Economic optimization of the cutoff wall depth for Nenskra dam, Georgia L. Canale, Stucky SA, Switzerland During studies for the foundation treatment at the Nenskra rockfill dam in Georgia, an economic analysis was carried out, as part of the feasibility study, to weigh the losses associated with seepage against the cost of a deep cutoff wall. A partial cutoff solution was found to represent a good compromise, reducing costs while limiting seepage to an acceptable level. Fig. 1. Nenskra s monthly energy and reservoir operating levels. Nenskra dam is a 135 m high asphalt-core rockfill dam (ACRD) which is part of the 210 MW Nenskra hydro project, currently under development in Georgia. The project is in the Svaneti region, on a major tributary of the Enguri river, upstream of the existing Enguri arch dam. The Nenskra dam foundation consists of a thick layer of fluvio-glacial deposits, 90 m deep, overlying bedrock of jointed gneisses and granites. Seepage beneath the dam is controlled by a plastic concrete cutoff wall, implemented from the foundation along the dam axis. The cost of the complete cutoff, down to the bedrock, is significant and represents more than 30 per cent of the overall cost of the dam. During the feasibility study it was decided to construct the concrete diaphragm down to a sufficient depth so that the expected leakage under the cutoff would be economically viable and would occur in controlled conditions. The adopted procedure for economicl optimization is described here. Methodology Two-dimensional seepage analyses were carried out using the program SEEP/W for various maximum depths of the cutoff (D). The hydraulic conductivity values in the foundation were derived from in-situ permeability tests from about 20 boreholes. The annual water volume lost by seepage was expressed as an annual loss of energy/revenue. The provisional unit prices of the diaphragm for the various depths were obtained with the support of international contractors and supplemented with unit prices taken from other projects successfully implemented worldwide. The costs of the seepage barrier and annual revenue losses Table 1: Monthly generation of the Nenskra powerplant Reservoir level (masl) Energy (GWh) Tariff (US$/kWh) Revenue (US$ mil) Jan 1398 91 0.0570 5.19 Feb 1378 85 0.0570 4.85 Mar 1350 88 0.0570 5.02 Apr 1339 99 0.0570 5.64 May 1365 140 0.0715 10.01 Jun 1396 145 0.0715 10.37 Jul 1414 155 0.0715 11.08 Aug 1413 157 0.0715 11.23 Sep 1422 63 0.0715 4.50 Oct 1428 45 0.0570 2.57 Nov 1430 36 0.0570 2.05 Dec 1416 90 0.0570 5.13 Average 1396 1194 0.0650 77.63 from energy production were then actualized in a discounted cash-flow analysis over a 35-year period. The most economic cutoff depth was considered to be that which corresponds to the minimum net present value (NPV). The risk of piping of the foundation material has been thoroughly assessed for the proposed option. Piezometers and a measuring section in the riverbed immediately downstream of the dam toe are envisaged, to monitor the total seepage flow continuously during operation. Energy production profile The main purpose of the Nenskra hydro scheme is to generate and deliver electricity to the national grid during the winter, and to export to neighbouring countries (mainly Turkey), with a profitable market, during the summer. Strict requirements on the firm energy were established by the owner on the monthly production over the three months of December, January and February. Because of the seasonal variation in the natural river discharges, some of the summer runoff has to be stored in the Nenskra reservoir. The average monthly production profile and reservoir operations are shown in Table 1 and Fig. 1. The Nenskra plant is designed to generate nearly 1200 GWh/year. The reservoir is drawn down from December to April and gradually filled again from May through to September. During October and November, the reservoir level is maintained at the maximum elevation until the beginning of December when it is lowered again. The reservoir can oscillate between els. 1335 and 1430, the annual average level being el. 1396. This level was con- 70 Hydropower & Dams Issue Six, 2013
sidered to be constant in all the seepage-flow analyses carried out for the purpose of the economic optimisation. The internal market price of US 5.70/kWh is considered in the study for the winter tariff (October to April); the Turkish market price, net of the transmission cost up to the border with Turkey, is paid for the summer export period (May to September), being US 7.15/kWh. The average annual revenue amounts to US$77.63 million, which corresponds to an annual electricity selling price of US 6.50. This price has been adopted in the economic comparison of options, for estimating the price value of the annual revenue lost by seepage. Design of foundation treatment The Nenskra dam is founded on a deep layer of compressible and pervious fluvio-glacial materials in a U-shaped glaciated valley. The foundation was investigated based on borehole drillings (full recovery), geophysical profiles and permeability tests. Fresh and strong granite-gneisses are exposed on both abutments, and were found at depths of nearly 90 m in the riverbed in the centre of the valley. The deeper soil layer, QG, comprises glacial alluvial outwash and is characterized by sands and sandy gravels with cobbles. This layer is 40 to 60 m thick, is fairly dense and has a lower permeability. The upper soil layer, QF, is an alluvial fan resulting from the deposition of alluvial material transported by the Nenskra river and the numerous side creeks. This layer, 20 to 30 m thick, comprising sands and sandy gravels with cobbles and boulders (as small as 50 cm), shows a low degree of compaction and is more pervious. The base rock, P Z, is composed of early Palaeozoic gneisses and migmatites with diorites and minor granites. The rock joints are generally tight. The watertightness of the rockfill dam will be provided by a vertical core in the bituminous concrete. The asphalt core will be connected to a concrete plinth running along the axis from one abutment to the other. Along the abutments, the concrete plinth will be placed on the excavated rock, treated with consolidation grouting and a 40 m-deep grout curtain, while in the central part, the plastic concrete diaphragm will be executed from a horizontal platform down to the optimal depth. The diaphragm will be directly connected to the asphalt core with asphalt mastic. The horizontal platform is designed to be 20 m wide, to allow for easy movement and operation of the equipment. To reduce the excavation volume along the dam axis, the riverbed is to be filled for 15 m, in the narrow part, with adequately compacted gravel, so that the cutoff wall will be provided from el. 1320. The total length of the platform will be about 600 m. Because of the particular rock profile, for the first 150 m on the left bank, a total cutoff will extend down to the bedrock, and will have depths of between 10 and Fig. 2. Longitudinal section along the dam axis: (a) geological features; (b) foundation treatment. Hydropower & Dams Issue Six, 2013 71
30 m; the seepage barrier will be supplemented by a grout curtain 50 m deep, down to el. 1240. The remaining length of the cutoff is part of the subject of the economic optimization described here. The length of the cutoff wall ranges from 450 m at 20 m depth, to 250 m at 90 m depth, and will be excavated by hydro-mills (or trench cutters). The geology and foundation treatments envisaged for controlling seepage are shown in Fig. 2. Seepage analysis The seepage analyses were carried out in steady conditions using the GEOSTUDIO (SEEP/W) software, which is suitable for analysing filtration in saturated and unsaturated porous media. The upstream reservoir level is assumed to be at el. 1396, corresponding to the average operating level in the annual reservoir operation. Two different geometries were considered in the 2D finite element model to estimate the total quantity of seepage. One transverse section was taken in the central part of the dam axis where the dam height and the alluvium depth are at the maximum, and the other was taken from the lateral part, where the bedrock is 70 m deep. To obtain the total quantity of seepage, the unit flow was then multiplied by an average seepage front (W) of 135 m and 250 m, respectively for section 1-1 and 2-2. The two sections of the model are shown in Fig. 3. Table 2: Hydraulic conductivities and anisotropy ratios Hydraulic conductivity k h k v/k h Material (m/s) (-) Shell 1E-02 1.0 Transition 1E-04 0.2 Filter 1E-05 0.2 Asphalt core 1E-12 1.0 Plastic concrete 1E-08 1.0 Upper alluvium 1E-03 0.1 Lower alluvium 5E-05 0.1 Bedrock (ungrouted) 1E-07 1.0 The dam embankment is constructed of rockfill and the hydraulic conductivity values (k) are well known from international experience and the technical literature. For the asphalt concrete and the plastic concrete, k values were provided by specialized contractors. As mentioned above, permeability values for the upper and lower alluvial layers and the bedrock were derived from site investigations. The shell material in the dam body can be considered as isotropic, but, as a result of the considerable layering, a significant anisotropy is expected for the alluvium in foundation. The perme- Fig. 3. SEEP/W Model: (a) Section 1-1 at chainage 0+500; (b) Section 2-2 at chainage 0+700 72 Hydropower & Dams Issue Six, 2013
Table 3: Seepage flow results Max cut-off depth D (m) 30 40 50 60 70 80 90 Unit flow (m 3 /s/m) Cross-section 1-1 (W = 135 m) Cross-section 2-2 (W = 250 m) 1.09E-02 3.13E-03 1.17E-03 8.77E-04 6.60E-04 4.83E-04 2.43E-04 9.25E-04 7.12E-04 5.61E-04 4.22E-04 2.05E-04 2.05E-04 2.05E-04 Total seepage (m 3 /s) 1.709 0.600 0.299 0.224 0.140 0.116 0.084 ability in the vertical direction is assumed to be one tenth of that in the horizontal direction. The cutoff forces seepage water to move along in a vertical direction down the wall and back up to the ground surface. The seepage path around the concrete cutoff wall will be twice the depth of the cutoff, and therefore the cutoff is more effective as the ratio kv/kh decreases. Hydraulic conductivity values adopted in the model are given in Table 2. The results of the seepage flow analysis for the various cutoff depths are given in Table 3. The seepage barrier starts to become really effective for D > 40 m, or in other words when the more pervious upper alluvium layer is entirely cut off by the concrete diaphragm. Unit cost of the cutoff The unit price of a plastic concrete cutoff wall depends greatly on the geological conditions and the maximum depth to which the diaphragm has to extend. In overburden material, the cutoff excavation process is further complicated by the presence of boulders and is sensitive to the size, frequency and strength of the boulders. Deeper cutoffs require tighter tolerances for the panel verticality and more elaborate equipment is required. Over the last two decades, the rapid development in excavation techniques, panel concreting and quality control has made this solution more and more effective and cost competitive, especially for depths greater than 20 30 m. With the latest generation of hydraulically operated trench cutters, plastic concrete cutoff walls can extend to a depth of 150 m and cut through steep and strong bedrock or large boulders. At the Nenskra project there is no technical limitation to prevent implementing a total cutoff wall down to the rock base; the only limitation may be the cost. A trench cutter, together with the concepts for the cutoff excavation and concreting works are shown in Fig. 4. During the feasibility study, consultations took place with the most qualified contractors worldwide to define reasonable unit prices for the cutoff at Nenskra (including equipment, concreting works and all ancillary works). In addition, intensive research was carried out on the final cost of cutoff walls at international projects successfully completed over the last 30 years in North and South America (for example in the USA, Chile and Argentina). This cost analysis was also aimed at identifying the critical aspects encountered at each project during the implementation of the cutoff works, to make sure that the range of prices adopted for classes of maximum depth would cover the largest number of unexpected and unforeseeable conditions. For those projects completed some time ago, prices were escalated. The resulting price list is given in Table 4. Unit rates are given per square metre of wall, because most of the price represents the machine excavation and auxiliary costs, and is payable per square metre. The cost of concreting works (materials, casting, and so on), generally payable per cubic metre, is minor, and typically represents less than 10 per cent of the total cost for deep cutoffs and up to 20 per cent for shallower cutoffs. The thickness of the concrete diaphragm ranges from 50 to 120 cm, as a function of the differential water head and the seepage gradient through the wall. Table 4: Unit rates of plastic concrete cutoff Maximum depth D (m) Unit rate (US$/m 2 ) 0-20 < 250 20-40 250-400 40-60 400-600 60-80 600-1000 80-100 1000-2000 100-120 > 2000 Fig. 4. The Trevi Hydromill, and conceptual scheme for the implementation of the cutoff. Hydropower & Dams Issue Six, 2013 73
Table 5: Economic comparison for different cutoff depths Cutoff max. depth (m) Cutoff area (m 2 ) Under seepage flow (m 3 /s) Annual energy loss (GWh) Energy loss Cutoff cost Total Annual revenue loss (US$ million) NPV (US$ million) Unit cost (US$/m 2 ) Cutoff cost (US$ million) NPV (US$ million) NPV (US$ million) 30 12 660 1.709 85.3 5.54 53.4 315 4.0 5.3 58.7 40 16 630 0.600 30.0 1.95 18.8 394 6.6 8.7 27.5 50 20 490 0.299 14.9 0.97 9.3 493 10.1 13.4 22.8 60 24 190 0.224 11.0 0.71 6.9 618 14.9 19.9 26.8 70 27 750 0.140 7.5 0.49 4.7 774 21.5 28.6 33.2 80 30 840 0.116 5.8 0.38 3.6 969 29.9 39.7 43.4 90 33 390 0.084 4.2 0.27 2.6 1213 40.5 53.9 56.5 Fig. 5. NPVs for various cutoff depths. Unit rates refer to the maximum depth to be reached, because this governs the selection of the machinery and equipment to be used. In general terms, down to depths of 50-60 m, the cutoff works are accomplished with equipment and methodologies which are currently quite standardized. On the other hand, a noticeable cost increase is observed for maximum depths greater than 60 m. For the purpose of the economic optimization, the following equation, expressing the unit price UP as a function of D, has been derived by regression from the values in Table 4: UP($ m 2 0.0225 D(m) ) = 160.17 e (1) Economic optimization The seepage loss and the consequent loss of revenue from energy sales decrease with the cutoff depth. Conversely, the cost of the cutoff wall increases with depth. Future revenue losses during operation and the cost of the cutoff can both be considered as outgoing cash-flows for the same investment, and they have to be discounted appropriately to the same time value. The economic optimization procedure of the cutoff wall consists of finding the depth corresponding to the option with the minimum net present value (NPV). Annual energy loss is proportional to the annual volume lost by seepage and the net generation head and is expressed by the following equation: E (GWh) = c ηqh (2) where: c is a constant (8.59E-02 for the Nenskra scheme); η is the global efficiency of the generators (0.88); Q is the under-seepage flow in m 3 /s; H is the net head (m). The 210 MW Nenskra powerhouse will be equipped with three 70 MW Pelton units with a setting level at el. 705, fed by a 15 km-long power tunnel and a high head pressure shaft. Therefore, assuming the reservoir level at el. 1396, the net head for generation including head losses in the power waterway will be approximately equal to 660 m. Annual revenue loss (EL) is obtained by multiplying the energy loss from eq. (2) by the average tariff of 0.065 $/kwh. The annual shortfalls of future cash inflow are discounted over the 35-year generation period and the net present value is obtained by the following expression: (3) where: t is the time in year and i is the discount rate; i is normally considered as the weighted average cost of capital (WACC), and for the purpose of the present analysis is taken at a time constant value of 10 per cent. The cost of the plastic concrete cut-off for a given depth is calculated by multiplying the unit price from eq. (1) by the total area of the diaphragm, graphically measured from the longitudinal section, see Fig. 2b. In accordance with the provisional five-year Nenskra general construction schedule, the cutoff works will start one year after commencement date for the project t (once excavation of the dam foundation and the diversion works have been completed) and it will not last more than 10 months. The net present value of the cutoff cost is given by the following equation: (4) where: t is the time from the reference year at which the cutoff cost (CO) is paid. The reference year, t = 0, for discounting past and future cash-flow is considered as the last year of the investment; therefore, t in eq. (4) equals -3. The results of the economic comparison are shown in Table 5. The curve of the total NPV (loss of revenue + cost of cutoff) is plotted in Fig. 5. The minimum NPV is obtained for a maximum cutoff depth of 50 m. The annual loss of energy caused by the under-seepage is about 15 GWh, representing nearly 1 per cent of the total annual generation. For deeper cutoffs, the seepage flow reduction does not economically justify the extra cost of the anti-seepage barrier. The cost of a total diaphragm extended down to the bedrock would amount to US$41 million, equivalent to about 32 per cent of the total cost of the Nenskra rockfill dam. 74 Hydropower & Dams Issue Six, 2013
cutoff. Therefore, any possible increase in the design depth, as a result of more severe permeability conditions, could only marginally affect the design of the cutoff system and will not compromise the technical and economical feasibility of the proposed solution. Various electricity selling prices have been taken into account within the expected range of US 6.5 to 8.5/kWh. The economic loss caused by the seepage under the dam increases with the electricity price, so the optimized cutoff depth will increase accordingly to limit this loss. Nevertheless, the sensitivity analysis has shown the very limited influence of the market price on the optimization study. Fig. 6. Total NPV curves for various anisotropy ratios (a) and electricity selling prices (b). This outcome is based on the assumption k v/k h = 0.1 in the foundation soil, which seems fairly realistic for bedded deposits, and considering an annual electricity tariff of US 6.5/kWh. Two sensitivity analyses were carried out to cover the main uncertainties of the study: the first was done by varying the anisotropy ratio of the alluvium layer within a reasonable range; the second was done by varying the electricity tariff within the expected range. The evolution of the minimum NPV is shown in Fig. 6. Soil deposits with considerable layering are characterized by a marked anisotropy, generally being k v/k h= 0.1. More pessimistic values of the ratio k v/k h as large as 0.25, 0.5, 0.75 and 1.0, have been considered, to check how rapidly the optimal depth moves. The sensitivity analysis has revealed that the optimal depth varies in a range of 10 m for the case of the Nenskra Risk of piping Apart from economic issues, to be technically feasible the partial cutoff solution will not expose the foundation to the risk of piping or migration of finer material, during the long-term operation of the dam. In general, gravelly soils where fine sand or cohesionless silt is not present in persistent or homogeneous layers are less susceptible to piping. Similar conditions occur in the foundation alluvium of the Nenskra dam. In addition to that, the strong layering and interbedding represents an obstacle to the migration of the fines, especially in the vertical direction, where the seepage path is forced by the presence of the cutoff. The flow-net for the selected cutoff depth of 50 m and with the reservoir maintained at full supply level is shown in Fig. 7. The exit gradients at the downstream toe of the dam, where the backward erosion could initiate, are not significant (that is, less than 0.34). The factors of safety against piping or blow-up, defined as the ratio of the vertical stress to the pore pressure at any point, are more than 2.3 greater than the minimum allowable 1.5, as suggested by Fell [2005 1 ]. The hydraulic gradients at the toe of the wall can be as high as 10; however, they will be very unlikely to cause erosion since the possible lenses of finer material will be bounded by graded alluvium. Locally, only small movements of the fines are expected, which would not affect the overall stability or performance of the plastic concrete diaphragm. During first filling and the long-term operation of the reservoir, the under-seepage flow will be monitored to control the exit gradients and the risk of erosion continuously. Fig. 7. Flow net for D = 50 m and the reservoir at full supply level. Hydropower & Dams Issue Six, 2013 75
Conclusions During the feasibility study for the Nenskra project, a partial cutoff wall of plastic concrete was proposed as foundation treatment for the dam, to reduce excessive under-seepage. A complete cutoff of 90 m, extending for a few metres into the bedrock could practically eliminate leakage beneath the dam, but on the other hand would be a very expensive solution, representing nearly one-third of the total cost of the rockfill dam. The optimal depth of the partial cutoff has been assessed by economic comparison of the value of hydropower which could be lost versus the cost of the diaphragm. The best economy is achieved for cutoff depths of around 50 m. The seepage quantity would be economically acceptable, and the expected leakage conditions in the foundation would be safe against potential erosion or piping. The full methodology, which has been described here, is supported by field data on hydraulic conductivity, consultations with cutoff contractors and reference projects. The main uncertainties are related to the anisotropy ratio of the alluvium layer and the electricity tariff; nevertheless, the sensitivity analyses have shown that the optimal cutoff depth is only mildly sensitive to the first parameter and hardly affected at all by the second one. It is clear that the final selection of the cutoff depth to be adopted for the Nenskra dam may change based on new factors which could emerge later during the detailed design or construction, but the approach remains valid and is considered to be very supportive for the early stage of design and highly influential in the decision-making process. Acknowledgements The author is grateful to JSC Partnership Fund and CEO, Mr Irakli Kovzanadze and to JSC Nenskra, the Client, and CEO, Mr Teimuraz Kopadze, for assistance in the preparation of this paper. The author would also like to thank Mrs Raffaella Granata of Trevi SpA for support on the definition of design requirements for the plastic concrete cutoff works. Reference 1. Fell, R. et al., Geotechnical Engineering of Dams, Taylor & Francis, UK; 2005. Bibliography Bruce, A., di Cervia, A.R., and Amos-Venti, J., Seepage remediation by positive cut-off walls: A Compendium and Analysis of North American case histories, Canadian Dam Association Conference, Canada; 2006. Pinilla, L., CFRD on deep alluvium, Blanket-Concrete face combination, MN Ingenieros, Chile; 2001. ICOLD, Bulletin No.150, Cutoffs for dams, Internaitonal Commission on Large Dams, Paris, France; 2010. L. Canale Luciano Canale is a Hydraulic Expert and Project Manager at Stucky Ltd. He graduated with an MSc in hydraulic structures from the Technical University of Naples, Italy. He has 13 years of experience in the design and implementation of hydroelectric schemes, hydraulic structures, gravity and rockfill dams. As a Project Manager, he has headed the design works for several large dams and hydro projects in Europe, Asia and Africa. He was the Project Manager for the feasibility study and the basic design of the Nenskra scheme, completed by Stucky in 2012. He is currently involved in the design and site supervision of two large hydropower projects under construction in India. He is a member of the Swiss Committee on Large Dams. Stucky SA, Rue du Lac 33, 1020-Renens, Switzerland. 76 Hydropower & Dams Issue Six, 2013