An appraisal of component reliability

With the support of the field evidence quoted in the previous chapter we can summarise our whole findings on the reliability behaviour of non-maintained items or components as follows. Failure occurs when a load on any item exceeds its strength at the instant that load is applied. We have shown that this can arise even when there is no strength degradation with age, because of the distribution of individual loads and strengths about their means. Such failures are easily identified physically: they are of a stress-rupture nature in which no wear-dependent or time-dependent phenomenon is involved. Excluding, in some cases, a short early life regime in which the failure rate decreases with time, these failures are essentially characterised by a constant failure rate which applies over the major part of the life before being terminated by wear-out. They are commonly referred to as random failures. It has been shown theoretically, and substantiated to some extent by field evidence, that this failure rate is critically dependent on the loading. Failure rates may vary by several orders of magnitude in different loading environments. Because of this sensitivity, it is postulated that no rational approach is possible if the existence of random failures is allowed. Consequently it is a fundamental axiom of the philosophy we are following that all random failures must be eliminated by good design. The conditions to achieve this have been set out in section 4.5. Items which have zero random failures (pedantically and exactly — not ‘zero’, but so small that they can be regarded as an ‘act of God’ and are neglected) are said to be intrinsically reliable. There is considerable field evidence to show that a high proportion of existing mechanical components are intrinsically reliable. Whether this is achieved by accident or intent is difficult to say, but if achieved by intent it was most likely done so as a result of experience and empirical design. As a result we have found that there is a significant proportion of items which are not intrinsically reliable. So room for improvement still exists. It is recognised that for some loading conditions a non-intrinsically reliable design may be acceptable, though it may not be inappropriate to note at this point that if any allowance is made for wear, the design will most likely be initially intrinsically reliable, regardless of intent.