On the Influence of Ultimate Number of Cycles on Lifetime Prediction for Compression Springs Manufactured from VDSiCr Class Spring Wire

For the generation of fatigue curves by means of fatigue tests, an ultimate number of cycles must be chosen. This ultimate number of cycles also limits the permissible range of the fatigue curve for the design of components. This introduces extremely high costs for testing components that are to be used in the Very High Cycle Fatigue regime. In this paper, we examine the influence of the ultimate number of cycles of fatigue tests on lifetime prediction for compression springs manufactured from VDSiCr class spring wire. For this purpose, we propose a new kind of experiment, the Artificial Censoring Experiment (ACE). We show that ACEs may be used to permissibly extrapolate the results of fatigue tests on compression springs by ensuring that a batch-specific minimum ultimate number of cycles has been exceeded in testing. If the minimum ultimate number of cycles has not been exceeded, extrapolation is inadmissible. Extrapolated results may be highly non-conservative, especially for models assuming a pronounced fatigue limit.

Semiparametric estimation of structural failure time models in continuous-time processes

Summary Structural failure time models are causal models for estimating the effect of time-varying treatments on a survival outcome. G-estimation and artificial censoring have been proposed for estimating the model parameters in the presence of time-dependent confounding and administrative censoring. However, most existing methods require manually pre-processing data into regularly spaced data, which may invalidate the subsequent causal analysis. Moreover, the computation and inference are challenging due to the nonsmoothness of artificial censoring. We propose a class of continuous-time structural failure time models that respects the continuous-time nature of the underlying data processes. Under a martingale condition of no unmeasured confounding, we show that the model parameters are identifiable from a potentially infinite number of estimating equations. Using the semiparametric efficiency theory, we derive the first semiparametric doubly robust estimators, which are consistent if the model for the treatment process or the failure time model, but not necessarily both, is correctly specified. Moreover, we propose using inverse probability of censoring weighting to deal with dependent censoring. In contrast to artificial censoring, our weighting strategy does not introduce nonsmoothness in estimation and ensures that resampling methods can be used for inference.