TRADE-OFF MODELING FOR CASCADE RESERVOIR OPERATION USING GENETIC ALGORITHM (Case Study: Cascade Reservoir Saguling, Cirata, and Djuanda)

• Main Author: (NIM 35004001); Tim Pembimbing: Prof. Dr. Ir. Hang Tuah, MOc.E., Dr. Ir. Iwan Kridasantausa, , AZMERI
• Format: Dissertations
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/14031
• Institution: Institut Teknologi Bandung

The operation of cascade reservoirs with different vision and mission as well as different management may emerge conflict of interests. As are Saguling, Cirata, and Djuanda Reservoirs starting to operate in cascade in 1998, the technical operation problems for each reservoir began to rise. To determine an appropriate pattern of operation for the three reservoirs, various softwares have been applied. The pattern of reservoir operation produced within 1988-1997 was susceptible to any dissatisfactions or complaints from one of the reservoir managements, as in the RESOP (Reservoir Operation) model developed in 1998. There was no review to the model whether or not the operation had contributed an equal service level between the energy demands and the fixed water demnads. As a result, the model did not fully serve the happening conflict of interests with solution.<p>This research is aimed at developing the model of operation; softwares that can be operated interactively (user friendly) using Genetic Algorithm (GA). The GA model is a fully applicable model for the problems of system in cascade reservoirs, in which the function of its complex goal can be achieved by inserting dependability between the energy and the release for fixed water.<p>The model development is performed by applying the function of goals to maximize the average energy and the release for fixed water. The improvement of the constraint factors applied is related to the turbine capacity, the secured reservoir and the characters of reservoirs, suddenly drawdown, and upraising. The analysis of trade-off is then performed by means of analyzing the service level for every demands of the reservoir managements, either for the energy demand or for the fixed water demand. Eventually, this may minimize the happening conflict of interest.<p>The pattern of slope sediment depends on the inflow, rule curve, and the physical condition of the reservoirs, whereas sediment of water basin depends only on the inflow condition. For the relatively small inflow in Cirata and Djuanda Reservoirs, the sediment volume is most affected by the sediment originating from the sliding reservoir slope. The sediment slope of Cirata Reservoir is 200-300% greater in a year, while The Djuanda is 2*106-3*106% in a year compared to the sediment of water basin. With a relatively greater inflow in Saguling Reservoir, the dominant sediment volume is that of the sediment of water basin and is greater reaching the value of 6 *104-7 *104% in a year compared to that of the slope sediment.<p>At the trade-off curve based on the equal service level, the service level for the demands of energy gets smaller due to the raising service level of fixed water. According to the trade-off analysis for the dry periode the service level achieved is 82,4% using GA model, and 82,2% using PNL model for the two demands. Whereas, according to the trade-off analysis in the normal periode, the service level achieved is 116,7% using GA model and 116,5% using PNL model for the two demands. Based on those analysis, the dry periode of Energy Demnad Ration (EDR) is in the safe condition, but the dry periode of Downstream Water Demand Ratio (DWDR) is in the critical condition.<p>The trade-off that is based on the equal service level for the accumulated sediment occurs in the 40th year, whereas it does not occur in the following year. Based on the trade-off analysis using GA model in dry periode, the service level achieved is 81,7% and 81,5% using the PNL model for the two demands. While the trade-off analysis in the normal periode using GA model, the service level achieved is 116,1% and 115,6% using PNL model for the two demands.<p>The service levels achieved in dry and normal periodes from the trade-off percentages using GA model may reduce 50%-54% of executing time rather than that of the PNL model. The saving of this executing time is such a superior for GA model. In addition to that, GA model has also the ability to determine the initial value performed randomly and it may lead the fitness function to the central superiority of GA model in finding the optimum global solution.<p>The improved trade-off graphics may contribute to the Decision Support System, particularly the operation of Citarum Cascade Reservoir. Then the fixed release for the three reservoirs is ascertainable in order to fulfill the demands of water in the downstream of Djuanda Reservoir and to produce the optimum energy production from the system of Citarum Cascade Reservoir as well. In addition to that, the optimum operation of cascade reservoir and the improved trade-off analysis may significantly contribute to the optimizing model as well as to the existed hydrology model. <br />