High-Fidelity Aeroelastic Simulations of Future Airplane Concepts

Overview

The introduction of lightweight materials and the emergence of energy-efficient aircraft concepts with flexible wings have spurred the development of aeroelastic modeling capabilities. These models should be both efficient and accurate enough to support design and control optimization tasks. High-fidelity computational tools play an important role in evaluating new aeroelastic aircraft design concepts and model assumptions in simulation environments. In this project, we investigated the effect of viscosity on a distributed flap system subject to aeroelastic deformations in the transonic cruise regime, and the significance of viscous drag for the purpose of design optimization.

Project Details

The aerodynamic drag performance of NASA's Generic Transport Model aircraft with flexible wings and the Variable Camber Continuous Trailing Edge Flap (VCCTEF) system is studied at fixed lift. The VCCTEF system includes 19 flaps, with 3 camber segments on each flap that can be rotated relative to each other. The flap deflection settings were adapted to the considered fuel loads. A "piano-finger" plain flap configuration without elastomer between the flaps was also analyzed. The wing deformations were obtained by an inviscid method that couples a 1D beam-stick finite element model to NASA's Cart3D computational fluid dynamics solver. Next, the aerodynamic loads from the inviscid solver were compared with those obtained with NASA's Launch Ascent and Vehicle Aerodynamics (LAVA) curvilinear Reynolds-Averaged Navier-Stokes (RANS) solver.

Results and Impact

An inviscid design optimization approach may be acceptable if the drag contribution by viscous effects does not vary much in the design space. In this study, the viscous drag was found to be consistent across the cases, while the pressure drag significantly varies with flap configuration and fuel load. For given lift and wing shape, viscosity increases the pressure drag by offsetting and smoothing the pressure shock and damping pressure gradients over the flap hinge lines. The drag improvement provided by the distributed flap system with elastomer could be predicted within 25% agreement between the viscous and inviscid methods. A design version with only one camber segment per flap performed similarly as with three segments. For the piano-finger configuration, flow separation in the gap between the flaps degrades both performance and the validity of the inviscid assumption. In future work, smaller gaps will be examined, and the effect of viscosity on the wing deformations will be studied by coupling the finite element model to the RANS solver.

Why HPC Matters

The wing deformation computations required an average of 5 aeroelastic iterations and 17 Cart3D calls, each iteration taking approximately 2.5 hours on 72 cores of the Endeavour shared-memory supercomputer at the NASA Advanced Supercomputing (NAS) facility. A LAVA viscous simulation typically took 8-12 hours using 320 Ivy Bridge cores of NAS's Pleiades supercomputer.