Advantages of Stochastic Bilevel programming in the generation capacity expansion problem

This paper proposes a bilevel model to assist a generation company in making its long-term generation capacity investment decisions among a number of alternative technologies considering uncertainty regarding the investments of the other generation companies. The bilevel formulation allows for the uncoupling of investment and generation decisions, as investment decisions of the single investing generation company are taken in the upper level for maximizing its profit, and generation decisions by all companies are considered in the lower level, that represents the market via a conjectured-price response to model oligopolistic behavior. The bilevel model is formulated as an MPEC, replacing the lower level by its KKT conditions, and approximated by a MILP and resolved as such. To evaluate the usefulness of the proposed model, results from a study case are presented and the stochastic solution is compared to perfect competition and an intermediate oligopolistic market situation between perfect competition and the Cournot oligopoly. Finally some stochastic measures are calculated.