PSAM 16 Conference Paper Overview
Welcome to the PSAM 16 Conference paper and speaker overview page.
Lead Author: Marco Arndt Co-author(s): Philipp Mell, M.Sc.; E-mail: philipp.mell@ima.uni-stuttgart.de
Dr.-Ing. Martin Dazer; E-mail: martin.dazer@ima.uni-stuttgart.de
Prof. Dr.-Ing. Bernd Bertsche; E-mail: bernd.bertsche@ima.uni-stuttgart.de
| Generic effects of deviations from test design orthogonality on test power and regression modelling of Central-Composite Designs |
|---|
| In the context of design of experiments (DoE), for many cases the quantitative dependency of a nonlinear target parameter on a few factors is to be determined for the related parameter prediction. For these cases, from the group of response surface designs, test plans are used following the structure of Central-Composite Design (CCD). Based on full-factorial test plans, they feature additional test runs in the center of the design space (center run) as well as along the main axes (star run), which yield the required information for a quadratic model while still being highly efficient. The leverage value α predefines the relative directional distance beyond the center run for the star runs. The individual value determination of α as well as the specific arrangement of the test runs in the design matrix follow a generic mathematical approach to match required DoE properties. Here the most essential respective property is orthogonality. It is sufficiently required in order to consider uncorrelated and independent coefficients separately and to establish regression models, guaranteeing the narrowest possible confidence intervals for parameter prediction. It can be complied and determined analytically based on α and the relative amount of individual run types. However, in the current state of research it remains unclear to what extent renewed adjustments in the amount and arrangement of test runs or further deviations from orthogonality have a practicable effect on design efficiency, test power and precision in regression coefficient estimation. This paper presents a parameter study regarding generic orthogonality deviations in CCDs. For this purpose, various orthogonality deviations are mathematically identified, quantified and performed. Subsequently, potentials and deviations in the effect detection are calculated. Finally, tendencies and first recommendations for design adaptations are presented under consideration of parameter prediction and design efficiency. This includes the categorical exclusion of possible orthogonality deviations as well as the quantification of tolerance limits for minor orthogonality deviations. |
Paper MA19 Preview
Author and Presentation Info
Lead Author Name: Marco Arndt (marco.arndt@ima.uni-stuttgart.de)
Bio: Marco Arndt studied Mechanical Engineering at the University of Stuttgart in Germany and received his M.Sc. in 2020. Since 2021, he has been a researcher in the Reliability Engineering Department at the Institute of Machine Components at the University of Stuttgart and pursues his PhD studies. In his research, he investigates experimental designs and testing strategies with a focus on efficiency improvement to develop highly efficient and adaptable experimental designs.
Country: Germany
Company: PhD candidate at the University of Stuttgart, Germany
Job Title: Research Assistant in Reliability Engineering
Bio: Marco Arndt studied Mechanical Engineering at the University of Stuttgart in Germany and received his M.Sc. in 2020. Since 2021, he has been a researcher in the Reliability Engineering Department at the Institute of Machine Components at the University of Stuttgart and pursues his PhD studies. In his research, he investigates experimental designs and testing strategies with a focus on efficiency improvement to develop highly efficient and adaptable experimental designs.
Country: Germany
Company: PhD candidate at the University of Stuttgart, Germany
Job Title: Research Assistant in Reliability Engineering