Predictive power dispatch using imperfect generation forecasts of variable

This thesis investigates the impact of high shares of variable renewable energy systems in an electricity grid regulated by a model predictive power dispatch controller. The main goal is to assess the variations in the operating reserve capacity in order to accommodate the stochastic variability of wind and solar generation plants. In order to achieve this result, a literature study for this particular field is presented and commented. This was necessary to frame the conditions and the aspects to consider in the modelling process: a concatenated MPC approach was chosen, divided in two different time-scale layers. In the first one, the power grid dispatch is created every hour in order to optimize the ecological usage of the storage in function to the two-day-ahead forecasts. In the second one, grid stability is ensured in face of the sudden changes in renewable power production. This implementation needed the development of a forecast generator tool completely scalable and adaptable for both solar and wind infeed. Results indicates that the strongest impact take place when the shares of RES in the electricity grid are equal or under the total demand capacity. The regulation needs tend to stabilize and even decrease with larger shares because forecast errors cease to play a decisive role. This means that during the transition phase large investments in the grid are required to ensure stability of operation, while after the threshold of 100% of covered demand the control problematic almost disappear, reducing the effectiveness of the regulation means installed.