As a pivotal driver for achieving China’s “dual-carbon” goals, renewable energy is primarily derived from wind and solar power. However, their intermittent and uncontrollable nature necessitates the development of efficient energy storage technologies that can support the large-scale integration of renewable sources. The lithium-ion battery, a cornerstone of electrochemical energy storage, faces challenges due to the uneven global distribution of lithium resources and high import dependency. In contrast, sodium-ion batteries (SIBs) have emerged as a promising alternative, offering abundant resources, high safety, excellent performance under extreme temperatures, high-rate charge-discharge capabilities, and low maintenance costs. This paper presents a comprehensive study on the integrated technologies for a novel sodium-ion battery energy storage system, focusing on thermal management, power conversion, and system monitoring to address the unique characteristics of SIBs.
1. Thermal Management for Sodium-Ion Battery Systems
Understanding the thermodynamic behavior of SIBs is fundamental to designing effective thermal management solutions. Our experimental analysis of the temperature variation during charge and discharge cycles, conducted at an ambient temperature of 20°C with a 0.25 C-rate, revealed distinct characteristics. The battery energy storage systems exhibited a slight temperature drop during charging, indicating an endothermic process, while a significant temperature rise of approximately 9.2°C was observed during discharge, confirming an exothermic reaction. This behavior differs markedly from lithium-ion batteries, particularly during charging. Further tests across a range of temperatures from -30°C to 55°C showed that the temperature rise during discharge is minimized at 25°C and increases substantially at lower temperatures due to higher internal impedance.
To evaluate the suitability of different cooling methods for SIB battery energy storage systems, we compared air cooling and liquid cooling. Tests on both air-cooled and liquid-cooled battery clusters highlighted the superior performance of liquid cooling. Key metrics such as voltage consistency and temperature uniformity were significantly better in the liquid-cooled system.
Table 1: Comparison of Voltage and Temperature Consistency in SIB Clusters During Charge/Discharge
| Parameter | Air-Cooled System | Liquid-Cooled System |
|---|---|---|
| Voltage Difference During Charging (mV) | 17 – 67 | 9 – 42 |
| Voltage Difference During Discharging (mV) | 10 – 92 | 7 – 43 |
| Temperature Difference During Charging (°C) | 3 – 6 | 1 – 3 |
| Temperature Difference During Discharging (°C) | 1 – 6 | 1 – 4 |
Based on these findings, we adopted a non-contact liquid flow cooling system with a multi-hole flat tube structure for the battery module. This design maximizes heat exchange surface area within a compact space. Thermal simulation of the cooling system, using varying flow rates (5 L/min), demonstrated that the temperature difference across the entire battery module could be maintained within 3°C.
$$ \Delta T_{\text{module}} \leq 3 \, ^\circ\text{C} $$
2. Power Converter Technology for Sodium-Ion Battery Characteristics
The unique voltage characteristics of sodium-ion batteries, particularly their wide voltage range, necessitate specialized power conversion technology. To mitigate the “barrel effect” and circulation losses common in centralized PCS topologies, especially given the consistency challenges of SIBs, we proposed an intelligent string-type energy storage system architecture.
Table 2: Performance Metrics of the Proposed Bidirectional Power Converter (PCS)
| Test Item | Technical Requirement | Test Result |
|---|---|---|
| Three-phase Voltage Unbalance | GB/T 15543-2008 | Compliant |
| Harmonic Distortion | GB/T 14549-1993 | Compliant |
| Voltage Deviation | -10% to +15% | Compliant |
| Rectification/Inversion Efficiency | ≥94% | Compliant |
| DC Component | ≤0.5% of rated current | Compliant |
| Voltage Fluctuation and Flicker | GB/T 12326-2008 | Compliant |
| Current Stability & Ripple | ≤±5% accuracy, <5% ripple | Compliant |
This architecture creates a one-to-one correspondence between battery clusters and PCS units. For battery energy storage systems, this de-couples the clusters, enabling per-cluster management and eliminating circulation currents. In a traditional system, a single battery pack failure can disable an entire cluster. With this intelligent string topology, a single PCS failure only reduces the system’s total capacity, as the power is redistributed among the remaining units. This design improves system availability and reduces the required initial battery configuration capacity by 6%.
To address the performance issues of traditional PCS under wide voltage inputs, such as poor response and large voltage ripple, we designed a novel power converter integrating a circuit selection scheme. This scheme allows the LLC resonant circuit’s configuration to be optimized for different operating conditions, reducing bus voltage, switching losses, and core losses, thereby boosting overall efficiency and response speed.
3. Monitoring and Control Technology for SIB Energy Storage Systems
The accurate estimation of system state and intelligent power coordination are critical for the safe and efficient operation of battery energy storage systems. The OCV-SOC curve of sodium-ion batteries, unlike that of lithium iron phosphate batteries, shows a linear relationship, which is advantageous for state estimation.
We developed a capacity estimation method that accounts for cell inconsistency. This process involves extracting features from individual cell charge curves, including those representing current state capacity and those representing cell-to-cell inconsistency. A feature selection algorithm based on the distance correlation coefficient (dCor) identifies features strongly correlated with the overall pack capacity.
$$ \text{dCor}(X, Y) = \frac{\text{dCov}(X, Y)}{\sqrt{\text{dVar}(X) \cdot \text{dVar}(Y)}} $$
Using these selected features, a Gaussian process regression (GPR) model with a squared exponential kernel was trained. The model establishes a mapping between the features and the battery pack’s total capacity. This machine learning approach provides a more precise capacity estimation compared to methods that ignore cell-to-cell differences.
Table 3: Performance of Intelligent String Branch (Single Cluster + Single PCS)
| Parameter | Test Result |
|---|---|
| DC Charging Energy (kWh) | 134.69 |
| DC Discharging Energy (kWh) | 130.73 |
| AC Charging Energy (kWh) | 138.22 |
| AC Discharging Energy (kWh) | 127.80 |
| DC Energy Efficiency (%) | 97.06 |
| AC Energy Efficiency (%) | 92.46 |
| Max Voltage Difference During Charging (mV) | 44 |
| Max Voltage Difference During Discharging (mV) | 46 |
Furthermore, we proposed a multi-string power coordination control method based on the SOC of each battery cluster. The control algorithm calculates the total power demand from the energy management system and allocates it to individual PCS units. Power is first distributed based on each cluster’s SOC, and the remaining undelivered power is then redistributed in a second step. This ensures that the active and reactive power of each PCS matches its respective SOC and charge/discharge capability, balancing the state of the entire battery energy storage systems.
The SIB energy management system (EMS) we constructed uses a decentralized architecture with a station-level monitoring host and local monitoring units. This structure reduces the load on the central network and improves operational reliability. The EMS integrates the proposed SOC estimation and power coordination methods to dynamically balance energy among battery clusters and control system charging/discharging.
4. Application and Validation in a 2.5 MW/10 MWh System
The integrated technologies were successfully applied in a 2.5 MW/10 MWh SIB energy storage system. This system comprises two 1.25 MW/5 MWh sub-systems. Each sub-system is built from 44 energy storage strings, each consisting of a battery cluster (with a 1P260S configuration) and a dedicated PCS module. The system uses 88 PCS units, each connected to a single battery cluster. Commissioning tests for the intelligent string branches, static system components, and grid connection were all passed successfully.
Table 4: System Performance Parameters of the 2.5 MW/10 MWh SIB System
| Parameter | Value |
|---|---|
| Rated Charging/Discharging Duration | 4 Hours |
| Rated Charging/Discharging Power | 2.5 MW |
| Rated Charging/Discharging Energy | 10 MWh |
| Battery System Nominal Voltage | 780 V |
| Cell Charging Termination Voltage | 3.75 V |
| Cell Discharging Termination Voltage | 2.35 V |
| Battery Cluster Charging Termination Voltage | 975 V |
| Battery Cluster Discharging Termination Voltage | 611 V |
| Energy Conversion Efficiency at Rated Power | ≥90% |
| Charge/Discharge Response Time | <0.5 s |
| Charge/Discharge Transition Time | <1 s |
| Operating Temperature Range | 15 – 35 °C |
5. Conclusion
Our research on integrated technologies for novel battery energy storage systems has yielded several key results. The non-contact liquid cooling system, utilizing a multi-hole flat tube structure, effectively manages SIB thermal profiles, maintaining module temperature differences within 3°C. The intelligent string-type topology improves system efficiency and reliability by enabling per-cluster management and eliminating circulation currents. A purpose-built PCS design addresses the wide voltage range of SIBs. Finally, the developed energy management system, with its SOC-based capacity estimation and multi-string power coordination control, successfully balances charge-discharge states and ensures stable operation. These integrated technologies have been validated through the successful deployment and operation of a 2.5 MW/10 MWh SIB energy storage system, demonstrating their practical viability for large-scale energy storage applications.