Hydropower Bidding Using Linear Decision Rules

Abstract

This thesis investigates the Linear Decision Rule (LDR) approach applied to the bidding problem of a Nordic hydropower producer with reservoir capacity. A stochastic programming model with piecewise LDR in the spot prices is developed. A comprehensive case study with uncertain spot prices conducted for the fall of 2012 shows that the LDR model performs equally well as a scenario based model on expectation, yet with a smaller standard deviation in the profits. The runtime of the LDR model is substantially longer than the runtime of the scenario based model. Therefore, promising techniques to reduce the runtime are developed and presented.