Temperature-sensor dynamics and timing limits for early warning of lithium-ion battery thermal runaway
Abstract
Timely warning is important for reducing the fire and propagation risks associated with lithium-ion battery thermal runaway. However, a warning method that performs well with ideal temperature data may become too slow when real sensor and communication limitations are considered. This study analyzed 58 Main mechanically induced thermal-runaway tests, including 50 reference-positive events and a 24-event high-severity subgroup, together with a 16-event strong-event cohort originating from Sandia National Laboratories (SNL). Recorded temperature histories were modified to represent sensor response delay, measurement error, finite resolution, communication delay, packet loss, and short data interruptions. Warning performance was evaluated using a causal temperature-and-rise-rate alarm. Sensor response was the main sensing limitation: in the complete Main cohort, an additional response-time constant of 0.7 s was the largest tested value retaining at least 95% timely-warning reliability. Dispersed packet loss had little effect, but a 1 s continuous data gap near alarm confirmation reduced reliability to 70.83% and 75.0% in the two strong-event cohorts. Combined analysis identified a candidate region of response-time constant ≤0.9 s and communication/processing delay ≤0.5 s, with 95.69% nominal conservative reliability. These results show that fast sensing and uninterrupted data delivery should be designed together for reliable thermal-runaway early warning.