Charge behavior

The analysis in the previous sections was done during a constant discharge current. In a PHEV or an EV, it is more desirable to have the capacity estimation performed during charge. This is because the discharge behavior is very dynamic while the charge one is usually a constant one. Discharging the cells in a PHEV or an EV depends on the driving pattern while charging is only done by connecting the cells into a power source. Thus to investigate whether the cells exhibit the same ICF behavior in charge and discharge, two charge/discharge experiments were performed on two different cells with different degraded capacities. Cell 1 has a capacity of 4.47 Ah (new cell) while cell 2 has a capacity of 3.23 Ah (degraded cell). Both discharging and charging experiments were performed using a C/3 constant current. The results are shown in Fig.5.13and5.14.

3.4 3.5 3.6 3.7 3.8 3.9 4 4.1 -15

-10 -5 0 5 10 15

dQ/dV [Ah/V]

Voltage Force

Voltage [V]

-0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4

dQ/dF [Ah/lbs]

0.15 0.16 0.17

Discharge Charge

Figure 5.13: C/3 constant current charge and discharge ICV and ICF curves for a new cell (capacity = 4.47 Ah).

3.4 3.5 3.6 3.7 3.8 3.9 4 4.1

Figure 5.14: C/3 constant current charge and discharge ICV and ICF curves for a degraded cell (capacity = 3.23 Ah).

There are two main takeaways from the data shown in Fig.5.13and5.14:

• In both cells, the charge and discharge ICV peaks do not have the same value. In the charge case, the ICV peak is at a higher voltage than the ICV peak in the discharge case. In the ICF case, however, both peaks in the charge and discharge case align.

The ICF peaks during charge, however, are less distinct.

• In the charge case, and in the case of the new cell in Fig. 5.13, there seems to be another peak in the ICF curve which occurs at at even higher voltage (around 3.95 V shown in dotted grey line). This phenomena needs further investigation during the charging case.

Although the ICV peaks in charge and discharge do not match, initial results indicate that in the ICF case, these peaks are matching. However, further research and analysis is needed in the charging case to investigate whether the ICF peaks in the charing case follow the same trend as the discharging case as the cell degrades. In the discharging case, the ICF peaks moved to higher voltage as the cell degraded. The same procedure would have to be repeated to parameterize the degradation model in charge similarly to the discharging case

as shown in Figs. 5.3 through5.6 and using Eq. 5.2 for capacity estimation. Also, more work is needed to understand the ICF behavior in charging, especially at high voltage, where the cell shows another ICF peak.

5.4 Conclusion

A novel method of using force in the incremental capacity analysis has been introduced.

The method shows promising results since it could be used in tandem or instead of the ICV method where the differentiation of voltage with respect to capacity can result in low single to noise ratios. Also, it is able to monitor and estimate capacity fade of a battery at higher SOCs as compared to using ICV method. This means that the stack does not have to operate at low SOCs to get an estimation of the capacity fade. For an NMC cell, results using the ICF method have shown that the peaks of the dQ/dF versus V curves occur at around 70% SOC while those using the ICV method occur at around 40% SOC. Also, 4 different fixtures were tested under different SOC and preloading conditions. All fixtures seem to exhibit similar behavior with a linear decrease of capacity with increasing ICF peak voltage value. Results show that the mean capacity of each fixture can be estimated with a maximum error of 2.5% over 95000 miles of cycling. Also, it has been shown that bulk force measurements can be used to estimate individual cell capacities. Results show that the maximum error is 3.1% with an absolute mean and standard deviation on the error of 0.42% and 1.14% respectively. However, more data and further investigation is required to study the adequacy of a linear fit. Results also showed that the above proposed method can work for estimating individual cell capacities if they are closely balanced. However as the cells drift apart in capacity (results shown here for cells that have 27% difference in capacity), the method cannot estimate individual cell capacities. Finally, the dependance of the ICF curves on C-rate is shown for different C-rates. For C-rates up to C/3, the shift in the ICF curve peaks is minimal which agrees with the strain behavior in [112].

However, at higher C-rates (1 C-rate), the peaks start to shift. Finally, some preliminary work has been done on comparing the charging and discharging behavior of the ICV and ICF curve peaks for two cells with different capacities. It is shown that although the ICV peaks do not align, the peaks in the ICF cases align. The ICF peak during charging is less pronounced and possibly harder to detect. More investigation is required during charging and discharging to determine the applicability of the ICF method in charging as there seems to be another prominent peak located at the higher voltages. Future work would also include investigating the sensitivity of the ICF curves to the applied C-rate, and implementing this estimation method in on-board state of health monitoring prognostic algorithms.

CHAPTER 6