For years I’ve been arguing that arresters do not fail from lightning – period. But that view has changed now that my partner, Deborah, convinced me to re-consider how I approach the issue. Her argument: if a surge arrester has already degraded due to moisture ingress or other factors and is later subjected to a lightning surge that causes it to fail, is it not reasonable to regard this as lightning induced? In the past, I would have said ‘no’ and labeled it rather due to moisture ingress or whatever other source of degradation. Her second point: if lightning causes a line-to-ground fault on phase A and an arrester then fails on phase B due to the resulting voltage rise, should this not also be considered a lightning failure?
I have to agree with her simple logic. Perhaps the message that arresters do not fail from lightning should be modified to: arresters do not fail due only to high current lightning surges.
As a longtime arrester designer and forensics specialist, I’ve seen many arresters that failed presumably due to lightning because this happened during a storm. However, subsequent teardown and forensic analyses seldom revealed the tell tale asymmetry in the VI characteristic of MOV disks that occurs during any high current surge. Indeed, this asymmetry has been verified in nearly all know varistor formulations.
For example, Fig. 1 shows such asymmetry in disk currents. In this case, the damage was severe with 20% difference between peak positive and peak negative currents. Such asymmetry signals permanent change has taken place in the MOV disks and is a valuable clue whenever trying to identify the root cause of arrester failure. However the lack of this ‘footprint’ in almost all arrester failures I’ve analyzed leads me to conclude that high current lightning is not the typical cause.
There is even a stronger rationale that guides understanding this topic. An EPRI study completed in 1981 (Report EL1140) undertook the challenge of quantifying the magnitude of currents flowing through some 2800 arresters, representing 32,808 arrester-years service and coming from 14 different utilities. The study was prompted by high failure rates of distribution arresters during the 1970s and which some thought was due to lightning surges. In 1981, the dominant arrester type in service was the gapped SiC unit and the gap was an excellent recorder of current flow through the device. Researchers painstakingly examined the gap etchings and compared these to etchings created in the laboratory. Based on these comparisons, the graph shown in Fig. 2 was created. By the way, similar studies had been carried out in the 1930s and 40s and this study was an update. Moreover, the EPRI data closely matched findings in a CIGRE paper by SC 33 from 1978.
Fig. 2 indicates that, for urban lines (i.e. partially shielded by trees) and where ground flash density (GFD) is 5, the time between 100 kA surges is about 800 years. In a GFD zone of 20, a 100 kA surge through an arrester occurs only once in about 200 years. However, if the arrester is installed on exposed lines in a GFD region of 10 (as is much of Florida), the arrester can expect to experience a current of 100 kA about every 60 years.
As illustrated by this data, surge arresters are clearly not normally exposed to high lightning currents. Since all Class 1 or DH Class of distribution arresters manufactured today are designed to withstand 100 kA 100% of the time, they are fully capable of handling the greater percent of lightning strokes.
It is these two sets of data that drive the argument that arresters do not fail solely from high current lightning. While they may fail during electrical storms, it is surely not due solely to lightning surge currents. By corollary, should an arrester fail during a lighting surge, it is safe to assume that it was well on its way to failure for some other reason or due to a long TOV – but not from the sudden high current.
Co-Convenor of IEC TC37 MT4
responsible for IEC 60099-4