The Operation of Lake Kariba: A Bayesian Analysis

David Rios Insua, Kazimierz A. Salewicz
Universidad Politencia de Madrid, IMES, Strobl Gruppe

Nov 30 1992

The paper describes a successful application of Bayesian decision analysis to a complex decision making problem: the operation of lake Kariba hydropower scheme. This management problem in complicated by: high uncertainty of the inflow process, multiple and conflicting objectives and influence of time on some of the parameters in the management task. Inflows to the reservoir are foretasted through dynamic linear models. Managerial preferences are modeled through a multiattribute utility function. Since the solution of the exact model is computationally too demanding, a heuristic method is applied to find a feasible control strategy. A comparison with results obtained for methods used previously to solve Lake Kariba management problem demonstrates the superiority of the methodology presented in this paper.


Resevoir operation, Bayesian decision analysis, dynamic linear models, multiattribute utility, dynamic programming, sensitivity analysis


PDF icon 1993-14.pdf