We propose a novel string-type battery energy storage system to address the critical issues of parallel mismatch between battery clusters, low charge-discharge efficiency, and high operational costs in conventional centralized architectures. This system adopts a double-stage configuration combining single-cluster management with modular power conversion systems (PCS), augmented by a refined battery management system (BMS) and pack-level optimizers. Extensive testing demonstrates that the optimizer improves the average charge-discharge capacity of a battery cluster by 2.37% and the overall round-trip efficiency by 2.26%. The proposed battery energy storage systems have been successfully deployed in extreme environments—including high temperatures, severe cold, and high altitudes—and across diverse applications such as frequency regulation, shared energy storage, peak shaving, and commercial backup. This work provides a new pathway for the advancement of battery energy storage systems in modern power grids.
Introduction
As the transformation of energy structure accelerates, battery energy storage systems are becoming increasingly vital in modern power grids. Current battery energy storage systems are mainly classified into centralized and string-type topologies. In a centralized battery energy storage system, multiple battery modules are connected in parallel and then linked to a large PCS. However, electrical parameter mismatch among modules leads to capacity degradation and reduced round-trip efficiency. Conventional string-type battery energy storage systems typically place a controller only at the single battery cluster level, allowing each cluster to output its maximum capacity, but the electrical imbalance among battery packs within a cluster remains unresolved. To overcome these limitations, we introduce a new string-type battery energy storage system that integrates an optimizer at each battery pack and a controller at each battery cluster, forming a double-stage architecture of single-cluster management plus modular PCS. By thoroughly analyzing the technical principles, performance advantages, and application prospects of this battery energy storage system, we aim to enhance overall performance, optimize electrical balance among battery packs, and provide a novel direction for the evolution of battery energy storage systems.
Analysis of Energy Storage Technologies
Advantages of String-Type Battery Energy Storage Systems
Compared with centralized battery energy storage systems, string-type configurations offer superior performance in discharge capacity, investment cost, operation and maintenance, safety, system design, wiring, and management.
Discharge performance: The string-type battery energy storage system employs pack-level optimization technology that effectively avoids capacity loss caused by series mismatch within a battery cluster (typically, 18 cells in series form a battery pack, and 21 packs in series form a battery cluster). The intelligent single-cluster controller further prevents circulation-induced capacity loss. As a result, the total discharged energy over the entire lifecycle of the string-type battery energy storage system is more than 6% higher than that of a centralized system. In contrast, centralized battery energy storage systems lack pack-level equalization design, leading to series mismatch that prevents full charge and full discharge, thereby reducing the actual usable capacity below the installed capacity. Moreover, because battery clusters are directly paralleled without cluster-level voltage regulation, inter-cluster differences further degrade the available capacity and power of the system.
Investment cost: String-type battery energy storage systems adopt a high-density factory pre-installation solution, reducing on-site delivery cost to 0.01–0.03 RMB/Wh. They also support mixing new and old batteries and flexible battery replenishment, allowing the initial battery capacity to be reduced by 30% compared to centralized battery energy storage systems, which require an initial over-provision of more than 30% due to infeasibility of replenishment.
Operation and maintenance: String-type battery energy storage systems enable extremely simple maintenance—replacing a battery pack does not require manual adjustment of state of charge (SOC) or periodic SOC calibration. Technicians can directly swap packs on site, reducing operation and maintenance costs by more than 90%. In contrast, centralized battery energy storage systems demand manual SOC adjustment of spare packs and regular SOC calibration, incurring high labor and time costs.
Safety: Our string-type battery energy storage system incorporates an advanced internal short-circuit detection function using AI outlier algorithms. This algorithm detects both sudden and derived internal short circuits, providing accurate early warning and protection, thereby improving system safety by over 90%. In comparison, centralized battery energy storage systems struggle to identify internal short circuits, posing a higher fire risk.
System design: The string-type battery energy storage system employs modular design for temperature control, battery system, and PCS, achieving system availability exceeding 99.9%. By contrast, a PCS failure in a centralized battery energy storage system can affect the entire container, limiting availability to 97%–98%.
Wiring and management: The string-type battery energy storage system automatically generates AC and DC wiring topology diagrams, eliminating manual configuration. The design visually presents connection status and anomaly alerts, facilitating fault diagnosis. In terms of management, 3D graphical technology displays cluster-level and pack-level data—including real-time SOC and charge-discharge power of every pack—and monitors cell SOH, SOC, temperature, and voltage. Centralized battery energy storage systems only display the highest and lowest module SOC values, with coarse granularity and no quick visual fault localization.
Optimizer Performance Testing
To validate the enhancement provided by the battery pack optimizer, we selected two battery clusters from a string-type battery energy storage system for a comparative test. Each battery pack includes the case, cells, a battery management unit (BMU), and an optimizer. The parameters of a single battery pack are listed in the following table.
| Parameter | Value |
|---|---|
| Cell configuration | 1P18S |
| Nominal voltage (V) | 57.6 |
| Nominal capacity (Ah) | 280 |
| Nominal energy (kWh) | 16.13 |
| Dimensions (L×W×H) (mm) | 442×307×660 |
Each battery cluster consists of a rack and 21 battery modules (including BMS and optimizer). The cluster parameters are given below.
| Parameter | Value |
|---|---|
| Nominal voltage (V) | 1075.2 |
| Nominal capacity (Ah) | 280 |
| Nominal energy (kWh) | 338.7 |
| Dimensions (H×L×W) (mm) | 1565.0×2475.0×787.5 |
| Optimal operating temperature (°C) | 10–40 |
We collected charge-discharge data over six months for the two clusters, comparing performance with the optimizer enabled versus disabled. The average charge-discharge energy results are shown in the next table.
| Optimizer State | Nominal Energy (kWh) | Average Charge-Discharge Energy (kWh) |
|---|---|---|
| Disabled | 338.7 | 320.1 |
| Enabled | 338.7 | 327.7 |
From the table, enabling the optimizer increases the average charge-discharge energy by:
$$\text{Improvement} = \frac{327.7 – 320.1}{320.1} \times 100\% \approx 2.37\%$$
We also recorded the grid-connected energy (supplied to the grid) and grid-consumed energy (including auxiliary consumption) over one month for the same clusters, and computed the comprehensive round-trip efficiency. The results are presented below.
| Optimizer State | Energy Supplied to Grid (kWh) | Energy Consumed from Grid (kWh) | Round-Trip Efficiency (%) |
|---|---|---|---|
| Disabled | 46236.6 | 58527.4 | 79.00 |
| Enabled | 48224.8 | 59346.7 | 81.26 |
The efficiency improvement is:
$$\text{Efficiency Improvement} = 81.26\% – 79.00\% = 2.26\%$$
These experimental results clearly confirm that the optimizer significantly boosts both the energy throughput and the overall efficiency of the battery energy storage systems.
Mainstream PCS Architectures
Currently, large-scale battery energy storage systems employ three mainstream PCS architectures: centralized PCS, single-cluster management with modular PCS double-stage, and single-cluster management with modular PCS single-stage. Since the double-stage architecture encompasses all functions of the single-stage, we focus on the first two. A schematic representation of these architectures is provided below.
In the centralized PCS architecture, battery modules are connected in series to form a cluster, each cluster is equipped with a high-voltage box, multiple clusters are paralleled and then connected to a single large centralized PCS, after which the AC voltage is stepped up to 35 kV. In the double-stage architecture, each cluster is connected to a cluster controller (DC/DC converter), multiple clusters are paralleled and then connected to a modular PCS, forming the single-cluster management plus modular PCS double-stage topology, followed by AC step-up to 35 kV.
Experimental Results and Discussion
Comparison of PCS Architectures
We compare the centralized versus the double-stage architecture across several key metrics: discharge energy, system configuration flexibility, grid performance, maintainability, availability, and safety. The comparison is summarized in the following table.
| Aspect | Centralized PCS | Single-Cluster Management + Modular PCS (Double-Stage) |
|---|---|---|
| Discharge energy | Low – cluster parallel mismatch causes available capacity and power loss. | High – eliminates cluster parallel mismatch; discharge energy >6% higher than centralized. |
| System configuration flexibility | Poor – cannot mix old and new batteries; capacity expansion simultaneously requires both battery and PCS changes, affecting AC capacity. | Excellent – allows different clusters with different configurations, mixing old and new batteries; expansion does not impact AC capacity. |
| Grid performance | Meets GB 51048-2014 standard. | Meets GB 51048-2014; additionally, charge-discharge power variation before and after high-voltage ride-through remains below 10%. |
| Maintainability | Poor – large PCS requires professional on-site maintenance. | Good – both DC/DC converters and modular PCS are easily maintainable. |
| Availability | Low – centralized PCS failure cannot be isolated individually. | High – modular design allows fault isolation; availability >99.9%. |
| Safety | Low – no individual isolation; inter-cluster circulation risk. | High – eliminates inter-cluster circulation. |
Clearly, the double-stage architecture offers significant advantages for modern battery energy storage systems.
Advantages of the Battery Management System (BMS)
BMS structure: The BMS is the core of our string-type battery energy storage system. It consists of four hierarchical levels: Battery Management Unit (BMU), Battery Cluster Unit (BCU), Cluster Management Unit (CMU), and Smart Array Controller Unit (SACU). Each battery pack is equipped with one BMU (Level 1), responsible for real-time monitoring of physical parameters, SOC estimation, online diagnosis, charge/discharge and pre-charge control, balancing, and thermal management, while communicating with the BCU. Each battery cluster has one BCU (Level 2), which collects operational data, performs fault diagnosis, data storage, and protection. The BCU independently controls each cluster to avoid bias and the “short plank effect,” and it adjusts voltage flexibly to ensure stable and continuous power output. The system includes one set of CMU+SACU (Level 3), where the CMU manages data analysis, alarming, and protection (e.g., overvoltage, undervoltage, overcurrent, insulation faults), and communicates via fiber optic with the SACU. The SACU (Level 4) serves as the advanced control layer, coordinating the entire sub-array of the battery energy storage system. It connects to the CMU and PCS, receives dispatch commands from the EMS, and manages energy flow, including inter-cluster balancing. It also interfaces with local APP for near-field control.
BMS functions: The BMS provides high-precision acquisition of cell temperature, cell voltage, pack voltage, cluster voltage, and current, enabling optimized operation. It estimates SOC and SOH with automatic calibration, and displays cycle count, depth of discharge (DOD), SOC, and SOH in real time. Based on SOC planning, it accurately controls charging/discharging and performs fault diagnosis. The distributed thermal management system—with door-mounted air conditioners per cluster and adjustable fans inside packs—keeps the temperature difference among packs within the container below 3 °C (at 0.5C charge rate). The BMS also has multi-level balancing functions (cell, pack, cluster) to mitigate capacity loss and circulation risks. It supports Ethernet, fiber optic, and RS485 communication with the energy management system, and can store up to 5000 events and 30 days of historical data, satisfying GB/T 34131-2023 requirements.
PCS Advantages
The PCS in our string-type battery energy storage system features single-cluster management, enabling independent SOC calibration for each cluster and eliminating inter-cluster circulation that compromises safety and battery life. It also allows flexible DC-side battery replenishment without directly paralleling old and new clusters. The PCS is modular, with each unit rated at ≤250 kW, enhancing maintainability and overall system availability.
System Configuration and Application Cases
Our string-type battery energy storage system has been successfully deployed in extreme environments—including Hainan (high temperature), Jilin (severe cold), and Qinghai (high altitude)—and across diverse application scenarios such as spinning reserve and frequency regulation in Singapore, shared energy storage in Hubei, grid peak shaving, and commercial backup in Jiangsu. A typical containerized configuration integrates 2.032 MWh of battery capacity in a 0.3048 m standard container. The system components are listed below.
| Component | Model/Specification |
|---|---|
| Container | LUNA2000-2.0 MWh |
| Battery | 3.2 V/280 Ah cells, 1P18S module, 21 modules per cluster, 6 clusters |
| BMS | Includes BMU, BCU, CMU, and communication harnesses |
| Cluster controller | DC/DC converter (each with two independent inputs) |
| HVAC | Distributed air conditioning |
| Fire suppression | Smoke detectors, temperature sensors, fire alarm panel, FK-5-1-12 gas, alarm bell, gas release indicator, manual/input modules |
The layout of the containerized battery energy storage system is designed for scientific safety and green operation, with distributed air conditioners placed at each battery cluster position.
Conclusion
We have proposed a novel string-type battery energy storage system that overcomes the limitations of traditional centralized architectures. By employing a double-stage topology of single-cluster management combined with modular PCS, and integrating pack-level optimizers, the system significantly mitigates series mismatch among battery clusters. Test results demonstrate that the optimizer increases the average charge-discharge energy by 2.37% and the round-trip efficiency by 2.26%. The system has been validated in extreme environmental conditions and diverse application scenarios, proving its robustness and high efficiency. This work offers a new practical approach for the advancement of battery energy storage systems. Future efforts will focus on further optimizing system design, expanding application domains, and promoting the widespread adoption of high-performance battery energy storage systems.