Numerical and data-driven uncertainty quantification of process parameters on clinching joint geometries

Abstract

Clinching is a prevalent mechanical joining method involving clamping and locking two sheet work pieces using a punch and die. Numerical simulation offers a cost-effective alternative to time-consuming experiments for design. Our study employs an ABAQUS explicit dynamic model with remeshing to accurately replicate clinching joint geometry. However, the challenge of hard-to-measure parameters and the investigation of these parameters is limited leading to different simulation results compared to the millimeters-scale experiments. To address this, we apply PyMC3-based uncertainty quantification (UQ) to explore the material parameter effects on clinching dimensions. Our data-driven Bayesian inference models highlight friction and high-plastic-strain flow stress as significant geometry influencers. We propose estimating challenging-to-measure parameters from experiments, leveraging UQ for confident parameter intervals. Through PyMC3, our research offers insights into parameter impact on clinching dimensions, enhancing numerical simulations for process design optimization.