Robust optimization of seasonal, day-ahead and real time operation of aggregated energy systems

This work proposes an approach for the robust operational optimization of Aggregated Energy Systems (AES) on three key time scales: seasonal, day-ahead and real-time hourly operation. The evaluation of all the three timescales is fundamental for AESs featuring seasonal storage systems and/or units (such as combined heat and power plants) with yearly-basis constraints on relevant performance indexes or emissions. The approach consists in a rolling horizon algorithm based on an Affine Adjustable Robust Optimization model for optimizing both day ahead schedule (commitment and economic dispatch) and the decision rules to adjust the real-time operation. The robust optimization model takes as input (i) the day-ahead forecasts of renewable production and energy demands with their corresponding uncertainty, (ii) past and future expected performance of the units with yearly constraints, and (iii) target end-of-the-day charge levels for the seasonal storage system. These long-term targets are estimated by optimizing the operation over representative years defined on the basis of the past measured data. The proposed methodology is tested on three real-world case studies, featuring up to four short-term uncertain parameters (energy demands and non-dispatchable generation), yearly constraints and seasonal storage. Results shows that the proposed methodology meets the yearly constraints and safely manages the seasonal storage without shortages, while always meeting the energy demands (no shedding). In addition, the cost of short- and long-term uncertainty were evaluated by comparing the results of the robust rolling horizon with other two deterministic approaches, proving a limited increase.

Automated summary

This work proposes a robust operational optimization approach for Aggregated Energy Systems (AES) on three different time scales. The approach includes a rolling horizon algorithm based on an Affine Adjustable Robust Optimization model. The proposed methodology is tested on three real-world case studies and shows promising results in meeting yearly constraints and managing seasonal storage.