Artificial neural network modeling and VFT correlation of experimental dynamic viscosity of the 2-(2-ethoxyethoxy)ethanol + 2-propanol binary mixtures at high pressures and temperatures

Abstract

This work reports experimental dynamic viscosity (n) data for the binary mixtures of 2-(2-ethoxyethoxy)ethanol (Diethylene glycol monoethyl ether) and 2-propanol at high temperatures and high pressures. A falling-body viscometer was employed to carry out the measurements for seven compositions of the studied binary mixture, as well as for pure 2-propanol, over a temperature range of 293.15–353.15 K and pressures up to 70 MPa, with an experimental uncertainty of +-3%. At 0.1 MPa, the dynamic viscosity was measured using a classical Ubbelohde capillary viscometer with an uncertainty of +-1%. The high-pressure dynamic viscosity experimental data were correlated using the Vogel-Fulcher-Tammann (VFT) equation. To describe the viscosity of the binary system as a function of composition, temperature, and pressure, a combining rule based on the logarithm of viscosity was employed. Additionally, an Artificial Neural Network (ANN), a machine learning approach known for its accuracy in capturing nonlinear relationships, was applied to model the dynamic viscosity of the studied binary mixture. The dynamic viscosity behaviour of the analysed binary system was attributed to variations in free volume and the disruption or weakening of hydrogen bonds.