Integrating ISPH simulations with machine learning for thermal radiation and exothermic chemical reaction on heat and mass transfer in spline/triangle star annulus

Abstract

This work integrates incompressible smoothed particle hydrodynamics (ISPH) with machine learning (ML) to examine the uses of two different domain shapes, spline star-domain, and triangle star-domain, on double diffusion of nano-encapsulated phase change material (NEPCM). The effects of thermal radiation and exothermic chemical reactions are conducted in this work. The hexagonal-shaped is located inside the center of the star-shape carrying (Th&Ch). The left/right walls of the star shape are kept at (Tc&Cc) and the other walls have an adiabatic condition. It is recommended to examine the smoothness of closed curves in the heat and mass transfer by changing the triangle to spline curves. The present investigations of transmission of heat and mass of NEPCM within a complicated star shape can be employed in cooling systems of electronic/battery, heat exchangers, food processing, and chemical engineering. The ISPH simulations are executed for the pertinent parameters, Darcy number, Frank-Kamenetskii number, Rayleigh number, Dufour number, Soret number, and thermal radiation parameter. The performed simulations showed the significance of the spline star domain compared to the triangle star domain in smoothening the nanofluid flow and double-diffusive convection in an annulus. Frank-Kamenetskii number is a promising factor in the enhancement of heat transfer and controlling the phase change zone within an annulus. At a spline-star cavity, the maximum velocity rises by 32.32 % and 54.95 % when the Frank-Kamenetskii number grows from 0 to 5, and thermal radiation increases from 0 to 5. Soret number works effectively in enhancing the distribution of concentration at large temperature differences. Machine learning plays a vital role in many engineering applications. Hence, ML's powerful capabilities to predict the average Nu‾ and Sh‾. In the proposed ML model, the dimensionless time and thermal radiation parameter are used as input to the ML model, while their corresponding numeral estimated values of Nu‾ and Sh‾ are used as output targets for the ML model. The values of Nu‾ and Sh‾ are predicted by the suggested ML model with a comparatively small error less than 1 %.