The effect of factor interactions in Plackett-Burman experimental designs: Comparison of Bayesian-Gibbs analysis and genetic algorithms

Abstract

A genetic algorithm has been developed in order to estimate not only the main effects but also the association of terms when analyzing the influence of experimental factors through a Plackett-Burman design of experiments. The results for a series of simulated systems as well as experimental examples show excellent agreement with a Bayesian-Gibbs approach. The Plackett-Burman design is usually employed for screening, but its performance depends on the assumption that the interaction effects are negligible. Simulations allow one to analyze the effect of increasing interactions on the significance of main factors when Plackett-Burman designs are processed by neglecting factor associations.