This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Dr. Matthew Forster Research Scientist, Computational Engineering, BAE Systems 15th October 2019
Developments towards fluid-structure interaction simulations in BAE Systems’ corporate CFD suite Aerodynamics Tools and Methods in Aircraft Design, Royal Aeronautical Society
Towards FSI simulations in BAE Systems’ Computational Fluid Dynamics suite Overview
2
• A need to incorporate higher fidelity aerodynamics into flutter and structural coupling toolsets was identified within BAE Systems. We want:
• Capability to predict transonic flutter
• Ability to simulate complex geometries, e.g. unconventional planforms, inlaid controls (spoilers)
• Accommodate structural models in multiple formats
• Tool to fit into existing flutter and structural processes
• Interchangeable results between CFD and DLM (Doublet Lattice Method)
• Two approaches being developed for unsteady aeroelastic predictions:
• Linearised Frequency Domain CFD
• Coupled CFD-CSM
• Preliminary LFD results presented here
• Dr. Sebastian Timme from the University of Liverpool, via the Royal Academy of Engineering Industrial Secondment Scheme, helped materialise part of this work
• Fast response mesh generation for complex configurations – “bottom up” approach using an unstructured, quadrilateral-dominant surface mesh (generated in Mercury) and a grown boundary layer mesh interfaced to a Cartesian farfield mesh via cut-cells (generated by Venus)
• Surface meshing:
• Iso/Anisotropic quad-dominant, triangular or structured surface meshes
• User-defined spacing fields, or automatic spacing fields based on model characteristics such as curvature
• Volume meshing:
• Wake plane and “multiple normal” capabilities for better treatment of sharp trailing edges
• Fully parallel, using MPI, with both implicit and explicit solvers
• Capable of performing single phase simulations in air or water, over a large speed range – including incompressible flows using the Artificial Compressibility Method (ACM) and Low-Mach-Number preconditioning
• RANS simulation using a range of turbulence models – including Spalart-Allmaras, Menter SST and Reynolds Stress Transport (RST) models
• Unsteady simulation using scale-resolving methods – including DES and wall modelled LES
• Range of boundary conditions including jets, intakes and propellers/rotors
Coupled CFD-CSM approach using OpenFSI with MSC Nastran. Figure modified from MSC Nastran 2017.1 User Defined Services Guide.
• Development ongoing for two CFD-CSM approaches: • Multi-physics code-coupling using MSC Nastran’s OpenFSI
interface • Data transfer between Flare and Nastran (see right) • Forces from CFD applied to wetted nodes of CSM • Displacements from CSM applied to CFD surface
• Modal-based aeroelastics solver contained within Flare
• Approximates deflected shape as a superposition of mode shapes
• Aeroelastic equations are solved to determine modal amplitudes
• For both LFD and CFD-CSM a method is needed to transfer information between structural and fluid domains.
• CFD mesh and structural mesh points often not co-located, need to transfer mode shape deflections, structural displacements and forces
• Radial Basis Functions (RBF) are used as an interpolation for:
• Interpolation between structural grid and CFD surface grid
• Efficient CFD volume mesh deformation
• Algorithm used from: Kedward, et al, “Efficient and exact mesh deformation using multiscale RBF interpolation”, J. Comp. Phys., 2017.
• Allen showed that same RBF matrix can also be used to interpolate forces back to structure for CFD-CSD, in “Unified Approach to CFD-CSD Interpolation and Mesh Motion using Radial Basis Functions”, 2007 AIAA Applied Aerodynamics conference.
Reference solution from DLR Tau solver, see right. Note that the LFD predictions overpredicted the extent of the shock oscillation compared with the URANS.
• LFD solution used in BAE Systems’ flutter and structural coupling toolset
• Direct replacement for DLM Q matrices
• Mode shapes interpolated from Nastran grid to CFD surface using RBF
• CFD volume mesh deformation also using efficient RBF methods
• Low speed M=0.3, flutter solution comparable between LFD and DLM
• Transonic M=0.8, differences in results possibly due to transonic deficiencies of DLM • Nastran SOL145 solution for 10 frequencies, 4 mode shapes: 83 seconds
• LFD simulation on 100 cores for 10 frequencies, 4 mode shapes: 1114 seconds (18.5 minutes)
Towards FSI simulations in BAE Systems’ Computational Fluid Dynamics suite Use cases
32
Typhoon and Hawk manufacture and capability development
Unmanned and future air system capabilities
• In-house CFD-based aeroelastics capability still relatively early in development
• LFD capability is maturing and is intended to be used in the near future on our products and projects such as • Typhoon • Hawk • Tempest • Future projects
Permission to reproduce any part of this document should be sought from BAE Systems. Permission will usually be given providing the source is acknowledged and the copyright notice and this notice are reproduced.