kWh Analytics Feature: Thermal Anomalies in BESS: 75% of Sites Show Early HVAC-Related Risk Signals

In operating Battery Energy Storage Systems (BESS) assets, PowerUp’s digital twin analysis of 8 utility-scale sites totaling 596 MWh and 149 containers shows that 75% of sites experienced at least one HVAC-related thermal anomaly, while 21% of containers were affected. This corresponds to ~0.5 events per container per year. Thermal management system failures are not uncommon; Clean Energy Associates found failures in 15% of audited BESS units in a 2024 study.

HVAC issues can stem from multiple sources: software configuration errors, hardware failures (such as insufficient coolant pressure), or external factors (such as power outages). Temperature deviations reduce performance and can trigger safety shutdowns, and are likely to become more common as global temperatures fluctuate. If unmanaged, they increase system stress and contribute to thermal runaway.

Thermal Runaway Mechanisms

Lithium Iron Phosphate (LFP) chemistry is widely adopted in BESS for its safety profile. To better understand its thermal behavior, PowerUp conducted thermal‑runaway tests with CEA‑Liten to identify each cell’s onset temperature, where exothermic reactions begin, and escalation to thermal runaway may occur.

Onset temperature marks the boundary for intervention. Although LFP cells exhibit higher onset temperatures than other chemistries (Figure 1), they remain at risk of thermal runaway if temperatures continue rising. Cell-level testing shows that 90°C is the typical thermal-runaway onset temperature for LFP chemistry under adiabatic conditions (no heat exchange). Any temperature higher than 70°C requires immediate attention.

Figure 1: Temperature evolution during an overtemperature test, according to a Heat-Wait-Seek procedure in an adiabatic calorimeter; source: [PowerUp / CEA-Liten].

To address the limitations of threshold-based monitoring, PowerUp deploys a thermal digital twin of each BESS. This model combines electrochemical principles with machine learning to establish the expected BESS temperature behavior under operating conditions, including rack-level current, temperature, and cooling system setpoints.

Observed anomalies are frequent and persistent: 70% lasted more than one day, extending up to 43 days. After a short learning period, the model detects deviations from expected temperature patterns. These appear days or weeks before safety thresholds are reached, providing early visibility into cooling and thermal-management anomalies.

The model adapts to seasonal changes, operating regimes, and aging, enabling corrective intervention before shutdowns occur or thermal stress accumulates toward critical limits. Earlier visibility into these thermal deviations minimizes disruption and safety exposure, allowing asset teams to investigate cooling drift or abnormal temperature behavior while systems remain stable.

Operating further from safety limits lowers overall safety exposure while protecting availability and performance. Early detection supports risk mitigation and long-term asset economics.

The full kWh Analytics 2026 Solar Risk Assessment can be downloaded here.