Jonathan Woodworth
As the demand for higher quality power increases, the development of more lightning resistant distribution systems is required. This article is an overview of the challenges and opportunities we now face in the lightning protection of electric power distribution systems. Specific topics are lightning protection of underground circuits, overhead circuits, system equipment, and lines of all configurations.
If you have ever personally witnessed a lightning strike, you can definitely understand how daunting the task of lightning protection turns out to be. In 1979 after just starting in the power industry, I witnessed my first direct strike close by, I said to myself at the time “What kind of fool thinks they can do something about that?” Well, here I am 43 years later still working on it. My first 20 years, in this very interesting area, were mostly about designing surge arresters, while the last 23 have been more about their application. Every IEEE or International Electrotechnical Commission standards meeting I attend, I find myself invigorated by the discussions and always come away with new and interesting perspectives, I hope this article can do the same for you.
For the purposes of clarification, an electric power distribution system in this article refers to systems with voltages above 1 kV and below 40 kV. Many places in the world refer to this space as medium voltage. The systems can be subdivided into overhead and underground. The overhead systems are highly exposed to lightning in certain parts of the world, while the underground systems are only indirectly affected. Both need consideration when discussing lightning protection. Equipment on both systems is typically the primary target of protection; however, the distribution lines can also be a focus of lightning protection efforts, though typically with less commitment. The reason distribution lines are less of a concern is because the insulators are self-restoring after a lightning-caused flashover. Self-restoring insulation immediately recovers its insulative properties and quickly goes back into service after the external flashover with only a momentary outage to the system. The rationale for more concern for equipment is that their insulation is not self-restoring and if it fails due to lightning, it needs to be replaced resulting in the possibility of a long outage. In underground circuits, the cable is often the primary target for protection again because the cable insulation is nonrestorative and always needs repair which includes a long-term outage at considerable cost.
It is often valuable to look back before you look forward, so for that reason a bit of history on protection of power lines may be useful here. The first lightning protection systems were introduced to the world shortly after Ben Franklin invented the “lightning rod” or Franklin terminal as it is often referred to these days. They were used mostly to protect churches, ships, barns, and other important structures. Though not a surge arrester, they are a close cousin. The first lightning arresters were not used on power lines, but instead used on telegraph lines nearly 100 years after the lightning rod was introduced. In 1847, the now famous Princeton University professor and namesake of the unit of inductance, Joseph Henry, wrote an article in the American Journal of Science and Arts, entitled “Wires of the Electric Telegraph.” In the article, he described a simple device that could help protect the lines from lightning strikes and the dangers resulting from a strike. A rendition of the protective device described by Henry in the article is shown in Figure 1.
A very early and perhaps the first patent to use the term lightning arrester was number 29,355 issued in 1860. Of course, it was related to the telegraph since power systems were not invented yet. The first arresters to protect power systems were likely the same as they had been for telegraph lines, however since there was ac power present on the power lines, an ac fault was created each time the arrester operated. This ac power arc, often referred to as a power follow current, was very difficult to extinguish, and if allowed to continue for too many cycles, it would destroy the arrester. Several designs were tried unsuccessfully for almost two decades. In 1887, E.A. Sperry patented a lightning arrester that physically increased the size of the gap immediately after the arrester operated to help reduce the effect of the ac fault that occurred after the lightning had passed through the arrester. Thomas Edison got in on the action with an 1892 arrester patent, but this design also didn’t address the termination of the power follow current after the lightning surge through the unit. Finally in 1907, E.E. Creighton revealed a solution when he applied for a patent while in the service of General Electric that used a complex arrangement of aluminum plates, dielectric fluid, and insulating oil, which solved the power follow current problem for the first time. Some of these electrolytic arresters were used up until the middle of the 20th century.
In 1915, C.F. Frank patented the oxide film pellet arrester. These arresters utilized a series gap and a chamber full of pellets with oxide film on the surface. These pellets had a nonlinear resistance characteristic that would provide some mitigation of the power follow current. The first expulsion-type arrester was patented in 1918, and this type of arrester is still in use today all over the world though they have not been produced for many decades. The first silicon carbide gapped arrester was introduced in 1926 by John Robert McFarlin. This type of distribution arrester was produced until the 1990s, and millions are still on distribution systems protecting valuable assets.
The modern-day metal–oxide surge arrester (MOV) for system protection was introduced first for high-voltage substation applications in 1976–1977 and the distribution version in 1980 (see Figure 2). This solid-state n-type power electronic component completely eliminated the one or two cycles of ac current that always continued to conduct after the lightning surge in the silicon carbide gapped arrester. It only took 90 years to solve that technical problem in the lightning arrester industry. In 1987, the rubber-housed distribution arrester that is produced today was introduced by the Hubbell Company and is the form that now dominates lightning protection options for distribution systems.
The lightning flash is a highly studied phenomena, and there are many critical treatises available for the reader; however, from a distribution system perspective there are a few aspects that should be noted. First, lightning is not evenly distributed around the world. There are many people on Earth that have never seen lightning or heard thunder. The primary source of lightning around the world is cumulonimbus clouds. These clouds are formed in climates that are warm and high in moisture. They form over land or water, but mostly over land. When fully formed, they reach from a few thousand feet at the bottom, up to 40,000 ft in the air. The flat tops we can sometimes see are where they touch the stratosphere.
The lightning flash is not what it seems; it is typically not a simple stroke to Earth. It almost always comes as a series of strikes. The human eye is capable of sensing this multistroke parameter. The next time you are in a lightning storm, notice the flicker of the lightning in the clouds. It is not as obvious if you see a stroke, it only becomes more visible when the stroke is hidden by the clouds. Figure 3 shows the typical lightning flash parameters.
The ground-to-cloud lightning that makes up 20–30% of total lightning strikes are the strokes that challenge distribution power systems. The other intracloud and cloud-to-cloud strokes just light up the sky and make long thunder rolls. An important parameter of lightning is the ground flash density (GFD), which is a measure of how often lightning strikes the ground for a given area. When running lightning studies of power systems, this parameter is quite important in determining the outage rate of a line or area. The GFD of most places on Earth is well-known and is provided by numerous companies. Vaisala Inc. provides an interactive map on their website that shows the GFD of every country and or subregion in the world. If more precise information is required, it can be obtained through their services.
When a lightning flash hits the ground, it is either a direct strike to the distribution system or more likely an indirect strike. The indirect strikes terminate on trees, buildings, or the ground. If the flash is close enough to the distribution line, it can still create a significant surge on the line that requires mitigation. The direct strike will almost certainly cause a surge that will damage equipment if not properly mitigated.
The modern day metal–oxide varistor (MOV) as shown in Figure 2 has been the only style of distribution arrester manufactured in the United States for the last 30 years. Prior to that, the gapped silicon carbide (SiC) arrester had been the primary arrester type for several decades. There are different types and classes of distribution arresters used on the various distribution power systems. For the overhead system, the most common type of arrester is the MOV heavy-duty (HD) arrester that was introduced in the early 1980s. This arrester is designed and configured to protect distribution transformers. This distribution type arrester is by far the model of choice in the United States and world markets.
The very similar, but lower duty, MOV normal duty arrester is not nearly as widely applied to distribution circuits as the HD. However, it is found to be very adequate in the few utilities that use this model. Per the arrester test standard, this model is tested to a charge level 35% lower than the HD model.
The riser pole (RP) arrester is another widely used arrester and applied primarily at the transition poles between overhead systems and underground systems. The RP arrester is typically tested and certified as an HD arrester, but with special selection of MOV disks, better protection (i.e., lower discharge voltage) is offered. This lower discharge voltage is advantageous to underground circuits as will be described later.
The elbow arrester is a special design that is used to protect deadfront switchgear, pad-mounted transformers, underground cable in underground residential distribution circuits, and wind farm/solar farm collection circuits. Since this arrester is only attached to underground circuits where surge currents seldom exceed a few kiloamps, it is tested and certified as a light duty arrester.
The distribution “line arrester” is widely used in areas of higher lightning GFD. This arrester is typically an HD type arrester and is used to reduce the insulator flashover rate of exposed distribution lines. This arrester is often found suspended from the line with a hot line clamp but can also be mounted on a bracket similar to ones used with arresters protecting equipment as shown in Figure 2.
Since arresters are typically installed between the phase conductor and earth, the voltage that appears across the arrester for its life is the phase-to-phase voltage divided by 1.73 (the square root of three). Therefore, a 13.8-kV system typically uses an 8.4-kV maximum continuous operating voltage (MCOV) arrester where the line-to-ground voltage is 8.0 kV. Unfortunately, old habits are difficult to change, and the SiC arresters that preceded the MOV technology did not have an MCOV parameter. Instead, it just had an ac rating that was equal to the highest system voltage that could appear on a distribution system to which it was applied during a fault. So, for the same 13.8-kV system, the arrester rating was 10 kV. When the MOV-type arrester was introduced in the late 1970s, it had two ac ratings, the new MCOV rating and the old SiC rating. This dual rating system carried on for 40 years, and only in 2020 did the arrester standard C62.11 finally put the old SiC arrester rating system away. We now only label and reference the ac rating of an arrester as its MCOV.
The primary purpose of surge arresters on distribution systems is to protect equipment. The way they protect equipment is to clamp the overvoltage caused by a lightning strike. The modern day MOV type arrester does this by lowering its resistance as the voltage rises and conducting the excess current to earth. The protective capability of an arrester is quantified by their protective level (discharge voltage at 10 kA). This protective level is a competitive parameter, and arrester suppliers offer various protective levels with the premier products offering the best levels. When selecting an arrester, it is important to know the level of protection needed so that the risk of failure is reduced. The best 10-kA lightning protective level (see Figure 4) on the market today is about 3.0 x MCOV and are most often used in RP-type arresters. A protective level of 3.3–3.6 x MCOV are used for distribution transformer protection.
A second parameter of importance is the temporary overvoltage capability (TOV). When arresters are applied to distribution circuits that are not effectively grounded, it is possible for the system voltage to rise to high levels should a ground fault occur on the line. If an arrester with a low TOV capability is installed, it is quite possible it will fail. If a higher TOV-capable arrester is used, it will ride-through the event and continue to protect its equipment as designed.
A third parameter considered to be important is in fact not so important. The energy handling capability of a distribution arrester is quantified by its charge transfer rating. When a distribution system is hit with a lightning strike, it flashes over the nearest insulator it encounters unless there is a very close arrester. Because of the inevitable flashover of a nearby insulator, the distribution arrester seldom carries the majority of the lightning strike (IEEE Std 1410–2010). This is why a normal-duty and HD arrester both have excellent service history in terms of lightning-caused failures.
The ground lead disconnector is an important add-on component of a distribution arrester. This component was first introduced in the late 1930s and has become a very important part of distribution system protection. The disconnector is designed to operate only if the arrester has failed. When an arrester fails, it typically becomes a short circuit and if not removed from the circuit will cause the entire circuit to be disabled. So, when an arrester fails short, the disconnector operates and disconnects the ground lead from the arrester effectively removing the faulted arrester from the distribution circuit. Most disconnectors today contain a small explosive charge of black powder used to eject the ground lead away from the arrester to attain the disconnection function. Unfortunately, black powder is a hazardous material and must be handled as such during transport on public highways. Manufacturers, however, have fortunately learned how to render the black powder nonhazardous and these new arresters (since 2005) do not need to be transported as a hazardous material. But there are still millions of hazardous arresters in the field that if transported on public highways should be handled as a hazardous material.
When distribution power systems came into existence in the late 1880s, power engineers already knew they would need to manage lightning from the experience of the telegraph engineers. Can you imagine the steep learning curve they were on in these early power system years? Surely, they learned quickly that the voltage levels that appeared on the line from the lightning were so high it could not be insulated at a high enough level to avoid an insulator flashover. Also, by the 1880s, they had learned that some insulation would recover from a flashover, so it became known as self-restoring insulation. Insulation that does not recover and is permanently damaged was referred to as nonself-restoring insulation and these terms are still used today. Examples of self-restoring insulation is the post insulator found at the pole top that insulates the line from the pole. An example of nonself-restoring insulation is the oil and paper insulation found inside most distribution transformers. Since most porcelain insulators are self-restoring, protection of these components is limited and usually only found in areas of high lightning occurrences. A great deal of effort is put forth to protect nonself-restoring insulation since the consequence of failure is so much higher.
Insulators have power frequency and impulse withstand ratings. The power frequency rating is what determines what ac voltage can be applied long term. The impulse withstand rating indicates at which impulse voltage it will cease to be an insulator and flashover. Lightning protection of power systems is focused on this impulse rating that is either called its basic lightning insulation level (BIL) or critical flashover voltage (CFO). The BIL rating is the highest voltage that will never cause a flashover, whereas the CFO is the lowest voltage that will cause a flashover 50% of the time when applied. When working with self-restoring insulation, the CFO of the insulator is used. If we are working with nonself-restoring insulation, it is important to never flashover so the BIL of an insulator is used. Another way to look at it is that equipment insulation usually has a BIL rating and standard insulators more often have a CFO rating. Typical distribution transformer BIL ratings are 95–200 kV and typical line insulators range from 95–250 kV CFO. It is with these parameters in mind along with the system voltages that surge arresters are selected to offer protection from lightning.
The margin of protection analysis as shown in Figure 4 is a commonly used method to quantify the level of protection that surge arresters offer to nonself-restoring insulation. It is actually a very simple form of an insulation coordination study. The blue discharge voltage curve is produced by impulsing the arrester at different current amplitudes. The 10-kA lightning protective level is the voltage measured across the arrester when impulsed at 10 kA. This is the primary protective level used to evaluate a system’s level of protection. This level is typically 3–4 times the MCOV rating of an arrester. In other words, if the MCOV rating of an arrester is 8.4 kV, the arrester’s lightning protective level will be roughly 30 kV.
Since insulators have three fundamental impulse withstand characteristics, and arresters have three fundamental protective levels, these three parameters are compared to each other and quantified as a percent of the insulation withstand levels. Typically, a margin of protection of greater than 15% is acceptable. The margin of protection offered to distribution transformers today are very good if the arresters are properly mounted near the transformer. For example, on a 13.8-kV solidly grounded system with a design BIL of 95 kV, the margin of protection level is 95 kV/30 kV = 216% which means the insulation curve is 2.16 times higher than the arrester discharge voltage. The distribution circuit that is most difficult to protect these days is the 35 kV impedance grounded underground circuit particularly if long lead lengths are used at the riser pole.
The primary focus of lightning protection of distribution systems is on equipment found on the system because most insulation associated with it is nonself-restoring. Even if there is nearby self-restoring insulation, it is typically ignored. A case in point is the external insulation of a transformer bushing. The internal insulation of a bushing is always the focus of its protection since the external insulation is much less likely to fail even if it does flashover. It is easy to assume that the protection of the distribution transformer has been an issue since the first power systems. It is very interesting to see distribution transformers that are clearly 50–70 years old in rural parts of the United States with equally old surge arresters attached. One has to wonder if these transformers made it this long because of the surge protection, or because of very robust insulation.
There are a few general rules to follow when applying arresters on distribution equipment:
Distribution transformers as shown in Figure 5 have long been the main target for surge arrester protection on distribution systems where lightning levels merit the investment. There are many locations on the west coast of the United States and in northern Canada/Alaska where lightning protection is not necessary and not applied. In Del Norte County, California, where the GFD is 0.03 events/km2/year, the value of an arrester is nearly zero and you will not find arresters on distribution transformers. On the other hand, there are areas of Florida that have a GFD as high as 40 events/km2/year, and in this case a single transformer-mounted arrester can save the transformer from 10–20 times in its lifetime. In this case, the value of the arrester is very high and regularly used.
Reclosers and breakers that are often used for sectionalizing the distribution system, and mounted out on the lines, are at the same risk level as distribution transformers. If they are in a distribution substation, the risk is much lower. To protect those mounted on the line, arresters are mounted on both sides of the equipment as a lightning surge could occur in either direction on the line. This bidirectional issue leads to double coverage if the device is in a closed state but is necessary to provide the protection if the equipment in an open state.
Pole-mounted shunt capacitor banks are subjected to the effects of lightning just as all other equipment on the line. What makes them a bit different is that the arrester serves two purposes. They are used primarily to protect the capacitor switch insulation as the capacitors themselves do not need lightning protection. The second purpose of mounting arresters on these banks is to again protect the switches but this time from switching surges caused by the capacitor being switched in certain conditions.
Underground systems can serve both residential and industrial circuits. The very common underground residential distribution systems typically consist of a point on the overhead network where it transitions to underground. This point is referred to as a transition pole or riser pole. At this point, the system is converted from air-insulated conductors to underground cable. The cable carries the power to transformers that are mounted on a concrete pad where the cables connect to the windings through elbows and bushings.
Lightning protection of these systems focuses on the cable insulation as well as the pad-mounted transformer insulation as shown in Figure 6. Both applications have nonself-restoring insulation and need special attention. The reason these systems need special attention is because of the traveling-wave phenomena that occurs with lightning surges on power systems.
When lightning strikes a line directly or nearby, a surge is induced onto the line at the point of the surge. This surge does not appear simultaneously on all parts of the system but instead it travels from the strike point in both directions at a very fast rate. For overhead lines, the speed of travel is very near the speed of light and, for underground systems, about one half the speed of light. So, on an overhead system it travels about 900 ft/µs and on an underground system ∼450 ft/µs. Not only does the surge travel at a finite speed down the line, if it should run into an open point (or point of very high impedance), it reflects all or some of the surge back up the line from where it came. The reflected surge and the still incoming surge interact with each other causing an increase in the amplitude of the surge. This occurrence is referred to as the voltage doubling phenomenon even though it may sometimes increase as high as three times. For underground systems, this voltage-doubling phenomena is an always-present issue to mitigate.
Because of the voltage doubling issue, the arresters used at the point of entrance into the underground circuit are typically chosen to clamp the surge voltage at its lowest possible level. This lower discharge voltage arrester in some cases can adequately protect the open point transformer and cable, but most of the time the riser pole arrester at the underground residential distribution entrance is not enough protection. When the riser pole arrester is not adequate, the next level of protection to install is a deadfront elbow arrester at the endpoint of the underground line. In almost all cases, an arrester at the riser pole and at the open point will provide all the needed protection of the pad-mounted transformer and connecting cable.
The distributed energy resource collector circuits on large photovoltaic or wind farms are a unique distribution system that has emerged in the past 20 years. This circuit can be difficult to protect from lightning when there are overhead and underground sections to the circuit. They seldom exceed a 35-kV system voltage so are often considered a distribution system from a lightning protection perspective. At the present time, 35-kV distribution collector circuits fed by dc-to-ac inverters are very difficult to protect. If this type of circuit is isolated from the grid due to a fault in the circuit, and the power sources are not shut off soon enough, large overvoltage swings can occur that may damage or completely fail arresters. It is important to use grounding transformers or inverters that can instantaneously shut down to not overstress the arresters. A general rule of thumb for protecting distributed energy resource circuits is that arresters need to be used at all riser poles, at endpoints, and in the substation.
Protecting distribution lines as opposed to distribution equipment is a very different task. In most cases, the lines are insulated with air insulators that fall in the class of the self-restoring insulation. In North America, northern electric utilities tend to do less line protection than southern utilities because the GFD is lower in the north and higher in the south. Lightning protection of lines comes in three forms for distribution systems: arrester installation is the foremost application, overhead shield wires is the second most common, and installing underground systems is a distant third means of lightning protection. Figure 7 provides an idea of the many different configurations and methods used to protect distribution lines.
In Figure 7, Config 1 is the most common line type worldwide. It is also the hardest to protect from lightning because all three phases are directly exposed to lightning unless a double shield is installed. The easiest distribution line to protect is Configs 3 or 4. In this configuration, either a shield wire can be used as shown in 3s or 4s, or a single arrester on the top phase can be used as shown in Configs 3a and 4a. Isolating the down ground conductor as shown in Config 4sig with an isolated down ground and shield wire is also an excellent tool for improving the performance of the line to lightning.
Figure 8 shows the possible improvements in a system outage rate by adding arresters, a shield wire, an isolated down ground, or any combination of such. With just 125-kV CFO insulators on this configuration, the annual outage rate is the highest for this area of analysis at 9.90 per 100 km/year. The second column shows the configuration that was evaluated. Each configuration from top to bottom show the possible improvement in the outage rate. With arresters on all phases of all poles, the line becomes lightning proof. Similar results were achieved when analyzing all configurations shown in Figure 7.
For many years, it has been the practice of electric utilities to install arresters every fourth or fifth pole in areas where there is little tree or structure protection of the lines. It is not known where this ineffective practice originated, but it offers a false sense of lightning protection. IEEE 1410–2010 discusses this false sense of security in detail. It also shows in detail that even skipping one pole will result in a 100% outage rate from a direct strike. There may be an improvement in the indirect outage rate, but this is very small. Users of this strategy should consider an alternate arrester deployment for systems other than four per mile.
Spacer cable circuits as shown in Figure 9 are a space-saving, tree-resistant configuration for distribution lines. It is very effective in achieving these two improvements. However, if improved lightning protection is desired, this is not a means to that end.
A spacer-cable design has a messenger wire that may appear to act as a shield; however, due to the very short spans between spacers, it is not an effective shield wire. Back flashovers will likely occur because of the low insulation level of the spacers and the short distance from any direct strike to a spacer. Arresters installed only at the pole offer no protection in reducing midspan flashovers of a spacer. The only effective way to improve the lightning outage rate of a spacer cable circuit is a shield wire of the circuit and ample CFO of the spacer cable at each pole.
Since surge arresters are the predominant means of protecting distribution transformers from lightning, this analysis is focused on the value of a surge arrester. The use of surge arresters on power systems is so commonplace that sometimes it is not clear why they are really needed and if they are worth the cost associated with their purchase and installation. If you are one of the skeptics, read on.
In lightning prone regions of the world, it is almost a universal rule to install lightning arresters on distribution transformers to protect the high-voltage windings. For regions where lightning is a rare event, the transformers remain unprotected. Certainly, this author does not recommend arrester installation at locations where it is not well justified, but sometimes knowing the locations where they are needed is not obvious. If you wish to analytically determine the value of a distribution arrester, here is a method you can use. To determine the value of an arrester in this case, determine the cost to replace unprotected equipment in the absence of the arrester.
To make this calculation possible, let’s make a few simple assumptions:
First, a direct strike to the line where a transformer is mounted and there is no arrester installed will always result in a transformer failure. Second, the number of direct strikes to a line is a function of the line height, width, length, and GFD. The collection area can be 1–10 spans because beyond that the surge voltage will be reduced by corona and not damage the transformer.
As you can see in Figure 10, a transformer that initially costs US${\$}$1000 installed in a low lightning density area (GFD =1) only saves the unit 1.089 times from lightning in 18.4 years which yields a lifetime savings of US${\$}$1089. This value barely exceeds the cost of the transformer bringing to question if the use of an arrester in this case is the right choice. If the GFD is higher however, the arrester value increases. An arrester installed in an area with a GFD of 12 will save the transformer owner US${\$}$13,074 over the life of the arrester. Here, it is an easy decision to install arresters.
The energy and cost of protecting equipment on distribution systems and reducing lightning induced outages on the lines themselves is a daunting task. Thousands of power engineers from Thomas Edison to today, have devoted countless hours and even careers on this task. This article, I hope, offers a glimpse into what this task involves. However, as power systems evolve and distribution of power becomes unnecessary, this task may disappear. Until then, enjoy it and hopefully this article will empower you to do it better.
I would like to thank Deborah Limburg Cofounder of ArresterWorks for her significant editing and technical assistance in the writing of this article.
“What is the value of a distribution arrester.” ArresterWorks. Accessed: 2022. [Online] . Available: www.arresterworks.com/arresterfacts/arresterfacts.php
“History of surge arresters.” ArresterWorks. Accessed: 2022. [Online] . Available: www.arresterworks.com/history/articles.php
Jonathan Woodworth is with ArresterWorks, Olean, NY 14760 USA.
Digital Object Identifier 10.1109/MPE.2022.3230890