Theses and Dissertations

Advisor

Shinde, Vilas

Committee Member

Sescu, Adrian

Committee Member

Khare, Vivek

Date of Degree

5-16-2025

Original embargo terms

Immediate Worldwide Access

Document Type

Graduate Thesis - Open Access

Major

Aerospace Engineering

Degree Name

Master of Science (M.S.)

College

James Worth Bagley College of Engineering

Department

Department of Aerospace Engineering

Abstract

This research presents a novel integrated framework combining computational fluid dynamics (CFD) with machine learning techniques to enhance the analysis of backward-facing step flows. A high-fidelity OpenFOAM CFD simulation was developed and validated against experimental data, accurately predicting the reattachment length (��/�� ≈ 6.18) and velocity profiles throughout the domain. Machine learning models, including Random Forest, Gradient Boosting, and Support Vector Regression, were integrated through a robust data pipeline, with the ensemble approach demonstrating superior performance (RMSE of 1.18 m/s, ��2 of 0.951). Feature importance analysis revealed pressure (32%) and turbulent kinetic energy (28%) as the dominant physical parameters governing the flow behavior. Physics-informed feature engineering significantly boosted the model’s performance by incorporating knowledge in the domain, which illustrated the strength of hybrid approaches. The combined framework exhibited speedups of 140–240 times compared to traditional CFD while maintaining the error in the prediction at the level of 5%. The strategy overcomes long-standing CFD challenges, particularly for separated flows, by leveraging the strengths of physics-based and data-driven approaches. The framework efficiently accommodates rapid design iteration, parametric exploration, and optimization for aerospace, automotive, and energy systems engineering design.

Download

Share

COinS