A Tempo-Spatially Clustered Unit Commitment with Long-Cycle Storage Dispatch

Abstract

Unit commitment that addresses the flexibility restrictions of power system operation that other routines like (optimal) power flow cannot, has been given more expectation for large-scale long-term scenarios, while its computational efficiency hinders its deployment, especially with a large number of binary variables. The typical reduction method — temporal clustering method cannot be applied to systems with long-cycle storages because of the loss of chronological information. A new framework of reformulated unit commitment composed of both temporal clustering with clustering index and unit clustering in the spatial dimension is proposed here, which aims at combining the advantages of the two clustering methods for unit commitment to suffice as the operation routine in large-scale long-term long-cycle storages equipped power system optimizations. The entire tempo-spatially clustered unit commitment with long-cycle storage dispatch is evaluated on a 118-node 54-unit system.