Experimental Verification of Numerical Computation with Evolved Material Property Model and Sensitivity Analysis on WAAM Distortion using P-GMAW

This paper is dealing with an investigation using nonlinear numerical computation for manufacturing to predict substrate distortion induced by a hollow-shaped Wire Arc Additive Manufacturing (WAAM) process. The WAAM component is designed in a thin-walled rectangular form consisting of twenty-five deposited layers with austenitic stainless-steel wire SS316L as feedstock and SS304 as substrate material. In this investigation, two different material models were investigated which was directly retrieved from numerical computation or virtual manufacturing software database and evolved component material composition obtained from experiment and characterized using Scanning Electron Microscopy (SEM) with Energy Dispersive X-ray (EDX) analysis as well as generated using advanced modelling software JMATPRO for creating new properties including the flow curves. The hollow shape is modelled based on simplified rectangular mesh geometry with identical specimen dimension during experiment. A coupled thermomechanical solution is adopted including phase-change phenomena defined in latent heat, whereby temperature history due to successive layer deposition is simulated by coupling the heat transfer and mechanical analysis. Further, transient thermal distribution is calibrated from experiment obtained from thermocouple analysis at two reference measurement locations. New heat transfer coefficients are to be adjusted to reflect actual temperature change. Prior to simulation execution, a sensitivity analysis was conducted in order to find the optimal number of elements or mesh size towards temperature distribution. Based on these boundary conditions, WAAM distortion is verified using robotic welding system equipped with pulsed power source. The experimental substrate distortion is measured on various points prior to and after the process. It can be concluded based on the adjusted model and experimental verification that using nonlinear numerical computation, the prediction of substrate distortion with evolved material property of component yields closer average result within the relative error less than 11% compared to database material giving doubled inaccuracy up to 22%.