An Improved Algorithm for Constructing Large Fractional Factorial Designs

Abstract

Fractional factorial designs (FF-Designs) are widely used in various engineering, industrial and scientific areas for their run size economy and cost-effective.A complete catalogue of FF-Designs provide a helpful way for experimenters to choose best designs,in this paper we introduce an improved algorithm for constructing the set of all non-isomorphic 2-level regular FF-Designs by developing a new sequential generation procedure that reduce significantly the number of candidate designs from which isomorphs need to be removed, to illustrate the efficiency of the proposed method some comparisons with existing generation procedure are given.The present algorithm is able to enumerate all 16384-run and all 32768-run designs with resolution 9, we extend the catalog by all 65536-run designs with resolution 10, all 131072-run designs with resolution 10 up to 22 factors, all 262144-run designs with resolution 11 and all 524288-run designs with resolution 12, which were not generated in literature