{"id":37053,"date":"2026-04-25T00:38:00","date_gmt":"2026-04-25T00:38:00","guid":{"rendered":"https:\/\/trumonytechs.com\/?p=37053"},"modified":"2026-04-25T00:38:53","modified_gmt":"2026-04-25T00:38:53","slug":"prevent-thermal-runaway-in-bess","status":"publish","type":"post","link":"https:\/\/www.trumonytechs.com\/sv\/prevent-thermal-runaway-in-bess\/","title":{"rendered":"Hur man f\u00f6rhindrar termisk rusning i BESS: En guide till systemteknik"},"content":{"rendered":"

Preventing thermal runaway in a Battery Energy Storage System is not one design choice. It comes from six engineering layers working together: cell selection, BMS architecture, active cooling, inter-cell barriers, fire suppression, and operational monitoring. Each layer needs real parameters, real standards, and a real service life behind it.<\/p>\n

This guide is written for BESS engineers, integrators, and procurement teams. It is built around specification decisions on grid-scale, commercial, and industrial projects. If you want the basics of what thermal runaway is and how a single cell fails, start with our primer on thermal runaway in lithium-ion batteries<\/a>. This article picks up from there. The question it answers is different: how do you specify each safety layer so that single-cell failures stay contained?<\/p>\n

The thesis is simple. Prevention in a BESS works through two goals. First, keep cells inside their safe envelope during normal service. That is the job of the BMS and thermal management. Second, contain a cell-level failure when the envelope is breached anyway. That is the job of inter-cell barriers, suppression, and cabinet design. Both goals need parameter-level specification. Neither is optional. The rest of this article walks through each layer in order of when it acts \u2014 from upstream controls, to event-time barriers, to long-term operations.<\/p>\n

One structural point worth flagging upfront. Grid BESS runs 15\u201325 years of daily cycling, often outdoors, often at high module density, inside permitting frameworks like NFPA 855, GB\/T 42288, or the EU Battery Regulation. Prevention specs have to land on both the physics and the applicable code. A spec that satisfies one but not the other does not get built.<\/p>\n

Layer 1 \u2014 Cell-Level Risk Reduction Through Selection<\/h2>\n

The first prevention layer is the cell itself. Chemistry, supplier qualification, and incoming inspection together set the baseline for every event downstream safeguards have to handle.<\/p>\n

Chemistry sets the event envelope.<\/strong> LFP now dominates new stationary storage deployments. The reason is specific to prevention work. LFP cells show thermal runaway onset around 270\u00b0C in the literature, with lower peak heat release and slower propagation than NMC. NMC cells run away earlier, in the 150\u2013200\u00b0C range, and release more energy per event. Exact values shift with cell design, state of charge, and test method. They should always be checked against the actual cell datasheet. But the ranking is consistent. For grid storage on daily partial cycling at moderate C-rates over 15 years, LFP’s thermal robustness usually beats its lower energy density. For space-constrained commercial projects with heavier cycling, the NMC-versus-LFP trade needs a real analysis of cabinet cooling capacity and propagation margin.<\/p>\n

Supplier qualification is a prevention spec, not paperwork.<\/strong> Post-incident reviews keep surfacing manufacturing root causes. Internal contamination. Separator defects. Electrode misalignment. None of these can be fully fixed by any downstream layer. Procurement specs for grid-scale projects should require, at minimum: IEC 62619 or GB\/T 36276 documentation, cell-level abuse test data under specified conditions, lot traceability for every cell shipped, and access to process audit records. A certificate without the underlying reports is weak evidence.<\/p>\n

Incoming inspection catches lot-level drift.<\/strong> Even qualified suppliers drift. Filler batches change. Separator suppliers change. Electrode coating varies. Incoming inspection for storage-grade cells should cover, at minimum: open-circuit voltage distribution, capacity sampling, and DC resistance distribution. High-count projects benefit from adding IR imaging during formation cycling. That catches self-heating outliers before cells reach module assembly.<\/p>\n

The cells that enter a module set the ceiling on everything downstream. A spec that skips qualification and inspection to save cost almost always pays for it later in warranty and incident cost.<\/p>\n

\"Heat<\/p>\n

Layer 2 \u2014 BMS Architecture for Prevention<\/h2>\n

BMS specification for BESS prevention goes beyond the cell-voltage-and-temperature scope common in EVs. Grid storage BMS also needs early-warning detection, rate-based trend analysis, and integration with the facility EMS and fire panel.<\/p>\n

Sensor density decides what the BMS can see.<\/strong> Temperature sensor count per module is a common under-spec. A module with four sensors cannot reliably tell a uniform 5\u00b0C ambient rise apart from a local 15\u00b0C hot spot on one cell cluster. Prevention-grade BMS for grid BESS typically specifies one temperature sensor per cell or per small cell group. Position matters. Sensors should cover both mid-height cell surfaces and inter-cell gaps, where early propagation signals appear. Voltage sensing at cell or parallel-group level is equally important. Internal short-circuit signals show up as small, sustained voltage divergences long before temperature rise becomes obvious.<\/p>\n

Rate-of-change detection catches events before absolute thresholds trigger.<\/strong> Conventional BMS protection works on fixed thresholds. Prevention-grade BMS adds rate-of-change detection on top. A cell rising at more than 1\u00b0C per minute while neighbours stay flat is an actionable signal, even if absolute temperature is still in range. Rate thresholds need tuning against real duty-cycle data. The commissioning phase should include a baseline exercise, not just default vendor values.<\/p>\n

Action authority and latency matter as much as detection.<\/strong> A BMS that detects a precursor in 5 seconds but waits 30 seconds for EMS coordination has given up the prevention window. Prevention-grade specs should define the BMS’s autonomous action authority directly. Which faults trigger local action? Which escalate? Latency on each path should be measured during commissioning, not read off a datasheet.<\/p>\n

Integration with gas and fire detection.<\/strong> Off-gas detection (VOCs, CO, H\u2082, CO\u2082) often shows up several minutes before any measurable temperature rise. A prevention BMS pulls gas-sensor data into the same logic stream as electrical and thermal data. Many current BESS specs now require off-gas detection as a formal safety interlock. The BMS, EMS, and fire panel should share one event-classification layer at the cabinet level, not run as independent systems.<\/p>\n

BMS capability is the most frequently under-specified prevention layer in BESS projects. That happens because it shows up as a BOM line-item, not as a real engineering spec. The right procurement question is not “does this BMS meet IEC 62619?” It is: “what is the detection latency, the sensor density, the rate alarm config, and the autonomous fault matrix?” The answer reveals whether prevention is actually being specified.<\/p>\n

Layer 3 \u2014 Thermal Management as Prevention<\/h2>\n

Keeping cells inside their safe envelope is the strongest prevention mechanism available. When all cells stay in the 20\u201335\u00b0C band across every cycling condition, the chance of chemistry-driven runaway drops to its baseline. That baseline is set by manufacturing defects and external mechanical events. Thermal management is how you hold that band for 15\u201325 years.<\/p>\n

Liquid cooling has displaced air cooling in grid BESS for real reasons.<\/strong> Liquid systems deliver about an order of magnitude more heat removal per unit volume than forced-air systems. They hold tighter cell-to-cell uniformity (often 3\u20135\u00b0C versus 8\u201315\u00b0C for comparable air systems). They also scale better as module density rises. Three parameters drive the design: cold-plate geometry matched to cell form factor, flow-rate distribution verified by CFD, and TIM specification at the cell-to-plate interface. Our broader treatment of engineered ESS cooling solutions<\/a> covers the decisions that separate cooling systems built for thermal uniformity from those built only for average-temperature control.<\/p>\n

Uniformity is the prevention metric \u2014 not average temperature.<\/strong> Commissioning reports usually capture average module temperature. But propagation risk responds to cell-to-cell variance, not the average. A module at 28\u00b0C average with a 15\u00b0C spread carries much more risk than one at 30\u00b0C average with a 3\u00b0C spread. Prevention-grade specs should include a cell-to-cell \u0394T limit. Typical targets sit at 5\u00b0C or lower. The target should be verified by measurement at commissioning, not by simulation alone.<\/p>\n

Duty cycle shapes the cooling requirement.<\/strong> A BESS running frequency regulation cycles harder than one doing capacity firming. The heat profile follows. Cooling sized for average duty fails during aggressive cycling \u2014 when cells need cooling most. Specs should be built against peak-hour cycling at peak ambient, not annual averages. This is a first-principles point addressed in industrial energy storage thermal management optimisation<\/a>. Cooling sized on the wrong duty-cycle assumption under-delivers exactly when prevention matters most.<\/p>\n

Coolant and leak containment are service-life specs.<\/strong> Glycol-water coolants degrade over time. Inhibitor depletion, pH drift, and biological growth are the common failure modes. Serious BESS projects specify coolant testing intervals (usually annual), replacement protocols, and pressure-drop monitoring for flow-path fouling. Leak detection belongs in the base spec \u2014 both active sensors and passive drip trays, not a retrofit.<\/p>\n

Thermal management specified only for average heat removal meets a cooling goal but misses the prevention goal. The difference shows up at the parameter level, and it shows up in real-world fire statistics.<\/p>\n

Layer 4 \u2014 Inter-Cell Barrier Specification<\/h2>\n

Even with good cells, a capable BMS, and enough cooling, cell-level failures happen. Sub-detection manufacturing defects. Rare mechanical events. Long-tail aging pathologies. Inter-cell barrier materials are the containment layer. They keep a single-cell event from cascading. This layer acts in a regime where everything upstream has already been exceeded. The spec bar is higher as a result.<\/p>\n

Barriers are a different material class from conventional TIMs.<\/strong> A standard TIM moves heat between a cell and a cooling plate. High conductivity is good. An inter-cell barrier slows heat transfer between adjacent cells during a runaway event. Low conductivity is good. The two have opposite conductivity targets and should be specified separately, even when they sit in the same module. The foundational material decisions behind thermal interface material selection<\/a> apply to the cell-to-cold-plate interface, not to the inter-cell gap.<\/p>\n

Three temperature tiers drive barrier specification.<\/strong><\/p>\n