Thermal and solidification modeling of welding: A design tool approach.

摘要

Modeling and simulation tools can be used in welding to prevent faults such as insufficient melting or undercuts, non homogeneity or undesired phases in the solidified region, corrosion susceptibility and stress reduction in the heat affected zone, or buildup of residual stresses. A successful modeling tool should allow optimization in the design stage via a time efficient virtual experiments and avoid the need for a time consuming series of trial and errors of actual welds. The method developed using an integrated mechanical engineering and materials science approach consists of three stages: the prediction stage, the analysis stage and the refinement stage. The prediction tool, which was developed in LabVIEW™ and ANSYS™ Parametric Design Language (APDL), conducts weldability checks and calculates an initial guess for the welding conditions. The analysis tool is a parametric finite element model of transient heat transfer in the weld, developed in ANSYS™ that additionally provides information for the solidification morphology in the fusion zone. Finally, the refinement tool utilizes control theory criteria in order to optimize the selection of the process parameters. The optimization variables are calculated by the analysis tool and compared with target values; the error between them drives the selection of a better set of process parameters. The analysis and refinement tool are run in a loop until the error between the predicted and desired characteristics is minimized.; The focus of this work was to develop the required methodology and tools, prove the feasibility, and evaluate the efficiency of the suggested approach. Verification experiments were conducted for Plasma Arc and Laser Beam welding. Good agreement was observed between the predicted by finite element analysis and experimental values of the fusion zone geometry as well as the solidification morphology in the case of plasma welding. The predictions for the laser welds were less satisfactory. It is believed that this inaccuracy is related to uncertainty of the actual power density profiles in the experiments and lack of incorporation of liquid convection in the model. The stability of the overall Design Tool was good, and improvements can be achieved with the use of a more robust control scheme.