Construction of Orthogonal Arrays without Interaction Columns

In this paper a new class of orthogonal arrays (OAs), i.e., OAs without interaction columns, are proposed which are applicable in factor screening, interaction detection and other cases. With the tools of di erence matrices, we present some general recursive methods for constructing OAs of such type. Several families of OAs with high percent saturation are constructed. In particular, for any integer λ ≥ 3, such a two-level OA of run 4λ can always be obtained if the corresponding Hadamard matrix exists.