IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 18, NO.

1, MARCH 2003

157

Testing the Islanding Protection Function of Photovoltaic Inverters

Achim Woyte, Ronnie Belmans, Senior Member, IEEE, and Johan Nijs, Senior Member, IEEE
AbstractA major safety issue in grid-connected photovoltaics is to avoid nonintentional operation in islanding mode when the grid is being tripped. Worst-case conditions under which islanding can occur are specified analytically. The circuit that is commonly used for testing is described. The issue of appropriate test conditions with regard to reactive-power injection to the grid is discussed and the stabilizing impact of rotating machines and resonant circuits is evaluated in detail. Islanding test results for small inverters are presented. They confirm that very simple islanding protection methods that are commonly used, are likely to fail, if inverters are loaded with considerable capacitance. The obtained results support the assessment of the islanding protection function. They emphasize important points when defining new certification procedures for upcoming guidelines and standards. Index TermsCertification, inverters, islanding, photovoltaic power systems, protection, reactive power, safety, standardization, testing.

Fig. 1. Test circuit for islanding protection, as proposed by Hberlin [2].

I. INTRODUCTION N A grid-connected photovoltaic (PV) system, the PV generator is connected to the public low-voltage grid via an inverter. As every distributed generation unit being connected to the public grid, the PV system has to comply with common safety standards. A major issue is to avoid nonintentional operation in islanding mode when the grid is being tripped at fault conditions or for maintenance purposes [1]. A self-commutated inverter can detect grid outage by means of frequency and voltage monitoring. As a consequence of the disturbed power balance in the grid section that becomes isolated from the main power supply, frequency and voltage obtain new values, mostly being situated beyond the allowed limits. However, if the local load matches the power delivered by the local PV generator, the power balance can be in equilibrium in the isolated grid section. Frequency and voltage remain constant and the inverter keeps on operating. The isolated grid section, still being energized by PV or other embedded generation, is then referred to as an island. In order to overcome such a potentially dangerous situation, most inverter manufacturers implement additional islanding prevention measures beside frequency and voltage monitoring. The proper functioning of such islanding protection algorithms

is usually checked by a laboratory test procedure according to local guidelines and standards. The purpose of this work is to assess different test procedures for the islanding protection function as they are applied in different countries. Parameters of typical control algorithms for PV inverters that can have an influence on their islanding behavior are identified. The impact of the applied test circuit on the efficiency of the islanding detection scheme is studied. Finally, laboratory tests are carried out with four small inverters in order to study their islanding behavior under typical test conditions. II. CONDITIONS FOR ISLANDING In principle, every self-commutated inverter can operate in islanding mode. In the laboratory, this can be shown by using the test circuit from Fig. 1. Resonant circuit and testing load together represent the disconnected grid section. If no particular control algorithm for islanding prevention is implemented, the load conditions under which islanding occurs, depend only on the inverters frequency and voltage limits. Assuming constant active- and reactive-power output before and after grid tripping, voltage and frequency during islanding can be determined from the power balance (1)

(2) In these equations, and indicate the inverter operating is the reactive power supplied by the capacitor of point. and are active- and reactive-power the resonant circuit. mismatch, describing the power that is supplied to the grid before tripping. When inductive reactive power is supplied to the is positive. grid, and can be adjusted by tuning the testing load. For a given capacitance and inverter power, a so-called nondetective -domain where zone (NDZ) can be determined in the

Manuscript received December 7, 2001. This research was supported in part by IMEC vzw., Leuven, and by the European Commission under Contract ERK5 CT199900014. The authors are with the Department of Electrical Engineering, Katholieke Universiteit Leuven, B-3001 Leuven, Belgium (e-mail: [email protected]). J. Nijs is also with the IMEC vzw., B-3001 Leuven, Belgium. Digital Object Identifier 10.1109/TEC.2002.808410

0885-8969/03$17.00 2003 IEEE

158

IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 18, NO. 1, MARCH 2003

Fig. 2. Calculated NDZ of a 200-W inverter at different power levels and for different resonant circuits with fixed voltage and frequency limits.

Fig. 3. Grid voltage and current waveform of a PV inverter applying AFD.

B. Reactive Power by Active Antiislanding Techniques an inverter with predefined voltage and frequency limits operates in islanding mode [1], [3]. In Fig. 2, the NDZs of a 200-W inverter are indicated for different operating points, being deter. mined by combinations of , , and Theoretical examinations on the probability of nonintentional islanding show that matched load may occur with high PV grid penetration (i.e., installed PV power normalized to the load of the distribution grid [4], [5]). Probability of occurrence and duration of nonintentional islanding in real low-voltage grids are still under examination by the International Energy Agency (IEA) in task V of the photovoltaic power systems program (PVPS) [6][8]. The stochastic examinations of matched load conditions do not take into account the effect of additional active antiislanding measures. Islanding can be prevented even under matched load conditions by applying additional active measures. Possible algorithms are different types of current pulse [9], [12], [19], [41] and frequency shift methods [3], [10][12], [14], [17], [42]. An issue that still has to be solved is the question of mutual disturbances of different antiislanding schemes on one section of the distribution grid [9], [12]. III. INFLUENCE OF REACTIVE POWER Earlier studies on islanding of self-commutated inverters for grid-connected PV assumed an inverter power factor close to unity [2], [3]. The testing load was purely resistive. Reactivewas assumed to be negligible. More power mismatch recent measurements found reactive-power output values of up to 20% of active power at full load and 60% of active power at 30% of full load [13]. A. Reactive-Power Consumption of the Inverter Capacitance The main reactive power is introduced by the grid filter being part of the grid interface. While the inverter output stage is controlled to generate active power only, the subsequent grid filter introduces a capacitive component proportional to the square of the grid voltage. In practice, manufacturers apply more sophisticated filter designs in order to ensure the high power factor over a wide power range, requested by most guidelines and standards [24][39]. The reactive power introduced by active antiislanding techniques depends on the technique applied. Here, the impact of active frequency drift (AFD) on the reactive-power balance is illustrated. A detailed discussion of AFD is given in [14]. Inverters with AFD generate a slightly distorted current waveform (Fig. 3). In this example, the first current half cycle is shorter than half of the period of the grid voltage. The time is called zero time difference between both zero crossings or dead time. For the second half-cycle, the current of the first is half-cycle becomes inverted and the control bias for measured. to half of the period of the grid voltage is The ratio of referred to as the chopping fraction (3) If islanding occurs with purely resistive loads, voltage and current have the same shape. As a consequence, in order to maintain a constant chopping fraction, the control algorithm increases the frequency of the output current. Again, the voltage follows and the frequency drifts until the frequency limit is met. However, this method fails for loads with considerable capacitance [13], [14], as present in most European cable distribution VAr, grids. With the circuit shown in Fig. 1 and all higher-order current harmonics virtually become short-circuited. As a consequence, the voltage remains approximately sinusoidal. For AFD, the power factor is given by the chopping fraction as calculated from (3) (4) Since the power factor of the fundamental is predetermined by , and is enforced by the PV array, the reactive power can also be assumed constant. This means that for realistic values of , AFD has no impact neither on the size, nor on the location -domain. of the NDZ in the C. Description of Nondetective Zones In order to take into account the influence of a resonant load, phase criteria have been proposed for describing nondetective

WOYTE et al.: TESTING THE ISLANDING PROTECTION FUNCTION OF PHOTOVOLTAIC INVERTERS

159

zones substituting the power equations in the -domain [14][16]. The NDZ is defined in a parameter space of resistance , inductance , and capacitance normalized on reswithin the isolated onant capacitance at grid frequency part of the grid. However, the positive impact of reactive power introduced by active antiislanding measures becomes overestimated by this criteria. For determining the worst-case testing load, the reactive-power portion according to (4) is simply neglected and only reactive power supplied by the grid filter is taken into account. In practice, for an islanding test procedure, this definition of the NDZ implies that the reactive-power portion originating from AFD is not compensated by the testing load. Reacis not adjusted to zero, implying that tive-power mismatch the tests do not represent a worst-case scenario. Reactive power originating from AFD might therefore be misinterpreted: a higher chopping fraction would lead to decreasing size of the NDZ not being true when assuming a precisely matched load and a large parallel capacitance. , and directly result from Unlike , , and the power balance. Therefore, it is better to keep on describing -domain with as an additional parameter, NDZs in the of the resonant circuit or, alternathe stored reactive power as to be defined tively, the resonant circuits quality factor by (5). IV. STABILIZING ELEMENTS The introduction of stabilizing factors to the islanding test procedure renders detection more difficult. First islanding tests had been carried out by applying the circuit from Fig. 1 in the absence of any stabilizing component. Islanding under these circumstances only occurred with pure frequency and voltage monitoring in the absence of any further antiislanding technique [2], [17], [19]. In existing grids, an island in equilibrium would be stabilized by rotating machines, other embedded generators, and reactive power being stored in the grids inductive and capacitive elements. Therefore, a large variety of approaches to introducing stabilizing elements to the islanding test is being discussed worldwide. A. Rotating Machines and Resonant Circuit Induction machines add inertia to a separated grid section, considerably retarding the frequency shutdown. As these machines are present in existing grids, the need was felt to connect a rotating machine in parallel to the test setups [19], [20]. The islanding test circuit as it is currently requested in Germany where the grid is not disconnected but only weakened by introduction of a series resistance is still a relic from these approaches [2], [18], [19], [21]. In other countries, the island was stabilized by a resonant circuit tuned to grid frequency rather than by the inertia of a rotating machine. Although physically, the impact of rotating machines is different from the one of a resonant circuit, their effect is comparable as both of them, in the seconds after a grid disconnection, increase the tendency to maintain the grid frequency, thus preventing the frequency relay from tripping.

Further research, recently carried out in the U.S. has demonstrated that rotating loads not necessarily result in worse disconnection properties of the islanding detection scheme than matched resonant loads with high quality factor [17]. A practical difficulty with the application of rotating machines lies in the matter of reproducibility of test conditions. In order to create reproducible test conditions, the machine properties, including inertia and friction, must be specified and test laboratories all over the world would have to apply them. Hence, in Europe, the U.S., and Australia, solely a resonant circuit is applied. B. Sizing of the Resonant Circuit Regarding the application of a resonant circuit, it is discussed whether this resonant circuit should have a specified size in absolute terms of reactive power or a specified quality factor, implying that reactive power of the resonant circuit is linked to the inverter operating point. Specifying the absolute size of the circuit, meaning a fixed in VAr renders islanding detection more difficult value for for small inverters and at operating points with partial load [13]. The resonant circuit represents reactive power that under real conditions is stored in the grid. The value of 100 VAr as requested in Australia [26], Belgium [28], and Germany [37] has been agreed upon after discussion on what would be a reasonably high value that still could be realized with simple laboratory equipment. It must therefore be seen rather on a historical and practical background than on a technical one. is reIn the U.K. and U.S., a specified quality factor quested. The quality factor of a resonant circuit is defined as the ratio of reactive power in the circuit to active power conand matched sumption. At resonant frequency with , according to Fig. 1), the relation between active load ( quality factor and size of the resonant circuit is (5) With being the inverters phase angle and assuming no higher order harmonics, (2) becomes

(6) Equations (1) and (6) describe the NDZ only by relative terms and . The NDZ is now scalof power mismatch able to all different inverters and operating points. is proposed as being a typical quality In the U.S., factor for a distribution grid being fully compensated by capacitance [17]. In the according standard and test procedure [24], . [25] therefore, it has to be tested with In the U.K., a quality factor above 0.5 has to be applied. The reasoning is different from the one in the U.S. The resonant circuit is applied in order to simulate a second PV inverter within the disconnected grid section, introducing a tendency to keep has the grid energized at rated frequency. The value been chosen as this represents the aperiodic limit for a resonant circuit to perform free oscillations. The quality factor of the total grid section does not play a role in this argumentation [16], [22], [23], [29].

160

IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 18, NO. 1, MARCH 2003

TABLE I SUMMARY OF THE TEST RESULTS AND CONCLUSIONS

Fig. 4. NDZ for inverter A recorded at P

=03
:

1 P

, Q = 0:1 1 P .

Fig. 5. NDZ for inverter D recorded at P = 0:3 1 P , Q = 0:18 1 P .

The application of a constant quality factor rather than a resonant circuit of constant size is useful as this provides scalability of the test procedure and can also be justified technically. From a practical point of view, it is beneficial to test with not too high of a quality factor. For typical single phase inverters of leads already to considup to approximately 5 kVA, erable values of inductance and capacitance. Higher values for might make testing unreasonably expensive due to the large equipment that would be required. V. LABORATORY TESTING OF MODULE INVERTERS The NDZs of four small, so-called module inverters of maxare recorded by applying the cirimum 200-W rated power VAr. cuit presented in Fig. 1 with The tests demonstrated that each of the four inverters can be forced into islanding under certain load conditions. Partly, the NDZs correspond to the calculated areas, indicating that some of the applied algorithms work insufficiently with loads containing a capacitive component. However, the NDZs are not al-domain as (1) ways located around the origin of the and (2) imply. Figs. 4 and 5 show the results of two of the measurements. Inside the inner zone indicated by triangles, the inverter remaines islanding. Outside the outer zone indicated by circles, the inverter immediately switched off. The border of its NDZ is thus

located in between both zones. The theoretical NDZ as calculated from (1) and (2) from the settings of the frequency and voltage relays is indicated by the solid border. From the location of the NDZ and the causes for shutdown at its boundaries, conclusions on the effectiveness of the particular protection algorithm can be drawn. Table I gives a summary of the different inverters islanding behavior. Comprehensive results of the conducted test series have been presented in [13]. The failing of inverter D has to be interpreted as a prototype problem. The results have already been discussed with the manufacturer. The implemented algorithm will be rechecked. The tests on inverters A and B verify the investigations made in [14]. Their recorded NDZs correspond to the ones computed for a situation without further antiislanding measures. The frequency drift algorithm implemented in inverter B fails with considerable capacitive loads. Inverter A remains islanding because the voltage-control loop does not become unstable as intended. Inverter C shows excellent results with regard to islanding. The manufacturer has apparently made a good effort in order to prevent islanding. Information about the islanding prevention schemes applied is, however, not publicly available. VI. CONCLUSIONS In order to ensure worst-case test conditions independent of the antiislanding algorithm, the active- and reactive-power misand , must be adjusted to zero with regard to the match, current and voltage fundamentals. The introduction of stabilizing elements makes the test conditions more realistic. For practical reasons, the introduction of a resonant circuit with a predefined quality factor is favorable , the rather than a rotating load. With regard to the value for practical boundary conditions for implementation in the laboratory should be taken into account. Regarding the low probability for matched load conditions, the choice of a relatively low quality factor between 0.5 and 1 would be appropriate. Test results show that small inverters are still sensitive to islanding if tested in a worst-case scenario and loaded with ca-

WOYTE et al.: TESTING THE ISLANDING PROTECTION FUNCTION OF PHOTOVOLTAIC INVERTERS

161

pacitance. In order to avoid the need for oversized and expensive protection equipment, the applied islanding protection algorithms still have to be improved. Algorithms based on instability of voltage and frequency, while the grid is tripped, can play a major part. However, those algorithms must be implemented with care. The particular stability limits should be checked thoroughly by theoretical examination, simulation, and worst-case tests. Still, an issue for future antiislanding schemes is the avoidance of mutual disturbances by a high number of different types of inverters in an isolated grid section. Research about this subject is still ongoing. ACKNOWLEDGMENT The authors wish to thank J. Appelbaum for his useful criticism and comments during the preparation of this paper and L. Conings who carried out part of the measurements described. REFERENCES
[1] H. Van Reusel, R. Belmans, W. Coppye, L. De Gheselle, M. Dussart, and P. Lauwers, Adaptation of the Belgian regulation to the specific island behavior of PV grid connected inverters, in Proc. 14th Eur. Photovoltaic Solar Energy Conf. Exhibition, Barcelona, Spain, 1997, pp. 22042206. [2] H. Hberlin and J. Graf, Islanding of grid-connected PV inverters: Test circuits and some test results, in Proc. 2nd World Conf. Exhibition on Photovoltaic Solar Energy Conversion, Vienna, Austria, 1998, pp. 20202023. [3] M. E. Ropp, M. Begovic, and A. Rohatgi, Prevention of islanding in grid-connected photovoltaic systems, Progress in Photovoltaics: Res. Applicat., vol. 7, pp. 3959, 1999. [4] M. Begovic, M. E. Ropp, A. Rohatgi, and A. Pregelj, Determining the sufficiency of standard protective relaying for islanding prevention in grid-connected PV systems, in Proc. 2nd World Conf. Exhibition on Photovoltaic Solar Energy Conversion, Vienna, Austria, 1998, pp. 25192524. [5] H. Kobayashi and K. Takigawa, Statistical evaluation of optimum islanding preventing method for utility interactive small scale dispersed PV-systems, in Proc. 1st World Conf. Photovoltaic Energy Conversion, Waikoloa, HI, 1994, pp. 10851088. [6] A. Kitamura, Rokko island testing center (results of demonstration tests), in Proc. Int. IEA-Workshop on Existing and Future Rules and Safety Guidelines for Grid Interconnection of Photovoltaic Syst., Zrich, Switzerland, Sept. 1997. [7] , Demonstration Test Results for Grid Interconnected Photovoltaic Power Systems, New Energy and Industrial Development, Tokyo, Japan, Rep. IEA-PVPS T50-02: 1999, 1999. [8] T. Yoshioka and T. Ishikawa, Grid-Connected Photovoltaic Power Systems: Summary of IEA/PVPS Task V, New Energy and Industrial Development, Tokyo, Japan, Rep. IEA-PVPS T5-03: 1999, 1999. [9] K. Kln , A. Grabitz, P. Kremer, and B. Kress, Five years of ENS (MSD) islanding protectionWhat could be the next steps?, in Proc. 17th European Photovoltaic Solar Energy Conf. Exhibition, Munich, Germany, 2001, p. 2544. [10] D. Schulz and R. Hanitsch, Netzkopplung von PhotovoltaikanlagenRealisierungsmglichkeiten der Netzberwachung, in Proc. 16. Symp. Photovoltaische Solarenergie, Staffelstein, Germany, 2001, pp. 218226. [11] G. A. Smith, P. A. Onions, and D. G. Infield, Predicting islanding operation of grid connected PV inverters, Proc. Inst. Elect. Eng. Elect. Power Applicat., vol. 147, no. 1, pp. 16, 2000. [12] O. Tsukamoto, T. Okayasu, and K. Yamagishi, Study on islanding of dispersed photovoltaic power systems connected to a utility power grid, Solar Energy, vol. 70, no. 6, pp. 505511, 2001. [13] A. Woyte, R. Belmans, and J. Nijs, Islanding of grid-connected AC module inverters, in Proc. 28th IEEE Photovoltaic Specialists Conf., Anchorage, AK, 2000, pp. 16831686. [14] M. E. Ropp, M. Begovic, and A. Rohatgi, Analysis and performance assessment of the active frequency drift method of islanding prevention, IEEE Trans. Energy Conversion, vol. 14, pp. 810816, Sept. 1999.

[15] M. E. Ropp, M. Begovic, A. Rohatgi, G. A. Kern, R. H. Bonn Sr., and S. Gonzalez, Determining the relative effectiveness of islanding prevention techniques using phase criteria and nondetection zones, IEEE Trans. Energy Conversion, vol. 15, pp. 290296, Sept. 2000. [16] J. M. Thornycroft, R. J. Hacker, E. He, T. Markvart, and R. J. Arnold, Type testing of inverters for grid connection of photovoltaics, in Proc. 16th Eur. Photovoltaic Solar Energy Conf. Exhibition, Glasgow, U.K., 2000, pp. 24572461. [17] J. Stevens, R. Bonn, J. Ginn, S. Gonzales, and G. Kern, Development and Testing of an Approach to Anti-Islanding in Utility-Interconnected Photovoltaic Systems, Sandia National Laboratories, Albuquerque, NM, SAND 20001939, 2000. [18] A. Woyte, R. Belmans, J. Nijs, P. Heskes, and F. Phlippen, Mains monitoring and protection in a European context, in Proc. 17th European Photovoltaic Solar Energy Conf. Exhibition, Munich, Germany, 2001, pp. 580583. [19] S. Marte and P. Kremer, Methods against islanding for small grid connected inverters, in Proc. 13th European Photovoltaic Solar Energy Conf. Exhibition, Nice, France, Oct. 1995, pp. 18391841. [20] A. Mizorogi and K. Takigawa, Islanding prevention performance test of Japanese manufacturers inverters, Solar Energy Materials & Solar Cells, vol. 67, pp. 583589, 2001. [21] D. Schulz and R. Hanitsch, Islanding detection in Germany: Current standards and development, in Proc. 17th European Photovoltaic Solar Energy Conf. Exhibition, Munich, Germany, 2001, pp. 520523. [22] Co-Ordinated Experimental Research Into PV Power Interaction With the Supply Network-Phase 1, ETSU, ETSU Rep. 1, 1999. [23] Co-Ordinated Experimental Research Into PV Power Interaction With the Supply Network-Phase 2, ETSU, ETSU Rep. 2, 2000. [24] IEEE Recommended Practice for Utility Interface of Photovoltaic Syst., IEEE Std. 929-2000, Jan. 2000. [25] Appendix A: Proposed First Edition of the Standard for Static Inverters and Charge Controllers for use in Photovoltaic Power Syst., UL Subject 1741 Draft, Mar. 1999. [26] Grid Connection of Energy Syst. Via Inverters, Draft Australian Standard for Comment DR 01212-01214, July 2001. [27] Characteristics of the Utility Interface for Photovoltaic (PV) Syst., EN 61 727, 1996. [28] Technical Requirements for Connection of Dispersed Photovoltaic Gen15 kW Operating in Parallel on the Distribution Neterating Syst. work, Lastenboek BFE-FPE CCLB 06/06.00, June 2000. [29] Recommend. for the Connection of Inverter-Connected Single-Phase Photovoltaic (PV) Generators up to 5 kVA to Public Distribution Networks, Engineering Recommend. G77, 2000. [30] Installations Photovoltaiques Productrices denergie, Prescriptions Provisoires de scurit, June 1990. [31] Proyecto De Orden de del 2000 por el que se Establecen las Condiciones Administrativas y Technicas de Conecion de Las Instalaciones Fotovoltaicas a la red en Baja Tension, Draft, Jan. 2000. [32] Guidelines for the Elect. Installation of Grid Connected Photovoltaic Syst., Dec. 1998. [33] Supplementary Conditions for Decentralized Generators, Low-Voltage Level, Nov. 1997. [34] Dienst Uitvoering En Toezicht Elektriciteitswet Netcode (DTE), Dec. 1999. [35] Design Qualification and Type Approval of Inverters for Grid-Connected Oper. Photovoltaic Power Generators, Nov. 1999. [36] Eigenerzeugungsanlagen am NiederspannungsnetzRichtlinie fr Anschlu und Parallelbetrieb von Eigenerzeugungsanlagen am Niederspannungsnetz, VDEW Guideline, Jan. 2001. [37] Selbstttige Freischaltstelle fr Photovoltaikanlagen einer Nennleistung 4.6 kVA und einphasiger Paralleleinspeisung ber Wechselrichter in das Netz der ffentlichen Versorgung, E DIN VDE 0126, Apr. 1999. [38] Photovoltaic Semiconductor Converters, Part 1: Utility Interactive Fail Safe Protective Interface for PV-Line Commutated ConvertersDesign Qualification and Type Approval, Draft prEN 50 330-1, Feb. 1999. [39] Utility Interactive Fail Safe Protective Interface for PV-Line Commutated ConvertersDesign Qualification and Type Approval, NWIP Draft 82/234/NP, Dec. 1999. [40] Safety of MachinerySafety-Related Parts of Control Syst.Part 1: General Principles for Design, EN 954-1:1996, 1996. [41] K.-W. Kln , Method and Device for Measuring Impedance in Alternating Current Networks and Method and Device for Preventing the Formation of Separate Networks, World Patent WO 96/10 188, Apr. 4, 1996. [42] E. Masaki, K. Hirokazu, T. Tsukasa, and N. Hirofumi, Apparat. Method for Detecting an Inverter Islanding Oper., Eur. Patent EP 0 810 713 A2, Dec. 3, 1997.

<

162

IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 18, NO. 1, MARCH 2003

Achim Woyte received the electrical engineering degree from the University of Hannover, Germany, in 1997. He spent half a year working and studying in Venezuela and Italy. He is currently pursuing the Ph.D. degree at K.U. Leuven, Leuven, Belgium. He worked for more than three years in electroheat and high-voltage engineering at University of Hannover. His research interests include the grid interface and grid interaction of residential photovoltaic systems.

Ronnie Belmans (S77-M84-SM89) received the M.S. degree in electrical engineering in 1979, the Ph.D. degree in 1984, and the Special Doctorate in 1989 from the K.U. Leuven, Belgium, and the Habilitierung degree from the Rheinisch-Westflische (RWTH), Aachen, Germany, in 1993. Currently, he is Full Professor with K.U. Leuven, teaching electrical machines and variable speed drives. He is Appointed Visiting Professor at Imperial College, London, U.K. He is also President of the Union Internationale pour les applications de Electricit (UIE). He was with the Laboratory for Electrical Machines of the RWTH, and was a Von Humboldt Fellow from October 1988 to September 1990. He was Visiting Associate Professor at McMaster University, Hamilton, ON, Canada. During the academic year 1995-1996, he occupied the Chair at the London University, U.K., offered by the Anglo-Belgian Society. His research interests include variable speed drives, vibrations and noise in electrical machines, electrical energy systems, and power quality. Dr. Belmans is a fellow of the IEE U.K.

Johan Nijs (SM97) received the university degree in electrical engineering, and the Ph.D. degree in applied sciences from the K.U. Leuven, Belgium, in 1977 and 1982, respectively. After having worked at Philips, Leuven, and, K.U. Leuven, Belgium, and I.B.M. Thomas J. Watson Research Center, Yorktown Heights, NY, he joined the Interuniversity Micro-Electronics Center (IMEC) in Leuven, in 1984 where he became groupleader and recently associate vice-president/department director of the Packaging, MEMS, and Photovoltaics department. In 1990, he was also appointed Part-Time Associate Professor at K.U. Leuven. Dr. Nijs is a full-member of ISES-Belgium.