For more than 35 years, the medium-voltage (MV) rural distribution network of Electricit, de France (EDF), Paris, France, has grounded its neutrals with a resistance that limited the single-phase short-circuit current to 150-300 A. Powered by high-voltage (HV) substations, 80% of the network operates at 20 kV and 20% operates at 15 kV.
Used since the 1960s, the resistance solution, although well suited to the overhead rural networks of the time, had reached its limits. The increased use of underground cables for distribution had increased the zero-sequence capacitive current to the point where fault currents of 600 A could be found. This situation led to overvoltages in the area of faults that were no longer compatible with existing standards. Limiting these values, while still keeping within the guidelines of the existing protection plan, would require building new substations to reduce the total length of overhead lines and underground cables connected to the same substation. This would require considerable investment.
A more economical solution would compensate for the undesirable effects of the phase-to-ground capacitance of the cables and lines by replacing the neutral-point resistance with a properly tuned coil. The crew would ground the neutral with a compensation coil that limits the fault current to 40 A compared to 100-1000A with the traditional grounding methods. The coil enables the overvoltages to be controlled by reducing the fault current by approximately 10 to one at the ground-fault site. The quality of service is then improved as three faults out of four self extinguish without disturbing customers.
New Equipment Required EDF was forced to develop new concepts because the compensated neutral modifies the ground-fault characteristics (strong reduction of the current at 50 Hz). In fact, EDF has developed a set of four new compatible procedures (Fig. 1), which have been patented by EDF's Research and Development Div. and which have won the WIJMANS prize awarded by the Dutch company, KEMA.
GENEPI The first procedure involves the compensation impedance (coil) and its automatic tuning system, GENEPI (GEstion du NEutre Par Injection - neutral management through injection). The GENEPI system tunes the coil automatically. Its principle is based on the injection of current in the zero-sequence system of the network. This injection allows the calculation of the compensated network parameters required for system tuning. The coil can be either a plunger core coil or a step coil.
For a few seconds (Fig. 2), a current (Iin) is injected in the neutral point via a transformer, which is inserted into the zero-sequence system. This injection causes the zero-sequence voltage (Vo) to vary. Analysis of the variation ratio between the zero-sequence voltage and the injected current allows the zero-sequence impedance of the circuit to be determined and, therefore, the mismatch of this circuit can be calculated. Once the system has acquired this information, it may then directly adjust the position of the compensation coil.
This system is compatible with all network configurations, especially if the network has a slight natural unbalance. It minimizes the adjusting operations of the compensation impedance during the tuning operations and it does not require the network to be mismatched during the measurement.
Protection Device Using a compensation coil makes the presently used current protection scheme ineffective. On the other hand, the utility must be able to detect restriking faults that appear on compensated networks. So a wide-bandwidth zero-sequence wattmetric protection device is installed on each feeder to acquire its zero-sequence active power. The sign of this power and the occurrence of a fault allows the faulted feeder to be determined. The devices' sensitivity allows detection of ground faults up to one to 2 kOhms on the French network where they are installed (average feeder length: 10 miles). Only the residual current of the faulty feeder has an active component (which is in phase opposition to the zero-sequence voltage). Thus the zero-sequence active power is negative in the faulty feeder and null in the sound feeders.
Monitoring this feature provides fault detection on highly capacitive feeders. For this reason, the method is particularly convenient for fault detection in combined overhead-underground networks.
DESIR The DESIR (DEtection Selective par Intensites Residuelles - selective detection through residual current) algorithm compares the vectors of the zero-sequence residual currents of each feeder (Fig. 3). After the 50-Hz component vectors of feeder residual currents have been extracted, the projections of these currents can be compared on a reference axis (Fig. 4). This axis is perpendicular to the sum (In) of the 50-Hz residual currents. The projection of the faulted feeder's residual current (Irdd) is in phase opposition to the unfaulted feeders. Thus, the faulted feeder can be detected using this comparison. DESIR's sensitivity allows accurate detection of resistive faults with a fault resistance ranging from 6 to 20 kOhms.
The DESIR system allows one to ensure selectivity in case of resistive fault currents. Its detection method may be adapted to the resistive neutral point connection. DESIR can be used with or without residual voltage measurement.
Fault Detectors The fault detectors are based on the analysis of the variation of the zero-sequence voltage and the zero-sequence current at the occurrence of the fault. The detectors installed on overhead lines (Fig. 6) use magnetic-field sensors (acquisition of the zero-sequence current) and electric field sensors (acquisition of the zero-sequence voltage). The detectors installed in cable networks (Fig. 5) are current transformers (acquisition of the zero-sequence current) and existing capacitive voltage dividers (acquisition of zero-sequence voltage).
This new method of detecting phase-to-ground faults is based on the analysis of the sign of the relative variation of the residual current and voltage signals when a fault occurs. The sensors immediately know if a fault is upstream or downstream from their location, thus allowing quick separation of the faulted section from the rest of the network.
This method allows simple and inexpensive sensors to be used. Magnetic and electric field sensors are mounted in overhead networks. And, inexpensive torus and capacitor voltage dividers fitted on MV cables as detectors can be used in underground networks.
Project Results The raw results have come up to expectations: a reduction of 34% in short-term outages and rate of self-eliminating faults equal to 75%. The main objectives of these experiments were to optimize the solutions in terms of cost and performance, checking compatibility with the existing system and the reliability of the system.
A detailed progress check was kept on the operation of the various equipment. Each event trapped by the network signal recorders was analyzed in detail. This analysis, completed by monitoring changes in the quality of supply and overvoltages, confirmed the technical utility of this new protection plan and its economic valuation.
The experimental feedback has not only allowed the specifications to be upgraded (refining of adjustments, taking into account of new operating constraints) and validated, but has allowed a more detailed study of integration into the existing substations.
The heavy involvement of the operator and his satisfaction with the results is a token of success for the deployment of this new stage.
This procedure ensures the supply of tailored and optimized products that meet the cost and performance requirements of the operator and significantly contribute to Electricite de France's moves toward improving safety and quality of service.
Fifteen hundred HV/MV substations are involved and modifying them all will take 10 years.
Note: EDF has registered engineering licenses for GENEPI, DESIR and a new concept of fault detectors. GENEPI and DESIR are also registered trademarks of EDF.
Jean Pierre Clerfeuille is head of the Power System Dynamics and Control Branch, Research & Development Div., Power Systems Dept., Electricite de France, which he joined in 1978. He has the degree of electrical engineer from the Ecole Superieure d'Electricite. Clerfeuille is a member of IEEE and CIGRE.
Philippe Juston is a research engineer in the Power System Dynamics and Control Branch, Research & Development Div., Power Systems Dept., Electricite de France, which he joined in 1984. He was graduated from the Ecole Superieure d'Electricite. Until 1993, he worked on power system stability and was project manager on the power system stabilizer. From 1994 he worked on neutral grounding of MV networks and on connection of embedded generation.
Michel Clement is head of the MV neutral grounding project, Distribution Div., Electricite de France, which he joined in 1981. He has the degree of electrical engineer from the Ecole Superieure d'Electricite. He was previously responsible for planning and operation of HV/MV substations and MV networks. He is the author of numerous international papers on neutral grounding and overvoltages.