Cooperation, perseverance yield progress in aerodynamic prediction reliability
By Jennifer Abras and Mehdi Ghoreyshi|December 2023
The Applied Aerodynamics Technical Committee emphasizes the development, application and evaluation of concepts and methods using theories, wind tunnel experiments and flight tests.
Milestones achieved this year focused on improving core predictive technologies. Midyear, researchers at the Institute of Aerodynamics and Flow Technology of the German Aerospace Center, DLR, carried out validation activities to improve the reliability of peak lift predictions for high-lift aircraft configurations. Improved reliability reduces the need to overdesign aircraft, leading to more efficient designs. Numerical simulations were executed via the in-house code TAU, and the low-speed wind tunnel in Braunschweig, Germany, provided experimental data. The researchers selected the 5.2% scale High-Lift Common Research Model, or CRM-HL, as a representative baseline configuration. As part of a DLR-NASA cooperation, NASA’s Langley Research Center in Virginia provided the wind tunnel model and personnel during almost four months of model preparation and a test campaign. DLR also provided an ultra-high bypass ratio nacelle and a Krüger leading edge system for the CRM-HL model to investigate a configuration variation. During the test execution, NASA Langley and DLR researchers evaluated preliminary data and shared it between agencies and provided subsets of the data to a larger community of researchers. The tests and simulations fall under the DLR project ADaMant, or Adaptive Data-driven Physical Modeling towards Border of Envelope Applications.
In June, the University of Cambridge reported preliminary results of an ongoing study of aircraft coatings. Comprised of fibers aligned in the streamwise direction to reduce turbulent skin friction, these coatings were designed to increase the drag-reducing mechanisms present in existing technologies. Preliminary work indicated that these coatings can yield up to 20-25% drag reduction, though these estimates are based on macroscopic models and must be assessed via fully resolved simulations. The resolution required must be fine enough to resolve all the scales of turbulence and the fibers resulting in simulations that are significantly more computationally intensive than conventional simulations. With current algorithms, each simulation would take up to a year to run. Researchers developed new algorithms to execute the same simulations in about a week, so the study of these fibrous coatings can be expedited.
Aerodynamic research at the University of California, Los Angeles in May and July focused on the influence of extreme levels of gusts on aircraft wings. The study of gust-airfoil interactions involves a large parameter space defined by gust- and flight-characteristic parameters, necessitating a gigantic experimental and computational campaign. The UCLA researchers applied data-driven analysis to show that the aerodynamic flow field data for a large number of computational and experimental cases can be compressed to only three variables using autoencoders while retaining the dynamics of the entire flow field. Their work revealed that the seemingly complex dynamics of violent gust-airfoil interactions are low-rank in nature, offering new avenues for data collection, characterization, modeling, sensing and control of flows around small aircraft that must operate in adverse weather and extreme aerodynamic conditions.
In July, researchers at Pennsylvania State University and Wright State University posted their machine-learned turbulence model on GitHub. This model captures the physics pertaining to pressure gradient forces and yields improved skin-friction predictions for separated flows, both of which are common in aerospace engineering but are difficult to model. The machine-learned model leverages artificial neural networks and the high-fidelity data accumulated over the past decade. Furthermore, the researchers preserved the existing calibrations of the baseline Spalart-Allmaras turbulence model, which gives the model the ability to generalize to unseen Reynolds numbers and geometries. This improves the ability to explore new designs and the reliability of tools when applied to unfamiliar
geometries and flow conditions.
Contributors: Ricardo Garcia-Mayoral, Ralf Rudnik, Kunihiko (Sam) Taira and Weixing Yuan
