Testing Prevents Surge in Arrester Failures

For more than 30 years, Manitoba Hydro has been testing surge arresters on delivery, at installation and during operation.

SURGE ARRESTERS ARE OFTEN OVERLOOKED WHEN PERFORMING POWER FACTOR TESTS on transformers, breakers and other apparatus in a substation. Oftentimes, the testers are aware of how a transformer or a breaker functions but unaware of the intended purpose of the surge arresters. Because there are no moving parts to maintain or an oil sample to pull, it is often policy not to perform any arrester testing.

Manitoba Hydro (MH; Manitoba, Canada) has been following a surge arrester testing program for more than 30 years. Under the program, typically 30 to 40 arresters are identified each year as requiring replacement. This can be due to an increase in 10-kV, 60-Hz watts-losses; a decreased voltage to produce 1 mA and 2 mA of resistive current; a reduced sparkover withstand voltage; or an increased sparkover operating voltage. MH's arrester testing program has been especially successful in identifying moisture ingress concerns because of design inadequacies specifically related to the newer generation of polymer-housed metal-oxide arresters.

MH purchases an average of several hundred surge arresters per year. These are mainly distribution and station class but also include some specialty units for installation at generation and converter stations. The arrester test program requires that arresters be acceptance tested prior to the supplier being paid. MH's acceptance test includes the Doble (Boston, Massachusetts, U.S.) watts-loss test, a partial-discharge test and a characterization of the voltage-current (V-I) response of each unit. The maintenance tests are limited to the Doble watts-loss test and, when practical, the V-I characterization or sparkover test. When reviewing test results, limits are applied as specified in standards IEC 60099 and IEEE C62. Also, the methodology for the qualification of test results requires that results for similar models and sister units be comparable.


All new station class and specialty arresters are acceptance tested by MH upon delivery. Typically, on new orders of both the station and distribution classes of arresters, MH experiences failure rates below 0.1%, although problem orders have been discovered at times. An example of this was for an order of 54 polymer-housed station-class arresters for installation at system voltages of 115 kV, 138 kV and 230 kV. The arresters originally passed acceptance tests, but several of them failed the Doble watts-loss test on-site prior to energization. The problem was revealed to be moisture ingress due to a vulnerable weather seal. The manufacturer replaced the entire order with porcelain-housed units and addressed the cause of the vulnerability.

In another instance, an unusual number of polymer-housed distribution class arresters were failing acceptance tests. These suspect units showed increased watts-losses, low voltages at the 1-mA resistive current and excessive partial discharge. The entire order was returned to the manufacturer, which identified the cause of the failures as having been related to the assembly of the units.


All service-aged arresters are field tested by MH staff either on an 8- or 10-year cycle, depending on the arrester's make and history. The 10-kV, 60-Hz watts-loss test is performed and, when practicable, either the V-I response is characterized or a sparkover test is performed. The V-I characterization is performed on metal-oxide arresters, while sparkover tests are performed on gapped arresters. The V-I and sparkover tests confirm an arrester's protective characteristics are at acceptable values. It is important to note that gapped arresters more than 25 years old are generally recommended for replacement because of their age. Also, all arresters undergo an infrared scan biannually. Arresters with temperature variations between adjacent units above 5°C (41°F) are investigated.

All arresters kept in stock for unplanned replacement purposes are tested every five years, while salvaged station class arresters are re-acceptance tested and returned to stock.


Through this arrester test philosophy, MH has found many units that either have had design vulnerabilities or suffered from the effects of deterioration. Both of these situations generally lead to unintentional arrester operation, which, because of the proximity of the arrester, can cause damage to important equipment. This test program has the added benefit of improving MH's electrical systems reliability as well as reducing the costs associated with forced outages. Additionally, the acceptance test program has enabled MH to extend the time between outages for maintenance tests on arresters. This acceptance test program also has the added benefit of fostering the development of in-house expertise by making available the necessary test equipment to thoroughly investigate problem arresters.

Reg J. Gamblin (rgamblin@hydro.mb.ca) is a technical officer in the Insulation Engineering & Testing department of Manitoba Hydro, where he is responsible for the condition assessment of generators, transformers, insulating oil, surge arresters, bushings and instrument transformers. In 1998, he earned a diploma in electrical engineering technology from Red River College. He has authored and co-authored several papers for industry conferences on condition assessment and high-voltage testing.

Keith Hill (khill@doble.com) has been employed at Doble Engineering since 2001 and currently works as a principal engineer in the Client Service department. He is secretary of the Doble Client Committee on Arresters, Capacitors, Cables and Accessories. Prior to joining Doble, Hill had over 25 years of testing experience, with the last 18 years as the electrical supervisor of engineering services at Lyondell-CITGO Refining (formerly ARCO). At ARCO, Hill started the testing group that was responsible for the plant power distribution system, electrical testing, power quality, plant-wide electrical reliability and infrared. Hill received his bachelor's degree, with a major in power, from the University of Houston. Hill is a member of IEEE, a Level I and II thermographer, and a former NETA-certified technician.

please send additional review pdf to Liisa Colby - lcolby@doble.com

Surge Arresters

Surge arresters help prevent damage to apparatus due to transient overvoltages. The arrester provides a low-impedance path to ground for the current from a lightning strike or switching overvoltage and then restores to a normal operating condition. A surge arrester may be compared to a relief valve on a boiler or hot water heater. It releases high pressure until a normal operating condition is reached. When the pressure is returned to normal, the safety valve is ready for the next operation. When a high voltage (greater than the normal line voltage) exists on the line, the arrester immediately furnishes a path to ground and thus limits and drains off the excess voltage. The arrester must provide this relief and then prevent any further flow of current to ground. The arrester has two functions: it must provide a point in the circuit at which an overvoltage pulse can pass to ground, and it must prevent any follow-up current from flowing to ground.

The technology of surge arresters has undergone major changes in the last 100 years. In the early 1900s, spark gaps were used to suppress overvoltages. In the 1930s, silicon-carbide arresters replaced the spark gaps. In the mid-1970s, zinc-oxide gapless arresters, which possessed superior protection characteristics, replaced silicon-carbide arresters.


Silicon-carbide arresters are currently less popular than the metal-oxide varistor (MOV), yet a great number of them are still in service.

In the October 1996 issue of IEEE Transactions on Power Delivery, Dr. M. Darveniza recommended all silicon-carbide arresters that have been in service for more than 13 years be replaced because of moisture ingress. His tests revealed that degradation was evident in 75% of arresters tested.

To determine if a silicon-carbide arrester warrants replacement, field testing must be performed. The ideal method is to determine the protective level of the arrester; this is not practical since an impulse generator is required. An effective and more-practical method is to determine the watts-loss of the arrester and compare to like arresters.


The MOV arrester is the arrester usually installed today. Doble (Boston, Massachusetts, U.S.) documentation reveals that MOV-type arresters entered the market in the United States around 1976. The metal-oxide arresters are without gaps, unlike silicon-carbide arresters. This gapless design eliminates the high heat associated with the arcing discharges. The MOV arrester has two voltage ratings — duty cycle and maximum continuous operating voltage — unlike the silicon-carbide arrester, which just has the duty-cycle rating. A metal-oxide surge arrester using zinc-oxide blocks provides the best performance, as surge voltage conduction starts and stops promptly at a precise voltage level, thereby improving system protection. Failure is reduced, as there is no air gap contamination possibility; however, there is always a small value of leakage current present at the operating frequency.

It is important for the test personnel to be aware that when a metal-oxide arrester is disconnected from an energized line, a small amount of static charge can be retained by the arrester. As a safety precaution, the tester should install a temporary ground to discharge any stored energy.


Polymer arresters are gaining in popularity over the porcelain arresters. When a reclose operation occurs and the fault has not cleared, the arrester is subjected to a second fault current. This second operation can lead to arrester explosion, if the porcelain had already been weakened by the first fault. If the pressure-relief rating of the arrester is exceeded, the arrester may fail violently, since it cannot controllably vent the excess gasses. This type of failure can lead to other equipment being damaged or injury to personnel who may be in the vicinity of the failure. Because of the ability of the polymer station arrester to vent out the side, the housing is not weakened when exposed to the fault current. Therefore, a polymer arrester can be reclosed on multiple times without risking a violent failure. The polymer arresters are less expensive than porcelain arresters and appear to avoid some of the in-service problems, such as moisture ingress, often occurring in porcelain arresters. One manufacturer reports that moisture ingress is the direct cause of failure in 86% of all failures.


Two of the most common tests to perform in the field on surge arresters are the Doble power factor test and infrared analysis. Some manufacturers state that no single test will indicate the complete operating characteristics of an arrester. Reference service advisories from some manufacturers recommend power factor testing and infrared as methods to detect possible problems caused by moisture ingress. Field testing of arresters by power factor, infrared or other methods is used as a reference.

Different models and makes of arresters will have different watts-loss readings. The tester is attempting to identify a variance in the past watts-loss readings. Power factor is not calculated because the current is so small. The arrester should have a visual inspection to detect cracks in the porcelain, abnormal rust staining and any abnormal physical condition of the arrester that is observed. Incorrect factory installation of arrester gaskets has been detected by visual inspection upon receipt of the arresters.


A damaged seal-gapped arrester should be handled with care. Because of increased pressure caused by the destruction of internal elements, a defective arrester could be an explosive hazard. If the decision is made to perform an internal inspection of a failed arrester, ensure the arrester has vented properly. Likewise, do not throw away a defective arrester; the arrester should be properly vented before disposing.

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.