Software / AI

Hazard & Defect Detection Identifies lithium plating during operation

Improves the detection of temperature anomalies and its risk factors

Lithium plating is one of the most detrimental aging phenomena in batteries, as it can lead to short circuits and in the worst case thermal runaway. High currents at high cell voltage are known conditions to increase this risk (as well as low temperature). Lithium plating can be identified through a change in the voltage profile during charging (figure 1. An additional peak can be found when differentiating the data in incremental capacity analysis (figure 2. AI algorithms can be applied to help identify this peak during operation, or the change in resistance resulting from the lithium plating reaction.

The battery temperature is crucial to consider when evaluating the safety of a battery asset. High temperature can be a sign of abnormal behavior and in the worst-case lead to thermal runaway. Battery analytics can improve the detection of temperature deviation and resolve the underlying problem.

Fast charge increase the risk of lithium plating Changes in voltage and resistance identified AI algorithms improve detection

Figure 1. Cell voltage during charging

Figure 2. Incremental capacity analysis on cell voltage during charging

Example 1 Cooling system failure The thermal management system HVAC for heating, ventilation and air conditioning) is responsible for keeping the batteries at a good temperature for safety and decreased aging. While temperature variations are expected, as a result of the battery operation, this example shows a HVAC failure leading to unexpected temperature rise. A combination of an electrothermal model with machine learning enables early and accurate detection of these incidents.

Example 2 High resistance outlier Monitoring resistance over all racks in an asset enables the detection of resistance outliers. In this example a high resistance outlier was identified, also resulting in an extreme temperature rise. A deeper analysis found an electrical connection fault to be the reason for the abnormal behavior.

Temperature profile comparing container with and without management issues

Individual rack DC resistance evolution

Source: Chen et al. “Operando detection of Li plating during fast charging of Li-ion batteries using incremental capacity analysisˮ, Zhu et al. “Intelligent lithium plating detection and prediction method for Li-ion batteries based on random forest modelˮ, Safeguarding Battery Storage: AIDriven HVAC Anomaly Detection - PowerUp, Impact of anomalies on uptime and safety - PowerUp

Extracted from the PDF; charts and images may read imperfectly.