A Smart Bi-Directional Model of Heat Transfer and Free Surface Flow in Gas-Metal-Arc Fillet Welding for Practising Engineers

Abstract

In recent years, numerical heat transfer and fluid flow models have provided significant insight about fusion welding processes and welded materials. A major problem in their practical use is that several input parameters cannot be easily prescribed from fundamental principles. Available inverse models of fusion welding for the determination of these unknown parameters have ignored important physical processes such as convection in the weld pool to make computational tasks tractable. As a result, these models are not very different from the neural network type models that are not required to obey any physical law. A smart, bi-directional, numerical model has been developed to determine three-dimensional temperature and velocity profiles, weld geometry and the shape of the solidified weld reinforcement surface during gas metal arc (GMA) welding of fillet joints. Apart from the transport of heat from the welding arc, additional heat from the metal droplets was also considered in the model. The model is capable of estimating unknown parameters such as the arc efficiency, effective thermal conductivity and effective viscosity from a limited number of data on weld geometry based on multivariable optimisation. Alternative strategies for the optimisation are examined. The calculated shape and size of the fusion zone, finger penetration characteristic of the GMA welds and the solidified free surface profile were in fair agreement with the experimental results for various welding conditions. In particular, the computed values of the leg length, the penetration depth and the actual throat agreed well with those measured experimentally for various heat inputs. The weld thermal cycles and the cooling rates were also in good agreement with the independent experimental data. The research presented here shows that advances in computational hardware and software have now made construction of smart, bi-directional, large transport phenomena based phenomenological models a useful undertaking.