ABSTRACT
Historically, design of experiments (DoE) has been the most important statistical tool for process development, and it continues to be of vital importance today. In this chapter, classical DoE is introduced with fractional factorial design as its foundation. Designs are defined and categorized into three types: Screening, characterization, and optimization. Screening and characterization designs are chosen based on their resolution. Center points are added to allow for replication and test for linear lack-of-fit. Optimization designs (response surface methods) are developed to ensure curvature in the response is properly modeled. Case studies from bioassay, bioprocess, and purification development are provided. A checklist of important considerations in the design phase is given and a demonstration of the projection property is provided. Finally, regression analysis is introduced and the analysis for one or more responses variables using a linear model is detailed.
