Gaps in Battery Data are Jeopardising Energy Transition

­The energy transition is in many ways a story of the rise of batteries. A recent International Energy Agency reportnoted that delivering COP28 climate targets will depend on our ability to scale up battery deployment, which could treble renewable energy capacity and double the pace of energy efficiency improvements by 2030. 

Stationary lithium-ion batteries will be the cornerstone of clean energy, storing surplus power to fill gaps in supply from intermittent renewable energy sources and smooth out spikes in demand. Utility-scale batteries will provide essential flexibility to renewable electricity grids and help store and integrate more clean power into networks. On the demand side, batteries will also be essential to reduce the burden on electric grids by providing backup power for everything from EV charging stations to data centres, flattening out peaks in demand. Batteries are also the core component of the Uninterrupted Power Supply systems that guarantee uptime and business continuity across mission-critical services from utilities to hospitals.

Cumulatively, this means the success of the energy transition and the continued supply of dependable power across society will hinge on the continued reliability and longevity of batteries.

Future Risks

Yet as society places ever growing demands on batteries, there are rising risks of battery degradation, fire hazards, and even failures. Batteries are highly vulnerable and failure-prone, and their performance is affected by a range of factors like temperature, charge cycles, and age. Battery Energy Storage Systems (BESS) are composed of several battery modules, containing cells connected in series and in parallel, which makes them extremely complex and difficult to manage.

There are persistent technical challengesaround keeping lithium-ion and sodium-ion batteries in a prolonged high state of charge, as required in certain industries such as data centres or utilities. Battery failure is now the leading cause of downtimefor BESS.

As society becomes more reliant on battery performance, poor battery management could affect the safety, durability, and reliability of critical applications from hospitals to the electric grid itself, while jeopardising the energy transition. For example, arbitrage strategies that involve rapid charging and discharging of BESS to store renewable energy at low prices and sell it during peak demand could quickly deplete battery capacity. Other use-cases that involve fast, unpredictable charging and discharging such as providing long-term flexibility or ancillary services for renewable grids can overtax batteries and lead to faster degradation, reducing capacity and lifecycles. In this way, poor battery management could reduce our ability to provide dependable clean power to society.

The Key Battery Blind Spot

A key cause of battery degradation and safety hazards is the inability of current methods to predict battery health and lifespan with sufficient confidence. Current Battery Management Systems can only predict state of health or state of charge with around 90% accuracy. Battery data is often captured retrospectively instead of in real-time or the data quality is too poor to allow proper analysis. This means red flags such as abnormal temperature or voltage levels are often overlooked or detected too late. Battery data is also often incomplete or unstructured which renders it unusable for data analytics and machine learning systems.

Click image to enlarge

An incomplete picture of battery conditions is masking critical risks and leading to batteries being overused, causing overheating,accelerated degradation and premature repair or replacement costs. This is creating uncertainty around everything from fire risks to maintenance and insurance costs. Inaccurate historical data also prevents operators understanding how overuse may be reducing battery lifespans, and even storing up the risk of potentially catastrophic failures. This also means operators cannot tap into the full range of potential battery services and revenue streams from energy trading and ancillary to frequency regulation and grid flexibility services. This is due to being unable to predict how each of these use-cases might affect battery performance and lifetimes. In this way, inaccurate or incomplete data is reducing the potential profitability of batteries,creating risks and uncertainties that could deter investors.

AI-Powered Batteries

The predictive power of AI has helped other industries from telecoms to retail create smarter product development, pricing, and maintenance. New innovations in AI are similarly being used to turbocharge battery safety, performance, and lifespans.

Data analytics innovations can now quality-check battery data automatically across its lifetime for everything from missing files to incorrect timestamps, transforming raw data into refined material for AI algorithms. Battery data can be processed and visualised with digital dashboards displaying everything from battery operating status to monitoring parameters. The data can be used to define the optimal patterns of usage balancing organisational goals such as discharging power during peak periods with the need to preserve battery health and extend its lifespan.

AI algorithms can then harness this quality-assured data to help predict and prevent causes of degradation or safety hazards, enabling predictive maintenance and smarter battery management. For example, AI insights are helping optimise battery use to significantly extend their lifetimes. Some AIs can now predict at installation stage how many cycles of charging and discharging batteries can tolerate without experiencing degradation. These algorithms can generate flexible lifetime forecasts that change based on the latest state of health updates or patterns of use, ensuring operators can plan for upgrades or replacements. 

Similarly, machine learning is yielding new insights on the causes of degradation or battery safety such as cell imbalances or lithium plating causing thermal runaway. Whereas conventional systems could only detect thermal runaway minutes to hours before the incident, these AI systems can predict potential thermal runaway several months in advance.

As a result, AI is opening a new window into battery performance, helping estimate the true state of health and state of charge with 98% accuracy. This could help optimise future arbitrage strategies, ancillary services, or off-take agreements to maximise profits while maintaining battery capacity, and life expectancy.  AI data could also help operators demonstrate that batteries can provide reliable future capacity, attracting fairer prices for battery storage in the capacity markets and incentivising more investments in battery storage. For example, it’s more difficult to estimate the extent to which batteries can provide reliable power than other flexible generating capacity such as gas power plants, which makes it harder to give a fair capacity value to batteries in the capacity markets. AI-driven data on long-term battery capacity could enable battery operators to prove their long-term capacity to regulators.

This could even enable battery lifespans to be predicted at installation stage and help optimise charging methods to increase performance and extend lifecycles. Ultimately, the data could be used to refine battery designs, chemistries and materials to boost performance or provide specific services such as energy trading. The data could even help find substitute materials that reduce supply chain costs. By improving end-to-end processes from design to operations, AI algorithms can drive a 30% increase in overall battery system lifetime value.

A Sea-Change in Transparency

With the clean energy transition increasingly hinging on the health of batteries, we will need to understand the complex factors governing battery health, performance, and lifetimes with greater accuracy than ever before.

Click image to enlarge

New AI innovations have the potential to bring unprecedented transparency to battery lifetimes and performance, so that everything from design to operations is powered by smart data. This could enable operators to predict the lifespan and optimal usage for each type of battery at design stage to create and operate more long-lasting, safe, and high-performance batteries.

For example, battery AI could predictively optimise arbitrage strategies requiring rapid discharging during peak demand, to improve battery capacity and reduce maintenance costs. Utilities or EV chargepoint operators could predict the impact of surges in demand on battery lifespans and create smarter demand management and battery storage. Ultimately, this could unblock the energy transition by enabling batteries to provide safe, sustainable, and dependable backup power across society.

Socomec