Market-ready artificial intelligence (AI) is a key feature of battery management to deliver sustainable revenues for a more competitive renewables market, writes Dr Adrien Bizeray of Brill Power.
With 2GW of renewable power having come online in the UK in 2023 and the pipeline expected to deliver yet more capacity in the short term, system operators might be forgiven for concerning themselves with increasing supply at a time when wholesale electricity prices are softening.
Enjoy 12 months of exclusive analysis
- Regular insight and analysis of the industry’s biggest developments
- In-depth interviews with the industry’s leading figures
- Annual digital subscription to the PV Tech Power journal
- Discounts on Solar Media’s portfolio of events, in-person and virtual
Or continue reading this article for free
But there are significant technical efficiencies in battery systems that have yet to be realised by most operators. And these efficiencies provide the solution to unlock very positive commercial viability despite challenges presented by supply-side capacity increases and reduced prices.
I believe the answer points to the value that AI can bring to the market. What’s more, my conviction is not based on some far-off expectation of the benefits of artificial intelligence, but on market-ready technology that has the potential to significantly reduce the investment quantum of new energy storage systems as well as optimise operational revenues.
Market maturity
I spent my research years investigating battery management at Oxford University and helped to found Brill Power to tackle what I considered to be profligacy in the way grid storage batteries have been specified and managed.
My colleagues at Brill have developed hardware, firmware and data analytics to solve the problem of oversizing batteries in the system design phase, by enabling up to 60% longer cycle life for batteries.
Together, we have more recently turned our attention to AI solutions to tackle the sub-optimal operation of systems during their lifetime of operation, thereby providing operators with both a reduction in capital asset costs as well as an increase in operational revenues.
While the market in its emergent phase may have provided sufficient financial insulation for inefficient energy storage solutions for renewable power, maturing market conditions have now thrust system optimisation to the forefront of operators’ considerations.
I believe that AI, if appropriately embraced, is well-placed to ensure continuing financial appeal to those interested in funding new renewables capacity.
Data helps, but AI is the defining feature
As storage capacity has exponentially increased, so the industry has started to collect an ever-increasing volume of data related to the cycling of battery systems and the performance of battery packs at discrete cell level.
That data quantum has already surpassed the levels of information that can be meaningfully interrogated by manual means, so almost by default, automation of data processing has become a necessity. But to achieve what?
There are two levels of application where machine learning and AI tools can help.
At the first level, there is the assessment of multiple sources and types of information to generate useful customer insights, for instance, battery degradation analytics, system lifetime projections, operational anomaly detection and so forth.
This is the insight that enables informed human decision-making as to how best to run a storage system.
The step on from this is the wholesale automation of the system management where essentially we take the human out of the loop and AI collects the data, makes the optimisation decisions and then ultimately executes changes to the system to deliver the most efficient operation of the asset.
For instance, in a scenario where battery management is tightly integrated with cloud analytics, an AI agent can detect drifts in state of charge or state of health accuracy computed by the local battery management and automatically trigger a cloud optimisation algorithm to re-parametrise battery mathematical models that underpin the SOX estimation; the new optimal models are then automatically updated back on the local BMS, thereby unlocking any curtailed capacity arising from SOX inaccuracies.
This is an actionable reality for today’s system operators.
Even better still
The opportunities for driving efficiencies into stationery storage systems are exponential. Once AI is executing changes to optimise systems operation, a feedback loop allows the code to self-learn and ultimately continuously improve operational parameters.
And small incremental improvements, for instance, a 1% increase in available energy across a 1-hour, 100MW system, could increase annual revenue by £51k* (US$65,340), representing a significant financial yield over the lifetime of BESS.
Brill’s analytics, coupled with AI can provide the context for seeking increased yields from systems relative to possible asset degradation. In a market with increasingly tight margins, using AI to understand the cost of cycling, future degradation and increased availability provides a holistic view for asset owners and investors to establish a model for asset management that provides tailored financial returns and greater insight to inform contract negotiations.
This greater level of transparency is key to stimulating more investment liquidity in the sector, not just from a revenue perspective from the current generation of systems, but also to better inform the capital decision-making for future system sizing decisions.
Right-sized and well run
Between the readiness of AI to help drive welcome new margins into stationery energy systems and the 60% longer battery life delivered by Brill hardware, there is a verified 30% reduction in system lifetime costs.
Essentially, everyone either oversizes battery provision at the start of a systems commissioning or augments systems through life to provision for degradation. We minimise the requirement for that through our patented active loading system that makes systems last longer.
This reduces the capital outlay from the get-go. The AI then provides the operational optimisation and together these combined benefits deliver the 30% reduction in cost – or seen another way, a significant increase in revenue potential.
*Modo Energy; based on GB BESS revenues (excl. capacity market)
Read Adrien Bizeray’s co-authored 2021 technical feature article, ‘How to design a BMS, the brain of a battery storage system,’ with your ESN Premium subscription, or read an exclusive extract here on Energy-Storage.news.
About the Author
Adrien Bizeray DPhil, is chief data scientist and co-founder at Brill Power, a startup aiming to make batteries smarter, cleaner, safer and longer-lasting as a critical element of the shift to zero emissions energy, through products including its Battery Intelligence Platform. Bizeray holds an MSc degree from Imperial College London and his PhD was funded by Samsung for the development of physics-based battery models.