A model study was developed to analyse the total risk of large dams for a specified basin in Turkey. This paper describes the main principles of the model and introduces the results of a case study including the Euphrates basin – the largest of 26 basins in the country


Turkey covers an area of 77.95M ha and has twenty-six river basins. The Euphrates basin is the largest of basins in the country and its water yield resources is about 31.6km3 per year. The basin includes the main part of the Southeast Anatolian project, a multi-sector and integrated regional development effort designed in the context of sustainable development. The project encompasses such sectors as irrigation, hydraulic energy, agriculture, rural and urban infrastructure, forestry, education and health. The water-resources development component of the project envisages the construction of 22 dams and 19 hydro power plants and irrigation of 1.7M ha of land. The total cost of the project is estimated as US$32B. The total installed capacity of power plants is 7476MW and the annual energy production reaches 27BkWh.

The Euphrates river, formed from the Karasu and Murat tributary rivers, is the main river of the basin and has a length of 2800km. The river crosses Iraq to join the Tigris, where it flows into the Persian Gulf. Its average discharge upstream and at the Syrian border is 650m3/sec and 950m3/sec, respectively. In the basin, thirty-two large dams have been designed to exploit the energy and irrigational potential of the basin (Figure 1). Twenty-three of them have been completed, seven are under construction with the last two dams in the design stage. A series of dams on the main river, namely Karkamis, Birecik, Ataturk, Karakaya, Keban and Ozluce, have been completed. The large ones, such as Alparslan I, Kigi, Uzuncayir and Yazici, are under construction (DSI, 2002). It is estimated that they will be entirely finished in 2007. Alparslan II and Konaktepe are currently at the design stage, with construction expected to start this year.

It is commonly thought in Turkey that embankment dams, which are well compacted according to the specification, are suitable types for regions with high seismic activity. In general, strong ground shaking can result in the instability of the embankment and loss of strength at the foundations (Jansen, 1988 and Castro et al., 1985). Active faults, which are very close to the foundation of dams, have the potential to cause damaging displacement of the structure. There are some examples of dams in Turkey which were damaged during past earthquakes (Tosun, 2002). This paper deals with an evaluation of seismic hazard and local site effects and evaluates 32 large dams, which have a structural height greater than 30m, in the Euphrates basin (Table 1).

Geological setting and structural geology

The Euphrates basin is situated in a region with very complex geology and also a very active seismicity. The geological setting and the data used in the analysis are based on previous works. Brinkman (1976) carried out detailed geological investigations in this region. Ertunç (1999) conducted a detailed engineering geological study for the basin. For this study, the geology of the basin was simplified and represented by seven separate units ranging in age from Precambrian to Quaternary. The oldest units, which are commonly made up of various metamorphic schists, are found at the central part of basin. There is no problem with water leakage and foundation stability in the dam sites, but small landslides can be observed along the main bank of the river. The younger units are mainly composed of sedimentary rocks including marl, mudstone and limestone, and shale alternation.

The North Anatolian Fault and East Anatolian Fault Zone are the common structural features in the basin. The North Anatolian Fault zone is one of the best-known strike-slip faults in the world, because of its significant seismic activity and well-developed surface features. It is approximately 1500km-long and extends from eastern Turkey in the east to Greece in the west. Its width ranges from a single zone of a few hundred meters to multiple shear zones of 40km. The North Anatolian Fault Zone joins with the East Anatolian Fault Zone at Karliova in the basin and forms a typical triple junction. This zone produces very large earthquakes, which have resulted in the death of a thousand people and very severe structural damages (Bozkurt, 2001).

The East-Anatolian Fault Zone, which has a 550km length, is a northeast trending, sinistral strike-slip fault zone. It is comprised of a series of faults arranged parallel to the general trend. It is a transform fault forming parts of boundaries between the Anatolian and Arabian Plates. Within the basin it extends from triple junction in the northeast to the Kahramanmaras area in the southwest, where it meets and forms triple junctions with the Dead Sea Fault Zone. It is known that the left-lateral slip along the fault zone contributes to the westward extrusion of Anatolian. Several pull-apart basins complicate the structure of the fault zone; conjugate fractures, folding and a considerable thrust component (Westaway and Arger, 1996; Hempton, 1987). This system has produced large earthquakes in the past, even if they were not high intensity. There are a great number of dams located on or close to this shear zone.

Methods of analysis

For the seismic hazard analysis of a particular site, all possible seismic sources are identified and their potential is evaluated in detail, as based on the guidelines given by Fraser and Howard (2002) and the unified seismic hazard modelling for the Mediterranean region introduced by Jiminez et. al (2001). The study of seismic activity includes deterministic and probabilistic seismic hazard analyses.

The deterministic seismic hazard analysis considers a seismic scenario that includes a four-step process. It is a very simple procedure and gives rational solutions for large dams because it provides a straightforward framework for evaluation of the worst ground motions. Due to the unavailability of strong motion records, various attenuation relationships were adopted to calculate the peak ground acceleration (PGA) acting on dam sites (Boore et al, 1997). For this study two separate predictive relationships for horizontal peak ground acceleration were considered (Ambraseys, 1995; Campbell, 1981). Ambraseys (1995) introduced an attenuation relationship based on data collected from shallow earthquakes in Europe.

The probabilistic seismic hazard analysis is widely used and considers uncertainties in size, location and recurrence rate of earthquakes. Kramer (1996) states that the probabilistic seismic hazard analysis provides a framework in which uncertainties can be identified and combined in a rational manner to provide a more complete picture of the seismic hazard. The seismic hazard parameters were statistically estimated in the computer program DAMHA, which was developed at the Earthquake Research Center, Eskisehir Osmangazi University.

The seismic hazard parameters were statistically estimated by using the method developed by Gumbel (1958) and the Gutenberg-Richter (1942) methods. The Operating Basis Earthquake (OBE) was defined by means of the probabilistic methods mentioned above. It is known as the earthquake that produces the ground motions at the site that can reasonably be expected to occur within the service life of the project. Maximum Credible Earthquake (MCE), which is the largest earthquake magnitude that could occur along a recognised fault or within a particular seismotectonic province or source area under the current tectonic framework, was obtained for each zone, and the most critical framework for the dam site was selected as Controlling Maximum Credible Earthquake (CMCE). The Maximum Design Earthquake (MDE) was then defined. FEMA (2005) states that MDE is normally characterised by a level of motion equal to that expected at the dam site from the occurrence of deterministically evaluated MCE. Most of the large dams in Turkey were analysed using these definitions (Tosun and Seyrek, 2005; Tosun and Savas, 2005).

For all analyses throughout this study, the peak ground acceleration was deterministically obtained by considering the Maximum Credible Earthquake (MCE). All procedures mentioned above can be executed by the DAMHA program that is working on the basis of geographic information system (GIS). The authors believe that seismic criteria and analysis parameters for dams should be selected more conservatively than for conventional structures since the failure consequences are more disastrous. It is clear that the main requirement of an earthquake-resistant design is to protect public safety and property.

Analyses and discussions

For the seismic hazard analyses of the dam site in the basin, a detailed study was performed to identify all possible seismic sources, as based on the seismic zonation map of Turkey. The National Disaster Organization and other Institutes prepared the map for general use. This was modified by the authors for use on dam projects at the Earthquake Research Center in Eskisehir Osmangazi University. Local geological features and seismic history were used to quantify the rate of seismic activity in the basin. As a result of detailed evaluation, the total area covering all basins was separated into eleven seismic zones.

The number of earthquakes having a magnitude on the basis of surface wave (Ms), which is greater than 4.0, is 264. The earthquakes which occurred in the basin is given in Figure 2 with structural features. The numbers of earthquakes with Ms that are greater than 5.0 and 6.0 are 55 and 7, respectively. There are only two earthquakes with a magnitude greater than 7.0 in the basin. The first one is the Erzincan earthquake with Ms of 7.9 in 1939. The second one is the Karliova earthquake with Ms of 7.0 in 1949.

The seismic hazard analyses were performed for 32 dams in the basin. The results of this analysis are given in Table 2. The results indicate that peak ground acceleration (PGA) changes within a wide range (0.011g and 0.564g). Thirteen dam sites have low hazard rating and are identified as hazard class of I. Fourteen dam sites are classified as II with moderate hazard rating. According to icold (1989) classification, if the PGA value is greater than 0.25g and the energy source is closer than 10km from the dam site, it is classified as hazard class IV with hazard rating of extreme. Erzincan dam, which is located at the northern part of basin, has a hazard class of IV. It is very close to the epicenter of the catastrophic 1992 earthquakes and can be subjected to a PGA of 0.564g with a Maximum Credible Earthquake (MCE) of 7.9. Surgu dam, which was damaged by the Dogansehir earthquake with Ms of 5.8 in 1986, also has a hazard class of IV with a PGA of 0.256.

For Alparslan I, Kigi and Uzuncayir dams, the value of PGA is greater than 0.25g, but they are no closer than 10km from the energy source. Therefore, they are identified in the hazard class of III. Alparslan I dam’s risk is similar to Erzincan dam, when the adverse geotechnical properties of foundation soil are considered. It is well known that the foundation material of this dam is composed of claystone and siltstone. The large dams, which are under the influence of the near source zone, are located very close to the North Anatolian Fault – the most active zone in Turkey.

There are important dam structures in the basin such as Karkamis, Birecik, Ataturk, Karakaya, Keban and Ozluce dams. These dams, which are located on the main river of the basin, could cause very serious damage to downstream life and property if they failed. For Birecik and Karkamis dams, the PGA values were estimated very low, meaning they are seismically safe sites. Atatürk dam, which is the largest dam in Turkey with a storage capacity of 48700hm3, is also located on a site with very low seismicity. On the other hand, Karakaya dam, which is the unique concrete dam of the basin, is subjected to relatively high earthquake loads of a value of 0.132g with a MCE value of 7.0.

Polat and Cat dams are located in the western part of basin. Their location is frequently jointed, fractured, and faulted. The dam sites are identified as hazard class II with a moderate value of PGA, even if the MCE value is low. The PGA value is 0.170g for Polat dam, while it is 0.211g for Cat dam. The Surgu dam discussed above is also included within this zone. However, the authors point out the fact that these dams are under the influence of local near-source zone and have high risk rating for earthquake conditions. The damage caused by Dogansehir earthquake with Ms of 5.8 on Surgu dam confirms the authors’ idea (Tosun, 2002). As a result of seismic hazard analyses performed for the large dams within the basin, a map showing the equivalent PGA values was developed as given in Figure 3.

MAPINFO program – which is a geographic information system (GIS) defined as an information system designed to work with spatially referenced data or geographic coordinates – was used to obtain this map. The PGA values, which were estimated by deterministic method as based on MCE values, were plotted on the map geographically, and these values were then interrelated to estimate equivalent curves.

This map shows that the most critical area in the basin is very close to the North Anatolian Fault zone. Because the greater PGA values are seen on this part, there are also some isolated areas showing different seismological behaviour. One of them – the most critical one – is the area that includes the large dams of Surgu, Polat and Cat, along with numerous small dams. The other one is located on the eastern part of the basin, which includes the Alparslan I and Alparslan II dams. This area is also important for the dams to be subjected to high PGA value because of the potential impact of local near-source zone.


For this study, thirty-two large dams, which are located on different seismic zones of the Euphrates basin, were analysed to estimate their seismic hazard and risk classes. The authors state that the model considers the actual earthquakes which occurred within the basin. At the end of this study, the following results were concluded:

•The seismic performance of embankment dams closely depends on the nature and strength of the fill material and foundation. Stability problems for embankment dams can occur due to earthquake shaking. An earthquake can cause great danger to dams. The most critical zone for the basin is North Anatolian Fault. Erzincan, Kigi, Tercan, Uzuncayir and Alparslan I dams are earth structures which have critical PGA values. The MCE value for the first three dams is greater than 7.2. Erzincan and Alparslan I dams are also under the impact of near source zone.

•Most of the dams, which are located near active seismic zone, have damaged or failed during earthquakes. There is a shear zone, which was frequently jointed, fractured and faulted at the west part of the basin where the Surgu, Cat and Polat dams are located. The PGA values range from 0.170g to 0.256g (moderate to extreme hazard classes). A damage or failure risk is high for these dams, when subjected to a loading of an earthquake with moderate magnitude (M=5.9-6.1). The damage on Surgu dam under the 1986 Dogansehir Earthquake with Ms of 5.8 is a good example for this case. The authors state that all dams within the shear zone of the basin are under the influence of local near-source zone and their seismic positions must be re-evaluated in detail.

•There are three dams, which were constructed in composite type as a combination of embankment and concrete body. They are Karkamis, Birecik, and Keban dams. The first two will be subjected to very low PGA for the scenario earthquake with MCE value. The PGA value for Keban dam is relatively high (0.115g) and the dam is identified as risk class III (high). It is well known that this dam has a leakage problem in the abutments and foundation materials. This condition increases the risk for downstream life and property. Therefore, Keban dam should be evaluated with highest priority as a part of the National Dam Safety Program and its static and dynamic stability should be performed using a 3D numerical method.

•The main requirement of earthquake resistance design for dam structures is to protect public safety and property. It is clear that seismic criteria and analysis parameters for a dam structure should be selected more conservatively than those for conventional structures. As a result of this study, fifteen large dams, as discussed above, are identified as the dams with high-risk class. Therefore, these structures must be analysed while selecting the appropriate seismic parameters and redesigned, if necessary.

Table 1 Table 2 Ataturk dam Ataturk dam Erzincan dam Erzincan dam Karakaya dam Karakaya dam Keban dam Keban dam Ozluce dam Ozluce dam Author Info:

The authors are Hasan Tosun, Professor, President of DSA and Eskisehir Osmangazi University Turkey, email: htosun@barajguvenligi.org; and Murat Turkoz, Hasan Savas and Evren Seyrek, Research Assistant, Eskisehir Osmangazi University Eskisehir, Turkey


Table 1
Table 2