A genetic algorithm-assisted inverse convective heat transfer model for tailoring weld geometry

Although heat transfer and fluid flow models have provided significant insight about the welding processes and welded materials, currently they are not widely used, mainly because of two difficulties. First, the model predictions do not always agree with experiments because the values of energy absorption efficiency and other parameters cannot be prescribed from scientific principles. Second, the available models are unidirectional and cannot currently predict welding variables necessary to attain a target weld attribute. Here we provide a rigorous proof that the heat transfer and fluid flow models can be combined with an appropriate genetic algorithm (GA) to enhance reliability of computational results and achieve inverse modeling capability. The new capability enables systematic tailoring of weld attributes based on scientific principles. In particular, the GA-based optimization of arc efficiency, arc radius, effective thermal conductivity, and effective viscosity using a limited volume of experimental data led to superior weld geometry computations for a wide variety of welding conditions. Furthermore, the inverse model's ability to calculate multiple combinations of arc current, voltage, and welding speed needed to achieve a target weld geometry was developed and rigorously tested by welding experiments.