As the “new engine” of China’s dual-carbon goals, renewable energy primarily comes from wind and solar power, which suffer from intermittency and uncontrollability. Therefore, efficient energy storage technologies must be developed to complement renewable energy expansion. Electrochemical energy storage, dominated by lithium-ion batteries, is the most widely applied and promising technology. However, global lithium resources are unevenly distributed, and China faces a significant supply gap with high external dependence. Sodium-ion batteries (SIBs) offer abundant resources, high safety, excellent high- and low-temperature performance, superior high-rate charge/discharge capability, and low maintenance costs, effectively addressing lithium resource shortages.
In this work, we focus on the integrated technologies of a novel sodium-ion battery energy storage system. We optimize the liquid cooling plate structure and thermal management strategy, propose an intelligent string-type topology for the battery energy storage system, design a bidirectional converter suitable for the wide voltage range of SIBs, develop a capacity estimation method considering cell inconsistency, and implement a SOC-based power coordination control for multiple battery strings. These technologies are applied to a 2.5 MW / 10 MWh sodium-ion battery energy storage system.
Thermal Management of Sodium-Ion Battery Energy Storage System
Thermodynamic Characteristics During Charge/Discharge
We tested the temperature variation of a sodium-ion battery during charging and discharging at room temperature (about 20 °C) with a 0.25 C constant-power cycle. During charging, the battery temperature gradually dropped to ambient temperature, indicating a weak endothermic effect, which differs from lithium-ion batteries. During discharging, the temperature rose significantly by about 9.2 °C, indicating an exothermic reaction similar to lithium-ion batteries. We further tested discharge temperature rises at various ambient temperatures. Table 1 summarizes the results.
| Temperature (°C) | Discharge Temperature Rise (°C) |
|---|---|
| -30 | 23.4 |
| -20 | 25.3 |
| -10 | 21.2 |
| 0 | 18.2 |
| 10 | 15.7 |
| 25 | 11.2 |
| 35 | 12.2 |
| 45 | 13.7 |
| 55 | 13.5 |
The minimum temperature rise occurs at 25 °C; at low temperatures, the rise increases due to higher internal impedance. Above 35 °C, the rise is slightly higher than at room temperature.
Comparison of Air Cooling and Liquid Cooling
We tested both air-cooled and liquid-cooled sodium-ion battery clusters under cyclic charge/discharge to evaluate voltage and temperature consistency. In the air-cooled cluster, the voltage range during charging varied from 17 mV to 67 mV, and during discharging from 10 mV to 92 mV; temperature differences ranged from 3 °C to 6 °C. In the liquid-cooled cluster, the voltage range was significantly narrower: 9 mV to 42 mV (charging) and 7 mV to 43 mV (discharging); temperature differences were only 1 °C to 4 °C. Liquid cooling dramatically improves both voltage and temperature uniformity, making it more suitable for the battery energy storage system.
Liquid Cooling System Design Based on Thermal Simulation
We adopted a non-contact liquid cooling system using multi-port flat tubes arranged around the battery modules. Each module uses a 1P13S configuration with cell OCV difference within 3 mV and internal resistance 0.15–0.25 mΩ. Each cell has one temperature sensor, and 2 mm aerogel is used for thermal insulation between cells. The battery pack uses a 1P52S configuration with 52 cells and 57 temperature sensors. We performed thermal simulations for the liquid cooling system at different flow rates (minimum, nominal, and maximum). Figure 1 shows a representative thermal simulation result.
The simulation shows that the temperature rise from inlet to outlet is about 2.4 °C, and at the recirculation zone near the inlet, the temperature rise is about 2 °C higher. The overall temperature difference across the battery pack remains within 3 °C, confirming the effectiveness of the liquid cooling design for the battery energy storage system.
Power Converter Technology for Sodium-Ion Battery Energy Storage System
Intelligent String-Type Topology
Conventional centralized PCS topology connects battery clusters in parallel, leading to current imbalance and the “bucket effect”. For SIBs, inconsistency is more pronounced. We propose an intelligent string-type topology where each battery cluster is paired with a dedicated PCS. This architecture decouples clusters, eliminates circulating currents, improves safety and cycle life, and supports mixing of new and old clusters. The system efficiency increases and the required battery capacity is reduced by about 6% compared to conventional designs. In the string-type battery energy storage system, if one PCS fails, the total power command is redistributed to other units, maintaining high availability.
PCS Design for Wide Voltage Range of SIBs
Sodium-ion batteries have a wide voltage range (e.g., 2.35 V to 3.75 V per cell). Conventional PCS designs suffer from poor response, high voltage ripple, and low efficiency under such conditions. We developed a configurable LLC circuit that adapts to different operating conditions via external switching. This reduces bus voltage, lowers switching losses, and improves efficiency. Key performance indices of the PCS were tested and all meet standard requirements, as shown in Table 2.
| Test Item | Requirement | Result |
|---|---|---|
| Three-phase voltage unbalance | GB/T 15543—2008 | Pass |
| Total harmonic distortion | GB/T 14549—1993 | Pass |
| Voltage deviation | -10% ~ +15% | Pass |
| Rectifier & inverter efficiency | ≥ 94% | Pass |
| DC component | ≤ 0.5% of rated current | Pass |
| Voltage fluctuation & flicker | GB/T 12326—2008 | Pass |
| Current accuracy & ripple | Accuracy ±5%, ripple ≤ 5% | Pass |
Monitoring Technology for Sodium-Ion Battery Energy Storage System
Capacity Estimation Considering Cell Inconsistency
The OCV-SOC curve of SIBs (layered oxide cathode) is nearly linear without a plateau, facilitating SOC estimation. However, cell inconsistency affects system capacity. We propose a method that extracts features from each cell’s charging curve: first-class features representing state capacity, and second-class features representing inconsistency. The distance correlation coefficient is used to select features strongly correlated with battery pack capacity:
$$
dCor(X,Y) = \frac{dCov(X,Y)}{\sqrt{dVar(X) \, dVar(Y)}}
$$
where X is the feature matrix and Y is the pack capacity vector. We then build a Gaussian process regression model with a squared exponential kernel to map the selected features to pack capacity. The model is trained on data under various temperatures and charge rates. Online, real-time features are extracted from charging curves to estimate the pack capacity of the battery energy storage system.
SOC-Based Power Coordination Control for Multiple Strings
Due to SOH differences among battery clusters, power distribution becomes challenging. We developed a two-step coordination method. First, based on the total power demand from the energy management system, each PCS receives an initial power allocation proportional to its cluster’s SOC. Second, the remaining unassigned power is redistributed to match the charging/discharging capabilities and SOCs of each cluster. The algorithm ensures balanced operation and improves consistency of the battery energy storage system.
Energy Management System for Sodium-Ion Batteries
We constructed a decentralized energy management system (EMS) for the sodium-ion battery energy storage system, comprising a station-level monitoring host and local monitoring units. The EMS integrates data cleaning, screening, and analysis to optimize real-time power allocation. It achieves unified control of battery cluster energy balancing, system charge/discharge power, and grid fault ride-through. The architecture relieves the host computer’s burden and improves reliability.
Application Verification of the Sodium-Ion Battery Energy Storage System
We integrated the proposed technologies into a 2.5 MW / 10 MWh sodium-ion battery energy storage system. The system consists of two 1.25 MW/5 MWh subsystems. Each subsystem contains 44 strings, each string being a battery cluster (1P260S, 5 battery packs) connected to a dedicated PCS. In total, 88 PCS units are used. The overall electrical connection is shown schematically in the system design. Key system parameters are listed in Table 3.
| Parameter | Value |
|---|---|
| Rated charge/discharge hour rate | 4 h |
| Rated charge power | 2.5 MW |
| Rated discharge power | 2.5 MW |
| Rated charge energy | 10 MWh |
| Rated discharge energy | 10 MWh |
| Nominal battery system voltage | 780 V |
| Cell charge termination voltage | 3.75 V |
| Cell discharge termination voltage | 2.35 V |
| Cluster charge termination voltage | 975 V |
| Cluster discharge termination voltage | 611 V |
| Energy conversion efficiency at rated power | ≥ 90% |
| Charge/discharge response time | < 0.5 s |
| Charge/discharge regulation time | < 0.5 s |
| Charge/discharge transition time | < 1 s |
| Operating temperature range | 15–35 °C |
| Rated AC current | 3788 A |
| Power factor | ±0.95 |
| Total harmonic distortion (full load) | < 5% |
Intelligent String Branch Performance Test
We tested a single string (battery cluster + dedicated PCS) for energy efficiency and voltage consistency. Results are in Table 4.
| Item | Result |
|---|---|
| DC charge energy (kWh) | 134.69 |
| DC discharge energy (kWh) | 130.73 |
| AC charge energy (kWh) | 138.22 |
| AC discharge energy (kWh) | 127.80 |
| DC energy efficiency (%) | 97.06 |
| AC energy efficiency (%) | 92.46 |
| Max cell voltage difference during charging (mV) | 44 |
| Max cell voltage coefficient during charging (%) | 1.7371 |
| Max cell voltage difference during discharging (mV) | 46 |
| Max cell voltage coefficient during discharging (%) | 1.9339 |
The branch achieves over 90% AC energy efficiency and maintains excellent voltage consistency, demonstrating the effectiveness of the intelligent string architecture for the battery energy storage system.
System Static Test
We conducted factory static tests on the battery compartments, PCS compartments, and fire protection system. All items passed, as summarized in Table 5.
| Test Item | Result |
|---|---|
| Battery cluster power-on check | Pass |
| Battery cluster communication test | Pass |
| Battery cluster fault simulation | Pass |
| Battery compartment liquid cooling system check | Pass |
| PCS power-on check | Pass |
| PCS communication test | Pass |
| PCS fault simulation | Pass |
| Fire system sensor check | Pass |
| Fire system fire simulation test | Pass |
System Grid-Connection Test
After interconnection with the grid dispatch center, we performed grid-connection tests. All test items met requirements, as shown in Table 6. The 2.5 MW / 10 MWh sodium-ion battery energy storage system was successfully commissioned and operates reliably.
| Test Item | Result |
|---|---|
| Power quality test | Pass |
| Fault ride-through capability test | Pass |
| Voltage adaptability test | Pass |
| Frequency adaptability test | Pass |
| Active power control capability test | Pass |
| Reactive/voltage control capability test | Pass |
| AGC/AVC test | Pass |
| Primary frequency regulation test | Pass |
Conclusion
We have systematically developed integrated technologies for a novel sodium-ion battery energy storage system. The non-contact liquid cooling system with multi-port flat tubes maintains module temperature differences within 3 °C. The intelligent string-type topology and the wide-voltage-range PCS achieve over 90% energy efficiency and excellent dynamic performance. The capacity estimation method using Gaussian process regression accurately reflects cell inconsistency, and the SOC-based power coordination control ensures balanced operation. The decentralized energy management system unifies battery balancing, power control, and grid fault ride-through. These technologies were successfully validated on a 2.5 MW / 10 MWh sodium-ion battery energy storage system, which has been commissioned and operates smoothly, demonstrating the viability of sodium-ion battery energy storage system for large-scale grid applications.