rolf sestriere

Matthew N. Godo, Ph.D.
FieldView Product Manager, Intelligent Light
David Corson,
Program Manager - AcuSolve, Altair Engineering
Steve M. Legensky
General Manager, Intelligent Light
Yves-Marie Lefebvre
Sales & Support Engineer, Intelligent Light
Examining the Aerodynamic Performance of
Commercial Bicycle Racing Wheels using CFD
Intelligent Light
• Established in 1984
27 years in July 2011!
Global Customer Base
• Two components to
our business:
– FieldView Software
– Applied Research Group
• Customer-driven R&D
• CFD & Post-processing
Research
• CFD for Wind Energy
Our Mission
To help our customers using CFD
to do more with less and make
better decisions
How we accomplish our mission:
• CFD post-processing products & methods
• Workflow automation
• Development of new CFD
methodologies
The Right Wheel?
Background
• Wind Tunnel testing used extensively in
cycling for over 20 years
– Typical for Zipp, 85h at $850/h,
run 3 or 4 times per year
• Benefits to cyclists from Wind Tunnels
– Improved rider positioning for lower drag
– Significant performance improvements in
equipment design
Advertisement ca
2007
• Current status
– Still considerable component variations
– UCI rule changes & enforcement can be
rapid & unpredictable
– Wind Tunnel reaching its limit today
– Interpretation of results ‘controversial’
Zinn, L., “Spoked aero’ wheels catching up with discs”,
Inside Triathlon, 1995, 10(4), p 36-37
How much does it matter?
Tour de France 2008
Stage 20 Individual Time Trial
1
2
3.0%
3

4
5
6
Finish Position
From Greenwell et.al.
 Wheel drag is responsible for 10% to
15% of total aerodynamic drag
Rider makes up the drag majority
7
8
9
10
11
12

Improvements in wheel design can
reduce drag between wheels by as
much as 25%
Overall reduction in drag can be on
the order of 2% to 3%
13
14
15
0
1
2
3
4
5
Percentage Time Difference
IronManTM Lake Placid Triathlon 2008
Male 45-49 Age Group
Q
Q
Q
Q
Q
1
3.3%
2
3
4
5
6
Finish Position

7
8
9
10
11
12
13
14
15
0
1
2
3
4
5
Percentage Time Difference
6
7
8
Boundary Conditions
inner wheel
(incl. spokes)
hub
mesh for inner wheel
is created separately
Realistic spoke rotation
outer wheel
(incl. tire)
non-conformal interface
Mesh Displacement
(unsteady)
Moving Reference Frame
ground plane
Boundary Conditions
top tube
Surrounding
domain
head tube
down tube
ground plane
ground plane

Ground plane


yaw angle

Far Field



no-slip surface
translational speed of 20 or 30mph
uniform velocity profile
yaw angles from 0o to 20o
Fork, frame, caliper, brake pads


no slip surface
zero relative velocity
Postprocessing Objectives
Performance Metrics
• Resolved Forces
• Turning Moments
• Aerodynamic Torque
Top View
Turning Moment
Wind Velocity
(effective)
Axial Drag Force
Side (Lift) Force
𝑅
𝑀=
Bike Velocity
(relative)
𝑟 ∙ 𝑇𝜃𝑥 𝑑𝐴
0
• Power to Overcome Aero Resistance
P = 𝐹𝐷 𝑉 +
𝑏𝑜𝑡ℎ 𝑠𝑖𝑑𝑒𝑠
Side View
Vertical Force
Wind Velocity
(effective)
𝑀𝜔
Requirements
Axial Drag Force
• Quantitative & Qualitative
• Easily automated & scalable
• Batch compatible on remote clusters
Direction of Wheel Rotation
CFD Results vs Wind Tunnel Data
Circumferential Variation, Side Force
Direction of Flow
Turning Moments, All Wheels
Turning Moment vs. Yaw Angle at 20mph
Turning Moment vs. Yaw Angle at 30mph
1
0.2
0.4
0
0
Moment [N·m]
Moment [N·m]
2
0
0
-0.2
Rolf Sestriere
Zipp 404
Zipp 808
Zipp 1080
HED TriSpoke
Zipp Sub9 Disc
(right axis)
-2
Rolf Sestriere
Zipp 404
Zipp 808
Zipp 1080
HED TriSpoke
Zipp Sub9 Disc
(right axis)
-1
-0.4
-0.4
-4
0
2
4
6
8
10
12
14
Yaw Angle [degrees]
16
18
20
0
2
4
6
8
10
12
14
Yaw Angle [degrees]
16
18
20
Wheel Only Studies
Rolf Sestriere
Zipp 404
Zipp 1080
HED TriSpoke
Zipp 808
• Configurations: 6
• Speeds : 1
– 20mph
• Yaw Angles: 1
– 10o
• Design Points: 6
• Time steps: 256
– For each design point
• Total steps: 1536
• ~1.2TB of data
– ~200GB per wheel
Zipp Sub9
Streaklines revealed strong periodic shedding,
distinctive for each wheel studied
Strouhal No., All Wheels
(20mph, 10 degrees yaw)
Strouhal No.
6.0
5.0
3.0
2.0
1.0
Strouhal range obtained from resolved drag, side, vertical forces and
moments
Expanding the Scope…
– 0o, 5o, 10o, 15o, 20o
• Design Points: 90
• Time steps: 256
– For each design point
• Total steps: 23040
• Numbers of merit (for each step)
–
–
–
–
Drag & Side Force
Turning Moment
Aerodynamic Torque
Total Power
Wheel only
• Yaw Angles: 5
Reynolds Carbon
– 20mph, 30mph
Zipp 404
Blackwell Bandit
• Configurations: 9
• Speeds : 2
Zipp 1080
HED TriSpoke
Solver & Coprocessing
AcuSolveTM






Based on stabilized
Galerkin/Least Squares
Second order accuracy
 Time and space
Equal order interpolation
for all variables
Globally & locally
conservative
Fully coupled pressure/
velocity iterative solver
Fully parallelized for
shared mem & clusters
Volume Mesh
Parallel AcuSolve
Processes
Socket Communication
User Requests
(variables, elements, etc.)
Python
Script
Prism layers
FieldView UNS file
mini-grids
Batch Postprocessing Workflow
cluster or cloud system
minigrids
forces
forces
forces
Batch
XDB
XDB
Solver



Parallel, 54 cores
4-6h elapsed per design
point
Only mini-grids saved
Batch Postprocessing



FVXTM scripts used for all
performance metrics
Concurrent, typically 40
jobs in queue
Less than 1h per job
Batch Postprocessing Workflow
cluster or cloud system
Solver
runs
Batch
XDB
XDB
Solver



Parallel, 54 cores
4-6h elapsed per design
point
Only mini-grids saved
FTP
Batch Postprocessing



FVXTM scripts used for all
performance metrics
Concurrent, typically 40
jobs in queue
Less than 1h per job
XDB Data Reduction



46X smaller files
Full numerical fidelity
FTP to local desktops for
interactive postprocessing
Power vs Time vs Yaw Angle
TriSpoke
Industrial Relevance
“For Zipp, working with Matt on this paper [AIAA-20101431] was largely what spurred the Firecrest rim shape
development on the handling side. Before this, we had
some super fast shape concepts, but realized from the data
that there was just so much more to be done on the
handling side, that we spent a few extra months in
development chasing favorable handling characteristics
(rearward center of pressure and shedding behavior).
Ultimately we still can't replace the wind tunnel with CFD,
but the ability to understand and predict so many aspects of
performance and handling is pretty awesome!
And that's just the beginning...”
Zipp 404 Firecrest
cross section profile
Josh Poertner, Category Manager,
Zipp Speed Weaponry, Indiana