Plackett-Burman Design

Highly efficient screening design for many factors in multiples of 4 runs — only main effects identifiable.

Also known as: Plackett-Burman, PB design

Plackett-Burman designs (1946) are two-level screening plans whose run count is a multiple of 4 — e.g. 12, 20, 24, 28 runs. With $N$ runs one can screen up to $N - 1$ factors. They shine when many factors (10–20) must be checked for significance with minimum effort.

Efficiency: $N - 1$ factors in $N$ runs, where $N \in {12, 20, 24, 28, \dots}$
Resolution III — main effects are partially aliased with many two-factor interactions
Complex alias structure — each interaction is spread across many main effects rather than 1:1 as in classical 2ᴵᴵᴵ designs

Used in

DoE Planner
DoE Advisor

In the Algorithm Lab

doe/doe-planner

Sources

Plackett, R. L. & Burman, J. P. — The Design of Optimum Multifactorial Experiments, Biometrika, 1946
Montgomery, D. C. — Design and Analysis of Experiments, 10th Edition, Chapter 8