Electrothermal model predicts loss of life

As a part of its pursuit of new technologies that will increase its reliability and efficiency, Entergy Transmission has installed a transformer diagnostic system on four power transformers. In the new electric utility environment, Entergy's goal is to operate its power equipment closer to its rating without compromising its integrity, longevity or its reliability. Within this framework, routine, scheduled maintenance will be replaced with condition-based maintenance where diagnostics play a critical role.

On-line transformer monitoring and diagnostics will constitute one facet of the proposed system. Pertinent information will be collected to permit intelligent loading decisions in the event of contingencies that may occur during normal operations. Early warning of incipient transformer failure is provided by the system's coil integrity functions, which will enable operators to switch load and schedule maintenance. In addition, the system's harmonic spectrum analyzer serves as a useful tool for harmonic analysis and power quality monitoring.

The diagnostic system consists of hardware and software that continuously monitors transformer operation and displays the history of the transformer. Input data consist of phase currents and voltages at both ends of the transformer, top and bottom oil temperatures, and load tap changer position. The collected data are fed into an electrothermal model of the transformer to predict loss of life and coil integrity.

System Description As shown in Fig. 1, the hardware consists of a 16-channel data acquisition unit, a personal computer and an uninterruptible power supply. Data are collected at the rate of 8196 samples per second per channel, which are simultaneously sampled by dedicated A/D converters. Each A/D converted channel is optically isolated from the rest of the system.

The data, continuously down-loaded to the personal computer via a 16-bit parallel interface at the rate of 262 kBytes per second, are used to calculate rms values of voltages and currents, transformer loss of life, an index indicating internal integrity of the coils and total harmonic distortion of voltages and currents. Input data are obtained from existing substation instrumentation, which include potential transformers and current transformers. The system software is multi tasking in the sense of allowing a user to make inquiries and to view data without interrupting the data acquisition function.

Transformer loss of life is evaluated from the recorded terminal voltages and currents, tap position and the oil temperatures. The internal heat sources, which drive the model, are losses in the iron core legs and in each winding. The system uses a state estimation method to compute the temperatures of the coils from the measurements.

Transformer coil integrity is evaluated via an index, which determines the consistency of the measured terminal voltages and currents. At intervals, which are usually every two minutes, a subset of measured waveforms is taken to compute the phasors of voltages and currents. The significance of the computed solution is based on the consistency of the computed estimates with the measured phasors. If an internal flaw exists, such as an intercoil discharge or shorted turns, there will be a large calculated error and the estimated values will not be consistent with the measured values. This discrepancy between measured and estimated values is expressed with a probabilistic measure that describes the goodness of fit between the measurements and the model. This probability is termed as the confidence level or index of coil integrity.

Fault Recording Capability In addition to transformer diagnostic functions, the monitoring system also functions as a fault or disturbance recorder. The system can be programmed to record and to store all monitored waveforms during disturbances. These waveforms are stored on disk and can be recalled and displayed on the screen or downloaded via modem. Downloaded data are continuously stored in a circular buffer while checking to ascertain if a trigger condition is met. The circular buffer (Fig. 2) always contains the latest samples from each channel. When a disturbance occurs, the systems continue to collect data for a user-specified time interval and the entire buffer contents are then stored in the computer disk in COMTRADE format. This process yields a waveform table containing data collected over a period of one second with a user specified pre-trigger section.

A flexible trigger specification system allows the user to specify criteria for initiating disturbance/fault recording. Specifically, the triggering can be set to occur when any monitored waveform, or a function of one or more waveforms, exceeds a specified threshold: -When the rms current on any phase exceeds twice the nominal current. -When the rms voltage on any phase is above or below specified thresholds.

The quantities needed to determine whether a user-specified condition or triggering spec is met are continuously computed over a sliding time window that is, typically, one cycle. Disturbance waveform tables are stored in the transformer monitoring system computer hard disk, which retains the data from the latest events. The number of events, which are retained, are specified by the user. These data can be viewed in graphical form on the system's monitor or can be transferred to the central office for viewing, printing and analysis.

Remote Monitoring With remote monitoring, it is possible to access any number of transformer diagnostic systems using a general purpose IBM compatible personal computer with a modem running Microsoft Windows 95. After establishing the modem connection, several operations can be performed to: -Download disturbance data (COMTRADE files). -Download historical data. -Inspect and modify site configuration parameters. -Perform on-line monitoring of the thermal and coil integrity model states. -Perform on-line monitoring of harmonics on any measured or computed channel.

-Achieve manual triggering for waveform recording.

Downloaded data can be viewed via a digital oscilloscope-type interface with any number of channels simultaneously displayed, zoomed and panned. Functions of the waveforms can also be computed and displayed using the waveform calculator interface to define the desired operation on the waveform data.

Examples of typical data displayed by the transformer diagnostic system are illustrated in Figs 3, 4 and 5. Figure 3 illustrates a snapshot of transformer performance in real time. In the upper left corner of the form, the measured phasors of the terminal voltages and currents are displayed. In the upper right corner, a graph of the phasors is shown. Any of the phasors can be displayed to show high-side values, low-side values and fluxes. The form also displays the measured tap setting as well as the computed system frequency, which is computed from the captured waveforms. In the lower part of the form, the measured temperatures and heat sources as well as the estimated temperatures of the coils and core legs are displayed. Finally, in the lower part of the form, the computed loss of life and the probability of coil integrity are displayed. This form permits the user to view the status of the transformer in real time.

Figure 4 shows the spectrum of one specific measurement; in this case, the low-side current of phase A with harmonic content displayed.

Figure 5 illustrates the performance history of a specific transformer showing the power output, low-side phase A current, transformer tap position, top of oil temperature, incremental loss of life, cumulative loss of life and probability of coil integrity. The data span a period of two days, for which no events have occurred for the moderately loaded transformer. Figure 6 illustrates a snapshot of transformer performance obtained remotely.

The Diagnostic System With several transformers equipped with the transformer diagnostic tool, it is possible to compute loss of life as an indication of the cumulative thermal stress on the transformer insulation, which can be used as a guide for designing maintenance schedules. The coil integrity index can be used as an early warning system of internal problems.

The disturbance recording capability is used to evaluate the effects of through faults and the computed hot spot temperature can be used for determining the limits of transformer loading before the permissible temperature level is exceeded in real time. In addition, data on harmonics and the total harmonic distortion of the transformer load can be monitored. All of the information provided by the system will result in the ability to optimize transformer maintenance and utilization.

P. Sakis Meliopoulos received the ME and EE diploma from the National Technical University of Athens, Greece in 1972 and the MSEE and Ph.D. degrees from Georgia Institute of Technology in 1974 and 1976, respectively. He joined Georgia Tech, where he is professor in the faculty of Electrical Engineering, active in teaching and research in modeling, analysis and control of power systems. He is a member of the Hellenic Society of Professional Engineering and the Sigma Xi.

G. J. Cokkinides received the BS, MS and Ph.D. degrees in electrical engineering from Georgia Tech and is associate professor at the University of South Carolina. His work includes power system simulation and control, electromagnetic system modeling, measurement instrumentation and CAD software development. He is a member of IEEE.

Rowland I. James received the BS degree in engineering science (electrical engineering) from Louisiana State University in New Orleans in 1972. He served in the U.S. Navy, worked as a design engineer in the telecommunications and petrochemical industries and joined Entergy at Louisiana Power & Light Co. He is now a member of Entergy Transmission, where he has done extensive work in substation ground grid design, testing and analysis and substation equipment monitoring, testing, inspection and analysis. He is a senior member of IEEE, past chairman of the New Orleans Section of IEEE and a member of IEEE/PES Transformer Committee.