m is defined for the study area. Probabilistic seismic hazard analysis is carried out to evaluate PGAs and spectral accelerations at T = 0.2 s and T = 1 s for each cell on the engineering bedrock outcrop (Erdik et al., 2005). A regional time dependent Poisson model for the return period of 475 years that corresponds approximately to 10 % probability of exceedance in 50 years is considered in the analysis (Erdik et al., 2004). 24 real acceleration time histories compatible with the earthquake hazard in terms of probable magnitude (M = 6.5 - 7.5) epicenter distance (R = 20 - 40 km) and fault mechanism (strike slip) recorded on stiff site conditions with average shear wave velocities (Vs30) larger than 420 m/s are selected as the probable input acceleration time histories from the PEER strong motion data bank (Ansal and Tönük, 2007). Selected acceleration time histories are scaled with respect to PGAs estimated from the seismic hazard analysis for each cell before being used as outcrop motions in site response analyses.

The local site conditions are characterised based on an extensive site investigation study conducted in the area with at least one soil boring conducted at each cell location along with in-hole PS-Logging, surface seismic wave measurements and laboratory index tests (OYO, 2007). All available information on geological and geotechnical conditions is evaluated to determine one representative soil profile with shear wave velocities extending down to engineering bedrock (Vs = 750 m/s) for each cell.

Site response analyses are performed for 24 acceleration time histories for the representative soil profiles in each cell using Shake91. The averages of 24 values of PGA, PGV and acceleration response spectra from 24 Shake91 runs for each cell are determined to define the variation of ground shaking parameters due to the probabilistic seismic hazard scenario. The short and long period spectral accelerations (Ss and Sl) are obtained through optimisation for the best-fit NEHRP envelope spectrum.

The variations of PGV and Ss can also be calculated empirically based on Vs30. Ss and Sl can be obtained by using the procedure proposed by Borcherdt (1994) as suggested in NEHRP. PGV can be determined using Hazus formulation that relates Sl to PGV.

Comparisons of variations show that the use of equivalent shear wave velocity to estimate the effects of site conditions may yield very different Ss and PGV amplitudes that may not always be on the safe side. The observed differences indicate the importance of methodology employed in estimating the effects of site conditions.

A detailed building inventory from street surveys for approximately 16 000 buildings is considered in the evaluation of seismic vulnerability of Zeytinburnu (Aydinoglu and Polat, 2004). Building inventory is divided into groups based on the construction type, number of stories and construction year of buildings (Erdik et al., 2002, 2003). All buildings are classified according to a 'Bijk' matrix where 'I' shows the construction type as:

(1) reinforced concrete frame building
(2) masonry building
(3) reinforced concrete shear wall buildings
(4) precast building.

The number of stories ('j' dimension of the matrix) is defined as:

(1) low rise (one to four stories, including basement)
(2) mid rise (five to eight stories, including basement)
(3) high-rise (eight or more stories, including basement).

The construction date ('k' dimension of the matrix) is defined as:

(1) construction year: pre-1980 and
(2) construction year: post-1980.

The available inventory in Zeytinburnu indicates that almost all of the buildings are mid-rise reinforced concrete frame buildings.

Region-specific vulnerability relationships (Aydinoglu and Polat, 2004) that relate spectral displacements to building damage for each building type are used to estimate damage in Zeytinburnu.

The distribution of number of buildings at each damage state for all building types in the area are computed and displayed in maps showing number of buildings at each cell for a given type of building and damage state. Numbers of buildings at each damage state estimated using NEHRP amplification factors are estimated to provide direct comparisons.

The natural gas pipeline inventory of Zeytinburnu area is compiled based on information provided by Istanbul Gas Distribution Industry and Trade Co. Inc. (IGDAS). The inventory is consisted of length, diameter and material properties of the main steel pipeline system. Empirical correlations that relates PGV to pipeline damage is used to estimate repair rate and number of repairs in the pipeline system due to wave propagation. Numbers of expected repairs at each cell are calculated as the product of repair rate and total pipeline length.

Application of the methodology to Zeytinburnu demonstrates that there are significant variations in the ground motion parameters within the investigated region which cannot be detected when the site conditions and their effects are evaluated using NEHRP site classification and related amplification coefficients. Therefore it appears essential to incorporate site-specific response analysis in order to have more accurate information on ground shaking characteristics for microzonation and for the estimation of seismic damage in buildings and lifeline systems.

References:

American Lifelines Alliance (ALA, 2001) Seismic Fragility Formulations for Water Systems, Part 1 Guideline, http://www.americanlifelinesalliance.org.

Ansal, A. and Tönük, G. (2007) Source and Site Effects for Microzonation. Theme Lecture, Fourth International Conference on Earthquake Geotechnical Engineering, Earthquake Geotechnical Engineering, Editor: K.Pitilakis Chapter 4, pp.73-92. Springer.

Ansal, A. Akinci, G. Cultrera, M. Erdik, V. Pessina G. Tonuk, G. Ameri (2009) 'Loss Estimation In Istanbul Based on Deterministic Earthquake Scenarios of the Marmara Sea Region (Turkey)', Soil Dynamics and Earthquake Engineering, 29(4):699-709

Ansal, A., A. Kurtulus, and G. Tönük (2010) 'Seismic Microzonation and Earthquake Damage Scenarios for Urban Areas', Soil Dyn. Earthquake Eng., Vol. 30, pp. 1319–1328.

Ansal, A., Durukal, E. and Tönük, G. (2006a) 'Selection and scaling of real acceleration time histories for site response analyses', Proceedings of ISSMGE ETC12 Athens Workshop, Athens Greece.

Ansal, A., Erdik, M., Studer, J., Springman, S., Laue, J., Buchheister, J., Giardini, D., Faeh, D. and Koksal, D. (2004b) 'Seismic Microzonation For Earthquake Risk Mitigation In Turkey'. Proceedings of the 13th World Conference of Earthquake Engineering, Vancouver, CD paper No.1428.

Ansal, A., Laue, J., Buchheister, J., Erdik, M., Springman, S M., Studer, J. and Koksal D. (2004a) 'Site Characterisation and Site Amplification for a Seismic Microzonation Study in Turkey'. Proceedings of 11th International Conference on Soil Dynamics and Earthquake Engineering and 3rd Earthquake Geotechnical Engineering, San Francisco, US

Ansal, A., Tönük, G., Demircioglu, M., Bayrakli, Y., Sesetyan, K. and Erdik, M. (2006b) 'Ground motion parameters for vulnerability assessment', Proceedings of the First European Conference on Earthquake Engineering and Seismology, Geneva, Switzerland, Paper Number: 1790.

ATC and SSC - (1996), 'ATC-40 Seismic Evaluation and Retrofit of Concrete Buildings'. Report SSC 96-01, California, EUA.

Aydinoglu, N. and Polat, Z. (2004) 'First level evaluation and assessment of building earthquake performance', Report for the Istanbul Master Plan Zeytinburnu Pilot Project, Metropolitan Municipality of Istanbul, Planning and Construction Directorate (in Turkish).

Bendat, J. S. and A.G. Piersol (1985) 'Random Data: Analysis and Measurement Procedures'. John Wiley& Sons, New York.

Beresnev, I. A., E.H. Field, P. A. Johnson, and den KE-A. Abeele (1998) 'Magnitude of Nonlinear Sediment Response in Los Angeles Basin during the 1994 Northridge, California, Earthquake', Bull. Seism. Soc. Am., Vol. 88, pp. 1079-1084.

Borcherdt, R.D. (1994) 'Estimates of site dependent response spectra for design (methodology and justification)' Earthquake Spectra 10(4), 617-654.

Darendelli, M. B. (2001) 'A new family of normalised modulus reduction and material damping curves', PhD. Thesis, Civil Engineering Dept., University of Texas at Austin.

Eidinger J., Avila E. (1999) 'Guidelines for the Seismic Upgrade of Water Transmission Facilities'. Monograph No. 15, TCLEE/ASCE.

Elgamal, A. W., M. Zeghal, H .T. Tang, and J.C. Stepp, (1995) 'Lotung Downhole Array, I: Evaluation of Site Dynamic Properties', Jour. Geotech. Engrg., ASCE, Vol. 121, No. 4, pp. 350–362.

EPRI (1993) 'Guidelines for Determining Design Basis Ground Motions', Palo Alto, CA, Electric Power Research Institute, Vol. 1, EPRI TR-102293.

Erdik, M. and Fahjan, Y. (2005) 'System analysis and risk' in Assessing and Managing Earthquake Risk Geo-scientific and Engineering Knowledge for Earthquake Risk Mitigation: Developments, Tools, Techniques, Part 3, Book Series: Geotechnical, Geological, and Earthquake Engineering, V2, eds.

Oliveira C. S., Roca A, Goula X., Erdik, M., Demircioglu, M., Sesetyan, K. and Durukal, E. (2005) 'Assessment of earthquake hazard for Bakirköy, Gemlik, Bandirma, Tekirdag and Körfez', WB MEER Project -A3 Component, Microzonation and Hazard Vulnerability Studies For Disaster Mitigation in Pilot Municipalities, Bogazici University, Kandilli Observatory and Earthquake Engineering Research Institute.

Erdik, M., Demircioglu, M., Sesetyan, K., Durukal, E. and Siyahi, B. (2004) 'Earthquake hazard in Marmara region', Soil Dynamics and Earthquake Engineering 24, 605-631.

Erdik, M., N. Aydinoglu, Y. Fahjan, K. Sesetyan, M. Demircioglu, B. Siyahi, E. Durukal, C. Ozbey, Y. Biro, H. Akman and O. Yuzugullu (2003) 'Earthquake Risk Assessment for Istanbul Metropolitan Area', Earthquake Engineering& Engineering Vibration, 2(1):1-25

Erdik, M., Y. Fahjan, O. Özel, H. Alcik, A. Mert, and M. Gül (2003) 'Istanbul Earthquake Rapid Response and the Early Warning System', Bull. Earthquake Eng., Vol. 1, pp. 157-163.

Erdik, M., Aydinoglu, N., Barka A., Yüzügüllü, Ö., Siyahi, B., Durukal E., Fahjan, Y. Akman, H., Birgören, G. Biro, Y. Demircioglu, M., Özbey, C.and Sesetyan K. (2002) BU-ARC, Earthquake Risk Assessment for Istanbul Metropolitan Area, Project Report, Bogazici University Publication.

FEMA 356 (2000). 'Prestandard and Commentary for the Seismic Rehabilitation of Buildings', Federal Emergency Management Agency, Washington, D.C. US.

FEMA273 (1997). NEHRP Commentary on the Guidelines for the Seismic Rehabilitation of Buildings. Federal Emergency Management Agency, SW Washington, D.C. USA.

Fukushima Y., K. Irikura, T. Uateke, and H. Matsumoto (2000) 'Characteristics of Observed Peak Amplitude for Strong Ground Motion from the 1995 Kobe Earthquake', Bull. Seism. Soc. Am., Vol. 90, pp. 545-565.

HAZUS (2003) 'Earthquake Loss Estimation Methodology', Technical Manual, Federal Emergency Management Agency and National Institute of Buildings Sciences, Washington, D.C. US.

Idriss, I. M., and JI. Sun (1992) 'Shake91', A computer program for conducting equivalent linear seismic response analysis of horizontally layered soil deposits, modified based on the original SHAKE program by Schnabel, Lysmer and Seed, 1972.

KoeriLossV2 (2007) 'Earthquake Disaster Scenario Prediction and Loss Modelling for Urban Areas', Spence, R. ed. EU FP6 Project on Risk Mitigation for Earthquakes and Landslides Report.

Kurtulus, A., E. Safak, A. Strollo, S. Parolai, and A. Ansal (2008) 'Sensor Azimuth Determination at a Seismic Vertical Array in Istanbul, Turkey', European Seismological Commission ESC 2008, 31st General Assembly, Crete, Greece.

Mehta, K., R. Snieder, and V. Graizer (2007) 'Downhole Receiver Function: a Case Study', Bull. Seism. Soc. Am., Vol. 97, No. 5, pp. 1396-1403.

Menq, F. Y., K. H. Stokoe, E. Kavazanjian (2003) 'Linear dynamic properties of sandy and gravelly soils from largescale resonant tests', International Symposium IS Lyon 03, Deformation Characteristics of Geomaterials, (Lyon, France) 22-24 September 2003.

NEHRP (2003) 'Recommended Provisions For New Buildings and other Structures', FEMA-450, prepared by the Building Seismic Safety Council for the Federal Emergency Management Agency, Washington, DC.

O’ Rourke, M. and Ayala, G. (1993) 'Pipeline damage due to wave propagation', Journal of Geotechnical Engineering 119 (9), 1490-1498.

O’ Rourke, M. and Deyoe, E. (2004) 'Seismic damage to segmented buried pipe', Earthquake Spectra 20, 1167–1183.

Safak, E. (1995) 'Discrete-time Analysis of Seismic Site Amplification', Jour. Engrg. Mechanics, ASCE, Vol.121 No.7 pp.801-809.

Snieder, R., and E. Safak (2006) 'Extracting the Building Response using Seismic Interferometry; Theory and Application to the Millikan Library in Pasadena, California', Bull. Seism. Soc. Am., Vol. 96, No. 2, pp. 586-598.

Trifunac, M. D, and M. I. Todorovska (1998) 'Nonlinear Soil Response as a Natural Passive Isolation Mechanism- the 1994 Northridge, California, Earthquake', Soil. Dyn. Earthquake Eng. (17): 41-51.

OYO Inc., Japan (2007) 'Production of Microzonation Report and Maps on European Side (South)', Final Report to Istanbul Metropolitan Municipality.

Seed, H. B., Wong, R. T., Idriss, I. M., and Tokimatsu, K. (1986) 'Moduli and damping factors for dynamic analysis of cohesionless soils', Journal of Geotechnical Engineering 112(11), 1016-1032.

Vucetic, M. andDobry, R. (1991) 'Effect of soil plasticity on cyclic response', Journal of Geotechnical Engineering, ASCE 117(1), 89-107.

Potential impact

During the 20th century, earthquakes claimed over 130 000 lives in the countries of today's EU alone (and over 400 000 in the wider European-Mediterranean area), as well as vast but uncalculated damage to property and economic activity. Over the last 20 years, improved understanding and the experience of earthquake loss has driven the progressive development of new and better codes and regulations for building in earthquake areas; and buildings and facilities constructed according to current codes are less likely to be heavily damaged by expected earthquakes. But throughout the European area, most of the built environment was created before these codes were formulated and enforced, and without the benefit of present understanding of the effects of earthquakes. Many of these buildings and facilities (which include schools, hospitals, and highway structures used continuously by the public) are unsafe by current standards and are liable to be seriously damaged or to collapse during probable earthquakes.

The major impact of the project today and in future is its unique potential to generate invaluable earthquake data which is today regarded as one of the most-needed type of information to develop, modify and verify the current analysis and design procedures in earthquake engineering. Scientific studies that would be supported by the information collected at the URBANQUAKE will enable more accurate identification of the earthquake ground motion characteristics thus leading to more realistic estimation of the vulnerability of the building stock. Accurate assessment of structural vulnerabilities in earthquake prone cities could provide effectiveness in dealing with urban retrofit and rehabilitation projects and in mitigating the earthquake damages and causalities. In the case of retrofit and rehabilitation projects, the findings could lead to more efficient and economical solutions.

In terms of state-of-the-art research in the field of local site effects, the URBANQUAKE is unique in Europe with respect to instrumentation and accumulated data. The existing high seismic activity in the region creates a unique opportunity for scientific observations. These findings and the analyses of these data by the European engineers and scientists could lead to significant progress in the field of earthquake engineering, site response and vulnerability evaluations in the long run. The scientific achievements will enhance significantly the competitiveness of EU in the field of earthquake engineering since very few similar test sites at similar scales with similar active seismicity exist in the world.

The impact potential of project and the achievements attained in the first project period has already been acknowledged by CORDIS; the project has been selected for special promotion on Technology Marketplace and is announced as an exploitable technology on official website.

Dissemination activities:

During the course of project, special attention has been given to the dissemination of project results and the exploitation of foregrounds developed within the project. URBANQUAKE has been presented in various national and international conferences and workshops as invited lectures and as articles in peer-reviewed publications.

As a result of these efforts for dissemination URBANQUAKE has gained the attention of the research community throughout the world. Collaborative FP7 projects as well as collaborations with other tests sites in US and Europe such as NEES and the Euroseistest Site have been initiated to share and exploit foregrounds of the project.

Use and dissemination of foreground

Significant efforts have been spent so far to present URBANQUAKE in various international conferences and workshops as invited lectures and as articles in peer-reviewed publications which have brought international recognition to Istanbul Strong Motion Network. Collaborations have been formed with various US and European based organisations and consortiums as described above. In the coming years, these efforts will be continued; as more data is accumulated within URBANQUAKE and as the foreseen capacity developments are achieved, the opportunities for collaborative projects on international scale and for attracting world-class researchers to the Institution will be inevitable. A unique set of earthquake data that is generated at the urban earthquake test site will be available to the community. It is strongly believed that the end result of all the scientific activities initiated with the grant rewarded by EU for URBANQUAKE will lead to significant progress in the field of earthquake engineering.