Proceedings of the

The 33rd European Safety and Reliability Conference (ESREL 2023)
3 – 8 September 2023, Southampton, UK

Analysis of Efficiency in Response Surface Designs Considering Orthogonality Deviations and Cost Models

Marco Arndta and Ing. Martin Dazerb

Institute of Machine Components, University of Stuttgart, Germany.


Orthogonality in DoE favors non-correlated effects and minimizes their confidence intervals. However, accidental or deliberate deviation from orthogonality is often possible on the one hand, and sometimes even desirable within reliability demonstration testing. Based on an investigation of orthogonality deviations (errors in factor levels) with respect to performance quantities such as a priori power and scaled prediction variance, the result of a study for DoE costs is put into perspective in this paper. In the course of this, the accuracy of regression models in CCDs as well as their likelihoods of detecting them correctly are compared to a cost model, which ultimately gives orthogonality deviations a price tag. Thus, favorable combinations of design modifications regarding orthogonality are measured within a trade-off.

Keywords: Design of experiments, Reliability engineering, Cost modelling, Test efficiency, Orthogonality deviations, Statistical power analysis, Response surface methodology.

Download PDF