Chronological Time-Period Clustering for Optimal Capacity Expansion Planning

Abstract

To reduce the computational burden of capacity expansion models, power system operations are commonly accounted for in these models using representative time periods of the planning horizon such as hours, days or weeks. However, the validity of these time-period aggregation approaches to determine the capacity expansion plan of future power systems is arguable, as they fail to capture properly the mid-terms dynamics of renewable power generation and to model accurately the operation of electricity storage. In this paper we propose a new time-period clustering method that overcomes the aforementioned drawbacks by maintaining the chronology of the input time series throughout the whole planning horizon. Thus, the proposed method can correctly assess the economic value of combining renewable power generation with interday storage devices. Numerical results from a test case based on the European electricity network show that our method provides more efficient capacity expansion plans than existing methods while requiring similar computational needs.