Robust optimal design of building cooling sources considering the uncertainty and cross-correlation of demand and source.

• A Bi-Bin method is proposed to reduce uncertainty scenarios and applied in the robust model. • The robust method can be solved efficiently under thousands of uncertainty scenarios. • The uncertainties and cross-correlations of the demand and source are investigated. • The deterministic and the proposed robust optimization models are conducted and compared. • Two schemes are conducted to test the proposed robust optimization model. The demand and renewable energy uncertainties introduce significant complexity to the design of renewable cooling sources. Conventional design method may introduce a risk of suboptimal solution due to overlooking the multiple uncertainties. The stochastic programming model is widely used for optimal design of building cooling sources, while it may encounter computational burden when considering a large number of uncertainty scenarios. Therefore, it is necessary to develop a computational model for design of renewable cooling sources, which can account for multiple uncertainties and can be solved efficiently. This paper proposes a robust optimization model for sizing a renewable building cooling source when there are cooling demand and renewable energy uncertainties. A scenario reduction method, named the Bi-Bin method, is proposed to improve computational efficiency. To illustrate the model's application, the design of a renewable cooling source for a factory building is investigated. The results indicated that the robust model for the design of a building cooling source with 3000 uncertainty scenarios could be solved efficiently in 372 s. The equivalent annual cost optimized by the robust model was slightly higher, at 1.5%, than that optimized by the deterministic model; however, the renewable cooling source optimized by the deterministic model could be unreliable considering multiple uncertainties, with a the maximum failure rate up to 3.9% and a maximum shortage of cooling energy of 595 kW. The results showed that the proposed approach can result in an economical and reliable cooling source under demand and renewable energy uncertainties. [ABSTRACT FROM AUTHOR]