In 1993, LILCO initiated a study to determine the effectiveness of surge arresters at different spacings on feeders that were selected on the basis of their history with lightning problems. Five circuits, monitored with a variety of equipment for several seasons of lightning, are composed of three feeders with different arrester spacings and two control feeders that have no extra arresters. Lightning flashes are correlated with recorded faults by substation optical and magnetic sensors, National Lightning Detection Network data and lightning-activated camera systems. The tests used arresters on every phase of every pole for one circuit, arresters on every phase at four poles per mi on another circuit and arresters on every phase at eight poles per mi for the third circuit.
Three lightning-caused faults, recorded on the circuit with arresters on every pole and every phase, were significant because theory suggests that lightning-caused flashovers should not have occurred. Each fault occurred on a lateral and caused a fuse to operate. Although there should have been arresters on all poles, field inspection revealed that some arresters were missing, illustrating a common construction problem that is difficult to achieve in practice for using arresters on every phase at every pole. In any case, flashovers occurred even with considerable arrester usage. Several lightning-caused faults were recorded on the circuits with arresters spaced at rates of four and eight per mi. In each case, arresters were present within one pole span, yet flashovers still occurred across insulators to guy wires. On one line, the lightning hit a pole with arresters on all phases, but a flashover still occurred at the next pole down, across two of the phase insulators. In this case, one of the MOV arresters failed, raising the concern with distribution-class arresters and posing the question as to whether MOVs used for line protection can accommodate the energy in a direct strike. Based on preliminary results, even arresters at every pole have not been found to eliminate flashovers. The results also verify that arrester spacings at wider intervals are not effective for reducing flashovers from direct hits. This observation contrasts with reports from some southeastern utilities who report good results with wider arrester spacings and suggests that their improved performance may be the result of induced flashovers from nearby strikes and not from direct hits. It appears that arresters should not be used for lines in the open, which are exposed to direct hits and to few induced flashovers. On the other hand, for areas that are shielded by trees and buildings, where induced voltages are the biggest concern, flashovers could be handled by improved insulation or with surge arresters.