Abstract :
The paper discusses the influence of statistical considerations on the design of tests intended to lead to an assessment of the explosion risks arising from the normal operation of electrical equipment on aircraft The statistical approach implies that a finite amount of testing cannot establish a nil risk, the amount of testing depends upon the degree of assurance required. It is shown that an apparently `reasonable¿ degree of assurance leads to an unmanageable amount of test effort if the tests are made under fully simulated service conditions. Test effort can be reduced by orders of magnitude if it is possible to assess separately each of several factors that are simultaneously necessary for an explosion Testing can be further reduced when it is possible to speed up the operation of the unit for test purposes. The proposition that test effort may be reduced by using `enhanced¿ test conditions is considered. In this Connection, dependence on extrapolation procedures is particularly criticized. It is noted that standardization of enclosures for equipment would much reduce test effort. The advantage in relieving the test situation, of dependence on sealed enclosures that need only mechanical testing is emphasized.
