2012 IEEE 7th International Power Electronics and Motion Control Conference - ECCE Asia June 2-5, 2012,

Harbin, China

An Analysis of Shunt Active Power Filter Based on Instantaneous Reactive Power Theory
Zhang Juhai
Chongqing Design and Research Institute of China Coal Technology Engineering Group Chongqing, China 467095085@qq.com
AbstractShunt Active Power Filter is an ideal device that could manage the harmonic compensation in an adjustable and dynamic manner. This article reviews the basic principle of shunt active power filter, along with the current tracking circuit based on the instantaneous reactive power theory and the main circuit performing as an inverter with PWM hysteresis control. A comprehensive control method is analyzed and a harmonic compensation simulation is conducted, the result of which verifies the harmonic detection algorithm is well-proposed and the power quality of the grid is overall-enhanced. Keywords-harmonics; reactive power; current tracking; active power filter

II.

THE PRINCIPLE OF SHUNT ACTIVE POWER FILTER

A. Basic Structure Fig 1 demonstrates the simplified configuration of shunt APF[2], which could be dichotomized into reference current calculation section and its counterpartcompensating current generator (consisting of current control, drive and main circuit). The core function of the reference current calculation circuit is to detect the harmonics as well as the reactive power contributed by the non-linear load.

I.

INTRODUCTION

Harmonic suppression and reactive power compensation are essential for the construction of smart grid, which gathers, distributes, and acts on the behavior of suppliers and consumers in order to improve the efficiency and sustainability of electric services. With the increasingly widespread application of power electronic device, such as electronic appliances, switching model power supplies and electronic ballasts, the harmonic contamination they produce in grid has given rise to a variety of safety concerns. In a comparative study, the traditionally proposed LC passive filter could only absorb harmonic waves of a specific frequency, whereas the active power filter (APF) is capable of handling harmonics ranging between fluctuating frequencies and simultaneously maintaining a good compensative performance [1]. The performance of APF is primarily compatible to the interaction of current tracking control circuit and main circuit, the latter of which acts as an inverter consisting of various power electronic devices that produce the targeted compensating current under the PWM control determined by the reference circuit. The most widely accepted control mode is to initially detect and calculate the harmonics in the load current, then the negative value of which represents as the reference current that triggers the control of inverter by PWM current tracking method. Consequentially, a high-pass filter (HPF) is required to eliminate the high-order harmonic generation (a byproduct of PWM hysteresis control) with frequencies close to the carrier wave.

ic*

Figure 1. simplified shunt APF configuration

B. The Calculation of Reference Current The circuit of reference current calculation is essentially based on three-phase instantaneous reactive power theory[3], which suggests a novel approach to detect the three-phase harmonic current. As shown in Fig 2, ia, ib and ic are initially detected and then served as the input for the next stage of algorithm, which involves specific calculation of instantaneous active current ip as well as reactive current iq:

ia i p i = C ib q ic
Where :

(1)

3000

978-1-4577-2088-8/11/$26.00 2012 IEEE

 sin t 2 C= 3 cos t

2 2 sin t sin t + 3 3 (2) 2 2 cos t cos t + 3 3

In addition, a three-phase PLL section is prerequisite in order to grasp the angle of phase A. When ip and iq flow through the low-pass filter (LPF), the harmonics contained could be eliminated, and thus give rise to the DC component of the load current known as ip and iq . With inverse transformation of the transpose matrix of CT, hence the threephase fundamental current of iaf, ibf and icf could be readily obtained. Therefore, the reference current ia*, ib*and ic* is obviously the subtraction of those fundamental three-phase currents with the correspondingly original load currents.
ip iq
i *ca i*cb i*cc

acting on S4, which may not be necessarily switched on, for the current flows through diode D4 is not reversed in direction but rather begin decreasing. When the hysteresis control is applied, the real output of the current from inverters could maintain a fluctuation value only within h and h, bouncing up and down in zigzag forms, as one of the examples illustrated in Fig.4.
2h

Reference Current

Compensation Current t

PWM
Figure 2. Algorithm of the reference current calculation circuit

t
Figure 4. One example of the hysteresis control of the reference and compensating current

C. Current Tracking Control Circuit Current tracking is the first step for the production of targeted compensating current, under the basic function of obtaining the PWM signals that are responsible for the switch modes of each device in the main circuit. In addition, the PWM signal is generated by the comparison between the reference current and the real compensating current.

D. Main Circuit Parameter Design The working mode of the main circuit is determined by the condition of its six switches. Generally speaking, there is always a single device switched on in all three groups (S1/S4, S3/S6 or S2/S5), constructing six different combinations alltogether. The main circuit is shown in Fig.5.

* ia

ia

Figure 3. Circuit of current tracking control

The current-tracking-controlled PWM inverters possess various forms, the most common of which is the hysteresis current tracking control shown in Fig 3. ia* represents the reference value and meanwhile the tracking target of the load current. A hysteresis error h is intentionally established in case of the high frequency of switches in inverters. While ia*-ia h, the hysteresis controller produces high level which drives the upper bridge arm S1 in working mode condition, and thus increasing load current ia even when ia is outstripping ia*. The increase of load current will not be affected unless ia is exactly greater by h comparing with the reference current, then follows the reverse working mode of hysteresis controller by switching off S1 and simultaneously

Figure 5. Main circuit

Suppose the sum of three-phase voltage source ea+eb+ec=0 along with the current ia+ib+ic=0, then the differential equation of phase A could be described as: di L a = ea + K aU c (3) dt in which Ka denotes the switching coefficient, while ea represents as the instantaneous value of the ac side voltage and

3001

ia as the compensating current. In addition, when the upper bridge arm of phase A is switched on, the absolute value of Ka is -2/3; similarly, Ka is 1/3 when the lower bridge arm is working. Provided that the sample working time is long enough, the average effect of AC voltage ea in (2) would be zero. As the possibility of Ka=1/3 is 66.7% while that of Ka=2/3 is 33.3%, the mean of K a is hence 4/9.

Within one period of time, (2) could be reshaped as:

L
*

i * c max 4 = Uc 9 tc

(4)

I/A

where i cmax denotes the maximum value of the command signal of compensating current; the most appropriate value of coefficient varies from 0.3 to 0.4 in terms of the overall effect of compensation, which could be obtained by simulation. Taking function (3) and safety concerns into consideration, the prerequisite of Uc 3Em should be granted, in which Em is the peak value of phase voltage. Assuming that the DC side capacitor is always in transition from charging to discharging states within a single switching cycle, the maximum acceptable deviation of DC side capacitor voltage is [4]: (5) Ct c * In the simulation designed below, if the value of |i cmax| is 40A, and U c max is 2V, then C=0.12F could be calculated. III. SIMULATION RESULTS

Figure 6. Load current of phase A

U c max =

* ic max

Figure 7. FFT analysis of load current

Based on the configuration of shunt APF, a simulation system is established by Matlab/Simulink, with the simulation time set to 0.2s. Phase A is mainly selected as a sample for analyzing since the three-phase system is symmetric [5]. In addition, Table. 1 provides the basic parameters for simulation.
TABLE I. PARAMETERS OF ACTIVE POWER FILTER

According to Fig.7, the proportion of both 5th and 7th harmonic wave is much higher than the rest of the counterparts, accounting for nearly 24% and 13% respectively. The Total Harmonic Distortion (THD) is up to 39.25%, suggesting a severe contamination of load current resulted from the non-linear loads in the circuit.
B. Load Current after Compensation

Line voltage & frequency Line inductor Resistor Active power filter parameters Switching Frequency

Us = 380V, f=50Hz L = 6mH R=5 L c = 4mH, C dc=12mF f = 11kHz

The graph of load current after compensation is demonstrated in Fig.8, which is more compatible to the general shape of sine wave compared to that of Fig.6.

A. Analysis of Load Current

Fig. 6 shows the load current of phase A, and Fig.7 demonstrates the corresponding FFT analysis. Observing Fig.6, the wave is not sine-shaped, indicating a large proportion of harmonics of various orders have distorted the original smooth sine wave of the source current.

Figure 8. Load current of phase A

When the simulation time is rearranged to 1s, the THD of the compensated load current reduces to 3.13% rather than the original 39.25%, stating an effective compensation of harmonic waves resulting from the non-linear load.

3002

[4]

[5]

Hongda. Li, Cao Kang, Active Power Filter Simulation Based on Instantaneous Reactive Power Theory and the PWM Hysteresis Control Mode, ICEMI2011, Chengdu: IEEE, 2011, pp.95-100. J. Yang, Harmonic & Reactive Current Detecting Methods and a Study of Shunt Active Power Filters, Ph.D. dissertation, Xian Jiaotong Univ., Xian, 1996.

s,
Figure 9. FFT analysis of load current

C. Analysis of Harmonic Current

Fig. 10 displays the harmonic current detected in simulation, the mixture of which to the sine wave adversely affects the power quality in the grid.

Figure 10. Harmonic current of phase A

IV.

CONCLUSION

On the basis of the simulation, the article utilizes the instantaneous reactive power theory (ip-iq calculating approach) to detect harmonic wave and reactive load current efficiently, compensate the load current effectively by reshaping the source current to that similar to sine wave and reducing the THD value to below 5%. The compensation performance of the simulation result provides a practical approach to the basis of APF application in smart grid system. REFERENCES
[1] [2] H. Akagi, New Trends in Active Filters, Proceedings of EPEs 95, Sevilla, 1995. pp. 17-26. Jun Yang and Zhaoan Wang, Harmonic Suppression and Reactive Power Compensation, 3rd ed. Beijing: China Machine Press, 2005, pp. 257-258. H. Akagi, Y. Kanazawa and A. Nabae, Generalized theory of the instantaneous reactive power in three-phase circuits, IEEE & JEEE. Proceedings IPEC, Tokyo: IEEE, 1983, pp.1375-1386.

[3]

3003