STATISTICAL CHARACTERIZATION OF RANDOM CRACK PATTERNS CAUSED BY THERMAL FATIGUE

Thermal fatigue experiments performed with an austenitic steel give rise to an irregular pattern of cracks on the specimen surface. The investigations are aimed at characterizing the patterns in quantitative terms and at discerning trends with increasing numbers of cycles. Statistical methods based on different models of stochastic geometry are applied to take into account the random influence in pattern formation. Descriptive statistics as well as stochastic models are used to characterize the damage level. The stochastic models, such as fibre processes and random mosaics, will finally correlate the physical damage process with the random crack pattern observed. It is shown in this paper how the statistical characteristics of the random crack patterns can be related to the statistical characteristics of the stochastic models and the way the physical nature of the damage process influences these quantities.