Introduction of ensemble based forecasts to the electricity dispatch simulator SimSEE

In the context of the Uruguayan electricity market, the Administration of the Electricity Market manages the economic electricity dispatch using the software SimSEE for optimization and simulation of the optimal operation. The Uruguayan electricity mix includes a majoritary portion of time variable renewable sources, namely hydropower, wind and solar.Inside SimSEE, stochastic synthetic realizations of diverse variables such as wind velocity, solar irradiance, electricity demand and water inflows to hydropower centrals, are generated and integrated to the simulation by means of CEGH models. A CEGH model is built of two components. One is a non-linear mapping transformtion that link the space of the represented set of signals to a corresponding Gaussian Space. The other is, a linear system that represent the correlation between signals in the present with their previous states.The assimilation of forecast information is naturally done handling the inputs of the linear system in the Gaussian Space by settings biases and noise amplitude tuners per signal channel at each time step. This work shows a new methodology to optimally compute the biases and noise amplitude tuners, using maximum likelihood optimization, for assimilation of sets of ensemble forecast. An example of the developed method, applied to the assimilation of the forecast ensemble of water inflows to the hydroelectric dams of the Uruguayan system, already in use at ADME, is shown.