VOLTAGE DIPS are generally considered a power-quality

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 19, NO. 2, APRIL 2004 783 Assessment of Voltage Dips in HV-Networks: Deduction of Complex Voltages From the Measured RMS Voltages Math H. J. Bollen, Senior Member, IEEE, Philippe Goossens, and Alain Robert Abstract Many network operators are doing important efforts to collect relevant and precise information about voltage dips in HV networks. This information is however not relevant for the end user, who is connected at a lower voltage level, often behind a Dy transformer and who will not experience the same voltage dips as recorded by the power quality monitors. In most cases the power quality monitors only records rms voltages during dips and the deduction of the characteristics at lower voltage levels is not possible. The voltage phasors are required to derive this kind of information. In this paper a new method is proposed which enables to obtain valuable statistical information for the network users from existing monitors and databases, which only contain information about rms voltages. Algorithms are proposed to estimate the dip type from the relation between the three rms voltages during the dip. Knowing the dip type the phasors can be calculated. Good results are obtained for faults in transmission networks, especially on overhead lines. The more severe the voltage dip, the better the performance of the algorithm. Index Terms Electromagnetic compatibility (EMC), power quality, voltage dips (sags), power quality monitoring. I. INTRODUCTION VOLTAGE DIPS are generally considered a power-quality problem of equal importance as long and short interruptions in the supply [1], [2]. In a liberalized electricity market, network operators should be able to give precise and relevant information on supply reliability and voltage quality to the network users. Recently many power quality monitors were installed in distribution and transmission networks to obtain statistical information on voltage dips. This statistical information is, in most cases, not relevant for the network user. The equipment of the network user is connected at a different voltage level, often behind a Dy transformer, and as a consequence it does not experience the same dip as recorded by a power-quality monitor. To obtain valuable information for the end user, the voltage dips should be translated to the equipment terminals at the lower voltage level. Therefore, the recording of the complex voltage is mandatory. However, at this moment, almost all installed monitors only record the rms voltages during the dip and most power Manuscript received November 25, 2002. M. H. J. Bollen was with the Department of Electric Power Engineering, Chalmers University of Technology, Gothenburg, Sweden. He is now with the STRI AB, Ludvika, Sweden (e-mail: m.bollen@ieee.org). P. Goossens and A. Robert are with the Elia, Brussels, Belgium (e-mail: philippe.goossens@elia.be; alain.robert@elia.be). Digital Object Identifier 10.1109/TPWRD.2003.823202 quality databases only contain information about the rms voltages and the dip duration at the monitor terminals. In this paper, a method is proposed to extract the phasors information from the three rms voltages only. The method can be used to translate voltage dip statistics of the network into valuable information for the end user or to compare voltage dip measurements realized at different voltage levels or with different monitor connections (star or delta). With star-connected monitors, the presence or absence of star-point grounding also affects the results. In this paper, we will assume a grounded star-point, even though the classification in Section II also holds for nongrounded star. The solution method presented in this paper is based on the characterization method proposed in [5]. After a short discussion of a classification of three-phase unbalanced dips into seven types, the proposed algorithm is described in detail. The algorithm is next applied to a number of measured dips. II. CLASSIFICATION OF VOLTAGE DIPS To collect statistics on voltage dips it is important to be able to describe voltage dips through a small number of parameters. Short and long interruptions are described by their duration, voltage dips are typically characterized by a duration and a retained voltage. This is being formalized in a number of standard documents, including IEC 61 000-4-30 and IEEE 1564 [4]. Such a two-dimensional event characterization is very acceptable for a single-phase supply but for a three-phase supply it only covers three-phase faults. The commonly used method, being formalized by IEC 61 000-4-30, is to take the longest duration and the lowest retained voltage to characterize the event. The somewhat strange consequence of this is, e.g., that the dip due to a single-phase fault in a high-impedance-grounded system will be quantified the same (or even more severe) as the dip due to a three-phase fault. However a customer behind a Dy-transformer only experiences a severe event in the second case. Some utilities solve this by connecting their monitors phase-to-phase, but that will limit the amount of information to be obtained from the monitors. To overcome this and similar problems an improved dip characterization has been proposed in [5], which will be summarized below. This characterization is based on the way in which the dips due to different fault types propagate through the system. A generalized magnitude of the event is defined, referred to as characteristic voltage. In [6], a method is proposed to extract the type of dip, the characteristic voltage, and other characteristics of three-phase 0885-8977/04$20.00 2004 IEEE

784 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 19, NO. 2, APRIL 2004 unbalanced voltage dips. This characterization can be used to present dip statistics, for testing of sensitive equipment, but also for extracting additional information about dips [7]. For the methods as described in [6] and [7], the complete waveform information of the voltage is needed, whereas most monitors only store rms voltages to save memory. Even with this limited amount of information it remains possible to extract information about the underlying event [8] but information about the type of dip (and thus the type of fault) could not be extracted from rms voltages only. The method for classification of three-phase unbalanced voltage dips as proposed in [2] and [6] is summarized in Table I. For details of the origin and propagation of the different types, the reader is referred to these two publications. A recapitulation is given in Table II. The expressions for the complex phase voltages in Table I are given under the assumption that positive-sequence, negative-sequence, and zero-sequence source impedance are equal. The method for distinguishing between the different dip types based on rms voltages is grounded on this assumption. The consequences for the method when this assumption is not being met are discussed further down in this paper. All voltages in Table I are complex values. The characteristic voltage can be expanded as TABLE I COMPLEX VOLTAGES AND PHASOR DIAGRAMS FOR THE DIFFERENT TYPES OF THREE-PHASE UNBALANCED DIPS where is the (characteristic) magnitude and the (characteristic) phase-angle jump. Type B and type E contain a zero-sequence component and can thus only be observed for measurement devices connected in star. The other types can also be observed in delta configuration. For the same fault, we will obtain a different dip type depending on the measurement connections and the voltage level of the measurement. The transfer of voltage dips across transformers depends on the winding connections. Star-delta en Delta-star transformers swap line and phase voltages [5]. (1) III. PROPOSED ALGORITHM The proposed algorithm consists of two steps. First, the type of dip is obtained from the relation between the three rms voltages. The main information used in the first step is that the number of possible faults, and thus the number of possible dips is limited. The dip type that best fits with the three measured rms voltages is chosen. A number of approximations are made for this. In the second step, the three rms voltages and the known dip type are used to calculate the characteristic magnitude and phase-angle jump. From the characteristic magnitude and phase-angle jump the complex phase voltages are calculated. Expressions like in Table I can be used for this. A. Estimating the Dip Type When the characteristic phase angle jump is neglected, a relation between the large and small voltage drop can be determined for the different types. The characteristic phase angle jump is the result of the difference in X/R ratio of source and feeder impedance [9]. In HV-transmission networks this phase angle jump is in most cases small enough to be neglected in the first step of the algorithm. The seven dip types according to Table I

BOLLEN et al.: ASSESSMENT OF VOLTAGE DIPS IN HV-NETWORKS 785 TABLE II FAULT TYPE AND DIP LOCATION THREE-PHASE UNBALANCED DIPS are grouped based on the number of phases with the most severe voltage drop. Type A: the three voltages drop the same amount. These events will be referred to below as three-phase drops. Type C, E and G: two retained voltages are much smaller than the third one, these voltage dips are referred to as two-phase drops. Type B, D and F: one voltage drops much more than the two other voltages (single-phase drops). Note that a single-phase drop does not imply that only one phase experiences a drop in voltage. For measured voltage dips, the three rms voltages will often be different, due to the characteristic phase angle jump and other phenomena. For the distinction between single-phase and two-phase drops the three retained voltages are sorted in ascending order after which the events are split into two groups., the highest retained voltages are close to each other. This group contains single-phase and three-phase drops., the lowest retained voltages are close to each other. This group contains two-phase and three-phase voltage dips. Three-phase drops fall in one of the two categories based on small random differences between the retained voltages. They will be separated from the other types in the next step. 1) Single-Phase Drops: Voltage dips with a single-phase drop are characterized by the fact that one phase shows a larger voltage drop than the other two. This group includes types A, B, D, and F. To distinguish between the dip types the relation between the lowest voltage and the highest voltage is used. Considering a zero phase-angle jump, relations between highest and lowest rms voltages can be obtained from the expressions in Table I. This relation is, for type A and for type B for type D and for type F (2) (3) (4) (5) Fig. 1. Relation between small and large rms voltage for single-phase and three-phase drops. Expressions (2) (5) are shown as thick dashed curves in Fig. 1. The thin solid lines give the demarcation between the different types of dips as used in the classification algorithm. Three-phase drops (Type A) are separated from single-phase drops only at this stage of the algorithm. Typically the two largest rms voltages are not the same. The characteristic phase angle jump will lead to a difference between the two highest rms voltages. Therefore, the arithmetic average is taken Consider, as an example, a dip with the following measured rms voltages:. Ranking the rms voltages in ascending order gives:. Using the arithmetic average, we get which corresponds to the indicated cross in Fig. 1. The type of dip is in this case estimated as D. 2) Two-Phase Drops: The same method as for single-phase drops can be applied to two-phase drops. This includes voltage dips of type A, C, E, and G. With the lowest rms voltage, and the highest rms voltage, considering zero characteristic phase-angle jump given for type A for type C for type E and for type G (6) (8) (9) (10) (11) The value for can again be calculated from the arithmetic average of the two lowest rms voltages. These curves are plotted

786 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 19, NO. 2, APRIL 2004 Fig. 2. Relation between small and high rms-voltage for two-phase and three-phase drops. as thick dashed lines in Fig. 2; the demarcation between the dip types is indicated by thin solid lines. From the figure it follows that it is not possible to distinguish between type C and type E from rms voltages only when the smallest voltage exceeds 0.5 p.u. As two-phase-to-ground faults (leading to type E) are less common than single-phase and phase-to-phase faults, it is safest to classify these events as type C. Consider as a second example an event with the following rms voltages:. After ranking the rms voltages, this reads as., and ; they are close so that this falls into the twophase drops. Using the arithmetic average gives after which the dip will be classified as type C. This event is indicated through a cross in Fig. 2. B. Limitations to the Classification Method 1) Phase-Angle Jumps: The presence of phase-angle jumps will make it harder to distinguish between single-phase drops and two-phase drops. A nonzero phase-angle jump affects the two highest or lowest phases in a different way. The result is that all three rms voltages become different. For minor phase-angle jumps, there will still be two rms voltages close to each other so that the method still works. In certain circumstances, e.g., for cable faults, the phase-angle jump may become too large to distinguish between single-phase and two-phase drops. Similar problems were observed when testing an algorithm for characterization of three-phase unbalanced voltages dips using six rms voltages (three phase-to-ground and three phase-to-phase). For shallow dips with relatively large phase-angle jumps, the algorithms gave incorrect results [10]. The impact of the phase-angle jump is shown by the example in Fig. 3. At the left, the phasor plot is given for the phase voltages as measured in a 70-kV network; at the right, the phasor diagram is shown for the line voltages. In a star connection, the type is clearly B (probably due to a single-phase fault at the same voltage level). In delta connection (or behind a Yd or Dy transformer), the event must be type C. Nevertheless, the proposed algorithm will classify the event as a dip type D. The phase-angle Fig. 3. Type C dip (right) which is incorrectly classified as Type D. The type B dip from which this dip originates is shown on the left. jump is too large and the magnitude of the drop is too small to correctly distinguish the type. The effect of nonzero fault impedance will in most cases be an additional phase-angle shift. For more complex faults (two-phase to ground, three-phase), nonsymmetrical fault impedances may blur the distinction between the different types. 2) Zero-Sequence Impedance: It has also been assumed in the derivation of the classification algorithm that positive-sequence, negative-sequence, and zero-sequence impedance are the same. Typically, the zero-sequence impedance in HV-networks is larger than the positive-sequence impedance. This will cause a rise in the rms voltage of the nonfaulted phases for dip type B (single-phase fault). This will make the distinction with other types even easier. For two-phase-to-ground faults (type E), the effect of the zero-sequence impedance is similar. When the zero-sequence impedance is larger than the positive-sequence impedance (as is normally the case), the voltage in the nonfaulted phase will rise. This will make it possible to distinguish between type E and C. Type C does not contain a zero-sequence component, so when the voltage does not rise in the nonfaulted phase, the dip should be classified as type C because type E is more rare than type C (the example in Section III-A-2 will be classified as type C). For faults close to the secondary side of a distribution transformer, the zero-sequence impedance may be lower than the positive-sequence impedance. This will lead to a drop in the rms voltage in the nonfaulted phases, which makes it harder to distinguish between types B and D. 3) Load Effect: The effect of the load on the voltage dip is merely that the rms voltages show a slow decay. This may make it harder to distinguish between types D and F, and between types C and G. However, at the start of the event, the load effect is small and the above-given expressions are rather accurate. The algorithm could be further extended to estimate the dip type

BOLLEN et al.: ASSESSMENT OF VOLTAGE DIPS IN HV-NETWORKS 787 from the rms voltages immediately after fault initiation and next track the load effects during the remainder of the event. C. Estimating the Characteristics Knowing the dip type, the characteristics of the event can be estimated. Knowing all the characteristics the three complex voltages can be calculated. Three-phase unbalanced voltage dips can be characterized by the characteristic voltage and the PN-factor (the definition of this rather complex quantity [6] is not recalled here). As both are complex numbers, this gives four degrees of freedom where we have only three known parameters (the three rms voltages). It is thus not possible to estimate all complex parameters. Here, we will only estimate the characteristic voltage. For type B, one of the most frequent types in transmission networks, we will take account of the ratio between the zero- and positive-sequence impedance. 1) Type A: For type A, the extraction of the characteristic magnitude is straightforward: the characteristic magnitude equals the mean rms voltage of the three phases. It is not possible to obtain any information on the characteristic phase-angle jump (12) 2) Type B: Type B dips, due to single-phase faults at the same voltage level, may be a large part of the total number of events for star-connected monitors. To accurately estimate the characteristic voltage, it is essential to consider the zero-sequence voltage component. The effect of a zero-sequence impedance that differs from the positive and negative-sequence impedance, is that the voltages in the nonfaulted phases are also affected by the single-phase fault. The more general expression for the Type B dip becomes (instead of the expression in Table I). A general solution is again not possible as there are four degrees of freedom with only three measurements. If we assume that both and are directed along the positive real axis (no phase-angle jump and equal X/R ratio for zero and positive-sequence), the two parameters are estimated from the rms voltage ranked in ascending order, as follows: (15) where is the arithmetic average of the two highest rms voltages. 3) Type C: Type C dips show a two-phase drop with the two lowest voltages and close to each other and the highest rms voltage close to 1 p.u.. The following expressions can be obtained for the lowest rms voltages [2]: (16) where is the characteristic voltage. From these two expressions, the characteristic magnitude and the characteristic phase angle jump can be calculated (17) 4) Type D: For a dip of type D, the two highest rms voltages are close. The lowest rms voltage gives the characteristic magnitude (18) (13) The characteristic phase angle jump is obtained from the difference between the two highest rms voltages where is the effect of the fault on the nonfaulted phases. Note that no longer is the characteristic voltage as defined before. The effect of the fault on the nonfaulted phases is related to the voltage in the faulted phase through the following expression: (14) resulting in (by taking the difference) (19) (20) and being the zero-sequence and positive-sequence source impedance (at the point-of-common coupling between the fault and the measurement location). Knowing the fault location it is possible to determine (14) from the system parameters. However, the fault location is not always known and the ratio is different for different voltage levels and (to a lesser degree) for different locations. The zero-sequence voltage in some cases is also damped by YYd-transformers between the fault and the measurement location. Therefore, two parameters are estimated from the rms voltages: and 5) Type E: Type E dips are due to two-phase-to-ground faults at the same voltage level as the measurement location. They will be only rarely detected since two-phase-to-ground faults are rare and as the majority of them will be (incorrectly) classified as type C (see Fig. 2). It is not possible to obtain any information on the characteristic phase-angle jump. The characteristic magnitude is the average of the two lowest rms voltages (the characteristic phase angle jump is neglected) (21)

788 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 19, NO. 2, APRIL 2004 Fig. 4. Evaluation of proposed algorithm to deduce the complex voltages from the measured rms voltages. 6) Type F: Type F is related to Type D but with a larger drop in the highest rms voltages. The characteristic magnitude is obtained from the lowest rms-voltage, the characteristic phase angle jump from the difference between the highest rms voltages (22) 7) Type G: For dip type G, the characteristic magnitude can be estimated with (23) The two lowest rms voltages can be used to estimate the characteristic phase angle jump (24) For dip types F and G, the characteristics are affected by the ratio between zero-sequence and positive-sequence source impedance at the fault location. For deriving (22) (24), this ratio has been considered equal to one. For transmission lines, the ratio is typically in the range of 2 4. The magnitude estimation for type F dips is independent of this ratio, but all other expressions are no longer exactly valid. Knowing the exact ratio revised expressions can be derived but the applicability of those remains dubious. More tests of the current version of the algorithm are needed to assess the accuracy of the algorithms for two-phase-to-ground faults. IV. CONDITIONS FOR POWER QUALITY RECORDERS The algorithm can only be applied when all three rms voltages are known. It is preferable to connect the monitoring equipment in star, as it is not possible to derive the phase voltages from the line voltages with the algorithm (the zero-sequence voltage is missing). As shown above, the opposite (from phase voltages to line voltages) is in many cases possible. Some power quality monitors do not store the voltages on all phases, each individual phase voltage must exceed 110% or drop under the 90% to be recorded. If, during a single phase drop, Fig. 5. Comparison between estimated and actual complex voltages for a voltage dip of type B. only one phase drops under the 90%, the voltages on the other phases will not be stored and it will not be possible to apply the proposed algorithm to distinguish between the types. Voltage dips are often characterized by the duration and the lowest retained voltage on each phase. It is preferable not to use the voltages, obtained with this characterization, to estimate the type: the instant in which the minimum is reached can be different for the three phases, and errors may occur when estimating the dip type. For a more accurate estimation, it is preferable to record the development of the rms voltages during the event and apply the characterization algorithms to the rms values obtained at the beginning of the dip. The load effect on the voltages may lead to incorrect estimations after a few cycles. V. CASE STUDY FOR MEASURED VOLTAGE DIPS It is recommended to adapt the software of power quality monitors to take account of these considerations. The processing as proposed in this paper can be performed on-line by the monitor or as post-processing applied to earlier-recorded events.some voltage dips measured in the Belgian transmission grid were used to evaluate the proposed algorithm. The waveform was stored during the dip and in post-processing the development of the rms values and the phasors was calculated. The proposed algorithm was applied to measured rms-values to make an estimation of the phasors, the voltage dip type and characteristics. The results are compared with the phasors as obtained directly from the measured waveform, as shown in Fig. 4. Examples for types B, C, and D are given in Figs. 5, 6, and 7. In general, the approach was proven to work correctly and the errors remained small. The example in Fig. 8 shows the phasors for a voltage dip type E with phase angles that make it hard to estimate complex phasors from the rms voltages only. While the estimation of the dip type with the proposed algorithm worked quite well, it was impossible to estimate this phase-angle jump. Phase angle jumps cannot be obtained for voltage dips of type E. The proposed approach can be used to obtain the dip magnitude between phases from phase-to-ground rms measurements or to estimate the dip magnitude behind a Yd or Dy transformer.

BOLLEN et al.: ASSESSMENT OF VOLTAGE DIPS IN HV-NETWORKS 789 TABLE III ESTIMATION OF DIP MAGNITUDE BETWEEN PHASES OR BEHIND DY OR YD TRANSFORMER FROM THE THREE RMS VOLTAGES ONLY Fig. 6. Comparison between estimated and actual complex voltages for a voltage dip of type C. Fig. 7. Comparison between estimated and actual complex voltages for a voltage dip of type D. that a higher accuracy will require a significantly higher amount of data storage. The algorithm made only one error in the type estimation. A type C was estimated as a type D dip. The explanation is the reduced magnitude of the dip in combination with the significant phase shift. In fact, type C and D become very similar in this circumstance: the wrong estimation leads to an error of only 0.04 p.u. For a type E dip, an important error of 10% was made. As mentioned before, the performance of the algorithm for two-phase-to-ground faults is questionable. Fig. 8. Comparison between estimated and actual complex voltages for a voltage dip of type E. The examples in Table III demonstrate that the error in the estimations remains, in most cases, less than 0.04 p.u. This will be acceptable for most applications, especially when considering VI. CONCLUSION Most power quality monitors only store the rms voltages during a dip. The approach described in this paper makes it possible to extract the phasors from the three rms voltages only and to obtain information concerning the dip type and the characteristic voltage. Good results are obtained for small phase angle jumps, this is generally the case for faults in transmission networks, especially on overhead lines. The more severe the voltage dip, the better the performance of the algorithm. However, the phase angle jump for type A, B, and E cannot be obtained and the accuracy of the algorithm for type F and G (two-phase-to-ground faults) is affected by the ratio between zero-sequence and positive-sequence source impedance at the fault location. This is not really a problem; types E, F, and G are uncommon in HV-networks and the type recognition works well. The approach can be used for statistical purposes, e.g., to obtain dip statistics for the phase-to-phase voltages from phase-to-

790 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 19, NO. 2, APRIL 2004 ground measurements. The algorithm can also be applied to compare the measurement information obtained from monitors installed at different voltage levels or with different connections (star and delta) and to obtain information of the dip as experienced by equipment behind a Dy-transformer. The monitoring equipment should always store the rms voltages on the three phases, even if one or two voltages during the dip remain above 90%. The very small voltage drops on the nonfaulted phases contain precious information about the dip type and are required to estimate the type correctly. A connection in star is generally preferable (measurement of phase-to-ground voltages). Further work on this algorithm includes a more thorough testing than was possible within the scope of this paper. A large database with detailed event recordings may be used to statistically test the proposed algorithm. Synthetic events and simulations in different types of networks (with different types of load) may be used to further study and quantify the limitations of the algorithm. Important issues to study include the suitability of the method for two-phase-to-ground faults and its accuracy for voltage dips due to faults in the MV distribution networks (i.e., dips with a large characteristic phase-angle jump). [7] E. Styvaktakis, M. H. J. Bollen, and Y. H. Gu, Expert system for classification and analysis of power system events, IEEE Trans. Power Delivery, vol. 17, pp. 423 428, Apr. 2002. [8], Automating classification of power system events using rms voltage measurements, in Proc. IEEE Power Eng. Soc. Summer Meeting, Chicago, IL, 2002. [9] M. H. J. Bollen, P. Wang, and N. Jenkins, Analysis and consequences of the phase jump associated with a voltage sags, in Proc. Power Syst. Comput. Conf., Dresden, Germany, Aug. 1996. [10] M. H. J. Bollen, Algorithms for characterizing measured three-phase unbalanced voltage dips, IEEE Trans. Power Delivery, vol. 18, pp. 937 944, July 2003. Math H. J. Bollen (M 94 SM 96) received the M.Sc. and Ph.D. degrees from Eindhoven University of Technology, Eindhoven, The Netherlands, in 1985 and 1989, respectively. He is a currently responsible for the product area EMC and Power Quality at STRI AB, Ludvika, Sweden. Before joining STRI in 2003, he was a post-doc at Eindhoven University of Technology, a Lecturer at the University of Manchester Institute of Science and Technology, Manchester, U.K., and a Professor in the Department of Electric Power Engineering, Chalmers University of Technology, Gothenburg, Sweden. His contribution to research includes the development of methods for voltage dip analysis, which resulted in a text book on power quality. Dr. Bollen is active in IEEE and CIGRE working groups on voltage dip analysis and statistics. REFERENCES [1] M. F. McGranaghan, D. R. Mueller, and M. J. Samotyj, Voltage sags in industrial power systems (check title!), IEEE Trans. Ind. Applicat., vol. 29, pp. 397 403, Mar./Apr. 1993. [2] M. H. J. Bollen, Understanding Power Quality Problems Voltage Sags and Interruptions. New York: IEEE Press, 1999. [3] J. Arrillaga, M. H. J. Bollen, and N. R. Watson, Power quality following deregulation, Proc. IEEE, vol. 88, pp. 246 261, Feb. 2000. [4] Voltage Sag Indices Draft 2, Working Document for IEEE P1564, Dec. 2001. [5] M. H. J. Bollen, Characterization of voltage sags experienced by threephase adjustable-speed drives, IEEE Trans. Power Delivery, vol. 12, pp. 1666 1671, Oct. 1997. [6] L. D. Zhang and M. H. J. Bollen, Characteristic of voltage dips (sags) in power systems, IEEE Trans. Power Delivery, vol. 15, pp. 827 832, Apr. 2000. Philippe Goossens received the B.Sc. degree at the K.I.H. De Dayer, Mechelen, Belgium, in 1995. He is presently responsible for connection management in the Quality Monitoring Department at Elia, Brussels, Belgium, the Belgian transmission system operator. He was previously Project Engineer in the Power Quality product line at Laborelec. Alain Robert received the M.Sc. degree from the Montefiore Institute at the University of Liège (ULg), Liege, Belgium, and the Ph.D. degree from the Catholic University of Louvain (UCL), Louvain-La-Neuve, Belgium. He is responsible for the Quality Monitoring Department at Elia and is an Invited Professor at UCL. He His past employment experience includes ULg, Laborelec, Electrabel, and CPTE. He is Chairman of international study committees CIGRE 36 and CIRED 2, dealing with power quality and EMC.