Deriving local Nusselt number correlations for heat transfer of nanofluids by genetic programming

Created by W.Langdon from gp-bibliography.bib Revision:1.8051

@Article{GUZMANURBINA:2023:ijthermalsci,

• author = "Alexander Guzman-Urbina and Kazuki Fukushima and Hajime Ohno and Yasuhiro Fukushima",
• title = "Deriving local Nusselt number correlations for heat transfer of nanofluids by genetic programming",
• journal = "International Journal of Thermal Sciences",
• volume = "192",
• pages = "108382",
• year = "2023",
• ISSN = "1290-0729",
• DOI = "doi:10.1016/j.ijthermalsci.2023.108382",
• URL = "https://www.sciencedirect.com/science/article/pii/S1290072923002430",
• keywords = "genetic algorithms, genetic programming, Nanofluids, Nusselt number, Machine learning",
• abstract = "The application of nanoparticles in the design of heat exchange systems is anticipated to significantly enhance energy efficiency across the industrial, commercial, and residential sectors. A nanofluid results from mixing a base fluid and nanoparticles. These nanoparticles generally are made from various materials, including metals, metal oxides, and polymers. To design nanofluid-assisted energy systems, it is necessary to estimate the heat transfer achieved by the nanofluids. Most models developed to describe the heat transfer of fluids require estimating the Nusselt number (Nu), representing the ratio of convective heat transfer to conductive heat transfer in a given system. Previous work in nanofluids has focused on modeling equations for the average Nusselt number (Nuavg) in tube systems. However, an equation that allows the local Nusselt number (Nux) to be calculated is more versatile than one which calculates the Nuavg since it can be applied to any length of circular tubes. In this study, we derive equations using Genetic Programming (GP) to estimate the local Nusselt number of nanofluids (Nux,nf) that flow through horizontal circular tubes. The method uses an evolutionary algorithm to generate correlation equations for exploring the interaction of the flow regime, flow properties, system configuration, and nanoparticle properties. It comprises creating a population of candidate equations, selecting the best ones through fitness evaluation, and combining them through genetic operators (Selection, crossover, and mutation) to create a new generation of equations. Acceptable Nux,nf models (R2>0.9) were obtained with GP tree structures larger than three levels (depth 3). Findings from the Nux,nf correlations obtained show that the determinant variables for the model were the Reynolds and Prandtl numbers. This implies that the effect of the nanofluids is driven mainly by alterations made by the nanoparticles to the inertial forces of the flow and the thermal diffusivity. The results of this study highlight the potential of this machine-learning-based approach to provide insight into the physicochemical mass and heat transfer mechanisms",

}

Genetic Programming entries for Alexander Guzman-Urbina Kazuki Fukushima Hajime Ohno Yasuhiro Fukushima

Citations