Real-Time, High-Fidelity Analysis of Thermal Fluids and Effects on Materials for High-Speed Aircraft Using Data--AI-Driven Methods
Abstract
Stability and controllability (S and C) of the high speed aircraft is critical to ensuring high maneuverability and enhanced safety of innovative aircraft during various missions. While the wind tunnel experimental data are not sufficient throughout the entire flight envelope, the numerical methods to calculate the stability and their derivatives, both static and dynamic, have been mostly based on the simple interpolation formula from the look-up tables constructed by wind-tunnel data or low-fidelity linear models based on the lifting line or potential flow methods. The DATCOM is one of such models and popular in predicting the stability and controllability of the aircraft However, the corresponding accuracy is greatly limited, and the nonlinear physics including shocks, turbulent wakes, and thermal fluid effects on the materials and structures cannot be solved at a desired accuracy. On the other hand, the high-fidelity CFD method of the Reynolds-Averaged Navier-Stokes (RANS) solver or large eddy simulation (LES) can resolve those nonlinearities and complex multiple-physics problems, but requires several hours to days to compute even with massively parallel computation architecture. The future technology for the innovative aircraft design should enable fast and accurate evaluations in real time to explore wide range of design space in the stability and controllability as well as the effects on the advanced materials. The motivation for the research is to overcome those limitations and enable real-time, high accuracy estimation of complicated physics problems for the future high-speed aircraft, as well as S and C, materials and structures. With the recent AI methods proven to be powerful in almost all engineering areas, the issues related to accuracy vs. computational cost can be effectively handled by machine learning techniques. However, the direct application of the popular AI method to the CFD has many challenging issues to resolve, including the topology of the non-cartesian CFD mesh dissimilar to computer graphics, spatio-temporal structure of the flow solution data, and strongly coupled multi-physics of aerothermo-structure-materials. The merit and the novelty of the proposed project is resolving those issues effectively by modifying the AI methods to the CFD calculations, and predicts in real time static and dynamic stabilities of the high-speed aircraft and their derivatives.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Feb 16, 2024
- Source ID
- FA23862314022
Entities
People
- Seongim Choi
Organizations
- Air Force Office of Scientific Research
- Gwangju Institute of Science and Technology
- United States Air Force