Deep learning-driven heat transfer prediction in irregular ternary hybrid nanofluid flow over fin geometries via the Adam optimization algorithm

Abstract

Nanofluids' enhanced thermal and heat transmission qualities have piqued the curiosity of several researchers. Recently, ternary nanoparticles of different shapes have been combined to generate a unique nanofluid with outstanding thermal characteristics; This work examines the Ternary hybrid nanofluid flow dynamics and the effects of radiation, ambient temperature, and natural convection heat transfer on the transient thermal performance of a porous fin that is rectangular, convex, and triangular, Using Darcy's model, this study creates a heat transport equation, Triangular fin exposed, convex, and rectangular are the three case styles considered while assessing thermal performance. Through the use of PDSolve in the Maple 2024 version program using the finite difference technique, the transformed dimensionless partial equations are solved. Deep Neural Network (LSTM with Adam algorithm) was able to predict the heat transfer rate accurately. By using MATLAB software, the present study model represents the accuracy. The study produced groundbreaking findings that the fins' efficiency is increased when a ternary hybrid nanofluid is present. In wet conditions, three fins of different forms have been compared. Compared to convex and triangular fins, the radiative, thermo-geometric, and convective transfer characteristics have more heat in rectangular geometries, The analysis's conclusions greatly impact enhancing heat transmission in industrial processes.