Physics‐Informed Neural Networks for Exothermic Reactions in Porous Media

The main aim of this chapter is to study the exothermic reactions model, which has a fixed heat source in the porous media. Heat flow profiles are crucial for efficient heat transfer in various types of thermal insulation. This investigation centers on the driving force problem, specifically focusing on scenarios where a temperature gradient is considered. The governing equation of the model is restricted to an energy balance equation that gives the temperature value in the conduction state in the presence of a constant heat source. To solve this ordinary differential equation (ODE), we utilize a scientific machine learning approach known as physics-informed neural networks (PINNs). The PINNs techniques primarily encode the underlying physical rule (i.e., the ODE) into the neural network as prior knowledge. We conducted the entire experiment in Python on Google Colab. For validation purposes, we compared the PINNs solution with existing solutions in the literature. The results indicate that the PINNs approach aligns well with established solutions, confirming its accuracy and reliability.