Why battery safety needs to catch up with electrification

As electrification expands across EV fleets, charging hubs and stationary storage, Joe Holdsworth, Founder and CEO of Metis Engineering, explores why battery monitoring must evolve beyond the conventional BMS.

The scale and diversity of battery energy storage deployment has changed in a very short period. A few years ago, the risk conversation was largely centred on consumer electronics. Now we are talking about multi-megawatt stationary storage installations, high-duty-cycle EV fleets, construction plant and charging hubs operating in environments that batteries were never originally designed for.

Each of those contexts introduces its own stress profile: thermal cycling, vibration, humidity, dust and partial charging patterns. As the installed base grows, so does the statistical likelihood of faults, and the consequences of those faults become more serious because the energy densities involved are higher and the environments are often confined or otherwise critical. Battery safety has moved from a product-level concern to an infrastructure-level one, and the industry is still catching up.

What early warning signs can indicate a developing battery fault before it becomes a thermal runaway event?

Thermal runaway rarely appears without precursors. The challenge is that those precursors are often subtle and may develop over hours or days before the event itself. Temperature deviation is the most commonly monitored signal, but relying on it alone means operators may already be fairly late in the progression.

Voltage divergence between cells, internal resistance changes and anomalous self-discharge rates can all indicate mechanical damage, contamination or early-stage dendrite formation. Beyond the electrochemical signals, gas release is one of the most useful early indicators: lithium-ion cells can begin off-gassing specific volatile organic compounds (VOCs) before temperatures reach critical thresholds.

That is why environmental sensing inside a battery pack can support earlier warning than electrochemical monitoring alone. Combining multiple signal types gives operators a higher-confidence alert and more time to respond.

Where do conventional BMS systems provide enough visibility, and where can additional environmental sensing be useful?

A well-implemented BMS is essential and does its job effectively within the cell and pack. It monitors voltage, current and temperature at the cell or module level, enforces charge and discharge limits, and manages balancing. For most standard operating conditions, that is sufficient.

The limitation of a conventional BMS is that it monitors what is happening inside the designed electrical circuit. It does not sense the atmosphere around the pack, it cannot detect a failing cell in an adjacent system, and it may not catch the early VOC gas signature of a fault that has not yet produced a measurable electrical anomaly.

In applications where packs are co-located, in confined spaces, or operating in environments with variable ambient conditions, supplementary environmental sensing can add a layer of situational awareness that the BMS architecture was not designed to provide. It is complementary rather than competitive.

How much difference can earlier detection make to incident response?

The difference between a manageable incident and a catastrophic one often comes down to minutes. Thermal runaway in a lithium-ion cell can transition from initial off-gassing to full propagation across a pack extremely rapidly once it takes hold.

If a detection system provides a reliable alert at the off-gassing stage rather than at the point of temperature exceedance, it may provide ten to thirty minutes of additional response time. For a fixed installation with suppression systems, that can be the difference between automated intervention being effective or not. For an EV fleet operator, it may be the difference between a controlled shutdown and a vehicle fire. For a construction site with a mobile battery unit, it can determine whether personnel have time to evacuate safely.

Earlier detection does not just reduce asset damage; it can make emergency response protocols more workable in practice.

What does good battery health monitoring look like for stationary storage, EV fleets, charging infrastructure or electric plant?

Across all of those applications, the common thread is continuous monitoring rather than periodic inspection, and data integration rather than isolated alerts. For stationary storage, cell-level electrical monitoring can be combined with enclosure-level environmental sensing, with all data feeding into a system that can flag trends rather than just threshold breaches.

For EV fleets, health monitoring needs to account for the operational cycle: charging, discharging, idle periods and the cumulative effects of route profiles. Battery-based charging infrastructure is an area that often receives less attention than it should; battery and charger hardware are under the repeated thermal and electrical stress of charging and discharging cycles.

For electric plant and construction equipment, the environmental context is particularly demanding, and ruggedised sensing that can tolerate vibration, dust and temperature extremes should be considered as part of the wider safety strategy.

What are the main challenges when integrating safety sensors into existing systems?

Retrofit is genuinely harder than designing monitoring in from the outset. The physical installation challenges are manageable if the sensor can be installed inside a battery pack or close to the breather port, but the integration challenge is usually about data: how does the sensor system communicate with existing BMS or fleet management platforms, and how does the operator reconcile alerts from multiple sources?

CAN bus compatibility is an important consideration because many industrial and automotive systems already communicate over CAN. A sensor that can sit natively on the CAN network and output data in a format that existing control systems can interpret can reduce the integration burden.

Power availability, enclosure ratings and the need to avoid introducing ignition risks in potentially gas-rich environments are also practical considerations that influence what hardware is viable for a given installation.

What should engineers and operators be asking when assessing battery safety risk on a project?

Start with the environment, not the battery. What are the ambient temperature and humidity ranges? Is the installation enclosed? What are the consequences of a thermal event in that specific location?

Then ask what signals your current monitoring architecture actually captures, and where the gaps are. If your BMS gives you cell-level electrical data but nothing about the atmosphere around the pack, you may have a blind spot. Ask what your detection-to-response time is under your current setup, and whether that is realistic given the thermal runaway progression characteristics of the chemistry you are using.

Finally, ask what you would need in order to demonstrate due diligence to insurers, regulators or site owners. Battery safety is increasingly a compliance and liability question, not just a technical one, and the standard of evidence required is rising alongside the scale of deployment.