Application of CVaR risk aversion approach in the expansion and operation planning and for setting the spot price in the Brazilian hydrothermal interconnected system

Long-term hydrothermal generation planning (LTHTP) problems have been traditionally conceived as minimum cost-based optimization models. However, such policy may lead to unacceptable amounts of load curtailment in critical inflow scenarios, which are likely to be avoided. This paper describes a direct approach for the implementation of a Conditional Value-at-Risk (CVaR) version of stochastic dual dynamic programming for the LTHTP problem. We also present the main results of the validation studies for determining the values of the key parameters of the model. The proposed methodology has been officially used in Brazil since September 2013 for the following activities: operation planning and dispatch, setting the spot prices and for expansion planning.