Electric utilities are under increasing pressure to address the issue of SF6 (sulphur hexafluoride) leaks from substations. SF6 is widely used as an insulator in high voltage circuit breakers and switchgear, but it is also a greenhouse gas. In the USA, the Environmental Protection Agency has recently begun asking utilities for a voluntary reduction in SF6 emissions. There is also the added incentive that SF6 lost through leakages is expensive to replace, at $1500 per bottle. In addition frequent topping up of leaking systems increases the risks of human error as well as particle and moisture ingress.

The conventional technology for locating SF6 gas leaks, halogen detectors and soapy water, is time consuming and also entails the costly and burdensome effort of de-energizing substation equipment.

Now a new leak detection system is available which gets round these problems. Developed by Laser Imaging Systems of Punta Gorda, Florida, with support from EPRI, the US collaborative utility research organization, the new system uses a CO2 laser camera. Called GasVue, it “represents a significant breakthrough in SF6 leak detection – without the need to take equipment out of service,” claims EPRI.

The GasVue laser camera works by illuminating the scene with infrared energy at a wavelength that SF6 strongly absorbs. An infrared detector then forms an image on a screen, which can be recorded on a VCR. The leaking SF6, which is normally invisible, appears as black smoke against a lighter background. The laser camera is sensitive to very small leaks.

Utility experience positive

Over the past 18 months, over two dozen utilities have put the GasVue camera through its paces in spotting SF6 leaks during demonstrations at their substations. Experience to date has been very positive. For example, Jim Vandegraft of Illinois Power describes the camera as being “precisely the type of technology we need to support efficient substation maintenance work,” while Eskom of South Africa has reported on its trials in some detail* and concludes that “the camera was highly effective in detecting and locating SF6 leak sites.”

The demonstrations at Eskom, which were actually done by people from US utility Public Service Electric and Gas who are trained and experienced in operation of the camera, took place in September 1998.

EPRI member Eskom, responsible for about 98 per cent of the electricity in South Africa, makes extensive use of SF6 as the primary insulation in GIS and outdoor circuit breakers.

Based on the successful demonstration of the camera, Eskom decided to do a complete scan of its Invubu GIS. This substation has been completely refurbished to repair extensive SF6 leak sites.

In the Eskom trials the camera was tested at four substations:

  • Craighall 275kV GIS (which has a complex configuration of SF6 ducting with flanges in awkward locations and known leaking compartments);
  • Pieterboth 88kV outdoor substation;
  • Nevis 275kV outdoor substation; and
  • Koeberg 400kV GIS (which has known leaking compartments but from unknown locations).

    The latter GIS is a critical link between the Koeberg nuclear power station and the Eskom grid and demonstration here provided the basis for a planned future scan of the entire station.

    The GIS has known leaking compartments, but from unknown locations.

    The Craighall demonstration began with a simulated leak then focussed on existing leaks. A careful scan was conducted on a link compartment that had been topped up three times in 1998 and GasVue successfully located the site of the leak: a threaded hole on the bursting disc. Previous attempts to locate this leak using the traditional halogen detector had failed. The leak was found to be intermittent, “essentially puffing SF6 at lengthy intervals” – difficult to detect conventionally unless the sensor is at the leak site at the right moment. With GasVue the puffing action of the SF6 could be readily seen.

    The outdoor demonstration of the camera began at Pieterboth, but due to rain, was moved to Nevis because there was a known leak at head-height. It was planned to try to detect the leak site while keeping the camera within the vehicle.

    This was indeed successfully done. The SF6 was found to be leaking from the control cable conduit into the control box. The leak had originated on the 275kV breaker, with SF6 building up in the control box at the breaker and being forced down the cable conduit to where it picked up by the camera.

    GasVue easily detected the leak and pinpointed its exact origin. A halogen detector could detect the leak in the control box, but much more time would be needed to determine the source of the SF6 at the breaker.

    At the Koeberg 400kV GIS, the GasVue camera accurately located a leak on a feeder link compartment. While the station staff was aware that the compartment was leaking, GasVue pinpointed the location. Interestingly, in this case the distance between the camera and the leak site was rather large, 9m (28 feet). Even at this distance, the camera could detect and locate the SF6 leak. Detection of such a leak conventionally would have been very difficult due to its inconvenient location and the fact the SF6 was being vented from a number of locations simultaneously.

    With the laser camera other leaks were detected and pin-pointed to specific blanking bolts. Again the camera demonstrated the ability to accurately detect and locate leaks at large distances.


    The Eskom tests confirmed that a scan of a GIS or outdoor station could be completed in a fraction of the time it would take using a halogen detector or soapy water, the most significant benefit being the ability to detect leaks on live, in-service apparatus.

    Despite the complex and compact geometry, the camera still proved highly effective in the GIS environment, although a smaller camera and a longer cable between the camera and cooling unit would have been helpful.

    Another striking feature was the ability to detect and locate leaks at a distance. In the Koeberg 400kV GIS case, using the zoom function, the position of the leak was determined to an accuracy of about 1cm at a distance of 9m. This long range capability avoids having to negotiate one’s way round the station on high scaffolding. It also allows leak detection on live components above ground without having to take the equipment out of service.

    On the negative side, the camera as tested at Eskom was found to be heavy for prolonged use and the power supply and cooling unit large. In an outdoor yard, it needed to be transported in a vehicle while in an indoor GIS station it had to be transported on a trolley. EPRI reports that it expects to begin prototype field trials of a new camera design at the end of this year. Size and weight will be noticeably reduced and a simpler cooling system will make it more rugged.