Creating clear designs: a graph-based algorithm and a catalog of clear compromise plans

Abstract

A graph-based algorithm is proposed for creating regular fractional factorial designs with two-level factors such that a pre-specified set of two-factor interactions is clear of aliasing with any main effects or two-factor interactions (clear design). The Clear Interactions Graph (CIG) used in the algorithm is unique for each design and different in nature from the well-known Taguchi linear graph. Based on published catalogs of two-level fractional factorials, enhanced by the CIG, a search algorithm finds an appropriate clear design or declares its non-existence. The approach is applied to the creation of a catalog of minimum aberration clear compromise plans, which is also of interest in its own right. [Supplementary materials are available for this article. Go to the publisher’s online edition of IIE Transactions for additional discussions on run times and implementation of the algorithm for larger designs.]

Keywords:

• Clear two-factor interactions
• compromise plan
• design of experiment
• linear graph
• clear interactions graph