Effect of load forecasting uncertainties on the reliability of North American Bulk Power System

This paper highlights the importance of demand uncertainties and their associated risks to system planning and operations. It presents different approaches to modeling peak demand uncertainties. The work presented is part of a larger effort by the North American Electric Reliability Corporation (NERC) to demonstrate the benefits of applying a probabilistic approach to all NERC Assessment Areas to better understand emerging Bulk Power System (BPS) trends and reliability factors to promote optimum planning. NERC-wide aggregated actual and forecasted seasonal peak demands, along with their associated forecasting errors, are presented within this paper. A detailed individual assessment has been examined, applying statistical characterization and forecasting models used in a case study originally released by the Electric Reliability Council of Texas (ERCOT). For ERCOT, the results show higher forecast accuracy for the summer season compared to the winter season. This outcome is due to tail events in the peak demand normal distribution functions developed and unpredictable weather patterns. For the purpose of risk assessment validation, a Linear Predictive Coding (LPC)-based method is used to predict long-term demand. As a result, the developed distribution error has been utilized to construct scenario analysis for ERCOT's demand uncertainty. The findings of our analysis suggest the use of more event-driven outcomes, including extreme weather, unexpected load increases, and other worst-case scenarios.