A contribution to the study of the number of tests required to establish the rupturing capacity of an oil circuit-breaker

The paper is based on an investigation of the degree of variation of results obtained in repeat tests in the simplest case of a single-break circuit breaker and fixed conditions of supply. It is shown that the difficulty of rating a circuit breaker is closely analogous to the difficulty a student would have if asked to determine the tensile strength of a specimen on a testing machine in which a concealed fulcrum of the lever system was moving in an unknown manner and varying the leverage over a range of the order of, say, 1 to 50. Curves are given, based on a large mass of data, showing the variations which occur in the individual maxima and the means of arc energy in groups of repeat tests, and startling conclusions are drawn as to the prohibitively large number of tests necessary to establish a rating with any reasonable degree of accuracy by the generally accepted commercial methods of testing. The investigation also throws light on the probable cause of some of the mysterious failures of circuit breakers that have been reported from time to time. On three-phase testing a slight reduction in the number of tests may be allowed, but this does not materially affect the general conclusion that the number necessary would be prohibitively large. A thorough investigation of all possible causes of variation, the development of formulÿ and the determination of constants by test, appears to be the only way of solving the important and pressing problem of the rating of oil circuit-breakers.