One-Step Screening and Process Optimization Experiments

It is often recommended that process optimization experiments be performed sequentially, starting with a screening design and followed by additional experiments to deconfound interactions, to proceed along the path of steepest ascent, or to fit a response surface. In many situations, however, experimenters do not have the luxury of follow-up experiments, and they must base their improvement recommendations on the result of one experiment. Better than choosing between an optimization or screening design, recent literature has shown that certain orthogonal array experiments can be used simultaneously for screening and optimization. This article describes these experiments and gives examples of one-step screening and optimization.