Figs. 1 and
2 Shows discharge (1 A current) and subsequent charging capacity (2 A current) of LFP battery. As observed and discussed in our previous paper [
12] five distinct regions are seen in voltage-time graph both in charge and discharge operation. The capacity of the battery was 10.82 Ah at 1 A discharge current which is higher than the rated capacity 10 Ah. For Peukert’s study the battery was discharged at various currents, namely 1, 2, 4, 6, 8 and 10 A to find out the impact of rate of discharge on capacity output. As expected, the rate of discharge hardly made an impact on discharge capacity. Though rate of discharge increased by 10 times the capacity drop is just 2.77% indicating the discharge kinetics is very fast in LFP system. Voltage-time plot at various discharges current is shown in
Fig. 3. As the discharge current increases, the number of distinct regions is getting reduced from 5 to 2.
Fig. 4 explains charging behavior of battery after various discharge currents. Since charging current is the same, the previous discharge current does not have any specific effect on charging behavior and capacity.
Fig. 5 shows typical Peukert’s plot. The slope of 1.0117 indicates the reaction is very fast and independent on rate of discharge. The preliminary capacity and Peukert’s study indicate that the battery quality is good and can be subjected for life cycle test.
3.1 Life cycle test as per IEC-61427-1 (2013)
This test is very robust and accepted by international community for solar-off grid application. As per the standard one endurance unit corresponds to 1 year life in the field and comprises of Phase-A, 49 cycles at low SOC (10 to 40%), Phase-B, 99 cycles at high SOC (75 to 100%) and one capacity health check. The test protocol was designed to take care of seasonal variations such as, for more than 2 weeks the battery might not get charged due to absence of solar radiations. Cumulative Ah output in Phase-A is 14.7 times and Phase-B is 24.75 times of rated capacity. Hence total Ah output in one endurance unit is 39.45 times of rated capacity (C-10 in case of lead-acid and C-1 in case of Lithium battery). Since the test is conducted at 40°C, the ambient temperature Ah turnover is double the value of 40 C and in this case, it is 78.9. At maximum, solar battery is discharged for 20% of rated capacity in a day. Hence one endurance unit corresponds to 395 days,
i.e., little more than one year in the field. In the present study the test was conducted at room temperature (25 to 28°C), hence two endurance units correspond to one year life in the field.
Fig. 6 is the real time voltage
vs. cycle evolution plot. In Phase-A, as the life cycle progresses the end discharge voltage is becoming more positive because the charge input is 1.03% of previous discharge output which increases SOC by 0.03%. In Phase-B, during change over stage between 14.6 V charge and 1.25 A CC discharge, voltage spike is taking place due to cell balancing act. The changeover is taking place in less than 50 ms interval. Since two cells are in parallel in each module (4S/2P configuration) the balancing current is equal to difference in voltage between the cells and its internal resistance. This huge inflow of current results in occasional voltage spike and as the cycle progresses more spikes should occur due to cell divergence.
Fig. 7 explains this phenomenon. In four endurance units 7 spikes has occurred out of which 3 are up to 15 V, 3 are up to 16 V and 1 is up to 17 V. In 5
th and 6
th endurance units, more than 30 spikes occurred in the voltage range 17 to 20 V. This is an indication of cell divergence due to inherent variations in raw material, design and process [
13]. Since voltage excursion is up to 20 V in the battery and 5 V at cell level, the impact will be huge in oxide-based cathode cells. Solar off-grid application needs battery to perform its duty close to 100 and 0% SOC, the inherent variation and voltage spike due to cell balancing is a challenge in real life applications. To confirm that the cell voltage excursion is only due to cell balancing, after the life cycle the battery was dismantled and two individual cells were subjected to discharge/charge study. The voltage-time behavior is depicted in
Fig. 8 and fluctuation is not happening at all during changeover from charge to discharge. This clearly shows that the battery after reaching 14.6 V, during change over time (less than 50 ms) from charge to discharge, cell balancing is happening and during this gap the current which is flowing between the cells results in polarizing the cells in an uncontrolled way.
Fig. 9 explains health check capacity data after each endurance unit. The plot shows nearly 4% capacity decay per unit for first two units and sudden drop in capacity by 14% and further a drop of capacity by 3% per unit. After sixth endurance unit battery lost 30% of its original capacity indicating that LFP battery can reliably work in solar off-grid application for just 2 to 3 years. This is startling and reasons to be known for this poor performance.
To find out the reason, the discharge capacity curve as in
Fig. 10 was plotted for every health check study after each endurance unit. It is clear that up to 2 cycles, the end voltage drop started at 12.6 V and from 3 to 6 cycles, the voltage drop started at 12.8 V indicating the phenomena happening between 12.6 to 12.8 V is missing in the later endurance units. To probe further, the battery was discharged and cells were dismantled and individually charged at 0.5 A current till the voltage becomes 3.65 V and
Fig. 11 shows charging capacity of individual cells. All the cells are diverging after 3.4 Ah input, and cell No. 2 and 5 are polarized at 60 mV higher than the other cells. The individual capacity of the cells is shown in
Fig. 12. Assuming that the original capacity of the cells is 5.4 Ah, out of 8 cells, 5 cells lost 25 to 30%, one cell 20% and 2 cells around 10% of rated capacity. IR (Internal resistance) of the cells measured using Hioki instrument is shown in
Fig. 13. Variation in the IR is in the range of 0 to 46%. Capacity and IR variation between individual cells clearly indicating that the cells are diverging during life cycle test. To unravel more, the data in
Fig. 11 was rearranged in the form of dV/dQ
vs. Ah and dV/dQ
vs. voltage plots as in
Figs. 14 and
15 [
14,
15]. They demonstrated that incremental capacity study (dQ/dV
vs. voltage) can be effective for identifying the root causes for capacity fading such as loss of active material, change in cell chemistry, undercharging and under discharging phenomena. dV/dQ
vs. Ah plot was used in [
16] for studying the degradation mechanism. In the present study dV/dQ is used in y-axis because the change in voltage during charging and discharging for LFP system is zero or very minimum due to involvement of two-phase systems. In addition, the 10 minutes’ data collection frequency made dV very small and as a result dV at the denominator is distorting the data and hence dV/dQ is used.
Fig. 14 shows that all the cells behave more or less same till the Ah input during charging is 3.4 Ah. Divergence is occurring after 3.4 Ah and the second peak position also appeared at different Ah for the cells and the capacity range varies from 4.57 to 5.07 Ah. Appearance of second peak at different Ah indicates that the amount of lithium intercalated in graphite structure is different for the cells under study. The original rated capacity for the cell is 5.4 Ah. Hence the individual cells reduced capacity by 6 to 16% of its rated value.
Fig. 15 reveals that above 3.4 V the slope of dV/dQ
vs. voltage plot is varying for all the cells. This charging voltage regime corresponds to lithium plating reaction on carbon electrode. Variation in the slope indicates indeed lithium is plating on carbon edge at different rate [
17,
18]. For cell No. 2 and 5, the shift in the dV/dQ
vs. voltage plot by nearly 60 mV indicates higher internal resistance which reflects in IR
vs. cell plot in
Fig. 13. The variations in (a) cell to cell capacity, (b) IR, (c) appearance of second peak in dV/dQ
vs. Ah plot and (d) slope in dV/dQ
vs. V plot after 3.4 V suggests loss of active lithium due to plating on negative electrode and formation of additional SEI layer due to plated lithium.
Solar off-grid application demands battery to be operated mainly between 0 to 30% SOC (peak winter and rainy seasons) and 75 to 100% SOC (peak summer). When batteries are operated in low SOC (0 to 30%), preferential lithium plating is occurring due to higher reactivity of carbon edge atom than graphene layer. Repeated cycling at this range preferentially plates out lithium on negative electrode.
Figs. 14,
–
16 show the above-described phenomena. Variations in lost capacity and capacity up to 2
nd peak as in
Fig. 16 indeed explains the challenges the battery faces when it operates at low SOC regime. At high SOC (75 to 100%) the charge balancing due to inevitable inherent variation leads to voltage excursion even up to 5 V per cell (
Fig. 7). This may lead to destruction of positive active material. In case of LFP it is not seen but in vase of NMC it may lead to fire hazard. When similar study was conducted using NMC, the battery was gutted into fire in author’s laboratory. The voltage excursion at high SOC and lithium plating at low SOC results in poor life cycle performance of LFP battery in solar off-grid applications. Hence to make lithium battery successful in solar off-grid applications, it has to be operated between 30 to 90% SOC window and hence the battery has to be over designed at least by 40%.