Numerical investigation of temperature distribution of autogenous tig welding process

Abstract

It is common to find welded steel structures in the industry, either in producing of parts or in cases that involve repairing them. There is several welding processes used to join metal parts and the study of these techniques is essential, both for choosing welding parameters, as well as for choosing the most appropriate method for each industrial application and the type of structure that will need repair by the welding process. This work focuses on studying the autogenous TIG welding process. The large thermal input in a welding process produces microstructural alterations and residual stresses that can harm the welded region. For this reason, the tireless search for the improvement of welding processes and the main effects caused to the material after the execution of the weld is justified. Understanding the temperature variation throughout the welded material can prevent failures from occurring. Experimental research processes are quite time consuming and costly. The use of numerical simulation steps is recurrent to assist in experimental welding processes for acquiring thermal cycles. Closed software such as ANSYS®, ABAQUS®, SYSWELD®, among others, are commonly used for numerical simulation stages of thermal cycle surveys. Numerical methods like; difference finite (FDM), finite element (FEM), finite volume (FVM) and meshless Element-Free Galerkin (EFG) are widely used to survey the welding thermal cycle. Although they are quite robust methods, they need to be complemented. Aiming to develop a simulator in the Fortran language aimed at numerical analysis of temperature distribution due to a welding process. This work uses the element-based finite volume method (EbFVM) as a mathematical pillar, together with structured and unstructured meshes. Four different heat sources; circular, semi-ellipse, semi-ellipsoid and double ellipsoid were used in this work as a numerical investigation target that best suited the autogenous TIG welding process. Parameters such as: electric current, voltage, welding speed, specific heat, thermal conductivity, welding efficiency and plate temperature are fundamental variables for the results to be compared with real experiments. A first test with data collected from the literature was performed to validate the numerical steps. After validating the results and checking the robustness of the code written in FORTRAN®, an experimental stage controlled in the laboratory was necessary. In the experimental stage, the steel used was AISI 409 with dimensions of 160mm x 50mm x 6.3mm. K-type thermocouples were used to acquire local thermal cycles. The thermocouples were fixed at points close to the fused region. Four weld passes were performed on the steel piece with inter-pass temperature

after the first weld pass. Still in the experimental stage, metallographic tests were carried out to measure the dimensions of the weld bead accurately and to be used in the heat sources of the numerical model. The results of the thermal cycles acquired in the experimental process for thermocouples fixed close to or within the thermally affected zone showed results in agreement with the numerical results when comparing the values of peak temperature, heating and cooling time and heating and cooling rate for the four welding passes. For thermal cycles acquired in regions farther from the thermally affected zone, the experimental and numerical results were concordant in the heating and cooling rates and in the heating and cooling time for the two values of electric current used.