Machine learning based modeling of pollutant and heat transfer in ternary nanofluid flow through porous media

Significance: This study provides a novel approach to optimizing thermal and mass transfer in magnetically controlled ternary hybrid nanofluids using an Artificial Neural Network (ANN) framework. The ANN-based predictive model improves computational efficiency, and it makes the findings highly relevant for applications in environmental pollution control, industrial heat exchangers, and advanced cooling systems. Purpose: The present study investigates the thermal and mass transfer characteristics of a ternary hybrid nanofluid within a magnetically controlled porous medium while accounting for pollutant dispersal. In this study nanofluid with three unique nanoparticles are taken between two parallel sheets and cross mathematical fluid model has been utilized. Furthermore, an inclined magnetic field is incorporated to assess its influence on heat and mass transport dynamics. The thermophoretic velocity effect is considered to examine its role in nanoparticle migration, further impacting pollutant distribution. The study evaluates how the combination of these factors influences the efficiency of pollutant removal and overall transport properties within the fluid system. Methodology: Bvp4c scheme has been used on obtained ODEs and initial solution is taken for each parameter. Training is initiated on this data and a comprehensive computational framework is developed using an Artificial Neural Network (ANN) to optimize the thermal and mass transfer processes. After training ANN predicts the solution for different parameters. Findings: Weissenberg number and squeezing parameter reduce velocity due to dominant elastic forces and compression resistance. Additionally, an inclined magnetic field and higher magnetic intensity impede motion, whereas suction enhances velocity while injection suppresses it. Mean squared error (MSE) consistently declined over epochs, and regression plots confirmed a strong correlation between predicted and actual values.