An efficient Monte Carlo algorithm in probabilistic operational planning

Renewable energies are taking a growing part in electricity generation. The uncertainty on their generation forecasts, due to their weather dependency, as well as the inability to control their generation, makes them challenging for the grid. The deterministic approach used so far in operational planning does not allow taking these uncertainties into account. There is thus a great need to switch from the deterministic approach to a probabilistic one. The latter aims to consider the distribution of the errors on forecasts of the continuous variables, such as the wind speed and the load, as well as the probability of failure of the grid elements. In previous works, the reliability of the grid, i.e. the performance of a grid plan a given time ahead real time, is evaluated thanks to several risk indicators in particular the Mean Cost of Operations (MCO). This indicator can be easily computed by Monte Carlo (MC) sampling, but an analog algorithm turns out to be way too time-consuming, considering the time constraints in operational planning. This work aims to implement a more effective way to compute the MCO, which allows to cover all significant contingencies of the grid as well as the possible variations of the continuous variables. This efficient MC algorithm is illustrated on an adaptation of the 69-bus IEEE test case.