The linear decision rule approach applied to the hydrothermal generation planning problem

Abstract

We use the linear decision rule approach to develop a model for a stochastic multi-stage generation planning problem in the Nordic region. By developing both the primal and the dual versions of the program, the loss of optimality incurred by the linear decision rule approach can be estimated. Uncertain parameters take values in an uncertainty set defined by upper and lower bounds. Alternative modelling methods for stochastic problems of comparable size and structure either suffer from the curse of dimensionality, or have to rely on unrealistic simplifying assumptions to achieve tractability. We show that the linear decision rule approach gives a good trade-off between tractability and accuracy for a stochastic generation planning problem.