A Power Control Method for Hybrid Electrical Accommodation Systems

Abstract

This paper presents a power control method for a hybrid electrical system, which enhances a conventional main AC grid by adding a DC sub-grid in parallel. In this system, each prosumer connects to the sub-grid via a couple of half-bridge inverters of a device called a power router. All of the power routers connected to the sub-grid are controlled such that they maintain the voltage of the sub-grid equally, which is an essential feature of peer-to-peer power accommodation without synchronization. This paper proposes an adaptive hysteresis current-control technique for controlling sub-grid-connected inverters that maintains the voltage of the sub-grid while accommodating a given power. The proposed control method also balances currents between positive and negative lines of the sub-grid simultaneously. The proposed control method is implemented on a high-speed FPGA-based digital controller and is evaluated on three prototypes of the power router. The experimental results show that the proposed method yields fast and stable responses and enables the power router to accommodate power with a high accuracy.

1. Introduction

Today, various major changes in the electrical power system are being stimulated by rapid increases in photovoltaic (PV), wind-turbine, and other modes of distributed generation, together with digital transformations [1,2,3,4]. Microgrids, which are composed of distributed generation systems, energy storage, and loads, are being developed in various areas [5,6] for both low and medium voltage ranges [7]. They exist in various sizes, from simple systems that serve a few customers to complex systems with high levels of generation and large loads [8]. Interconnecting main grids with sub-grids allows microgrids to operate in grid-connected or islanded modes more flexibly. In various microgrid structures, conventional consumers, who previously have only consumed energy, will change into prosumers, who not only consume but also provide energy [9,10]. Electrical trades between prosumers in microgrid systems are increasing alongside conventional passive trades between consumers and power suppliers in grid-connected systems. Peer-to-peer (P2P) power trading systems based on new platforms, such as blockchain technology, produce a freer power market, where power can be traded between prosumers directly [11,12,13]. While accommodating electrical power via the main grid may lead to instability in the system and impose a countable wheeling charge, sub-grids enable prosumers to accommodate electrical power more flexibly and can support the main grid [14]. As a result, studies on the design and stability analysis of the sub-grid system have attracted increasing attention from both academia and industry [8,15,16,17] in the last few decades.

There are two main methods for controlling sub-grid systems: droop control and master–slave control [18,19,20,21]. Droop control enables parallel multiple inverters to share consumed power without exchanging information. However, this method has a well-known inherent limitation—namely, a trade-off between the regulation of common voltage and the accuracy of current-sharing. Droop control is unable to control the accommodated power freely. On the other hand, master–slave control enables the slave to send or receive a specified power via the sub-grid. However, the latter control method requires a sufficiently large power source to be operated as a master in order to maintain the voltage of the sub-grid. These two methods cannot be adapted to microgrids based on the P2P accommodation system, in which prosumers can send or receive traded power while contributing to voltage regulation evenly.

Another issue is that, while many control techniques for AC sub-grids have been studied, control techniques for DC sub-grids are still rare [6,18,22,23,24,25]. This is understandable, since AC grids play a dominant role in traditional distribution systems, and the techniques used in traditional AC grids can be applied to sub-grids. Recently, however, interest in DC sub-grids has risen, due to the proliferation of renewable generation systems, such as fuel cells, PV generations, and batteries, which naturally are DC. Controlling parallel inverters in DC sub-grid systems does not require synchronization, which is still a knotty problem in AC sub-grids. Moreover, various high-tech electronic devices, which need DC power rather than AC power, also increase the appeal of DC sub-grids. To the best of the authors’ knowledge, research on P2P accommodation systems using DC sub-grids, which can combine the advantages of droop control and master–slave control, has not been presented in the literature.

In a short paper we published previously [26], we presented the preliminary concept of such a DC sub-grid accompanied by a simple simulation. In this study, we extend that idea to propose a hybrid electrical system, which enhances the conventional main AC grid by adding a DC sub-grid in parallel. The main contribution of this paper is to propose a method that uses adaptive hysteresis current control technique to control accommodated electrical power between the prosumer nodes while maintaining a constant DC voltage in the sub-grid.

This paper is organized as follows. The hybrid electrical power accommodation system using a DC sub-grid is presented in Section 2. The control method for the sub-grid using an adaptive hysteresis current-control technique is proposed in Section 3. Experimental results for the evaluation of the proposed system are presented in Section 4. Section 5 presents our conclusions.

2. A hybrid Electrical Accommodation System

Figure 1 depicts the structure of the hybrid electrical system considered in this study. This system included a conventional main AC grid and a DC sub-grid. Compared to an AC sub-grid, the DC sub-grid does not require synchronization with prosumer nodes connected to it. Thus, it enables us to control the nodes independently without using information shared between the nodes. The sub-grid is designed to accommodate power between the prosumer nodes connected by devices called power routers. The structure of the power router, which comprises multiple inverters [27] connected to a common DC bus, is depicted in Figure 2. The power router controls the power flow between the main grid, the sub-grid, PV generators, and batteries, or else supplies power to the load. In the power router, there is an inverter, which is controlled so as to maintain a constant voltage in the common DC bus—i.e., the sum of input and output power of the router is maintained at zero. This inverter may connect to the AC grid or the battery. For instance, let us consider a power router connecting to a PV and an AC grid. When the PV generates power, it causes the voltage of the DC bus to increase. The inverter connecting to the AC grid sends the same power to the grid to maintain a constant voltage in the DC bus. In this case, the power router plays the role of a conventional power conditioner.

In general, the main grid and the sub-grid can be operated simultaneously. When the main grid is unavailable or fails—i.e., the voltage or frequency is over or under a standard range, the power router detects the problem automatically. When a fault in the main grid is detected, the power router decides either to operate the grid-connected inverter continuously by a fault ride-through (FRT) function or to open the breaker $K$ to prevent the power router leaving from the main grid. In this case, the system can be operated as an islanded DC microgrid system using the DC sub-grid. The power router also enables the battery or PV to supply the electrical power for the load without being connected to the main or sub-grids as a stand-alone system.

A controller of the power router monitors all of the currents and voltages of the inverters. When a measured value of the current or voltage is over a limited value, the power router judges a fault to be occurring within itself and stops the inverters until it is restarted by the maintainer.

Electrical power trading between prosumers may be based on the demand and supply ability of each prosumer node, where each communicates with trading systems via a communication network. Smart contracts using new technologies, such as blockchain, may be used to implement electricity transactions automatically as a P2P power trading system. Via the DC sub-grid, the power router directs the inverters to transfer electrical power based on transaction results received from the trading system.

The DC sub-grid is composed of positive-voltage and negative-voltage lines. The outputs of two inverters within the power router connect to the lines of the sub-grid without sharing a neutral line, as depicted in Figure 3. Capacitors $C_1$ and $C_2$ comprise the common DC bus of the power router depicted in Figure 2. The neutral line of the power router is not shared, and this prevents uncontrollable loop-currents, which may be caused by the difference between the ground voltages at each power router.

Let the voltage of the sub-grid be $v_{sub}$. The sub-grid-connected inverters are controlled to preserve the line to neutral voltages of $\pm v_{sub}/2$. These inverters are controlled independently. In this study, the sub-grid is designed such that all prosumers connecting to the sub-grid contribute to maintaining the voltage of the sub-grid equally, which is an essential feature of peer-to-peer power accommodation. The electrical accommodations are carried out between pairs of prosumers, who trade the power. This execution-based electrical trading preserves the demand-supply balance of the sub-grid. The proposed system allows multiple prosumers to connect to the DC sub-grid asynchronously.

Figure 4 shows an example of the power flow in the power router, which connects to the battery, PV, AC main grid, and DC sub-grid. The PV-connected inverter is controlled so as to track the maximum power point (MPPT) of the PV. The battery-connected inverter is controlled so as to maintain the voltage of the DC bus. The AC main-grid- and DC sub-grid-connected inverters send power of 1.0 kW and 1.5 kW to the grids, respectively. At 0.8 s there is an outage in the AC main grid. Simultaneously, the power router is disconnected from the main grid by opening the breaker K and is controlled so as to maintain the voltage at the load, consuming 2 kW of power. At 1.1 s, the main grid recovers and the power router is controlled so as to reconnect to the main grid. The battery is charged and discharged depending on the total input and output power in order to maintain the voltage of the DC bus at 175 V.

3. Adaptive Hysteresis Current Control for DC Sub-Grid-Connected Inverters

3.1. Adaptive Hysteresis Current Control

Compared to the sine-triangle pulse-wide-modulation (SPWM) technique, the hysteresis current-control technique has a stable and fast response; moreover, the controller design is simpler since its system parameters are not required in the design process. However, the current band in conventional hysteresis control is usually fixed at a constant. As a result, the switching frequency varies to maintain the ripple current within the band, which may lead to undesired harmonics. In this paper, we used adaptive hysteresis current control [28], where the current band is calculated adaptively in order to keep the switching frequency constant while exploiting the advantages of the conventional hysteresis current-control technique.

Figure 5 depicts the LCL half-bridge inverter circuit used in this study. Switch devices $S_1$ and $S_2$ are used to control the output current $i_o$ of the inverter tracking a given reference current $i_{ref}$. The current band $\Delta i_b$ is calculated adaptively, depending on the value of the DC voltage and the output voltage, in order to maintain a constant switching frequency, as shown below [28]:

$$\Delta i_b\left(t\right)=\frac{{V_{dc}}^2-v_o{\left(k{T}_{sp}\right)}^2}{4L{f}_{sw}{V}_{dc}}$$

(1)

where $k{T}_{sp}\le t<\left(k+1\right)T_{sp}$ and $T_{sp}$ is a sampling period of analog/digital converters. In Equation (1), $V_{dc}$ and $v_o$ are DC bus and output voltages, respectively, $L$ is output-inductance and $f_{sw}$ is switching frequency.

In this study we considered the power router used in households with a low-voltage application. Thus, the two-level half-bridge circuit is used due to its simplicity in both circuit and controller design. However, the power router may also be used for a larger residential area with a higher voltage. In such a case the multilevel inverter circuit topology may be considered in order to reduce undesirable distortions, common-mode voltage, and switching frequency [29].

3.2. Reference Current Calculation for the DC Sub-Grid

A challenge for the proposed sub-grid system is controlling the accommodated electrical power while simultaneously maintaining the voltage of the sub-grid. Another important issue is that the currents flowing along the positive and negative lines need to be balanced almost exactly, because an imbalance of the currents may conduct an undesired DC component flowing into the AC main grid. Isolated transformers can be used to clear this imbalance issue; however, they may make the system more complex and increase its cost. In this study, we tried to control the accommodated power, maintain the sub-grid voltage, and balance the current simultaneously. Each power router was controlled independently without using information exchange from the others.

Our target was to calculate the reference current $i_{ref}$, which is able to achieve the above requirements of sub-grid control used for hysteresis current control presented in subsection A. The proposed calculation for the reference current is as follows:

$$i_{ref}=\frac{v_{ref}-v_o}{C{f}_{sw}}+sig\left(v_{ref}\right)\frac{P_{ac}}{v_{sub}}-\left(k_P\epsilon +k_I{\displaystyle \int \epsilon }dt\right)$$

(2)

where $v_{ref}=v_{sub}/2$ is the positive line-connected inverter and $v_{ref}=-v_{sub}/2$ is the negative line-connected inverter. $P_{ac}$ is the given accommodated electrical power. If $P_{ac}>0$, the inverter sends power to the sub-grid; otherwise, if $P_{ac}<0$, the inverter receives power from the sub-grid. The signal function $sig\left(v_{ref}\right)$ takes the value of 1 for positive $v_{ref}$ and $-1$ for negative $v_{ref}$.

The current difference $\epsilon$ is defined as:

$$\epsilon =i_p+i_n$$

(3)

where $i_p$ and $i_n$ are the output currents of the positive and negative line-connected inverters, respectively. Parameters $k_P$ and $k_I$ are the positive control gains designed for a PI regulator.

Figure 6 details the calculation of the reference current given by Equation (2). The calculation includes three elements. The first element of the reference current controls the output voltage of the inverter [27]; the second element specifies the power accommodation; the PI regulator in the third element balances the positive and negative currents.

Remark 1.

The voltages of the capacitors composing the common DC voltage of the power router are maintained at the constant value of $V_{dc}$ by another inverter, which may connect either to the AC main grid or battery [27].

Remark 2.

Each power router is controlled independently without using shared information from the others.

4. Experimental Results

The proposed control method was evaluated using a prototyping hybrid system, which is composed of three power routers. Figure 7 depicts the setup of the experiment. Each power router was composed of three inverters. Each inverter had a rated power of 300 W. The first inverter connected to a 100 V/50 Hz AC main grid. This inverter was used to maintain the DC voltages of the power router at 175 V. The two other inverters were connected to the sub-grid, the voltage of which, $v_{sub}$, was set at DC 250 V. Each half-bridge inverter circuit depicted in Figure 4 had a rated power of 300 W with parameters given by $C=6.8\mathsf{\mu }\mathrm{F}$, $L=2.2\mathrm{mH}$, and $L_g=1.1\mathrm{mH}$. The electrical circuit of each power router is depicted in Figure 8.

The switching frequency and sampling frequency of the analog/digital converter were 20 kHz and 4 MHz, respectively. The control algorithm was implemented on a digital controller, which was an FPGA board with a clock frequency of 160 MHz. The control gains of the PI regulator were set at $k_P=0.1$ and $k_I=10$.

Figure 9 depicts: the voltages of the AC main grid and DC sub-grid; the output currents of the grid-connected inverter $i_{grid}$; the sub-grid-connected ($i_p$: positive voltage line; $i_n$: negative voltage line) inverters of the power routers 1–3 when the power was not accommodated ($P_{ac\_1}=P_{ac\_2}=P_{ac\_3}=0W$). It can be seen that the voltages of the DC sub-grid were maintained at 250 V, while the output currents $i_p$ and $i_n$ of the sub-grid-connected inverters were maintained almost exactly at 0 A.

Figure 10 depicts the same responses of the inverters when the accommodated power was set at $P_{ac\_1}=200W$ and $P_{ac\_2}=P_{ac\_3}=-100W$—i.e., when the first power router sent 100 W of power to each of the second and the third power routers. Figure 11 depicts the responses when $P_{ac\_1}=200W$, $P_{ac\_2}=-200W$ and $P_{ac\_3}=0W$, i.e., when the first power router sent 200 W of power to the second router, while the third one only connected to the sub-grid and did not accommodate power. The figure indicates that the proposed method was able to control the accommodated power accurately while maintaining the voltage of the sub-grid at a constant and balancing the current between the positive and negative lines almost exactly.

Figure 12 and Figure 13 depict the voltages of the main grid and sub-grid, as well as the output currents of the grid-connected inverter $i_{grid}$ and the sub-grid-connected ($i_p$: positive voltage line; $i_n$: negative voltage line) inverters of power router 1 when the accommodation power $P_{ac\_1}$ increased from 0 W to 200 W and decreased gradually from 200 W to 0 W. The output currents of the sub-grid-connected inverters with the proposed control responded rapidly to the step changes of accommodation power almost without any delay and oscillation. The voltage of the sub-grid was stably maintained at 250 V. The main-grid-connected inverter had a slower response; however, its output current became stable within approximately 10 periods.

As mentioned in the introduction, to the best of the authors’ knowledge no research previous to this study has controlled the electrical power accommodated between prosumer nodes while maintaining the DC voltage of the sub-grid. Thus, it is unable to compare the experimental results presented in this paper with conventional work as a benchmark.

The power accommodation carried out in our simulations is summarized in

Table 1. The accommodation powers used in the simulations.

Accommodation Powers			Figure
Pac_1 [W]	Pac_2 [W]	Pac_3 [W]
0	0	0	Figure 8
200	−100	−100	Figure 9
200	−200	0	Figure 10

.

5. Conclusions

A power control method for a DC sub-grid in a hybrid electrical accommodation system, which is an enhancement of conventional main AC grids, is proposed in this paper. The power router, which also connects to the load, the main AC grid, PV generators, and batteries, connects to the sub-grid via several half-bridge inverters. A control method using an adaptive hysteresis current-control technique is proposed for controlling the sub-grid-connected inverters. The proposed method enables the sub-grid-connected inverters to maintain the DC voltage of the sub-grid while accommodating the given power. The proposed control method also balances the currents between the positive and negative lines of the sub-grid simultaneously. In the proposed system, the prosumers connected to the sub-grid contribute to maintaining the voltage of the sub-grid equally, which is an essential feature of P2P power accommodation without synchronization. The proposed control method was implemented on a high-speed FPGA-based controller and evaluated on three prototypes of the power router. The experimental results show that the inverters with the proposed control method yielded fast and stable responses. The power could be accommodated between the power routers via the sub-grid exactly. Although the proposed control method has been experimentally verified for the inverters, which have a low rated power, it can also be expected to be valuable for higher-rated power systems.