In modern battery energy storage systems, precise State of Charge (SOC) calibration and effective balancing control are fundamental to maximizing system capacity utilization. Series-connected battery packs inevitably suffer from inconsistencies caused by manufacturing tolerances, temperature gradients, and varying aging rates. These inconsistencies lead to divergent voltage operating windows among cells, which ultimately limits the usable capacity of the entire battery energy storage system. To address these critical challenges, we propose a novel Self-Calibration Balancing Strategy (SCBS) based on dynamic voltage characteristics at the charge and discharge terminals. This strategy enables full-time adaptive control aimed at fully utilizing the cell with the minimum capacity within the battery energy storage system.
Our proposed methodology is validated on three 200kWh lithium iron phosphate (LFP) energy storage cabinets. Each cabinet comprises 224 cells connected in series. The performance of the SCBS is rigorously compared against two conventional balancing strategies: the Minimum Voltage Balance Strategy (MVBS) and the Average Voltage Balance Strategy (AVBS). The experimental results demonstrate that the SCBS achieves superior balancing efficiency and effectiveness, significantly enhancing the operational reliability of the battery energy storage system.
| Parameter | Value |
|---|---|
| Rated Energy | 200 kWh |
| Rated Power (0.5P) | 100 kW |
| Cell Chemistry | LFP |
| Nominal Cell Voltage | 3.2 V |
| Nominal Cell Capacity | 280 Ah |
| Number of Cells in Series | 224 |
| Total Rated Voltage | 716.8 V |
System Topology and Balancing Circuit
The control topology of the distributed battery energy storage system includes the battery system, a Battery Control System (BCS), and various environmental monitoring subsystems. The battery system itself consists of 224 cells connected in series. The system utilizes a voltage cut-off strategy for protection; charging stops when any cell reaches 3.6V, and discharging stops when any cell reaches 2.8V. This operating mode makes SOC consistency critical for capacity utilization.
The balancing circuit is a switched shunt resistor architecture. Each cell is equipped with a 33Ω shunt resistor controlled by a MOSFET. When activated, the resistor dissipates energy from the cell with a balancing current of approximately 80mA. The timing of the Analog Front-End (AFE) voltage sampling and balancing is mutually exclusive to prevent interference. The balancing efficiency θ, representing the duty cycle of effective balancing within a period, is a key parameter in our calculations. If the AFE chip temperature exceeds 75°C, the system switches to an odd-even interleaved balancing mode. If the temperature surpasses 80°C, balancing is paused to ensure safety and stability of the battery energy storage system.
The SCBS Adaptive Control Algorithm Model
The SCBS strategy is built upon a feedback-based adaptive control framework. It features a real-time SOC self-calibration mechanism that leverages the dynamic voltage curves at the end of charging and discharging, followed by a full-time adaptive balancing process.
Real-Time SOC Self-Calibration at Terminal Stages
The LFP cells in the battery energy storage system exhibit a near-linear voltage change in the 95%-100% SOC charging region and the 0%-5% SOC discharging region. This characteristic is exploited for self-calibration.
During Discharge: When the voltage of the lowest cell drops below 3.1V, we begin sampling voltage and accumulated Ah data at fixed intervals. The terminal voltage curve of the cell that first discharges to 2.8V is used as the discharge reference curve. The reference SOC for this reference cell at the cut-off moment is 0%. The reference SOC for other cells at time t is calculated as:
$$SOC_{ref,dis}(t) = \frac{Th_{Ah}(t) – Th_{Ah}(t_1)}{Q_{e}} \times 100\%$$
Where \( Th_{Ah} \) is the accumulated Ah throughput calculated via the Coulomb counting method, \( t_1 \) is the moment the system reaches 0% SOC, and \( Q_e \) is the rated capacity of the system.
$$Th_{Ah}(t) = Th_{Ah}(t_0) + \int_{t_0}^{t} I(t) dt$$
During Charge: When the voltage of the highest cell exceeds 3.4V, we begin sampling data. The terminal voltage curve of the cell that first charges to 3.6V is used as the charge reference curve. The reference SOC at the cut-off moment is 100%. The reference SOC for other cells is calculated as:
$$SOC_{ref,cha}(t) = \left(1 – \frac{Th_{Ah}(t) – Th_{Ah}(t_2)}{Q_{e}} \right) \times 100\%$$
Where \( t_2 \) is the moment the system reaches 100% SOC. Using these reference curves, the SOC at the cut-off moments (\( SOC_{end,cha} \) and \( SOC_{end,dis} \)) for each cell can be interpolated from its voltage. The actual maximum dischargeable capacity for each cell \( i \) is then:
$$Q_{a}(i) = \frac{Q_{dis}}{SOC_{end,cha}(i) – SOC_{end,dis}(i)}$$
The minimum value in this capacity sequence, \( Q_{cell,min} \), represents the maximum capacity that the entire battery energy storage system can theoretically achieve after balancing.
| Cabinet ID | Initial Discharge Capacity (Ah) | Max Voltage Deviation at Charge Cut-off (mV) |
|---|---|---|
| 1 | 277.9 | 145.7 |
| 2 | 278.5 | 141.2 |
| 3 | 277.4 | 148.2 |
Full-Time Adaptive Balancing Control Strategy
The SCBS aims to achieve SOC consistency at the charging cut-off moment and full utilization of the minimum capacity cell. The equilibrium target is defined as:
$$max(SOC_{end,cha}) – min(SOC_{end,cha}) \le 1\%$$
When calibration conditions are met, the balancing strategy is triggered. It first identifies the cell with the minimum SOC at the discharge cut-off moment. The SOC of this same cell at the charge cut-off moment is defined as the baseline \( SOC_{base} \). All cells with a charge cut-off SOC higher than this baseline are identified as requiring balancing. The required balancing time for each target cell \( i \) is calculated as:
$$T(i) = \frac{(SOC_{ref,cha}(t_1, i) – SOC_{base}) \times Q_e – \Delta Q}{\theta \times I_b}$$
Where \( \theta \) is the balancing efficiency, \( I_b \) is the balancing current, and \( \Delta Q \) is a reserved capacity margin (e.g., 1 Ah) to prevent ineffective balancing near the target. The system adaptively corrects the balancing strategy in subsequent cycles based on the latest reference curves and cell states, forming a closed-loop feedback regime.
| Operating Condition | Terminal Voltage (mV) | True SOC (%) | Offline Curve SOC (%) | Self-Calibration SOC (%) |
|---|---|---|---|---|
| Charging (0.2P) | 3445 | 94.7 | 98.7 | 94.4 |
| Charging (0.2P) | 3488 | 96.7 | 99.4 | 96.5 |
| Discharging (0.5P) | 2988 | 7.8 | 2.5 | 7.7 |
| Discharging (0.5P) | 2821 | 0.6 | 0.2 | 0.5 |
Experimental Validation and Strategy Comparison
An unbalanced state was artificially constructed in three cabinets. Cabinet 1 was managed with MVBS, Cabinet 2 with AVBS, and Cabinet 3 with the proposed SCBS strategy. The cabinets operated under a standard 2-cycle-per-day industrial load profile.
The MVBS strategy only activates balancing during charging when a voltage difference exceeds 50mV, turning off when the difference drops to 30mV. The AVBS strategy activates balancing when a cell’s voltage exceeds the average voltage by 50mV, turning off at 30mV. The SCBS strategy performs full-time balancing according to its calculated schedule, iterating and correcting based on dynamic calibration.
| Strategy | Effective Balancing Time (12 Days) | Pre-Balance Capacity (Ah) | Post-Balance Capacity (Ah) | Capacity Improvement (%) | Pre-Balance Max Voltage Deviation (mV) | Post-Balance Max Voltage Deviation (mV) |
|---|---|---|---|---|---|---|
| MVBS | 5.7 h | 262.4 | 262.8 | 0.2% | 192.5 | 191.5 |
| AVBS | 34.9 h | 263.1 | 266.2 | 1.2% | 212.4 | 205.1 |
| SCBS | 220.0 h | 261.5 | 277.1 | 5.9% | 211.2 | 150.5 |
The results clearly indicate the superiority of the SCBS. The MVBS showed almost no improvement, as the large voltage plateau of LFP batteries offers very little time for voltage-based balancing. The AVBS performed better by balancing across all states, but its reliance on average voltage failed to target the specific cells that were limiting capacity in a large 224-cell string, leaving significant capacity untapped. The SCBS, by directly targeting the cells that limit capacity based on SOC dynamic calibration, achieved a 5.9% capacity improvement, nearly five times that of the AVBS. It successfully restored the system back to its initial balanced state before the artificial unbalance was introduced.
Scalability and Stability Analysis
We further validated the SCBS strategy on battery energy storage systems with different numbers of cells to demonstrate its scalability. The results confirm that the strategy is agnostic to the module size and cell capacity, making it a robust solution for utility-scale battery energy storage systems.
| Module Size (Cells) | Cell Capacity (Ah) | Pre-Balance Max Voltage Deviation (mV) | Post-Balance Max Voltage Deviation (mV) | Pre-Balance Capacity (Ah) | Post-Balance Capacity (Ah) |
|---|---|---|---|---|---|
| 8 Serial | 314 | 176.2 | 100.3 | 275.4 | 281.1 |
| 128 Serial | 314 | 162.3 | 66.8 | 273.5 | 280.6 |
| 416 Serial | 280 | 117.5 | 50.2 | 272.1 | 277.5 |
Long-term stability is a primary concern for the deployment of any algorithm in a commercial battery energy storage system. We monitored Cabinet 3 for 120 days after the SCBS completed its initial 12-day balancing phase. The discharge capacity and charging cut-off voltage deviation remained stable and consistent with the balanced state, showing no signs of abnormal performance or degradation.
Conclusion
We have developed and validated a highly effective SOC dynamic self-calibration and adaptive balancing strategy for series-connected battery energy storage systems. The SCBS strategy overcomes the fundamental limitations of voltage-based balancing methods by directly targeting the underlying SOC imbalance. Its key innovation is the real-time, self-learning capability to calibrate the SOC-to-voltage curve for the specific aging state of the battery energy storage system, eliminating the inaccuracies of static offline curves. This enables precise identification of capacity-limiting cells and optimal allocation of balancing time. Our comprehensive experiments on 200kWh cabinets demonstrate that the SCBS achieves a capacity improvement of 5.9% within 12 days, significantly outperforming conventional MVBS and AVBS strategies. The strategy is robust, scalable, and ensures long-term stability, making it an ideal solution for maximizing the performance and economic return of modern battery energy storage systems.