Battery State of Health Estimation for Uninterruptible Power Supplies

Uninterrupted power supplies are an increasingly popular technology that are employed to protect a load, which can range from a simple workstation or an entire building, from grid failures. Good maintenance of these systems is important, and it involves the monitoring of the state of health (SOH) of its batteries. The calculation of the SOH of the batteries is not easy in UPS applications due to their limitations, as we can’t just discharge the batteries to test their capacity since they need to be ready to act as a power source for the load if the grid is disconnected, therefore alternative ways to estimate it should be studied. Different methods described by existing papers were considered for this experiment: coup de fouet, relaxation curves and impedance spectrum. The first involves a very well documented phenomenon that happens right at the beginning of the discharge, meaning the capacity lost is minimal, while the second observe the behaviour of the open circuit voltage after a recharge, requiring no discharge whatsoever, but it has much less research backing it up regarding lead-acid batteries. Impedance analysis employs EIS (electrochemical impedance spectroscopy), a non-intrusive procedure, to gain information on the impedance spectra of the batteries, from which certain parameter can be extracted to gain information on its SOH. Adequate measurements UPS batteries were taken for each method in other to carry out the necessary observations. The results were overall mixed. The analysis on the coup de fouet confirmed the method to be a simple and effective to calculate way to estimate SOH with a simple voltage measurement through a linear relationship. Positive results were also found with the impedance spectra, both looking at single frequency resistance measurements and with full spectrum parameterization. On the other hand, the study of the relaxation behaviour was not fruitful, and no correlation between the behaviour of the battery open circuit voltage (OCV) and SOH was found. The tests also highlighted some shortcomings in these methods. In particular, for the relaxation curve observation, some improvements to the measurement procedure are pointed out that could be taken into consideration to try and find a correlation between OCV and SOH. For the EIS method, it is acknowledged that there is still a lot we do not know about what influences the shape of the impedance, which makes it harder to interpret it correctly.