Load Aggregation Management Strategies for Demand Response: A Dual Forecasting Approach for Cost Minimization

Pokpong Prakobkaew¹Somporn Sirisumrannukul^2*

• ¹Department of Instrumentation and Electronics Engineering, Faculty of Engineering, King Mongkut’s University of Technology North Bangkok, Bangkok, Thailand
• ²Department of Electrical and Computer Engineering, Faculty of Engineering, King Mongkut’s University of Technology North Bangkok, Bangkok, Thailand

The rapid increase in electricity demand poses significant challenges to the stability of power grids. Demand response programs provide a viable solution by incentivizing consumers to either reduce or shift their energy consumption during peak periods. Load aggregators play a critical role in the effective management of these programs by consolidating load reductions from multiple participants. However, aggregators encounter difficulties in maintaining profitability while minimizing operational costs, particularly in markets where demand response participation is still in its nascent stages. This study can be summarized into 3 parts. First, we examine three participant selection strategies for optimizing load aggregation: time-slot-based, price-based, and forecast-based approaches. Second, by utilizing Thailand's pilot demand response program as a case study, where the pool of participants is limited, we propose a dual forecasting methodology. This approach integrates short-term load profile forecasts derived from XGBoost with long-term load duration curve predictions generated by SARIMAX, thereby enhancing both accuracy and efficiency. Lastly, the results demonstrate that the dual forecasting strategy outperforms the alternative methods, leading to reduced costs and improved reliability of load reductions. This strategy proves particularly effective in optimizing the use of a limited number of participants, rendering it highly suitable for emerging demand response markets.