Original scientific paper

Estimating a PDF of parameters of oxide inclusions on fatigue fracture surface of standardized specimen

Ana Bižal ; University of Ljubljana, Faculty of Mechanical Engineering, Askerceva 6, SI-1000 Ljubljana, Slovenia
Jernej Klemenc ; University of Ljubljana, Faculty of Mechanical Engineering, Askerceva 6, SI-1000 Ljubljana, Slovenia
Matija Fajdiga ; University of Ljubljana, Faculty of Mechanical Engineering, Askerceva 6, SI-1000 Ljubljana, Slovenia

Abstract

Studies show that, if present, inhomogenities should not be neglected when predicting the fatigue life of cast parts. Accounting for their presence presents a problem due to random nature of their location, distribution and size. Knowledge of statistical characteristics of inhomogenities is fundamental for describing their effect on fatigue life. This paper presents a statistical analysis of oxide inclusions which caused fatigue failure of AlSi9Cu3 specimens. Geometrical characteristics were described using geometry parameters and L. Three different mixed PDFs were used for modelling the distribution of the selected geometry parameters and multivariate Gaussian function was found to be the most appropriate. For estimating unknown parameters of the mixture, modified EM algorithm and REBMIX were used and compared. Results identified modified EM algorithm as more suitable. Bootstrap analysis was carried out for the chosen joint PDF and algorithm, for assigning a measure of variation of the estimated PDF’s parameters.

Keywords

EM algorithm; finite mixture distributions; inclusions; REBMIX algorithm; statistical distribution

Hrčak ID:

112317

URI

https://hrcak.srce.hr/112317

Publication date:

20.12.2013.

Article data in other languages: croatian

Visits: 2.033 *