By Claude F. Kane
Management is always looking for ways to improve its reliability of their production assets while reducing maintenance costs and downtime. More often than not, these goals seem to conflict and the maintenance manager is caught in the middle. Production and budget constraints do not always allow sufficient time to shut equipment down for maintenance and testing, but when the production line or process unexpectedly shuts down, enormous cost figures are frequently quoted ranging from hundreds of dollars per hour to hundreds of thousands of dollars per hour.
Many maintenance professionals are setting up programs to monitor and trend off-line testing results such as insulation resistance, leakage current and insulation power factor. These are all excellent tests, but it takes years to compile enough data to provide useful information. Depending on the maintenance program, there may be anywhere from 6 months to 5 years between obtaining test data for comparison purposes. A lot can go wrong within that time span.
In order to narrow the gap of these conflicting goals, maintenance professionals are actively searching for on-line predictive maintenance methods that will allow them to determine the health of the equipment in their facilities and to provide additional information in order to prioritize expenditures of a limited maintenance budget. In the case of electrical equipment, the maintenance professional have had a limited number of resources to draw upon to meet these new goals. To date, cost effective predictive maintenance tools used on energized electrical equipment are limited to the following:
- The human senses of sight, smell, sound and touch
- Recording and tracking of various electrical quantities such as voltage, current and power
- Fluid testing of transformers and inspection of gauges
- Vibration monitoring equipment for rotating apparatus
- Ultrasonic corona detection
- Thermographic inspection
When applied safely and correctly all of the above can provide the maintenance department valuable information, but each has its limitation.
The use of the human senses is an excellent tool, but frequently when a problem has reached the point where one of these senses indicate a problem, a failure is imminent and one needs to act quickly. The recording and trending of the various electrical quantities can assist in prioritizing maintenance requirements. Usually equipment with higher loads will degrade faster and will require more frequent maintenance. Performing periodic fluid tests on liquid filled transformers has proven to be a valuable source of information as well as frequent inspection of the various gauges. Monitoring and trending the vibration signatures of motors and generators is useful in providing information as to the mechanical health of a machine. Vibration signatures have little to do with the electrical health of a machine.
Over the past four years, use of ultrasonic sensors to detect the presence of corona in medium and high voltage equipment has become common. Use of this equipment is somewhat limited to exposed electrical equipment and outdoor substations. There has been limited success in use on metal enclosed switchgear since more often than not, maintenance personnel cannot obtain access to the various higher voltage compartments. For example, in metalclad switchgear, the main bus is not accessible, therefore you only have access to the main breaker and termination compartments. Opening the breaker compartment door and/or removing the termination compartment panels presents serious safety concerns. Another example is a medium voltage fused disconnect switch. Due to safety interlocks, the test equipment operator has no safe access to the energized components. The ultrasonic sound created by corona cannot and does not travel through the sheet metal. You may have some success using the detector through vents, but the results will be of limited value. This technology is not applicable to the 600 volt class equipment since the corona phenomena does not occur at these lower voltages.
The application of thermography technology to surveying electrical equipment for heat is probably the best known and most widely used of any on-line predictive maintenance technology. There is no doubt that the introduction of thermography provided the most exciting advancement in the electrical maintenance industry in 25 years. As with corona detection, the applicability of thermography to many medium voltage systems is limited. Having direct access to the energized areas of the medium voltage equipment is difficult at best and presents many safety concerns.
Thermography is best suited for 600 volt class equipment and below and for outdoor substations. It provides limited value for use with the various metal enclosed switchgear systems.
A recent addition to the lineup of available on-line predictive maintenance technologies is the measurement of Partial Discharges (PD) on distribution systems and equipment greater than 2,000 volts AC. Measurement of PD will be the most exciting predictive maintenance technology since the advent of thermography. A partial discharge occurs when there is a small void, crack or irregularity in an insulation system, where an electric field can build up. This electrical field will eventually cause enough stress and degradation of the insulation system to cause complete failure of the insulation. This complete failure is called a Full Discharge. An electrical discharge that has not completely bridged an insulation system is called a partial discharge. The magnitude of these discharges is usually very small and difficult to detect. The time span from when partial discharge begins to the point of a full discharge depends on many factors. Voltage levels, the shape of the void, ambient temperatures, systems losses all can have an effect on how quickly the insulation will fail. As with all electrical tests, trending of data is very important.
Partial discharge phenomena occurs in all types of medium voltage electrical equipment including: switchgear, bus duct, transformers, arresters, bushings, potheads, switches, motor starters, motors, generators, cable terminations, cable splices and the cables themselves. PD has been known for years to be a leading indicator of insulation breakdown. The problem has always been in finding a reliable, cost effective way of measuring PD. PD technology has finally found its way from the laboratory to the field. For PD measurements to be of any value, it must be able to detect the presence of breakdowns in a non-destructive manner and with the equipment energized. The goal desired by all maintenance professionals is to have a method of measuring insulation systems integrity without shutting down equipment and performing tests.
For example, in the case of switchgear, a special sensor is connected between the grounded side of the metering CT circuit and the cubicle sheet metal. On cables, the sensor is connected around the ground wire that connects to the cable shield or placed around the insulated conductor. On motors. sensors can be placed on the motor frame or around the equipment ground connection or around the insulated motor lead. Placement of the sensors is very important for sensitivity reasons. Each installation will have different requirements and the testing engineer will need to assess each application in order to provide the best possible results.
Current available on-line technology cannot locate the exact location of the partial discharge while the equipment is energized. It can only tell that there is PD activity and the severity of the activity. Through a process of elimination, the PD area can be localized, but the exact area of concern cannot be pinpointed. Use of special sensors off-line can determine the exact location of activity. As with all electrical measurements, one single data point provides little to no information. You need several data points over time to provide information. Initial judgments can be made by comparison of data from identical or similar equipment. For example, Figure 1, illustrates a lineup of four cubicles of 15 kV switchgear. Data from each cubicle is obtained. By comparing data between cubicles, information as to the severity and relative location of the PD activity can be determined.
Measurements On Switchgear
Figure 2 shows recommended locations to place sensors. By placing sensors in these various locations, you will be able to quickly localize the source of the PD activity. Again, you will not be able to pinpoint the exact location of PD activity while the equipment is energized, but you certainly can narrow it down to a small area.
Currently there are no set industry standards in determining maximum acceptable levels of partial discharge in various equipment except for cables. Medium voltage cables are tested for PD at the cable manufacturing facility. The maximum allowable PD magnitude is 5 pico-coulombs. As experience is gained, maximum levels of PD activity will be established for switchgear, bus duct and other electrical equipment. Once PD is found, there is usually no need to panic and immediately shutdown the equipment unless unusually high results are obtained. It is recommended that you initially survey all equipment and make >