Practical approach to SOH estimation of LifePo4 cells

Source: Tursan PPS

In current time most of the evs are equipped with LFP cells due to their inherent advantages over NMC cells. While their usage increases its also important to ensure that BMS is able to accurately manage those battery packs made from LFP cells. For doing that it is necessary that BMS is well aware about condition of each of the cells present in the battery pack, then only it can manage them over a longer period of time.

In this article we are going to discuss about the technique which can be used for accurately judging the SOH of an lfp cell with least possible errors. Most of the manufacturers suggests for doing a charge from <10 to 100% for accurately judging the SOH, but is it the only way of judging SOH? No, we will discuss an alternative way which can calculate SOH with an error of 1 to 2% max.

Voltage as a parameter for SOH estimation

This might sound confusing as voltage is not even an accurate indicator for SOC then how can it accurately specify SOH of an LFP cell. But when voltage vs cell capacity graphs were plotted during charge and discharge of an LFP cell at speeds of C/5 it showed changes in the voltage plateau with aging of the cell. 2 specific zones were discovered which showcases the difference based on SOH difference of the cell. These zones were 30 - 60% and 80 - 90% where voltage plateau was visible during both charge and discharge of LFP cells.

Source: https://doi.org/10.1016/j.energy.2016.02.163

This C rate is specifically the one which is showing the voltage plateau more clearly than other charge/discharge rates. These tests were performed on consistent basis until the SOH reached 80%, and were carried out at temperature of 25 degree Celcius. Also the DV curves exhibit certain characteristic showcasing the state of health of LFP cells. Below mentioned are the DV curves for lfp cell with different soh readings.

Source: https://doi.org/10.1016/j.energy.2016.02.163

The distance indicated by Ach0, Ach1,... or Adch0, Adch1,.... can be used to identify the soh of respective cells. When there is significant reduction in SOH of LFP cell one can simply detect the peak slope level and identify the SOH. It is as shown in the figure below.

Source: https://doi.org/10.1016/j.energy.2016.02.163

While it has also been observed that once SOH of cell reduces below 85% one can observe 2 peaks as shown below.

Source: https://doi.org/10.1016/j.energy.2016.02.163

These peaks can further help in SOH estimation for lower SOH of the cells. 

Source: https://doi.org/10.1016/j.energy.2016.02.163

Now various testings were performed based on 18 used cells taken as reference. These were the parameters of the cells considered under study.

Results of the tests are as follows.

Source: https://doi.org/10.1016/j.energy.2016.02.163

Error is SOH estimation is shown in the form of bar graph here.

Source: https://doi.org/10.1016/j.energy.2016.02.163

Conclusion

1. Complete discharge and full charging the cell is not the only method to have soh estimation of an LFP cell. 

2. This method requires quite low computational power. Hence, its very easy to implement in BMS of Electric Car.

3. Its very reliable method showing only 0.8% of average deviation.

4 It didnt exceed the 3.5% of deviation from the real SOH, and an average error of 1.5% which can be elimated by keeping 3-4% of buffer below 0% of the battery percentage.