A Data-Informed Framework for Predicting Unsteady Aerodynamic Loads for Advanced Air Mobility Vehicles
Faculty Mentor Name
Sarasija Sudharsan
Format Preference
Poster
Abstract
Advanced Air Mobility (AAM) requires aircraft to operate within urban environments, which are characterized by unpredictable gusts and building-induced turbulence. These conditions produce highly unsteady aerodynamic loads that challenge vehicle stability, control, and structural design. This research addresses the challenges of predicting unsteady aerodynamic loads, where traditional low-fidelity aerodynamic models struggle to accurately predict nonlinear, unsteady phenomena such as dynamic stall and gust-induced separation. High-fidelity methods such as Large-Eddy Simulations (LES) provide accurate results but are computationally expensive and impractical for design iterations and control-oriented applications.
The objective of this work is to develop a hybrid, physics-informed aerodynamic modeling framework that bridges the gap between computationally expensive Large-Eddy Simulations and efficient but limited classical unsteady models. This study compares panel method and dynamic stall formulations against high-fidelity LES databases. Data-driven corrections are implemented to identify systematic modeling discrepancies. The resulting reduced-order models provide the accuracy necessary for system-level analysis, integration with flight dynamics, and control system development while maintaining the computational efficiency to support safer and more reliable AAM vehicle design.
A Data-Informed Framework for Predicting Unsteady Aerodynamic Loads for Advanced Air Mobility Vehicles
Advanced Air Mobility (AAM) requires aircraft to operate within urban environments, which are characterized by unpredictable gusts and building-induced turbulence. These conditions produce highly unsteady aerodynamic loads that challenge vehicle stability, control, and structural design. This research addresses the challenges of predicting unsteady aerodynamic loads, where traditional low-fidelity aerodynamic models struggle to accurately predict nonlinear, unsteady phenomena such as dynamic stall and gust-induced separation. High-fidelity methods such as Large-Eddy Simulations (LES) provide accurate results but are computationally expensive and impractical for design iterations and control-oriented applications.
The objective of this work is to develop a hybrid, physics-informed aerodynamic modeling framework that bridges the gap between computationally expensive Large-Eddy Simulations and efficient but limited classical unsteady models. This study compares panel method and dynamic stall formulations against high-fidelity LES databases. Data-driven corrections are implemented to identify systematic modeling discrepancies. The resulting reduced-order models provide the accuracy necessary for system-level analysis, integration with flight dynamics, and control system development while maintaining the computational efficiency to support safer and more reliable AAM vehicle design.