Experimental and ensemble machine learning analyses of heat transfer, friction factor and thermal performance factor of rGO/water nanofluids in a tube

Abstract

This paper explains about the experimental evaluation of thermophysical properties, heat transfer coefficient, Nusselt number, friction factor, and thermal performance factor of water-based reduced graphene oxide (rGO) nanofluids passes through a tube under turbulent flow conditions. The rGO nanoparticles in this study was prepared based on the Hummer's procedure. The thermophysical properties were estimated at different volume loadings from 0.5 % to 2.0 % and at different temperatures from 20 °C to 60 °C, respectively, similarly, the heat transfer and friction factor experiments were conducted in the Reynolds number ranging from 1800 to 21,000. The obtained data of heat transfer coefficient, Nusselt number, friction factor and thermal performance factor was predicted based on the advanced machine learning algorithms of boosted regression tree (BRT), and extreme gradient boosting (XGB), respectively. Results indicated that, at 2.0 % vol. of rGO/water nanofluid the thermal conductivity is raised by 26.95 % at a temperature of 60 °C, viscosity is raised by 63.29 % at a temperature of 20 °C, over base fluid. Moreover, the results are further shown that, at 2.0 % vol. of nanofluid the Nusselt number, heat transfer coefficient, factor in friction, drop in pressure, and pumping power raise of 34.80 %, 44.19 %, 16.73 %, 10.48 %, and 4.55 % at a Reynolds number of 13,705, against water. The correlation coefficient (R2) results predicted from BRT and XGB models for Nusselt number data is 0.9831 and 0.994, respectively. It is understood that the XBG algorithm predicts very accurate values compared to BRT algorithm with a less mean square error. The overall performance of the system was raised by 1.28 -folds than water. Using the experimental data, new Nusselt number and friction factor correlations were proposed.