Artificial intelligence prediction of natural convection of heat in an oscillating cavity filled by CuO nanofluid

Abstract

BackgroundIn the last five years, a new generation of advanced numerical methods was developed to reduce computational costs and improve the prediction process. The combination of Artificial Intelligence and traditional computational methods is the best sample of this generation. Employing this method can expand the horizon of numerical modeling. This method is also suitable for complex processes such as natural convection and oscillating heat transfer. This case study tries to peruse the effects of variable boundary conditions such as magnetic field, angle, and nanofluids volume fraction on the heat transfer. Also, the impact of oscillation has been studied. MethodsTo accomplish these aims, numerical modeling of natural convection based on Boussinesq approximation was used. Techniques from machine learning were employed to develop a predictive tool that utilizes the data generated by computational analysis of the problem. The Ra and Ha were considered in the ranges of 103<Ra<106 and 0<Ha<40, respectively. Oscillation frequencies between 20 and 50 Hz and amplitudes of 0.5, 1, and 2 mm were studied. FindingsThe result showed that Nu increased in initial steps and then decreased at a constant Ha by increasing the rotation angle. It appeared that the amplitude of oscillation has a pronounced effect on the heat transfer rate. The results show that the Ha could affect streamlines of fluid. By increasing Ha=0 to Ha=52, the maximum velocity point in the close enclosure moved 14% closer to the hot side.