As energy storage devices become more important in the management of the world’s power supply, safeguarding large battery assemblies against catastrophic failure (CF) has become a top priority. Although lithium-ion batteries are found in a wide array of applications, from mobile phones to commercial airliners, the continued expansion of lithium-ion batteries is hindered by safety, durability, and reliability concerns. One of the reasons for CF is battery internal short-circuit (SC), which causes thermal runaway. Unfortunately, no fast industry-accepted method to forecast CF exists. Recently, we proposed a new electrochemical approach to detect battery SC far before CF can occur.1 The approach is based on the application of a constant voltage (VTEST) to a short-circuited cell (battery) at a slight discharge overvoltage from fractions of mV to several mV, depending on the battery system and desired speed of SC detection. Under this slight polarization, causing minimal battery disturbance, the response signal (measured current) transitions from an initial high negative value to a stabilized positive value through a current zero crossing point. A stabilized, positive current under this potentiostatic condition is a metric for SC detection and is SC current at absolute value in the case when the cell’s genuine self-discharge is negligible. For a short-circuited cell with notable genuine self-discharge, the stabilized positive current is a sum of two currents: genuine self-discharge current and SC. The current zero crossing point time is another useful and practical metric for SC detection. This method was validated on rested, single 18650 Li-ion cells with shorts of different severity and demonstrated high precision. Lower-resistance shorts result in shorter time to current zero crossing point and higher SC current. Diagnostics at field applications, such as electric drive vehicles, require fast determination of SCs in battery architectures consisting of numerous cells in serial-parallel connections. In order to determine the usefulness of the method, we tested a dozen SC scenarios for one-string, consisting of three 18650s cells, and dual-string, consisting of six cells. In order to mimic field scenarios, the study was conducted on resting architecture and when the architecture was connected to the external known resistive load. In the field, the load is battery-controlling circuitry, such as battery management system, with vehicle equipment connected to the battery. It is assumed that a vehicle is parked, not driven, during this type of diagnostic. The SC resistor was applied across a single cell or several single cells randomly positioned in battery architectures, while potentiostat was connected to short-circuited cell, not-short-circuited cell, and a string or double-string containing short-circuited and not-short-circuited cells. If a potentiostat is connected to a healthy not-short-circuited cell in the battery architecture consisting of other short-circuited cells, there is no SC current running in the diagnostic circuit. In contrast, short-circuited cells produce SC currents detectable on all levels of architecture. Using this methodology, it is feasible to first detect problems on the full battery architecture level, then problematic string(s) can be found and, in the string, problematic cells with high self-discharge (and potential shorts) can be identified. This method appears to be a universal diagnostic tool used for a battery over its life span. Initially, it can be used for the detection of faulty cells under production, and then it can be applied at the beginning of life during initial characterization (mapping) of battery performance. It can also be used for periodic state-of-health checks during the operational life of the battery in comparison to the beginning of life. When the battery reaches the end of life, which is 80% of state-of-charge for electric-drive vehicle applications, it may be repurposed for secondary use in residential or industrial storage. This is the time for checking for shorts to avoid problems in larger battery architectures. Lastly, the approach can be used for first responders on battery-related accidents. This patent-pending method is accurate, fast, non-invasive, and applicable to any battery chemistry or design with capability to detect nascent shorts. The method can be adapted for battery management systems for monitoring battery state-of-health at any time and in any state of charge. The technology based on this method can be used in electric-drive vehicles; stationary energy storage; military, aeronautic, and portable electronic devices; and many other applications.
Read full abstract