2016 optimisation a rear wing endplate in a rotating domain

Optimisation of End Plates in a Rotating Domain
Hashan Mendis
October 21, 2016
Contents
1 Introduction 2
2 Objectives 2
3 Methodology 2
4 Pre-processing 2
4.1 Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
4.2 Domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
4.3 Mesh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
4.4 Set up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
4.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
4.5.1 Straight line analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
4.5.2 Rotating Domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
5 Conclusion 11
6 Appendix A 14
7 Appendix B 14
1

1 Introduction
RMIT has a Formula-SAE team that designs a race car for the international competition. An aerodynamics package
has been designed for this years car consisting of a front and rear wing. An initial analysis has already designed the
profiles dimensions and angles for the rear wing using straight line analysis. The rear wing has yet to be simulated
in a rotating domain to determine its efficiency and how the flow structure changes.
2 Objectives
Optimize the end plate design to reduce side force in yaw.
3 Methodology
Using ANSYS CFX the end plate designs were simulated in a straight line to evaluate downforce, drag and establish
a base line.
The designs were then simulated in a rotating domain of an 8.2 m radius at 40 km/hr. The rear wing was set at
at 10o
yaw angle and 1.5o
of roll, simulating the race car in an event taking a constant radius corner. In order to
reduce computational time the rear wing was simulated in isolation. As wind tunnel models are not present using
the current wing design a mesh study will be conducted to ensure mesh independence.
4 Pre-processing
Pre-processing set up overviews the geometry, fluid domain, mesh and set up.
4.1 Geometry
The different end plate geometry is seen below. Figure 1a refers to the initial design of a square end plate with 50
mm radius edges. Figure 1b shows the end plate with a section taken off the top. The bottom cut out was taken
out to represent the actual end plate design at the request of the chief engineer.
2

(a) Initial End Plate Geometry
(b) Modified End Plate Geometry
Figure 1: End Plate Designs
In both situations the bottom edge of the end plate is 150 mm from the lowest point of the main element and
then top edge is 30 mm from the gurney flap.
4.2 Domain
Two fluid domains were created to simulate both conditions. The rear wing were placed in a straight line to
understand the flow structure and it establish a base line. The straight line domain is shown in Figure 2a
The designs were then placed a in an 8.2 m radius rotating domain at 40 km/hr as seen in Figure 2b, to understand
the change in flow structure.
(a) Straight Line Domain
(b) Rotating Domain
Figure 2: Straight Line and Rotating Domains
3

A domain study was conducted to ensure the walls did not interfere with the wings and the outlet was placed
far enough to resolve the wake. The results of the domain study is shown in Table 1 and used for the simulation.
Parameter Straight Line Rotating Domain
Outer Upwind 5 m 5 m
Outer Downwind 20 m 25 m
Outer Side 4 m 5 m
Outer Top 5 m 6 m
Inner Upwind 0.2 m 0.2 m
Inner Downwind 1.4 m 1.4 m
Inner Side 1 m 1 m
Inner Top 2 m 2 m
Table 1: Straight and Rotating Domain Dimensions
4.3 Mesh
The mesh had a tetrahedron structure as seen in Figure 3in the inner domain to conform to the complex shape,
swept bias on faces touching the inner domain and hex elements used elsewhere as seen in Figure 4. The swept and
hex elements would improve computational resources compared to a tetrahedron elements.[2] The bias of the swept
elements downwind increased in length, dampening the swirling ﬂow and making it easier to resolve the ﬂow at the
outlet.
Figure 3: Surface mesh around the rear wing
4

Figure 4: Swept Elements set up
Sizing controls were placed on the profile being an area of importance and on the inner volume mesh. They
were also placed on small edges such as the end plates and wing trailing edges.
A mesh study was conducted on the initial end plate geometry as seen in Figure 1a to ensure the results were
mesh independent. The study was conducted by halving the mesh parameters, Coefficient of lift (Cl) was monitored
to ensure values were within 1%. A k-w SST turbulence model was being used and a Y+ of 1 was used for the
profile to capture the boundary layer.[1]
After the volume and surface sizing was established the Y+ of the profile was studied.The Y+ was tested to de-
termine what level of resolution is required to accurately capture the boundary layer. A Y+ 1 required 2 million
elements in the inflation region, using a larger Y+ would reduce the mesh count and computational time.
The results of the mesh study in a straight line is seen in Figure 5, with data in Appendix A. The mesh study
found mesh independence when the volume sizing was at 100 mm, 10 mm sizing on the profile with a first layer
height of 0.01 mm, using 30 layers.
5

Figure 5: Mesh Study
The results of the Y+ study is shown in Appendix B, with the first layer and number of layers changed. The
results show a small change in the over all Cl of the rear wing, however a flow structure change is evident.
During the study the boundary layer was checked by using the Turbulent Viscosity Ratio. It is the ratio between
the turbulent viscosity and molecular viscosity . The ratio was plotted against the symmetry plane as seen in Figure
6. The model is using a Y+ of 1 with 30 layers, adjacent to the wall is the laminar sub layer, showing laminar
flow resulting in no turbulence viscosity. As the flow progressed into the logarithmic region, turbulence viscosity
increases in the boundary layer, then dissipating approaching the free stream. The transition is capture in the
inflation layer showing the boundary layer height is appropriate. This approach was checked during the Y+ study
to ensure the boundary layer is captured at all times. [3]
6

Figure 6: Turbulent Viscosity Ratio on symmetry
The velocity contour is shown in Figure 7a. It shows separation of the second element at the same place on all
cases. Looking at the Turbulence Kinetic Energy (TKE) in Figure 7b, the energy of the system changes after a Y+
of 7. As the first layer height of the boundary layer increase, after a Y+ of 7 it does capture the energy within the
layer. This will result in errors in the flow structure being unable to predict separation in other places.
(a) Velocity on Symmetry Plane
(b) Top left:Y+1,Top right:Y+7,Bottom left:Y+30,Bottom
right:Y+60
Figure 7: Contour on symmetry plane
Mesh study of the rotating domain using the initial end plate design was also conducted concluding the mesh
used for the straight line analysis was appropriate. As before an inner refinement zone had tetrahedron elements,
local refinement around, profiles and small edges as seen in Figure 8a. The outer domain contained swept elements,
increasing in length towards the outlet as seen in Figure 8b.
7

(a) Surface mesh of the wing in the rotating domain
(b) Volume mesh of the wing in the rotating domain
Figure 8: Rotating Domain Mesh
4.4 Set up
The domains were set up using the following parameters. (Table 2)
Parameter Straight Line Set Up Rotating Domain Set up
Inlet 40 km/hr 40 km/hr - Theta Component
Outlet 0 Relative Pressure 0 Relative Pressure
Ground Free Slip Wall Free Slip Wall
Wall Free Slip Wall Free Slip Wall
Rear Wing No Slip Wall No Slip Wall
Turbulence Model k-w SST k-w SST
Monitor Points Cl, Cd Cl, Cd, Cx
Table 2: Domain Set up
The k-w SST turbulence model has been used to capture the boundary layer and according to [3] it has the
ability to accurately capture the boundary layer.
The force coefficients were calculated using the following equation:
ForceCoefficient =
2 × Force
Density × V elocity2 × Area
(1)
Where:
Density = 1.225 kg/m3
Velocity = 11 m/s
Area = 0.3255 m2
Coefficient of Lift = Cl Coefficient of Drag = Cd Coefficient of Side Force = Cx
8

4.5 Results
4.5.1 Straight line analysis
The results of the straight line analysis is shown in Table 3. They show a 6% decrease in Cl and a 3% increase in
Cd compared to the original design.
End Plate Design Cl Cd
Initial Design 2.27 0.75
Modified Design 2.14 0.77
Difference 6% 3%
Table 3: Results from straight line simulation
The pressure contour on the end plate designs are shown in Figure 9. They show that the cut taken from the
modified end plate in Figure 9b takes away a portion of the high pressure region. As the bottom section of the
end plate has not changed the difference in downforce is due to the reducing of the high pressure region on the top
surface.
(a) Pressure on Initial Geometry (b) Pressure on Modified Geometry
Figure 9: Pressure on End Plate Design
The flow structure around the end plates create two vorticies at the top and bottom edge. There is a high
pressure region on the upper wing surface, pressure is lower on the other side of the end plate creating a clockwise
vortex. The opposite can be seen on the bottom side with a counter rotating vortex due to the low pressure region
of the profile. The original geometry vortex core can be seen in Figure 10a. The vortex grows down the end plate
and expands due to the rotating flow. Figure 10b shows the vortex of the modified end plate, it grows down the
end plate and expands but due to the angle of the cut the core grows vertically. As the end plate of the original
design is high and further away from the profile, the top vortex does not effect the original design as much.
9

(a) Vortex Core of Initial Geometry
(b) Vortex Core of Modiﬁed Geometry
Figure 10: Vortex Core on End Plate Design
The vortex generated by the end plate bleeds pressure from the upper and lower surface of the rear wing, taking
a section oﬀ the top of the end plate however has a small on the overall Cl and Cd.
10

4.5.2 Rotating Domain
The results for the Rotating Domain can be seen in Table 4. There is a 1% increase in Cl and no change to
Cd of the modified design. As the modified end plate has a smaller area, the force exerted by the air is smaller
resulting in a smaller side force. This can be seen in the results as a 20% reduction in side force of the modified
design. Compared to the straight line simulations the modified design was more efficient under yaw conditions.
The initial geometry reduced Cl by 11% and Cd by 3%. The modified design had a 5% reduction in Cl and 6% in Cd.
End Plate Design Cl Cd Cx
Initial Design 2.03 0.73 0.41
Modified Design 2.01 0.73 0.33
Difference 1% 0% 20%
Table 4: Results from rotating domain simulation
The change in the flow structure changes the pressure exerted by the rear wing changing the force coefficients.
The pressure distribution of the lower wing surface is similar, pressure on the upper wing surface show a larger high
pressure region of the initial geometry, resulting in more downforce as seen in Figure 11. As the modified end plate
is closed to the high pressure region of the profile, the vortex created has a larger low pressure region. This bleeds
pressure reducing downforce, accounting for the small reduction in downforce between the two designs.
(a) Pressure Contour on Initial Design (b) Pressure Contour on Modified Design
Figure 11: Pressure Contour on rear wing
5 Conclusion
The objective of the report was to optimised the end plate design to reduce side force in yaw. A model of the initial
design and modified end plate was placed in a straight line and rotating domain.
11

The straight line analysis created a base line showing the original end plate making 6% more Cl and 3% more Cd.
Due to the large end plate the flow was contained with vortexes far away from the profile surface.
The designs were then placed in a rotating domain at a 1.50
roll and 100
yaw. The results found the modified design
making similar Cl and Cd, however saw a 20% reduction in Cx. The report found that the modified geometry was
more efficient in yaw.
12

References
[1] ANSYS. ANSYS Theory Guide. 2016
[2] ANSYS. Meshing User Guide. 2016
[3] LEAP Australia, http://www.computationalﬂuiddynamics.com.au/tips-tricks-turbulence-part-4-reviewing-how-
well-you-have-resolved-the-boundary-layer/. 2013
13

6 Appendix A
Mesh Type Number of Elements x106
Cl
200 mm volume sizing , 40 mm proﬁle sizing, Inﬂation: Y+ 1 2.60 1.97
Table 5: Straight and Rotating Domain Dimensions
7 Appendix B
Y+ Number of Elements x106
Cl
1 3.34 2.23
7 2.03 2.17
30 1.98 2.22
1.16 1.16 2.22
Table 6: Y+ Study
14

2016 optimisation a rear wing endplate in a rotating domain

More Related Content

Similar to 2016 optimisation a rear wing endplate in a rotating domain (20)

More from Hashan Mendis (8)

Recently uploaded (20)

2016 optimisation a rear wing endplate in a rotating domain