Industrial batteries used within a typical battery energy storage system (BESS) are designed to last for a certain number of cycles or years before they need to be replaced. The expected lifespan of an individual battery varies depending on the type and the manufacturer. For example, lead-acid batteries typically last less than 1,000 cycles on the grid, while lithium-ion batteries can last about 5 times longer*. Metal hydrogen batteries are gaining in popularity due to their high energy density and even longer lifespan, with warranty promises emerging of up to 30,000 cycles.**
The degradation of these batteries is a natural process that occurs over time and is influenced by various factors such as temperature, state of charge (SOC), depth of discharge (DOD), and cycling frequency as well as ambient factors such as temperature. Understanding how your batteries are going to degrade in different conditions is essential for predicting their end of life (EOL) and ensuring that they operate efficiently throughout their lifespan. Knowing the EOL then allows calculation of important economic metrics, like remaining useful life (RUL) and the net present value (NPV) of BESS assets.
Battery analytics can assist in evaluating degradation and predicting EOL by analyzing data from the battery management system (BMS) and other sources such as environmental sensors and other edge devices. By ingesting, aggregating, tagging and threading this data, battery analytics solutions can provide valuable insights into the health of the battery and predict EOL with increased accuracy. This ensures both less operational disruptions and reduced maintenance costs. There are of course other ways of doing this. The literature is replete with electrochemical models of batteries in both BESS and vehicle applications (e.g. L. De Pascali, F. Biral and S. Onori, “Aging-Aware Optimal Energy Management Control for a Parallel Hybrid Vehicle Based on Electrochemical-Degradation Dynamics,” in IEEE Transactions on Vehicular Technology, 69(10) 10868–10878 2020) as well as machine learning approaches (e.g. X. Hu, Y. Che, X. Lin and S. Onori, “Battery Health Prediction Using Fusion-Based Feature Selection and Machine Learning,” in IEEE Transactions on Transportation Electrification, 7(2) 382–398 2021). Perhaps somewhat curiously, the role of manufacturing variability is not as well studied, though still inspires some academic interest within various contexts (e.g. P. Dechent, et al. “Estimation of Li-Ion Degradation Test Sample Sizes Required to Understand Cell-to-Cell Variability” Batteries & Supercaps 4(12) 1821–1829 2021).
An important aspect of understanding battery degradation is threading battery data down to a unique identifier. Particularly when some amount of manufacturing data (or at least commissioning data) is available, this allows for a robust machine learning approach to not only monitor and predict degradation, but also more importantly to positively impact safety.
Threaded data, by nature, contains the entire operational (and in some cases manufacturing) history of the individual battery, combined with telemetry data often gathered from multiple sources. When dealing with such large amounts of data, a robust battery analytics solution can ensure that your critical data queries are the lowest latency possible, if not instantaneous, across potentially gigabytes of data. These are critical factors to consider in evaluating battery performance.
Why not just use the vendor-provided BMS readout of state-of-health (SoH) you might ask? While it’s certainly sensible to include this SoH in any analysis, many of our customers treat this parameter as somewhat uncertain and often ask us to include other metrics of health. For instance, almost all customers ask us to calculate capacity for individual charge and discharge segments using a Coulomb counting approach. This, in turn, is used as a basis for SoH measurements with the relation SoH = 100% Capacity Current (fully charged) / Capacity Rated (fully charged)). While some customers go much further using ML-derived models, they wish to keep it proprietary.
Augmenting the OEM-supplied SoH with these other metrics is necessary because batteries often perform differently in the field once deployed compared to vendor specs that are often derived from laboratory based cycler data. We all know the real world just isn’t the lab. For example, a battery may perform well in a laboratory at 25°C but may degrade much faster when exposed to higher temperatures in real-world use. Cell series resistance (Rs), for example, can increase faster in some climates compared to others.
Remote monitoring, a module within Peaxy Battery Analytics, is an important tool that can be used to better track the overall state of health of your batteries by allowing you to monitor the performance of your batteries from a distance or across geographic sites in real time and identify and address any issues that may arise. By identifying issues early on with automatic alerts, you can take corrective action before these issues become more serious and potentially lead to premature failure or safety concerns.
Peaxy Battery Analytics can ingest, normalize, feature extract and report on data from any available data channels to help you better understand how your batteries are going to degrade in different conditions, from manufacture to field deployment. It can be extended with customer specific scripts to provide critical KPIs not supplied by vendors out of the box. (See our previous issue on how custom KPI’s can be used in warranty scenarios.)
Understanding how your batteries are going to degrade in different conditions is essential for predicting EOL and ensuring that they operate efficiently throughout their lifespan. Threading battery data down to a unique identifier, remote monitoring, understanding how operating conditions impact battery performance, and solutions like Peaxy Battery Analytics are all tools that can help. Armed with such highly accurate and detailed insights, you can achieve a competitive edge in both streamlining operations and reducing costs.
* Fan, X., Liu, B., Liu, J. et al. “Battery Technologies for Grid-Level Large-Scale Electrical Energy Storage.” Trans. Tianjin Univ. 26, 92–103 2020; Dufo-López, R.; Cortés-Arcos, T.; Artal-Sevil, J.S.; Bernal-Agustín, J.L. “Comparison of Lead-Acid and Li-Ion Batteries Lifetime Prediction Models in Stand-Alone Photovoltaic Systems.” Appl. Sci. 11, 1099 2021