Publication

Abstract : The precise forecasting of thermal and mass transportation properties in a non-Newtonian nanofluid circulation is vital for the design and development of efficient thermal management systems, especially in micro-scale electronic devices, polymer processing, and biomedical equipment. In this context, the current work inspects the thermal and mass distribution of Casson nanofluid composed of SWCNT nanoparticles and sodium alginate-based liquid over an exponential stretching surface in the presence of inclined magnetic field, chemical reaction, slip impact, and non-uniform heat source/sink physical phenomena. The effective transport properties of these nanofluids strongly depend on their molecular structure, necessitating the use of topological indices. Similarity transformations are utilized to alter the governing partial differential equations (PDEs) to a system of ordinary differential equations (ODEs), and solutions are obtained using Runge Kutta Fehlberg - 4th 5th scheme and the shooting technique. The outcomes of the numerical calculations are presented and visualized with the aid of graphs. To improve the predictive capability, a Multi-Task Neural Network is developed and offers improved generalization across a wide range of parameter values. The numerical outcomes show that improving Casson, inclination angle, and magnetic parameter values slows down the velocity, while thermal slip declines the temperature profile. The improvement in the rate of heat and mass transfer improves up to 2.3753% and 10.5201% in the presence of Casson nanofluid for changes in inclination angle and magnetic field. The outcomes of the neural network model show strong agreement among the numerical and MTNN predictions, with a total loss of 0.000, and R 2 for Cf, Nu, and Sh tasks are found to be 0.9999, 0.9997, and 0.9994, respectively, indicating a perfect fit of the data for predicted and target values with excellent convergence and effective numerical stability.