Modeling and uncertainty quantification of material properties in additive manufacturing

Determination of optimal process parameters for the additive manufacturing (AM) process requires use of simulation models. Quantifying the uncertainty in AM process plays an important role in the quality control of additively manufactured products. This work presents an uncertainty quantification framework to model and quantify the variability of macroscale material properties due to multiple uncertainty (aleatory and epistemic) sources present in the AM simulation process. A multi-scale multi-physics simulation model is developed first to simulate the additive manufacturing process. The melt pool profile obtained from macroscale finite element analysis (FEA) is coupled with a microscale cellular automata model to predict the microstructure evolution during solidification. Surrogate model is created to replace the expensive FEA model and surrogate model error is also considered. Based on the simulation model, various sources of uncertainty are aggregated to quantify the uncertainty in the grain size distribution of the microstructure. The contributions of the various sources of uncertainty to the uncertainty of microstructure grain size distribution are analyzed using variance-based global sensitivity analysis. The results show that the proposed approach can effectively perform UQ of the AM process and the uncertainty in the grain size distribution is mainly affected by material properties and grain growth parameters.