Aggregated inflows on stochastic dynamic programming for long term hydropower scheduling

This paper aims to present and analyze a different approach for long term hydropower scheduling. In opposition to the Markovian stochastic dynamic programming, where monthly inflows are modeled according to probability distribution functions conditioned to some occurrence of inflow in the previous month, in the proposed approach inflows are aggregated in groups of k months to establish the Markovian modelling. Initial tests were conducted on hypothetical singlereservoirs hydrothermal systems based on four real Brazilian hydro plants with distinct hydrological regimes. The performance of both regular and proposed methods was evaluated through simulation using the historical data available in Brazil, between January 1931 and December 2012. The results show that performance of both methods are very similar comparing mean spillage and mean power generation but with lower costs for the proposed method, with differences surpassing 1% in some cases.