To trade or not to trade: Simultaneously optimising battery storage for arbitrage and ancillary services

Abstract

This work presents a novel methodology for determining the value a battery storage system provides while participating in a competitive frequency response market, considering uncertainties. Battery storage systems are an attractive choice for power services in low-carbon electricity grids and their optimal operation are a commonly studied matter. However, the non-deterministic nature of competitive electricity markets is often overlooked. Here, we consider these market uncertainties for a storage device providing Great Britain’s Firm Frequency Response (FFR) and arbitrage services. We use a machine learning classifier to determine the set of all possible FFR market outcomes and their associated probabilities. These are then propagated through a linear optimisation model to generate a set of possible scenarios, from which the most likely can be ascertained. Several different classifiers and bidding strategies are compared, the most suitable classifier and bidding strategies which maximise revenue whilst minimising the probability of the worse-case scenario are identified. It is found that the mean expected income is overestimated by ∼28% when uncertainties in FFR market outcomes are not considered. Providing arbitrage over a tight band can still provide significant income and does not impede on the storage’s ability to provide FFR services in real time.