Reservoir drawdown and sluicing
Modify operating rule (focus or redistribute sediment)
The 144 MW Kali Gandaki run-of-river project was designed to manage the extremely high sediment load of the Kali Gandaki river. However, excessive turbine abrasion and rising flood levels in the upper reach of the reservoir are the two major challenges.
The Kali Gandaki project began operations in 2002. The project’s concrete gravity dam was designed with radial crest gates for operation in a sluicing mode to sustain about 3.5 Mm3 of storage capacity for power peaking. The general layout of the project’s headworks is seen in figure 1.
To minimise sediment entrainment during sluicing, the intake was designed as an overspill weir (figure 2). A desanding basin is also operated during the sluicing period.
During major flood events (> 2000 m3/s), the power plant is taken out of operation and the gates are opened to pass the flood through the reservoir at maximum velocity.
The operational mode is as follows:
inflow > 2x design power - sluicing mode (summer monsoon)
inflow <= 2x design power - impounding mode (winter)
inflow > 2000 M3/s - flood- stop power production, open spillway gates
The Kali Gandaki River originates in the high Himalayas carrying high sediment load. The river generates a suspended sediment load of 43 Mt/yr, of which around 25 per cent consists of sand. This sand has a high concentration of highly abrasive angular quartz. About 95 per cent of this suspended sediment load is delivered during the monsoon, between late May and late September and is large enough to completely fill the reservoir in a single monsoon season.
Data on discharge, suspended sand concentration in the river, and on the flow diverted into the turbines is shown in figure 3. The suspended sand concentration in the river and delivered to the turbines suddenly spikes in early June. This corresponds to the date that the reservoir level is lowered, thereby mobilising sand. The sand concentration drops again when the reservoir level is brought back up to its impounding level, reducing both the flow velocity through the reservoir and the rate of sand transport. This sluicing procedure has nearly stabilised reservoir capacity, producing a sediment balance across the reservoir (see figure 4).
As designed, the system has operated satisfactorily, and the plant has produced power on an almost continuous basis since construction. Although the project design initially contemplated periods of reservoir flushing, this was found not to be necessary. There are however two remaining sediment related problems at Kali Gandaki.
The first is that turbine abrasion is excessive. Damage to turbine runners was attributed to a combination of sand abrasion plus cavitation (figure 5). This is being dealt with through a reconfiguration of the intake weir to provide a uniform crest elevation of 516 m (as shown in figure 2) and improved geometry at the intake and desander inlet, based on physical model testing.
Secondly, flood levels are increasing in the upper reach of the reservoir due to the slow accumulation of coarse sediment. One-dimensional numerical sediment transport modelling indicated that there was no feasible modification to the operating rule that would reverse this process, given the increased river base level created by the dam.
A higher capacity trash rake will also be installed to minimise blockages during periods of high trash and debris loads.
To address sand abrasion, field tests and physical modelling were undertaken (figure 6) to determine how the efficiency of the intake and desander could be improved. It was also envisaged to improve the debris-shedding characteristics of the intake, since the trash rack was becoming overloaded by debris.
The as-built intake structure geometry contained a dog-leg section, which differed significantly from the straight intake wall originally specified. Model testing focused on identifying modifications which would optimise intake operation, while constructing on top of the existing dog-leg foundation. A modification to the inlet configuration of the desander basins was also identified, which would reduce the effect of hydraulic short-circuiting due to plunging of sediment-laden flow at the entrance to the basins.
An analysis of operational data was also undertaken, and it was found that water levels were not always being held at prescribed levels. Better operational control and operator training and monitoring were identified as essential factors for achieving the most efficient operation from the installed sediment management infrastructure.