Abstract
Reliability has been described as a parameter of design. Where there is a large amount of capital at stake, or a risk to human life, the characteristic of reliability becomes very important. The designer may base his estimates of reliability on: (1) field experience; (2) sample testing; (3) prediction. In high risk situations it is suggested that prediction techniques play an essential part in early design. The need is stressed for the quantification of reliability; this will tend to remove the subjectiveness and thereby help communication. The methods of prediction are considered, based on a suitable reliability definition. These methods range from obtaining a consensus of opinion from a body of people to the mathematical modelling techniques employed by individual designers and assessors. The reliability data available on component parts are surveyed and used as the basis for describing current techniques for predicting the reliability of equipments and systems. In particular, the techniques for the safety analysis of various modes of failure are considered. The correlation between predicted and practical results is described for some applications. Conclusions are drawn on the general applicability of the prediction methods and the requirements for an organized reliability data bank.

This publication has 0 references indexed in Scilit: