B.D. Russell

A digital signal processing algorithm for detecting arcing faults on power distribution feeders

Signal processing hardware and software that can be used to improve the detection of certain power system faults using computer relays are discussed. Integrated systems and architectures for monitoring several fault-sensitive parameters have been investigated. A suggested architecture utilizing several processors is presented. Several fault-sensitive parameters for the detection of arcing faults are presented. A detection methodology based on these parameters is described, and a partial solution to the problem of directionality is discussed. The use of a knowledge-based environment to modify protection criteria is suggested. >

Analysis of high impedance faults using fractal techniques

Phase currents and voltages in a distribution power system change with a certain degree of chaos when high impedance faults (HIFs) occur. This paper describes application of the concepts of fractal geometry to analyze chaotic properties of high impedance faults. Root-mean-square rather that instantaneous values of currents are used for characterization of temporal system behavior; this results in relatively short time-series available for analysis. An algorithm is presented for pattern recognition and detection of HIFs; it is based on techniques suited for analysis of relatively small data sets. Examples are given to illustrate the ability of this approach to discriminate between faults and other transients in a power system.

An arcing fault detection technique using low frequency current components-performance evaluation using recorded field data

Several low frequencies of the current waveform in distribution feeders exhibit modified behavior under fault conditions. Two frequencies, 180 Hz and 210 Hz, were selected for study owing to the strong magnitude variations associated with arcing faults at these frequencies. A hierarchical algorithm with adaptive characteristics is presented along with the performance results for its application at these low frequencies. The various parameters that affect the sensitivity of the algorithm are discussed. The results of tests using recorded field data are given, and the effects of using a digital filter front end for the algorithm are also discussed. >

Behaviour of low frequency spectra during arcing fault and switching events

Distribution feeder faults modulate primary current and generate noise through arcing phenomena. The variation and behaviors selected low frequencies during fault conditions are presented. These are contrasted with normal events such as feeder-switching and capacitor-bank operations. Recorded field data have been analyzed, and the statistical results are presented. Specific behavior characteristics such as arc duration, arc repetition rate, and magnitude of low-frequency spectra are presented. Comparisons are made for different soil types and conditions. >

Effects of conductor sag on spatial distribution of power line magnetic field

Traditional extremely low frequency (ELF) magnetic field computation techniques assume that the current carrying power line conductors are straight horizontal wires. This assumption results in a model whose magnetic fields are distorted from those produced in reality. An exact solution and an approximation are proposed for modeling magnetic fields produced by the sagged conductors of power lines, by taking advantage of the fact that the equation of the catenary exactly describes the line sag. The proposed approaches differ in the required computational burden and in the precision of the results. A field mapping measurement example illustrates the applicability and the need for this analysis. The relative importance of the catenary effect is discussed.

A parameter-based process for selecting high impedance fault detection techniques using decision making under incomplete knowledge

Using varying environmental parameters and nonadapting detection techniques, the authors propose a generic method for choosing the fault detection technique or techniques which are most appropriate for a given set of conditions. This result is achieved using a decision-making method under incomplete knowledge. The performance of each detection technique is derived under various conditions and environments with this method. The method is not yet complete because the performance outcome of all techniques is not fully defined. However, it provides a head start when attempting to optimize detection given consideration of a wide variety of techniques, parameters, and environmental parameters. >

Measurement of magnetic fields in the direct proximity of power line conductors

Modeling and managing of power frequency magnetic fields requires verification of theory with actual measurements. Measurements only at ground level are not always sufficient for comprehensive studies. The technique and the results of three-dimensional mapping of the power frequency magnetic fields high above ground level are presented in this paper. Comparative calculations illustrate relevance and approximations of the existing theoretical approach to field modeling. The influence of harmonics on the elliptical rotation of the magnetic field vector is illustrated. The possibility of use of the magnetic fields for the power line proximity detection is discussed. >

Non-intrusive sensing techniques for the discrimination of energized electric cables

The implementation of a piezoelectric acoustic sensor for nonintrusive detection of the energization status of 3-/spl phi/ electric power cables is presented. Simultaneous excitation of a piezoelectric crystal by acoustic vibrations and electric field may occur. The Fourier spectral analysis of the sensors output signal is used to determine the current loading status of the cable. Test results are included for both shielded and nonshielded 3-/spl phi/ cables. Also, the possible use of an optical fiber interferometer for acoustic wave detection is discussed.

Applications of microcomputer-based systems in power substations

Three applications of microprocessors in power substations are presented. The use of the processors for metering, high-speed data recording, and distribution feeder automation is described. Emphasis is placed on those functions which could not be performed if microprocessor-based design architectures were not utilized. Specific information is given concerning substation hardware. The advantages of microprocessor-based designs are discussed, and details are given of the performance of field installations. >

Evaluation of methods for discrimination of energized underground power cables

The maintenance and repair of underground cable power networks often requires the discrimination of energized cables among a group of unenergized ones. Presently, an invasive detection method is performed which requires utility personnel to puncture the protective cable shielding, creating a potentially hazardous situation. This paper presents a noninvasive approach for energized cable detection based on the electrically induced acoustic waves inside of the cable. The influence of background electric fields on the performance of this approach is studied. In addition, the use of a fiber optic acoustic sensor would significantly increase the safety of the detection procedure, borrowing on the intrinsic electrically insulating properties of optical fibers (i.e. dielectric, immunity to electromagnetic interference, etc.). Feasibility studies and preliminary experiments were performed using a fiber Fabry-Perot interferometric acoustic sensor. The performance of the fiber optic sensor is compared with that of a piezoelectric accelerometer. A detailed analysis of the test results is included. Recommendations are given for correct interpretation of test results based on the specifics of a power system and sensor type.