Wettability study of Developed Silicon based SMAW Electrode Coating Fluxes using SiO 2 -CaO-TiO 2 and SiO 2 -CaO-MgO Ternary System

Abstract

The current research aims to develop and investigate the wettability behaviour of the fluxes used to coat shielded metal arc welding electrodes for offshore applications. Weld characteristics on duplex stainless steel are significantly affected by the flux's wettability characteristics. The wetting properties of SMAW electrode coatings were measured using the sessile drop technique. Twenty-six flux constituents of coatings have been formulated using the mixture design approach. At a temperature of 1423 K, the contact angle, spread area, surface tension, and adhesion work are measured. X-ray diffraction (XRD) analytical techniques and Fourier transform infrared spectroscopy (FTIR) analysis were employed to determine the phases present in various types of fluxes. The influence of electrode coating flux compositions on different wetting parameters was examined using statistical models. It was observed that CaO, SiO2, and TiO2 individually have an increasing impact on contact angle, whereas MgO has a decreasing effect. CaO.MgO, SiO2.MgO, and TiO2.MgO is the only binary constituent increasing the contact angle. CaO and TiO2 exhibit increasing impacts, whereas SiO2 and MgO have a decreasing impact on the spread area. CaO.SiO2, SiO.TiO2, SiO.MgO, and TiO2.MgO interactions have increasing effects, while other binary interactions show decreasing effects on the spread area. Individual constituents have a positive effect on surface tension. MgO is the only constituent that increases the work of adhesion, and other constituents decrease it. Interactions like CaO, SiO2, CaO, TiO2, and SiO2.TiO2 has an increasing impact on adhesion work, while other binary interactions show a negative effect. Utilizing an artificial neural network approach, the mean square error (MSE) and mean absolute percentage error (MAPE) values for the predicted outcome were effectively minimized. ANN model prediction was compared to regression prediction.