Proceedings of the 5th World Conference on Climate Change and Global Warming
Year: 2025
DOI:
[PDF]
Optimal Operational Planning for Power Trading under Forecasting Uncertainty
Takamasa Urasawa, Kiyofumi Sato, Yoshikuni Yoshida
ABSTRACT:
In response to the growing role of renewable energy and the evolving landscape of Japan’s power markets, this study investigates the relationship between forecasting methods and economic returns in energy arbitrage operations within the electricity market. This investigation was carried out through integrated simulations targeting the Japanese market, covering the Day-Ahead, Hour-Ahead, and Supply and Demand Adjustment markets. Three forecasting methods—the Persistence model, the Prophet model, and the VAR model—were employed in the forecasting stage. Although the Prophet model achieved higher prediction accuracy for individual market prices, the VAR model yielded superior profits by effectively capturing the interdependencies among multiple time series and reflecting the relative changes between market prices; in contrast, the Prophet model was less capable of directly handling relationships among variables. Furthermore, simulation results indicate that the optimal operational strategy does not involve charging batteries with solar-generated power, suggesting that the synergistic effect of co-installing batteries and solar panels is minimal. Economic analysis estimated payback periods of approximately 10 years for battery systems and 7.5 years for solar installations. The significance of this research lies in demonstrating to power generation companies that relying solely on forecasting accuracy in multi-market optimization may lead to suboptimal performance, and it emphasizes the need for developing integrated evaluation metrics that comprehensively assess forecasting techniques across multiple markets.
keywords: electricity market, energy arbitrage, battery storage, solar energy, forecasting method