A Smart Model for Welding Engineers to Achieve Target Fusion Zone Geometry and Microstructure using Parallel Genetic Algorithm

A smart, bi-directional model has been developed to determine three-dimensional temperature profiles, weld geometry and the transformed phase fraction during gas tungsten arc (GTA) butt welding. The model is capable of estimating input parameters such as the net input power, welding speed, effective thermal conductivity, and, activation energy, preexponential factor and exponent in JohnsonMehl-Avrami (JMA) equation for phase transfer calculations, using limited number of experimental data and parallelized Parent Centric Recombination (PCX) based Generalized Generation Gap (G3) genetic algorithm as a multivariable optimization technique. Thus, the model can satisfy both the industrial and research requirements of specifying a window of input parameters to obtain specific weld geometry and calculating uncertain input parameters required during mathematical modeling. The calculated shape and size of the fusion zone and the austenite phase fraction were in fair agreement with the experimental results. The effect of various parameters on convergence and reliability of the proposed multiple deme based G3 model was also studied. The model gave increased reliability and faster convergence.