AI is a critical differentiator for energy storage system success

Whether in grid storage application, or to manage battery packs in EVs, AI is providing a new, powerful dimension to making batteries cheaper, more efficient and longer-lasting. In this article, Adrien Bizeray, chief data scientist and co-founder of Oxford-based Brill Power reflects on AI’s game-changing potential for clean energy.

Share on

(Previously published by Energy Storage News)

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.

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.

AI and 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)


At GroupWhistle, we help clients formulate and communicate leading thinking. If you’re at the vanguard of a technology breakthrough and want to share your ground-breaking news with the world, see our Thought Leadership services

Speak to us
GroupWhistle have a dedicated F1 studio design capability. Call us on +44 (0) 1865 727 654 or email