First-Hand:The Evolution of Isolated Phase Bus Duct, 1950 to 2000


Submitted by Robert Rehder P Eng, LSM3512357, Oct 3, 2018.

The aim was to maximize the reliability and resilience of bus ducts used in power stations and critical applications mainly above 3000 amperes.

Prior to 1950, the initial connections from generator to circuit breaker and from breaker to distribution transformer consisted of bare conductors mounted on porcelain insulators and with phase spacing sufficient to meet the industry standards at the time of manufacture. The spacing of the insulators along the length of the bus was dependent on the short circuit current available and the mechanical load capability of the conductor size and shape. For currents above 2000 amperes, losses due to skin effects dictated the use of tubular conductors or a box arrangement of two channel shapes. This simple system was subject to failure from dirt contamination of the porcelain insulators or from animals such as rats. Once a ground fault occurred it contaminated the air insulation between phases and the fault then became a serious phase to phase fault. As the arc continues, a force will act on it to move it away from the source of power and it will motor itself as far as it can go and still sustain itself, which is usually to the transformer bushings, and it will continue to melt and damage the transformer until protective relays cut off the source of power. Design engineers added enclosures to reduce the contamination of the insulator surfaces and keep out the rodents.

The first enclosures enclosed all three phases. This still allowed a ground fault to develop into a more severe phase to phase fault. Motoring along the bus can be limited by insulating the conductor surfaces. Some manufacturers add a thin skin of insulation to the conductors to minimize the chance of an arcing ground fault motoring away from the source of power, taking out terminal devices such as transformer bushings. The disadvantage to the insulation coverage of the conductor is the addition of combustible material. The advantage is the reduction of air clearances to ground and between phases to approximately half value.

The next step was to enclose all three phases with the main enclosure but add barriers between phases. This is known as segregated phase bus duct. If there is a ground fault it will take time for it to burn through the barrier and contaminate the air in the adjacent phase which is followed by a phase to phase fault. The speed of the ground fault relay is critical as is the design and materials in the phase barriers.

In the 1950s the demands for electricity had a major post war increase, and steam and hydro generators were increasing in size to meet the needs. Current ratings were increased to more than 10,000 amperes. To gain reliability, the phases were each totally enclosed in separate square aluminum sheet enclosures with an air space between each enclosure. The length of each enclosure was limited to approximately 10 feet to limit the build-up of voltage from the induced currents in the enclosures. Each enclosure section was grounded at only one point to prevent circulation of induced currents in the enclosure mounting hardware. Steel beams supporting the three phase enclosures had copper bar loops around them to prevent the induced voltage and associate currents causing thermal damage in the support members. At the ends, enclosures were insulated from each other. Aluminum covers on each phase provided access to the conductor joints. Hatch covers were located on the enclosures to provide access to insulators; there were more than one insulator at the support point.

The square enclosures were soon changed to cylindrical enclosures to reduce the magnitude and complexity of the induced currents in the enclosures. General Electric did extensive tests to prove that a single insulator was sufficient to support the conductor and under short circuit conditions the conductors wanted to move to a point of zero force which is about one half inch off the centre of the enclosure. The insulator was mounted on a spring plate to permit it to make this move. Using only one insulator at each longitudinal conductor support point instead of multiple insulators reduced the number of creepage paths to ground and therefore increased reliability.

The conductors were aluminum so joining the conductors required special attention because aluminum oxide could form and it is a good insulator. Anti-oxidant greases were used when making the bolted joints. This limited the temperature rise to 50 degrees C. For the desirable 65 degrees C rise, the joint surfaces had to be plated with tin or silver and the existing plating processes were not reliable. Some manufacturers had the plating company subject every plated joint to 300 degrees C and then check them for lack of bond of the plating material to the aluminum. Welding the joints, and welding in folded aluminum sheet or bellows to absorb thermal expansion and contraction, gave the most reliable joint.

Flexible copper braids were used at bus duct terminations at generators or transformers to absorb motion or vibration on some installations. Research proved that the sleeves pressed around the braids to contain the individual strands must be compressed by a pressure that is beyond the mechanical yield point of the copper strands so that there are no air spaces between strands. The steel bolts used to make the joint must be tightened to a high enough torque to stretch the bolts to be able to absorb the permanent yield of the joint under short circuit thermal effects and still have enough stretch left to provide minimum joint pressure after a fault had been cleared. Any air space between strands will allow the copper of the strand to move which will use up some of the stretch in the steel bolt and the joint can loosen and cause a hot connection.

In the 1960s the size of the power stations and current ratings were doubling. For space and cost reasons, forced air cooling became common. The heat losses in each of the outside phases are half of the losses in the center phase in a side by side arrangement. Most manufacturers blew air down the centre phase and had it return in the two outside phases. The air duct to pass the cooling air from the centre phase to the outside phases was equipped with a stack of narrowly spaced aluminum sheets to de-ionize the air should there be an arc in one of the phases. This was to prevent a phase to ground fault in one phase spreading to a phase to ground to phase fault in the system.

In order to determine the size of the cooling system the designer needed to quantify the I2R losses in both the conductors and the enclosures. The conductor losses were easy to establish from reliable formulae. The enclosure losses were difficult to establish because it was chalenging to determine the magnitude and path of the induced currents in the enclosures. At first each manufacturer had their own formulae and there seemed to be no test method to verify by test which formulae was closest to reality. GECanada, Peterborough Engineering Laboratory developed a calorimetric method to measure the losses at any point on the enclosure or the total losses around the circumference at any point on the enclosure length. It involved applying a fixed current (4000 amperes) to the bus for approximately 10 seconds and, using a chart recorder, record the temperature rise using a temperature sensor on a point on the enclosure. The loss at the point of measurement is the tangent of the angle at the start of the temperature rise. This angle is determined by moving up the curve and then tracing the curve back to the start to eliminate the effect of initial transients. The results of this procedure was used to establish an industry standard formulae for calculating the enclosure losses.

As total losses are a user’s cost for the life of the machine, this became a factor in evaluation of a proposal so the purchaser specified the cost of losses to be included in the pricing. It also was used to optimize the best self-cooled rating to be used and the amount of forced air cooling required. Through the 1960s the industry developed a continuous enclosure design. In this design the enclosures were electrically continuous by welded or bolted connections that could carry the same amount of current as the main conductor. At both ends of each phase enclosure there were welded or bolted electrical connections between all three phases so that the induced voltages in the enclosures could result in a circulating current approaching the magnitude of the main conductor current. The magnitude of this current is a function of the resistance of the enclosure’s total length and the resistance of the end connections. Current values of 95% of main conductor current can flow in the enclosures which means that the external magnetic fields from that phase is only due to 5% current. Therefore the short circuit force on the main conductor under these conditions is reduced by approximately 90%. The number of support insulators along the main conductor length can then be reduced and the creepage surfaces to ground are very much reduced, increasing reliability and resilience. The position of the insulators relative to external mechanical supports is no longer critical. Also, reduction of the external fields results in less heating in nearby iron conduits or structural beams or in steel reinforcement in structural concrete beams.

By the 1970s most of the designs used continuous enclosures and the enclosures had enough mechanical strength to span between structural support points and did not use separate longitudinal supports. Some manufacturers used aluminum in their support structures instead of galvanized steel to save weight and possibly reduce installation costs. For example, “A” frame supports could use the same tubes as the lower current conductors reducing inventory costs for the manufacturer and reduced weight for the installer.

On September 1, 1976, GE Canada performed a losses measurement test at a nuclear generating station to verify that the losses at full current rating were equal to or less than the number given in the contract documents. Temperature sensors were placed on 20 locations on the inside surface of the tubular main conductors. Wires were run from the sensors to a metering device inside the conductor tube. The meter’s screen filled an opening in the conductor and the screen sequentially displayed the temperature readings. There was a viewing port in the opposite enclosure so that a technician could safely read and record the numbers. As there is no magnetic field inside the conductor the instrument’s operation was not affected. There were two sources of power for the instrumentation, one a battery, and the other by a wire from each end of the total conductor length, which brought the voltage developed by the IR drop of the current flowing through the conductor. The voltage drop was sufficient to record temperatures provided the current flow was more than 50% of the total generator’s power rating. At the start of the test, the generator terminals were reconnected to have the generator and isolated phase bus circulating the full 27,760 amperes rating. With time, the temperature would rise and level off at a maximum value. The station operator phoned Peterborough and gave them the current value and ambient temperatures. Peterborough started working through their design data and tabulated their calculated value of the temperature rise at each of the points being monitored at site. When the temperatures had leveled off (18.3 hours after start) the site phoned Peterborough and they compared the measured temperature rise with Peterborough’s calculation results. All the temperatures calculated were greater than the measured temperatures by a few degrees and this verified that the bus duct and its forced air cooling met the losses statements in the contract documents. This same measurement procedure was used a few years later on another contract with the improvement of transferring the temperature readings across the high voltage air gap by using an infrared pulse signal that was translated back to digital information by a station computer. The equipment was left in service when the commissioning work was complete.

From 1970 to 2000 there was concern about the reliability of the plating on the conductors. Fixed and portable temperature sensor instruments were used to safely observe the temperature of conductor joints and circulating induced current heating in enclosures and adjacent structures. Designers continue to strive for increased reliability and resilience.