Short/Mid-Term Hydrothermal Dispatch and Spot Pricing for Large-Scale Systems-the Case of Brazil

The weekly hydrothermal dispatch and spot pricing of the Brazilian power system are officially obtained using a large-scale stochastic programming model called DECOMP. This optimization tool is capable of representing many aspects of the entire integrated generation system, such as: interconnected system areas, power demand per area divided in load bocks; stochastic water inflow scenarios; anticipated dispatch of some thermal plants; an accurate head-dependent hydro production function; pumped storage plants; system-wide, electrical, hydro and thermal operation constraints, among others. This paper describes in detail the mathematical aspects of the model, as well as the implementation of the dual dynamic programming solving strategy for this real-life application. We also show dispatch results, analyses of the accuracy in representing some nonlinear aspects, as well as convergence results and CPU time efficiency for an official study with the entire Brazilian system.