Anatomy of a Dark Lull: A Data Science Framework for Quantifying Compound Renewable Risks in Australia Using Vine Copulas

This paper presents a statistical framework to quantify the risk of Dark Lull events-compound renewable energy droughts-in Australia's National Electricity Market (NEM). We modelled the joint probability of wind and solar output across five states (NSW, QLD, VIC, SA, TAS) using Regular Vine (R-Vine) Copulas. Unlike traditional correlation matrices, our model captures the asymmetric dependence between regions during extreme lows. Our analysis of 1,000 years of synthetic data reveals that standard planning assumptions underestimate the risk of simultaneous failure. We also found strong "tail dependence" between wind generators in adjacent states, meaning that when one state fails, its neighbours often fail too. Specifically, we identified drought events lasting approximately 56 hours with a return period of 10 years. This is significantly longer than the 2-to-4-hour duration of currently planned batteries. This result suggests that the current market focus on short-duration lithium-ion batteries may leave the grid exposed to supply shortfalls.