The design of AC overhead lines requires considering many factors, including wind and ice loads, strength of poles and insulator attachment hardware, line voltage and transient overvoltage stresses as well as insulator pollution exposure. It is therefore difficult to define any uniform design methodology for such lines, sometimes even within a single power utility, as these factors may change from one service territory to the next.
An interesting example of this situation relates to lightning protection and involves a U.S. based power supply utility, the Bonneville Power Authority. BPA uses overhead groundwire (OHGW) protection on all its lines running east of the Cascades Mountains but not on lines that run westward toward the Pacific. The lightning flash density towards the west is apparently low enough that tripout rates are deemed satisfactory even though every flash terminates on an energized phase conductor. This is normally considered a shielding failure and at least one of the strokes in a flash will be strong enough to cause a flashover.
The voltage rise in this case is given by the peak current multiplied by the surge impedance of the phase divided by two – because the current splits equally in each direction away from the point of attachment. Once lightning is allowed to flow into the phase conductor’s surge impedance, there are two issues that need to be addressed: 1. possible stress on substation equipment as the current surge arrives and 2. the additional stress caused when the insulators flash over, with voltage collapse times as short as 100 nanoseconds. Since the rate of change of voltage can be sufficient to initiate or advance small punctures from cap to pin in porcelain insulators, some service areas require high resistance to puncture. In Canada, for example, where unshielded lines are common both in British Columbia and Newfoundland, a standard requirement is that ceramic insulators do not puncture when exposed to a rate of change of voltage of 2500 kV per microsecond.
I was close to the Cascades Mountains this past July, attending the IEEE’s Annual Meeting in Denver, Colorado where I also took the opportunity to make the scenic drive to Gateway Canyon. My interest was not only as a tourist but also to discuss additions to a document being prepared by the High Voltage Direct Current (HVDC) working group of the Overhead Lines Subcommittee. Rather than starting from scratch, the writing team felt that a document that highlights the differences between ac and DC line design would be more concise and useful. For example, CIGRE Technical Brochure 518 proved to be an excellent resource for selecting leakage distance based on unified specific creepage distance. Corrections for DC USCD include a table for preferential pollution accumulation rate due to electrostatic attraction. Detailed comments about electric fields, ion density, corona and audible nose also appear in the new report. My task was to assess the role of a metallic earth return wire as well as HVDC line voltage bias issues on the lightning backflashover calculation process.
The voltage bias issue is straightforward and has been part of IEEE 1243 since 1997. Lightning is predominantly negative in polarity. If the line has continuous dc voltage, this will either add (positive pole) or subtract (negative pole) from the impulse voltage that occurs at the ground ends of the insulators. Programs such as IEEE FLASH suggest that the positive pole is three times more likely to have lightning tripouts than the negative pole. Another part of the modeling, usually hidden inside code, is driven by the number and locations of OHGW. It turns out that insulated phase conductors located closer to the OHGW than to ground can pick up a significant impulse voltage even though there is no current flowing into them. The guide illustrates this with a series of matrix calculations using self and mutual surge impedance values [Z] multiplied by vectors of currents [I] to get vectors of voltages [V]. An example with two OHGW showed that an amp of current flowing into each OHGW caused a voltage rise of 677 V on the OHGW, and 195 V¬ on the energized poles. This was great news for the insulation since it meant that the impulse stress was reduced to (677 – 195) = 482 V. The ratio 195/677 = 0.29 is called a “coupling factor” and its model is also part of normal ac line backflashover performance calculation in the IEEE FLASH program.
I felt it important to set out these background details for HVDC line designers because they often face a choice whether to invest in expensive, substation-sized earth return electrodes at each end or, alternately, to provide an insulated earth return conductor from end to end. When one of the poles is out of service, the return wire can rise in potential to 50 kVdc, so its insulators can be much shorter and it can be placed below and between the two poles. Under lightning surge conditions, the ground return insulators will flash over immediately, bringing in some additional relief to the voltage stress across the two poles. The earth return will carry away a small added fraction of lightning current and more importantly will improve the coupling factor from 0.29 to 0.37.
Against the benchmark of a single OHGW with no earth return, as used for example in Manitoba, Canada, the twin OHGW with single earth return would have a 62% advantage. The closer that the ground return conductor is to the phases the better will be its effect in regard to lightning. An experienced designer will recognize other trade-offs and may need to control separation using inter-phase spacers or reduced span lengths.
The goal of the IEEE Task Force is to produce a PES “Technical Report” – a new format that allows greater scope than Transactions papers of limited length. The Overhead Lines subcommittee issued its first, TR-17, “An Introductory Discussion on Aeolian Vibration of Single Conductors” running 38 pages that will soon be available on the PES Resource Center. The completed HVDC line design guide should follow shortly, after completing its review process.
Dr. William A. Chisholm