CFD Methods for Aerodynamic Design

Transcription

1 CFD Methods for Aerodynamic Design Afandi Darlington Optimal Aerodynamics Ltd

2 Why CFD? Datasheet methods are still very relevant today (ESDU, USAF DATCOM) Validated estimates of lift, drag, moments, stability derivates BUTthey are less useful for: New clean 2D aerofoil & multi-element design Complex configurations Detail design trade-offs Complex intake flows Combination with optimisation methods CFD is a useful tool when intelligently applied It is not a black box (G.I.G.O) It has limitations depending on the formulation

3 Tools for the job 3-D N-S 3-D EULER 3-D PANEL METHOD 2-D MULTI ELEMENT 2-D PANEL METHOD & 3-D VORTEX LATTICE CFX, Fluent MGAERO, Flite3D VSAERO MSES Euler, 2-D N-S XFOIL, AVL (FREE!) COST CPU TIME CAPABILITY

4 CFD Method CAD import/cleanup Meshing divide up problem Post processing Flow solver B.C. s: M, α, {T, ρ} Solve problem in each cell Reassemble solution Navier-Stokes equations; simplifying assumptions Focus of this talk are STEADY methods i.e. no rotating or moving geometries; steady converged values of lift, drag, moment etc

5 XFOIL 2-D viscous-coupled panel method (incompressible) Single element aerofoils (but can do plain flaps) Industry standard code for low speed aerofoil design Quick to run Pressure plots show pressure coefficient, C p : C p = 1, stagnation C p = 0, free stream static pressure C p = -2, two dynamic pressures below freestream static Limitations poor at C lmax (overestimates by ~0.1); but good for C d and C m and lift-curve slope FREE! (google XFOIL for free download)

6 XFOIL DEMO HLLF29140 (Kestrel aerofoil)

7 Mesh sensitivity HLLF29140, M0.1, Re=3e6, α=4deg XFOIL, 280 max N pan =40 N pan =200

8 Vortex Lattice (AVL) Inviscid, lifting surface model quick to run Incompressible flow Good for spanwise lift distribution & induced drag FREE! (google AVL Drela )

9 Vortex Lattice (AVL) typical output 3.5% less induced drag than elliptical loaded wing without winglets

10 2-D N-S code Multi-element surfaces Unstructured grid Compressible flow Accurate (inc. C lmax ) Runs on a PC workstation Cost ~ 5k/yr Convergence in C l Convergence in C d

11 F1 high lift laminar aerofoil using MSES MSES F1 Fowler flap Viscous coupled Euler code (orthogonal grid) Compressible flow (slats..) Accurate for C lmax, can handle sizeable separations Cost ~ 5k/yr Note heavily separated wake on flap

12 Wind tunnel validation F1 Kestrel wing Aerofoil test model 2

13 a= α=-2, (α=0 similar) LAMINAR FLOW > 60% x/c LAMINAR SEPARATION BUBBLE TURBULENT WEDGES TURBULENT FLOW

14 F1 Wind Tunnel Tests IR Camera Trailing edge IR camera Transition from laminar to turbulent flow Leading edge FLOW F1 Kestrel wing upper surface Source : F1-2 2 nd Cranfield test, α=-2, configuration 25 (flaps stowed)

15 F1 Wind Tunnel Tests IR Camera FLOW FLOW Trip tape F1 Kestrel wing upper surface Source : F1-2 2 nd Cranfield test, α=-2, configuration 25 (flaps stowed)

16 Source : F1-2 2 nd Cranfield test, configuration 24 (flaps 0 )

17 VSAERO 3-D viscous coupled panel method Orthogonal grids skill reqd. for meshing Multiple elements Good for attached incompressible flows, complex geometry Needs large PC workstation Cost ~ 10k/yr Industry standard CFD code Data accepted by FAA

18 Centaur Run CFD time for this analysis case ~8hrs on PC workstation Actuator disk model available for props, rotors Note wake relaxation, produces accurate induced drag calculation

19 Downwash distribution at tailplane location

20 Gull 36 UAV Centaur 2 Customer feedback (James Labouchere, Warrior Aeromarine): The results from the CFD programme were right. They have been proven to be right (GULL 36 data logging). We can now proceed with scaling with utmost confidence to know we will get other products right.

21 VSAERO Design Methods and Tools for Light Aircraft Run time for this case ~6 hrs on PC workstation CFD data used to validate aerodynamic centre position (=> axle loads) and performance

22 Unstructured Navier-Stokes code with turbulence modelling for boundary layer Kestrel/Epic LT Solves compressible flow with large separations Heavy CPU requirements (PC cluster or mainframe) Limited application in light aircraft design unless resources are available Cost ~ 14k/year Great for detail design fairings, junctions, pylons etc. Kestrel onglet design Attachment line

23 Kestrel/Epic LT Flow separation at wing root without onglet

24 Kestrel/Epic Attached LT flow at wing root with onglet Attachment line

25 Kestrel CFX analysis of intake duct

26 Tools for the job 3-D N-S 3-D EULER 3-D PANEL METHOD 2-D MULTI ELEMENT 2-D PANEL METHOD & 3-D VORTEX LATTICE CFX, Fluent MGAERO, Flite3D VSAERO MSES Euler, 2-D N-S XFOIL, AVL COST CPU TIME CAPABILITY

27 Conclusions CFD tools are useful aids to the designer, but don t ignore datasheets/nasa reports/arc R&Ms etc Some useful codes are free (XFOIL, AVL) Large PCs can run all CFD codes apart from N-S cases was CRAY supercomputer 10 years ago! Pick you code to suit the problem Practice with the code against published validation cases to ensure you can drive it properly Ensure the mesh is fine enough and each case has converged If in doubt, contact the experts!