Thermal property characterization and modeling of SMAW electrode coating flux using ANN and regression analysis

The design and development of SMAW electrode coating fluxes, along with their thermal property characterization, are discussed. Using an extreme vertices design approach, 27 flux mixture combinations have been prepared. The TGA-DSC and hot disk analysis are done to quantify the thermal response of the fluxes. Results show that increase in CaF2 and SiO2 reduces weight loss. Increase in CaO increases weight loss, whereas increase in BaO shows minimal effect on weight loss. Enthalpy change of flux was found to influence mainly with CaF2 and CaO. Thermally stable flux has lower volatile compounds such as crystallization water and moisture, which helps to minimize gaseous entrapment in weld pool. An increase in CaF2 shows slight decreases and then increase in thermal diffusivity. The increase in thermal diffusivity leads to instant heat flow in the coating. Regression analysis is done, and the effect of individual and interaction of flux constituents on thermal properties is discussed. The artificial neural network is employed for predictive modeling in the domain of the mixture design approach, its prediction accuracy is compared with regression analysis. Water vapor solubility of slag was estimated between 3.169 and 3.947 and sulfide capacity of slag was − 2.479 to −1.892.