Design-in quality and reliability via experimental design

Abstract

Quality and Reliability by Design as a concept and guiding principle is not new. However, the requisite methodologies have not necessarily been well grasped by practitioners of technological design. In this paper, we use a study based on the design of a relatively simple appliance to examine an important dimension of quality and reliability enhancement at the design stage, namely optimization via statistical design of experiments (DOE). While the inadequacies of one-factor-at-a-time or OFAT are well known, it may not be commonly realized that standard DOE may not bring about straightforward results that the design engineers are looking for, and appropriate interpretations of DOE outcomes are necessary to bring about useful solutions. Furthermore, an orthogonal experiment may point to design solutions that are better than what actually has been tested in a series of prototypes. These points are illustrated in the realistic case study presented in this paper.