Correlation-Separated Copula Modeling of Regional Renewable Generation for Accurate Prediction of Inter-Regional Transmission Line Flows

Abstract

The rapid integration of regionally distributed wind and photovoltaic (PV) generation reshapes power flows on the major transmission corridors that connect load and generation regions. Scenario-based planning of these corridors therefore depends critically on how faithfully the spatial dependence of renewable output is represented. This paper investigates whether explicitly separating and modeling the inter-regional correlation of renewable generation yields more accurate predictions of inter-regional transmission line flows. Using one year of hourly measured wind and PV capacity-factor data for the Korean mainland, three scenario-generation models are constructed and compared: a system-wide single-distribution model, a region-resolved Gaussian-copula model that preserves the empirical inter-regional correlation, and an independent per-region model. PV is modeled on daytime samples only, and all dependence is estimated on time-aligned panels so that same-hour co-movement is retained. The models drive renewable injections on a 193-bus model of the national grid, and the resulting inter-regional flow distributions are compared against a ground truth derived from the measured series using a validated DC power-transfer-distribution-factor engine. Averaged over the 48 inter-regional lines, the region-resolved copula model attains the lowest flow-distribution error, the independent model is consistently the least accurate, and the advantage of correlation modeling grows monotonically with renewable penetration. The findings are confirmed against full AC power flow on representative scenarios, and validation against measured data from two subsequent years shows that the estimated correlation structure transfers out of sample. Cast in a chance-constrained transmission-sizing framework, the accuracy gain reduces the mis-sizing of required corridor capacity by 41–59% for the studied system and, in contrast to the conventional models, produces a reinforcement plan that is at once secure and free of stranded capacity, quantifying the value of correlation-separated renewable modeling for transmission planning in high-renewable systems.