Performance Improvement of Parallel Active Power Filters Using Droop Control Method

Ghazal Falahi

School of Electrical engineering

Sharif University of technology

Tehran, Iran 11365-9363

Email: Falahi@ee.sharif.edu

Hossein Mokhtari , Member IEEE

School of Electrical engineering

Sharif University of technology

Tehran, Iran 11365-9363

Email: Mokhtari@sharif.edu

Abstract-In this paper, a new method based on droop control scheme is proposed for controlling parallel operation of active filters. The harmonic components of the load current are extracted by an enhanced phase-locked loop (EPLL). In the parallel group, each filter operates as a conductance and the harmonic workload is shared among them. A droop relationship between the conductance and non-fundamental apparent power controls the operation of each unit. The non-fundamental apparent power has been calculated based on IEEE Std 1459. Principles of operation are explained in this paper and simulation results which are presented approve the effectiveness of this method. The results indicate a significant reduction in Total Harmonic Distortion (THD) in a rectifier application.

Keywords-Power quality; Parallel active filters; Power system harmonics; Droop

. INTRODUCTION

The increasing use of power electronic devices has resulted in harmonic pollution in power networks. The main problem stem from the flow of non-active energy caused by harmonic currents and voltages. Flowing non-sinusoidal current into the network has the drawback of deteriorating the harmonic pollution and degrading power quality [1].

Mitigation equipments such as Active Power Filters (APFs) are designed and used to improve power quality. APFs are one of the emerging solutions to surpass power system harmonics and enhance power quality. This is mainly due to technological progress in the power switching devices, Digital Signal Processors (DSPs) and new control algorithms [2].

In many cases, it is favorable to connect APFs in parallel instead of using filters with increased capacities. This results in higher reliability and minimizes the installed converters rating and cost. One of the most effective and reliable methods for controlling the operation of parallel inverters is droop method. The droop method is usually used to achieve good sharing among units when communication between the inverters is difficult due to their physical location or when a more reliable and flexible system is required. This method

avoids any control wire interconnection among different modules and that’s why it is often named as wireless or independent parallel control [3]-[5].

This paper proposes a control method for parallel operation of Active Filter Units (AFUs) based on droop strategy. In this case, each filter operates as a conductance and several units can share the non-fundamental load power without any interconnection. The droop coefficient of each unit is determined by its capacity and the harmonic filtering workload is shared among AFUs in proportion to their capacity. With increasing or decreasing of nonlinear loads in the system the droop control method will help AFUs to dynamically adjust their non-fundamental filtering capacity to acquire satisfactory compensation. The non-fundamental power is calculated based on IEEE Std 1459. Also a new approach for harmonic extraction based on EPLL has been introduced. Fig. 1 shows the power circuit of parallel connected AFUS and Fig .2 shows the proposed load sharing and control method.

As compared to existing harmonic extracting methods, EPLL-based method provides higher degree of immunity and insensitivity to noise, harmonic and other types of pollutions. Simple Structure of EPLL-based method simplifies its implementation in digital software and/or hardware environments as an integral part of digital control platform for power electronic converters.

Fig. 1. Power circuit of parallel connected AFUs

978-1-4244-2487-0/09/$25.00 ©2009 IEEE

Fig. 2 Load sharing and control method Fig. 3. EPLL structure

The dominant feature of the proposed method over conventional methods is the frequency adaptivity which permits desired operation when the center frequency of the base signal varies. This system is also capable of coping with the unbalanced system conditions [6].

II. Operation principals of parallel APF

This section explains the principals of operation of the proposed parallel APF system.

A. Harmonic detection method

An estimation of fundamental component is obtained by means of an adaptive nonlinear notch filter, i.e. EPLL. The overall structure of the EPLL is in accordance with a conventional PLL. The basic structure has three independent internal parameters K, KPKv and KiKv. Parameter K dominantly controls the speed of convergence of amplitude and parameters KPKv and KiKv control the rate of convergence of phase and frequency respectively. As compared with the conventional PLL, the EPLL method generates a more accurate angle and estimation of fundamental component in a polluted environment. An implementation of the EPLL is shown in Fig. 3. The input signal is compared with its extracted smooth version to generate an error signal which is used by low-pass filter (LF) to generate a driving signal for VCO. The EPLL is actually a band-pass filter and the continuous time differential equations governing it’s dynamic are derived from the block diagram of Fig. 3 as [6]:

(1)

(2)

Where y(t) is the fundamental component and e(t) is the error signal. Fig. 4 compares the phase angle extracted by an EPLL and a conventional PLL. Also an estimation of the fundamental component of input signal is presented.

B. Current controller

After detecting the harmonic component of phase current, a current regulator is used for accurate tracking of the reference current by the AFUs and the voltage commands are calculated as follows:

(3)

Where the Lx is the output inductor of the AFUx and T is the sampling period. The voltage commands are used as the reference voltages for a Pulse Width Modulator (PWM) and the gating signals are generated to provide an effective tracking of current commands [8].

EPLL(Rad) Ia1(A) PLL(Rad)

Fig. 4. Phase angle extracted by an EPLL and conventional PLL(Rad/sec) and an estimation of fundamental component

C. Control algorithm for parallel operation

A proper method is needed to contrononlinear loads among AFUs. The conventiparallel power converters requires intercoconverters to achieve balanced load sharinconventional methods uses a voltage controand several “slave” units. However a confimaster/slave strategy is not redundant due master unit. To achieve true redundancy, aable to operate independently [9].

To achieve true redundancy, each AFU canthat it behaves like a harmonic conductanoutput current of each AFU is related to tnode to which the AFU is installed, i.e.:

IAFx=Gx.EAFx

Where the IAFx is the output current of Aharmonic conductance and EAFx is the phase v

The proposed control method is a load-that will share the non-fundamental apparAFUs. Droop control method has been exuninterruptible power supply (UPS) systvarious units to share loads without anyTherefore the reliability of the system is enhoperation of AFUs, the droop control metThis control technique can be defined asbetween the conductance and the non-fundpower. Therefore, the harmonic workloadamong the AFUs. For the proposed powequations can be given as:

G1=G0+d1(SN1-SN10)

G2=G0+d2(SN2-SN20)

:

Gx=G0+dx(SNx-SNx0)

In the above equations, Gx is the conduG0 is the rated conductance, di is the slope ofis the non-fundamental power of AFUx annon-fundamental apparent power. The drooshown in Fig. 5. The base of compensation 1459 and the Non-fundamental apparent poaccording to:

Where S1x is fundamental apparent power othe apparent power of AFUx. Vabcx, Iabcx are and currents of the AFUx and Vabcx1, Iabcx1 arphase voltages and currents of AFUx respect

ol the sharing of ional approach for nnection between ng. One of these

oller as a “master” guration based on to dependency of

all units should be

n be operated such nce. Therefore the the voltage of the

(4)

AFUx , Gx is the voltages of AFUx.

-sharing technique ent power among

xtensively used in tems and allows y communication. hanced. In parallel thod can be used. s the relationship damental apparent d can be shared

wer system, droop

)

) (5)

uctance command, f the equation, SNx

nd SN0 is the rated op characteristic is

is IEEE Standard ower is calculated

(6)

of AFUx and Sx is the phase voltages re the fundamental tively.

Fig. 5. Droop

III. Simula

Simulations have been environment to investigate the in different conditions. The parin Table 1. Since the nonlinearand 7th harmonics are dominanAPFs are installed in parallel line. The THD of the load cusimulation results. The THD othe operation of APFs, is approand its fundamental componencompensated source current iinverter reference and output cin Fig. 8 (a) and (b), it is shocondition the filters currents acoefficient. The simulation reAPFs can effectively suppreunbalanced and distorted volta10 show the three phase systeunbalanced conditions and respectively and Fig. 11 showinstantly in response to the load

The simulation results verify leads to a better harmonic extraimproved compensation is accompensated source currencompensation signal generatioEPLL harmonic extraction metraditional synchronous refeunbalanced and distorted condi

Table1. Simula

Source voltage 220v (line-Transmission line parameters

R=0.05, L=

Active filter 2 AFUs, L=G0=0 , d1=8

PWM A sine/trian

Nonlinear load A diode rec

Fig. 6 Load current and it’s fundame10m

p characteristic

ation results

carried out in the PSCAD proposed droop control method

rameters of simulation are given r load is a 6-pulse bridge, the 5th

nt harmonics in the load current. on the same position along the urrent is 27% as shown in the of the source current, thanks to oximately 3%. The load current nt are shown in Fig. 6 and the is shown in Fig. 7. Also, the current of two AFUs are shown own that under different droop are proportional to their droop esults indicate that the parallel ss the harmonics even under age conditions. Fig. 9 and Fig.

em voltages under distorted and increase of nonlinear load

ws that the source current rises d increase.

the fact that the EPLL method action, and as a consequence, an chieved. Fig. 12 compares the nt THD for two different on methods. It is clear that the ethod is more accurate than the erence frame method under itions.

ation parameters

-line), 60Hz =4mH

=6mH, SN10= 800VA, SN20=800VA, 8x10-4, d2=4x10-4

ngle PWM, fpwm=10 KHZ

ctifier (6-pulse bridge), RL load.

ental component y-axis:2A/div, x-axis: ms/div

Fig. 7 System current after compensation y-axis:2.5A/div, x-axis: 10ms/div

(a)

(b)

Fig. 8 (a) Reference and output current of AFU1 , y-axis:0.5A/div, x-axis: 5ms/div (b) Reference and output current of AFU2 , y-axis:0.5A/div, x-axis:

5ms/div

Fig. 9 system phase voltages y-axis:100v/div, x-axis: 10ms/div

Fig. 10 Load current and it’s fundamental component in response to load increase y-axis:2.5A/div, x-axis: 20ms/div

VI. Conclusion

In this paper, a new system is proposed for better operation of paralleled APFs based on droop method and an EPLL. A droop relationship between conductance command and non-fundamental apparent power (G-SN) controls the sharing of nonlinear workload among various AFUs. This definition is based on IEEE1459 Std. A nonlinear load, i.e. a 6-pulse bridge, is considered to verify the performance of the proposed technique. Computer simulation shows the effectiveness of the proposed control technique for harmonic suppression. The droop characteristic adjusts the filters

Fig. 11 Transition of system current in response to load increase y-axis: 5A/div, x-axis: 20ms/div

(a)

(b)

Fig. 12. (a) THD in abc/dq detection method y-axis:1percent/div, x-axis: 50ms/div (b) THD in EPLL detection method y-axis:1percent/div, x-axis:

50ms/div

capacity based on non-fundamental power while keeping the source THD within desired limits.

References

[1] IEEE Trial-Use Standard Definitions for the Measurement of Electric power Quantities Under Sinusiodal , Non-Sinusiodal, Balanced or Unbalanced Conditions, IEEE Std 1459.

[2] LIU Jinjun, WANG Xiaoyu, YUAn Chang ,WANG Zhaoan, “On the Control of Active Power Filters, IEEE,The 7th International Conference on Power Electronics,Oct. 2007.

[3] J.M.Guerrero, L.Garcia de Vicuna, J.Matas, J.Miret, and M.Castilla, “A Wireless Load Sharing Controller to Improve Dynamic Performance of Parallel-connected UPS Inverters,” in Proc. IEEE-PESC’03 Conf. , pp.1408-1413.

[4] S.J.Chiang, W.J.Ai, ”Parallel Operation of Three Phase Four Wire Active Power Filters Without Control Interconnection,” IEEE PESC 2002 Vol. 3, pp. 1202-1207.

[5] Josep M. Guerrero, Nestro Berbel, Luis Garcia de Vicuna, Jose Matas, “Droop Control Method for the Parallel Operation of Online Uninterruptable Power Systems Using Resistive Output Impedance,”IEEE APEC 2006, pp.1716-1722

[6] .M.Karimi-Ghartemani, M.R.Irvani, “A Method for Synchronization of Power Electronic Converters in Polluted and Variable-Frequency Environments,” IEEE Trans. Power Syst., vol. 19, pp.1263-1270, Aug.2004.

[7] Lusian Asiminoaei, Lascu Cristian, Frede Blaabjerg, “Harmonic Mitigation Improvement With a New Parallel Topology For Shunt Active Power Filters,”IEEE PESC 2006,pp.1-7.

[8] Po-Tai Cheng, Tzung-Lin Lee ,”Distributed active filter systems (DAFs):A new Approach to Power System Harmonics,” IEEE Transaction on Industry Application, Vol. 42, No. 5, pp1301-1309 Sept./Oct. 2006,.

[9] Josep M. Guerrero, Luis Garcia de Vicuna,”A Wireless Controller to Enhance Dynamic Performance of Parallel Inverters in Distributed Generation System,”IEEE Transaction on Power Electronics , Vol.19, No.5, pp. 1205-1213, September 2004.