Metal oxide surge arresters (MOSAs) have high reliability. They can be expected to function as intended almost indefinitely – assuming they are well designed, properly manufactured and operated within their specified range of applied voltage, temporary overvoltage, surge magnitude and surge energy, (and also have not been damaged by some external force). However, experience suggests that arresters do occasionally fail in service, usually meaning that the arrester has suffered an internal short circuit, placing a fault (most often a line-to-ground fault) on the system.
Almost invariably, this results in operation of some circuit-interrupting device (whether circuit breaker, recloser, fuse, etc.), causing disruption of electrical service. If the arrester is equipped with a disconnector (e.g. as most distribution arresters installed in the U.S.), the disconnector should operate when the fault current flows following an arrester short circuit. This isolates the arrester either from the ground or from the line, depending on where the disconnector is installed. Then, upon re-energization, the failed arrester is no longer in the circuit electrically and service can continue – but now with some reduced level of surge protection. In the case of a station arrester, a failure will typically cause a system lockout until either that portion of the substation can be by-passed or until the arrester is physically removed from service.
While it is desirable to avoid arrester failures at all times, the consequence of a station arrester failure is typically much more severe than that of a distribution arrester. Often, a considerably greater disruption of service would accompany a station arrester failure. Moreover, the cost of replacing a station arrester is orders of magnitude greater than when replacing distribution arresters. This edited contribution by Steve Brewer of Hubbell Power Systems reviews current technologies for on-line monitoring of the condition of surge arresters, with a view to anticipating problems before they evolve into failures.
Goals of Field Tests
There are three primary reasons that a program of field-testing would be undertaken by end users:
- Determine if an arrester is near end of life;
- Predict when end of life will occur;
- Evaluate if the arrester is still providing protection to the insulation.
The first two goals can reasonably be realized but the third is not practical in a field environment. Fortunately, with today’s generation of MOV arresters and the design tests that must be passed, there is virtually no duty that can impact the arresters ability to protect.
To better understand end of life testing it is useful to have a working knowledge of how a surge arrester behaves under power frequency conditions – not only when the arrester has aged but also when it is tested in the factory prior to shipment. Under normal line to ground power frequency voltages, surge arrester current is nearly all capacitive. The relationship between voltage and current on a new surge arrester is seen from the following chart.
As voltage on the arrester increases, the current waveform becomes much more resistive in nature. This characteristic is also true if the voltage remains constant but internal temperature of the arrester increases for some reason. Monitoring the resistive component of current over time is key in efforts to evaluate the arrester’s internal state and predict ultimate failure. The MOV blocks exhibit a negative temperature coefficient and, as such, the resistive component of current will increase as block temperature increases. The two figures below demonstrate these characteristics.
Almost without exception, an arrester failure results in complete short-circuiting of the unit inside its external housing. While it is possible that an arrester could flashover externally without its external air gap being compromised (e.g. by animal, bird or vegetation), typically the MOV blocks prevent the terminal-to-terminal voltage reaching a point, under any circumstance, where such external flashover could occur. In all scenarios, failure ultimately occurs as a result of a dielectric breakdown whereby the internal structure of the arrester has deteriorated to the point where it is unable to withstand the voltage applied to its terminal – whether this is normal system voltage, temporary power frequency overvoltage (e.g. following external line faults or switching) or lightning or switching surge overvoltages. The possible ways in which an arrester could reach such a state are varied even though relatively rare.
Perhaps the most common root cause of failure is moisture ingress to the interior of the arrester, implying that it was not well designed and/or not properly manufactured and/or damaged by some external force – all resulting in a compromise to the sealing system.
Temporary Overvoltage (TOV)
Under normal operating conditions (arrester energized at its maximum continuous operating voltage or MCOV), the temperature of the MOV blocks rises just slightly above ambient to the point where heat being generated by the blocks is perfectly matched by that being dissipated by the arrester. If the power frequency voltage across the arrester increases (e.g. as a result of a system disturbance, fault or switching operation), the MOV blocks conduct more current and begin to rise in temperature. If the overvoltage is of a sufficient magnitude, the heat generated by the MOV blocks will always be greater than the heat that can be dissipated and a potential thermal runaway situation will result.
MOV Block Ageing
In the early days of MOSAs, MOV blocks from all manufacturers exhibited some ageing, whereby their power dissipation at a given voltage increased slowly but continually over time. The effect on arrester performance would be similar to that described above for TOV, i.e. after time in service the power (heat) generated by the blocks would be similar to that resulting from a TOV when the blocks were new. At a later point, heat generated would be equivalent to that generated by a higher TOV when the blocks were new. Ultimately, the heat generated could reach a level where no stable operating point could be maintained.
With significant improvement in MOV processing technology over the past 30 years, however, the MOV blocks produced by major manufacturers now exhibit a characteristic whereby power dissipation actually decreases over time at any given voltage. They become more rather than less thermally stable with time, and arrester failure due to block ageing is unlikely.
Thermal Runaway Resulting from Surge Duty
Surge duty referred to here is that resulting from relatively high current surges due to lightning, switching of long lines or capacitor banks. Some of these surges may have very high amplitudes but relatively short duration (e.g. lightning surges), while others may have much longer duration but with significantly lower amplitude (e.g. switching surges). All have a certain charge content that, when passed through the MOV blocks, results in a certain amount of energy being absorbed by the blocks. This absorbed energy results in an almost immediate (i.e. adiabatic) heating of the blocks and if the input of energy is too high, temperature rise of the blocks could be such that the arrester is pushed into a thermal runaway condition.
Damage to MOV Blocks from Surge Duty
One manifestation of energy absorbed by MOV blocks is rise in their temperature. However, if the energy is of a sufficiently high magnitude and is deposited into the blocks in a relatively short period of time, the blocks might become physically and irreversibly damaged. For example, the thermo-mechanical shock imparted could cause blocks to crack into pieces. In some cases, a block can be punctured in a localized area, partially or completely through its body. In yet other cases, a pinhole type failure can occur at the edge of the block, sometimes causing material to be removed from the outside peripheral surface. Typically, each such form of damage is accompanied by degradation of the blocks’ electrical integrity – either in their inability to sustain another energy event without complete electrical breakdown or reduction in their ability to support normal operating voltage. Both outcomes could, sooner or later, result in complete arrester failure.
The figure below displays the typical failure modes for MOV surge arresters in a flow charge format. In all cases, final failure mode will be increased conduction, ultimately leading to a complete dielectric breakdown. Therefore, it is clear that measuring temperature and/or resistive component of current will be key to effective monitoring of the internal condition of arresters.
On-Line Monitoring Methods
Utilities generally prefer on-line measuring methods since this does not require a substation to be taken out of service to perform evaluations. For all methods discussed, it is important that a baseline measurement be established and data taken over time to identify trends that may indicate increase in resistive current. As far as what measurements are acceptable, values will vary between manufacturers. As such, the supplier of any arrester being evaluated should ideally be contacted for their input on appropriate limits.
Increases in power losses that occur in a degrading arrester result in increasing temperature of the MOV blocks in all or at least part of the arrester. Since MOV block temperature is not a characteristic that can be measured directly in a conventionally constructed arrester, many utilities use indirect means such as infrared thermography. Heat generated by the MOV blocks inside an arrester is dissipated to the outside by conduction, radiation or convection. For sufficiently high heat generation, the external arrester housing can become elevated above ambient temperature. A heated surface emits radiation that can be detected by an infrared sensor, allowing an estimate of surface temperature. The device used for this could be an infrared thermometer that registers a temperature of the spot or small area on which the device is focused. Or it can be an infrared camera that captures a thermal image over a wider area, thereby providing a temperature profile over a portion or entire length of the arrester. This non-contact method of measurement clearly has appeal since it can be conducted from a safe distance, quickly and at almost any time. However, the technique is not without potential for errors and need for careful interpretation. The amount of radiation emitted from a heated surface depends on emissivity of the surface. A shiny surface with low emissivity will emit less radiation at a given temperature than a dull surface with high emissivity at the same temperature. The ‘apparent’ temperature registered by the infrared device could therefore be quite different, even though actual temperature is the same in these cases. Other problems can occur with reflected radiation from nearby sources or from the sun. These are not insurmountable but need to be taken into account. It is advisable that individuals taking such measurements have training in taking and interpreting of thermographic measurements.
Even with ‘accurate’ measurements of arrester temperature, it is not always easy to determine their significance. As for off-line techniques, there is no single hard-and-fast threshold delineating a ‘good’ from a ‘bad’ arrester. Judgment needs to be employed. Without knowing details about the thermal properties of the arrester ‘package’, it is not necessarily straightforward to translate temperature measured on the outside of the housing into temperature of the MOV blocks inside. Different types of arrester design and construction may also differ in this regard. Simply because an arrester has some region that appears hotter than others is not necessarily an indication that a problem exists or is emerging. For example, on high voltage multi-unit arresters, it is quite likely that temperature near the top will be higher than at lower points because there is inevitably some degree of non-linearity of voltage distribution along the MOV blocks column (even with grading rings employed). Blocks near the top of the arrester are typically being stressed above the average for the entire unit and likely to be running somewhat hotter than the overall arrester average.
Arrester manufacturers typically advise a utility to take thermal profiles of arresters on all three phase at a particular location, over a short span of time to avoid issues with changing ambient conditions and then to look for differences between profiles. It is low probability that a problem of degradation exists in all three arresters and even lower probability that degradation is progressing at exactly the same rate in all three arresters. If a relatively modest variation in temperature (e.g. less than 5-10°C) exists along the arrester but is the same for all three, then one can be assured that there is not an emerging issue. The key is examining for significant differences, e.g. an arrester having a distinctly different temperature profile versus the others. Even then, this does not always mean that the arrester in question is ‘bad’ but rather suggests that it be monitored regularly to see if the differences increase.
Resistive Component of Current
While it is impractical to monitor arrester power losses in the field and MOV block temperature cannot be directly measured, a parameter that can be monitored directly and continuously is arrester current. Such measurement requires that the arrester be isolated from ground by placing it on an insulating sub-base and then connecting the isolated base of the unit to ground via a conductive path that bridges the sub-base. A device connected in this conductive path can then be used to obtain a signal that is proportional to arrester current. This device could be a resistive shunt of several hundred to a few thousand ohms (e.g. a 1 kΩ shunt would provide a 2 volt signal for an arrester current of 2 mA). Alternatively, a suitably sized current transformer could be placed around the conductor (a current transformer with 1:100 turns ratio and a 10 kΩ burden would provide a 200 mV signal for an arrester current of 2 mA). In either case, protection needs to be provided (e.g. in the form of a small MOV across the terminals of the measuring device) to limit output voltages if and when the arrester is called upon to limit surge overvoltage – which could result in short duration impulse currents of many kA.
Most arrester manufacturers offer devices for measuring arrester current. These widely-used and relatively simple devices provide measurement of the total current flowing through the arrester, typically in terms of peak magnitude or peak/√2 (i.e. to provide a quasi-rms value).
The above might suggest that monitoring the resistive component of current would provide a better indication of arrester health than monitoring total current. This is true, but while total current can be directly and easily measured on in-service arresters, it is very difficult to extract the resistive component. As indicated, the peak of the resistive component occurs at the instant of the voltage peak across the arrester. If accurate information regarding voltage waveform were available, it would be possible to determine the value of the total current at the instant of voltage peak, thereby yielding the peak value of the resistive component. While this is possible in a controlled laboratory setting, it is virtually impossible in the field because obtaining an accurate measure of the waveform of the voltage applied to the arrester being monitored would require an expensive and extremely accurate voltage divider with zero phase-angle error.
3rd Harmonic Measurements
If at normal operating voltage an arrester conducts increasing amounts of current because of degradation of its volt-amp characteristic, the resistive component of the arrester current is a much more sensitive measure of the degradation than is total current. This is because total current is dominated by the significant capacitive component until the arrester has degraded substantially. However, the resistive component is not readily extractable from total current – especially in the case of in-service field measurements. Simulation and experimental data demonstrate that the 3rd harmonic component of arrester current provides a reasonable indirect means of estimating the resistive component. It can therefore serve as a means to monitor degree of degradation of the arrester’s conducting properties. In the field, measurement of the 3rd harmonic component of current can be undertaken quite easily on energized arresters and several instruments are commercially available for this purpose. (In considering a 3rd harmonic monitoring system, an important factor to take into account is system frequency since a device designed for a 60 Hz system will not function correctly on a 50 Hz system.)
The best monitoring method would appear to be direct measurement of internal block temperature since this would be the best reflection of change in the resistive component of current. Surge arresters with built-in thermal monitoring devices would not require use of insulating sub-bases and would not be as susceptible to the impact of external leakage currents. However, with current arrester technology, this monitoring method remains unattainable.
Power Loss at MCOV
In the case of almost all underlying causes of degradation in an arrester’s performance, the resistive component of current will increase over time. With steady applied voltage, this results in an increase in power loss of the arrester. However, measurement of arrester power loss in service is not easy and requires separate measurements of voltage applied to the arrester and of arrester current – these then being multiplied and integrated over time. For accurate determination of power loss, there must be no phase shift errors in the voltage and current measurements. That is, the signal that represents the applied voltage must be exactly 90 degrees out-of-phase with the capacitive component of the signal that represents the current. Even small phase errors can result in large percentage errors in power loss measurement. Moreover, measurement of voltage requires expensive equipment such as high voltage divider or potential transformer. For these reasons, measurement of power losses on assembled arresters at normal operating voltage can reasonably only be done in a high voltage laboratory or along a manufacturer’s assembly line.
A question often asked is how often an arrester should be evaluated. Since the rate of deterioration of a surge arrester (if any) can vary greatly, it is recommended that testing be performed as often as possible, given maintenance budgets or other constraints. It is also important to maintain records over time and compare current information with baseline data.
It is possible to damage an MOV arrester from an overvoltage and this can occur in the course of field testing. For example, high potential testing is not recommended for MOV surge arresters due to the possibility of accidental damage. Finally, in order to ensure that only internal arrester currents are measured, it is also important to be sure that the surface is clean and dry.