Surge arresters are used to protect equipment such as transformers and switchgear from over-voltages caused by lightning, switching and temporary voltages caused by power system faults. The operational reliability of a surge arrester is substantially influenced by its surge withstand capability. This is particularly important for surge arresters used in overhead distribution lines. The lightning stress withstand capability of a surge arrester is usually assessed in the laboratory by applying up to 20 impulse currents, normally in 4 groups of 5 impulses applied at 60 second intervals, with an interval of 30 minutes between groups. From the literature, it was known that about seventy (70) percent of lightning ground flashes are composed of multiple strokes, with an average number of 4 strokes per flash (range 1 to over 20) and an average inter-stroke time interval of about 40 ms (range 15 to 140 ms). The standard testing of these devices may therefore not be truly representative of the surges encountered in real applications.
It is also observed that only one group, Darveniza and Mercer , appears to have studied the effects of multiple-stroke lightning on distribution arresters. Darveniza and Mercer  studied the response of the silicone carbide arresters using multipulse currents.
With the financial support of the Australian Research Council, the Australian Electricity Industry Research Board and ASEA Brown Boveri Power Transmission (Australia), the Department of Electrical & Computer Engineering at the University of Queensland began a research project in 1990 aimed at assessing the withstand capability of a zinc oxide (ZnO) surge arrester using multiple-stroke lightning impulses. The tested arresters include 18 of 5 kA rated discharge current and 3 of 10 kA rating. The tested ZnO blocks include 39 of 5 kA rating and 6 of 10 kA rating. A summary of all the tests carried out on arresters and blocks and the test results can be found in Tables B.l-B.2 of Appendix B.
For the assessment of withstand capability of a zinc oxide arrester, two types of electrical tests were made in this research project. These include impulse tests without a.c. energisation and operating duty cycle tests. The effects of these tests were evaluated using several electrical methods. These include measurements of d.c. and a.c. reference voltages at a leakage current level specified by manufacturer, discharge residual voltage at rated nominal discharge current, maximum power at continuous operating voltage of the test specimen and harmonic analysis of the resistive current component.
Comparisons of the effects of multipulse and standard single impulse currents applied to arresters show important differences. The result of major significance is that multipulse currents can give rise to changes in ZnO block characteristics (including failure) that are not evident during single impulse currents. The multipulse test results reveal that 5kA and l0kA rated ZnO arresters could withstand their rated discharge currents without much degradation. However, some of the 5kA arresters are more vulnerable to damage at higher multipulse currents of 5kA to 9.5kA. In fact, 12 out of 18 arresters of 5kA rating failed at current levels between 5 and 9.5kA. The effects of these tests were found to be cumulative. The test results also indicate that test samples could be damaged in two ways by surface flashover and block shattering. The most common was damage by surface flashover. The probable causes of surface flashover in zinc oxide test samples were investigated. The more likely causes of surface flashover were plasma generation, manufacturing defects of the ZnO block surface coating, dielectric properties of the surface coating and the electrode contact system. Block surface coating was found to play a dominant role in surface flashover mechanism. Zinc oxide blocks with glass surface coating could be damaged by flashover at about 65 kA (4/10 µs) single pulses and 10 kA (8/20 µs) multipulses. This flashover could be inhibited up to 90 kA (4/10 µs) by an additional surface coating with silicone varnish or by enclosing blocks in a water-proof polymer housing. The improvement in the zinc oxide block performance was mainly due to the elimination of mono-molecular layers of gel (water) on the glass surface of a zinc oxide block. Two of three arresters returned from field service were found damaged by surface flashover.
Impulse tests (without a.c. voltage energisation) carried out with single and multipulse current impulses on ZnO arresters were successfully simulated using a computer program adopted from PSPICE and using numerical techniques. Operating duty tests conducted with single and multiple impulse currents were also simulated by using a combination of an electro-thermal model and a computer program adapted from PSPICE or by using numerical techniques. The simulated results were within 5% of the test data.
Metallographical examination studies on ZnO arrester blocks appeared to show no evidence of any additional phase formation. This study also indicates that that no relationship could be drawn between impulse current magnitude and level of degradtion of a ZnO block. "
The literature observes that the average currents discharged by a distribution arrester are I in the range 5 to 6 kA, with the mean current of direct strokes estimated at 31 kA. Distribution arresters fitted on overhead lines are exposed to the full severity of lightning strikes. Multiplicity is an_ inherent feature of lightning flashes to the ground, and the multipulse tests with 5 to 10 kA currents reported in this thesis are realistic laboratory simulations of what can happen in ' - the field; this is because lightning refuses to meet the classical guidelines set in the standards. In view of the test results, it can be recommended to use 10 kA current rating arresters rather than 5 kA rating arresters for overhead distribution protection purposes. A 10 kA arrester should give better performance, but the multipulse test is still capable of destroying arresters with blocks of poor homogeneity.