15 Modeling Of A Voltage Controlled Vsi And Current Controlled Vsi

782

IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 21, NO. 3, SEPTEMBER 2006

Application of Voltage- and Current-Controlled Voltage Source Inverters for Distributed Generation Systems Sung-Hun Ko, Seong R. Lee, Hooman Dehbonei, Member, IEEE, and Chemmangot V. Nayar, Senior Member, IEEE

Abstract—Voltage source inverters (VSI) have been widely used in uninterruptible power supplies, unified power flow controllers or unified power quality conditioners, and distributed generation systems (DGS). VSIs are inherently efficient, compact, and economical devices used to control power flow and provide quality supply. VSIs can be classified as voltage-controlled VSIs (VCVSIs) and current-controlled VSIs (CCVSIs), depending on their control mechanism. In this paper, a detailed comparison of VCVSIs and CCVSIs for DGS applications is presented. This paper examines the advantages and limitations of each control technique in a single-phase DGS, without incorporating additional hardware and/or extra complex control techniques. Discussions on the concepts, hypotheses, and computer simulations of different VSIs in the presence of different loads and conditions are presented. The experimental results confirm the validity of the analysis and simulations outlined. The paper provides design recommendations for the use of VCVSIs and CCVSIs in various applications. Index Terms—DC–AC power conversion, energy conversion, power conditioner, power electronics.

I. INTRODUCTION UE to improvement in technologies, electrical power can be generated more efficiently and closer to the point of consumption. Additionally, distributed generation systems (DGS) enable alternative energy sources (AES) to easily utilize and supplement fossil fuels. Renewable energy sources (RES) (e.g., solar, wind, biomass, wave, hydropower, etc.) can play a major role in the preservation of our underground resources and the reduction of air pollutants. DGS based on RES have been known to be one of the most cost-effective, reliable, and durable power systems to provide energy saving and noninterrupted power with high power quality [1]–[5]. DGS can be classified further into stand-alone and grid connected systems (series and parallel processing), according to the output of the voltage source inverters (VSIs) and connection to other ac sources and loads [6]. Typical examples of other ac sources are any available grid (strong, weak, or diesel grids) or other DGS sources. VSIs are inherently efficient, compact, and economical and offer

D

Manuscript received October 25, 2005; revised February 7, 2006. This work was supported in part by the Australian Research Council under Grant LP 0348994 and in part by the Postdoctoral Fellowship Program of the Ministry of Commerce, Industry & Energy (MOCIE). Paper no. TEC-00360-2005. S.-H. Ko and S. R. Lee are with the School of Electronic & Information Engineering, Kunsan National University, Kunsan 573-701, Korea. H. Dehbonei and C. V. Nayar are with the Department of Electrical and Computer Engineering, Curtin University of Technology, Perth 6854, Australia (e-mail: [email protected]). Digital Object Identifier 10.1109/TEC.2006.877371

numerous functions that require a minimum number of power conversions [7]–[11]. The parallel processing DGS controls power flow and quality by controlling the power conversion between the dc bus of bi-directional VSIs and the available grid [12], [13]. The bi-directional VSIs can be further classified into VCVSIs and CCVSIs, depending on their control mechanism [14]. In DGS, VCVSIs use the amplitude and phase of an inverter output voltage relative to the grid voltage to control the power flow [15]. In VCVSIs, the desired current flow is generated by controlling the voltage across the decoupling inductor. The CCVSI uses switching instants to generate the desired current flow in the VSI’s inductor filter, using instantaneous current feedback [16]. There are advantages and limitations associated with each control mechanism. For instance, VCVSIs provide voltage support to the load (the VSI operating as a voltage source), while CCVSIs provide current support (the VSI operating as a current source). The CCVSI is faster in response compared to the VCVSI, as its power flow is controlled by the switching instant, whereas in the VCVSI, the power flow is controlled by adjusting the voltage across the decoupling inductor. Active and reactive power is controlled independently in the CCVSI, but are coupled in the VCVSI. Generally, the advantages of one type of VSIs are considered as a limitation of the other type. In this paper, a detailed comparison of VCVSIs and CCVSIs is investigated for DGS under various conditions. The experimental results on a scaled-down version (1 kVA) of DGS, confirm the validity of theoretical and simulation studies. The design consideration and summary of different VSI controls is presented in Section V.

II. PARALLEL PROCESSING DISTRIBUTED GENERATION SYSTEMS A typical configuration of the parallel processing DGS using a VSI is shown in Fig. 1. This system consists of a VSI, which is connected in parallel to the grid for a CCVSI and through a decoupling inductor for a VCVSI. It is generally expected that the VSI performs the following functions in DGS [7], [17]–[19]: 1) Load voltage stabilization (±5% voltage regulation) in both parallel processing and stand-alone modes; 2) Uninterruptible power supply (UPS); 3) Reactive power support—grid power conditioning including power factor correction (>0.9) and harmonics mitigation (THD<5%) (only in parallel processing mode) as per IEEE standard 1159 [17];

0885-8969/$20.00 © 2006 IEEE

KO et al.: APPLICATION OF VOLTAGE- AND CCVSIs FOR DISTRIBUTED GENERATION SYSTEMS

Fig. 1.

783

Schematic diagram of a parallel processing DGS.

Fig. 3. Phasor diagram of a VCVSI-based DGS with resistive load and grid is responsible for supplying the active power.

Fig. 2.

The equivalent circuit diagram of a VCVSI-based DGS.

4) Active power support—load power conditioning including demand side management (DSM). In this mode of operation, a bi-directional VSI is responsible for controlling the active power flow between the dc bus and the ac grid. A. Voltage-Controlled VSIs in DGS Fig. 2 shows the simplified/equivalent schematic diagram of a VCVSI. For the following analysis it is assumed that the output low pass filters (Lf and Cf ) of VSIs will filter out high-order harmonics generated by pulse width modulations (PWMs). The decoupling inductor (Xm ) is an essential part of any VCVSI as it makes the power flow control possible. In a VCVSI, the power flow of the DGS is controlled by adjusting the amplitude and the phase [power angle (δ)] of the inverter output voltage with respect to the grid voltage. Hence, it is important to consider the proper sizing of the decoupling inductor and the maximum power angle to provide the required power flow when designing VCVSIs. Assuming the maximum permissible voltage fluctuation in the grid voltage (Vg ) is ±20% and the grid has to supply the active power demanded by a resistive load, the phasor diagram of the VCVSI-based DGS is shown in Fig. 3. In this figure, it is assumed that the voltage of the inverter has to be kept constant (Vc1 = Vc2 = Vc3 , load voltage stabilization). Fig. 3 shows that as the VCVSI voltage retains constant, any changes in the grid voltage to control the desired power flow, the power angle has to change in proportion. The power angle could be both lagging or leading, providing either the active power flow from the grid to the VCVSI or vice versa. Fig. 3 shows that the lagging power angles result in active power from the grid towards the inverter, regardless of the grid voltage’s amplitude and minimum power angle obtains when the grid and the VCVSI voltage are identical. This figure shows that reactive power always flows from the higher voltage source to the lower voltage source. Hence, the higher voltage source has to supply all the reactive power demanded by the decoupling inductor as well as load. In weak

grid applications, when the grid voltage drops considerably, the VCVSI has to supply both the rated active power and full reactive power, resulting in over sizing of the inverter (>100% of the rated power). Unity power factor operation (Ig p 1 = Ig 1 ) is only possible if the grid voltage is reduced to Vg 1 and at the specific power flow corresponds to Vx1 . This is a special case, which depends on the size of the decoupling inductor, the load and maximum permissible power angle. Therefore, power factor correction is not possible using VCVSIs in DGS. This is one of the main drawbacks of VCVSI-based DGS. Using Fig. 2 the fundamental grid current can be expressed as (1) Ig =

Vg
0 − Vc
δ Vc sin δ Vg − Vc cos δ =− −j jXm Xm Xm (1)

where Vg and Vc are, respectively, the grid and the VCVSI’s fundamental voltages, and Xm is the decoupling inductor impedance. Using per unit values (Sbase = 2 Vbase /Zbase , Vbase = Vc and Zbase = Xm ) where Vbase , Zbase , and Sbase are the base voltage, impedance, and complex power values, respectively, the grid apparent power can be expressed as (2):
2  − Vgpu cos δ . (2) Sgpu = −Vgpu sin δ + j Vgpu Using per unit values, the complex power of the VCVSI and decoupling inductor are Scpu = −Vgpu sin δ + j[Vgpu cos δ − 1]
2  Sxpu = j Vgpu − 2Vgpu cos δ + 1

(3) (4)

where Sgpu , Scpu , and Sxpu are per unit values of the grid, VCVSI and decoupling inductor apparent power respectively, and Vgpu is the per unit value of the grid voltage. As addressed above, since the load voltage must remain constant (load voltage stabilization), the only controllable parameter in the VCVSI is the power angle (δ). Hence, the power angle is used in a VCVSI for DSM operation. For DSM operation, it is important to extract the maximum power from RES and supply this power to the load or DGS. Assuming that both RES and the grid are supplying the demanded active power by the load, the

784

IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 21, NO. 3, SEPTEMBER 2006

Fig. 6.

Fig. 4. Control block diagram of the VCVSI-based DGS used in the simulations and experiments.

Fig. 5.

The equivalent circuit of a CCVSI-based DGS.

power angle can be calculated from (5) Pg = PL − PRES =

Vg Vc sin δ. Xm

Thus, the power angle (δ) is   (PL − PRES )Xm δ = sin−1 Vg Vc

(5)

(6)

where PL and PRES are the load and RES active power, respectively. Equation (6) explains that as the available RES energy is increased, the power angle is reduced. This means that the RES’ penetration will increase. Fig. 4 shows the block diagram of a VCVSI control system based DGS used in both the simulations and experiments. This control block diagram includes the DSM function (6). If the RES available energy is more than the load consumption, then the power angle can be leading to export this extra active power to the DGS. In Fig. 4,the phase locked loop (PLL) is responsible for synchronizing the inverter output voltage with the grid voltage. The sampling from the load current, RES voltage and current is also ∗ ) (for DSM opused to generate the required power angle (δref eration). After comparing the required/reference values and the actual variables, an error signal is generated to feed a PI controller. After generating the desired reference signal, it is given to the PWM generator block to generate the required switching signals. B. Current-Controlled VSIs in DGS Fig. 5 shows the equivalent schematic diagram of a CCVSI. As CCVSI controls the current flow using the VSI switching instants, it can be modeled as a current source and there is no need for a decoupling inductor (Fig. 5). As the current gener-

Phasor diagram of a CCVSI-based DGS with inductive load.

ated from the CCVSI can be controlled independently from the ac voltage, the active and reactive power controls are decoupled. Hence, unity power factor operation is possible for the whole range of the load. This is one of the main advantages of CCVSIs. As the CCVSI connects in parallel to the DGS, it follows the grid voltage. Fig. 6 shows the phasor diagram of a CCVSIbased DGS in the presence of an inductive load (considering the same assumption as VCVSI section). Fig. 6 shows that when the grid voltage increases, the load’s active power consumption, which supplied by the grid increases and the CCVSI compensates the increase in the load reactive power demand. In this case, the CCVSI maintains grid supply at unity power factor, keeping the current phase delay with respect to the grid voltage at a fixed value (Θ). Therefore, the CCVSI cannot maintain the load voltage in the presence of a DGS without utilizing extra hardware and control mechanisms. This limitation on load voltage stabilization is one of the main drawbacks of CCVSI-based DGS. Assuming the load active current demand is supplied by the grid (reactive power support function), the required grid current can be rewritten as follows:   SL ∗ (7) Ig = Re[IL ] = Re Vg where SL is the demanded load apparent power. For grid power conditioning, it is preferred that the load operate at unity power factor. Therefore, the CCVSI must provide the remainder of the required current (8) Ic = IL − Ig∗ .

(8)

In DSM, it is desirable to supply the active power by the RES, where excess energy from the RES is injected into the DGS. The remaining load reactive power will be supplied by the CCVSI. Hence, (8) can be rewritten as (9)   SL − PRES ∗ Ig = Re[IL ] − Re[Ic ] = Re (9) Vg Equations (8) and (9) show that in the worst case, the CCVSI has to supply both the active and reactive power demanded by the load. This means that the CCVSI sizing can be rated at full load without the need to oversize. This is an advantage of CCVSI-based DGS compared to the VCVSI. The control block diagram of the CCVSI-based DGS used in the simulation and experiment using (9) and (10) is shown in Fig. 7. The CCVSI control scheme samples the DGS voltage (Vg ) for synchronization using a PLL. The samples of load current (ILoad ), RES current (IRES ), and voltage (VRES ) are used to

KO et al.: APPLICATION OF VOLTAGE- AND CCVSIs FOR DISTRIBUTED GENERATION SYSTEMS

785

Fig. 7. Control block diagram of the CCVSI-based DGS used in the simulations and experiments.

TABLE I SIMULATION CONDITIONS AND SELECTED PARAMETERS WHERE L m IS THE DECOUPLING INDUCTOR, L f IS THE FILTER INDUCTOR, AND C f IS THE FILTER CAPACITOR

generate the desired inverter current amplitude (Id∗ ) (reference current) using (9) and (10). After generating the reference current signal (Ic-ref )), this current is compared to the instantaneous CCVSI current in order to generate the error current (Ierr ). This error current is then given to the current regulator block to generate the desired instantaneous switching PWMs. III. SIMULATION RESULTS To compare the performance of the parallel processing DGS using a VCVSI and CCVSI, a 1kVA system including linear and nonlinear loads was simulated using PSim software. Table I shows the different parameters and selected values identical with the experimental hardware used in simulation, to provide a foundation to compare results. A. Power Conditioning in DGS This simulation was conducted to evaluate the performance of the different VSIs in the presence of different loads, where Vg and Vc are the voltage waveforms of the grid and inverter, and Ig , Ic , and Iload are current waveforms of the grid, inverter and load, respectively. Fig. 8 shows the power conditioning of a DGS in the presence of an inductive load (Z = 40
36.7◦ [Ω]), using different VSIs. Fig. 8(a) shows that the grid can supply the load’s active power when the grid voltage is almost the same as the VCVSI voltage at a low power angle. In this case, the required reactive power (600 var) supplied by the VCVSI, the power factor is good as the inverter size and hence the decoupling inductor is relatively small. Fig. 8(b) shows that the CCVSI can supply the reactive power required by the load while

Fig. 8. Waveform results of power conditioning of a DGS in the presence of an inductive load (z = 40
36.7◦ [Ω]). (a) VCVSI. (b) CCVSI.

the active power is supplied fully by the grid in the same way as for VCVSI. Fig. 9(a) shows the VCVSI as a power conditioner for a DGS in the presence of a nonlinear load (a RLC diode bridge rectifier). The VCVSI cannot maintain pure sinusoidal voltage across the nonlinear load (Vc ). Hence, a portion of low-order current harmonics will be injected into the grid (Fig. 10(b), ITHD = 10.9%). Fig. 9(b) shows that the CCVSI can provide all the reactive power demanded by the nonlinear load and hence the grid supplies only the remaining active power (unity power factor operation). In this case, the CCVSI prevents any low-order harmonics from being injected into the grid (active filtering) (Fig. 10(c), ITHD = 1.1%). Fig. 10(a) shows the DGS in the absence of VSIs. In this case, all the reactive power associated with low-order harmonics from the nonlinear load must be supplied by the grid (ITHD = 60.8%). This figure also signifies that a VCVSI cannot meet the IEEE standard 1159 (less than 5% of THD) when a nonlinear load is presented, while a CCVSI can achieve unity PF and satisfies THD requirements of voltage and current for the full range of the load, without the need for an additional controller (assuming that the grid voltage is sinusoidal). B. UPS Function in DGS To study the performance of each VSI controller in UPS mode, the system was simulated in the presence of nonlinear

786

IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 21, NO. 3, SEPTEMBER 2006

Fig. 9. Waveform results of power conditioning of a DGS in the presence of a nonlinear load (a RLC diode bridge rectifier). (a) VCVSI. (b) CCVSI.

loads (Fig. 11). It was assumed that at 30 ms the grid fails and both VSIs had to supply the load. As it is shown, before grid failure the VCVSI supplied the reactive power demanded by the nonlinear load and the rest of the reactive power demanded by the decoupling inductor supplied by the grid (when both the grid and the VCVSI voltages were identical). This figure shows that the VCVSI picked up the load rapidly after the grid failed [Fig. 11(a)]. Fig. 11(b) shows that the CCVSI provides/absorbs almost all the nonlinear current to/from the load before grid failure and full load current afterwards to supply the load. It is shown in the presence of a nonlinear load that a CCVSI cannot provide and maintain a sinusoidal voltage waveform even with extra voltage feedback [Fig. 11(b)]. However, the VCVSI can provide the required voltage support and UPS function, even in the presence of a nonlinear load and without the need for extra feedback or load estimation control algorithms. It is shown that in the presence of a nonlinear load, the VCVSI can maintain the load voltage VTHD at 11.2% while the load current ITHD is 50.7%. These values can be read as high as 36.3% load voltage VTHD and as low as 0.7% load current ITHD in CCVSI and stand-alone operations. The significance of this data is that both the VCVSI and the CCVSI cannot compensate low-order harmonics of the load voltage in order to meet IEEE standards (eg., 1159 and 944) in the presence of

Fig. 10. The harmonic spectrum of the grid current in the absence and presence of VSIs supplying nonlinear load. (a) Without power conditioning. (b) VCVSI. (c) CCVSI.

nonlinear loads, without extra feedback and complex control algorithms [8]. C. DSM Function in DGS In RES-based DGS, it is required to give priority of supply to the RES and reduce the share of the grid to supply the load. If the RES is not enough, then both the RES and the grid will supply the load demand. Fig. 12 illustrates the case that the grid is the only available source to supply the required active power demanded by the linear load (100% of 1 kw) and suddenly, when the RES becomes available (50%), allows the RES to take part and supply 50% of the load demand. Fig. 12(a) shows that the VCVSI maintains load voltage while the RES begins to supply 50% of the load demand. In this case, the grid current is reduced to 50% while the load current is maintained at 100%. Hence, the load does not detect any abnormality in the supply. Fig. 12(b) shows that a CCVSI can perform DSM and supply the load (at 50%) in the same way as a VCVSI. The delay in the power waveforms in Fig. 12 are due to the existence of a low pass filter in the power meter used for power measurements in

KO et al.: APPLICATION OF VOLTAGE- AND CCVSIs FOR DISTRIBUTED GENERATION SYSTEMS

787

Fig. 11. UPS mode waveform results of a DGS in the presence of nonlinear load. (a) VCVSI. (b) CCVSI. Fig. 12. Waveform results of demand side management in DGS (grid supplies linear load from 100% to 50%). (a) VCVSI. (b) CCVSI.

PSim software. Both power diagrams in Fig. 12 show the DSM capability of CCVSIs and VCVSIs. D. Voltage Regulation in DGS In this simulation, the grid voltage was changed from its nominal value (here 200 V) to 160 V. It was assumed that the grid has to supply the load active power. Fig. 13(a) shows the voltage stabilization for the load when the grid voltage fluctuated in the presence of a VCVSI. After a step down in the DGS voltage from 200 to 160 V, the grid can still supply the active power while the VCVSI maintains the load voltage. In this case, the VCVSI has to supply the extra reactive power demanded by the decoupling inductor, which is dependent on the DGS voltage and the inductor’s size. In this case, the grid current increases due to an increase in reactive current flow from the VCVSI to the grid. Fig. 13(b) shows the voltage stabilization for the load when the grid voltage fluctuated in the presence of a CCVSI. Due to the direct connection of the CCVSI to the DGS, it cannot compensate the grid voltage fluctuations without additional hardware and control feedback algorithm [20], [21]. Hence, the load voltage cannot be maintained and the load will suffer from grid voltage fluctuations. As it is assumed in this case that the DGS must supply the active power demanded by load, the CCVSI current remains at zero even after the DGS voltage

drops. In this case, the load demand was reduced in proportion to the decreases in the DGS voltage, and hence resulted in less active power support by the DGS. IV. EXPERIMENTAL RESULTS Fig. 14 shows a photograph of a scale down version of a DGS that was prototyped to examine the analytical and simulation analysis. The experimental setup consists of a computer to monitor and program the desired control techniques (Figs. 4 and 7) into a digital signal processor (DSP), to provide a switching signal to a control board and a VSI. A variac is used to simulate a weak grid, while different loads are connected to the output of the VSIs. The scope and power analyzers were used to record the information for further evaluative comparisons of the analytical and simulation results. System specifications are given in Table II. As a battery bank of 180 V was used in this test, a low frequency transformer was utilized to step-up the output voltage of the H-bridge inverter to the required value (200 V). The Voltech (PM3000A) power meter measures a power factor of over 0.99 for a CCVSI and for a VCVSI, from 0.97 to 0.99, depending on the load and grid voltage fluctuation for linear loads. This relatively good power factor for the VCVSI

788

IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 21, NO. 3, SEPTEMBER 2006

Fig. 13. Waveform results of voltage stabilization when the grid voltage changes from 200 to 160 V in DGS in the presence of a linear load (R = 40 [Ω]). (a) VCVSI. (b) CCVSI.

Fig. 15. Experimental waveform results of power conditioning of a DGS in the presence of an inductive load (z = 40
36.7◦ [Ω]). (a) VCVSI. (b) CCVSI.

is due to the low power operation of the DGS (1 kVA). The Tektronix (TDS3054B) digital scope was used to capture the following results. A. Power Conditioning in DGS

Fig. 14.

Photograph of the prototyped DGS.

TABLE II THE SPECIFICATIONS OF THE PROTOTYPED DGS

Fig. 15 shows that, in the presence of an inductive load (Z = 40
36.7◦ [Ω]), when the grid voltage is almost the same as the VCVSI voltage, a slight adjustment to the power angle will enable the grid to supply the load active power, and the required reactive power (600 var) can be supplied by the VCVSI. Fig. 15(b) shows that at unity power factor, the grid supplies the load active power and the CCVSI supplies the load reactive power (reactive power compensation). These results also comply with the simulation results (Fig. 8). Fig. 16 shows the experimental waveform results of the power conditioning of a DGS in the presence of a diode bridge rectifier with RLC (nonlinear load). It confirms that the CCVSI has better performance in the presence of a nonlinear load for low-order harmonic mitigation and provides unity power factor operation to the DGS [Fig. 16(b)], compared to the VCVSI [Fig. 16(a)]. As was expected, the output voltage of the VCVSI (Vc ) will be distorted in the presence of a nonlinear load [Fig. 16(a)]. This deformation in the voltage waveform can be

KO et al.: APPLICATION OF VOLTAGE- AND CCVSIs FOR DISTRIBUTED GENERATION SYSTEMS

Fig. 16. Experimental waveform results of power conditioning of a DGS in the presence of a nonlinear load (a diode bridge rectifier with RLC). (a) VCVSI. (b) CCVSI.

corrected by adding a wave-shaping control algorithm and using extra feedback signals [8]. B. UPS Function in DGS Experiment results for the study of the UPS function of different VSIs are shown in Fig. 17. It observed that both the VCVSI and CCVSI can supply and maintain the pure sinusoidal voltage waveform when the grid fails, in the presence of linear loads. However, in the presence of nonlinear loads, the CCVSI cannot provide the proper voltage without additional hardware and control feedback algorithms [Fig. 17(b)]. Although the VCVSI cannot mitigate low-order harmonics from the nonlinear loads, it can maintain the load voltage after the grid fails [Fig. 17(a)]. These results verify the simulation results (Fig. 11). Fig. 17 illustrates that after grid failure, the CCVSI can provide sinusoidal current waveform to the load by adding predictive control algorithms, however, this voltage will be distorted as it tries to keep the load current sinusoidal. C. DSM Function in DGS Demand side management is an important function in any DGS which defines a load sharing among the suppliers. In this experiment, to study the DSM function of a VCVSI and CCVSI, it was assumed that these VSIs would suddenly have to take 50% of the load active power supply. Fig. 18(a) shows that in a DGS

789

Fig. 17. Experimental waveform results of UPS function of a DGS using different VSI in the presence of nonlinear load (a RLC diode bridge rectifier). (a) VCVSI. (b) CCVSI.

system using a VCVSI, where the grid and the VCVSI voltages are identical, grid supplies the reactive power demanded by decoupling the inductor, while the grid supplies the active power. For demand side management, the proportions of active power to be supplied by the VCVSI and grid respectively can be controlled by changing the power angle (δ). In this experiment, the power angle was modified in order that 50% of the active power to be supplied by the VCVSI. In this situation, after the transient it was observed that without changes in the load current the grid current decreased while the VCVSI current increased. Fig. 18(b) shows a DGS in the presence of a CCVSI. In this case, as the grid was subject to supply the full resistive load there was no reactive power compensation (no decoupling inductor), hence the converter current was maintained at zero. After a command from the control system to overtake 50% of the active power by the CCVSI, it was observed that with a very smooth transient (no change in the load current) the CCVSI supplied the remaining 50% of the load-demanded active power. These results support the simulation results (Fig. 12). D. Voltage Regulation in DGS Voltage regulation is another important feature required in most applications dealing with sensitive loads. Moreover,

790

IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 21, NO. 3, SEPTEMBER 2006

Fig. 18. Experimental waveform results of demand side management of a DGS in the presence of a resistive load. (a) VCVSI. (b) CCVSI.

Fig. 19. Experimental waveform results of voltage stabilization in a DGS in the presence of a resistive load. (a) VCVSI. (b) CCVSI.

A. Load Voltage Stabilization voltage stabilization would be one of the most important requirements in weak grid applications. The following tests were carried out in order to study the performance of the VCVSI and CCVSI in stabilizing the voltage. In this experiment, the voltage of the grid was changed from 200 V to 160 V (Fig. 19). Fig. 19(a) illustrates that the VCVSI shares in supplying the reactive power demanded by the decoupling inductor with the grid, when the grid voltage is the same as the VCVSI voltage. This can be done by adjusting the power angle to allow the grid to supply all the active power demanded by the load. This experiment confirms the findings from the simulation in Fig. 13, namely that the VCVSI can maintain the load voltage regardless of changes in the grid voltage. However, due to the parallel connection of the CCVSI to the DGS, the CCVSI follows the grid voltage and hence cannot provide voltage support to the load without extra hardware and complex control techniques [Fig. 19(b)].

V. DESIGN CONSIDERATION AND COMPARISON OF VCVSIS AND CCVSIS IN DISTRIBUTED GENERATION SYSTEMS The comparison of the VCVSI and CCVSI-based DGS is shown below.

It is shown that the VCVSI can regulate the load voltage within ±5% as per IEEE standards (1159 and 944). In contrast, as a CCVSI is connected directly to the grid it cannot compensate the grid voltage fluctuation. A decoupling inductor is essential to decouple the effect of grid voltage fluctuation, which can be achieved by using VCVSIs. Therefore it is suggested that a VCVSI be used to provide the required voltage support to the load in applications with a weak grid [22], [23]. B. Uninterruptible Power Supply As a VCVSI by nature performs the same as a voltage source, it can maintain voltage support for the load in the absence of a grid (stand-alone operation). It is shown that the VCVSI cannot provide a pure sinusoidal waveform in the presence of a nonlinear load without extra control mechanisms and feedback. However, as is shown [8], wave shaping of the VCVSI is possible with extra feedback and hence the sinusoidal output voltage is guaranteed even in the presence of nonlinear loads and in stand-alone operations. On the other hand, the CCVSI cannot provide proper voltage support as by nature it is a current source and voltage follower. Therefore, a VCVSI is recommended for those applications where a UPS function is of high priority [24].

KO et al.: APPLICATION OF VOLTAGE- AND CCVSIs FOR DISTRIBUTED GENERATION SYSTEMS

C. Reactive Power Support, Active Filtering and Power Factor Correction As the active and reactive powers are coupled in a VCVSI, it generally offers poor power factor correction performance at low load, or when the grid voltage is different from the voltage of the load/VCVSI. In contrast, a CCVSI provides good reactive power support and decoupling from the active power. This capability enables CCVSIs to perform at unity power factor and to mitigate low-order harmonics effectively. Therefore, CCVSIs are recommended for those applications where reactive power support, including unity power factor operation or active filtering is the main goal, (i.e., active power line conditioners (APLC) [25], [26]). D. Active Power Support or DSM Both the VCVSI and CCVSI offer effective bidirectional power flow between their dc and ac bus. The power flow control in a VCVSI is very sensitive, and depends not only on a limited power angle, but also on the size of the decoupling inductor. However, as the phase and amplitude are controlled separately in a CCVSI, the power flow in a CCVSI is smoother due to the decoupling of the active and reactive power in this control technique. Hence, for DSM operation when voltage support is not a priority, CCVSIs are recommended (i.e., photovoltaic grid-tied inverters). E. Sizing the PCU Due to the existence of a decoupling inductor, a VCVSI must supply both the active and reactive power demanded from the load as well as the reactive power required by the decoupling inductor. This means that the VCVSI has to be oversized (>100%). This could be worsened by an increase in the voltage of the grid and VCVSI. However, the CCVSI can be rated at full load (100%), as there is no decoupling inductor and it only needs to supply the active and reactive power required by the load. In practice, it is possible to change the control algorithm in VSIs with respect to the different functions required. However, there are some prerequisites as well as pros and cons associated with changing the control algorithm in VSIs which must be considered. VI. CONCLUSION Voltage source inverters have been widely used in many applications, including distributed generation systems. VSIs are inherently efficient, compact and economical devices, which are used to control power flow and the quality of power supply. In this paper, a detailed comparison of VCVSIs and CCVSIs for DGS applications was presented. It was shown that neither VCVSIs nor CCVSIs alone can offer all the functions required in a DGS. Hence, the most appropriate VSI should be chosen based on its application and priority. Alternatively, both types of VSIs must be used for those applications where voltage stabilization, unity power factor operation and active filtering are required. This paper examined the advantages and disadvantages of each

791

VSI’s control techniques in the presence of different loads and provides design recommendations for the use of VCVSIs and CCVSIs in various applications. The experimental results verify the theoretical analysis and computer simulations. ACKNOWLEDGMENT The authors are grateful to Curtin University of Technology for providing opportunities to carry out this research. This work was supported in part by the Australian Research Council under Grant LP 0348994 and partly by the Post-doctoral Fellowship Program of the Ministry of Commerce, Industry & Energy (MOCIE).

REFERENCES [1] W. Wongsaichua, W.-J. Lee, S. Oraintara, C. Kwan, and F. Zhang, “Integrated high-speed intelligent utility tie unit for disbursed/renewable generation facilities,” IEEE Trans. Ind. Appl., vol. 41, no. 2, pp. 507–513, Mar.–Apr. 2005. [2] T. Jin, L. Li, and K. Smedley, “A universal vector controller for threephase PFC, APF, STATCOM, and grid-connected inverter,” presented at the 19th Ann. IEEE Applied Power Electronics Conf. Expo. (APEC’04), 2004, vol. 1, pp. 594–600. [3] K. Dai, P. Liu, J. Xiong, and J. Chen, “Study on dual-DSP-controlled threephase series-parallel compensated line-interactive UPS system (deltaconversion UPS),” presented at the IEEE Int. Electric Machines and Drives Conf. (IEMDC’03), Jun. 1–4, vol. 1, pp. 436–442. [4] F. Blaabjerg, Z. Chen, and S. B. Kjaer, “Power electronics as efficient interface in dispersed power generation systems,” IEEE Trans. Power Electron., vol. 19, no. 5, pp. 1184–1194, Sep. 2004. [5] M. Dai, M. N. Marwali, J.-W. Jung, and A. Keyhani, “Power flow control of a single distributed generation unit with nonlinear local load,” presented at the IEEE PES, Power Systems Conf. and Expo., Oct. 10–13, vol. 1, pp. 398–403. [6] M. Prodanovic and T. C. Green, “Control and filter design of three-phase inverters for high power quality grid connection,” IEEE Trans. Power Electron., vol. 18, no. 1, pt. 2, pp. 373–380, Jan. 2003. [7] A. J. Baronian and S. B. Dewan, “An adaptive digital control of current source inverter suitable for parallel processing inverter systems,” presented at the IEEE 30th Ann. Meeting, Industry Applications Conf., Oct. 8–12, vol. 3, pp. 2670–2677. [8] H. Dehbonei, “Power conditioning for distributed renewable energy generation” Ph.D. Dissertation, Dept. Elect. Comput. Eng., Curtin Univ. Technol., Perth, 2003. [9] T. Kawabata, K. Ogasawara, N. Sashida, Y. Yamamoto, and Y. Yamasaki, “Parallel processing inverter system,” IEEE Trans. Power Electron., vol. 6, no. 3, pp. 442–450, Jul. 1991. [10] H. Dehbonei, C. V. Nayar, and L. Borle, “A combined voltage controlled and current controlled ‘dual converter’ for a weak grid connected photovoltaic system with battery energy storage,” presented at the IEEE 33rd Annu. Power Electronics Specialists Conf. (PESC’02), Cairns, Jun. 23–27, 2002, vol. 3, pp. 1495–1500. [11] H. Dehbonei, C. V. Nayar, and L. Chang, “A new modular hybrid power system,” presented at IEEE Int. Symp. Industrial Electronics (ISIE’03), Rio de Janeiro, Brazil, Jun. 9–11, 2003, vol. 2, pp. 985–990. [12] M. N. Marwali and A. Keyhani, “Control of distributed generation systems-Part I: Voltages and currents control,” IEEE Trans. Power Electron., vol. 19, no. 6, pp. 1541–1550, Nov. 2004. [13] G. Ledwich and A. Ghosh, “A flexible DSTATCOM operating in voltage or current control mode,” IEE Proc. Gen., Transm. Distrib., vol. 149, no. 2, pp. 215–224, Mar. 2002. [14] Z. Chen and E. Spooner, “Voltage source inverters for high-power, variable-voltage DC power sources,” Inst. Elect. Eng—Proc. Gen., Transm. Distrib., vol. 148, no. 5, pp. 439–447, Sep. 2001. [15] C. V. Nayar, M. Ashari, and W. W. L. Keerthipala, “A single phase uninterruptible power supply system using a bi-directional sinusoidal PWM inverter,” presented at the 1998 Int. Conf. Power Electronic Drives and Energy Systems for Industrial Growth (PESDES’98), Dec. 1–3, 1998, vol. 2, pp. 671–676.

792

IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 21, NO. 3, SEPTEMBER 2006

[16] L. J. Borle, M. S. Dymond, and C. V. Nayar, “Development and testing of a 20-KW grid interactive photovoltaic power conditioning system in Western Australia,” IEEE Trans. Ind. Appl., vol. 33, no. 2, pp. 502–508, Mar.–Apr. 1997. [17] Monitoring Electric Power Quality, IEEE Standard 1159, 1995. IEEE Standards Board. [18] The Application and Testing of Uninterruptible Power Supplies for Power Generating Stations, ANSI/IEEE Standard 944, 1985. [19] M. Ashari, W. W. L. Keerthipala, and C. V. Nayar, “A single phase parallel connected uninterruptible power supply/demand side management system,” IEEE Trans. Energy Convers., vol. 15, no. 1, pp. 97–102, Mar. 2000. [20] P. C. Loh, M. J. Newman, D. N. Zmood, and D. G. Holmes, “Improved transient and steady state voltage regulation for single and three phase uninterruptible power supplies,” presented at the IEEE 32nd Annu. Power Electronics Specialists Conf. (PESC’01), Jun. 17–21, vol. 2, pp. 498–503. [21] D. N. Zmood and D. G. Holmes, “Improved voltage regulation for currentsource inverters,” IEEE Trans. Ind. Appl., vol. 37, no. 4, pp. 1028–1036, Jul.–Aug. 2001. [22] H. Dehbonei, C. V. Nayar, and L. Borle, “A multifunctional power processing unit for an off-grid PV diesel hybrid power system,” presented at the IEEE 35th Ann. Power Electronics Specialists Conf. (PESC’04), Jun. 21–25, vol. 3, pp. 1969–1975. [23] R. L. Vasquez-Arnez and L. C. Zanetta, Jr., “Compensation strategy of autotransformers and parallel lines performance, assisted by the UPFC,” IEEE Trans. Power Del., vol. 20, no. 2, pt. 2, pp. 1550–1557, Apr. 2005. [24] H.-L. Jou, J.-C. Wu, C. Tsai, K.-D. Wu, and M.-S. Huang, “Novel lineinteractive uninterruptible power supply,” δ/b Proc. IEE Elect. Power Appl., vol. 151, no. 3, pp. 359–364, May 2004. [25] P. Salmeron and J. R. Vazquez, “Practical design of a three-phase active power-line conditioner controlled by artificial neural networks,” IEEE Trans. Power Del., vol. 20, no. 2, pt. 1, pp. 1037–1044, Apr. 2005. [26] H. C. Lin, “Intelligent neural-network-based adaptive power-line conditioner for real-time harmonics filtering,” IEE Proc. Gen., Transm. Distrib., vol. 151, no. 5, pp. 561–567, Sep. 13, 2004.

Sung-Hun Ko received the B.Sc. and M.Sc. degrees from the Department of Control and Instrumentation Engineering, Kunsan National University, Kunsan, Korea in 1998 and 2000, respectively, and he is currently pursuing the Ph.D. degree in the School of Electronics and Information Engineering, Kunsan National University, Kunsan, Korea. From 2000 to 2001, he was with Research Laboratory, Seo-Young Electronics, Inc., Korea. Currently, he is working as a Visiting Research Fellow with the Department of Electrical and Computer Engineering at Curtin University of Technology, Perth, Australia. His current research interests include renewable energy based distributed generation system, power factor correction, inverter control, and neural network.

Seong R. Lee received the B.Sc. and M.Sc. degrees in electrical engineering from Myong-Ji University, Seoul, Korea in 1980 and 1982, respectively, and the Ph.D. degree from Chonbuk National University, Jeonju, Korea, in 1988. From 1997 to 1998, he was the Visiting Professor with the Department of Electrical and Computer Engineering at Virginia Tech, VA. From 2002 to 2004, he was the Director of Engineering Research Institute at Kunsan National University, Kunsan, Korea. Since 1990, he is a Professor with the School of Electronics and Information Engineering at Kunsan National University Currently, he is working as a Visiting Professor with the Department of Electrical and Computer Engineering at Curtin University of Technology, Perth, Australia. His current research interests include soft-switching inverter, power factor correction, switch mode power supply, and renewable energy based distributed generation systems.

Hooman Dehbonei (S’01–M’03) received the B.Sc. and M.Sc. degrees in electrical engineering from the Iran University of Science and Technology, Tehran, Iran in 1992 and 1997, respectively, and Ph.D. degree from Curtin University of Technology, Perth, Australia, in 2003. Presently, he is an Australian Research Council Postdoctoral Fellow with the Department of Electrical and Computer Engineering at Curtin University of Technology. He is a Chartered Professional Engineer and National Professional Engineers Register with the Institute of Engineers, Australia. His current research interests include power systems (design, analysis, quality, and control), power electronics (its application in power systems and renewable energy), renewable energy, and hybrid/distributed generation systems.

Chemmangot V. Nayar (M’86–SM’90) received the B.Tech. degree in electrical engineering from the University of Kerala, India, in 1969, the M.Tech. degree in electronics from the Indian Institute of Technology, Kanpur, in 1976, and the Ph.D. degree in electrical engineering, specializing in wind electrical power generation, from the University of Western Australia, Perth, Australia, in 1985. He holds a Personal Chair in electrical engineering at Curtin University of Technology. Prof. Nayar is a Chartered Engineer and Corporate Member of IEE, and a Chartered Professional Engineer and Fellow of IEAust.