In modern power systems with high penetration of renewable energy, the battery energy storage system (BESS) plays a crucial role in enhancing power quality, grid stability, and economic efficiency. The widespread deployment of BESS demands cost-effective, efficient, and scalable power conversion architectures. Among existing configurations, the decentralized BESS architecture—where multiple battery clusters are each interfaced to a common DC bus through dedicated DC/DC converters—offers an effective solution to mitigate circulating currents among parallel battery clusters. However, conventional DC/DC converters in such architectures suffer from high cost, low efficiency, and complex topologies, especially when full-power converters are used. To address these challenges, we propose a novel multiport equalization converter (MEC) that achieves battery cluster power management through low-cost half-bridge chopper circuits while eliminating the need for an external power supply to maintain DC bus energy balance. This paper presents the topology, operating principles, control strategies, parameter design, and experimental validation of the proposed MEC.
The proposed MEC is specifically designed for decentralized BESS, where multiple independent battery clusters are connected to a single PCS through a common intermediate DC bus. By processing only a fraction of the total power (partial power processing), the converter reduces component stress and improves overall system efficiency. The MEC topology comprises one PCS-side half-bridge and multiple battery-side half-bridge choppers, all sharing a common intermediate DC-link capacitor. This structure enables bidirectional power flow and autonomous energy balance of the intermediate capacitor without any additional auxiliary power supply or isolation transformer. In the following sections, we detail the mathematical modeling, control design, and experimental verification of the MEC within a BESS context.
Topology and Operating Principle
The MEC topology, as shown in Figure 3 of the original work, consists of an intermediate DC bus capacitor and several half-bridge chopper circuits. One half-bridge (PCS-side) connects the intermediate capacitor to the PCS DC bus, while the remaining half-bridges (battery-side) connect the intermediate capacitor to individual battery clusters. By regulating the duty cycles of the switches, we can independently control the current of each battery cluster and simultaneously maintain the intermediate bus voltage at a stable level. The voltage of the intermediate bus, denoted $$v_{sc}$$, serves as the energy exchange hub. The battery cluster voltage equation is given by:
$$v_{pc} d_1 – v_{sc} = v_{OCVj} + r_{bat} i_j + L \frac{di_j}{dt}$$
where $$v_{pc}$$ is the PCS DC bus voltage, $$d_1$$ is the duty cycle of the PCS-side half-bridge upper switch, $$v_{OCVj}$$ is the open-circuit voltage of the j-th battery cluster, $$r_{bat}$$ is the equivalent internal resistance, $$i_j$$ is the cluster current, and $$L$$ is the filter inductance. The intermediate capacitor dynamics are described by:
$$C_{sc} \frac{dv_{sc}}{dt} = d_1 i_{be} – \sum_{j=1}^{n} d_{j1} i_j$$
where $$C_{sc}$$ is the intermediate capacitance, $$i_{be}$$ is the total BESS current, and $$d_{j1}$$ is the duty cycle of the j-th battery-side half-bridge upper switch. These equations form the basis for the average current control and cluster current control.
Control Strategy
We propose a hierarchical control framework consisting of three loops: average current control, battery cluster current control, and intermediate bus voltage balance control.
Average Current Control
The average current of all battery clusters, $$i_{avg} = \frac{1}{n}\sum_{j=1}^{n} i_j$$, is regulated by adjusting the PCS-side duty cycle d1. The small-signal transfer function from $$\hat{v}_{pc}$$ to $$\hat{i}_{avg}$$ is derived as:
$$G_{avg}(s) = \frac{\hat{i}_{avg}(s)}{\hat{v}_{pc}(s)} = \frac{1}{L s + r_{bat}}$$
A proportional-integral (PI) controller is used to track the average current reference. The bandwidth of this loop is designed to be significantly higher than that of the cluster current loops to achieve decoupling.
Battery Cluster Current Control
To equalize the currents among clusters, we define the deviation current $$\Delta i_j = i_j – i_{avg}$$ and control it via the battery-side duty cycle $$d_{j1}$$. From the small-signal model, the transfer function from $$\hat{d}_{j1}$$ to $$\hat{i}_j$$ is:
$$G_{I,j}(s) = \frac{\hat{i}_j(s)}{\hat{d}_{j1}(s)} = \frac{V_{sc}}{L s + r_{bat}}$$
Since there are n clusters but only n-1 independent deviation currents (the sum of deviations is zero), we treat one cluster as the balancing cluster. The duty cycle of this balancing cluster is determined by the sum of the others, ensuring overall consistency. This approach effectively decouples the cluster currents under the assumption that the average current loop is much faster.
Intermediate Bus Voltage Balance Control
The intermediate bus voltage $$v_{sc}$$ must be maintained constant to allow proper power exchange. By adjusting the PCS-side duty cycle $$d_1$$, we can control the net power flowing into the intermediate capacitor. The transfer function from $$\hat{d}_1$$ to $$\hat{v}_{sc}$$ is:
$$G_{cs}(s) = \frac{\hat{v}_{sc}(s)}{\hat{d}_1(s)} = \frac{I_{be}}{C_{sc} s}$$
where $$I_{be}$$ is the steady-state total BESS current. A PI controller with a hysteresis compensator to handle current direction is employed to regulate $$v_{sc}$$ to its reference value.
Parameter Design
The key parameters of the MEC, including the intermediate bus voltage $$V_{sc}$$, capacitance $$C_{sc}$$, and filter inductance $$L$$, are selected based on system requirements. The intermediate bus voltage must be sufficiently high to cover the variation range of battery cluster voltages. For a typical lithium-ion battery with SOC ranging from 10% to 90%, the voltage variation is about ±10% of the nominal voltage. Thus, we set:
$$V_{sc} = \max\left( \frac{v_{OCV,10} – v_{OCV,90}}{2}, \frac{v_{OCV,10} + v_{OCV,90}}{2} \right)$$
The capacitance $$C_{sc}$$ is designed to limit the voltage ripple within a specified tolerance $$\alpha$$. Considering the worst-case scenario where half of the clusters are sourcing current while the other half are sinking, the maximum voltage ripple is:
$$\Delta v_{sc} = \frac{n}{2} \frac{I \Delta t_{\max}}{C_{sc}} \leq \alpha V_{sc}$$
where $$\Delta t_{\max}$$ is the maximum energy transfer time within one switching period. With symmetrical carrier modulation, $$\Delta t_{\max} = \frac{|d_{21} – d_1|}{2} T_s$$. Hence, the required capacitance is:
$$C_{sc} = \frac{n I \Delta t_{\max}}{2 \alpha V_{sc}}$$
The filter inductance $$L$$ is chosen to limit the current ripple $$\beta$$:
$$L \leq \frac{V_{sc} \Delta t_{\max}}{\beta I}$$
For our experimental prototype, the parameters are listed in Table I.
| Parameter | Symbol | Value |
|---|---|---|
| Number of battery clusters | n | 3 |
| Switching frequency | fs | 10 kHz |
| Battery cluster nominal voltage | E | 100 V |
| Battery cluster rated current | I | 6 A |
| Intermediate bus voltage reference | Vsc | 10 V |
| Filter inductance | L | 2 mH |
| Intermediate capacitance | Csc | 10.8 mF |
Simulation and Experimental Verification
We validated the proposed MEC through both simulation and a scaled-down experimental prototype. The simulation model was built in MATLAB/Simulink with three battery clusters (LF280K cells). The system parameters are summarized in Table II.
| Parameter | Symbol | Value |
|---|---|---|
| Number of battery clusters | n | 3 |
| Battery cluster rated voltage | E | 1331 V |
| Battery cluster rated current | I | 70 A |
| Cell capacity | C | 280 Ah |
| PCS rated power | P | 1725 kW |
| Intermediate bus voltage | Vsc | 133 V |
| Intermediate capacitance | Csc | 9 mF |
| Filter inductance | L | 1.9 mH |
Simulation Results: Two cases were simulated: (1) identical SOC but different internal resistances among clusters, and (2) different SOCs (55%, 60%, 65%) with otherwise identical parameters. The simulation sequence consisted of five operating modes:
– Mode 1 (0–0.8 s): Average current 0.5 p.u., MEC inactive.
– Mode 2 (0.8–1.6 s): MEC activated, current equalization enabled.
– Mode 3 (1.6–2.4 s): Average current increased to 1 p.u.
– Mode 4 (2.4–3.2 s): Current reversed to –1 p.u.
– Mode 5 (3.2–4.0 s): Light load at –0.2 p.u.
The simulation waveforms show that, in Mode 1, cluster currents differed by up to 6.95 A due to natural distribution. After MEC activation, all cluster currents converged to the same value (35 A in Mode 2, 70 A in Mode 3, –70 A in Mode 4, –14 A in Mode 5), demonstrating the effectiveness of the cluster current control. The intermediate bus voltage remained stable at 133 V with a maximum ripple of 2.6 V during the most severe transient, confirming the energy balance control. In the SOC-difference case, the MEC successfully reduced the current imbalance from 9.7 A to zero, accelerating SOC convergence. The simulation results confirm that the MEC can handle bidirectional power flow and maintain both average current and intermediate bus voltage regulation under all tested conditions.
Experimental Results: A 1.8 kW scaled-down prototype was built, as shown in Figure 9 of the original work. The experimental parameters are as per Table I. The three battery clusters were emulated with the same SOC but different SOH, leading to natural current imbalance. The same five-mode sequence was applied. The experimental waveforms (Figure 10) show:
– In Mode 1 (before MEC activation), the maximum current difference was 1.9 A (0.316 p.u.).
– After activation (Mode 2), all cluster currents became equal at 3 A (0.5 p.u.).
– In Mode 3, the total current increased to 18 A (1 p.u.) while cluster currents remained balanced.
– In Mode 4, the current reversed to –18 A (–1 p.u.) within 22 ms, meeting the standard IEC 62898-2 requirement.
– In Mode 5, the current reduced to –3.6 A (–0.2 p.u.) with balanced distribution.
The intermediate bus voltage was maintained at 10 V with negligible drift, and the PCS DC bus voltage stayed at 100 V. The grid-side waveforms showed correct power reversal during Mode 4.
The experimental results closely match the simulation predictions, validating the proposed MEC topology and its control strategy. The MEC not only equalizes battery cluster currents but also ensures stable intermediate bus voltage without any external power supply, thereby reducing system cost and complexity.
Conclusion
In this paper, we have proposed a multiport equalization converter (MEC) tailored for decentralized battery energy storage systems. The MEC architecture, built from half-bridge choppers and a shared DC-link capacitor, enables partial power processing and eliminates the need for an external power source or high-frequency isolation transformer. Through systematic mathematical modeling, we derived the control laws for average current regulation, individual cluster current balancing, and intermediate bus voltage stabilization. Simulation and experimental results on a 1.8 kW prototype confirmed that the MEC achieves bidirectional power flow, maintains cluster current equalization under various SOC and SOH conditions, and robustly regulates the intermediate bus voltage. The proposed solution offers a low-cost, efficient, and scalable approach for large-scale BESS deployment. Future work will focus on fault-tolerant operation and integration with grid dispatch strategies.