Optimal control of a water reservoir with expected value–variance criteria

The article presents how to solve a reservoir management problem, which has been formulated as a two‐criteria stochastic optimal control problem. Apart from the expected value of a performance index, its variance is also considered. Three approaches are described: a method based on the Lagrange function; a method based on the ordinary moment of the second order (finite time horizon); and a method based on linear programming (infinite time horizon). In the second part of the article, they are assessed in a case study concerning a reservoir in the southern part of Poland.