Reliable prediction of thermophysical properties of nanofluids for enhanced heat transfer in process industry: a perspective on bridging the gap between experiments, CFD and machine learning

• Published in: Journal of thermal analysis and calorimetry 2023-06, Vol.148 (12), p.5859-5881
• Main Authors: Ullah, Atta, Kilic, Mustafa, Habib, Ghulam, Sahin, Mahir, Khalid, Rehan Zubair, Sanaullah, Khairuddin
• Format: Article
• Language: English
• Subjects: Analytical Chemistry; Artificial neural networks; Availability; Chemistry; Chemistry and Materials Science; Computational fluid dynamics; Fluids; Heat transfer; Inorganic Chemistry; Machine learning; Mathematical models; Measurement Science and Instrumentation; Nanofluids; Nanoparticles; Particle shape; Physical Chemistry; Physical properties; Polymer Sciences; Power plants; Shear rate; Thermal stability; Thermophysical properties; Working fluids
• ISSN: 1388-6150; 1588-2926
• DOI: 10.1007/s10973-023-12083-7

In recent years, traditional fluids are frequently being replaced by efficient heat transfer fluids showing physical and thermal stability. One such category of fluids is called nanofluids, in which solid nanoparticles (metals or their oxides, nitrides and so on) are suspended in a base fluid resulting in enhanced heat transfer characteristics. These nanofluids are increasingly used in low to medium temperature applications toward intensification of process and power plants by reducing the overall size and heat losses. However, as compared to a pure fluid, prediction of thermal and physical properties of nanofluids is a challenge due to unavailability of a general model. These thermal and hydraulic characteristics are strongly dependent upon multiple factor including particle size, particle volume concentration, particle composition, particle shape, temperature, base fluid material, pH and shear rate. Keeping these challenges in mind and availability of modeling tools, we first summarize and comment on popular correlations available to predict thermal and physical properties of nanofluids. Then, a general approach for carrying out reliable computational fluid dynamics (CFD) simulations is presented. The limitation of a general correlation of physical properties for input into CFD code can be overcome by use of machine learning (ML) tools such as artificial neural networks (ANN) taking advantage of the huge databank of physical properties of nanofluids. The use of ML to compliment CFD for accurate and reliable simulation of systems employing nanofluids as working fluids is highlighted at the end as potential emerging areas of research. Graphic abstract