Samanalawewa reservoir was created by damming the Walawe river, one of the five main rivers in Sri Lanka. The project is situated 160km southeast of the capital, Colombo, and is a key element of the 120MW Samanalawewa hydroelectric project. The project includes a 100m high rockfill clay core type dam, 4.5m in diameter, together with a 5.35km long tunnel, a 685m long steel penstock, surface power station housing two turbine generator units and a 110m long tail race canal. The general layout of the project is shown in Figure 1.
Construction works on the project were carried out between 1986 and 1991. On completion of the main works, a trial impounding was carried out to ascertain the reservoir water tightness due to the very poor geological conditions discovered across the right bank. Soon after impounding commenced, the reservoir started to leak in its right bank. The leakage appeared as a small spring located about 300m away from the dam. Impounding had to be suspended and remedial measures were planned.
Geologically the site is in a highly complex region subjected to a number of tectonic movements. Signs of karstification are also apparent. The site features a highland series of the Sri Lankan Precambrian complex comprising crystalline metamorphic rocks. Deep weathering was identified throughout the right bank.
After detailed geological evaluations, a 100m deep, 300m long grout curtain was implemented as a remedial measure. On completion of the grout curtain construction in early 1992, reservoir impounding continued. However, despite the presence of the grout curtain, the reservoir continued to leak in the same location. The leakage and the ground water pressure in the right bank gradually increased with the rising water levels, indicating the inadequacy of the grout curtain.
In 1998 a second attempt was made to remedy the leakage. This involved laying a wet clay blanket in ingress zones in the reservoir bed. The ingress zones were identified following advanced geological and geophysical investigations and water chemistry studies.
However, even after the blanket laying was completed, the reservoir continued to leak – although there was a slight reduction in the flow rate and a considerable reduction in the right bank ground water pressure. Today the reservoir leakage is at a maximum rate of 2m3/sec at full supply level and varies according to the reservoir levels.
The isotope study
After these attempts to seal the leakage proved unsuccessful, an isotope study was conducted to discover the source of the leak. During the initial phase of this study, environmental Isotopes were used as a tracer. In this regard the natural isotopes present in the different water sources are monitored and the isotope signature is established. Comparison of different signatures is then undertaken to identify hydraulic relationships that may exist among different water sources.
The first step of the study was to establish a proper sampling plan. Reservoir configuration and the site’s geological structure was taken into consideration when deciding on the sampling plan, which initially included about 20 locations representing the whole reservoir system. Thus it included river water, leakage outlet, ground water monitoring wells, drain pipes and rain water precipitating into the reservoir area. Because the dam is located just downstream of the confluence point of Walawe river and its main tributary Belihul river, measurements were taken separately for the two rivers. All locations were monitored on a regular basis over a 12-month period to establish the stable isotope signatures of the reservoir and the surrounding ground water system. The sampling locations are shown in Figure 2.
The RBS2, RBS6, RBS13, RBS23 and SP58 are standpipe piezometers installed in the grouting adit in the right bank while GW2, GW11, GW18 and MS3 are surface ground water monitoring wells in the right bank. This grouting adit, which is 1.8km in length and driven in to the right bank at river bed level, was used during construction of the 100m deep grout curtain.
The monitoring well GW18 – which is located 2km away from the dam, making it the furthest sampling point away from the structure – is not shown in the figure. The monitoring well GW19 is located in the left bank. Sampling of water from the above locations was carried out on a monthly basis, with testing conducted in a laboratory abroad. In this study the concentrations of the following stable isotopes were monitored:
• Deuterium 2H or D
• Oxygen – 18 18O
• Tritium 3H
• Carbon – 14 14C
• Sulphur – 34 34S
Results
The results obtained from laboratory tests have been analysed and the results are presented graphically.
Figure 3 gives a general picture of the isotopic distribution in the reservoir area. All the measured isotope signatures are almost lying on the Global Meteoric line (GMWL) and are widely spread along it.
Figures 4 and 5 show the variation of 18O and 2H respectively of leakage water (marked as main leak), river water and ground water from several selected locations over time.
Due to the complex nature of the leakage phenomenon, the test results do not indicate any straightforward information on the leakage flow paths, however the following observations could be made. The isotope signatures of the leakage water and most of the ground water wells fall in between that of the two rivers, suggesting that leakage water could be considered a mixture of waters from the two rivers. According to the graphs, the leakage water and ground water represent a very similar behaviour with respect to both isotopes. Further, it should be noted that the isotopic signatures shown by the monitoring wells near and away from the dam are almost similar, indicating the existence of a widespread underground aquifer. This is further supported by the fact that all monitoring wells in the right bank show a flat water table, which is fluctuating with the reservoir water level. It is a possibility that the ground water in the aquifer has been replaced by the reservoir water. The reservoir water which infiltrated the right bank aquifer through ingress zones may have initially mixed with ground water locally, and then replaced the ground water as the leakage outlet continue to discharge from right bank. Considering the time elapsed between the initial leakage initiation in 1991 and the sampling period 2001 this possibility cannot be ruled out. Furthermore, geologically this area had been subjected to karstification and the underground cavernous nature prevailing has been confirmed during drilling and grouting operations.
During planning of the original remedial measures, it was established that the major leakage paths originating from the reservoir run across the area where the stand pipe piezometers RBS4, RBS6 and RBS10 are located. This was based on the fact that several major geological faults were cutting across the right bank in the same area as well as on the water chemistry studies. However according to the isotopic signatures, no difference in RBS2 and RBS6 and other ground water wells could be seen.
Unlike most geochemical tracers, 2H and 18O are inert and conservative in mixing relationships. Therefore stable isotopes can serve to quantify groundwater mixing at the local to catchment scale where mixing between ground waters of different recharge origins, from different aquifers and flow systems can take place. Mixing between two distinct water sources A and B could be represented by simple linear algebra as shown below:
dmixture = X dA + (1-X) dB
Where dmixture is the d2H or d18O of the mixed water. X is the fraction of water source A in the mixture of A and B.
In this study the leakage water was considered as a mixture of original groundwater in the right bank and the reservoir water entering into the right bank, and an attempt was made to derive the mixing proportions using the above equation. The isotopic concentrations of monitoring wells which are considered to be intact from leakage phenomenon were used to represent the original ground water in the right bank. Several computational trials were carried out using different wells representing original ground water conditions. However none of them resulted in a realistic value for the mixing proportion.
Analysing the data
After analysing obtained data the following conclusions can be made:
• The isotope signatures of the leakage water and the ground water fall in between that of the two rivers feeding the reservoir. This indicates that over the years ground water in the right bank has been replaced by the reservoir water.
• In all the ground water wells monitored the leakage outlet shows a similar variation of isotopic signature with time. This confirms the existence of an underground aquifer extending along the right bank. It also indicates that river water movement takes place throughout the right bank and any specific areas or paths cannot be clearly identified by this method. Also the previously established leakage paths through RBS4, RBS6, and RBS10 by other methods are no longer valid. Furthermore the ground water holes located further away from the dam show a similar variation with those near the dam.
• No information on ground water and river water mixing could be derived, suggesting that the ground water in the reservoir precincts has already been replaced with river water.
• No conclusion could be made on the possible leakage paths and ingress zones which are necessary in planning the remedial measures. Therefore the second phase of this study will involve carrying out artificial isotope injection in the reservoir bed and in selected bore wells to trace the leakage paths and the ingress zones. It is also intended to carry out an estimate of the internal cavernous spaces by monitoring the 3H variation in the leakage water and ground water over time.
Author Info:
The authors are: Kamal Laksiri, Ceylon Electricity Board, Sri Lanka, kamallaksiri@hotmail.com; D.G.L.Wickramanayake, Atomic Energy Authority, Sri Lanka, dglwiki@yahoo.com; and Professor Yushiro Iwao, Saga University, Japan, iwaoy@cc.saga-u.ac.jp