Statnett SF, the Norwegian power grid company, is using video and image processing technology for remote monitoring. Fighting adverse weather and its impact on transmission lines that cross mountainous terrain is the task facing design and operational staff at Statnett SF. Extreme wind forces and ice loads can cause immediate breakdowns while vibration, conductor galloping and other dynamic effects cause wear and fatigue of conductors and hardware, or even clashing of conductors. Ice, polluted with sea salt or sulfates and nitrates transported over long distances, causes strong leakage currents over the insulators, flashover and degradation. It is important to have exact knowledge and "see" what is going on at these remote locations to take the appropriate countermeasures-a difficult task when the transmission line is not accessible for months at a time.

Remote Monitoring Systems The main objective of remote monitoring is to improve the reliability of the transmission line system. Statnett SF considers the following phenomena to be the most important: conductor galloping and other dynamic phenomena; ice accretion on earth wires and conductors; and discharge activities on insulators in icing areas.

The information collected by the remote monitors may serve to: - Alert the operational staff. - Assist load dispatch centers and field crew in identifying the exact cause of transmission outages. - Provide information on the effectiveness of countermeasures. - Provide information that can be used for improving the design of future lines.

Blaasjoe Test Site In 1994, the Blaasjoe test site was established to study the potential of using video and image processing technologies on the route of the 420-kV Kvilldal-Holen transmission line. This circuit is routed on the top of the mountain range in the middle of southern Norway, 1100 m (3600 ft) above sea level. This line, which has a direction of about 330 deg clockwise from the north, was chosen because it seems to have the most severe exposure to galloping conditions of any transmission line in Norway. Conductor galloping with peak-to-peak amplitudes up to 10 m (33 ft) were observed in the 1980s, resulting in severe structural damage. The line is designed for 30 kg/m (20 lb/ft) ice loading in this area, and wind speeds of 50 m/s to 55 m/s (110 mph to 125 mph) may be expected. In the undulating landscape, the snow depths may be from an insignificant amount to 15 m or 20 m (50 ft or 66 ft) or more in the hollow parts, making it impossible or dangerous to access this area from the end of October to the middle of January. Strong winds and fog make the use of helicopters very unreliable.

The line has a horizontal phase configuration and V-string insulation. Because of conductor galloping on circuits with V-string suspensions, the test span now has one phase with I-strings of non-ceramic insulators. One of the other two phases is equipped with de-tuning pendulums for galloping control while the middle phase remains as constructed for reference purposes.

De-tuning pendulums are masses attached to the overhead power lines to modify their mechanical properties and reduce their susceptibility to galloping. The arm length is chosen to separate the frequency of torsional motion away from the vertical natural frequency with wind and ice on the conductors. The amount of mass is chosen to maintain this frequency separation up to a preset level of ice and wind moment. Usually the masses are located at four positions along the span dividing the length into unequal subspans. Extensive field experiences in North America show that on both single and bundle conductors, the pendulums reduce maximum galloping amplitudes to about one-third of the amplitudes of lines without pendulums. Figure 1 shows the de-tuning pendulums used at the test site.

The monitoring system now installed comprises five cameras, V1 to V5, observing different parts of the line as shown in Fig 2. Camera V1 gives an overview towards Tower M98, and camera V2 monitors the suspension strings on Tower M99. Cameras V3, V4 and V5 are focused on the phase conductors and earth wires 52.5 m (170 ft) from Tower M99. Figure 3 shows the monitored span on the remote circuit, the weather station and camera V4. The cameras are connected to a PC, which is located in a heated building. Information is transferred via a radio link, which is connected to Statnett's regular telecommunication system and to the main office in Oslo (Fig. 4). The facility is powered from a nearby low-voltage line.

In the digital image processing software, up to 16 separate task windows may be defined for each camera. Each window may be given a different setting, with individual parameters for all aspects of the detection. Rapid changes of intensity and other values in a predefined number of these windows are used toidentify the positions of the conductors and their movements. The system uses light-intensive cameras, extending its ability to operate in low-light conditions.

The remote PC runs Astraguard software to detect movement of the conductors, analyzing the images looking for actual movement of cables. When motion is observed, the existing monitoring system operates: - A picture sequence from the detecting camera is saved on the PC, and the system quits Astraguard. - Video sequences are recorded from all the other cameras; the number from each camera may vary. - The system saves original high-resolution pictures to a remote disk. - The system creates low-resolution highly compressed pictures for transfer (2 kB to 3 kB each). - Compressed images (more than 50 times) are transferred and inspected. - Original pictures may be collected at any time on request. The system is supplemented by a weather station near camera V4 that records air temperature, relative humidity, wind speed, wind direction and icing rate (ice detector). The wind sensor is heated and has no moving parts. The weather station transmits data every hour via the radio link to the central PC located in Oslo.

Difficulties have been experienced in automatically separating image changes because of conductor movements from variable sunshine, clouds, rain, snow and birds. However, the software has shown the ability to "learn" the normal conditions and has reduced the number of false alarms to an acceptable level.

Galloping Events During the 1997/98 winter, this system detected three major galloping events, in addition to several minor ones. The information collected demonstrated that there is still a great deal to learn about galloping phenomena. Table 1 shows the temperature, relative humidity and icing rates on the days of the major events. The wind data was takenfrom the Midtlaeger weather station-which is located approximately 30 km to 35 km (19 miles to 22 miles) from the site-because of local anemometer malfunction.

Table 2 summarizes the conductor movement during the three events that occurred during differing wind and ice loading conditions. This data set confirms that no firm conductor behavioral patterns can be established. Manual picture analysis is labor intensive, and Statnett SF, therefore, supports further development of the image processing. The intention is that the software will automatically derive complete time series data similar to that shown in Table 2 for each individual conductor in a multicircuit system. Furthermore, if the conductors are equipped with markers, the torsional movements-which are an integral part of the galloping motions-could be examined as well. To accomplish all these intentions, it may be necessary to use two synchronized cameras at different locations.

Other applications For the 1998/99 winter weather period, the detection system on the Kvilldal-Holen line was expanded to record discharge activity on the insulators in the nearest tower. The same insulators were already equipped with instruments for leakage current measurements (Fig. 5). Possible discharges may be visible during non-daylight hours, but to date, no event has been detected, even though these discharges may easily be detected with this system.

Statnett SF also is testing the application of video and image processing technology for verifying the lightning detection system. A device for ice detection and quantitative measurements of ice accretion is being evaluated as well. Although video cameras cannot be installed everywhere, studies such as these may be performed according to the needs and special problems of particular locations. This technology represents an automation of the eye, because, in principle, it is possible to identify and react to any event that can be seen by the human eye. Therefore, a new dimension is being introduced for automatic remote monitoring in general.

Leakage Current Measurements In Norway, outages caused by sea salt deposits on the insulatorsare frequent along the coast, especially on the high-voltage distribution system. During the last two decades, contaminated snow and ice accretions on the insulators also have caused severe problems on the extra high-voltage transmission system in mountainous areas. The pollution may be both from sea salt and the long-range transportation of sulfates and nitrates from other European countries.

Measurement of leakage currents may serve to monitor the presence of polluted snow and ice accretion on the insulators. A newly developed technique uses the time variation of the leakage current amplitude, or the accumulated leakage current charge during a certain time interval as the most direct way of determining the behavior of the insulator during pollution. This pollution monitor consists of a self-contained instrument with the integrator electronics and circuitry for data processing. The leakage current signal is measured via a measuring resistor (shunt) with protection circuitry.

The SCOUT Concept The project detailed above has led to the idea of a smaller, automatic video-analyzing station being developed, called the self contained observation unit (SCOUT). The main unit contains PCs for image processing and other data-handling purposes, telecommunications (based on cellular or satellite phones) and a battery package powered by wind turbine, solar cells or other systems. Four to eight video cameras can be mounted on the container or externally. All PCs and cameras can be operated remotely from a host PC, and pan-tilt-zoom cameras can be controlled remotely from the host computer to check details of the site whenever required. The main unit is rugged and may be transported to the site by helicopter or snowmobile.

By using advanced image compression, it is possible to reduce the amount of data by 100 to 1000 times, allowing for online, live video analyses and detailed video coverage of alarm events with standard PC capacity and regular telecommunication speeds. After transmission the video frames may be displayed with true speed.

Alarm messages may be sent to several preprogrammed addresses, including PCs in operating centers, fax machines, pagers, cellular phones, or other recording devices and operational personnel. The staff also will have the online facilities to download present status data from the local station for inspection.

In addition to video systems, this unit will handle conventional instruments, such as a weather station, leakage current measurement and strain gauges.

One major advantage with a "pure" video-analyzing system is that it will give valuable information on the behavior and physical condition of remote transmission lines without interfering with the line itself, neither for installation nor maintenance of the instruments. The cost of a complete unit with diesel engine and satellite communication currently is estimated to be US$100,000 to US$150,000, depending on the number of cameras and serial production of the system. It is important that this concept is valued in terms of a cost-benefit analysis that takes into consideration the costs of manual inspections including travel, helicopters and the potential cost savings that arise when a condition-based maintenance strategy is developed.

These systems are too expensive to install in large numbers, but relevant utilities have the knowledge and experience to identify those locations where the transmission systems are exposed to the most onerous climatic conditions. It should be an easy task to "tailor" one or a few stations, which could provide the information required to design and construct transmission lines capable of withstanding all the forces of nature. Such a system is now proposed for an exposed transmission line in the subarctic region in northern Norway, where climatic conditions are worse than at the Blaasjoe test site. A 420-kV line along the coast outside the glacier Svartisen typically has 3 to 13 outages each year-mostly because of conductor and loop clashing-each lasting from a few minutes up to a day or two. This line often is inaccessible for up to three or four months during winter. The lost revenue may beabout US$150,000 a day. The planned regime for compensation of nondelivered energy in the Norwegian system may imply future losses for Statnett could be 10 times this value. Although many actions are taken to reduce the number of failures on this line, it is still important to learn about the mechanical and electrical behavior under extreme weather conditions to improve their performance efficiently with a minimum of "trial and error."

In other cases, lines on the western side of the mountains fail because of polluted ice on insulators. In 1993, there was a total breakdown of the system when 11-kV, 300-kV and 420-kV lines failed simultaneously for this reason. Dramatic economic losses for the metallurgic and oil industries were barely avoided, as an increase in air temperature resulted in the subsequent melting of the ice from the insulators.

Monitoring pollution and leakage current also is important in understanding more about insulation levels, especially if HVDC mountain lines are introduced on the Norwegian transmission system in the future.

Acknowledgments Statnett SF is grateful for the active and valuable engagements and support from Rune Stenseth of Barnard AS and David G. Havard of Havard Engineering, Toronto, Canada, for the video and galloping studies, as well as from Asle Schei and Vegard Larsen of TransiNor AS who have taken care of the CPI development and leakage current studies. The project is partly funded by the Norwegian Research Council.