Investigation of perpendicular blade-vortex interaction using computational fluid dynamics and machine learning

Date

2026-1

Author

Arıcan, Mahmut

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

48
views

0
downloads

Cite This

Simulation of complex flow phenomena, such as blade-vortex interaction (BVI), demands high-fidelity approaches in Computational Fluid Dynamics (CFD). Although CFD can capture intricate flow features, achieving sufficient resolution requires highly refined grids, especially for vortex interactions with aerodynamic surfaces. This leads to high computational costs, particularly in three-dimensional simulations. To address this challenge, a methodology is proposed in this thesis to predict high fidelity steady Reynolds-averaged Navier–Stokes (RANS) solutions using a Graph Neural Network (GNN) algorithm trained on detailed CFD data. The approach is demonstrated through the case study of a perpendicular blade–vortex interaction over an airfoil. The interaction phenomenon is investigated using high-fidelity simulations, and the resulting data are used to train the GNN model. Results show that plausible flow predictions can be obtained on the blade surface at reduced computational expense.

Subject Keywords

CFD, BVI, Machine learning, GNN

URI

https://hdl.handle.net/11511/118997

Collections

Graduate School of Natural and Applied Sciences, Thesis