Theses and Dissertations

ORCID

https://orcid.org/0009-0004-4445-4992

Advisor

Sescu, Adrian

Committee Member

Bhushan, Shanti

Committee Member

Burgreen, Greg

Committee Member

Narsipur, Shreyas

Date of Degree

12-12-2025

Original embargo terms

Immediate Worldwide Access

Document Type

Dissertation - Open Access

Major

Engineering (Aerospace Engineering)

Degree Name

Doctor of Philosophy (Ph.D.)

College

James Worth Bagley College of Engineering

Department

Department of Aerospace Engineering

Abstract

This dissertation develops and validates a machine‑learned (ML) actuator line model (ALM) as an advancement in rotor modeling that can be applied to diverse engineering applications. The model addresses two standard ALM limitations: dependence on pre‑tabulated lift and drag coefficients and the inability to represent unsteady inflow. The ML‑ALM is trained on a hydrokinetic blade‑resolved simulation database of blade‑element forces under unsteady conditions, and is queried at run time to supply axial, tangential, and spanwise forcing at each actuator element. Validation covers solitary rotor performance and wake predictions against experimental data, and verification in an inline configuration against blade‑resolved simulations for downstream rotor performance and wake. Engineering applicability is demonstrated with an eight‑device array simulation. Across cases, the ML‑ALM predicts rotor performance and wakes within 10% of blade‑resolved results, reproduces advection and breakdown of tip vortical structures and the associated turbulent kinetic energy burst, and reduces computational time by about 92% relative to corresponding blade‑resolved runs. These results support ML‑ALM as a practical mid‑fidelity rotor model when unsteady wake physics are required at tractable cost.

Download

Share

COinS