Electronic International Standard Serial Number (EISSN)
Models consider ideal and simplified situations that will never be met in the real case. The process of comparing model predictions and experimental observation is in the basis of scientific research. This comparison is however complicated because of the uncertainties of the model input data and the difficulty to control the accuracy of the tests and to obtain a significant statistical sampling. Moreover, there isn't yet a consensus on a validation parameter. This paper presents a three-step validation procedure that allows quantifying the application limits of a two-dimensional stress model in a three-dimensional situation. A global uncertainty model is calculated comprising the uncertainty of the model and also the uncertainty coming from the experimental results. The EN number, a statistical magnitude for interlaboratory comparisons, is used to analyse the compatibility between the experimental and theoretical results. Finally, a bootstrapping method is proposed to calculate the coverage interval of the sampling and determine if new experiments should be carried out. Numerical results of this new validation procedure are provided for the case under study. It is also demonstrated that the computed uncertainty budget can be a useful tool to enhance the two-dimensional model by enlarging its uncertainty limits.
Model validation Strain gauge measurement Stress concentration factor Uncertainty analysis