Voltage Compensation and Energy Supplement-Based Power Conversion System and Its Experimental Verification

Abstract

A voltage compensation and energy supplement-based power conversion system (VC-PCS) is proposed in this paper. A VC-PCS is a series connected to a battery that only processes fractional battery power; therefore, it has lower rated voltage and power in comparison with traditional PCSs. Compared with results of previous studies, the VC-PCS proposed herein can output bipolar voltage to meet voltage variation requirements and is a necessary and comprehensive improvement to previous research. First, its topology structure, working principle, and control strategy are comprehensively examined in this paper. Next, results of the simulations and physical experiments carried out are presented; the results show that the proposed VC-PCS is feasible in principle, and can be realized in physical applications.

1. Introduction

A power conversion system (PCS) is used to connect a power supply with a battery energy storage system (BESS); it requires voltage conversion and bidirectional power transmission. A PCS can be used to realize ‘power peak shaving and power valley filling’ in a power system [1,2,3,4,5]. With the development of distributed generation, a percentage of a BESS can eliminate power fluctuation, and it can also improve the grid’s power quality and stability [6,7].

When a PCS is connected to an AC grid, it needs a single-stage DC/AC converter or a two-stage DC/DC converter cascaded with a DC/AC converter. Whether in a single-stage or two-stage converter, a PCS processes full battery power. A single-stage converter has the advantages of simpler implementation and lower cost; the tree-phase two-level converter has been widely used in PCS applications [8]. To improve its rated voltage and power quality, the neutral point clamped (NPC) three-level converter and multilevel converters, such as a flying capacitor multilevel converter and a modular multilevel converter (MMC), were subsequently adopted [9,10].

A two-stage converter should be used when the battery voltage cannot match the converter’s requirements. In many references, resonant soft switching technology has been studied to improve a PCS’s efficiency [11,12,13]; however, all these studies were based on full-power PCS. The full battery power is transmitted through the PCS, and power loss is still non-ignorable, no matter what type of soft switching technology is adopted.

Traditionally, a PCS transmits full battery power regardless of battery voltage variation. On the other hand, when the battery’s state of charge (SOC) changes significantly during the charging or discharging process, the battery voltage change is relatively small. For example, in a lithium iron phosphate battery, when the battery’s SOC changes from 80% to 20%, the battery’s voltage changes only 0.09 V, which is approximately 2.82% of its rated voltage (3.2 V) [14]; this means that the difference between the converter’s input and output voltages is very small.

Considering the battery behavior of voltage and SOC, a fractional power processing or partial power processing architecture was proposed [15,16,17,18]. A converter was series connected with a battery to process a fraction of the DC voltage; hence, the converter power was also greatly reduced. In [15,16,17], unipolar biasing voltage converters were studied, but applications were restricted by their unipolar characteristics. In [18], a bipolar voltage-based fractional power converter achieved lower power processing; however, as only some theoretical analysis was provided, it requires a strong physical verification.

So far, there is no comprehensive research regarding bipolar partial power converters, including theoretical analysis, simulation, and experimental verifications. A voltage compensation and energy supplement-based PCS (VC-PCS) is studied in this paper. As shown in Figure 1, it was series connected with a battery to compensate for voltage variation, and only processed fractional battery power. In addition, it needed auxiliary power to supplement energy.

During the charging process, when the battery voltage was higher than the DC grid voltage, the VC-PCS output negative voltage to compensate for the battery voltage increase and supplemented energy loss at the same time. During the discharging process, when the battery voltage was lower than the DC grid voltage, the VC-PCS output positive voltage to compensate for the battery voltage decrease and supplemented energy loss at the same time. In other words, the VC-PCS could charge and discharge the battery whether the battery voltage was higher or lower than the DC grid voltage.

This remainder of this paper is organized as follows. In Section 2 and Section 3, the VC-PCS topology and operation principle are introduced and thoroughly analyzed. Section 4 presents the controller design and the control diagram. In Section 5 and Section 6, results of simulations and prototype experiments are presented, respectively. Finally, Section 7 concludes the paper.

2. PCS with Voltage Compensation and Energy Supplement

Figure 2 shows the proposed VC-PCS topology; the VC-PCS consisted of a voltage compensation modular converter (VCM) and a dual active bridge converter (DAB). The VCM was composed of a full bridge (FB) converter (S1–S4) and an inductor L; the VCM generated bipolar voltage, and the inductor filtered current. The DAB provided stable DC voltage for the VCM, and functioned as a current commutation path and energy supplement unit.

According to working state and voltage differences between the battery and the DC grid, the VCM operated in four different states:

(1) CCC: constant current (or voltage) charging state.

(2) CCD: constant current discharging state.

(3) CCC-NVC: CCC with negative voltage compensation.

(4) CCD-PVC: CCD with positive voltage compensation.

3. Principle Analysis

Clearly stated, the DAB could be equivalent to a constant bidirectional DC power v_dc. According to the working state and battery voltage, the principled analysis is given as follows.

3.1. CCC State Analysis

In charging state (i₁ < 0), if the battery voltage is lower than normal (v₂ < v₁), the CCC can be realized without voltage compensation, and it only needs to control the charging current.

As shown in Figure 3a,b, when S₂ is on, the current flows along L-S₂-S₄D, and the FB voltage is 0. When S₂ is off, the current flows through L-S₁D-v_dc-S₄D, the current is commuted to reduce current ripple, energy is fed back to the DC power, and the FB voltage is v_dc. Figure 3c shows the FB voltage; it is a positive pulse wave, and the VCM voltage satisfies

$$L\frac{\mathrm{d}{i}_1}{\mathrm{d}t}=v_1-v_2-v_{\mathrm{dc}}+D_1{v}_{\mathrm{dc}}$$

(1)

where D₁ is the duty ratio of S₂, and the current increases with an increase in the duty cycle; so, the current can be positively controlled by the duty cycle.

3.2. CCC-NVC State Analysis

In charging state (i₁ < 0), if the battery voltage is higher than normal (v₂ > v₁), a negative voltage is needed to realize battery charging, i.e., v₃ < 0 and v₂ + v₃ = v₁; the VCM power is positive, so the VCM supplements energy to the battery.

As shown in Figure 4a,b, when S₂ and S₃ are both on, the current flows through L-S₂-v_dc-S₃, the DC power supplements energy to the battery, and the FB voltage is -v_dc. When S₂ and S₃ are both off, the current flows through L-S₁D-v_dc-S₄D, the current is commuted to reduce current ripple, energy is fed back to the DC power, and the FB voltage is v_dc.

Figure 4c shows the FB voltage in the CCC-NVC state; the FB voltage is a positive or negative pulse wave, and the VCM voltage satisfies

$$L\frac{\mathrm{d}{i}_1}{\mathrm{d}t}=v_1-v_2-v_{\mathrm{dc}}+2D_2{v}_{\mathrm{dc}}$$

(2)

where D₂ is the duty ratio of S₂ and S₃, and the current increases with an increase in the duty cycle; so, the current can be positively controlled by the duty cycle.

3.3. CCD State Analysis

In discharging state (i₁ > 0), if the battery voltage is higher than normal (v₂ > v₁), the CCD can be realized without voltage compensation.

As shown in Figure 5a,b, when S₁ is on, the current flows along S₃D-S₁-L, and the FB voltage is 0. When S₁ is off, the current flows through S₃D-v_dc-S₂D-L, the current is commuted to reduce current ripple, energy is fed back to the power, and the FB voltage is -v_dc. Figure 5c shows the FB voltage in the CCD state; the FB voltage is a negative pulse wave, and the VCM voltage can be described by (1). Similarly, the current can be positively controlled by the duty cycle.

3.4. CCD-PVC State Analysis

In discharging state (i₁ > 0), if the battery voltage is lower than normal (v₂ < v₁), a positive voltage is needed to realize battery discharging, i.e., v₃ > 0 and v₂ + v₃ = v₁; the VCM power is also positive, so the VCM supplements energy to the DC grid.

As shown in Figure 6a,b, when S₁ and S₄ are both on, the current path is S₄-v_dc -S₁-L, the DC power supplements energy to the DC grid, and the FB voltage is v_dc. When S₁ and S₄ are both off, the current is commuted to reduce current ripple, energy is fed back to the DC power, and the FB voltage is -v_dc. Figure 6c shows the FB voltage in the CCD-PVC state; the FB voltage is a positive or negative pulse wave, and the VCM voltage can be described by (2). Similarly, the current can be positively controlled by the duty cycle.

4. Controller Design

According to the above analysis, switches S₁ and S₄ are always off in the charging state; if the battery voltage is lower than normal, only S₂ is controlled to realize constant current (or constant voltage), charging while S₃ remains off. Otherwise, S₂ and S₃ are controlled to compensate for the negative voltage and realize constant current (or constant voltage) charging.

Switches S₂ and S₃ are always off in the discharging state; if the battery voltage is higher than normal, only S₁ is controlled to achieve constant current discharging. Otherwise, S₁ and S₄ are controlled to compensate for the positive voltage and realize constant current discharging, as shown in

Table 1. State and control analysis.

	State	Battery Voltage	Switch Control	FB Voltage Polarity	Power Supply’s Function
Charging (S1 = S4 = 0)	CCC	v2 < v1	S3 = 0, S2	-	current commutation
	CCC-NVC	v2 ≥ v1	S2, S3	Negative	supplement energy /current commutation
Discharging (S2 = S3 = 0)	CCD-PVC	v2 < v1	S1, S4	Positive	supplement energy /current commutation
	CCD	v2 ≥ v1	S4 = 0, S1	-	current commutation

.

PI controls are used in practical applications, as shown in Figure 7. In the control mode selection, when F_chg1 is 0, the constant voltage control is selected, and the battery voltage v₂ and reference v₂* are used in the PI2 controller to obtain current reference i_ref. Next, the current i₁ and reference i_ref are used in a P controller to obtain the duty ratio D₁; the final duty ratio is equal to D₁. When F_chg1 is 1, the constant current control is selected, and the current i₁ and reference i₁* are used in the PI1 controller to calculate the duty ratio D₂ directly; the final duty ratio is equal to D₂.

In the control state decision, when F_chg2 is 0, battery charging is selected. If v₂ < v₁, the converter operates in the CCC state; otherwise, it operates in the CCC-NVC state. When F_chg2 is 1, battery discharging is selected. If v₁ < v₂, the converter operates in the CCD state; otherwise, it operates in the CCD-PVC state. According to the analysis in Section 3, the switches S₁, S₂, S_3, and S₄ are used in this order.

5. Simulation

In order to verify VC-PCS’s principle and control strategy, the SIMULINK based simulation model was established, as shown in Figure 8, and simulation verifications were carried out. The main circuit parameters used in the simulations are shown in

Table 2. Simulation parameters of main circuit.

Number	Parameter	Value
1	Grid Voltage (V)	50
2	Battery Voltage (V)	40–60
3	Power Supply Voltage (V)	20
4	Inductance (µH)	165
5	Switching Frequency (kHz)	25
6	Proportional Parameter (kp2)	0.1
7	Integral Parameter (ki2)	10.0
8	Proportional Parameter (kp1)	1.2
9	Integral Parameter (ki1)	20.0

.

5.1. Constant Current Charging Simulation

As shown in Figure 9, at first, the battery voltage (40 V) was lower than the DC grid voltage, and the VC-PCS charged the battery in the CCC state with a constant current of −15 A. At time t₁, the current reference was set as −30 A, and the current tracked the reference quickly. At time t₂, the battery voltage was suddenly set as 60 V; the battery voltage increased and was higher than the DC grid voltage, the VC-PCS compensated negative voltage v_FB and operated in the CCC-NVC state, the current underwent a transient process, and then quickly returned to the steady state.

This shows that the VC-PCS could quickly track the charging step command and charge the battery whether the battery voltage was lower or higher than the DC grid voltage.

5.2. Constant Current Discharging Simulation

As shown in Figure 10, at first, the battery voltage (60 V) was higher than the DC grid voltage, and the VC-PCS discharged the battery in the CCD state with a constant current of 15 A. At time t₃, the current reference was set as 30 A, and the current tracked the reference quickly. At time t₄, the battery voltage was suddenly set as 40 V; the battery voltage was lower than the grid voltage, the VC-PCS compensated positive voltage and operated in the CCD-PVC state, the current underwent a transient process, and then quickly returned to the steady state.

This shows that the VC-PCS could quickly track the discharging step command and discharge the battery whether the battery voltage was higher or lower than the DC grid voltage.

5.3. Charging and Discharging Simulation

As shown in Figure 11a, the battery voltage (40 V) was lower than the DC grid voltage. At first, the VC-PCS charged the battery in the CCC state with −30 A. At time t₅, the VC-PCS changed to discharge the battery in the CCD-PVC state, and the current reference was 30 A. This showed that the current tracked the command quickly.

As shown in Figure 11b, the battery voltage (60 V) was higher than the DC grid voltage. At first, the VC-PCS discharged the battery in the CCD state with 30 A. At time t₆, the VC-PCS changed to charge the battery in the CCC-NVC state, and the current reference was −30 A. This shows that the current tracked the command quickly.

In all, whether the battery voltage was higher or lower than the DC grid voltage, the VC-PCS could charge or discharge the battery arbitrarily and realize a fast transition between charging and discharging states with good transient performance.

6. Experiment

To further verify the principle and control strategy, a physical prototype was built, as shown in Figure 12. An electronic load APL-II (Myway company) simulated a DC grid, and a bidirectional power PSB9750 (EA company) simulated the power supply. Moreover, five lead-acid battery cells (LCPA33-12) were used as an experimental battery. In the experiment, two THDP0200 (TEK company, ratio 100/1) voltage probes and one P5200A (TEK company, ratio 50/1) voltage probe measured voltages, and a current probe TCP0150 (TEK company) measured current.

6.1. Experiments during Charging Process

Figure 13 shows voltages and current waveforms during the charging process. As shown in Figure 13a, when the battery voltage was lower than the DC grid voltage, the VC-PCS charged the battery with −3 A; it was operating in the CCC state, and the FB voltage was a positive pulse wave.

As shown in Figure 13b, when the battery voltage was higher than the DC grid voltage, the VC-PCS charged the battery with −3 A; it was operating in the CCC-NVC state. The FB voltage was a negative or positive pulse wave, and the negative voltage was greater than the positive voltage, so a negative voltage was compensated.

The experimental results showed that whether the battery voltage was lower or higher than the DC grid voltage, the VC-PCS could charge the battery with the same currents; the waveforms were always consistent with the theoretical analysis.

6.2. Experiment during Discharging Process

Figure 14 shows the voltages and current waveforms during the discharging process. As shown in Figure 14a, when the battery voltage was higher than the DC grid voltage, the VC-PCS discharged the battery with 3 A; it was operating in the CCD state, and the FB voltage was a negative pulse wave.

As shown in Figure 14b, when the battery voltage was lower than the DC grid voltage, the VC-PCS discharged the battery with 3 A; it was operating in the CCD-PVC state. The FB voltage was a positive or negative pulse wave, and the positive voltage was greater than the negative voltage, so a positive voltage was compensated.

The experimental results show that whether the battery voltage was higher or lower than the DC grid voltage, the VC-PCS could discharge the battery with the same currents; the waveforms were consistent with the theoretical analysis.

6.3. Experiment during State Transition Process

Figure 15a shows state transition waveforms when the battery voltage was lower than the DC grid voltage. At first, the VC-PCS discharged the battery with 3 A, and it operated in the CCD-PVC state. At time t₁, a charging battery command was set, the VC-PCS began to charge the battery with −3 A, and it operated in the CCC state. At time t₂, a discharging battery command was set again, the VC-PCS began to discharge the battery with 3 A, and it operated in the CCD-PVC state again.

Figure 15b shows state transition waveforms when the battery voltage was higher than the DC grid voltage. At first, the VC-PCS charged the battery with −3 A, and it operated in the CCC-NVC state. At time t₃, a discharging battery command was set, the VC-PCS began to discharge the battery with 3 A, and it operated in the CCD state. At time t₄, a charging battery command was set again, the VC-PCS began to charge the battery with −3 A, and it operated in the CCC-NVC state again.

This shows that whether the battery voltage was higher or lower than DC grid voltage, the VC-PCS could charge or discharge the battery arbitrarily; it could also realize a fast transition between charging and discharging states with good transient performance, and was always consistent with the simulation results.

7. Conclusions

Compared with traditional PCS, the proposed VC-PCS has lower-rated voltage and power, and is an alternative to power energy storage applications. Its theoretical analysis, simulation, and experimental verification are analyzed in this paper; it is a necessary and comprehensive improvement to previous research, especially with regard to physical verification. This paper shows that a VC-PCS can operate normally regardless of voltage variation and state transition, and that it is feasible in principle. In the future, we will conduct further research, including a technical economy analysis and an efficiency comparison, and will try to solve technical problems regarding practical applications.