Benchmarking methods for identifying representative load periods for fast distribution network analysis

Analyzing distribution system planning problems, such as hosting capacity (HC) evaluation, using high-resolution annual load profiles, is computationally expensive for researchers and operators. To address this challenge, representative periods (RPs) are employed to reduce computational demand while maintaining analytical accuracy. This paper benchmarks clustering-based and optimization-based methods to identify RPs for fast distribution network (DN) analysis. A framework is proposed to evaluate the performance of these methods using metrics that assess chronological alignment, energy approximation, and extreme load representation. The results demonstrate that clustering approaches that use dynamic time warping (DTW) for similarity measurement provide superior chronological alignment accuracy, while optimization-based methods excel in representing extreme loads. The inclusion of extreme periods further enhances the clustering performance, making it a preferred approach for HC evaluation.