DOI
https://doi.org/10.25772/9ARP-FT66
Defense Date
2021
Document Type
Dissertation
Degree Name
Doctor of Philosophy
Department
Systems Modeling and Analysis
First Advisor
David J. Edwards
Abstract
Design of experiments is used to study the relationship between one or more response variables and several factors whose levels are varied. Response surface methodology (RSM) employs the design of experiment techniques to decide if changes in design variables can enhance or optimize a process. They are usually analyzed by fitting a second-order polynomial model. Some standard and classical response surface designs are $3^k$ Factorial Designs, Central Composite Designs (CCDs), and Box-Behnken Designs (BBDs). They can all be used to fit a second-order polynomial model efficiently and allow for some testing of the model's lack of fit. When performing multiple experiments is not feasible due to time, budget, or other constraints, recent literature suggests using a single experimental design capable of performing both factor screening and surface response exploration. Definitive Screening Designs (DSDs) are well-known experimental designs with three levels. They are also named second-order screening designs, and they can estimate a second-order model in any subsets of three factors. However, when the design has more than three active factors, only the linear main effects and perhaps the largest second-order term can be identified by a DSD. Also, they may have trouble identifying active pure quadratic effects when two-factor interactions are present. In this dissertation, We propose several methods for augmenting definitive screening designs for improving estimability and efficiency. Improved sensitivity and specificity are also highlighted.
Rights
© The Author
Is Part Of
VCU University Archives
Is Part Of
VCU Theses and Dissertations
Date of Submission
5-13-2021