Thermal

First-Pass Fire Tests for Battery Storage Developers

Independent analysis by Starro. Not client work.

A battery storage container cutaway with one warm module held apart from its neighbors

Objective

Barrier Required to Stop Cell-to-Cell Propagation

Large-scale fire testing is now required for stationary battery systems in the United States, and a failed test costs a unit, a redesign and months of schedule. The first design question is whether one cell's thermal runaway reaches its neighbor.

How much barrier between large LFP cells stops a single cell failure from spreading, and how much margin does it leave?

Findings

About 1.6 mm of Aerogel Keeps the Neighbor Below Venting

With cells separated only by a 0.1 mm film, the neighbor reaches runaway 3.4 minutes after the failed cell. With convective edges, a barrier of 0.020 m²·K/W (about 0.6 mm of aerogel) stops runaway, 0.052 m²·K/W (1.6 mm) keeps the neighbor below 145 °C, and 0.096 m²·K/W (2.9 mm) keeps it below 120 °C.

Edge heat loss is the dominant uncertainty. With adiabatic edges, a bound no real module reaches, 2.5 mm is needed to stop runaway and 7.3 mm to stay below venting. With 2 mm of aerogel the neighbor peaks at 135 °C more than an hour after the event.

Peak neighbor temperature against barrier thermal resistance for three edge-loss assumptions, with aerogel thickness on the upper axis.

Approach

Transient Conduction Model of a 280 Ah LFP Cell Stack

A face-to-face stack of 280 Ah lithium iron phosphate prismatic cells, modeled with transient finite-volume conduction. The failed cell releases the heat implied by its measured rise, 1.46 MJ over 300 s, and a neighbor goes into runaway when any point reaches 180 °C, the lower end of the measured range for this cell.

This study scales up our companion battery module study, where 1.5 mm aerogel sleeves held the hottest neighbor cell at 84 °C, with the vent gas ducted away, instead of 224 °C with potting alone. Barrier resistance and edge heat loss are swept.

Section through the modeled stack: the failed cell, its neighbor and a third cell joined through barrier pads, with heat loss from the cell edges.

Validation

Energy Balance, Contact Solution and Convergence Checks

How we checked the model:

  • All the heat the failed cell releases is accounted for, to 1 part in 10¹⁴.

  • An isolated cell reaches exactly the energy-balance temperature, 424.9 °C.

  • Two touching cells match the error-function solution within 0.032 K on a 77.5 K rise.

  • Refining the grid and time step eightfold changes the neighbor peak by 0.1 K.

  • Cell data: Jia et al., Applied Energy 327 (2022).

Recommendations

Specify Barrier Resistance and Design for Edge Cooling

Specify the barrier by its insulating value at high temperature, not by thickness alone, and design the module to let heat out around the cells. Duct the vent gas away. A model like this lets a team judge a new cell supplier or a thinner pad in days and walk into the certification test expecting to pass.

Tools used:

Transient finite-volume conduction model; exact energy-balance and contact-solution checks