1. Application of a
Lattice-Boltzmann Code in
Automobile and Motorcycle
Aerodynamics.
Dr.-Ing. Norbert Grün
Aerodynamics Simulation
Lecture Series on Road Vehicle Aerodynamics
von Karman Institute for Fluid Dynamics, Brussels
May 30 – June 03, 2005
2. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 2
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Outline
Aerodynamic Development Process
Physics Overview
Simulation Process
Validation Examples
Various Applications
Conclusion
3. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 3
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
CFD in the Aerodynamic Development Process
Simultaneous Usage of Experimental & Virtual Tools
Serial Development PhaseConcept Phase
Prototypes
100%
Windtunnel Model
CFD-Models
A
C
D
F
C
C
F
A
B
C
D
E
F
Styling-
Freeze
Styling–Competition
A
C
D
F
C
C
F
A
B
C
D
E
F
40%
Windtunnel Models
CFD-Models
Proportion-Studies
CFD-Models
StylingProcessAeroAnalysisTools
4. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 4
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Requirements on CFD as a Productive Tool
• Accuracy (∆CD <±0.005, ∆CL <±0.010), at least for trends
• Geometry input preparation minimized
• Ability to handle complex geometries (underhood & underbody)
• Deliver results in a reasonable timeframe (over night)
• Easy to use (by non-numerics specialists)
5. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 5
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
CFD Resources
285
222
95
48
248
253
0
50
100
150
200
250
300
1997 1998 2000 2001 2002 2004 2005
TotalNumberofProcessors
SUNSUN
SGI
2 x SGI
(95+127)
1 x SGI
(127)
2 x HP
(je 63)
SUN
1 x SGI
(160)
2 x HP
(je 63)
decicated PowerFLOW servers
NumberofProcessors
Speed-U
p
Efficiency on
Parallel Computers
6. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 6
New Goal :
• Construct simplified microscopic description (mesoscopic)
that still contains the essential micro-physics to achieve
desired macroscopic behaviour.
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Motivation for Lattice-Boltzmann Methods (LBM)
Microscopic
______________
Mesoscopic
______________
Macroscopic
Microscopic
______________
Mesoscopic
______________
Macroscopic
Kinetic Theory
Lattice Methods
Navier-Stokes
Kinetic Theory
Lattice Methods
Navier-Stokes
• Simulate fluid at microscopic level since the physics is simpler
and more general than macroscopic, continuum (PDE) approach.
• However, complete reproduction of molecular dynamics
is much too expensive (today and also in the „near“ future).
7. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 7
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
LBM vs. Traditional CFD Methods
Real Fluid
Free molecules in continous space
Kinetic Theory
Microscopic particles (Boltzmann Equation)
Real Fluid
Free molecules in continous space
Kinetic Theory
Microscopic particles (Boltzmann Equation)
Traditional CFD Methods
___________________________
Chapman-Enskog Expansion
Statistical Method applied to real gases
Navier-Stokes Equations
Conservation of Mass, Momentum and Energy
Numerical Methods
Discrete Approximation of
Partial Differential Equations
Traditional CFD Methods
___________________________
Chapman-Enskog Expansion
Statistical Method applied to real gases
Navier-Stokes Equations
Conservation of Mass, Momentum and Energy
Numerical Methods
Discrete Approximation of
Partial Differential Equations
Lattice-Boltzmann
_________________________________
Simulation of Particle Dynamics
• No integration of partial differential eqn.
• Movement & collisions conserve
mass, momentum and energy
• No numerical instabilities
Lattice-Boltzmann
_________________________________
Simulation of Particle Dynamics
• No integration of partial differential eqn.
• Movement & collisions conserve
mass, momentum and energy
• No numerical instabilities
Results
Fluid dynamic quantities at discrete points in space and time
Results
Fluid dynamic quantities at discrete points in space and time
8. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 8
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Basics of Kinetic Theory
Boltzmann Equation ),,(),,(),,(),,( tcxCtcxfctcxf
t
tcxf
dt
d rrrrrrrrr
=∇⋅+
∂
∂
=
Describes the rate of change of the velocity distribution function due to nonequilibrium
Velocity Distribution Function ),,( tcxf
rr
Gives the number of particles at time t per unit volume in phase space around x and c
Collision Term C satisfies the necessary conservation laws
∫ = 0)()( cdcCc
rrr
ξ
Mass
Momentum
Energy
1)( =c
r
ξ
cc
rr
=)(ξ
2
2
1
)( cc
rr
=ξ
Describes fluid behaviour using the interactions of air molecules
∫= cdtcxftx
rrrr
),,(),(ρDensity
∫= cdctcxftxutx
rrrrrr
),,(),(),(ρMomentum
∫ −= cductcxftxE
rrrrrr 2
))(,,(),(Energy
9. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 9
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Basics of Lattice Methods
Replace the continuous velocity
distribution function by a discrete
set of particle velocities defined on
a lattice of equal shaped cubic cells
Vtxiftxintcxf ∆≡→ ),(),(),,(
rrrr
},...,1;{ miicc =∈
rr
Particle dynamics is now described by the Lattice Boltzmann Equation
),(),(),( txiCtxintticxin
rrr
+=∆++
The collision operator C determines if a lattice system
produces a physically meaningfull fluid behaviour
During an elementary time interval particles can only hop from
one center of a cell to one of the m near neighbouring cells
according to their velocity
)1(=∆t
x
r
ticx ∆+
rr
ic
r
10. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 10
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Macroscopic Quantities
• Macroscopic quantities, such as density, pressure, velocity, etc.
are computed by statistical methods from the state vectors
DENSITY
MOMENTUM
ENERGY
• Higher order moments (Energy Flux, Stress Tensor)
are also available locally (do not involve derivatives)
∑=
j
j txntx ),(),(
rr
ρ
∑=
j
jj txnctxu ),(),(
rrrr
ρ
[ ]∑ •=
j
jjj txnccmtxE ),(),( 2
1
rrrr
0=u
r
ρ 0≠u
r
ρ
Vector length
denotes number
of particles moving
in that direction
m
T
kVRMS 3= ≈ 1000 m/s for oxygen at 20° C
Particle velocities can be much higher
than the resulting macroscopic velocity
11. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 11
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Transport Coefficients in Lattice Methods
Kinetic Theory allows to compute viscosity and thermal conductivity
from the velocity distribution function !
T
D
D
a MFPMFP
2+
== λλν
The molecular viscosity depends on the
mean free path between collisions and
the speed of sound (temperature).
Viscosity is set by adjusting the relaxation parameter of the collision operator
{ }
( )
( ) jc
eq
jc
eq
jjcj
jjjjj
nn
nntxn
txnCtxntcxn
ωω
ω
−+=
−−=
+=++
1
),(
),(),()1,(
r
rrrr
Lattice-Boltzmann Equation
Viscosity is reduced by reducing the mean
free path or equivalently the time
between collisons
Collision frequency for2<cω 0>Lattcν
cω/1
−=
2
11
cT ω
ν
Chapman-Enskog Expansion
−
+
=
2
11
2
2
c
D
T ωρ
λ
Viscosity Thermal Conductivity
12. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 12
Simple 2D Model with 4 directions and 3 speeds
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Concept of Particle Models
• The fluid is composed of a very large number of particles
(not molecules, this is a mesoscale model)
• Particles are only allowed to move in certain directions on the lattice
with limits on how far they can get in a single time step (their speed)
• The state of the fluid is represented by the number nji
of particles moving with speed (energy) j in direction i
1
Possible
Directions
2
3
4
Particle with
speed 1 in
direction 4
Particle with
speed 1 in
direction 4
Particle with
speed 2 in
direction 3
Particle with
speed 2 in
direction 3
A model allowing 3 speeds
(0,1,2) and 4 directions re-
presents the particle popu-
lation by 9 state vectors nji
n0 ( = n01 = n02 = n03 = n04 )
n11 , n12 , n13 , n14
n21 , n22 , n23 , n24
State vectors are integers !
Particle with speed 0
Particle with speed 0
The maximum number of particles per state
depends on the number of bits for state vectors !
13. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 13
• Repetition evolves time (t -> t+1) and forms an inherently transient solver
• The process of evolving (solving) the update equation is inherently
parallel (computationally efficient) and stable (computationally robust)
is the collision operator that exactly conserves
local mass, momentum and energy
jC
• Also drives local distribution to equilibrium (entropy maximized)n
eq
j
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Fluid – Fluid Interaction
• Dynamics in the fluid consists of two steps : MOVE & COLLIDE
• Update equation { }),(),()1,( txnCtxntcxn jjjjj
rrrr
+=++
Time t
Time t+1
n1
n2
n‚2
n‘1
Example : Mass Conservation
n'1 + n'2 = n1 + n2
14. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 14
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Fluid – Surface Interaction
Facets
Solid Body
Voxels
Surfels
Automatic discretization
The intersection of voxels with the facets
representing solid bodies creates surfels
which define the computational surface
resolution.
In each timestep surfels gather and scatter
particles, altering their momentum according
to the boundary conditions
Surface forces depend on the momentum
exchange between fluid and wall
Vin
Vout
Specular Reflection
Vtin
Vnin
Vtout
Vnout
Vin Vout
Bounce Back Reflection
Slip Condition
Normal component inverted
Tangential component unchanged
Momentum balance →→→→ normal force only
No Slip Condition
Normal component inverted
Tangential component inverted
Momentum balance →→→→ normal & tangential force
15. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 15
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Reynolds-Number Regimes
Regime Reynolds Number PowerFLOW
__________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________
Low Re < 10,000 Direct Simulation
Transitional 10,000 < Re < 100,000 currently not applicable
High Re > 100,000 Boundary Layer Simulation
approximate values,
actual values problem dependent
Regime Reynolds Number PowerFLOW
__________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________
Low Re < 10,000 Direct Simulation
Transitional 10,000 < Re < 100,000 currently not applicable
High Re > 100,000 Boundary Layer Simulation
approximate values,
actual values problem dependent
Solid Wall Solid Wall
Modeled Flow
Direct Simulation
Using a large number of voxels the
boundary layer is resolved down to the wall
with zero velocity at the wall.
Particles are bounced back from the wall
exactly canceling their momentum.
Boundary Layer Simulation
The presence of the wall is modeled by a
shear stress at the slip surface.
Particles loose momentum at the slip surface
according to the (modified) law of the wall.
16. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 16
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Turbulent Wall Model
( ) Byu += ++
ln
1
κ
Assumption: Universal velocity profile of a turbulent
2D boundary layer with dp/dx=0
++
= yu
:505for ≤≤ +
y
:5for ≤+
y
0.5
4.0
≈
≈
B
κ
τu
u
u =+
υ
τu
yy =+
ρ
τ
τ
w
u =
PowerFLOW Extension:
• include the effect of a longitudinal pressure gradient
∂
∂
+=→ ++++
x
p
fAAyUyU 1mit)/()(
The wall model provides the wall shear stress
to alter the momentum of scattered particles.
wτ
17. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 17
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Approaches to Turbulence Modeling
Dissipation
dl LLength
ν/2
dl UL /Time
Turbulent Scales
4/3
Re/ ≈dlLRange
( ) ( ) 2/12
Re/// ≈νdlULRange
RANS = Reynolds Averaging
All scales of motion are described by statistical methods (time averaged)
LES = Large Eddy Simulation
Alle Skalen werden berechnetmodeled computed via modified unsteady Navier-Stokes equations
Filter Width (Grid Size)
DNS = Direct Simulation
All scales of motion in space and time are computed
VLES = Very Large Eddy Simulation
modeled computed unsteady
Coherent anisotropic eddiesUniversal eddies
18. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 18
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
CFD Simulation Process
CAD/CASModel
CATIA/ALIAS
CAD/CASModel
CATIA/ALIAS
Clay Model
POLYWORKS
Clay Model
POLYWORKS
Simulation Model
(Surface Facetization)
ANSA, QUICKMESH, PowerWRAP, ...
1-5Days
Simulation Model
(Surface Facetization)
ANSA, QUICKMESH, PowerWRAP, ...
1-5Days
Simulation
PowerFLOW
1Day
Simulation
PowerFLOW
1Day
Postprocessing
PowerVIZ
Postprocessing
PowerVIZ
ResultResult
Shape Modification
of CAD/CAS Data
Shape Modification
of CAD/CAS Data
Morphing of the
Surface Mesh
(PowerCLAY)
Morphing of the
Surface Mesh
(PowerCLAY)
Turnaround
2-6 Days
Turnaround
2-6 Days
19. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 19
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
CFD Process: Geometry Preparation (Wrapping)
Complete STL Data
(imperfect facetization)
Gaps & holes
Overlaps & intersections
Interior details
Wrap
Wrapped Surface Facetization
(ready for simulation)
Water-tight single solid
Controlled granularity
Interior details removed
Preparation timereduced
from daysto hours!
Complete Set of CAD Data
Export or
facetize without
cleanup or de-
featuring
20. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 20
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
CFD Process: Geometry Morphing
Modification of the surface facetization instead of changing the CAD data
which would require a re-facetization.
21. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 21
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
CFD Process: Geometry Input
The surface
facetization
represents the
geometry only.
It does not set
the resolution
for the simulation.
Depending on the
level of detail up to
2-3 million facets
are used.
22. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 22
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
CFD Process: Modular Assembly
The complete
configuration may
be composed of
any number of
components.
Components may
be arranged in an
arbitrary fashion
and also intersect
each other.
23. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 23
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
CFD Process: Automatic Discretization
Voxels
(Fluid Cells)
Solid Body
Facets (Geometry)
Surfels
(Surface Elements)
Typical voxel counts
for external aerodynamic
cases range from 20-100
milion cells.
Geometry representation
embedded in a lattice of
cubic cells (with different
levels of resolution).
24. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 24
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
CFD Process: Simulation Timestep
Simulations are always run in transient mode
The physical time per timestep is determined by resolution and test conditions
[ ]epsec/timest
V
x
Maa
V
x
Vt LatticeLattice
∞∞
∆
⋅⋅=
∆
⋅=∆
Strictly there is no room left for the user to control the timestep
Artificially elevating the Mach number increases the time step
Example:
mmx
epsec/timesttMaCT
smV
2
10515.020
/50
6
=∆
⋅=∆⇒=⇒°=
=
−
∞
∞
That means 1 second of physical time requires 200,000 timesteps
Using Ma=0.30 instead of Ma=0.15 cuts the run time in half !
25. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 25
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
CFD Process: Transient Simulation
No explicit convergence criterion, user monitors key quantities to decide when to stop the simulation.
100,000 Timesteps
(1 Timestep = 4.7 10-6
sec.)
Averaging Window
26. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 26
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Validation Models (Scale 1:2.5).
5series touring
Open Convertible
5series Limousine with/without Mirror
Calibration Motorcycle
27. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 27
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Validation: Aerodynamic Forces
CZ2
0.114 0.105
CZ1
0.067 0.070
PowerFLOW 3.4: 0.252
CX
BMW Windtunnel: 0.252
CZ1
CZ2
-0.038 0.009 -0.027 0.006
PowerFLOW 3.4: 0.276
CX
BMW Windtunnel: 0.292
28. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 28
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Validation: Surface Pressure Distribution
Top Centerline
(Geometry not to scale)
PowerFLOW
Experiment
Bottom Centerline
(Geometry not to scale)
29. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 29
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Validation: Near Surface Flow Topology
30. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 30
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Validation: Reynolds Effect
31. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 31
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Validation: Motorcycles (Windshield Variations)
0,300
0,320
0,340
0,360
0,380
0,400
0,420
0,440
Serie LT Sport
Cx*A
Windkanal
(Aschheim)
PowerFLOW
Hot Wire Measurement
PowerFLOW
Serie
LT
Sport
32. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 32
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Analysis of Drag Generation
-0,06
-0,04
-0,02
0,00
0,02
0,04
0,06
0,08
0,10
0,0 0,1 0,3 0,4 0,5 0,6 0,7 0,9 1,0
0,00
0,05
0,10
0,15
0,20
0,25
0,30
0,35
0,40
0,0
Cx(x)Verteilung
Cx(x)Integral
PowerFLOW 3.4: 0.371
CX
BMW Windtunnel: 0.382
33. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 33
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Analysis of Lift Generation
-0,03
-0,02
-0,01
0,00
0,01
0,02
0,03
0,0 0,1 0,3 0,4 0,5 0,6 0,7 0,9 1,0
-0,40
-0,30
-0,20
-0,10
0,00
0,10
0,20
0,0
Cz(x)Verteilung
Cz(x)Integral
CZ1
CZ2
0.011 0.013
0.143 0.123
PowerFLOW 3.4
BMW Windtunnel
34. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 34
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Visualization : Transient Surface Pressure
35. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 35
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Visualization : Near Wall Streamlines
36. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 36
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Visualization : 3D Streamlines
Colors represent Near Surface Velocity Distribution
37. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 37
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Visualization: Transient Isosurface VX=0
Reverse flow (Vx<0) inside the isosurface
38. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 38
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Visualization : Transient Flow Field Slices
39. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 39
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Visualization : Transient Isosurface Cpt=0
For Cpt=0 the total pressure loss is equivalent to the dynamic free stream pressure
40. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 40
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Applications: NASTRAN Interface (Structure)
Select parts per PID
Match NASTRAN parts
with the PowerFLOW model
PLOADs [N/mm2]
on the PowerFLOW model
Map area loads onto
the NASTRAN model
Forces on
individual parts
41. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 41
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Applications: ABAQUS Interface (Heat Transfer)
Mapping heat transfer coefficients
from a PowerFLOW simulation
onto an ABAQUS structure model
42. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 42
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Applications: Detail Optimization (Mirror)
43. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 43
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Applications: Underhood Flow
44. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 44
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Applications: Motorcycles - Transient Flow Field
Near Surface Velocity
45. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 45
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Applications: Motorcycles - Transient Flow Field
Reverse Flow (Vx<0)
46. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 46
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Applications: Motorcycles - Transient Flow Field
Different Windshields
TouringTouring StandardStandard
SportSport
Helmkraft
StandardStandard
TouringTouring SportSport
47. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 47
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Applications: Motorcycle Acoustics
Punkt 92
0
20
40
60
80
100
120
140
10 100 1000 10000
Hz
dB(A)
Experiment Berechnung_grob
Punkt 101
0
20
40
60
80
100
120
140
10 100 1000 10000
Hz
dB(A)
Experiment Berechnung_grob
Punkt 110
0
20
40
60
80
100
120
10 100 1000 10000
Hz
dB(A)
Experiment Berechnung_grob
Small timesteps enable high sampling rates for
pressure and velocity fluctuations.
The upper frequency limit is determined by the
background noise of the Lattice-Boltzmann method.
The lower frequency limit depends on the physical time
(number of periods) covered by the simulation.
48. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 48
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Applications: Electronics Cooling
Goals: qualitatively – Heat Transfer Distribution
quantitatively – Surface Temperatures
Heat Transfer Coefficient
Velocity Magnitude
49. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 49
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Applications: Windtunnel Design
50. BMW Group
Dr. Norbert Grün
Lecture Series on
Road Vehicle
Aerodynamics
VKI, Brussels
May 30-June 3, 2005
Page 50
Application of a Lattice-Boltzmann Code in Road Vehicle Aerodynamics.
Conclusion
+ Maturity level sufficient for external aerodynamics
+ Short preprocessing phase due to automatic meshing.
+ Capability to handle complex geometries
(underhood/underbody).
+ Steep learning curve due to ease-of-use.
+ Does not require a numerics expert.
- Optimization loops still slower than the wind tunnel.
- Hardware requirements high for rapid turnaround.
- Thermal management capabilities still under development.