Various methods are used to incorporate electricity price data into optimisation models for techno-economic assessments of energy systems. While many studies rely on time-series electricity prices, they often optimise the system over an entire year to reduce computational complexity — an assumption that may not reflect practical decision-making conditions. This study evaluates the validity of that assumption by comparing three modelling approaches: the Basic model, which uses a full-year electricity price dataset; the Dynamic model, which operates with a 36-hour rolling window of known prices; and the Adaptive model, which augments the rolling window with a 12-hour AI-based forecast. The case study examines a hot water tank integrated into a Finnish district heating system, tested across storage capacities from 100 to 20,000 m3 and multiple heat demand scenarios.
Results indicate that compared to the Basic model, the Dynamic and Adaptive models estimate lower annual profits. At optimal storage sizes, profit deviations remain below 4% for the Adaptive model and under 7% for the Dynamic model. However, for large storage volumes and low demand, deviations increase to 40% and 70%, respectively. The gap between the Adaptive and Dynamic models also grows with storage size, reaching up to 17%. Rolling-window models require significantly more computation time - approximately 14 and 25 times longer than the Basic model.
These findings emphasise the importance of choosing a modelling approach aligned with study goals. Simplified models suit early-stage assessments, whereas for more serious evaluations, the implementation of rolling-window approaches is advised, given the volatility of current electricity markets.
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号