Accelerating Benders decomposition for short-term hydropower maintenance scheduling

Maintenance of power generators is essential for reliable and efficient electricity production. Because under maintenance are typically inactive, optimal planning activities must consider the impact outages on system operation. However, in hydropower systems finding a minimum cost schedule challenging optimization problem due to uncertainty water inflows nonlinearity hydroelectricity Motivated by an industrial application problem, we formulate scheduling as two-stage stochastic program, implement parallelized Benders decomposition algorithm its solution. We obtain convex subproblems approximating production using linear inequalities indicator variables, which account nonlinear effect number active For speeding up execution our algorithm, tailor test seven techniques, including three new applications special ordered sets, presolve warm start acceleration. Given large possible configurations these acceleration illustrate statistical methods computational experiments identify best performing configuration, achieved fourfold speedup algorithm. Results setting confirm high scalability scenarios implementation.