Price-Based Unit Commitment Electricity Storage Arbitrage with Piecewise Linear Price-Effects
Price-Based Unit Commitment Electricity Storage Arbitrage with Piecewise Linear Price-Effects
复制标题
具有分段线性价格效应的基于价格的单位承诺电力存储套利
DOI：
--
复制
发表时间：
2016
期刊：
影响因子：
--
通讯作者：
R. Belmans
中科院分区：
文献类型：
--
作者：
T. Brijs;F. Geth;Sauleh Siddiqui;B. Hobbs;R. Belmans
电力储存工厂可以用于许多应用，其中研究最多的应用之一是在日前市场套利。虽然套利价值与价差的存在有关，但它也取决于（反）收费行为对价格的影响，因为套利通常通过在收费时增加非峰值价格和在放电时减少峰值价格来减少价差。因此，在基于价格的单位承诺套利模型中有两个重要的假设：第一，是否假设存储运营商对未来价格有完美的了解，第二，他们是否认识到他们的（不）收费行为可能会影响这些价格，即定价或定价假设。本文提出了包含详细操作约束的套利问题的综合表述，并通过考虑现实世界的价格效应数据来放松价格承担假设，这些数据以每小时分段的基于提交投标的数量和价格之间的线性关系的形式发布，称为“市场弹性函数”。这些可以用于(1)评估基于简化价格效应的定价和定价假设，以及(2)在假设价格和价格效应在决策阶段已知的情况下提供套利价值的上限。此外，为了减少计算时间，提出了分段线性函数的逐步逼近，即从混合整数非凸二次规划到混合整数线性规划，同时提供了套利值的下界和上界逼近。将所建立的模型应用于2014年比利时日前市场，结果表明，价格效应对大型储能系统的运行和套利价值有很强的影响。
Electricity storage plants can be used for many applications, with one of the most studied applications being arbitrage in the day-ahead market. Although the arbitrage value is related to the presence of price spreads, it also depends on the effect of (dis)charge actions on prices, as arbitrage generally reduces price spreads by increasing off-peak prices when charging and decreasing peak prices when discharging. As such, there are two important assumptions in price-based unit commitment arbitrage models: first, whether the storage operator is assumed to have perfect knowledge of future prices, and second, whether they recognize that their (dis)charge actions may affect those prices, i.e., the price-taking or price- making assumption. This article proposes a comprehensive formulation of the arbitrage problem including detailed operating constraints, and focuses on relaxing the price-taking assumption by considering real-world price-effect data, published in the form of hourly piecewise linear relationships between quantity and price based on submitted bids, which are referred to as “market resilience functions". These can be used to (1) evaluate the price-taking and price-making assumptions based on simplified price-effects, and to (2) provide an upper limit to the arbitrage value under the assumption that prices and price-effects are known at the decision stage. In addition, a stepwise approximation to the piece- wise linear functions is developed to reduce computation time, i.e., from mixed-integer nonconvex quadratic programming to mixed-integer linear programming, while providing lower- and upper bound approximations to the arbitrage value. The developed models are applied to the Belgian day-ahead market for 2014, and show that the price-effect has a strong impact on the operation and arbitrage value of large-scale storage.