3.2.4. Frequency Regulation Duty Cycle Test

Two modules have been subjected to the 24-h frequency regulation (FR) duty cycle. The 24-h duty cycle power is normalized with respect to the power corresponding to the C1 rate at 50% SOC of the battery module, estimated at 1090 W for both FR duty cycles. This corresponds to a discharge energy throughout of 3.9 kWh per cycle for this energy neutral signal. Note that positive power typically represents discharge from the battery module, and negative represents charge into the battery module from a grid perspective. In this work, since a battery cycler is used to charge and discharge the battery, charge is associated with a positive power signal, and discharge with a negative power signal. A reference to the raw data based on which the duty cycle is generated is given in Appendix B of [29]. Table 3 shows the operating parameters for the frequency regulation duty cycles, with the DOD for the 24-h duty cycle set at 20%.

**Table 3.** Frequency regulation characteristics.


Frequency regulation tests are done at two different initial SOC levels (80% and 57%) to determine the effect of different operating SOC ranges on performance and degradation. The charge acceptance of lead acid batteries is impacted by the preceding step. If the preceding step is a discharge step, the formation of small lead sulfate crystals offers sufficient surface area for the subsequent charge, whereas, if the preceding step is a charge step, the newly formed low surface area spongy lead reduces charge acceptance [35]. While optimization of expander properties is expected to mitigate this issue, for this work, starting an FR duty cycle after a preceding charge step to the initial SOC resulted in voltage

excursions above the upper limit at the start of the duty cycle. Hence the batteries were brought to an SOC equal to target start SOC + 5% (62% for target SOC of 57% and 85% for target start SOC of 80%), and subsequently discharged to the target start SOC. This finding also indicates there is room for further improvement in expander research and development for duty cycles with volatile power signals.

Since the frequency regulation duty cycle is energy neutral, the total discharge Ah exceeds the total charge Ah for each cycle, since RTE < 1. That is, for the condition of charge energy equal to discharge energy, the average charge voltage is higher than average discharge voltage. Since energy in Wh is the product of Ah and battery voltage, the discharge Ah is greater than charge Ah for a 24-h FR duty cycle. At the end of each frequency regulation cycle, the battery modules are charged to bring them back to their starting SOC using a 10-h rate., followed by a constant voltage charge until the desired charge Ah is achieved. The desired charge Ah is calculated by measuring the difference between charge Ah and discharge Ah for the preceding frequency regulation cycle and applying a suitable overcharge to address gassing during charge. Note that no overcharge is applied for the FR duty cycle starting at 57% SOC, while a 0.75% overcharge is applied during charge back for the FR duty cycle starting at 80% SOC. The charge back step is followed by a rest period of 1 h before the next frequency regulation cycle. The total duration for frequency regulation, including charge back and rest, is about 29 h.

#### **4. Results and Discussion**

#### *4.1. Reference Performance Test Results*

Reference performance tests and pulse resistance tests are conducted at regular intervals to evaluate the performance stability of the batteries performing four different duty cycles. Performance stability assessment is done by monitoring changes in the battery capacity and internal resistance as a function of elapsed test time, number of cycles and cumulative energy throughput for each of the four duty cycles.

Figure 3a shows the voltage and current profile of one discharge and charge cycle for the reference performance capacity test. Figure 3b shows the current and temperature profiles for two charge and discharge cycles. Resistive and reversible heat contribute to heat generation during operation [36]. Lead acid batteries have a positive temperature coefficient for open circuit voltage of 0.2 mV/◦C, which corresponds to endothermic contribution of reversible heat during discharge (and exothermic contribution of reversible heat during charge) [37,38]. The battery temperature rises during charge as expected, while during discharge, it rises during the first cycle and drops during the second cycle. At the start of the first discharge, the battery temperature was low at 22 ◦C, resulting in resistive heating that overwhelms the endothermic effect. In contrast, at the beginning of the second discharge, the temperature was higher at 24 ◦C, with associated lower resistive heating, such that the endothermic effect dominates. The exothermic behavior during charge has the potential for cell dry-out during constant voltage charging, hence it is prudent to set the CC current low to avoid significant temperature increase. None of the modules exhibited significant temperature increase during the course of testing, indicating that the charge profile is suitable, and the electrolyte dry-out is not a major failure mode.

Figure 4a shows battery capacity retention as a function of time elapsed since test start for different duty cycles. Capacity retention is calculated using the initial measured capacity as 100%. Over the same elapsed time, batteries subjected to FR duty cycles batteries degraded more than those subjected to PS duty cycles. The battery subjected to FR duty cycle operating in the 57–37% SOC range has the longest testing period and degraded the most. For PS duty cycle batteries, the battery capacity recovered after initial degradation. Performance enhancement can occur due to various factors such as improved electrode wetting during initial cycles, rearrangement of active material resulting in a greater active area, better electrolyte distribution among battery cell components [39], and reversal of irreversible sulfation. This recovery indicates that the initial capacity drop is not related to positive active mass shedding. Figure 4b shows capacity retention as a function of cycles, while Figure 4c shows capacity retention as a function of cumulative discharge energy. The sharp drop in capacity for FR (57–37)% in Figure 4b,c was observed after the battery was in an idle state for four months stored in a fully charged condition and is probably related to irreversible sulfation of the electrodes [40].

**Figure 3.** (**a**) Voltage and current profile of the capacity test; positive current is charge; (**b**) Temperature and current profile of the capacity test.

**Figure 4.** Capacity retention vs. (**a**) months (**b**) number of cycles and (**c**) cumulative discharge energy for four different duty cycles.

Typically, irreversible sulfation due to the growth of lead sulfate crystals occurs when batteries are stored for prolonged periods without being charged or subjected to low SOC during cycling [40,41]. The poor electronic conductivity of these lead sulfate crystals also isolate portions of the electrode, making them inaccessible, further reducing battery capacity. The battery subjected to FR (57–37)% duty cycle degraded more during the same period than FR (80–60)% duty cycle. The degradation may be related to irreversible sulfation of the battery electrodes caused by operation at lower SOC levels.

Since the frequency regulation duty cycle has higher discharge energy throughput per cycle than the peak shaving duty cycle, Figure 4b shows higher degradation per cycle for FR, while the gap decreases when degradation is normalized per unit MWh discharged (Figure 4c), especially for the FR (80–60)% duty cycle. Note that the average SOC for this duty cycle is 70%, close to the average SOC of 75% for the PS duty cycles.

The greater rate of capacity loss for FR duty cycles, coupled with less of a capacity recovery appears to indicate that positive active material shedding, in addition to irreversible sulfation, may play a bigger role for the FR duty cycle, which involves multiple charge-discharges and the associated contractions and expansions of the active mass within a 24-h duty cycle. This needs to be taken into consideration while using VRLA batteries for frequency regulation. This failure mechanism appears to be absent for PS duty cycles. Note that converting cumulative energy throughput to number of 100% DOD equivalent cycles, FR (57–37)% has completed 728 100% DOD equivalent cycles, while the corresponding numbers for FR (80–60)%, PS (2-h) and PS (8-h) are 494, 362 and 289 respectively.

#### *4.2. Pulse Resistance Test Results*

Figure 5a shows internal resistance vs. SOC for discharge pulse durations from 100 (ms) to 6 s, while Figure 5b shows corresponding results for the charge pulse. The ohmic resistance, corresponding to the 100 ms measurement, includes contribution from charge transport resistance as well, and decreases as SOC increases. Ohmic resistance is determined by electronic conductivity of the current collector and the electrode active mass, along with electrolyte ionic conductivity and tortuosity of the separator and electrode pores. The discharge product at both electrodes, PbSO4, has lower electronic conductivity than the corresponding charge states of PbO2 at the positive and spongy Pb at the negative, while the electrolyte ionic conductivity increases with increasing state of charge [42]. While ohmic resistance should not depend on pulse direction (charge vs. discharge), for the 0.1-s discharge pulse, the ohmic resistance decreases more steeply with increasing state of charge compared to the charge pulse. The resistance at 10% SOC is nearly the same regardless of direction of the pulse, indicating the charge transfer resistance at this SOC is nearly the same for the charge and discharge pulse. As the SOC increases, the discharge pulse resistance decreases more than the charge pulse. This is expected, as the charge transfer resistance is expected to increase with state of charge for the charge pulse, negating the decrease in ohmic resistance at high SOC.

**Figure 5.** Internal resistance vs. SOC (%) profile (**a**) for discharge pulse (**b**) for charge pulse.

For the discharge pulse, at high SOC, the internal resistance increases with pulse width. While it is difficult to separate out charge transport and mass transport effects, assuming a 2-s cut-off for charge transport, it appears that mass transport effects can be estimated all the way up to 6-s pulse width. On the other hand, at low SOC, after 2–3 s, the resistance does not increase significantly. This indicates that due to low acid concentration at low SOC, mass transport effects are almost fully captured within 2–3 s. The data for the charge pulse is a mirror image. At low SOC, since there is plenty of water in the electrolyte, mass transport effects are observed for the entire 6-sec pulse duration. At high SOC, due to

higher concentration of the sulfuric acid product, a majority of the mass transport related effects are captured within 2–3 s.

#### *4.3. Degradation Analysis*

Figure 6 shows all four batteries' capacity retention vs. total internal resistance. The total internal resistance is averaged from 80% SOC to 30% SOC for charge and discharge pulses. The capacity degrades linearly with an internal resistance increase. VRLA batteries are typically positive electrode limited. As the battery cycles, the positive electrode grids undergo corrosion, causing an in-plane mechanical stress on the grid, leading to a decrease in adhesion between the active material and the grid. The active material is separated from the grid because of the expansion and shrinkage of the active material during the discharge and charge cycles, reducing the capacity [43]. The capacity recovery for PS cycles in Figure 4c indicates that positive active material shedding is not the main reason for capacity loss. As stated earlier, use of appropriate alloys and compressive force appears to have mitigated this failure mode. This indicates that irreversible sulfation at the bottom of electrodes due to electrolyte stratification and at the low surface area negative during partial state of charge operation may be a reason for the linear decrease in capacity as resistance increases, while resistance increase may be associated with positive grid corrosion, water loss and irreversible sulfation. As irreversible sulfation at the negative increases, there is expected to be a transition from positive limit to negative limit, at which point gas generation increases, accompanied by further resistance increase due to water loss. The slight increase in capacity corresponding to an increase in resistance for FR duty cycles may be related to reversal of dense sulfation film (leading to increase in capacity), while grid corrosion and loss of water lead to an increase in resistance.

**Figure 6.** Capacity retention vs. internal resistance profile for four duty cycles.

Figure 7 shows internal resistance vs. elapsed test duration. Ohmic resistance increases with increasing duration. Some of the reasons could be (1) irreversible lead sulfate formation at the bottom of both electrodes related to electrolyte stratification and at the negative electrode due to partial state of charge cycling, (2) positive grid corrosion resulting in inplane mechanical stress on grid and loss of contact with positive active material (3) Positive active material shedding due to active material volume change during cycling resulting in loss of electronic percolation in active mass (4) Electrolyte dry-out resulting in an increase in ionic resistance.

For all duty cycles, the ohmic resistance for charge pulse is greater than that for discharge pulse, which is due to contribution from charge transfer in the 100 ms pulse. Depending on the duty cycle, the reason for ohmic resistance increase may differ. For example, for peak shaving, since the charge rate is the same for both duty cycles, if electrolyte dry-out is the predominant mode, electrolyte loss is expected to be nearly the same for the 2-h and 8-h duty cycles. The slightly lower resistance for the 2-h discharge may be related to less time spent in the 75% to 50% SOC range, resulting in less irreversible

sulfation at the negative electrode. The ohmic resistance for the FR (80–60)% run matches the results for the PS 8-h, whereas the corresponding values for the FR (57–37)% case is lower. Hence it appears that electrolyte dry-out could be the dominating reason for ohmic resistance increase for the PS duty cycles and the FR duty cycle operating in the 80–60% SOC range, with the PS duty cycles expected to have greater grid corrosion and water loss due to being subjected to a full charge after each cycle, while the FR (80–60)% duty cycle is expected to have more sulfation due to spending >3.5× duration at <70% SOC. Irreversible sulfation at the negative electrode could be the dominant reason for ohmic resistance increase for FR (57–37)%. Note that this battery was on open circuit for four months. During this time, some water loss is expected due to self-discharge, but the main degradation mechanism may be irreversible sulfate formation, which is corroborated by the steep increase in charge transfer resistance for the charge pulse at the 10-month mark (Figure 7b).

**Figure 7.** Resistance vs. elapsed test duration (**a**) ohmic, (**b**) charge transfer, (**c**) mass transport, (**d**) total.

The charge transfer resistance for the discharge pulse is relatively unchanged with a slight increase, while it decreases in the first few months for the charge pulse. This appears to indicate that the charge acceptance initially is not very high, probably related to sulfation formed over a long storage period of ~6 months prior to testing. The steep increase in charge transfer resistance for the charge pulse at the 10-month mark appears to be related to the 4-month rest on the open circuit, which leads to sulfation of the active material and the associated poor charge acceptance. As expected, the corresponding charge transfer resistance for the discharge pulse is not very high since the reacting species is the charged form of the active mass in each electrode.

Mass transport resistance for the charge pulse increases with time, while that for the discharge pulse increases to a much less degree (Figure 7c). During charge, sulfuric acid is produced in the pores, requiring transport of the acid away from the pores to allow reactant water access. It is difficult for bulk water to force its way into the pores while displacing the acid that is produced. During discharge, sulfuric acid is consumed at the electrode pores, which are now filled with water. It is easier for acid in bulk to enter the pores and displace the produced water. Irreversible sulfation results in the increase of active mass volume, with an associated decrease in porosity and pore size. Hence the pore size and porosity reduction associated with irreversible sulfate formation increase mass transport resistance, with more of an impact on the charge pulse.

This effect is maximum for FR (57–37)% duty cycle, where irreversible sulfation at the negative electrode is expected to be the largest, with corresponding greatest pore size and porosity reduction. The results are similar for FR (80–60)% duty cycle and PS 8-h, where both duty cycles spent significant time at <70% SOC and hence are expected to have a similar degree of irreversible sulfation, while the PS 2-h duty cycle spends less time at <70% SOC, and hence has less irreversible sulfation, accompanied by a lower increase in mass transport resistance.

It is our hypothesis that for the PS duty cycles and FR (80–60)% duty cycle, the main mode of degradation is water loss, with some irreversible sulfation at the bottom of both electrodes related to electrolyte stratification, with additional sulfation at the low surface area negative for FR (80–60%) and PS 8-h related to longer time spent at SOC < 70%, while for FR (57–37)% duty cycle, the main mode is irreversible sulfation at the negative electrode related to longer time spent at low SOC along with sulfation at the bottom of both electrodes related to electrolyte stratification, with water loss related effects expected to be lower. To validate these hypotheses, the aged battery modules, along with fresh modules to establish a baseline, need to be disassembled in the charged state and subjected to the recommended tests in the following section.

#### *4.4. Recommended Tests to Validate the Proposed Failure Mechanism*

	- a. Cut the electrode into three parts, top, middle, and bottom, measuring wet weight, followed by titration to determine acid content for each part, followed by measurement of dry weight.
	- a. Immerse each separator piece in deionized water after weighing and titrate vs. 1 M NaOH.

#### *4.5. Internal Resistance vs. SOC Trends for the Duty Cycles*

Figure 8 shows the internal resistance measured from discharge and charge pulse at different SOC levels for four different duty cycles. The internal resistance is averaged over time at the same SOC. From Figure 8a–d, it can be seen that the duty cycles affect the battery internal resistance similarly, with the shape of the resistance as f (SOC) similar for all duty cycles. For all four duty cycles, total internal resistance measured from charge pulse is higher than total internal resistance measured from discharge pulse. This is in line with our observation that charge transfer- and mass transport-related losses are higher for charge.

**Figure 8.** Internal resistance vs. SOC (%) for four duty cycles. (**a**) peak shaving 2-h rate (**b**) peak shaving 8-h rate (**c**) frequency regulation (80–60)% and (**d**) frequency regulation (57–37)%.

#### *4.6. Round Trip Efficiency Results for the Duty Cycles*

Figure 9 shows the round-trip efficiency (RTE) of (Figure 9a) peak shaving duty cycles and (Figure 9b) frequency regulation duty cycles. While the RTE is constant for 400 peak shaving duty cycles, it decreases subsequently, with significant fluctuations. This is due to a change in the charge procedure after 400 cycles. Initially, the charge back procedure consisted of a CC/CV charge with charge termination set at 3% overcharge or current decreasing to 0.8 A, whichever occurs first. Also, the constant current and constant voltage charge steps each had a 7-h time limit to reduce excessive gassing. After 400 cycles, the CC/CV charge is terminated when the current decreases to 0.8 A, with the 3% overcharge requirement removed. Note that in these subsequent cycles, the overcharge is more than 3% of the discharge capacity, thus reducing the RTE. Additionally, the charge capacity is different in each cycle as some of the charge back cycles reached the termination current of 0.8 A before the time limit, while for other cases, the time limit was reached before the current decreased to 0.8 A. Hence the decrease and fluctuation in RTE are not primarily due to battery degradation. The FR (80–60)% has slightly lower RTE than FR (57–37)%. For the energy neutral frequency regulation signal, as explained earlier, the total discharge Ah > total charge Ah for the FR duty cycle, since RTE < 1. The charge back Ah is determined by applying a 0.75% overcharge to the difference between total discharge Ah and total charge Ah for the FR (80–60)% duty cycle, while no overcharge was applied to the FR (57–37)% duty cycle since gassing is expected to be insignificant in this low SOC level. Hence FR (57–37)% has a slightly higher RTE.

**Figure 9.** RTE (**a**) peak shaving duty cycle (**b**) frequency regulation duty cycle.

#### *4.7. Mapping Hybrid Electric Vehicle (HEV) Degradation with Grid Services*

It was determined that batteries subjected to a micro hybrid electric vehicle (HEV) drive cycle operating at 5% DOD in the 90–85% SOC range experienced no sulfation of the negative electrode, while batteries subjected to a mild HEV drive cycle operating in the 90–70% SOC range experienced mild sulfation, and batteries subjected to a full HEV drive cycle operating in the 80–30% SOC range experienced rapid sulfation [20]. The PS duty cycle appears to map with the mild HEV cycle, with a slightly lower average SOC, but getting fully charged after each cycle. The PS 8-h duty cycle spends more time at SOC < 75%, hence may have more sulfation at the negative. The FR (57–37)% duty cycle has a slightly lower average SOC than a full HEV drive cycle, is charged to only 57% SOC at the end of each cycle and is expected to show greater sulfation of the negatives. The FR (80–60)% duty cycle has an average SOC halfway between mild HEV and full HEV and is expected to experience sulfation greater than mild HEV and less than full HEV. The PS duty cycle and the FR (80–60)% duty cycle are expected to see greater water loss compared to the FR (57–37)% duty cycle. Table 4 below provides a mapping of grid services degradation with that experienced by various HEV cycles.


**Table 4.** Mapping of Grid Services Degradation with HEV Cycles.

#### **5. Conclusions**

Leveraging upon work done on the degradation of lead acid batteries used for hybrid electric, degradation mechanisms of batteries used in micro, mild, and full HEV are mapped with the grid service duty cycles used in this work. Post-cycling tests have been proposed to validate the hypothesis for dominant failure modes for each duty cycle.

Assigning different durations within the charge or discharge current pulse, ohmic, charge transfer and mass transport resistances are estimated. Higher ohmic resistance for

PS 8-h relative to PS 2-h is explained by the longer duration spent at <70% SOC for the former, while similar ohmic resistance increase for the PS 8-h and FR (80–60)% is explained by higher water loss for the former balanced by the higher sulfation for the latter related to 3.5× longer time at <70% SOC for the FR (80–60)% grid service. The higher charge transfer resistance during the charge pulse at the start of testing may be related to extended storage prior to test, resulting in sulfation, which is subsequently reversed. This hypothesis is further validated by the spike in charge transfer resistance for the charge pulse for the FR (57–37)% duty cycle after a 4-month stand at open circuit, which is also reversed upon further testing. The higher mass transport resistance for the charge pulse also correlates with grid services with greater expected sulfation, with the value highest for FR (57–37)% and lowest for PS 2-h. Hence these measurements provide very useful insights into the degradation mechanisms for various grid services. The higher resistance during charge across all duty cycles places limitation on maximum power for energy neutral grid services such as frequency regulation.

Regardless of the duty cycle, capacity loss increased with an increase in internal resistance, with nearly the same slope for all duty cycles. For FR (57–37)%, while the capacity decreases as expected with increase in resistance, there is reversal in this trend with a slight increase in capacity as resistance increases, which appears to signal a switch from negative limit to positive limit related to reversal of the dense sulfation at the negative. Such signatures provide insights into battery management approaches for VRLA used in various grid services, taking into account the initial positive & negative electrode active material ratio and electrolyte content to maximize battery operating life.

While positive active material shedding does not appear to be the main failure mode for PS duty cycles. it may play a bigger role for the FR duty cycle, due to multiple contractions and expansions of the active mass within a 24-h duty cycle. This needs to be taken into consideration while using VRLA batteries for frequency regulation and other services such as renewable smoothing that are associated with volatile signals. The capacity loss per energy throughput is highest for grid services with the highest potential for sulfation—FR (57–37)% > FR (80–60)% > PS 8 h > PS 2 h—thus highlighting the need for the mitigation of sulfation to be a top priority. This is achieved by minimizing electrolyte stratification, which affects both electrodes, and by increasing negative electrode surface area to ensure uniform current distribution by addition of suitable expanders.

**Author Contributions:** Conceptualization, V.V.V., E.C.T., D.M.R. and V.L.S.; Formal analysis, N.S.; Funding acquisition, D.M.R. and V.L.S.; Methodology, N.S., V.V.V. and E.C.T.; Software, E.C.T.; Supervision, V.V.V., D.M.R. and V.L.S.; Writing—original draft, N.S. and V.V.V.; Writing—review & editing, N.S., V.V.V., E.C.T. and G.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the U.S. Department of Energy (DOE) Office of Electricity under contract number 57558. The APC was funded by the U.S. Department of Energy (DOE) Office of Electricity.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data is contained in the figures and tables of this article. Supporting data for the figures and tables are available from the corresponding author.

**Acknowledgments:** This work was funded by the U.S. Department of Energy (DOE) Office of Electricity under contract number 57558 through Pacific Northwest National Laboratory (Validated Safety & Reliability). Pacific Northwest National Laboratory is operated by Battelle for the DOE under contract number DE-AC05-76RL01830.

**Conflicts of Interest:** The authors declare that they have no known competing financial interest or personal relationship that could have appeared to influence the work reported in this paper.
