Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Detrimental Effects of Capacitors in Distribution Networks in The Presence of Harmonic Pollution

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

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

1, JANUARY 2007

311

Detrimental Effects of Capacitors in Distribution Networks in the Presence of Harmonic Pollution


Nicola Locci, Carlo Muscas, Member, IEEE, and Sara Sulis, Student Member, IEEE
AbstractThe main goal of this paper is to analyze the behavior of an electric power network in the presence of harmonic distortion, when capacitors are installed. The work starts from a case study, where failure and malfunctioning of an industrial plant are described with the help of experimental measurements. The theoretical and mathematical details of the phenomena involved are analyzed by suitable computer simulations. The study is performed by considering some important electrical quantities, evaluated according to the denitions proposed by the standard IEEE 1459. The results put in evidence that using some typical quantities, such as the power factor, could lead to ambiguous conclusions in evaluating the actual quality of the loads, and this can assume special importance in virtue of the economic relevance of such quantities. Index TermsCapacitors, harmonic distortion, power factor, power quality (PQ).

I. INTRODUCTION

APACITORS are widely used in distribution power networks to obtain reactive compensation of the inductive loads. The root-mean-square (rms) value of the current in the power line feeding the customers loads is reduced by employing capacitors, so that both the generation capability of the power plant and the losses in the distribution network are minimized. However, the behavior of such compensation systems is optimal only under sinusoidal conditions. These days, the actual operating conditions of power networks (in particular, distribution networks) may signicantly differ from the pure sinusoidal steady state, which is the reference condition for which the plants and the electrical devices are usually designed, realized, and applied. This paper focuses on some implications of the harmonic distortion existing in modern distribution networks, when capacitors are installed in some of the nodes, since there is growing evidence that, under nonsinusoidal conditions, the presence of capacitors is associated with the partial or overall malfunctioning of the plants (see, for instance, [1] and [2]). Furthermore, let us consider penalties for polluting loads and power-quality indexes needed to assign responsibility for disturbances in electric power systems. Doubts arise as to how the role of capacitors should be taken into account when such acknowledged indexes will be used to qualify the polluting behavior of loads. In particular, there is a possibility that some indexes may penalize polluting consumers with capacitors in their

plants [3][5], even if they are used for the reactive compensation of the fundamental component of the current. In order to prove the relevance of these topics, we will consider a case study consisting of an industrial plant with signicant harmonic pollution and the presence of capacitors. First, the drawbacks that arose during preliminary tests will be reported. It is shown that the nonsinusoidal conditions of the network seriously affect the performance of the portion of the plant where the capacitors are positioned, leading to intervention of the protection relays and, thus, inhibiting the normal plant operation. Then, the analysis of the network will be carried out by means of proper simulations in order to investigate different possible congurations by evaluating, for each test, several electrical parameters dened in the IEEE 1459-2000 trial-use standard [6]. In particular, a well known and commonly used parameter will be considered: the power factor, for which different denitions are proposed in [6], according to different measurement purposes and different network conditions. It is well known that the power factor has important economic implications since its value is related to the penalties applied by utilities to consumers. The results of this study show that, if the involved phenomena are not well understood and the proper parameters are not considered, it is possible to make signicant errors in the assignment of the responsibility for low-power factor. II. CASE STUDY A. Plant Layout The experimental investigation was performed on the network supplying the forge unit of a metallurgic plant for manufacturing steel pieces. Fig. 1 shows the general scheme of the plant. The industrial plant is supplied by two identical 1-MVA medium-voltage (MV)/low-voltage (LV) transformers (referred to as Tr1 and Tr2 in the following), having a common primary MV (15 kV) bus. In the studied situation, the low-voltage outputs of the transformers are independent of each other: the feeder leaving from Tr1 is dedicated to the forge division (by means of its distribution center, named DC1 in Fig. 1), while Tr2 supplies the other loads of the plant (distribution centers DC2 and DC3). The supply line from Tr1 to the distribution center of the forge mm unit is 140 m long and is made by an EPR cable ( for each phase plus the neutral conductor). The power supply to the forge is provided by a three-phase rectier bridge, realized by thyristors, followed by a dc/ac highfrequency converter (10 kHz) realized by means of a high-power insulated-gate bipolar transistor (IGBT) inverter. The nominal current of the device is 870 A, with a rated voltage of 380 V. This nonlinear load is named Nlin in Fig. 1.

Manuscript received April 1, 2005; revised February 13, 2006. Paper no. TPWRD-00185-2005. The authors are with the Department of Electrical and Electronic Engineering, University of Cagliari, Cagliari 9123, Italy (e-mail: sara.sulis@diee.unica.it). Digital Object Identier 10.1109/TPWRD.2006.877088

0885-8977/$20.00 2006 IEEE

312

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 22, NO. 1, JANUARY 2007

Fig. 1. Plant layout.

From the same distribution center, a different feeder supplies a linear load (Lin) representing the service circuits of the unit. These circuits mainly include lighting circuits, consisting of groups of 400-W metal halide lamps, equipped with capacitors for reactive power compensation. The protection for the lighting circuits is guaranteed by automatic circuit breakers (CBs). During the preliminary tests, performed before starting the normal operation of the industrial process, overload conditions always occurred in the lighting circuit, thus causing the automatic CBs to open in less than a minute after the forge was triggered, even if CBs were designed and installed according to standard requirements. Such problems have prevented the plant from becoming operational. B. Voltage and Current Measurement To examine the nature and origin of the reported malfunctioning, both the voltages and the currents in the crucial points of the plant have been measured. A digital phosphor oscilloscope Tektronix 3014 (sample rate 1.25 GSamples/s on each channel, 9 b for the vertical resolution) was used to acquire the waveforms. Voltage transduction is guaranteed by an active differential probe Tektronix P5200 with a bandwidth ( 3 dB) up to 25 MHz and accuracy 5%. Current transduction for currents up to 100 A is ensured by a Hall effect clamp-on probe Tektronix A622 with a frequency range up to 100 kHz and accu. The current feeding the distribution center racy DC1 (whose rated value is 870 A) was directly measured on the output of transformer Tr1, acquiring the voltage at the output terminals of the current transformer installed in the power center. The accuracy of the measurement system is quite low, owing to the uncertainty introduced by both the transducers and the vertical channels of the oscilloscope. According to Standard IEC 61000-4-30 [6], the instrument specications could be marked as class B performance. However, this accuracy class can be considered sufcient for the troubleshooting purpose of these measurements and for the technical analysis reported in Section II-C. As for the evaluation of the power quantities described in [7] and discussed in Section III, more accurate data are necessary. To calculate such quantities, the experimental tests have been substituted with suitable computer simulations that reproduce with good approximation the actual situation. In this way, the data processed are not corrupted by any measurement uncertainty. Fig. 2 shows the acquired waveform of the current supplying the forge distribution center and its spectral content.
Fig. 2. Current absorbed by the forge: (a) waveform and (b) frequency spectrum.

Fig. 3. Line-to-neutral voltage in the division distribution center.

The signicant distortion of the signal can be clearly observed. The relevant total harmonic distortion (THD) is about 23%. Fig. 3 shows the line-to-neutral voltage waveform measured on the terminals of the feeder supplying the forge in the distribution center. We can observe the signicant distortion arising from the highly distorted current absorbed by the nonlinear load. In particular, the sudden commutations occurring in the current cause voltage drops in the inductive impedance of the supplying network, thus leading to the noticeable voltage spikes in the waveform of Fig. 3. The THD of this waveform is 14%. In order to get a more exhaustive view over the voltage prole in

LOCCI et al.: DETRIMENTAL EFFECTS OF CAPACITORS IN DISTRIBUTION NETWORKS

313

Fig. 4. Line-to-neutral voltage on the low-voltage terminals of the transformer Tr1.

Fig. 6. Current absorbed by the three lamps lighting circuit.

Fig. 5. Line-to-neutral voltage on the low-voltage terminals of the transformer Tr2.

the overall plant, Fig. 4 shows the voltage waveform acquired directly on the low-voltage side of Tr1. The voltage distortion, caused by the harmonic current owing into the equivalent series impedance of both transformer Tr1 and supply network, is reduced with respect to the voltage at distribution center DC1, which was also affected by the voltage drop on the cable. This is conrmed by the fact that the THD of this waveform is 11.6%. Finally, Fig. 5 shows the voltage on low-voltage (LV) terminals of transformer Tr2. This waveform is even less disturbed than the one shown in Fig. 4, since it is not inuenced by the distorted voltage drop caused by the harmonic currents in Tr1. As a consequence, the voltage on the LV bars of Tr2 has a THD equal to 3%. C. Failure Report and Analysis In the distribution center DC1, as well as in the overall plant, the protection for the lighting circuits was designed following the usual rule of thumb. Malfunctioning and inefciency of the protection plant occurred during the tests as untimely operation. The experimental acquisitions, along with the following considerations, allowed us to explain the situation. Each 400-W metal halide lamp of the lighting circuits is compensated with a 40- F capacitor and the combination absorbs 1.7-A rms line current

with rated voltage 220 V. Each feeder supplies a group composed of six lamps directly from the division distribution center. The applied voltage is the one represented in Fig. 3. In order to perform the experimental survey without causing the untimely intervention of the protection, three of the six lamps supplied by a single feeder where purposely bypassed. Fig. 6 shows the current absorbed by this reduced load. The rms current is 15 A, while the fundamental harmonic has an amplitude of 5 A. There are current peaks up to 50 A, with . THD Therefore, the actual current in the feeder with the original load (six lamps) is about 30 A rms, with peaks up to 100 A. These values should be compared to the rms current that would be absorbed by the same load under sinusoidal conditions, . which is One can notice that the ratio between the rms value of the disand the nominal one torted current is around three. Let us consider a thermal-magnetic CB, following the normalized curve C (according to [9]) for the modular CBs, with . The same breaker in a polluted system rated current works with a ratio and, therefore, in this condition, trips in 1 min. These considerations correspond perfectly with that which was experimentally observed. III. EVALUATION OF IEEE 1459 QUANTITIES A. Simulations The aim of this section is to closely examine the situation illustrated in Section II by evaluating a few signicant parameters dened in the IEEE standard [7] in different realistic operating conditions. Therefore, different network congurations should be tested to compare the results. These tests were performed by simulating the distribution plant under examination by means of the PSCAD/EMTDC program, produced by Manitoba HVDC. As stated before, the use of simulations also allows possible problems arising from inadequate measurement accuracy to be avoided in the evaluation of the meaningful parameters. Nominal data from manufacturers, as well as experimental results, were used to obtain a realistic model of the actual system, so that a very good agreement exists between the acquired voltage and current waveforms and the simulated ones.

314

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 22, NO. 1, JANUARY 2007

In particular, each lamp was modeled with a model (where , , F). The nonlinear load was modeled by means and of an equivalent circuit consisting of a linear load shunted with harmonic current generators, whose values were extracted from the experimental waveforms. The IEEE Trial Use Standard 1459 provides denitions for the measurement of electric power quantities under sinusoidal, nonsinusoidal, balanced, or unbalanced conditions. Besides the mathematical expressions that were used in the past, this standard denes new expressions aimed at accounting for the important changes that have occurred in the electric distribution network during the last decades. In the introductive note of the document, there is the claim that the new denitions were developed to give guidance with respect to the quantities that should be measured or monitored for revenue purposes, engineering economic decisions, and determination of major harmonic polluters. The power factor denitions used here are as follows, with the variables as dened in the Appendix: ; 1) fundamental (50/60 Hz) power factor . 2) total power factor To achieve the desired quantities, suitable signal processing has been performed in the LabVIEW software package. One period of the steady-state waveforms has been extracted from the simulations to avoid spectral leakage problems in the frequency analysis performed by means of a discrete Fourier transform (DFT) [9]. B. Results and Discussion As shown in Fig. 1, the network has a linear load (Lin), namely the lighting circuits, and a strongly nonlinear load (NLin), the forge, connected to the bus. In this specic situation, both loads are managed by the same customer, but, for the purposes of this work, it is interesting to consider them as if they would be representative of different users supplied by the same bars, which could play the role of a point of common coupling (PCC). The measurement of both applied voltages and absorbed currents has been achieved at the departure of the relevant feeders for each load. The rst tests were conducted by considering only one load connected at a time. As for the linear lighting circuits, the two power factor denitions lead, as was expected, to the same numerical result. In particular, when capacitors are not present, the inductive nature of the load is clearly evidenced by a low power factor , while the reactive compensation achieved with the capacitors increases this value up to 0.954 for both indexes. As for the nonlinear load, it is characterized by the following and . values: These results should be compared to the ones obtained when both loads are supplied at the same time, which is the normal operating condition of the plant. When the forge is on, the noncompensated lighting circuits and . The similar values of feature

the two quantities can be explained by the fact that the inductive lamps smooth the harmonic currents and, therefore, the harmonic powers have little effect on the total power quantities. In any case, the value is so low that the use of capacitors for power factor correction seems to be mandatory. However, this correction is valid for quantities at the fundamental frequency, whereas it can introduce dramatic collateral effects, depending on the strong increase of the harmonic distortion on the network, as clearly shown in Section II-B. This is conrmed by the is increased again results: the fundamental power factor falls down up to 0.954, whereas the total power factor to 0.180. To complete the survey, it should be considered that on the forge side the effects of the reactive compensation on the lamps are signicantly less noticeable: when the lighting circuit is not and , compensated, the forge features while the introduction of the capacitors leads to and . The results achieved on this simple network emphasize the need for carefully stating the most suitable denitions to be implemented in measurement instruments designed to analyze the behavior of electric systems under nonsinusoidal conditions, especially when such results are used to establish penalties for loads that contribute to the power-quality (PQ) degradation. IV. CONCLUSION This work focuses on an engineering problem related to the effects of the capacitors in electric power networks in the presence of harmonic pollution. Starting from a real case in an actual low-voltage industrial plant, the study remarks that, owing to the pervasiveness in the use of capacitors in such systems, strong nonlinear loads could have serious implications on the operation of neighboring plants. In addition, by evaluating quantities purposely dened to be measured for either revenue purposes or determination of major harmonic polluters, it has been shown that customers with linear loads, besides suffering from malfunctioning of their plants, could also be penalized for the low-power factor, depending on the denition implemented in the utility measurement station. These considerations could be more complex when both linear and nonlinear consumers take responsibility for harmonic pollution. In the assumption of sinusoidal conditions, the authorities enforce penalties to avoid the low-power factor in the network. Under nonsinusoidal conditions, the same penalties should be enforced according to parameters solidly acknowledged, whose recognized reliability would make the possible economical effects acceptable from the customers point of view. APPENDIX Let us indicate the rms value of the line to neutral voltage , the as , the rms value of the line to line voltage as rms value of the line current as , and the rms value of the neutral current as , the fundamental positive-sequence voltage , the fundamental positive-sequence current component as component as , and the relevant phase angle as . On these base meanings, the standard IEEE 1459 provides the following denitions for power factors in three-phase nonsinusoidal and unbalanced four-wire systems.

LOCCI et al.: DETRIMENTAL EFFECTS OF CAPACITORS IN DISTRIBUTION NETWORKS

315

1) Fundamental positive-sequence power factor

where

is the fundamental positive-sequence apparent power

is the fundamental positive-sequence active power

[3] D. Castaldo, A. Ferrero, S. Salicone, and A. Testa, An index for assessing the responsibility for injecting periodic disturbances, in Proc. 6th Int. Workshop Power Denitions and Measurement Under Non-Sinusoidal Conditions, Milano, Italy, Oct. 2003. [4] E. J. Davis, A. E. Emanuel, and D. J. Pileggi, Evaluation of singlepoint measurements method for harmonic pollution cost allocation, IEEE Trans. Power Del., vol. 15, no. 1, pp. 1418, Jan. 2000. [5] N. Locci, C. Muscas, and S. Sulis, On the measurement of power quality indexes for harmonic distortion in the presence of capacitors, in Proc. IEEE IMTC, Ottawa, ON, Canada, May 1719, 2005, pp. 16001605. [6] Electromagnetic Compatibility (EMC)Part 4: Testing and Measurement TechniquesSection 30: Power Quality Measurement Methods, IEC Std. 61000-4-30, 2003. [7] Trial-Use Standard: Denitions for the Measurement of Electric Power Quantities Under Sinusoidal, Nonsinusoidal, Balanced or Unbalanced Conditions, IEEE Std. 1459-2000, Jan. 2000. [8] Electrical AccessoriesCircuit-Breakers for Overcurrent Protection for Household and Similar InstallationsPart 1: Circuit-Breakers for a.c. Operation, IEC 60898-1, 2003. [9] A. V. Oppenheim and R. W. Schafer, Digital Signal Processing. Englewood Cliffs, NJ: Prentice-Hall, 1975. Nicola Locci received the Laurea degree in mechanical engineering from the University of Cagliari, Cagliari, Italy, in 1974. Currently, he is Associate Professor of electrical measurements with the Department of Electrical and Electronic Engineering, University of Cagliari. He was Professor of communication systems and his research topics are coding, photovoltaic systems, losses measurement in power electronics, and high-frequency (HF) transformers, variable reluctance motor parameters measurement, and nonactive energy compensation. His research interests include measurement on power systems with distorted waveforms, transducers performance improvement, and accuracy evaluation in signal processing of data-acquisition systems.

is the fundamental positive-sequence reactive power. 2) Total power factor

where

is the total active power and

is the effective apparent power. The rms effective current and voltage are dened as

Carlo Muscas (M98) was born in Cagliari, Italy, in 1969. He received the M.S. degree in electrical engineering from the University of Cagliari, Cagliari, Italy, in 1994. He was Assistant Professor in the Electrical and Electronic Measurements Group with the University of Cagliari from 1996 to 2001. Currently, he is Associate Professor of electrical and electronic measurement. His research activity mainly focuses on the study of power-quality phenomena, including the denition of electrical quantities used to characterize the behavior of power systems under nonsinusoidal conditions, along with the metrological qualication of the relevant measurement processes. He is author or co-author of many scientic papers.

REFERENCES
[1] M. Mamdouh Abdel Aziz, E. El-Din Abou El-Zahab, A. M. Ibrahim, and A. F. Zobaa, Effect of connecting shunt capacitor on nonlinear load terminals, IEEE Trans. Power Del., vol. 18, no. 4, pp. 14501454, Oct. 2003. [2] R. H. Simpson, Misapplication of power capacitors in distribution systems with nonlinear loads-three case histories, IEEE Trans. Ind. Appl., vol. 41, no. 1, pp. 134143, Jan./Feb. 2005. Sara Sulis (S04) received the degree in electrical engineering from the University of Cagliari, Cagliari, Italy, in 2002 and is currently pursuing the Ph.D. degree in industrial engineering at the University of Cagliari. Her main research activity is in the eld of power-quality measurements, with particular attention to the denition and measurement of electrical quantities in power systems under nonsinusoidal conditions and to the metrological qualication of the measurement processes involved.

You might also like