The Lower Kihansi Hydropower Project in Tanzania is owned by the Tanzania Electric Supply Company (TANESCO), and was constructed between 1994-2000. Located in the Rufiji basin, some 550km south west of the capital Dar-es-Salaam (figure 1), the plant, with its three 60MW Pelton turbines, contributes approximately 30% of Tanzania’s total electricity production.
The project, with a total construction cost, including installations, in the order of US$250M, was financed by the World Bank, the Norwegian Agency for International Development (NORAD), the Swedish International Development Agency (SIDA), the European Investment Bank (EIB) and Germany’s Kreditanstalt für Wiederaufbau (KfW).
Kihansi was the first project in the world with a water head as high as 850m in an unlined pressure tunnel. A complicating factor was the destressed rock conditions within the east African rift system. A comprehensive programme of rock stress measurements, using hydraulic fracturing tests, formed the basis for the detailed location of the powerhouse complex and the transition between the unlined and the steel lined sections of the headrace tunnel. The powerhouse needed to be relocated by some 800m to obtain a sufficient magnitude of rock stresses. Leakage reduction treatment of the headrace tunnel was based mainly on high pressure cement grouting, ahead of the tunnel face.
Excavation works included 9.8km of tunnels in addition to the powerhouse cavern with transformer niches of 26,600m³.
Project development
Based on a 1990 feasibility study, the World Bank and TANESCO subsequently agreed to incorporate the implementation of the Lower Kihansi project into a sector loan package – the Power VI Project Programme.
In December 1991, Norwegian consultant NORPLAN was selected to carry out a feasibility review of the scheme, and if this proved positive, to continue with the final design. Tendering and supervision of construction followed as an extension to NORPLAN’s initial contract.
Feasibility review and field investigations were conducted in 1992, followed by the detailed design and preparation of tender documents. Considering the ground conditions, it was found that a deeply sited tunnel system would be advantageous, compared to the shallow tunnel system previously proposed. The newly suggested design included an unlined headrace tunnel with a maximum head of close to 850m and an underground powerhouse.
Actual construction started in July 1994, with the mobilisation of the Chinese contractor SIETCO, who won the bid for preparatory works. In July 1995 the Italian contractor Impregilo SpA mobilised for the underground works. Construction was successfully completed in February 2000, within budget and time schedule.
Project description
The Lower Kihansi Hydropower Project exploits the steep Udzungwa escarpment where the Iringa plateau drops some 1000m to the Kilombero flats. This topographic feature permits the economic development of a hydro scheme by providing the high head needed for the relatively low streamflow (figure 2).
A 25m high concrete gravity dam impounds a small reservoir with a total storage of 1.6Mm³. The intake connects to the headrace tunnel via a circular unlined vertical headrace shaft (25m²), some 500m deep. The unlined headrace tunnel slopes at a gradient of 1:7. At the downstream end are the stonetrap and the transition section to the steel penstocks. The tunnel is 2,200m long and has a cross-section of 30m², except in the last downstream 600m where its cross-section is 37.5m² to allow for lining, if necessary, in the zone of the highest pressure. No surge shaft or chamber had to be provided due to the relatively short tunnel length and the low water velocities, combined with the use of Pelton turbines. The lower end of the headrace tunnel is located about 750m below ground.
The tailrace tunnel has a length of 2,100m and a cross-section of 34m². It slopes gently downstream at an inclination of 1:940 and connects to a 620m open-cut canal that transfers the water into the existing Kilombero river system.
The powerhouse cavern is excavated deep into the mountain massif and is 12.6m wide, 98m long and 32m high, with space for possible future installation of two additional turbines. The powerhouse is connected to the outside by a 1,900m access tunnel (40m² in cross-section) and also by a separate cable tunnel, which was chosen as an extra security measure for the 220kV cables passing from the underground power house to the outdoor switchyard.
A low-gradient, unlined water tunnel with a head of 846m had not been constructed before anywhere in the world. In two other hydropower plants in Norway, Tjodan and Nyset-Steggje, higher heads were obtained in 45o inclined shafts, 875m and 964m respectively. The design at Kihansi is more vulnerable with regard to leakages, as the high head section in a 1:7 inclined tunnel is longer than in a 45o shaft.
Excavation and support
All tunnel excavation was carried out by drill and blast through the area’s competent Gneiss with the blast holes drilled by two-boom jumbos (Tamrock and Atlas Copco). Loading was done partly with Schaeff ITC 112 and Schaeff ITC 312 electric chain loaders and partly with Caterpillar 966 wheel loaders. Mucking was carried out using IVECO 20t road dumpers.
The preferred cut type in the early phase was the V-cut, considered by the contractor to be more suitable in fractured rock. With experience, parallel hole cut became more common, mainly with two 102mm diameter holes. During excavation of the headrace tunnel, the number of cut holes was increased to three. The number of blast holes per round was normally 90-100 in the larger tunnels with drill rod lengths of 3.9m-4.6m. A wide range of explosives was used, mostly various cartridged slurry types. ANFO was also used, in the beginning mixed at the tunnel face and inserted into the blast holes in plastic bags.
Excavation of the 505m deep 25m² circular vertical headrace shaft was carried out by shaft sinking to the full size from the surface. This work was subcontracted to South African company Shaft Sinkers.
Drilling and charging constituted from 45%-54% and mucking 29%-34% of the construction time, except for in the bypass tunnel and in the lower part of the headrace tunnel where sounding and grouting was comprehensive. Installation of rock support consumed between 4% and 8% of the construction time. Support needs were in general low, corresponding well with the predictions in the tendering.
Relocation of powerhouse
The Udzungwa escarpment, belonging to the eastern branch of the east African rift system, was formed by large scale block faulting related to extensional tectonism. The rocks are mainly competent gneisses of the Pan African Mozambique Belt, subjected to high grade metamorphism. The degree of faulting and jointing is in general moderate to low with most of the joint/fault system oriented perpendicular to the tunnel system and with partly high permeability.
Although the overall concept of the scheme is simple, the high water pressures in an unlined headrace tunnel proved to be a real engineering challenge. The use of an unlined headrace tunnel is dependent upon the confining rock stresses being higher than the internal water pressure in the tunnel. SINTEF of Norway and SOLEXPERTS of Switzerland were engaged for conducting the complicated rock stress measurements. The primary aim of these was to ensure that there were sufficient internal stresses in the rock mass for adopting an unlined design, giving large cost savings within the concept of acceptable water leakages.
Stress measurements, by use of hydraulic fracturing methodology, were first conducted in deep, core drilled holes from the surface. After some costly and time-consuming attempts, where test equipment was lost in deep drill holes, a different approach was chosen. Hydrofracturing tests would be carried out from short holes drilled from within the tunnel during excavation. If the results from these tests were unsatisfactory, the layout would have to be modified.
The contract conditions were written to allow the power station to be sited deeper into the rock massif if necessary, since this would result in larger rock cover and probably improve the rock stress conditions for the critical part of the headrace tunnel. Hydrofracturing tests began when 800m of the access tunnel had been excavated and 300m remained to the original powerhouse location. Initial testing gave insufficient minimum principal rock stresses and it was decided to relocate the powerhouse. The encountered stress pattern was characterised by sub-horizontal minimum principal stresses oriented north to south, parallel to the tunnel axis and perpendicular to the main joint orientation. This pattern is used to reflect the original stress situation so that the low minimum principal stresses can be explained by extensional tectonism.
Continued stress measurements by hydrofracturing/ hydrojacking and by overcoring methods indicated an improved stress situation deeper into the rock massif. In total 124 hydrofracturing/hydrojacking tests and 22 triaxial tests by overcoring were conducted. After critical analysis of the test results, it was decided that the results were satisfactory and in general sufficiently high stresses were available within the rock mass to proceed with the unlined headrace tunnel design, provided that the power station complex was moved 780m into the mountain beyond the initial location.
A 120m long horizontal steel penstock liner from the power station to the headrace tunnel was designed to give an acceptable pore pressure gradient.
Leakage prevention
Economic analyses were made to obtain parameters for reducing leakage from the headrace tunnel. The net present value of production losses at the Kihansi project was estimated to US$24,000/l/s. Pre-grouting, performed at the face, was concluded to be the most cost efficient manner of reducing permeability in leakage zones. It was decided that pre-grouting was feasible and efficient when rock permeability, measured as a Lugeon value, exceeded 0.5-1.
Pre-grouting using both ordinary Portland cement and micro-cement, was conducted in sections where unacceptable permeability was detected by water pressure testing. The criteria for grouting was set to the permeability k> 2 E-7 m/s. Usually, sounding was performed with three holes of 18m length. The conductivity of the rock was determined by water pressure testing of each hole. If deemed necessary, post grouting was also conducted in leaking zones.
Various types of sectional linings and local structural measures in addition to grouting were considered, and designed for sections with unacceptable permeability. Type and length of such measures were selected based on a cost – benefit estimate. In the lower part of the headrace tunnel, at Chainage 66-90 and 111-144, sections of concrete linings were applied to reduce potential leakage through some pronounced joint systems.
Relatively high grout take was experienced up to chainage 150, with moderate permeability measured from the water pressure tests. The grout pump pressure was high, 60-90 bar, during most of the grout operation. In this area weathering was found along the joint planes. The relatively high grout take could be explained by ‘piping effect’ by the high grout pressure in this weathered material. Between chainage 390-480 the water pressure tests showed high permeabilities and the grout take was also very high. Water pressure testing of control holes in grouted zones showed k-values in the order of 2-8E-08 m/s.
Between chainage 500 and 1250 the rock was found in general to be tight with no grouting required. Fracture zones in the area chainage 1400-1450 gave very large water inflows through sounding holes (2-300l/min from one hole) needing as many as four grouting rounds before control holes gave satisfactory results (figure 3).
Amounts of leakage water into the tunnels during excavation were monitored weekly by collective measurements at the tunnel portals. Figure 4 shows the variation in leakage in Access and Headrace tunnels, influenced by various factors. The low ingress of water in the lower part of the Headrace tunnel is noteworthy. Even if the degree of fracturing of the rock here is in general low, the comprehensive grouting program prior to the excavation has very likely reduced the water ingress.
Water filling
Water filling of the headrace system was considered to be an especially critical operation and the response in the rock mass around the tunnel had to be analysed in detail. A monitoring system included automatic logging of pore pressures in strategically placed bore holes, both upstream and downstream of the concrete plugs. NGI (Geotechnical Institute of Norway) was engaged for the technical design and installation of the instrumentation. Pore pressure measurement was done in 17 holes, with lengths from 25m-140m.
Filling of the headrace tunnel and the shaft was planned for about one month’s duration to avoid high local hydraulic gradients in the joints close to the tunnel, and to avoid excessive changes in rock stresses around the tunnel. The plan included various stops at certain filling levels in the tunnel and in the shaft in order to monitor net inflow from rock to the tunnel or outflow from the tunnel into the rock at different water pressures.
During the filling operation, water inflow in the ‘dry zone’ of the tunnel system was followed closely. Several stations for manual monitoring were established. When the water head upstream of the plug passed in the order of 750m, serious inflows occurred inside a gallery of the bypass plug, and also downstream of the plug, with up to 50-60l/s. Subsequently, a grouting program was implemented and the inflow was reduced by 50%.
Concluding remarks
The water outflow from the tunnel after filling was higher than expected, in the order of 390l/s. Just above two years after filling, the leakage had been reduced to around 200l/s, in the order of 1% of the mean water flow in the Kihansi River.
The loss of water from the tunnel after filling was considered to consist of three components:
i) Water inflow into the ‘dry zone’ of the tunnel system, amounting to approximately 30l/s.
ii) Water outflow to terrain, observed as increase in water flow in creeks. The increase in the neighbouring creeks after filling was estimated to 100-150l/s.
iii) Filling up of a larger ground water reservoir, with higher ground water table. Two years after filling the rise of the ground water table along the tunnel was more than 40m. This component of the water loss is temporary, until the new water table has stabilised.
There are still unanswered questions regarding the hydrological behaviour of the rock mass along the pressurised tunnel. However, thorough investigations and analyses before, during and after the water filling has contributed to a much better understanding of the conditions for the design of a pressure tunnel in destressed rock.
Related Files
Figure 2b – Close up of the powerhouse complex
Figure 4 – Variations in leakages in the access and headrace tunnels
Figure 3 – Grout take in the headrace tunnel
Location map
Figure 2 – Longitudinal section
