Analysis of the Non-Newtonian Behavior and Viscosity of GNSs-CuO/Liquid EG Hybrid Nanofluid: An Experimental and Feed-Forward ANN Study

• Published in: International journal of thermophysics 2023-07, Vol.44 (7), Article 103
• Main Authors: Wang, Jing, Karimipour, Arash, Sajadi, S. Mohammad, D’Orazio, Annunziata, Bagherzadeh, Seyed Amin, Abdollahi, Ali, Inc, Mustafa
• Format: Article
• Language: English
• Subjects: Algorithms; Artificial neural networks; Classical Mechanics; Condensed Matter Physics; Copper oxides; Ethylene glycol; Geophysics; Graphene; High temperature; Industrial Chemistry/Chemical Engineering; Machine learning; Nanofluids; Nanoparticles; Physical Chemistry; Physics; Physics and Astronomy; Shear stress; Smoothness; Stability analysis; Temperature effects; Thermodynamics; Thermophysical properties; Viscosity; Zeta potential
• ISSN: 0195-928X; 1572-9567
• DOI: 10.1007/s10765-023-03196-0

The knowledge of thermophysical properties of heat transfer systems such as their dynamic viscosity is of interest for their efficiency improvement and miniaturization. Hybrid nanofluids are obtained by the suspension of nanoparticles in base fluids. In this study, the effects of temperature (at four levels, from 5 °C to 65 °C) and mass percentage (from 0.005 % to 5 %) were investigated on the dynamic viscosity of two types of graphene nanosheets-copper oxide/ethylene glycol (GNSs-CuO/liquid EG) hybrid nanofluids, i.e., 25 %GNSs-75 %CuO/liquid EG and 75 %GNSs-25 %CuO/liquid EG hybrid nanofluids. Transmission electron microscopy (TEM) and Zeta Potential were analyzed to investigate the shape, size, and stability of the nanoparticles. Results showed the non-Newtonian behavior and very high stability of the studied GNSs-CuO/liquid EG hybrid nanofluids. This behavior was more remarkable at high temperatures, high mass concentrations, and low shear rates. In similar temperatures and mass fractions, GNSs resulted in a higher increase in viscosity and relative viscosity compared to CuO nanoparticles. Furthermore, a feed-forward artificial neural network (ANN), as a robust machine learning algorithm, was used to predict the dynamic viscosity of both studied hybrid nanofluids having their mass percentage, temperature, and shear stress as network inputs. The acceptable precision of the ANN model and its generalization were observed through the comparison of experimental and predicted data and the smoothness of the obtained model surfaces, respectively.