A probabilistic diffusion weld modeling framework

Physics-based modeling of critical diffusion welds is problematic at best and, in practice, semi-empirical approaches are employed. This work reviews existing pore closure models identifying their shortcomings vis-a-vis actual manufacturing environments. A framework is developed that incorporates realistic manufacturing process attributes such as surface topography into pore closure models. Relevant quantities are represented as distribution functions instead of deterministic values, and manufacturing attributes are then correlated to parameters in these distribution functions. Using a Monte Carlo approach, the distribution of residual joint porosity as a function of both manufacturing attributes and bond process conditions (time, pressure, and temperature) can be derived. Existing models do not capture joint strength, so an additional objective of this work is to model the relationship between residual joint porosity and joint impact strength by applying probabilistic failure models. Finally, this overall approach is applied to model impact strength data of diffusion welds in Ti-6Al-4V.