Abstract
This paper proposes a stochastic bilevel program for strategic bidding of a hydropower producer. The price, quantity and ramp-rate bids are considered. The uncertainty of wind power generation, variation of inflows for the hydropower producer, and demand variability are modeled through the momentmatching scenario generation technique. Using discretization the stochastic bilevel program is reformulated as a stochastic mixedinteger linear program (MILP) with disjunctive constraints. We propose a modified Benders decomposition algorithm (MBDA), which fully exploits the disjunctive structure of reformatted MILP model. More importantly, the MBDA does not require optimal tuning of disjunctive parameters and it can be efficiently parallelized. Through an illustrative 5-node example, we identify possible strategies (specific to a hydropower producer) for maximizing profit, which in turn leads to market insights. We also use the IEEE 24-node, 118-node and 300-node case studies to show how our parallelized MBDA outperforms the standard Benders decomposition algorithm. The parallelized MBDA is also compared to the state-of-the-art CPLEX solver.

This publication has 28 references indexed in Scilit: