Short-term reservoir optimization by stochastic optimization to mitigate downstream flood risks

An important objective of the operation of multi-purpose reservoirs is the mitigation of flood risks in downstream river reaches. Under the assumptions of reservoirs with finite storage volumes, a key factor for its effective use during flood events is the proper timing of detention measures. Operational flow forecasting systems support this task by providing deterministic or probabilistic inflow forecasts and decision support tools for assessing optimum release strategies. We focus on the decision support component and propose a predictive control approach to optimize the release trajectories of reservoir systems over a finite forecast horizon of several days. This approach consists of a nonlinear gradient-based optimization algorithm and simulation components based on a pool routing model and kinematic or diffusive wave models for the downstream river reaches with a simulation mode and reverse adjoint mode for the efficient computation of first-order sensitivities. The framework is currently implemented in a reservoir system operated by the Brazilian CEMIG. We discuss preliminary results of the approach for the Queimado reservoir in the River Preto and show that release reductions can efficiently balance flood peaks in downstream tributaries. A requirement for this approach is the availability of sufficient free storage. Whereas the short-term optimization of release reductions primarily depends on observed data, the hydrological model and uncertainty in both, the creation of additional temporary storage by pre-releasing water relies on longer forecasting lead times and numerical weather predictions with larger amounts of forecast uncertainty.