Fault-Tolerant Multiport Active Bridge Converter for Resilient Energy Storage Integration in Zonal Shipboard DC System

Abstract

In this paper, the concept of a fault-tolerant multiport converter is proposed for a shipboard zonal DC system. The traditional zonal shipboard system offers a resilient power supply capability at the expense of increased cost and size. To solve this problem, the fault-tolerant multiport active bridge converter is proposed for shared energy storage between DC buses. When a short-circuit fault occurs on one bus, the energy storage can maintain uninterrupted power supply to the remaining healthy bus. With consideration of both normal operation and a fault-tolerant mode, the power transfer capability and ZVS region are analyzed. The proposed converter is compared with a traditional two-converter zonal system and multiport converter in terms of cost, volume, and efficiency. The performance of the proposed FT-MAB converter is tested through experimental verifications with the aim of validating the resilience of the power supply. The proposed FT-MAB converter achieves fault tolerance through topological reconfiguration, isolating the faulty port after the occurrence of a short-circuit fault and providing uninterrupted power supply to the healthy bus.

1. Introduction

The transition to DC shipboard power systems (SPSs) has become increasingly important due to advancements in power electronics which facilitate more efficient integration of energy storage systems (ESSs) and electric motor loads [1,2,3,4]. Unlike AC systems, DC allows for easier connection of distributed energy sources, contributing to a more compact shipboard architecture. It enhances system performance by enabling high-energy-density storage to balance power demands and provide backup during faults [5,6,7]. High-power-density storage, such as ultracapacitors, also serves to handle transient power imbalances. Moreover, DC systems also promote fuel efficiency by optimizing the operation of prime movers. With simple control, high efficiency, and better fault reconfigurability, DC microgrids are becoming the preferred choice over AC systems for modern vessels, ensuring reliable power supply for critical missions [4,5,6,7].

According to IEEE Std. 1709-2010 [1], the architecture of shipboard DC systems includes radial, ring type, zonal type, etc. Radial distribution is a simple and cost-effective architecture that is easy to redesign from a traditional AC system. However, it lacks flexibility when faults occur. In contrast, ring-type distribution offers better reliability and survivability [4]. It connects the buses in a loop, allowing the system to isolate faults without disrupting the entire network. However, since each load center is connected to only one bus, the critical loads are still prone to bus faults. In contrast, the zonal architecture offers the highest reliability, survivability, and reconfigurability [8,9,10]. As shown in Figure 1, the service loads on the ship are arranged into various zones stretching from the bow to the stern, with the DC buses installed longitudinally along the vessel. Each side of the DC buses connects with two separate bidirectional DC-DC converters, allowing for flexible reconfiguration during normal operation and fault conditions. However, the cost and size of the zonal DC system will be huge, due to the high number of DC-DC converters [9,10].

One classical bidirectional DC-DC converter is the dual active bridge (DAB) converter shown in Figure 2a. * represent the dotted terminals of the transformer. It features high power density and wide voltage range zero-voltage switching (ZVS) operation capability [11]. To gain additional benefits, such as a reduced size and cost savings, the dual active half-bridge (DAHB) converter in Figure 2b can be adopted [12]. Moreover, a CLLC resonant converter can also be adopted for both ZVS and ZCS operations [13]. However, the voltage regulation range would be narrow and voltage/current stress will also increase.

Considering the presence of multiple DC-DC converters in DC systems, one approach to reduce a system’s size and cost is to implement an integrated multiport design. Multiport converters have previously been investigated in applications such as aircraft, satellites, and electric vehicles [14,15,16,17], where compactness and efficiency are paramount. For zonal shipboard power systems, the potential multiport integration offers several advantages. On one hand, it reduces the total number of components and power conversion stages. Moreover, it allows for shared energy storage capacity. A comparison of the system’s structure is illustrated in Figure 3a,b.

Depending on the circuit topology, existing multiport converters can be classified into non-isolated [17,18,19], partially isolated [20], and fully isolated types [21,22,23]. For example, a non-isolated multiport converter has a simple topology and low component number. However, the converter efficiency and voltage range will be low. Based on the partially isolated converters in Figure 4b and fully isolated triple active bridge (TAB) converter in Figure 4c, both zero-voltage switching and a wide voltage regulation ratio can be achieved [22]. Most existing multiport converters focus on efficiency elevation and cost reduction [17,18]. However, the integrated design often comes at the expense of reliability.

The FT-MAB converter aims to tackle frequent bus short-circuit faults in shipboard DC systems using fault-tolerance. Short-circuit faults, caused by insulation degradation or aging in harsh environments, can lead to high-current surges and system collapse. Open-circuit faults, causing gradual power loss, are less urgent as the system can switch to alternate power sources. For instance, in Figure 4a, if the short-circuit fault occurs on v_bus2, it will result in the shutdown of the shared switches S₅ and S₆. Otherwise, the high fault current will flow through the antiparallel diode of S₃. Similarly, when a fault occurs on v_bus2 of Figure 4b, the shared power switches S₅ and S₆ will be blocked and the whole converter drops out. For the fully isolated TAB in Figure 4c, if a short-circuit fault occurs on v_Bus2, the primary side switches S₁₁~S₁₄ need to be blocked to limit the fault current. As a result, the healthy bus v_Bus1 will also drop out.

As shown in Table 1, to achieve both an integrated design and enhanced system reliability, the fault-tolerant multiport converter is first proposed for shipboard DC systems. In the existing studies on the control and modulation of DAB (Dual Active Bridge) converters [24,25], research has focused on optimizing control algorithms for traditional DAB or ISOP-DAB systems. However, the FT-MAB converter proposes a multi-port reconfigurable topology that achieves fault tolerance through hardware-level design, breaking through the limitations of current research, which is predominantly driven by control optimization.

According to IEEE Std. 1826-2020 [26],compared to traditional systems with independent energy storage units, the proposed FT-MAB converter ensures that at least one healthy energy path remains available during any short-circuit fault, thereby providing a robust solution to fault tolerance in shipboard DC systems.

Table 1. Contributions of this paper.

Integrated Multiport Design	Fault-Tolerant Operation	References
		[18,20,22]
		[8,21,27]
		Proposed work

Table 1. Contributions of this paper.

Integrated Multiport Design	Fault-Tolerant Operation	References
		[18,20,22]
		[8,21,27]
		Proposed work

The rest of this paper is organized as follows: in Section 2, the modulation of the proposed FT-MAB converter is analyzed in both normal and fault-tolerant modes. In Section 3, the power transfer capability and soft switching performance of the converter are evaluated in the two working modes. In Section 4, the proposed FT-MAB converter is quantitively compared with a traditional two-converter zonal system and multiport full-bridge converter, in terms of cost, volume, and power loss. The proposed converter is tested under normal operation and short-circuit conditions, verifying the proposed topology and modulation methods.

2. Operation Principles

The topology of the fault-tolerant multiport active bridge (FT-MAB) converter is shown in Figure 5. The converter consists of two half-bridges (HB), including switches S₁₁, S₁₂, S₂₁, and S₂₂, each connected to one of the two DC buses, Bus #1 or Bus #2, in the zonal DC power system. The two HBs are connected with the three-arm full-bridge (TAFB) through two transformers, respectively, including switches S₃₁, S₃₂, S₄₁, S₄₂, S₅₁, and S₅₂. The energy storage system is connected to the DC port of the TAFB. According to the operation condition, the working modes of the FT-MAB can be classified as the normal operation mode and fault-tolerant mode.

2.1. Normal Operation

During normal operation, bidirectional power can be transferred between energy storage and the two DC buses (i.e., Bus #1 and Bus #2). Moreover, mutual power support between Bus #1 and Bus #2 can also be achieved through the multiport converter, depending on the phase shift duty ratios d₁ and d₂, as analyzed in the following.

2.1.1. Same Power Direction of DC Buses

Depending on the phase shift duty ratios d₁ and d₂, the energy storage can simultaneously supply power to two DC buses or receive power from them. Without loss of generality, the case when d₁ > 0 and d₂ > 0 is considered. The key waveforms for this scenario are illustrated in Figure 6a, where the storage is charged from both buses. The switching states for this case are as follows:

Switching state 1 [t₀, t₁]: At t₀, S₁₂ is turned off and the inductor current i_Ls1 is of negative value. Within the switching dead time, the parasitic capacitor of S₁₂ is charged to bus voltage V_b and the parasitic capacitor of S₁₁ discharges to 0. As a result, S₁₁ can be turned on with zero voltage switching (ZVS). During this period, the inductor currents i_Ls1 and i_Ls2 are given as shown in (1).

$$\left\{\begin{array}{c}{i}_{Ls1}\left(t\right)=i_{Ls1}\left(t_0\right)+(t-t_0)(n{V}_b/2+V_o)/L_{s1}\\ i_{Ls2}\left(t\right)=i_{Ls2}\left(t_0\right)+(t-t_0)(-n{V}_b/2+V_o)/L_{s2}\end{array}\right.$$

(1)

where V_b is the DC bus voltage and V_o is the ESS voltage. n is the transformer turns ratio and L_s1 and L_s2 are the equivalent series inductance. At the secondary side of the converter, the current flowing through the shared switches S₄₁ and S₄₂ is the sum of the two inductor currents (Figure 5), as given by i_tro in (2).

$$i_{tro}\left(t\right)=i_{Ls1}\left(t\right)+i_{Ls2}\left(t\right)$$

(2)

Switching state 2 [t₁, t₂]: At t₁, S₂₂ is turned off and the inductor current i_Ls2 is negative. Within the switching dead time, the parasitic capacitor of S₂₂ is charged to bus voltage V_b and the parasitic capacitor of S₂₁ discharges to 0. As a result, S₁₁ can be turned on with zero voltage switching (ZVS). During this period, the inductor currents i_Ls1 and i_Ls2 are given as shown in (3).

$$\left\{\begin{array}{c}{i}_{Ls1}\left(t\right)=i_{Ls1}\left(t_1\right)+(t-t_1)(n{V}_b/2+V_o)/L_{s1}\\ i_{Ls2}\left(t\right)=i_{Ls2}\left(t_1\right)+(t-t_1)(n{V}_b/2+V_o)/L_{s2}\end{array}\right.$$

(3)

Switching state 3 [t₂, T_s/2]: At t₂, S₃₂, S₄₁, and S₅₂ are turned off. According to the direction of i_Ls1, i_Ls2, and i_tro, within the switching dead time, the parasitic capacitors of S₃₂, S₄₁, and S₅₂ are charged. On the other hand, the parasitic capacitors of the complementary switches S₃₁, S₄₂, and S₅₁ are discharged. As a result, ZVS-on of S₃₁, S₄₁, and S₅₁ can be guaranteed. In this period, the inductor currents i_Ls1 and i_Ls2 are given as shown in (4).

$$\left\{\begin{array}{c}{i}_{Ls1}\left(t\right)=i_{Ls1}\left(t_2\right)+(t-t_2)(n{V}_b/2-V_o)/L_{s1}\\ i_{Ls2}\left(t\right)=i_{Ls2}\left(t_2\right)+(t-t_2)(n{V}_b/2-V_o)/L_{s2}\end{array}\right.$$

(4)

2.1.2. Mutual Power Support Between DC Buses

When the phase-shifted duty ratios d₁ and d₂ have different signs, the system can achieve mutual power support between the two DC buses. Without loss of generality, we consider here the case d₁ > d₂ > 0. The corresponding modulation waveforms are shown in Figure 6b. In this scenario, the power is transferred from Bus #1 to Bus #2 as well as the energy storage.

Similar to the switching state analysis in the above Figure 6a, the switching states in Figure 6b can be divided into three stages, due to the circuit symmetry. The inductor currents i_Ls1 and i_Ls2 during each switching state are given as follows.

Switching state 1 [t₀, t₁]:

$$\left\{\begin{array}{c}{i}_{Ls1}\left(t\right)=i_{Ls1}\left(t_0\right)+(t-t_0)(n{V}_b/2+V_o)/L_{s1}\\ i_{Ls2}\left(t\right)=i_{Ls2}\left(t_0\right)+(t-t_0)(-n{V}_b/2+V_o)/L_{s2}\end{array}\right.$$

(5)

Switching state 2 [t₁, t₂]:

$$\left\{\begin{array}{c}{i}_{Ls1}\left(t\right)=i_{Ls1}\left(t_1\right)+(t-t_1)(n{V}_b/2-V_o)/L_{s1}\\ i_{Ls2}\left(t\right)=i_{Ls2}\left(t_1\right)+(t-t_1)(-n{V}_b/2-V_o)/L_{s2}\end{array}\right.$$

(6)

Switching state 3 [t₂, T_s/2]:

$$\left\{\begin{array}{c}{i}_{Ls1}\left(t\right)=i_{Ls1}\left(t_2\right)+(t-t_2)(n{V}_b/2-V_o)/L_{s1}\\ i_{Ls2}\left(t\right)=i_{Ls2}\left(t_2\right)+(t-t_2)(n{V}_b/2-V_o)/L_{s2}\end{array}\right.$$

(7)

2.2. Fault-Tolerant Operation

When a short-circuit fault occurs at one of the DC buses, the FT-MAB serves to isolate the fault and supplies power to the healthy bus. Without loss of generality, we assume here that the fault occurs at Bus #1.

After detected the fault, S₁₁, S₁₂, S₃₁, and S₃₂ are blocked. The remaining switches (S₂₁~S₂₂ and S₄₁~S₅₂) keep in operation. The equivalent circuit in the fault-tolerant mode is shown in Figure 7. It operates similarly to the single-input single-output dual active bridge converter.

The modulation waveforms in the fault-tolerant mode are depicted in Figure 8. Similarly to DAB, it consists of two switching states. The inductor current i_Ls2 in each state is given in the following.

Switching state 1 [t₀, t₁]:

$$\left\{\begin{array}{c}{i}_{Ls1}\left(t\right)=0\\ i_{Ls2}\left(t\right)=i_{Ls2}\left(t_0\right)+(t-t_0)(n{V}_b/2+V_o)/L_{s2}\end{array}\right.$$

(8)

Switching state 2 [t₁, T_s/2]:

$$\left\{\begin{array}{c}{i}_{Ls1}\left(t\right)=0\\ i_{Ls2}\left(t\right)=i_{Ls2}\left(t_1\right)+(t-t_1)(n{V}_b/2-V_o)/L_{s2}\end{array}\right.$$

(9)

3. Performance Analysis

With consideration of both normal operation and fault-tolerant conditions, the performance of the FT-MAB converter is analyzed in this section, including the power transfer capability and soft-switching performance.

3.1. Power Transfer Capability

Based on Figure 6a, when d₁ > 0 and d₂ > 0, the power transfer from Bus #1 (Bus #2) to the energy storage P₁ (P₂) is given as shown in (10).

$$\left\{\begin{array}{c}{P}_1={\displaystyle \frac{2}{T_s}}({\int }_{t_0}^{t_1}(n{V}_b/2-V_o)i_{Ls1}dt+{\int }_{t_1}^{T_s/2}(n{V}_b/2+V_o)i_{Ls1}dt)\\ P_2={\displaystyle \frac{2}{T_s}}({\int }_{t_0}^{t_2}(n{V}_b/2-V_o)i_{Ls2}dt+{\int }_{t_2}^{T_s/2}(n{V}_b/2+V_o)i_{Ls2}dt)\end{array}\right.$$

(10)

Due to the symmetry of the circuit, here we assume L_s1 = L_s2 = L_s. By combing (1)–(4) with (10), the converter output powers P₁ and P₂ are given as shown in (11).

$$\left\{\begin{array}{c}{P}_1={\displaystyle \frac{M{n}^2{V_b}^2{d}_1(1-2d_1)}{2f{L}_s}}{d}_1>0\\ P_2={\displaystyle \frac{M{n}^2{V_b}^2{d}_2(1-2d_2)}{2f{L}_s}}{d}_2>0\end{array}\right.$$

(11)

where f is the switching frequency, f = 1/T_s. M is the voltage conversion ratio defined by (12).

$$M={\displaystyle \frac{V_o}{n{V}_b}}$$

(12)

Based on (11) and (12), the total output power P_o is given by

$$P_o=P_1+P_2={\displaystyle \frac{M{n}^2{V_b}^2\left[d_1\right(1-2d_1)+d_2(1-2d_2\left)\right]}{2f{L}_s}}$$

(13)

Based on (11)–(13), the maximum power transfer between the DC buses and the energy storage can be obtained at d₁ = d₂ = 0.25. The corresponding maximum bus powers P_1max and P_2max and total power P_max are given by (14).

$$P_{1max}=P_{2max}={\displaystyle \frac{M{n}^2{V_b}^2}{16f{L}_s}},P_{omax}={\displaystyle \frac{M{n}^2{V_b}^2}{8f{L}_s}}$$

(14)

Similarly, based on (5)–(7), when d₁ > 0 and d₂ < 0, P₁, P₂, and P_o can be derived as shown in (15). Since d₂ < 0, P₂ < 0 always holds in (11), which indicates Bus #2 receives power. The maximum power (P₂) transfer from the energy storage to Bus #2 can be obtained when d₂ = −0.25. Moreover, the maximum power from Bus #1 is obtained at d₁ = 0.25. The maximum values of P₁ and P₂ are the same as given in (14).

Furthermore, based on (11) and (15), the input power of the two DC buses can be independently regulated by the two control variables d₁ and d₂. It can be observed that P₁ is irrelevant to d₂ and P₂ is irrelevant to d₁, which would facilitate decoupled power regulation.

$$\left\{\begin{array}{c}{P}_1={\displaystyle \frac{M{n}^2{V_b}^2{d}_1(1-2d_1)}{2f{L}_s}}{d}_1>0\\ P_2={\displaystyle \frac{M{n}^2{V_b}^2{d}_2(1+2d_2)}{2f{L}_s}}{d}_2<0\\ P_o={\displaystyle \frac{M{n}^2{V_b}^2\left[d_1\right(1-2d_1)+d_2(1+2d_2\left)\right]}{2f{L}_s}}\end{array}\right.$$

(15)

In the fault-tolerant mode, the converter output power can be developed based on (8) and (9), as shown in (16). Similar to DAB, during the fault-tolerant mode, the output power can be independently regulated with the phase shift ratio d₂.

$$P_o=P_2={\displaystyle \frac{M{n}^2{V_b}^2{d}_2(1-2d_2)}{2f{L}_s}}$$

(16)

Comparing (16) with (15), the maximum output power P_o in the normal operation mode is exactly twice the maximum output power in the fault-tolerant mode. On the other hand, the power rating between the healthy bus and the energy storage remains unchanged after fault-tolerant reconfiguration.

3.2. Soft Switching Performance

Based on the switching state analysis in Section 2.1, the ZVS conditions for each switch are listed in

Table 2. ZVS Constraints in normal mode.

Switches	S11, S12	S21, S22	S31, S32	S41, S42	S51, S52
ZVS Criterion d1 > 0 and d2 > 0 (in Figure 6a)	iLs1(t0) < 0	iLs2(t1) < 0	iLs1(t2) > 0	itro(t2) > 0	iLs2(t2) > 0
ZVS Criterion d1 > 0 and d2 < 0 (in Figure 6b)	iLs1(t0) < 0	iLs2(t2) < 0	iLs1(t1) > 0	itro(t1) > 0	iLs2(t1) > 0

.

Based on (1), (3), and (4), when d₁ > 0 and d₂ > 0, the inductor currents i_Ls1, i_Ls2, and i_tro at the instants when the corresponding switches are turned on are derived as shown in (17). By combing (17) and Table 1, the ZVS operation region can be developed.

Similarly, when d₁ > 0 and d₂ < 0, the expressions of i_Ls1, i_Ls2, and i_tro at the instants when the correlated switches are turned on are derived as shown in (18).

$$\begin{array}{c}{d}_1>0\\ d_2>0\end{array}\hspace{0.33em}\left\{\begin{array}{c}{i}_{Ls1}(t_0)=-{\displaystyle \frac{n{V}_b(8d_{1}M-2M+1)}{8f{L}_s}}\\ i_{Ls1}(t_2)={\displaystyle \frac{n{V}_b(4d_1-1+2M)}{8f{L}_s}}\\ i_{Ls2}(t_1)=-{\displaystyle \frac{n{V}_b(8d_{1}M-2M+1)}{8f{L}_s}}\\ i_{Ls2}(t_2)={\displaystyle \frac{n{V}_b(4d_2-1+2M)}{8f{L}_s}}\\ i_{tro}(t_2)=i_{Ls1}(t_2)+i_{Ls2}(t_2)\\ \hspace{1em}\hspace{1em} ={\displaystyle \frac{n{V}_b(2d_1+2d_2-1+2M)}{4f{L}_s}}\end{array}\right.$$

(17)

$$\begin{array}{c}{d}_1>0\\ d_2<0\end{array}\left\{\begin{array}{c}{i}_{Ls1}(t_0)=-{\displaystyle \frac{n{V}_b(8d_{1}M-2M+1)}{8f{L}_s}}\\ i_{Ls1}(t_1)={\displaystyle \frac{n{V}_b(4d_1-1+2M)}{8f{L}_s}}\\ i_{Ls2}(t_1)={\displaystyle \frac{n{V}_b(-4d_2-1+2M)}{8f{L}_s}}\\ i_{Ls2}(t_2)=-{\displaystyle \frac{n{V}_b(-8d_{2}M-2M+1)}{8f{L}_s}}\\ i_{tro}(t_1)=i_{Ls1}(t_1)+i_{Ls2}(t_1)\\ \hspace{1em}\hspace{1em} ={\displaystyle \frac{n{V}_b(2d_1-2d_2-1+2M)}{4f{L}_s}}\end{array}\right.$$

(18)

According to (17) and (18), the ZVS constraints of S₁₁, S₁₂, S₃₁, and S₃₂ are irrelevant to d₂. Similarly, the ZVS constraints of S₂₁, S₂₂, S₅₁, and S₅₂ are irrelevant to d₁. In addition, when the ZVS conditions of S₃₁, S₃₂, S₅₁, and S₅₂ are met, the ZVS operation of S₄₁ and S₄₂ is automatically achieved. The physical explanation is that the current flow through S₄₁ and S₄₂ is the sum of i_Ls1 and i_Ls2.

By combing (17) and (18) with Table 2, the soft-switching condition of the duty ratio with respect to the voltage conversion ratio M is given as shown in (19).

$$\left|d\right|>\mathit{max}\{{\displaystyle \frac{1-2M}{4}},{\displaystyle \frac{2M-1}{8M}}\}$$

(19)

By combing (19) with (13) and (15), the ZVS operation region of output power with the voltage conversion ratio M is shown in Figure 9.

In the fault-tolerant mode, the soft-switching conditions are given as shown in

Table 3. ZVS constraints in fault-tolerant mode.

Switches	S21, S22	S41, S42	S51, S52
ZVS Criterion (in Figure 8)	iLs2(t1) < 0	itro(t2) > 0	iLs2(t2) > 0

. By comparing Table 3 with Table 2, it is also found that the ZVS-on constraints given by (19) also applies to the fault-tolerant mode. As a result, the fault-tolerant operation region overlaps with the normal operation.

3.3. Control Design

The control aims of the FT-MAB include input power P₁ and P₂. The control diagram is shown in Figure 10.

With consideration of the bidirectional power transfer need, the forward/backward working mode as well as the total output power can be regulated with the phase-shift duty ratios d₁ and d₂. When the phase-shifted duty ratios d₁ and d₂ have different signs, the system can achieve mutual power support between the two DC buses. When the FT-MAB converter operates in the normal mode, the power is transferred from Bus #1 and Bus #2 to the energy storage. The specific values of d₁ and d₂ are determined by the controller PI_P1 and PI_P2. When the FT-MAB converter operates in the fault-tolerant mode, Bus #1 is short-circuited and P₁ = 0. The power is transferred from Bus #2 to the energy storage. The specific value of d₂ is determined by the controller PI_P2.

4. Circuit Comparison

In this section, the performance of the proposed FT-MAB is evaluated in comparison with a traditional dual-module system (Figure 3a). The classical DAB converter is adopted in the two-module system as an input-independent output-parallel DAB (IIOP-DAB). The detailed circuit is shown in Figure 11a. Moreover, if the full bridge is adopted at the bus side, the full-bridge dual-transformer asymmetrical triple-port active bridge (DT-ATAB) is illustrated in Figure 11b.

The conventional converters in Figure 10 are compared with the proposed FT-MAB under the same scenario, with the system parameters listed in

Table 4. Mode scenario for the topology comparison.

Parameter	Value
Bus voltage Vbus1, Vbus2	750 V, 750 V
Output power Po	15 kW
Output voltage Vo	400 V
Switching frequency fs	40 kHz

.

4.1. Cost Comparison

For a fair cost comparison, all semiconductor switches are selected from Infineon and the drivers are selected from Microchip products. Their prices are sourced from Digi-Key Electronics [28]. Considering the voltage/current stress margin, the semiconductor switches selected for each converter are listed in

Table 5. Component selection for case study.

Items		Proposed	IIOP-DAB	DT-ATAB
Peak current through switches/A	Input-side	27	17	17
	Output-side	60	30	60
Switches	Input-side	FF33MR12W1M1HP_B11
	Output- side	FF33MR12W1M1HP_B11
	Output- shared switches (S41,S42)	FF6MR 12W2M1H_B11	/	FF6MR12W2M1H_B11
Magnetic components	Transformer core	0P49925UC
	turns ratio n	1	0.5	0.5
	Inductor core	0077774A7
Capacitors	Bus-side	B32373B4107J080	C44USGT6120M81K
	Load-side	B32371A3806J030	C4DEIPQ6100A8TK	B25631B0127K800
Driver	2ASC-12A2HP

.

The switches used in the input stage of all three converters are FF33MR12W1M1HP_B11. For the FT-MAB converter and DT-ATAB converter, the output-side bridge current stress is increased to twice the value and the switch module used is FF6MR12W2M1H_B11.

The transformer turns ratio is designed as 15:15 for the proposed FT-MAB. As for the DT-ATAB converter and ISOP-DAB converter, the transformer turns ratio is selected as 30:15. All three converters use 0.1 mm × 1500 litz wire, with the core material selected as 0P49928EC. The series inductors utilize 0077774A7 magnetic cores. The prices for the magnetic components can be found in [28].

Under the condition of the same power rating, the cost of transformers and inductors is related to their quantity, current stress, and winding configuration. The bus-side and load-side capacitors for these three converter systems are determined with the same voltage ripple ratio of 1%.

A comparison of the total converter cost is shown in Figure 12. Due to the reduced number of switches, the cost of switches and drivers of the proposed FT-MAB converter is significantly lower. Considering the increased number of DC-link capacitors, their impact on the total capacitance and converter cost is relatively small. As a result, the proposed FT-MAB exhibits the lowest total converter cost.

4.2. Volume Comparison

Based on the component selection listed in Table 5 and the size parameters, the corresponding volume of each converter can be calculated. As shown in Figure 13, the main influencing factors for the converter volume are the capacitors and the transformer/inductors. The size of inductance is also influenced by the voltage rating. Since the full-bridge circuit has a higher voltage stress, the inductance volume is comparatively larger than that of the half-bridge in the FT-MAB. On the other hand, the proposed FT-MAB has a higher number of capacitors and the total capacitor volume is much higher. As a result, the total sizes of all three converters are similar.

4.3. Loss Comparison

The converter power loss mainly consists of switch losses, transformer losses, capacitor losses, and inductor losses.

Transformer/inductor power losses includes core losses and copper losses. The transformer core material is selected as 0P49925UC, with the assumption that the high-frequency transformer operates at a temperature of 80 °C and the maximum operating magnetic flux density is designed at 0.2 T. To mitigate the skin effect, litz wire is adopted. The winding current density of the high-frequency transformer is set to 3 A/mm². Based on the Steinmetz equation and transformer calculation formulas, the power losses of the high-frequency transformer are calculated. The inductor core material is selected as 0077774A7 and the total loss is determined by a calculation tool provided by the manufacturer [29].

The power loss of switches includes conduction losses and switching losses. Based on the datasheet provided by Infineon, the power losses of the switches are simulated using PLECS (4.8.4) thermal software. In comparison, the converter loss breakdown is shown in Figure 14a. Compared with IIOP-DAB and DT-ATAB, the proposed FT-MAB has the smallest power loss. Based on the PLECS simulation software, the efficiency curve of the proposed converter and traditional systems (IIOP-DAB) is plotted as shown in Figure 14b. It can be seen that the FT-MAB converter has a significant efficiency advantage. Specifically, the highest efficiency is achieved at an output power of 10 kW, which is consistent with the experimental results.

5. Experimental Verifications

The proposed FT-MAB converter and operation strategy are tested with an experimental prototype. Parameters of the prototype are listed in

Table 6. Parameters of the experimental prototype.

Parameter	Value
Input voltage Vbus1, Vbus2	48 V
Output voltage Vo	110 V
Rated output power Por	200 W
Switching frequency fs	40 kHz
Series inductance Ls	65 μH
Transformer turns ratio	1:1

. The switches and diodes are implemented with IRFP260N and MBR40200PT, respectively, and the control is implemented with a TI C6657 DSP controller. The photograph of the experimental platform is shown in Figure 15. Two regenerative DC power supplies are connected to the DC side of the half bridges, serving as the power supply for the two buses, respectively. Another regenerative DC power supply is connected to the DC side of the full bridge to simulate a bidirectional load. The experimental platform verifies the feasibility and characteristics of the FT-MAB operating in both normal and fault-tolerant modes.

5.1. Steady-State Operation and ZVS Characteristics

The steady-state waveforms of the FT-MAB converter in the normal dual-bus mode are shown in Figure 16.

As can be observed in Figure 16, by regulating the phase-shift duty ratios, the output voltage v_p1, v_p2 leads v_s1 in the forward mode and v_s1 leads v_p1, v_p2 in the backward power mode.

Moreover, by independently regulating the phase-shift duty ratios d₁ and d₂, the output power of the two DC buses can be controlled. The corresponding operation waveforms are illustrated in Figure 17a,b. The modulation waveforms are in accordance with the analysis in Section 2.1.

In comparison, in the fault-tolerant mode, the modulation waveforms are illustrated in Figure 16.

In Figure 18, when short-circuit fault occurs in Bus1, S₁₁, S₁₂, S₃₁, and S₃₂ are blocked and v_p1 is constantly 0. The modulation waveforms are in accordance with the modulation in Figure 8. By controlling the phase-shift ratio d₂, bidirectional power can still be regulated after a short-circuit fault. The fault-tolerant operation is in accordance with the analysis in Section 2.2.

In verification of the soft-switching performance, the zoomed-out waveform at the switching instant in normal mode and fault-tolerant mode are shown in Figure 19a and Figure 19b, respectively. Before the driving signal V_gsS11, V_gsS31 reaches, the voltage across the switch V_dsS11, V_dsS31 first drops to 0. As a result, the ZVS for power switches is guaranteed, in accordance with the analysis in Section 3.2.

5.2. Fault-Tolerant Operation

The fault-tolerant operation of the FT-MAB converter is tested in Figure 20. Before time t₁, the converter operates stably in the normal dual-bus mode. At time t₁, a short-circuit fault occurs on Bus 1 of the converter, which leads to a significant decrease in the bus voltage v_bus1.

When the bus voltage v_bus1 drops below 70% of the normal value at time t₂, the undervoltage protection is triggered. Consequently, the switching pulses of S₁₁, S₁₂, S₃₁, and S₃₂ are blocked. As a result, the primary-side current i_Ls1 of the corresponding transformer drops to 0. Meanwhile, the driving signals of S₂₁, S₂₂, and S₄₁~S₅₂ remain in operation. The fault tolerant operation is in accordance with the analysis in Section 2.2.

This paper experimentally validates the fault-tolerant capability and efficiency characteristics of the FT-MAB converter based on a low-power prototype. However, in high-power scenarios such as shipboard DC systems, several issues need to be further considered. For example, the conduction and switching losses of switching devices significantly increase at high power levels. The solution can adopt a modular multilevel converter (MMC) structure to reduce the voltage stress on the switches.

5.3. Efficiency

Precise measurements of the efficiency of the FT-MAB converter in both normal and fault-tolerant modes were conducted by collecting data. LEM LV 25-P voltage sensors and LEM LA 55-P current sensors were used to capture voltage and current signals at each port, respectively. The sampling frequency was set to ensure complete recording of high-frequency switching transients. Multi-channel signal synchronized acquisition was achieved using a National Instruments (NI) USB-9234 data acquisition card and instantaneous power was calculated in real-time using the LabVIEW platform. The efficiency curves under the normal dual-bus mode and the fault-tolerant single-bus mode are compared in Figure 21. Under normal operation with balanced bus power, the maximum efficiency reaches 97.54%. The efficiency decreases by 2% under full and light load conditions. Under fault-tolerant single-bus mode, the maximum efficiency reaches 97.33% and decreases by 2% depending on the load condition. The loss of the FT-MAB is broken down into the normal mode and fault-tolerant mode, corresponding to the experimental efficiency curves, as shown in Figure 22.

6. Conclusions

In the conventional zonal DC shipboard configuration, two sets of energy storage and power converters are required. As an alternative, a fault-tolerant multiport active bridge converter is proposed for reduced cost and power loss. The key findings of this paper are as follows:

■

The proposed FT-MAB converter enables bidirectional power regulation between the energy storage and the two DC buses, as well as mutual power support between the DC buses;

■

In the event of a short-circuit fault, the FT-MAB converter can isolate the fault port and maintain uninterrupted power supply to the healthy bus;

■

Compared to the traditional dual-converter zonal configuration, the proposed topology can maintain fault-tolerant operation and reduce the converter power loss as well as cost.

The proposed FT-MAB topology and modulation method are tested during normal operation and short-circuit fault condition, in verification of the resilient power supply aim for zonal DC systems. The FT-MAB converter achieves flexible energy interaction among multiple buses in a shipboard zone DC system through its multi-port shared energy storage design. By continuously supplying power to healthy buses through the energy storage unit, it significantly enhances the system’s power supply resilience, providing an innovative solution for high-reliability shipboard power systems.