Upon its completion in 2010, Deriner dam on the Coruh River will be the highest arch dam in Turkey. Dr Martin Wieland, Martin Aemmer and Roland Ruoss discuss how, during the dam design, the safety evaluation earthquake with a peak ground acceleration of 0.35g and the maximum flood of 10,110m3/sec had to be considered

The Deriner Dam and hydroelectric project is located on the lower course of the Coruh River in northeastern Turkey. The 249m high double-curvature arch dam has a crest length of 721m and a concrete volume of 3.5Mm^{3}. The maximum thickness of the dam at the base is 60m and the crest width of the crown cantilever is 1m. The underground powerhouse with a width of 20m, a length of 126m and a height of 45m will have four vertical Francis units with a total capacity of 670MW and is located on the right bank of the river close to the dam. The layout of the dam and main hydraulic structures is shown in Figure 1.

The catchment area of the Coruh River is 18,390km2, and the mean inflow is 154m^{3}/sec. The future reservoir has a volume of 1970Mm^{3}, a regulating storage volume of 960Mm^{3}, and the reservoir area at full supply level is 26.4km^{2}. The diversion tunnel with a horseshoe cross-section has a diameter of 11.6m and a length of 912m. With a discharge capacity of 1804m^{3}/sec a 100-year flood can be diverted during construction of the dam. This is important in view of the relatively long construction period.

The foundation rock consists mainly of granodiorite which is intersected by diabas dykes. This type of rock is generally very suitable for an arch dam. The rock was tectonically stressed. Above the sound rock there is a layer of decompressed rock, which is heavily jointed. Geophysical measurements show a clear difference in the wave propagation velocities of the fresh and decompressed rocks. The arch dam will be supported by the sound rock. This requires considerable foundation excavation works (Figure 2).

The Deriner project has been under construction since 1998. The excavation of the dam foundation was completed at the end of 2005 and concreting of the arch dam started immediately afterwards. However, concreting was suspended from September 2006 to September 2007. In January 2008 the dam reached a maximum height of 66m (Figure 3).

A consortium consisting of Turkish and Russian contractors as well as a Swiss supplier for the electro-mechanical equipment is responsible for the construction of the whole project. The project being implemented at the moment is based on the final design prepared by Poyry Energy from Zurich (formerly electrowatt Engineering) and Dolsar Engineering from Ankara in the early 1990s. The Poyry Energy and Dolsar consortium is the consultant to the dam owner DSI (General Directorate of State Hydraulic Works) and is in charge of quality assurance.

The main hazards from the natural environment which could affect the dam safety, and which needed special attention, are: (i) geological hazard, (ii) flood hazard, and (iii) earthquake hazard. These hazards are discussed in the subsequent sections.

**Dam foundation and rock excavation**

The volume of foundation excavation was 8.7Mm^{3} and was the result of the deep excavations needed to remove the decompressed rock. The following geological and geotechnical problems were encountered during the excavation works:

• (i) Unstable highly fractured rock masses, which had to be stabilised by rock anchors. At the left and right abutments 764 and 1232 restressable post-tensioned rock anchors were installed, respectively. The length of the anchors varied from 23 to 51m (including 8m bond length) and the capacity was 2064kN. The long-term safety will be secured by up-to-date monitoring. The anchors installed at the dam and diversion inlet areas are constructed as permanently monitorable by lift-off tests or permanently installed load cells. Overall, around 5% of the anchors will be monitored.

• (ii) Shear zones in dam foundations required special treatment. of the sheared zones.

• (iii) Slope deformations resulting from the rock excavation are monitored by geotechnical instruments (inclinometers) and geodetic survey. The first inclinometers were installed in May 2002 and the geodetic monitoring started in October 2004. By 2007 maximum slope displacements of 6cm and 4cm were observed at the left bank (elevation 542m asl, 145m above dam crest) and at the right bank (elevation 563m asl, 166m above dam crest), respectively.

**Spillways and flood safety**

For the flood discharge two types of spillways are provided: (i) two surface spillways located at the right and left abutments with a capacity of 2 x 1125m^{3}/sec, which can be regulated by one flap gate at each spillway intake, and (ii) an orifice spillway comprising eight gated orifices integrated in the dam body.

In the surface spillways the flood water is discharged by two inclined circular tunnels with a diameter of 8m, and lengths of 420 and 446m, respectively (Figure 1). At the end of the tunnels a flip bucket is provided on each river bank. The two surface spillways are arranged symmetrically and are operated simultaneously so the jets will meet above the plunge pool. The discharge capacity of the surface spillway is adequate for releasing the 100-year flood.

The eight orifice spillways have a discharge capacity of 7000m^{3}/sec. For peak discharges (e.g. 10,000-year flood: design discharge 9250m^{3}/sec) the operation of both the surface and orifice spillways is needed. The two orifice spillway outlets in the centre of the dam are used as bottom outlets.

The eight orifices are arranged symmetrically in such a way that the jets meet above the plunge pool thus improving the energy dissipation. It is therefore of great importance to always operate an even number of orifices.

For the design of the tunnels of the surface spillway, hydraulic model tests with a scale of 1:20 were carried out at DSI’s Hydraulic Laboratory in Ankara (Figures 4 and 5). In view of the high velocity flows, the basic question was whether cavitation effects were possible in the bends of the tunnels under full and partial filling of the tunnel. The model tests have shown that with a proper aeration system negative dynamic pressures can be avoided in these critical parts of the inclined tunnels.

**Seismic design aspects for worst case ground shaking**

*Seismic behaviour of large arch dams*

Linear-elastic dynamic analyses show that tensile stresses exceeding the dynamic tensile strength of mass concrete occur in the central upper portion of high arch dams during strong ground shaking, such as that caused by the maximum credible earthquake (MCE) or the safety evaluation earthquake (SEE). As the tensile strength of the grouted vertical contraction joints and the horizontal lift joints is inferior to that of the parent mass concrete, the contraction joints may open and horizontal cracks are likely to develop along the lift joints. Thus, the dynamic deformations of the dam will be mainly due to opening of contraction and lift joints, and the sliding movement of concrete blocks along these joints and cracks. These deformations limit the tensile stresses in the dam.

Joint opening and crack formation will cause a redistribution of the stresses in the dam and lead to an increase in compressive stresses. The higher dynamic compressive stresses of the cracked dam are unlikely to cause any damage, as the dynamic compressive strength of mass concrete is more than ten times larger than the tensile strength and the factor of safety for compressive stresses under usual loading conditions is usually at least 3. After an earthquake, the same (usual) loading conditions (dead load, water and silt loads, temperature effects) act on the dam as before the earthquake. Despite the fact that the dam is cracked after an earthquake and water pressure may develop in joints and cracks, the additional compressive stresses can be accepted. There have been no observations of earthquake damage in concrete dams caused by excessive compressive stresses.

It should be noted that field observations, and particularly shaking table tests, show only few seismic cracks in concrete dams [3]. This means that once a crack has formed, it will protect the rest of the dam from overstressing (in tension) and formation of additional cracks. Thus, for the analysis of the post-cracking behaviour of a dam, it is sufficient to look at the dynamic stability of concrete blocks separated by cracked joints, provided that the foundation stability is ensured.

*Methods of earthquake analysis of arch dams and dynamic stability of concrete blocks*

In practice, the seismic safety of a concrete dam is usually evaluated based on a linear-elastic dynamic analysis of the dam taking into account interaction with the reservoir and the foundation. Such analyses are representative for earthquake ground motions that do not cause high dynamic tensile stresses in a dam. This is usually the case for the ground motion caused by the operating basis earthquake (OBE), which has a return period of say 145 years [2, 5]. In general, high arch dams with relatively low fundamental eigenfrequencies experience higher dynamic responses than small arch dams. Strong motion data of the Punt dal Gall (130m) and Emosson (180m) arch dams in Switzerland have shown that amplification factors of the maximum crest acceleration with respect to the peak ground acceleration (PGA) can reach values of 11 or even higher for small earthquakes with a PGA of 0.01g to 0.05g. These large amplification factors are also due to the low damping of the fundamental modes of vibration of these dams, which is significantly less than 5% for small-amplitude oscillations.

Once the dynamic tensile stresses exceed the dynamic tensile strength at contraction and construction joints, then the dynamic response becomes nonlinear due to cracking, joint opening and joint displacements. A fully nonlinear dynamic analysis is not yet feasible for the design of an arch dam, as the methods of nonlinear dynamic analysis are still in the research and development phase. The reliability of the existing methods has also not been verified based on observed earthquake records. Therefore, there is a need for analysis models that can reflect the nature of the anticipated earthquake damage (i.e. cracks have fully developed along contraction and lift joints) and give a conservative assessment of the safety of the dam body.

A simple and logical concept going beyond the linear-elastic analysis is needed for the seismic safety assessment of arch dams. The proposed simplified method for the post-cracking analysis of detached concrete blocks in the central upper portion of an arch dam comprises the following steps:

• (1) Linear-elastic dynamic time history analysis of dam-reservoir-foundation system; calculation of envelopes of absolute accelerations and principal dynamic tensile stresses; analyses may be carried out for both the full and empty reservoir conditions.

• (2) Selection of detached concrete blocks for dynamic stability analysis based on envelopes of principal dynamic tensile stresses and absolute accelerations (two or three critical blocks formed by open contraction joints and horizontal lift joints may be selected).

• (3) Determination of time histories of radial and vertical components of absolute acceleration at the base of each detached concrete block from the results of step (1).

• (4) Modelling of the detached concrete block by a two-dimensional (2D) finite element (FE) model, in which the cracked horizontal lift joint is represented by contact elements.

• (5) Dynamic stability analysis of the detached concrete blocks using acceleration input from step (3); the concrete block undergoes combined rocking and sliding motions; due to the geometrical constraints, the 2D concrete block is allowed to move only in the upstream direction.

• (6) Evaluation of the maximum sliding displacement and the maximum crack opening displacements at the upstream and downstream faces for at least three different earthquake ground motions, and two or three critical concrete blocks; the largest sliding motions will be experienced by concrete blocks with the smallest height located close to the dam crest, where the accelerations in the up-/downstream direction are the maximum.

*Assumptions for simplified dynamic stability analysis*

The dynamic response of the detached concrete block depends on the dynamic response of the linear-elastic dam system. Besides the earthquake input, the most important factor governing the dynamic response of a dam is the damping ratio assumed in the linear-elastic model. For the seismic safety evaluation of high arch dams, it is prudent to assume damping ratios (including the effects of wave radiation into the reservoir and the foundation rock) not exceeding 5% to 7% for very strong ground shaking.

The other basic assumptions made for the dynamic stability analysis of detached concrete blocks are as follows (the critical block selected for the Deriner arch dam is shown in Figure 6.):

• The acceleration response in the central upper portion of the cracked dam with contraction joint opening can be obtained from a linear-elastic dynamic analysis.

• The detached concrete blocks can slide only towards the upstream due to the geometry of the arch dam; any movement beyond the downstream face is not allowed.

• The effect of shear keys and friction in the vertical contraction joints is ignored.

• Cracks will form along the vertical contraction joints and horizontal lift joints only. (Note: If inclined cracks sloping towards the reservoir would develop, then considerably larger sliding movements would occur than in the case of a horizontal crack).

• For the full reservoir condition, the uplift pressure in the lift joint is triangular and the hydrodynamic mass at the upstream face can be taken according to Westergaard (incompressible reservoir).

*Results of linear-elastic analysis of dam*

In Figure 7, the envelopes of the maximum along-stream crest displacements obtained from the linear-elastic analysis are plotted for the different input earthquake motions. The largest crest displacement due to the MCE is 220mm, which is about twice the water load deflection.

If contraction joint opening is taken into account, then these displacements will increase further. The crest deflection can also be used to estimate the maximum contraction joint opening near the crest of the dam by assuming that the concrete blocks are rigid and that deformations are possible only by rotations at the joints.

*Rocking-sliding analysis of detached concrete block*

The 2D rocking-sliding analysis of the 20m high concrete block (Figure 6) was carried out using the amplified acceleration at the base of this block as the input motion. Concrete spalling at the upstream and downstream faces of the lift joint was also taken into account by moving the centres of rocking motion about 1m inwards into the dam body, resulting in a higher rocking response. The main results of the rocking-sliding analysis are listed in Table 1 for a friction coefficient of 0.7 for both the empty and full reservoir conditions. Typical time histories of the sliding and rocking movements of the 20m high concrete block for empty and full reservoir are shown in Figure 8.

*Discussion of results of dynamic stability analysis of detached concrete block*

The results of the dynamic stability analysis of the 20m high detached block show the following:

• (1) The maximum sliding displacement of the 20m high block towards the upstream is equal to 1045mm in the empty reservoir condition for a horizontal PGA of 0.35g. The duration of the strong ground shaking plays an important role on the maximum sliding displacement in the empty reservoir condition, because the sliding movement towards the upstream side is of a cumulative nature.

• (2) In the full reservoir condition, the water pressure tends to push the block back to its original position, as shown in Figure 8. As a result, a much smaller maximum sliding displacement of only 346mm was computed for a horizontal PGA of 0.35g, and the duration of strong ground shaking did not have a significant effect on the maximum sliding displacement.

• (3) The earthquake shaking with a horizontal PGA of 0.35g produces crack opening displacements of up to 179mm and 76mm at the upstream and downstream edges, respectively, in the empty reservoir condition. In the full reservoir condition, the corresponding values are equal to 144mm and 96mm, respectively.

• (4) The maximum crack opening displacement at the upstream edge of the 20m block is much larger than that at the downstream edge in both the empty and full reservoir conditions.

• (5) The 20m high concrete block is still stable after a maximum upstream sliding displacement of 1m, which is acceptable as the dam is about 17m thick at the base of this block.

• (6) It can be noted that the maximum sliding and rocking displacements vary considerably. Therefore, it is important to perform the inelastic analysis using a number of statistically independent input earthquakes; otherwise, one may arrive at too optimistic results.

**Conclusions**

The Deriner arch dam is currently under construction. The project is behind schedule due to budget limitations.

The main hazard encountered during construction is related to the slope stability of excavations, which required the installation of more than 2000 restressable post-tensioned rock anchors.

The main hazards during normal operation are extreme floods and strong earthquake ground shaking.

During extreme floods the energy of the water discharged through the surface spillways and orifice spillways corresponds to ca. 20GW. This energy has to be dissipated in a controlled way. For this reason a plunge pool is provided and the flip buckets of the two surface spillways as well as the orifices in the dam body are arranged symmetrically in order to achieve additional energy dissipation by the impact of the jets already in the air above the plunge pool.

For the design of the surface spillways hydraulic model tests were carried out. By proper aeration any negative pressures in the critical bends of the inclined spillway tunnels could be eliminated. The tests proved that aeration facilities at two locations along each tunnel, approximately just before and after the horizontal bend of the tunnels, will be sufficient to prevent cavitation.

During very strong ground shaking, tensile stresses exceeding the dynamic tensile strength of mass concrete occur in the central upper portion of an arch dam. Since the tensile strength of the grouted vertical contraction joints and the horizontal lift joints is inferior to that of the parent mass concrete, the contraction joints may open and horizontal cracks are likely to develop along the lift joints.

A simple method has been presented for the dynamic stability analysis of detached concrete blocks. The following conclusions were drawn from the corresponding stability analyses:

• (1) In an arch dam, because of the kinematic constraints and the ‘radial’ arrangement of the contraction joints, detached concrete blocks can only slide in the upstream direction. Downstream movement is restrained by arch action.

• (2) If the base of the detached block is horizontal, the sliding movement due to strong earthquake shaking is relatively small.

• (3) As the effect of the shear keys in the vertical contraction joints has been neglected, the results of the proposed simplified rocking-sliding analysis can be considered to be conservative.

• (4) Because of the dimension of the detached concrete block at the sliding surface, an upstream movement of the order of 1m can be accepted for the maximum credible earthquake. The dam is still stable after the earthquake.

*Dr. Martin Wieland, Chairman, icold Committee on Seismic Aspects of Dam Design (martin.wieland.@poyry.com), Martin Aemmer and Roland Ruoss, Poyry Energy Ltd., Hardturmstrasse 161, CH-8037 Zurich, Switzerland*

*The authors are grateful to the General Directorate of State Hydraulic Works (DSI), Ministry of Environment and Forestry of Turkey for allowing publication of this paper.*

Author Info:

[1] ICOLD Bulletin 72 (1989), Selecting Seismic Parameters for Large Dams, Guidelines, Committee on Seismic Aspects of Dam Design, ICOLD, Paris.

[2] Malla S., Wieland M. (2003), Simple Model for Safety Assessment of Cracked Concrete Dams Subjected to Strong Ground Shaking, Proceedings of the 21st International Congress on Large Dams, ICOLD, Montreal, Canada.

[3] Wieland M. (2003), Seismic Aspects of Dams, General Report, Q.83, Proceedings of the 21st International Congress on Large Dams, ICOLD, Montreal, Canada.

[4] Wieland M., Aemmer M., Ruoss R. (2007), Die 249 m hohe Deriner-Bogenmauer in der Türkei (in German), Wasserwirtschaft, 97. Jahrgang, 10/2007, pp. 69-71, Germany

Tables

Table 1