Multiobjective dynamic programing with application to a reservoir

Abstract

The analysis of water resource problems often involves noncommensurable objectives. Where this is the case, no clear‐cut optimum solution exists, but rather a whole set of technologically efficient ‘noninferior’ solutions exists. A water resources decision maker, who reflects society's preferences, chooses the single noninferior solution preferred to all the others. Multiobjective dynamic programing is introduced as an extension of dynamic programing which can provide the decision maker with the entire noninferior set of solutions as well as the explicit trade‐offs between noncommensurable objectives at all objective levels to aid him in the decision‐making process. The technique makes possible the quantitative analysis of multiple‐objective problems which would provide great difficulties for the existing methods. An example of such a problem is presented and solved by multiobjective dynamic programing. The problem involved the generation of the noninferior set and trade‐off curves for the operation of a reservoir considering three objectives.