Lithium-ion batteries have been increasingly demonstrated in reuse applications for environmental and economic reasons, and stationary energy storage systems (ESS) and mobile ESS are emerging as reuse applications for electric vehicle batteries. Most mobile ESS deployments are at large scales, necessitating experimental data on thermal runaway (TR) to ensure comprehensive safety. In this study, TR induction and suppression experiments were conducted using fully charged NCM-based batteries at the cell (750 Wh), module (7.5 kWh), and pack (74 kWh) levels. The stepwise TR experiments measured changes in temperature, voltage, heat release rate, volatile organic compound concentrations, and vent gas composition. The suppression experiments assessed the effective water injection rate, timing, and volume required to mitigate TR propagation. The results demonstrate that in the case of TR caused by thermal abuse, early detection of battery abnormalities is possible through monitoring pre-TR indicators, such as temperature and vent gas concentration. It was also confirmed that CO₂ injections can effectively cool the battery without causing damage. Furthermore, it is proposed that rapid water injection, directly contacting the battery immediately after the onset of TR, can successfully prevent TR propagation.

Anti-islanding certification is generally reported as pass or fail, although response times across the active- and reactive-power (P-Q) matrix contain additional information. We analyzed 925 Condition-A measurements from 37 grid-connected photovoltaic (PV) inverters tested under Korean Industrial Standard KS C 8565 at 60 Hz. A signed 5 × 5 map and within-product contrasts used each inverter as its own reference. Relatively long responses followed the ΔQ = 0 row rather than the balanced cell alone. Mean response was 0.221 s at balance, 0.140 s at signed |ΔQ| = 10% endpoints, and 0.208 s at signed |ΔP| = 10% endpoints. The direct P- versus Q-axis difference was 0.068 s (95% confidence interval (CI), 0.040–0.095 s; p < 0.001). Across the matrix, the ΔQ = 0 row exceeded the |ΔQ| = 10% rows by 0.061 s (95% CI, 0.036–0.085 s; p < 0.001). All values met the 0.5 s criterion. Reactive-power balance, rather than a single balanced cell, therefore marked a corridor of relatively long compliant response. This analysis converts certification records ordinarily summarized as pass/fail outcomes into a directional performance profile. The analysis procedure, but not the numerical PV response times, is proposed for future grid-connected power conversion system (PCS) testing, which requires separate charge- and discharge-mode validation.