Pitfalls of the Typical Construction of Decision Matrices for Concept Selection

Abstract

The conceptual design phase of the engineering design process can be divided into three general activities (i) function specification, (ii) concept generation, and (iii) concept selection. In this paper we explore the pitfalls of one of the most popular approaches for concept selection – namely decision matrices. Although powerful in many cases, methods based on decision matrices may fail to aid designers in selecting potentially preferable designs. This is simply because decision matrices are based on an inadequate mathematical construct. In this paper, we show that some non-dominated design concepts (in particular, optimal concepts that lie on non-convex regions of the Pareto frontier) do not receive the highest total score when using decision matrices in their typical form; and as a result, may be prematurely eliminated. In practice, it is not uncommon for designers to use information from the decision matrix to classify certain design concepts as undesirable, when actually they may be desirable. This paper exposes significant risks associated with decision matrices, and suggests candidate alternative approaches. Interestingly, we also show that constructing decision matrices using approaches such as compromise programming does not necessarily solve the problem of misrepresenting the desirability of design concepts that lie on non-convex regions of the Pareto frontier.