Experimental Investigation and Modeling of Bead Geometry and Residual Stresses during Submerged Arc Welding of HSLA Steel

Abstract

Submerged arc welding (SAW) is a welding process used to join plates of higher thickness. In SAW process arc is formed between a continuously fed wire electrode and work piece. Weld is formed due to melting of wire and work piece and arc is shielded under a blanket of granular fusible flux. Flux melts during welding process and it provides protection to weld pool from atmospheric agents. Process parameters in SAW bear complex mathematical relationships and soft computing techniques are found to be useful for process modeling and optimization. Bead geometry and penetration of metal are some of the important physical characteristics of a weldment. Various welding parameters like high arc-travel and low arc-travel affect bead geometry as it results in poor fusion of metal. Current, voltage and arc-travel length are responsible for depth of penetration. Other factors affecting penetration is arc-length and arcforce. Bead geometry characteristics play an important rule to determine weldabilty of metal. Factors like open circuit voltage, trolley speed, contact tube to work piece distance, preheat temperature and wire feed rate are studied here. Soft computing is an emerging approach to computing that closely resembles the ability of human mind to reason and learn from an environment of uncertainty and imprecision. In this present work, soft computing based model Artificial Neural Network (ANN) has been developed and implemented to model and optimize the SAW process. Experimentations based on design of experiment have been carried out for different process parameters. Dataset generated from experimentation was employed to create ANN model. ANN model uses ANN algorithm called back prorogation neural network (BPNN) with Levenberg Marquardt (LM). A comparison of various ANN models using different number of layers and neurons was carried out and regression coefficient was found for it. Residual stresses are the type of stresses which exist in a body when no external loads are applied on it. These types of stresses are created in welded joints due to non-uniform heating and cooling and shrinkage of molten metal. There will always be some amount of residual stresses present in any welded component. Tensile residual stresses are detrimental for component as it increases susceptibility of a welded part to premature failure. They accelerate failure due to fatigue and stress induced corrosion cracking. Parameters should be selected such that so as to reduce residual stresses which minimize and avoid failures of welded components. There are various factors which influence residual stresses present in welded joint. They are type of welding technique used, parameters of welding process and properties of metal being welded. Study was undertaken to determine stresses in High strength low alloy steel plates by the help of X-ray diffraction (XRD) technique. It is believed that the work presented here will add significant contribution to existing literature from view point of industrial importance and academic interest.