Sulfur Hexafluoride (SF6) Insulating Gas Leak Detection with an IR Imaging Camera

Robert Madding and Robert Benson FLIR Systems, Inc.

ABSTRACT

For years, electric utility transmission thermographers have wanted a camera that could spot leaking sulfur hexafluoride, SF6. As an insulating gas, SF6 is widely used by the electric power industry in high voltage circuit breakers to prevent arcing. Early efforts met with limited success through the use of imagers that required active scanning with infrared lasers. The resulting systems were somewhat cumbersome and required specific conditions which limited their utility. Now there is an IR camera that can spot SF6 in very small amounts and is a completely passive system, requiring no infrared laser but for the smallest leaks. This paper gives a brief history of SF6 as an insulating gas, problems caused by leaking SF6, the theory behind the IR camera, and why it works as well as it does. Additionally, we present some sample findings from both the laboratory and actual operating circuit breakers in high voltage systems that use SF6.

INTRODUCTION

In 1933, just after receiving his bachelor's degree, Ray Herb worked with Glen G. Havens at the University of Wisconsin, Madison on a vacuum-insulated electrostatic generator of the Van de Graaff design, but this device could not be pushed above 300 kV. There was no understanding of the discharges that limited the attainable voltage. Ray therefore decided to try to use high-pressure insulation. Two other graduate students,

D. B. Parkinson and D. W. Kerst, joined Ray in this endeavor. Ray discovered accidentally that the dielectric strength of air could be greatly increased by the addition of carbon tetrachloride, an electronegative gas. According to Ray’s story, he tried other chemicals. When he threw a rag soaked in acetone into the tank, the first spark started a fire. He was easily able to reach 1 MV in a pressure tank 1 m in diameter and 2 m long filled with air and carbon tetrachloride. 1

The basic features of Ray's design, which have been incorporated into all modern electrostatic accelerators, include aluminum hoops surrounding the acceleration tube, a voltage gradient controlled either by corona points or resistors, a rotating vane generating voltmeter, high pressure insulation (originally air and carbon tetrachloride, but later nitrogen and Freon or sulfur hexafluoride).1

Having attended the University of Wisconsin graduate school in physics, Dr. Madding was fortunate to attend a colloquium that discussed Professor Herb’s discovery of carbon tetrachloride (CCl4) as an insulating gas. The story he remembers from 30 years ago was that as a graduate student, Ray was working long hours and quite frustrated with not being able to achieve sufficient vacuum to keep the Van de Graaff generator from arcing. The accidental discovery mentioned above occurred when he was checking the system for vacuum leaks. In those days, one squirted acetone on suspected areas and watched the vacuum gauge. If the pressure dropped initially due to the acetone liquid temporarily plugging the leak, then rose rapidly as the high vapor pressure acetone liquid evaporated, you had your leak. Apparently, when Ray tried this, he accidentally picked up a squirt bottle of CCl4. Not only did the pressure drop and then rise, indicating a leak, the system began to operate at higher voltages without arcing. Mystified by this event, Ray began experimenting with different gases as insulators. This eventually evolved into sulfur hexafluoride as the premier insulating gas, widely used today to insulate and prevent arcing in high voltage switches and circuit breakers.

In addition to electrical insulation, SF6 is used as a filler for bladders in athletic shoes, tennis balls, soundproof windows and tires. It is also used for the ultrasound measurements of tumors, as well as retinal eye repairs in the medical field. The US Navy also used SF6 as a propellant component in the Mark 50 Torpedo. It is also used as a cover gas in die casting to prevent oxidation of magnesium, and in the electronic industry for chip manufacturing2. According to reference 2, published in 2000, US electric utilities used over 1.5 million pounds of SF6 as refill for leaks. At the current price of about $10/lb, this amounts to $15,000,000 annual cost to our electric utilities just to replace leaking SF6. Not to mention the reliability costs associated with potential downtime, outages and expensive repairs.

As a greenhouse gas, SF6 is 22,200 times more potent than CO2 according to reference 3. According to the Intergovernmental Panel on Climate Change, SF6 is the most potent greenhouse gas it has ever tested. There is no doubt, from both a cost and environmental perspective, that finding and fixing SF6 leaks is in the best interests of our country and the planet.

Recognizing the cost of SF6, the environmental concerns and the reliability risk of SF6 leaks in high voltage equipment, electric power utilities and independent companies have spent years and invested thousands of dollars into developing technologies that can quickly, reliably and safely detect SF6 leaks. Power companies know when they have leaks as the pressure drops. And a one or two pound pressure drop on an 80 pound system can trigger an alarm, causing the system to default to an open condition. This creates reliability headaches, requiring power to be rerouted, and can even result in electric power outages.

Probes, or sniffers, work with limited success as the probe must be close to the leak to detect it. With 500 KV switchgear, this can be problematic. Low voltage compressor cabinets are a prime target for sniffers, as they are safe and accessible. But a big problem is often the gas floods the cabinet, and when the operator opens it, the sniffer is overwhelmed with gas. It can take considerable time to clear the enclosure and sniff individual components to find the leak.

IR camera technology has been implemented with limited success using active scanning technology. Here, the IR camera system emits a laser beam at the same wavelength as the absorption band of SF6, 10.6 micrometers. The camera is designed to receive a reflected beam and display an IR image. When the laser beam intercepts an SF6 cloud it is attenuated, twice normally, once on the way out and once on the way back. The problem with such a system is there must be something to reflect the beam. Pointed at the sky, nothing returns, and you cannot detect an SF6 cloud. The system was also very large, not very simple to operate, and prone to breakage.

Recently a new IR camera based on a robust design for military applications has proven quite adept at finding SF6 leaks. It is small, quite portable and extremely sensitive to SF6. The camera is completely passive and can find leaks as close as a few feet and as far away as tens of yards. A competent operator in a substation or high voltage yard won’t miss much, if anything with this camera.

HIGH VOLTAGE SUBSTATION EXAMPLE APPLICATIONS

Figure 1 shows a picture of the SF6 camera in operation at a 500 KV substation. Manufactured by FLIR Systems, Inc. the camera’s official name is the GasFindIR LW, as it finds gases with absorption bands in the long-wave portion of the IR spectrum. (There is a sister camera, the GasFindIR which works in the mid-wave band of the IR spectrum and is useful for detecting volatile organic compounds (VOC) gases.)

Based on laser rangefinder measurement, the rupture disks are 37 feet from the IR camera. Four of these 5 inch diameter disks can be seen in Figure 1, two on each tower. Figure 2 shows a close up image of one of these rupture disks with associated GasFindIR IR image. In a still picture, the leak is very difficult to see, so we highlighted the leaks in the black and white images for this paper. This helps compensate for the frozen image perspective necessary in the written document. With the live IR camera or its recorded video, the leaking gas plume, though small, is easy to spot as its motion gives it away.

Documentation with a digital visual camera is an important complement to the GasFindIR video. There are numerous good digital cameras available. We used a Nikon S10 Coolpix digital camera with 6 megapixel resolution, 10X zoom and vibration reduction. It is pocket size for a large pocket. The hand-held photos came out surprisingly well. We used an Archos digital video recorder to capture the standard video output of the GasFindIR.

Figure 1. GasFindIR LW viewing rupture disks on 500 KV live switchgear. Inset at right shows the GasFindIR mounted on tripod to exploit high sensitivity mode.

There was a light breeze when we found this leak blowing about 5 to 10 mph. The wind was slightly up-wind and cross-wind from the leak, somewhat in the direction of the arrow in the visual image in Figure 2. The breeze did not hamper our seeing the leak. It did cause the SF6 to disperse perhaps more quickly, making finding an individual still frame representative of the leak difficult. There is more gas than meets the eye in the image in Figure 2.

Figure 2. SF6 leak from rupture disk on 500 KV switchgear at 37 feet distance. Note the corrosion around the rupture disk bolts. In the video one can easily detect the leak coming from the bolt in the nine o’clock position.

Finding leaks in areas such as this would be very difficult and time consuming with hand held “sniffer” devices, as access is difficult and the power would need to be removed from the circuit for safety. With the GasFindIR LW, we were able to safely survey and document the entire substation in less than two hours with everything remaining fully operational.

Areas of a substation that have spare SF6 tanks strapped to the structure are places of “low hanging fruit” for SF6 leak detection. The electric power company usually knows it has leaks, as the pressure is closely monitored. For reliability, they maintain a state of readiness to replace leaking gas. In addition to rupture disks, compressor cabinets are a good place to look for leaks.

Figure 3. Using the GasFindIR to find and document leaks in an SF6 compressor cabinet.

Figure 4 shows a leak near a pressure switch. Though we were not authorized to perform repair work, we believe a simple tightening of a fitting could repair this leak. The gas is coming from one of the connections on the tee fitting just above the pressure switch. The opening to this cabinet was downwind from a 15 to 20 mph breeze. Even on the lee side, the wind caused a lot of swirling inside the cabinet. However, the leak was still quite obvious on the video. We also noticed a periodicity to this leak, not exactly puffing, but sometimes it appeared to leak more strongly than at other times. This is the area the GasFindIR was viewing in Figure 3.

Figure 4. SF6 leak above pressure switch in compressor cabinet. This cabinet services a 230 KV capacitor bank.

Other compressor cabinet opportunities are shown in Figures 5 and 6. This cabinet had two leaks, the larger being a cracked bellows, Figure 5. The cracked bellows leak was large enough to be visible in the normal mode. The other leaks were easily detected in the high sensitivity mode, but not initially so in the normal mode. High Sensitivity Mode, (HSM), is a newly developed feature specifically designed to find small gas leaks in the GasFindIR LW. It uses an adaptive motion filter that brings the leaking gas into plain sight and makes leak recognition far easier. After we found them, we tried the normal mode and could sometimes see them, though we already knew they were there.

Figure 5. SF6 leaking from a cracked bellows on a governor switch.

When we approached the leaking switch to take a digital photo, we noticed the utility personnel had already found the leak and labeled it. Our host was interested to see how long it took us to find it with the GasFindIR. It took us less than two minutes once the compressor cabinet doors were open. The smaller diaphragm leak in Figure 6 took only moments longer to locate. Of the leaks reported in this paper, these were the only two the electric utility had located.

Figure 6. Diaphragm leak.

Because one is looking for motion when detecting these leaks, having the IR camera stable is imperative. We used a professional photographer’s tripod. It had a regular camera head on it. For future work a video head would work better.

The primary author has years of experience in using IR thermography in substations to look for thermal anomalies, usually hot spots in electrical connections. The approach with the GasFindIR is different. You cannot go fast. You need to set up the camera on the tripod, focus on an area, get a good image and stare at it for a while. Heat is not the key here. Thermal contrast of moving SF6 gas against a stationary background is what you’re looking for.

SUMMARY

The new GasFindIR camera worked extremely well for finding SF6 leaks in both 500 KV and 230 KV substations. The technology has finally evolved to the point where a battery operated, rugged, small and lightweight camera can be easily taken into a substation and can rapidly find the leaks.

We found some operational tricks and environmental conditions that help optimize SF6 gas leak detection in substations:

REFERENCES

  1. Barschall, Henry H.; “Raymond George Herb”; Biographical Memoirs; National Academy of Sciences; http://books.nap.edu/readingroom/books/biomems/rherb.html
  2. Olivier, Jos. G.J., et al; “SF6 from Electrical Equipment and Other Uses”; Good Practice Guidance and Uncertainty Management in National Greenhouse Gas Inventories; http://www.ipccnggip.iges.or.jp/public/gp/bgp/3_5_SF6 Electrical_Equipment_Other_Uses.pdf ; pp 227-241; 2000
  3. U.S. Climate Change and Technology Program; “Research and Current Activities Reducing Emissions of Other Greenhouse Gases”; http://www.climatetechnology.gov/library/2003/currentactivities/othergases.htm