Long-Term Generation Expansion Planning Model Based on Intraday and Interday Renewable Energy Uncertainty Sets

CHEN Ruijie,LU Zongxiang,QIAO Ying,SHAN Baoguo,TAN Xiandong,JI Xingpei,XU Pengfei

DOI: https://doi.org/10.13335/j.1000-3673.pst.2024.0846

2024-01-01

Abstract:Generation expansion planning (GEP) models are frequently used to find feasible solutions for power systems with high renewable energy shares. Effective planning requires appropriate variable renewable energy (VRE) uncertainty representations such as model formulation, temporal resolution, and time duration. VRE uncertainties in GEP are usually modeled using a short time duration and a high resolution via representative scenarios and stochastic approaches. As the VRE penetration increases, VRE uncertainties create challenges for power systems at different time scales, ranging from intraday net load variability to longer periods of energy surpluses and deficits. Meanwhile, the frequency and intensity of extreme weather have escalated in the context of climate change, increasing the risk of large-scale VRE generation outages. Existing VRE models in GEP often focus on a certain time scale and lack a comprehensive consideration of multi-time scales and extreme scenarios. In this article, a long-term GEP model based on interday and intraday VRE uncertainty sets that can ensure the electricity balance at different time scales is proposed. On the one hand, long-term VRE uncertainty is characterized by interday sets via trend and deviation terms, capturing the seasonal VRE trend and intermittency, respectively. On the other hand, the intraday part is composed of typical and extreme uncertainty sets, which are implemented based on outliers and centers of several clustering results. The effectiveness of the proposed model is validated on China’s power system compared to the representative day method, time series aggregation method, and daily electricity balance method. Dramatically lower load loss and carbon emissions are achieved by the proposed method under different 8760-hour VRE outputs, but cannot be met by other methods mentioned above. Although the method focuses on modeling the VRE uncertainties, electricity supply reliability and carbon emission goals can also be met under different load demand and hydropower generation scenarios.