Stochastic optimization of a hydro-thermal system including network constraints

Abstract

A methodology for the optimal scheduling of hydrothermal power systems is described taking into account multiple hydro reservoir characteristics, inflow stochasticity, and transmission network represented by a linearized power flow model. The solution algorithm is based on stochastic dual dynamic programming (SDDP), which decomposes the multistage stochastic problem into several one-stage subproblems. Each subproblem corresponds to a linearized optimal power flow with additional constraints representing the hydro reservoir equations and a piecewise linear approximation of the expected future cost function. Each subproblem is solved by a customized network flow/dual simplex algorithm which takes advantage of the network characteristics of the hydro reservoirs and of the transmission system. The application of the methodology is illustrated in a case study with a Brazilian system comprising 44 hydroplants, 11 thermal plants, 463 buses and 834 circuits.<>