Probabilistic assessment of VHCF data as pertaining to concurrent populations using a Weibull regression model

Abstract

The Weibull regression model proposed by Castillo and Canteli to evaluate fatigue results represents a possible and adequate option to be used for the assessment and prediction of very high cycle fatigue (VHCF) lifetimes. Among others, this model provides a probabilistic definition of the S‐N field for determining failure mechanisms, ie, internal and surface ones, based on extreme value distributions of the Weibull family for minima, as well as the existence of an asymptotic fatigue limit and the capability to reduce the S‐N field to a single cumulative distribution function by considering the normalized variable V = (log N‐B) (log Δσ). In this way, both dual fracture mechanisms characterizing the VHCF data can be adequately interpreted and handled as independent distributions in such a particular and complex case of concurrent populations, known statistically as a confounded data problem. Once the model parameters of both normalized cumulative distribution functions are estimated, the probability of failure for any of both failure mechanisms at whatever stress range can be determined by applying a back normalization to the original Wöhler field. Two examples of application are presented, the first one to introduce the proposed methodology by means of an example with simulated data to show the reliability of the proposed approach to fit correctly the model parameters assumed for the simulation, and the second one, using real results from a former external experimental VHCF program. Finally, a test strategy, which optimizes the planning of the fatigue program, is suggested.

Fracture mechanical characterization of the initiation and growth of interior fatigue cracks

“…b), there is a higher probability of finding other smaller defects and interior cracks. This result points to the importance of arriving at a statistical analysis of interior cracking . At stress amplitudes between 140 and 185 MPa, the Mg alloy showed more interior cracks originating from smaller inclusions (between 40 μm × 10 μm and 120 μm × 40 μm), not consisting of Al 2 O 3 (Fig.…”

On the Influence of Control Type and Strain Rate on the Lifetime of 50CrMo4

“…This does not mean that the homoscedastic log-normal distribution is a good model overall. Actually, problems about model consistency and accordance to reality have been brought up in literature [81,[125][126][127][128][129][130][131]. At a significance level of 5%, the Shapiro-Wilk test rejects normality for three out of twelve data sets.…”

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

“…The most obvious explanation for this bias is the mismatch between the model employed and the true fatigue behavior of the springs under investigation. This mismatch has been discussed in literature [ 3 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 ] but no consensus has been reached yet. In future investigations, ACEs may be used to compare the predictive power of different models without prior assumptions regarding the true fatigue behavior.…”

“…A key reason for this is the presence of multiple competing failure mechanisms. Problems arising from this have been addressed in literature [ 3 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 ], however consensus over a suitable solution has not yet been reached. In this investigation, a log-normal distribution in the direction of the number of cycles is assumed above and below the kink point.…”