Balance hybrid/EV battery packs to extend operational lifetime: Pt. 2 – Algorithm and lab tests
Part 1 of this feature discussed the battery-monitoring and battery-balancing hardware, and the control strategy.
The goal of passive balancing is to adjust the state of charge (SOC) of all cells in the pack such that the maximum amount of energy can be safely extracted from the battery pack. Passive balancers do not create or contribute charge to the pack, which means that the cells with the lowest capacity in the battery pack will determine the pack’s useful capacity. To maximize the pack capacity the balancer needs to ensure that batteries with lower capacity and SOC are allowed to fully charge and discharge.
A battery’s total stored energy will only be used if it is allowed to be fully charged and fully discharged, meaning that the weakest cell should be the cell that finishes charging and discharging first. The main concern of a passive balancing scheme is to be able to identify the cells with higher capacity.
The SOC of a battery is reflected in the open circuit voltage of the battery and is an indicator of the percentage of the remaining energy. Two cells having the same SOC does not mean the cells are storing the same amount of energy; a higher capacity cell will always have more stored energy at a given SOC than a lower capacity cell.
The balancing software control algorithm is designed to coordinate balancing with the charger, and is enabled at the beginning of the charge cycle. Because passive balancing can only remove energy from the battery pack, it makes no sense to balance while the pack is discharging. This also eliminates the possibility that a cell with lower capacity will be equalized to the SOC of a higher capacity cell, which would decrease available capacity during discharge.
Once the charge cycle has started, the cell voltages are stored before the charger is connected; the balancer should determine which cell had the lowest cell voltage at the beginning of the charge cycle, this cell will be referred to as Clow. When the charge cycle is complete, indicated by one cell reaching the predetermined maximum voltage limit, the cell voltages are again stored. In both cases the cell voltages are measured with no load current and after a short period of settling.
Balancing is required if the Clow’s measured voltage isn’t the highest voltage after the charge cycle is completed. Clow’s voltage after the charge cycle is set to Vbalance. The bleed resistors should be activated for cells in the pack that have a measured voltage that is higher than Vbalance. Balancing switches should remain on until all individual cell voltages match the Vbalance voltage. After balancing has occurred, the batteries resume charging to completely charge the cells. To see the impact of passive balancing, two tests were conducted and the results follow.
Test results: Battery pack 1
Battery pack 1 was cycled through 100 charge/discharge cycles, Figure 3, below, shows the six cell voltages recorded after a number of cycles.
Figure 3: Pack 1 cell voltages after charge cycle
The figure shows the measured cell voltages at the end of a complete charge cycle after a short period of relaxation. The imbalance between cell voltages after charge is related to small variations in capacity and internal resistance. On the first complete cycle the battery pack’s capacity was measured to be 2.072 AHrs, after 100 cycles the capacity measured 2.043 AHrs, a small decrease in capacity as cycle count increased.
There is also a trend that the final voltage of the cells after charging is decreasing as the number of charge/discharge cycles increases; this is particularly noticeable after 100 cycles. This is most likely due to a small increase in the batteries’ internal resistance due to the battery aging. An increased internal resistance causes a battery to reach the end of charge threshold sooner. Despite having no balancing during operation, this particular battery pack maintained the same level of imbalance throughout the 100 cycles. It is quite rare to find a pack of cells that naturally match each other as well as this pack.
Test results: Battery pack 2, and conclusions
The second battery pack was evaluated with the passive balancing algorithm applied. Before any balancing was done the pack was charge/discharged cycled 10 times. The initial voltages for pack 2 are shown here in Figure 4.
Figure 4: Pack 1 cell voltages after charge cycle
Unlike pack 1 the SOCs of the batteries did not match very well from the manufacturer. This type of mismatch is far more likely to be encountered. Pack 2 required balancing before the pack could provide its total potential capacity. This is much more typical.
There is a large, greater than 100 mV, imbalance between cell 5 and the rest of the cells. This imbalance has a significant impact on the capacity of the battery. After one full cycle the pack showed a measured capacity of 1.765 AHr. After 10 cycles the imbalance remained and the balancing algorithm was activated. The balancers discharged all cells to match cell 5 and after the full charge cycle a SOC of 2.043 AHr is recorded, a 16% improvement from the original SOC. The balancing algorithm remained active, but made very few corrections over the next 50 cycles, and the capacity measured 2.044 AHrs after 50 cycles.
Even after a large number of balance cycles, the pack is still not making use of the total potential usable energy. The major limitation is that the balancing algorithm does not take into consideration the internal resistance of the battery. Cell 1 has a higher internal resistance consistently finished charging well before cell 5, preventing cell 5 from becoming fully charged.
A modification was made to the balancing algorithm after 50 cycles to see if the pack capacity could be improved. The balancing algorithm was modified to leave the discharge resistors connected across cells while the charger was connected if any cell’s voltage was greater than Clow. This allows weaker cells to obtain more charge before the charger disconnects and is an example of the steering charge current approach mentioned in Figure 2 (repeated here from Part 1 of this feature).
Figure 2: Two options for passive cell balancing. Resistor value determines the primary function.
This change in balancing strategy increased available capacity to 2.051 AHrs and improved balancing time. The battery pack charged and discharged 50 more times, for a total of 100 cycles the capacity after 100 cycles was measured to be 2.054 AHrs. The capacity of battery pack 2 remained constant over the test and increased when balancing strategy was improved. This improvement was obtained even though initially one cell was significantly mismatched from the others.
Conclusions
If battery packs are physically small with a low cell count, an initial conditioning step will allow the batteries to stay well matched over battery lifetime. In small battery packs the battery load and temperature conditions are generally well matched.
Testing showed little imbalance will develop over large number of charge/discharge cycles; battery pack 1 lost 1.4% of its capacity. The second battery pack demonstrated a need for balancing hardware from the start; without balancing hardware the battery pack effectiveness is completely at the mercy of the battery manufacturer and cannot correct any amount of errors in the battery pack.
With the balancing system, battery pack 2 was able to maintain its capacity throughout the test while pack 1’s capacity steadily declined. Overall the balancing system helps to extend the battery pack capacity throughout the operating lifetime.
Enhancements to the balancing algorithm could include using battery characterization data and specific cell modeling. This allows the controller to more accurately determine the energy level of individual cells in the pack, enabling the controller to more accurately balance cells and reduce balancing time even when using the same balance current.
Cuyler Latorraca is an applications engineer at Linear Technology.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
