Prediction of heat transfer in a rotor-stator cavity cooled by multiple jets: Integration of CFD models and machine learning for performance optimization
This work presents a detailed numerical study of heat transfer and turbulent flows in a rotor-stator cavity subjected to the influence of multiple impinging jets. Using Ansys Fluent and advanced turbulence models such as kϵRNGand RSM, the study accurately analyzes the complex fluid dynamics and thei...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-09-01
|
| Series: | Results in Engineering |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2590123025025861 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This work presents a detailed numerical study of heat transfer and turbulent flows in a rotor-stator cavity subjected to the influence of multiple impinging jets. Using Ansys Fluent and advanced turbulence models such as kϵRNGand RSM, the study accurately analyzes the complex fluid dynamics and their interactions with rotating and stationary surfaces. Specifically, the analysis covers a range of rotational Reynolds numbers from (Reω=2.32×108;Reω=5.4×105), while the jet Reynolds number varies from (Rej=16×103;Rej=50×103). The results highlight a significant improvement in heat transfer through the combination of central and concentric jets, enabling more uniform and efficient cooling of the rotor. The analysis of local Nusselt numbers revealed critical heat exchange zones, particularly at the jet impact points and in regions dominated by rotational effects. To refine predictions and optimize model performance, artificial neural networks (ANN) were integrated to capture the nonlinear and multivariate relationships between input parameters and heat transfer. The ANNs demonstrated exceptional performance, with high correlation scores for the training (R=0.95296), validation (R=0.94282), and test (R=0.9599)sets, confirming their ability to accurately predict local Nusselt numbers. Although the numerical simulations show good agreement with experimental data, discrepancies persist in transition zones, suggesting the need for more precise models or better experimental characterization. Furthermore, the development of reliable correlations to predict local Nusselt number behaviors remains a challenge due to the complex interactions between dynamic and geometric parameters. This study underscores the essential role of parameters such as jet spacing (G) and Reynolds numbers in optimizing heat transfer. The integration of artificial neural networks has improved prediction accuracy and opened perspectives for future research aimed at refining models and better understanding the complex local phenomena observed in rotor-stator systems. This work contributes to the improvement of cooling system designs in various industrial applications. |
|---|---|
| ISSN: | 2590-1230 |