Risk-averse medium-term hydro optimization considering provision of spinning reserves

This paper presents two algorithms for solving a medium-term hydro optimization. Considered are risk-averse operation, provision of spinning reserves as well as short-term production flexibility. Proposed is a variant of stochastic dual dynamic programming (SDDP) and stochastic dynamic programming as a benchmark. A risk measure is introduced in both methods. To deal with short-term production flexibility a decomposition of the problem into inter- and intrastage subproblems is performed. The provision of spinning reserves leads to non-convex value functions. To deal with it in SDDP a method based on Lagrangian relaxation was used which was further enhanced by locally valid cuts in order to find realistic water values.