Heat Transfer Neural Analysis of Bingham-Papanastasiou Fluid in Lid-driven Rectangular Cavity with viscous dissipation

Abstract

Abstract The study of Bingham-Papanastasiou fluids is conducted in lid-driven cavity with consideration of viscous dissipation. The left wall of the cavity is adiabatic while other walls are insulated. Numerical simulations are conducted to study the isotherms, temperature profile and Nusselt number. An inventive artificial neural network (ANN) model for precise predictions is made using the simulation data. Both COMSOL and ANN are used to find the best values for each input parameter to maximize the output parameters. The effectiveness of these two approaches in obtaining the best results is then evaluated through a comparison study. Plotting isotherms for larger Bingham numbers shows that the temperature distribution toward the left wall is behaving more and more rapidly. The rate of heat transfer rises with increasing Re values, but it exhibits the opposite behavior when Bn values are high close to the left adiabatic wall. After 406 epochs, the training state plot demonstrated convergence and optimization progress with gradient = 0.018363 and Mu = 0.0001. The novelty of this work is that the integrated approach involving Artificial Neural Networks (ANN) described in this study allows the prediction of flow behaviors for various cases without additional real-time CFD simulations once sufficient information is gathered through Computational Fluid Dynamics (CFD) simulations for a few flow cases. In real-world flow control applications where real-time CFD simulations might not be possible, this method seems to be beneficial.