Robust scheduling of building energy system under uncertainty

This paper proposes a robust scheduling strategy to manage a building energy system with solar power generation system, multi-chiller system and ice thermal energy storage under prediction uncertainty. The strategy employs a two-stage adjustable robust formulation to minimize the system operation cost, wherein a parameter is introduced to adjust the level of conservatism of the robust solution against the modeled uncertainty. Then a column and constraint generation algorithm with modified initialization strategy is adopted to solve this optimization model along with mixed-integer linear programming. Further, we evaluate the performance of the proposed strategy by hourly simulating the system operation of a practical project with Monte Carlo simulation. Numerical results show that the robust scheduling with a proper parameter can be superior to the deterministic strategy in all the studied cases. Additionally, the proposed strategy has similar results with the model-based predictive control strategy while the former only needs to be implemented once. Even in the highest load case, the relative deviation between the two strategies is less than 2%.