An application of a mixed-level OMARS design to a polymerization experiment

Orthogonal minimally aliased response surface or OMARS designs are a new family of experimental designs, bridging the gap between definitive screening designs and traditional response surface designs. Their technical properties render OMARS designs suitable to combine a screening experiment and a response surface experiment in one. In this article, we describe an application of a 24-run mixed-level OMARS design to study the effects of eight factors on the output of a polymerization experiment. The goal of the experiment is to optimize the production process of a new resin that will be used as a vascular stent adhesive. We analyze the data from the stent adhesive experiment using a design-based two-step analysis method that exploits the orthogonality properties of the OMARS designs and using a computationally efficient way of conducting an all subset regression, called MIO.