The solid state transformer (SST) is a combination of power electronic converter and high or medium-frequency transformer that connects two voltage zones. In addition to the normal functionalities of the conventional transformers, such as galvanic isolation and voltage matching, the SST benefits the power distribution system by enabling active power and voltage control, real time monitoring and control of its ports [1]- [4]. Due to the medium or high-frequency operation of the transformer, the size is much smaller than a conventional low frequency transformer (LFT) for the same voltage and power ratings. This size and weight reduction is a key motivation for designing high-frequency SSTs and isolated power electronic converters.
Recent studies have shown the importance of local installation of energy storage units, like battery energy storage systems, as one of the main requirements for the future architecture of the distribution system for supporting local demands. This provides many advantages in terms of reliability, resiliency, and power quality. However, independent power converters are required to operate with battery energy storage systems. Similarly, when photovoltaic (PV), battery energy storage and/or fuel cells systems are employed in a micro-grid or local power system, several two-port DC-DC converters are adopted to control and manage the system [5], [6]. This requirement decreases both power density and reliability [7]. Similar issues are valid for the electrical vehicle applications. Battery supported fuel cell vehicles or super-capacitor applications require multiple DC-DC converters. Although in some studies converters are designed as a multi-port structure to transfer energy to each other via a common DC bus, the requirement of electrical isolation between ports was not satisfied. Also, the conversion ratio between the ports in some cases are very high, and thus, these structures cannot be used. These requirements lead to the use of several isolated two-port DC-DC converters [6] and as a result, multiple transformers employed in the system will lead to a decrease in power density and reliability, and to an increase in the total cost [7].
Multi-port converters have been discussed and used for different applications including: thetraction (electrical trains) applications [8], [9], the electrical vehicle power distribution systems [10], and the hybrid distributed generation (DG) applications [7]. Multi-port solid state transformer or multiport isolated converter topology is an extension version of the Dual Active Bridge (DAB) converter that uses common transformer core for power sharing instead of common DC bus. Multi-port SSTs (MPSSTs) are investigated to improve the overall power density, cost and reliability [7]- [11]. The power of the MPSST flowing from one port to another is controlled by the phase-shift angle between these two ports. However, the other ports are affected as well in this phase shifting action; their corresponding phase-shift angles have to be changed in order to keep their power/voltage levels at the desired values. It is seen that increasing the number of ports also increases the control complexity of the converter. Therefore, galvanic isolated MPSSTs has been widely proposed for three-port systems [12]- [14]. A suggested topology of three-phase three-port systems was discussed in [15]. A limited number of control schemes have been suggested for MPSSTs. In [12], an open loop power sharing control scheme is introduced for three-port MPSST. This approach is based on the natural power sharing characteristics of a galvanically isolated three-port resonant DC-DC converter. However, only a sample operation mode (two ports are transferring power to the third one) is considered. In addition, non-ideal conditions like parameter mismatch and/or voltage fluctuations are not considered. A control scheme based on switch turn-on time is proposed for a flyback transformer based three-port MPSST [13]. However, this controller scheme is only designed for flyback transformer-based systems, and it is not discussed in much details. There are limited number of studies on controlling four-port SST [7]. Since power flow and voltage regulation are overly complicated by the increase in the number of ports, the controller design is a challenge for these converters. Therefore, a decoupled control scheme is an important requirement to control the power of each port.
In this paper, a four-port SST topology is developed to be used as a compact integration node for a distribution system with four voltage zones; DG, battery energy storage system, grid, and a load as shown in Fig. 1. A four-port MPSST and a decoupled control scheme is presented to control the output voltage and port power levels. There are three active converters interconnecting the grid, the energy storage, and the DG source. These devices could be in a different combination depending on the application in hand. The active switch-based converters are full-bridge single-phase converters enabling bidirectional power flow for grid and energy storage ports. The same converter is used for the unidirectional power conversion on the DG port. Finally, the load is connected to the fourth port. Due to the nongenerating nature of the load, a unidirectional single-phase diode bridge rectifier is used for this port. It is assumed that an energy management system is employed which determines the power references for all ports. Then, each active converter is controlled to track these reference signals.
The operation of the proposed system is validated with MATLAB/Simulink simulations and experimental studies performed on a 10 kW/port MPSST prototype. Simulation and experimental results have confirmed that the proposed control method is suitable for multi-port solid state transformer systems in terms of quality measures such as steady state and dynamic response.
Transformer design is one of the most important processes in SST design. In order to build an optimal transformer, some design parameters such as core material, cross sectional area, volume, shape, winding configuration, number of turns, wire type, maximum flux density, leakage and magnetizing inductances, parasitic capacitance between the four windings, core and winding losses and the efficiency of the transformer should be analyzed. Considering parameters such as size, energy density, etc., the operating frequency is selected to be 100 kHz.
As it is known, four core materials which are mainly silicon steel, amorphous, nanocrystalline and ferrite are used in medium and high frequency transformer design. In Fig. 2, these four different core materials are compared in terms of flux density, price, operating frequency and loss. Although the silicon steel provides higher saturation flux density values, its loss characteristics limit its use in high frequency applications. Today, using silicon steel may be suitable up to a few kHz operating frequency. As for the case of Amorphous and Nanocrystalline core materials; their saturation flux density is lower than silicon steel but higher than the ferrite. Therefore, they offer smaller core size and can provide more compact design. However, higher operating frequency conditions, like 100kHz as the case in this study, the increase in core losses is more dominant over the decrease in the core size being compared to ferrite [1], [3] [16]. As a result, 3C94 ferrite core was selected to be the most suitable material for our application.
Another important parameter to be looked after is the leakage inductance. It directly affects the efficiency and power transfer capability of the power converters. While higher leakage inductance values limit the power transfer capability and causes a decrease in efficiency, it also limits the peak value of the transformer current. Therefore, an optimal value should be determined to achieve the design objectives. In multi-port solid state transformer design, different winding placement also effects many parameters such as wire resistance, leakage and magnetization inductances, leakage capacitance of the windings, etc. After conducting extensive and thorough design studies, it is seen that in terms of leakage inductance and wire length, the winding placement given in Fig. 3(a) provides better results. Then, this design is modelled in a Finite Element Analysis (FEA) analysis tool (ANSYS-Maxwell) as shown in Fig. 3(b).
Transient and Electrostatic simulations have been done through ANSYS-Maxwell, and then a co-simulation (ANSYS-Maxwell and ANYSY-Simplorer simulating the model together) of the model is performed. The flux distribution obtained from transient simulation studies is given in Fig. 4. As it is seen from the figure, there is no saturation, so maximum flux density target is achieved. The voltage waveforms obtained from the co-simulation studies are given in Fig. 5. As a result of these studies, transformer parameters are obtained as given in Table 1. Then the designed transformer is implemented and the prototype of the transformer is shown in Fig. 6.
In this study, a four port SST control scheme is designed. The control of four-port SST is very similar to the control of a dual-active bridge (DAB) converter. Therefore, the equation for power delivered from port-x to port-y can be written as follows:
where Pxy is the power transferred from port-x to port-y; φx and φy are phase shift angles of the converters port-x and port-y, respectively; and are DC voltages referred to the AC side of port-x and port-y, respectively; fs is the switching frequency, and Lxy is the equivalent inductance between portx and port-y. In this study, a four-port SST is proposed as depicted in Fig. 7. Here, L1, L2, L3 and L4 are the corresponding winding leakage inductances combined with external inductor. For simplicity, mutual-leakage inductances presented in [11] are not considered here. The power flow between any two ports is calculated by (1).
The equivalent circuit given in Fig. 8 can be obtained from Fig. 7. Here, Lm is magnetizing inductance; L1 is port-1 leakage inductance; and L2 ' , L3 ' and L4 ' are leakage inductances of ports 2, 3 and 4, respectively, referred to port-1; v1, v2 ' , v3 ' , v4 ' and i1, i2 ' , i3 ' , i4 ' are port voltages and currents referred to port-1. It is clear from Fig. 8 and (1) that the equivalent inductances between the ports should be obtained to calculate the power flow. Thevenin equivalent circuit can be used for this purpose. The Thevenin equivalent circuit between port-x and port-y referred to port-1 is given in Fig. 9 (a) [7]. By neglecting the magnetization inductance, the equivalent inductances between the ports can be written as ( 2)-( 7): Power equations for the other ports can be written similarly. It can be seen that the power of each port depends on all of the phase angles of other ports. This makes the control of the system very complicated. To decouple the equations, a small gain (x+Δx) linearization method is applied to the power equations. If it is applied to (14), then power equation has two components Px and ΔPx as follows [17]:
Here, the power component Px represents the operation point of the converter which depends on Vx, Vy , Lxy, Vx, φx, and φy, while ΔPx component represents the power variation which depends on the variation of Δφx, and Δφy. This makes Δφx and Δφy perturbation control variables, and these perturbation variables can be used in control loop to control the power flow. Then, power variation components can be written as below:
If ΔP3 and ΔP4 are obtained by using the same method, they can be written in a linear system equation form as below:
The control variable can be calculated depending on the required power variations in accordance to (17).
As seen from ( 18), the power variation of any port only depends on its control variable, and thus decoupled power flow control is obtained.
The proposed four-port SST and the control scheme is designed and validated through simulation and experimental results. The simulation results are carried out with MATLAB/Simulink. The control scheme is implemented on Altera FPGA board. The switching frequency is chosen to be 100 kHz. The proposed system is designed for distributed generation systems. It is assumed that the four-port SST ports are connected to: a renewable power source (e.g. PV modules), a battery energy storage system, a grid, and a load. Many different scenarios can be designed and implemented to provide better energy reliability or decrease the energy cost. However, these scenarios are not discussed here, and it is assumed that one energy management system provides current references to the system. Then, the performance of the proposed decoupled control scheme and the four-port SST is tested for these conditions. The renewable energy port is controlled to obtain maximum available power from the renewable source. Therefore, it is assumed that a MPPT controller provides a current reference for the renewable port. Similarly, it is assumed that a current reference is provided for the battery port. Then, the current reference of the grid port is generated by a PI controller, and the output voltage is kept at its desired value. In Fig. 10, the output voltage and its reference signal waveforms, output current waveform, Port-2 (PV port) current and its reference signal waveforms, Port-3 (Battery port) current and its reference signal waveforms, and Port-4 (Grid port) current waveform are given. As explained before, the reference currents for Port-2 and Port-3 are randomly determined and the reference current waveform for Port-4 is generated with the PI controller. After the variation of reference signals are determined for Port-2 and Port-3 and a load change is applied, the performance of the proposed control scheme is tested. As seen from figure, the output voltage, Port-2 and Port-3 currents are tracking their reference signals and the performance of the proposed control of the proposed scheme is validated. It is seen that, proposed decoupled control scheme successfully tracks the reference signal for each port, and keeps the output voltage constant. It is also seen that it provides a good transient response.
The port currents waveforms of the four-port SST obtained from simulation and experimental studies are given in Fig. 11 and Fig. 12 for different operation conditions. In both figures, Port-2 is the most leading port; i.e., it provides more power to the other ports and load port. While Port-3 is the most lagging port in Fig. 11, the most lagging port is Port-4 in Fig. 12. The phase shift values are defined by (18) to track the power references of each port for any operation condition.
In this study, design of a four-port SST transformer and its decoupled control scheme is presented. The transformer includes two active bidirectional ports, one active unidirectional port, and one diode based unidirectional converter. The design details of the four-port transformer including core material selection, winding placement are investigated and validated with FEA simulations. The system is developed for fast, modular, and isolated utilization of DG and/or battery energy storage in the power system. The fourport SST is modelled and a control scheme is designed in order to control the power flow among the converter ports as well as ensure voltage regulation at the uncontrolled output port. In order to decouple the power sharing equations of the multiport system, a small signal linearization approach is applied to the power sharing equations. Simulation and experimental results have shown that the proposed system can control the power flow among the ports and the load port output voltage independently. It is seen that the proposed system provides a smooth control and power transfer between the ports as desired.
978-1-7281-0395-2/19/$31.00 ©2019 IEEE