Uncertainty quantification in performance evaluation of manufacturing processes

This paper proposes a systematic framework using Bayesian networks to integrate all the available information for uncertainty quantification (UQ) in the performance evaluation of a manufacturing process. Energy consumption, one of the key metrics of sustainability, is used to illustrate the proposed methodology. The evaluation of energy consumption is not straight-forward due to the presence of uncertainties in different variables in the process and occurring at different stages in the process. Both aleatory and epistemic sources of uncertainty are considered in the UQ methodology. A dimension reduction approach through variance-based global sensitivity analysis is proposed to reduce the number of variables in the system and facilitate scalability to high-dimensional problems. The proposed methodologies for uncertainty quantification and dimension reduction are demonstrated using two examples - an injection molding process and a welding process.