Science, engineering, and statistics

Symmetry and parsimony, together with unification through synthesis are important principles that govern the character of physical law. We show how these principles can be applied to engineering to develop an approach to reliability, and engineering in general, that centres on the identification, detection, and avoidance of failure modes through design. A definition of reliability, not presented in terms of probability, but rather based on physics, geometry, and the properties of materials, will be emphasized to support this approach. We will also show how the nature of the inductive-deductive learning cycle provides the framework for statistical science to be embedded into engineering practice, with particular regard to improving reliability through failure mode avoidance. Copyright (c) 2006 John Wiley & Sons, Ltd.