Dealing with Negative Inflows in the Long-Term Hydrothermal Scheduling Problem

The long-term hydrothermal scheduling (LTHS) problem seeks to obtain an operational policy that optimizes water resource management. most employed strategy such a is stochastic dual dynamic programming (SDDP). primary source of uncertainty in predominant hydropower systems the reservoirs inflow, usually linear time series model (TSM) based on order-p periodic autoregressive [PAR(p)] model. Although PAR(p) can represent seasonality and autocorrelation inflow datasets, negative inflows may appear during SDDP iterations, leading balance infeasibilities LTHS problem. Different from other works, focus this paper not avoiding but instead dealing with values cause infeasibilities. Hence, three strategies are discussed: (i) inclusion slack variable penalized objective function, (ii) truncation zero, (iii) optimal truncation, among which latter novel approach. compared individually combined. Methodological conditions evidence algorithm convergence presented. Out-of-sample simulations show choice significantly impacts performance resultant policy. combination reduces expected operation cost by 15%.