Predictive Modeling of Wettability Characteristics in Developed Shielded Metal Arc Welding Electrode Coating Fluxes: A Regression and Artificial Neural Network Approach
摘要
The coating fluxes used in SMAW electrodes for nuclear power plants have been specially developed. The study investigates various characteristics such as floatation coefficient, work of adhesion, spread area, and contact angle at a high temperature of 1323 K. The flux compositions, comprising Na3AlF6–CaO–Al2O3–SiO2, are developed using the extreme vertices design approach, resulting in 26 different compositions. Furthermore, the surface tension of these fluxes was calculated as part of the investigation. Regression analysis was employed to predict the influence of individual constituents and their interactions on the wettability properties of the flux coatings. Additionally, artificial neural network (ANN) models were developed and compared to regression analysis to assess prediction accuracy. The fluxes underwent further characterization through X-ray diffraction (XRD) and Fourier transform infrared (FTIR) analysis to identify the phases present. The structural analysis of the molten slag was conducted by examining the powdered samples obtained through quenching. The FTIR results indicate the presence of [SiO4]4− tetrahedral units and [AlO4]5− tetrahedral units, suggesting the formation of a network structure. Additionally, it is observed that CaO acts as a network breaker, causing the release of non-bridging oxygen (O−) at the expense of bridging oxygen (O0). The individual components CaO, Na3AlF6, Al2O3, and SiO2 were found to cause a decreased effect, while the binary interaction of the coating constituents shows a significant effect on the contact angle and floatation Coefficient. Individual interactions, CaO, SiO2, Al2O3, and Na3AlF6 exhibit a positive impact, while the binary interactions of Na3AlF6·SiO2, Na3AlF6·Al2O3, CaO·Na3AlF6, Al2O3·SiO2, CaO·SiO2, and CaO·Al2O3 have been found to have a decreasing effect on the spread area and work of adhesion.