A Methodology for Consideration of Uncertainty in Lifetime Design of Experiments

After a short introduction, approaches are presented which allow the use of Design of Experiments (DoE) in the case of non-normal distributed data (Weibull, etc.). Therefore that models are called Lifetime-DoE models ( L-DoE models). In order to take uncertainty into account, the classic DoE approach is also extended by integrating the calculation and simulation of confidence intervals (CI). Four methods were used: Fisher, Monte-Carlo, Pseudo and Bootstrap. Finally, the methodology was shown using a fictitious example. It turned out that the Pseudo-CI provides the largest and thus most conservative CI. While the Fisher and Monte-Carlo approaches have the smallest CIs.