Você está na página 1de 7

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 22, NO.

2, APRIL 2007

1049

Characterization and Compensation of Harmonics and Reactive Power of Residential and Commercial Loads
Jos Antenor Pomilio, Senior Member, IEEE, and Sigmar Maurer Deckmann
AbstractThis paper uses data obtained from laboratory measurements of typical home appliances to verify whether these nonlinear loads behave similar to current or voltage-type harmonic sources. Based on the measurements and additional simulations, it is stated that domestic and commercial electronic loads behave essentially like harmonic voltage sources. This behavior helps to explain why reactive shunt compensators and lters may even increase the harmonic current content in the feeders. Additionally, based on eld measurements in a residential low-voltage distribution secondary, the share between linear and nonlinear loads, as well as the prevailing kind of their nonlinearity is analyzed. The measurements, made just before and after the installation of a capacitor bank, conrm that the use of shunt compensation increases both voltage and current distortion in the feeder. Index TermsDiode, harmonic distortion, harmonics, nonlinear circuits, passive lters, power distribution.

I. INTRODUCTION

CCORDING to vast literature on the subject, the traditional approach for harmonics compensation in industrial applications assumes that the nonlinear loads can be modeled as pure harmonic current sources. Depending on the nonlinear load characteristics, the usual representation of nonlinear loads as a simple combination of harmonic current sources may greatly simplify the analyses of their effects on the overall system under similar conditions where distortion has been derived. However, additional conclusions may not be reliable if any modication is introduced in the circuit, such as the connection of a lter or a change in the loading condition. This certainly limits the usefulness of linear models to study nonlinear processes. A more general model should include not only the harmonic current or voltage sources but also their corresponding admittances or impedances. The main problem of this approach is that the equivalent admittance varies with the circuit operating condition, due to the nonlinear nature of the real load. Obviously, according to the NortonThevenin equivalence, it is possible to represent the harmonic source either as current or voltage

Manuscript received June 27, 2005; revised July 28, 2006. This work was supported by Elektro Energy and Services. Paper no.TPWRD-00373-2005. The authors are with the School of Electrical and Computer Engineering, State University of Campinas, Campinas 13081-970, Brazil (e-mail: antenor@dsce.fee.unicamp.br; sigmar@dsce.fee.unicamp.br). Color versions of one or more of the gures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identier 10.1109/TPWRD.2007.893179

sources. Nevertheless, this equivalence is not valid for pure current or voltage sources. Another important aspect is that some nonlinear loads should be modeled by harmonic voltage sources rather than by harmonic current sources. Among such loads are diode rectiers with capacitive output (dc) lter, which constitute the usual interface between electronic loads and the ac feeder [1], [2]. Typical applications of this case are TV sets, computers, electronic ballasts for uorescent lamps, and even the input rectier stage of inverters, if line reactors are not used [3]. The classication as harmonic voltage or current source depends on what is being imposed by the nonlinear load. In the case of motor-type loads, the motor inductances limit the current changing rate, which confers the behavior of harmonic current source to this kind of loads. In the case of rectiers with dc output capacitors and without signicant series impedance (such as electromagnetic-interference (EMI) lters), it is the dc capacitor that limits the voltage changing rate and confers the behavior of harmonic voltage source to these types of loads. This paper uses data from laboratory measurements and simulations of typical residential loads (refrigerator, TV set, and uorescent compact lamps) in order to verify whether they behave as current or voltage-source-type loads. To decide which type of source model best ts each case, a harmonic shunt lter is added to the circuit and the load response is then analyzed. For pure current source behavior, it is expected that the load imposes xed harmonic currents and the shunt lter absorbs part of this current according to the admittance, relative to the main feeder. For a pure voltage source type, it is expected that a shunt lter will increase the harmonic current emission of the nonlinear load, since the lter represents a new low-impedance path for the imposed harmonic voltages. It should be stressed that in the simulation studies of this paper, neither harmonic current nor voltage source models were used, but the electronic circuit models representing the respective loads were. The knowledge of the characteristic of nonlinear loads is important in order to understand the behavior of these kinds of loads under different operating conditions, and is especially useful to select the most effective compensation method to be used. The share between linear and nonlinear loads is evaluated considering eld measurements on a residential low-voltage distribution feeder. Finally, the prevailing behavior of the combined nonlinear loads is evaluated and tested by installing a capacitive bank at the transformer secondary.

0885-8977/$25.00 2007 IEEE

1050

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 22, NO. 2, APRIL 2007

Fig. 1. Refrigerators voltage, current, and power waveforms.

Fig. 3. AC source voltage, refrigerator current, and power absorbed after the lter connection.

TABLE I REFRIGERATOR CURRENT (RMS VALUE AT 127 V)

Fig. 4. Rectier with capacitive output lter.

Fig. 2. Tuned lter for current source load.

The lter capacitance (20 F) was selected to fully compensate the displacement factor (fundamental power factor), thus resulting in a lter inductance of 15.6 mH. Fig. 3 shows the waveforms after the lter installation. Table I shows that the load current remains almost constant when the lter is connected. The small increment of the fundamental current is due to the voltage increase produced by the reactive compensation (127 to 128.8 V). According to Fig. 2, the relationship between the ac supply current and the load current may be estimated by

II. NONLINEAR LOAD TYPES: CURRENT SOURCE OR VOLTAGE SOURCE (1) A. Current Source Load Type Typical home appliances, which impose current distortion, are the electromagnetic devices, such as motors and transformers. Refrigerators, freezers, washing machines, and air-conditioning devices can also be included in this group. In this case, the current distortion depends on the motors design and varies with the voltage level. In the rst laboratory test, measurements have been made with a low-consumption refrigerator (target 26.6 kWh/month). Fig. 1 shows the measured current, voltage, and instantaneous power. The voltage source is free from distortion. The apparent power resulted in 170 VA, while the power factor (PF) is 0.64 (131var and 108.5 W). The refrigerator current harmonics are shown in Table I. If this load effectively behaves as a harmonic current source, then a shunt harmonic lter, connected as shown in Fig. 2, should be capable of absorbing the loads harmonic currents, according to the relative admittances of the lter (1/Zf) and the feeder (1/Zi). To verify this, a shunt lter, tuned 5% below the 5th harmonic (285 Hz), was designed [4] and connected at the refrigerator (load) input terminals. With the lter connected, the 5th current harmonic at the main supply was reduced 10.8 dB (3.4 times). This result agrees with the relation obtained from impedance measurements. At 300 Hz, the measured lter impedance is 4.1 , the source impedance is 9.4 , and the expected 5th harmonic reduction is 3.3 times. Thus, one can conclude that the refrigerators nonlinearity behaves similar to a harmonic current source and, for this kind of load (refrigerator), the shunt lter was effective. B. Voltage-Source-Type Load Fig. 4 shows the circuit of a diode rectier with a capacitive output lter which, as will be shown, is an example of nonlinear load with harmonic voltage source behavior. As mentioned before, this kind of circuit is present in almost all residential and commercial nonlinear loads, such as computers, video monitors, TV sets, electronic lamp ballasts, etc. For the second test, a 20-in TV set was supplied by the same supply source used in the previous test. Fig. 5 displays the measured voltage and current waveforms. Such load usually absorbs constant power, and the current varies in order to compensate

POMILIO AND DECKMANN: CHARACTERIZATION AND COMPENSATION OF HARMONICS

1051

Fig. 5. AC source voltage, current, and power absorbed by a TV.

Fig. 7. Simulation results: with and without lter. Top traces: 5th harmonic lter and rectier currents (2 A/div). Bottom traces: Source voltage (100 V/div) and current (4 A/div).

Fig. 6. Load fed by rectier and harmonic voltage source equivalent, both with shunt lter.

the input voltage changes. The voltage distortion observed at %, while the current THDI reached the PCC is THDV 108%, and the resulting PF was only 0.66. However the fundamental power factor (displacement factor) resulted 0.97, capacitive. The capacitive behavior at the fundamental frequency of this kind of rectier is due to the capacitor loading process, which occurs just before the peak value of the mains voltage, so the fundamental component of the current results slightly in advance of the mains voltage. Since any LC harmonic lter also presents a capacitive behavior at the fundamental frequency, it is not possible to design a passive shunt tuned lter that simultaneously reduces the current distortion and improves the fundamental power factor. The compensation in this case is possible only with series lters or active solutions [3], which are out of the scope of this paper. In order to conrm this behavior, a shunt harmonic lter was considered at the input of a rectier circuit, as shown in Fig. 6. Equation (2) provides the ac supply harmonic current, including a shunt lter, related to a harmonic voltage load type [1]

(2) If the load behaves as a pure voltage harmonic source, the impedance is zero. This behavior will be different if considerable series inductance is connected to the rectier, and the equivalent model should include not only the harmonic source but also a series impedance. can be calculated Unfortunately the load series impedance only for a specic operating condition, because it varies with several parameters (voltage, power, dc capacitor, frequency, etc.). A modeling method has been proposed in [5], but it is

not valid for the discontinuous conduction mode, as is the case of the nonlinear loads considered in this paper. Even without in (2), it is evident that (1) also knowing the value of does not apply for loads fed through rectiers with an output is not supposed constant. capacitor, because Next, the harmonic voltage source behavior of this kind of nonlinear load will be veried by simulation and experimentally. Fig. 7 shows the waveforms obtained by simulation of a circuit feeding a linear inductive-resistive load (equivalent to the refrigerator at the fundamental frequency) and a rectier with capacitive-resistive load (equivalent to the TV set). A 5th harmonic lter, is initially present at the PCC, and is disconnected at the time assigned in Fig. 7. The shunt lter capacitance compensates the fundamental reactive power, so that the resultant displacement factor is close to unity, in spite of the highly distorted current owing into the mains feeder. Note that the rectier current increases while the lter is connected. This behavior can be explained if one considers that the nonlinear load behaves similar to a harmonic voltage source. If the load would have a current source behavior, its harmonic current emission would remain constant, even if the lter were connected. Fig. 8 shows a laboratory experimental result, obtained with the same kind of loads (refrigerator and TV) and 5th harmonic lter. The only difference is that the presented voltage is the real distorted network (PCC) voltage instead of the clean simulation voltage source. Considering that the TV power demand varies according to its image and sound adjustments, and in order to have a reproducible test, the following results were obtained for an equivalent rectier with capacitive lter and constant resistive load. According to Table II, all harmonic current of the nonlinear load increase when the lter is connected. A harmonic current source model for the load cannot explain such behavior. Table III indicates the corresponding voltage harmonics at the PCC. Note that, due to the 5th harmonic lter action, the 5th the 7th harmonics of the PCC voltage have been reduced, but the

1052

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 22, NO. 2, APRIL 2007

TABLE IV SUPPLY CURRENT COMPONENTS (RMS VALUE AT 127 V; LOAD: REFRIGERATOR RECTIFIER)

Fig. 8. PCC voltage and source current feeding a TV set; and a refrigerator, with the 5th harmonic shunt passive lter.

TABLE II RECTIFIER (NONLINEAR LOAD) CURRENT (RMS VALUE AT 127 V)

TABLE III MEASURED PCC VOLTAGE CHARACTERISTICS (LOAD: REFRIGERATOR RECTIFIER)

Fig. 9. Refrigerator TV load, with capacitive shunt compensation, and distorted supply voltage.

3rd and 9th harmonics increased, resulting a modest reduction of the total voltage distortion (THDV). The last column of Table III shows the effect of a simple capacitive compensation, intended to compensate the displacement factor of the load (refrigerator). In this case, the voltage distortion signicantly increases, especially due to the 7th and 9th harmonics growth. This can be explained by the parallel resonance between the shunt capacitor (20 F) and the feeder inductance (5 mH), occurring near 500 Hz. Thus, assuming that the nonlinear load behaves like a nonideal harmonic voltage source, it is necessary to consider also the resonance between the capacitor and the source impedance. , this amplicaFor a pure harmonic voltage source tion cannot be explained, but according to Fig. 6, the harmonic voltage component at the PCC can be estimated by

Table IV shows the ac mains current harmonics. Although the LC lter has reduced the 5th and 7th harmonic components in the mains feeder, it has increased the 3rd and 9th harmonic ows, thus increasing the overall supply current distortion. The fundamental current was reduced due to the reactive power compensation and therefore the THDI (%) resulted larger with the lter connected. It is important to mention that if the supply voltage already contains the harmonic component intended to be ltered, the use of shunt compensation can even produce the opposite effect in terms of waveform improvements. In this case, the resulting low-impedance path will drive current from the supply, increasing the current and voltage distortion on the feeder. This situation is illustrated in Fig. 9, showing the effect of the capacitor connected at the terminals of the load. If the supply voltage already contains harmonic distortion, the resonance amplies the correspondent harmonic current. In this case, the 7th and 9th harmonics are amplied. III. REACTIVE AND HARMONIC CHARACTERISTICS OF A RESIDENTIAL LOW-VOLTAGE DISTRIBUTION GRID The following data have been measured in a real distribution grid constituted by 70% of residential-type loads (141 consumers), 25% of commercial loads (8 consumers) and 5% of industrial load (1 consumer). This case represents a middle class

(3)

POMILIO AND DECKMANN: CHARACTERIZATION AND COMPENSATION OF HARMONICS

1053

Fig. 10 Reactive power measured at the transformer secondary along a week.

Fig. 12. Reactive power before and after capacitive compensation.

In the course of the day, until 9 or 10 P.M., this harmonic component increases, indicating that more nonlinear loads are turned on, such as TVs and compact lamps. Considering a TV set (56 W) and 3 uorescent compact lamps (3 15 W) per home, according to the laboratory measurements presented in the previous section, the total load would produce 700 mA of 3rd harmonic. This value is 7 times the harmonic current caused by a single refrigerator. Audience statistics indicate that at least 35% of the TV sets are disconnected, even during the evening. According to this, one may expect a maximum 3rd harmonic component of 15 A per phase, which is next to the value actually obtained by eld measurements.
Fig. 11. Measured 3rd harmonic current at transformer secondary.

IV. PASSIVE COMPENSATION IN LOW-VOLTAGE DISTRIBUTION GRIDS distribution area, which characterizes the kind of power consumption related to home appliances. The next gures show the measurements concerning one phase, but the other phases present similar behavior. Fig. 10 shows the behavior of the reactive power (only fundamental component of each phase) along a one-week measurekvar) occurs about 6 A.M. and ment. The minimum value ( can be attributed to the refrigerators presence, according to the next considerations. The apparent power demand of a typical two-door refrigerator is 300 VA. Considering the typical refrigerator duty cycle (1/3 on, 2/3 off) and the number of consumers, 5 kvar/phase is obtained, which is consistent with the eld measurements. Other reactive power sources are the public illuminations (40 lamps, ) contributing with about 1 kvar/phase. 200 W, According to Fig. 10, the reactive power quickly increases during the morning, probably due to the connection of other electric motor driven appliances, such as washing machines. In the afternoon, evening, and night, the reactive power gradually reduces. A similar analysis can be made for the current harmonics, as shown in Fig. 11 for the third-order component. The minimum value (2 A) per phase occurs early in the morning and is compatible with the measured component of the refrigerator, as shown in Fig. 2, and considering a duty-cycle of 1/3. Electric power distribution companies are interested in applying passive lters in low-voltage distribution grids to achieve an adequate voltage prole along the feeder, thus postponing higher cost modications. The choice of a passive lter could, in principle, improve the displacement factor, increase the voltage prole and minimize the harmonic circulation through the transformer, thus reducing its losses [6]. The lter or compensator could be installed near the transformer, and designed to compensate the lowest or base value of the reactive power demanded along a seven-days measurement (Fig. 10). However, as shown before, an increase of voltage and current distortions is expected since the typical nonlinear loads have a dominant behavior of a voltage source type. Therefore, the impact of passive compensation must be analyzed not only according to reactive power and voltage boosting objectives, but also taking into account the voltage and current distortions. Fig. 12 shows the reduction of the reactive power supplied by the mains after the installation of a 17.5-kvar three-phase (5.83-kvar/phase) capacitive bank near the transformer secondary terminals. As expected, after the capacitor connection, a small rootmean-square (rms) voltage increase is observed, as shown in Fig. 13.

1054

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 22, NO. 2, APRIL 2007

Fig. 13. Voltage at the transformer secondary before and after capacitive compensation.

Fig. 15. Current spectra at the transformer secondary, before (top) and after (bottom) capacitor bank installation.

Fig. 14. Secondary voltage THDV before and after capacitor bank installation.

Less reactive current in the feeder and a lower voltage drop for the customers are the expected benets of reactive compensation. On the other hand, voltage and current distortions are more accentuated after the capacitor connection, as shown in Figs. 14 and 15. The average voltage THDV increased from 2% to 2.5%, while the current THDI increased from 8.5% to 10% (not shown). This increase is more relevant for the 7th and 9th current harmonics, as depicted in Fig. 15. The frequency-response analysis of this case showed that there was a resonance between the capacitors and the supply system reactance in the range of the 7th and 9th harmonics. This explains the increase observed for those current harmonic at the transformer. V. CONCLUSION Residential and commercial supply networks exhibit dominant nonlinear loads that are better modeled as harmonic voltage sources. An example of a nonlinear circuit largely used in lowvoltage loads is the single-phase rectier with capacitive output lter. This type of converter does not determine the current; instead, it imposes the voltage at the PCC. Refrigerators and other motor-driven loads could well be represented by harmonic current sources but their impact on voltage and current distortions is much lower than that caused by the rectiers. The resulting current harmonics, related to nonlinear loads of the harmonic voltage source type, will not be satisfactorily

ltered out by shunt lters, since the load imposes the PCC harmonic voltage, and not the harmonic current. The use of capacitive compensation, to improve the fundamental power factor and the voltage prole along the low-voltage distribution grid, will increase the voltage and current harmonic contents in the circuit. Shunt compensation with passive LC lters for harmonic compensation is fairly effective because it can reduce the respective grid current harmonic (since the nonlinear load , is signicant). However, other equivalent impedance, harmonic components may increase. So if one were interested in reducing current harmonics in such feeders, the best alternative according to (2), would be to of the mains supply and the increase the series impedances . In either case, however, the fundaload series impedance mental (ac) and the dc supply voltages must be maintained under control in order to provide the necessary regulation at adequate levels. Thus, series lters or active compensators could be effective for improving the power factor and reducing the harmonics propagation to the mains. Since the use of series devices in the main feeder also presents installation difculties, active ltering seems to be a feasible solution. The other solution would be to have electronic loads with low-current distortions, as determined by IEC standard 61000-3-2 [7]. ACKNOWLEDGMENT The authors would like to acknowledge the R&D working group of the project on this subject, supported by Elektro Energy and Services and developed during 2004 by Unicamp and Lactec.

POMILIO AND DECKMANN: CHARACTERIZATION AND COMPENSATION OF HARMONICS

1055

REFERENCES
[1] F. Z. Peng, G.-J. Su, and G. Farquharson, A series LC lter for harmonic compensation of ac Drives., in CD-ROM IEEE Power Electronics Specialists Conf., Charleston, NC, Jun. 1999. [2] J. K. Phipps, J. P. Nelson, and P. K. Sen, Power quality and harmonic distortion on distribution systems, IEEE Trans. Ind. Appl., vol. 30, no. 2, pp. 476484, Mar./Apr. 1994. [3] T. Tanaka, Y. Nishida, and S. Funabike, A Method of Compensating Harmonic Currents Generated by Consumer Electronic Equipment Using the Correlation Function, IEEE Trans. Power Del., vol. 19, no. 1, pp. 266271, Jan. 2004. [4] S. M. Peeran and C. W. P. Cascadden, Application, design and specication of harmonic lters for variable frequency drives, IEEE Trans. Ind. Appl., vol. 31, no. 4, pp. 841847, Jul./Aug. 1995. [5] A. Emadi, Modeling of power electronic loads in ac distribution systems using the generalized state-space averaging method, IEEE Trans. Ind. Electron., vol. 51, no. 5, pp. 9921000, Oct. 2004. [6] J. A. Pomilio and S. M. Deckmann, Characterization and compensation of harmonics and reactive power of residential and commercial loads, in CD-ROM Brazilian Power Electronics Conf., Recife, Brazil, Jun. 2005. [7] Limits for Harmonic Currents Emission (Equipment input current up to and including 16 A per phase), in Int. Electrotech. Comm. 61000-321st ed. 1995.

Jos Antenor Pomilio (M92SM02) was born in Jundia, Brazil, in 1960. He received the B.S., M.S., and Ph.D. degrees in electrical engineering from the University of Campinas, Campinas, Brazil, in 1983, 1986, and 1991, respectively. From 1988 to 1991, he was head of the Power Electronics Group, Brazilian Synchrotron Laboratory. Currently, he is a Professor at the School of Electrical and Computer Engineering, University of Campinas, where he has taught since 1984. In 1993 and 2003, he was Visiting Professor at the University of Padova, Padova, Italy, and at the Third University of Rome, Rome, Italy, respectively. His main interests are switching-mode power supplies, power factor correction, electrical drives, and active power lters. Dr. Pomilio is currently an Associate Editor of the IEEE TRANSACTIONS ON POWER ELECTRONICS.

Sigmar Maurer Deckmann was born in Cruz Alta, Brazil, in 1946. He received the B.Sc., M.Sc., and Ph.D. degrees in electric engineering from the University of Campinas, Campinas, Brazil, in 1973, 1976, and 1980, respectively. Since 1974, he has been a faculty member with the School of Electric and Computer Engineering of the University of Campinas. He headed the Department of Systems and Energy Control from 1986 to 1988. He also conducted several projects about icker, harmonics analysis, and digital instrumentation. From 1999 to 2003, he was Vice Director of the School of Electric and Computer Engineering. The main research areas have been power system analysis, power system dynamics, power system instrumentation, icker and harmonic measurement, and power quality analysis.

Você também pode gostar