Uncertainty Impact on Aggregator Performance for Peak Shaving

Accurate peak period forecasting and load consumption profile prediction are critical for utility resource planning and emergency response preparedness. This article investigates the impact of uncertainties in both peak period characteristics (timing and magnitude) and thermostatically controlled loads (TCLs) behavior on capacity estimation and subsequent peak shaving performance. Uncertainty sources associated with common TCLs and challenges with forecasting peak period parameters are discussed. Conventional TCL models are updated to address and incorporate these inaccuracies and uncertainty factors. We design and implement a management system that aggregates TCLs using a comprehensive virtual battery (VB) model, ensuring adherence to all operational constraints. An optimal peak shaving algorithm is deployed during peak demand periods utilizing this VB framework, enabling comparative analysis of “OFF” and “ON/OFF” control scenarios. A novel metric, the shared load capacity percentage, is introduced to provide insights for peak shaving optimization and broader demand response (DR) applications. Findings indicate that employing an ON/OFF control strategy can potentially result in 20% more capacity compared to OFF control, particularly when baseline energy consumption is reduced by 10%. Furthermore, we present a detailed assessment of the impact of various uncertainties on the peak shaving process, considering both utility operations and end-user loads. The sensitivity of the proposed control system to forecasting inaccuracies is thoroughly analyzed, and the expected loss of loads’ hybrid reserved energy is quantified. A metric for aggregator reserved capacity loss is introduced, giving insight to the detrimental effects of forecasting errors on peak shaving efficacy, encompassing performance, cost, and temporal implications. The results reveal that inaccurate peak time forecasting and voltage level uncertainties can lead to up to a 50% variation of peak shaving capacity with electric water heaters (EWHs) and heat pumps (HPs). The study also demonstrates that ambient temperature forecasting inaccuracies lead to a 35% capacity provision change. These findings underscore the critical need for robust forecasting techniques and uncertainty management strategies to implement TCL-based peak shaving programs effectively.