CFD based neural network model framework for mixed convection in a lid-driven cavity with conductive cylinder
The current study investigates the application of an Artificial Neural Network (ANN) model to analyze and predict mixed convection heat transfer within a 2-dimensional square cavity with a conductive cylinder at the centre. The top lid of the cavity is maintained at a constant cold temperature and s...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-02-01
|
| Series: | Heliyon |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2405844025010175 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The current study investigates the application of an Artificial Neural Network (ANN) model to analyze and predict mixed convection heat transfer within a 2-dimensional square cavity with a conductive cylinder at the centre. The top lid of the cavity is maintained at a constant cold temperature and slides with a constant linear velocity, while the bottom wall is heated to maintain a constant temperature. The governing equations are discretized using Galerkin Weighted Residual Method and numerically solved using Gauss Quadrature procedure. The ANN model is trained using the data derived from CFD simulations and used to predict the heat transfer performance quantitatively and qualitatively. The setup is investigated for a wide range of Richardson numbers (0.1≤ Ri ≤ 10.0), Reynolds numbers (50 ≤ Re ≤ 250) and cylinder diameters (0.1≤ D/L ≤ 0.8). Heat transfer performance is evaluated from the average Nusselt number along the heated bottom wall. Correlations are established showing dependency of Nusselt number on Ri, Re and D/L. Velocity and thermal fields are expressed by streamlines and isothermal contours. The study shows that higher Richardson and Reynolds numbers lead to an enhancement in the overall heat transfer. But for larger cylinder diameters, heat transfer is mostly dependent on Reynolds number. It is also revealed that substantial improvements in computational efficiency are achieved by the ANN model as it has reduced 82 % of computation time and 83 % of storage requirements to predict the results by maintaining mean absolute error below 0.1 %. The study provides an insight that ANN modelling could open a new dimension to the heat transfer research field and significantly reduce the requirement of time and resources to solve complex problems. |
|---|---|
| ISSN: | 2405-8440 |