On rheological properties of environmentally friendly inorganic systems and their modeling by artificial neural networks

Abstract

This work aims to investigate the rheological properties, namely dependence of dynamic viscosity on temperature, chemical composition, and shear rate, of environmentally friendly inorganic systems using a high-temperature rotational viscometer up to 1550 °C. The liquidus and start and end softening temperatures of these systems were also studied. The environmentally hazardous calcium fluoride in the amount of up to 6 wt% was substituted by other components (B2O3, TiO2, and Na2O) to preserve the original utility properties of the investigated systems (low liquidus temperatures and viscosities). The effect of alternative additives ranging from 2 to 6 wt% on the required properties was more beneficial than that of fluoride. The most significant reduction in liquidus temperature of up to 185 °C was achieved by adding 6 wt% B2O3 while maintaining a low viscosity value. The addition of CaF2 (up to 6 wt%) had the least effect, lowering the liquidus temperature by only 22 °C as compared to the original system. In the case of TiO2 addition, the dependence of viscosity on chemical composition was non-linear and complex to predict with existing models. Therefore, it was modeled using artificial neural networks. The predicted viscosity values for a given temperature and chemical composition were in good agreement with the experimentally obtained values, as the maximum relative error between the measured and calculated viscosity values was less than 5%. The characterization of the internal structure of the investigated systems was performed by Energy Dispersive X-Ray (EDX), X-Ray Diffraction (XRD) analyses and Scanning Electron Microscopy (SEM).

Keywords

• Viscosity
• Oxide system
• Liquidus temperature
• Softening temperature
• Calcium fluoride
• Artificial neural network