CFD investigation and ANN prediction of heat transfer coefficient for fully developed turbulent air flow around double V-baffle turbulators

CFD investigation and ANN prediction of heat transfer coefficient for fully developed turbulent air flow around double V-baffle turbulators

This paper presents a numerical study of periodic fully developed turbulent airflow in a rectangular channel with double upstream V-baffles installed on the upper and lower walls of the channel in an in-line manner. Utilizing the OpenFOAM open-source software, this numerical research investigates th...

Full description

Saved in:
Bibliographic Details
Main Authors: Abdulaziz Alasiri, H.E. Fawaz
Format: Article
Language:English
Published: Elsevier 2025-07-01
Series:Case Studies in Thermal Engineering
Subjects:
Cross flow
Turbulent flow
Artificial neural network
Multilayer perceptron
Backpropagation algorithm
V-baffle turbulators
Online Access:http://www.sciencedirect.com/science/article/pii/S2214157X25003569
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This paper presents a numerical study of periodic fully developed turbulent airflow in a rectangular channel with double upstream V-baffles installed on the upper and lower walls of the channel in an in-line manner. Utilizing the OpenFOAM open-source software, this numerical research investigates the impact of Re from 10,000 to 40,000 and BR from 0.3 to 0.5 on flow structure and heat transfer performance. An ANN model is constructed to estimate the local heat transfer coefficient using results obtained from the current CFD simulations and utilizing axial local distance (X/P), Re, and BR as ANN input parameters. The process of training incorporates the analysis of the loss function on training and validation data for controlling the weights and biases using backpropagation while feed forward propagate the selected input parameters. A total of 11 hidden layers consisting of 24 neurons each has been used in constructing the ANN, and the training process is optimized using the ADAM algorithm to minimize the loss function. The Final layer uses the linear activation function while all the hidden layers use the rectified Linear Units Activation function (ReLU). The ANN model demonstrates excellent predictive performance, yielding values close to 1 for R2 and r, along with extremely low values for MSE, MAPE, MSLE, and log-cosh loss (0.01, 0.6 %, 0.001, and 0.01, respectively), demonstrating the ANN model's high predictive accuracy.
ISSN:2214-157X

Similar Items