Exemples d'application de la gestion optimale des reservoirs à buts multiples

Abstract

Click to increase image sizeClick to decrease image sizeThe "Decision-Risk" dynamic programming method is a mathematical tool designed to complement the simulation method in order to compare the relative priorities involved in the various uses made of a reservoir. The use of "money" as a common unit for measuring all the advantages and disadvantages related to reservoir exploitation requires that economic functions be established in relation to each of the various uses made of the reservoir (fig. 1 and 2). These functions make allowance not only for the advantages gained by programmed exploitation but also for the cost of failing to meet requirements. Since it is difficult to estimate these costs directly, it is often preferable to take a "permissible probability of failure" into consideration and to adjust the failure penalty introduced into the calculation in such a manner that the exploitation results respect this probability. In the case of the Jatiluhur Dam, in Indonesia, it was necessary to determine the power output guaranteed nine years out of ten and the reservoir operating procedure, so that the monthly deficit of irrigation water, did not exceed 10 % of the requirements except during one year in ten. The failure penalties were determined by successive approximation, complementing the operating procedure optimisation model with an operation simulation model based on this procedure. The same method was used to specify the program for equipping the hydroelectric plant of the Selingue dam in Mali, so that the power house can meet the country's electricity requirements up to 1990 whilst at the same time keeping within the limitations imposed by controlled discharge release related to irrigation and shipping needs. The model which combines the determination of optimum exploitation procedure according to the "Decision-Risk" process and the simulation of the procedures obtained, is just one necessary part of an overall planning study. The model is designed mainly to specify the performance of a multi-purpose dam for a given set of characteristics and targets. Due to the fact that the calculation response is very rapid and that the results are presented in a very comprehensive form (statistical distribution of reservoir levels throughout the year, failure probabilities with respect to each target, expectation of the overall benefit) a model of this type plays a very important synthesising role in multi-disciplinary studies.