individual
What campus are you from?
Daytona Beach
Authors' Class Standing
Mohammad Sanjeed Hasan, Graduate Student
Lead Presenter's Name
Mohammad Sanjeed Hasan
Faculty Mentor Name
Yongxin Liu
Abstract
The article focuses on an Artificial Neural Network (ANN) based thermal performance prediction for a trapezoidal-shaped closed enclosure in the presence of a magnetic field. The enclosure is filled with a hybrid nanofluid (Ag–SiO₂–H₂O) where a heated cylindrical pipe is placed at its center. The governing equations are solved numerically, and the resulting data are used to train the ANN model. The Nusselt number is calculated for different Rayleigh (Ra) numbers, Hartmann (Ha) numbers, and volume fractions (φ). The trained network then predicts heat transfer performance for new conditions such as different activation functions, loss functions, optimizers, data handling methods, and regularization techniques are tested to find the best ANN configuration. Result illustrates that the ANN model can predict the Nusselt number accurately if proper activation and optimization techniques are used. Even though minor variations exist, the ANN model performs well compared to numerical results. Overall, this study demonstrates that ANN-based prediction can be an effective approach for natural convection in hybrid nanofluids, which can help improve energy efficiency, thermal system design, and nanofluid applications.
Did this research project receive funding support from the Office of Undergraduate Research.
No
Deep Learning Based Prediction of Heat Transfer in a Nanofluid-filled Enclosure with a Magnetic Field
The article focuses on an Artificial Neural Network (ANN) based thermal performance prediction for a trapezoidal-shaped closed enclosure in the presence of a magnetic field. The enclosure is filled with a hybrid nanofluid (Ag–SiO₂–H₂O) where a heated cylindrical pipe is placed at its center. The governing equations are solved numerically, and the resulting data are used to train the ANN model. The Nusselt number is calculated for different Rayleigh (Ra) numbers, Hartmann (Ha) numbers, and volume fractions (φ). The trained network then predicts heat transfer performance for new conditions such as different activation functions, loss functions, optimizers, data handling methods, and regularization techniques are tested to find the best ANN configuration. Result illustrates that the ANN model can predict the Nusselt number accurately if proper activation and optimization techniques are used. Even though minor variations exist, the ANN model performs well compared to numerical results. Overall, this study demonstrates that ANN-based prediction can be an effective approach for natural convection in hybrid nanofluids, which can help improve energy efficiency, thermal system design, and nanofluid applications.