Conclusion
Metamodels and industrial applications

Substitute modeling methods have become indispensable in modern engineering, as they reduce the computational cost of uncertainty-related problems. In this article, the metamodel concept and probabilistic methods based on metamodels for calculating reliability have been presented. Several methods are available for sensitivity analysis. This article focuses on global methods based on variance decomposition and, more specifically, on Sobol sensitivity indices. This analysis was applied to a kriging model using a Matlab utility.

A probabilistic method for assessing the reliability of T-CSP technology packages is presented. Reliability is predicted by calculating the fatigue life of the package's solder joints when subjected to accelerated thermal loading. A finite element model of the T-CSP package is produced. By applying the Darveaux model to the crack growth rate,...