A computational study on the heat transfer mechanisms in different structural extended surfaces using physics based neural network

Abstract

Fins are widely used in many industrial applications including heat exchangers, combustors, computer processors, fuel cells, and gas turbine blades. Porous fins are substantial in industrial uses newlinedue to their high heat transmission characteristics, lightweight construction, energy efficiency, and adaptability. They are considered to be the major elements in many advanced technologies. Thus, newlinethe study of heat transfer mechanism using fins has become a significant topic in the engineering field. Meanwhile, artificial intelligence with deep neural network techniques is being utilized for newlinetackling applied mathematics problems including fractional equations, stochastic differential equations and partial differential equations (PDEs). Also, the differential evolution (DE) is prevalent newlinein various scientific and technical domains and has emerged as a highly effective population-based stochastic search strategy for addressing optimization issues across continuous space. On the other hand, the application of physics-informed neural networks (PINNs) has gained significant attraction in solving complex heat transfer problems in fins. Deep learning and physical laws are newlinecombined by PINNs to handle heat transfer equation nonlinearities. They can accurately solve real-world engineering issues by directly including boundary conditions and physical constraints newlineinto the model, making them attractive for advanced heat management systems. In this thesis, the features of different kinds of fin structure namely rectangular, concave and wavy shaped extended surfaces subjected to convective effects with the major factors influencing heat transmission are discussed. The relevant mechanisms are mathematically modelled in the form of governing equations and are represented by the corresponding ordinary differential newlineequations or partial differential equations. The highly nonlinear equations are subjected to nondimensional transformation with the help of dimensionless terms. The application of the diverse newlinesemi-analytical, and data-driven...