Team Number 12 CFD Simulation of Airflow Around Ahmed Bodies MECH4620

CFD Simulation of Airflow Around Ahmed Bodies with

Angled Rear Flap
MICHAEL FOGARASI (z5114174), JASON SENGMANY (z5061603),
JONATHAN D’CRUZ (z5114935), SAMHITA SOMAN (z5076367)
School of Mechanical and Manufacturing Engineering,
The University of New South Wales,
Kensington, NSW 2052, Australia

Abstract

The Ahmed body is a commonly used geometry in the computational analysis of aerodynamic flow over ground
vehicles. Using the CFD package, ANSYS Fluent, the coefficient of drag and lift has been analysed over a
modified Ahmed body with an attached flap at its rear. During modelling, it was found that semi-structured
quad-dominant meshes were more effective for solution convergence than triangular meshes. The results
showed that flaps on the trailing edge of the top face had significant impact on the lift and drag experienced
by the body. An optimal angle of 15° was observed at which drag was minimised and lift maximised. The main
contributing factor to the variation of results were the pressure gradients between the front and rear of the
body. For automotive design it was found a balance between lift and drag was required to reduce instability at
high speed and increase fuel economy, however, in this study only the drag forces were considered in
optimisation. At a second configuration, flaps on the trailing edge of the body were found to have little to no
effect on the body forces due to low fluid velocity in the region surrounding the flap.

Nomenclature
Alphabetical, greek/other after English.
A wetted area (m2)
CL Lift coefficient
CD Drag coefficient
𝑣 velocity (m/s3)
ρ density (kg/m3)

1 Introduction
Reducing fuel consumption of vehicles is of great importance to reducing operational costs and addressing the growing
concerns of climate change. Reducing the aerodynamic drag experienced by a moving vehicle is an effective means of
achieving greater fuel economy. It is therefore of great interest to research methods of reducing drag by changing the
geometric profile of the body. This report analyses the flow around a simplified vehicle geometry using Computational
Fluid Dynamics (CFD) and how it affects the drag and lift experienced by the object. Ahmed et al. [1] identified that
vehicles exhibit bluff bodies moving in close vicinity to a surface. The investigation considered how changing the base
slant angle of the geometrical body affects the experienced drag. The report found that the slant angle and vertical base
size were the two main factors contributing to pressure drag, which accounted for between 76 to 85% of the total drag.
Bayraktar et al.[2] studied the effects of Reynold’s number on the drag and lift coefficients on an Ahmed Body using
CFD and wind tunnel testing. The numerical solution utilised unstructured mesh with a mix of prismatic and tetrahedral
cells and time averaged Navier-Stokes equation (RANS) was used to account for turbulent flows developed in the model.
The CFD analysis was validated experimentally through wind tunnels with improving similarity with increasing mesh
density. The paper found that the coefficient of drag decreased with increasing Reynolds number and the coefficient of
lift remained relatively constant. It was noted however that this change was relatively small with the coefficient of lift
and drag varying by 3.5% and 2.0% respectively for Reynolds numbers between 2.2M and 13.2M.
Tian et al.[3] investigated how implementing flaps in various locations on an Ahmed body with a 25 degree slant
could reduce drag. The RANS model was used to reduce the required hardware and time resources available. The CFD
results were compared to experimental wind tunnel results, and the study concluded that the use of RANS methods in the
numerical solution was valid for determining the coefficient of lift and drag due to the small error between experimental

1
Team Number 12 CFD Simulation of Airflow Around Ahmed Bodies MECH4620

and numerical. The study also found that flaps could redirect flow to change the pressure distribution behind the body
resulting in detectable reductions in the pressure drag experienced by the body. The study further investigated the
optimum flap locations and angles with respect to drag reduction for a flow with Reynolds number of 4.29M. The
maximum pressure drag reduction for a top located flap was found to be 17.9% at an angle of 20 degrees with respect to
the Ahmed body slant.

1.1 Project Description

The optimisation of vehicle aerodynamics is often carried out in computational simulations prior to application. Previous
studies conducted by Tian et al [3] explored the reduction of turbulent flow at the rear slant of an Ahmed body. A similar
approach has been adopted in this study, however, using a 2D model to investigate the lift and drag of an Ahmed body
with various flap configurations. Due to the simplification of the geometry to a single plane, the longitudinal vortices
about the side edges of the rear slant, is assumed to be negligible. Therefore, the focus of the simulations will be on
reducing the pressure drag arising from the turbulent eddies in the wake of the body. In addition, the Ahmed body was
also assumed to be modelled as to achieve similarity with wind tunnel experimentation. The fluid domain, described in
Section 1.1.2, ensures minimal effect of wall conditions on the air flow around the body of interest.

1.1 Ahmed Body Geometry

The body used in this experiment is a geometry that differs from the original Ahmed model [1]. Figure 1 below displays
the model used in this study. The rear slant angle is 30 degrees with the body located 40mm from the ground plane. This
experiment considers two flap positions: configuration 1 investigates a flap located at the upper vertex of the slant and
the top plane of the body, the flap angle is taken from the slant angle shown in figure 1. Configuration 2 investigates a
flap positioned at the lower vertex of the slant with the angle also measured relative to the slant.

Figure 1. Geometry of modelled Ahmed body

2
Team Number 12 CFD Simulation of Airflow Around Ahmed Bodies MECH4620

1.1.2 Fluid Domain

To capture the flow around the Ahmed body without significant influence from surrounding structures, a fluid domain
extends the upper side and rear of the body. Using the dimensions recommended by Costa. B [4], the chosen fluid domain
will have a length 10 times the length of the Ahmed body, a height 3 times the length of the Ahmed body, and an inlet
positioned twice the length of the Ahmed body, as seen in Figure 2. As a result, the domain is to have the following
dimensions: 9000mm x 2700mm (L x H), and an inlet positioned 1800mm to the front of the Ahmed body. The trailing
edge of the domain is considered as a 0-pressure outlet.

10L

3L L

2L

Figure 2. Schematic of fluid domain relative to Ahmed body length

2 Model Description
The CFD analysis of the airflow around the Ahmed body was carried out in ANSYS Fluent with post processing done
using CFX-Post. The Reynolds number for the flow was found to be 1.8 million corresponding to turbulent flow.
Therefore, the 2D Reynolds Averaged Navier-Stokes (RANS) equations are solved using a Realizable 𝑘 − 𝜖 turbulence
model. Using these equations, the mean flow properties of the turbulence can be efficiently predicted.

The governing equations used in the flow simulations include the mass continuity equation. This relates the rate of change
of mass in a control volume, with the net rate at which mass enters the control volume. This property is depicted in the
time-averaged form below:

𝜕𝜌̅ 𝜕
+ (𝜌̅ 𝑢̅𝑗 + 𝜌̅′ ̅̅̅̅
𝑢′𝑗 ) = 0 (2.1)
𝜕𝑡 𝜕𝑥𝑗

To model the acceleration and motion of the fluid, the conservation of momentum is used as the governing equation. This
relates the sum of forces acting on a control volume with the product of the mass of the control volume and its acceleration.
For an incompressible fluid and inviscid flow, the time-averaged momentum equation used is:

𝜕 ̅̅̅̅̅̅ 𝜕 𝜕𝑝̅
(𝜌̅ 𝑢̅𝑖 + 𝜌 ′ 𝑢 ′) +
𝑖 (𝜌̅ 𝑢̅𝑖 𝑢̅𝑗 + 𝜌̅ ̅̅̅̅̅̅ ̅̅̅̅̅̅
𝑢𝑖′ 𝑢𝑗 ′ + 𝑢̅𝑖 𝜌 ̅𝑗 ̅̅̅̅̅̅
′𝑢 ′ + 𝑢
𝑗 𝜌′ 𝑢𝑖 ′ + ̅̅̅̅̅̅̅̅̅
𝜌′ 𝑢𝑖′ 𝑢𝑗 ′) = − (2.2)
𝜕𝑡 𝜕𝑥𝑗 𝜕𝑥𝑖

Using the Boussinesq approach, the Reynolds stresses, of the form 𝜌̅ ̅̅̅̅̅̅
𝑢𝑖′ 𝑢𝑗 ′, are solved as a function of mean velocity
gradients and turbulent viscosity. The turbulent viscosity is resolved using the 𝑘 − 𝜖 realizable turbulence model. Effects
of gravity and thermal effects due to skin friction are assumed negligible in the numerical modelling.

To reduce computational cost and ensure a mesh independence within the 500 000-element limit, a 2D symmetric model
is utilised to simulate an Ahmed body placed in a wind tunnel. This assumption is based off the planar symmetry of the
body, ignoring the effects of the rounded frontal area on the results being obtained. As the legs of the body were assumed
to have little effect on the air flow below the Ahmed body, a mid-section plane is used to ignore any protrusions below.

3
Team Number 12 CFD Simulation of Airflow Around Ahmed Bodies MECH4620

2.1 Grid Description and Refinement

The fluid domain was formed using a surface body with the Ahmed body as a hollowed-out section. The geometry is
sectioned into four faces to aid in local mesh refinement, as shown in Figure 3. The face surrounding the Ahmed body
has dimensions 2700mm x 800mm (L x H) and a leading edge 500mm away from the front of the body.

Figure 3. Modelled fluid domain geometry

A quad-dominant mesh is used on all faces to create a semi-structured grid. It was found that average meshing duration
was longer for a quad-dominant mesh than a triangular mesh, which was able to reach a solution convergence at a faster
rate. Additionally, the number of elements for identical sizing parameters was lower for a quad-dominant mesh. An edge
sizing of 0.3mm is applied to the rear of the Ahmed body about the flap as shown in figure 5. This allows refinement of
the flow separation about the rear of the body. The remaining front, top, bottom and lower back edges had an edge sizing
of 0.6mm applied. A total thickness inflation layer is applied to the Ahmed body, with a maximum thickness of 0.5 mm
and having 10 layers, as seen in Figure 5. The region surrounding the Ahmed body is a face sizing of 40mm. In the face
at the rear of the body is a face sizing of 50mm, used to capture the turbulent flow in the wake of the body. A total
thickness inflation is applied to the ground with a maximum thickness of 10mm. On the upper bound of the domain,
resembling the wall of the wind tunnel, a total thickness inflation of 50mm is used. The final mesh can be seen in Figure
4 below.

Figure 4. Baseline mesh using a quad-dominant method

Figure 5. Close-up of inflation layer about the flap region

4
Team Number 12 CFD Simulation of Airflow Around Ahmed Bodies MECH4620

A grid sensitivity study was conducted on the model with a 25° angled flap by incrementally reducing the edge sizing
around the Ahmed body wall. A 2:1 edge sizing ratio was maintained between the larger sized edge surrounding the front,
top and bottom of the body to the edges surrounding the flap. The velocity and pressure at a point above the flap, and the
lift on the Ahmed body were monitored for different mesh sizes.

From the h-convergence plots in Figure 6, the solution was found to be independent of mesh size at an edge sizing of
0.3mm around the flap and 0.6mm around remaining edges.

Figure 6. H-convergence of velocity (left) and pressure (right)

A mesh convergence test was also conducted with triangular meshing. From Table 1, with identical edge sizing of 0.25mm
around the flap and 0.5mm around remaining edges, the number of elements is significantly larger for the triangular
method. In addition to this, results for the quad-dominant mesh converged at a faster rate, reaching a lower value than the
triangular mesh, as seen in Figure 7. Refer to Appendix A for detailed images of the triangular mesh.

Table 1. Comparison of quad-dominant and triangular mesh

Mesh Method Quad-Dominant Triangular

Number of Elements 204729 306895
Velocity (m/s) 38.5195 38.6069
Pressure (Pa) -259.885 -265.607

Figure 7. Comparison in velocity convergence between quad-dominant and triangular mesh

5
Team Number 12 CFD Simulation of Airflow Around Ahmed Bodies MECH4620

2.2 Boundary Conditions

The boundary conditions applied in the domain are summarised in Table 2. The 2D Ahmed body acts as a void to the
fluid flow and is given a no-slip wall boundary condition. The ground plane and top-plane is also given this condition to
replicate the conditions experienced in wind tunnel testing. The inlet (front boundary) is given a constant velocity that
replicates the uniform air velocity experienced by a vehicle moving with a constant speed. The rear boundary is chosen
as a zero-pressure boundary to replicate the atmosphere conditions of the domain.

Table 2. Boundary Conditions used in the experiment

Boundary Location Boundary Type Condition

Ground Plane Wall No-slip

Top Plane Wall No-Slip

Ahmed Body Wall No-Slip

Front Plane Inlet v=30m/s

Rear Plane Outlet Zero Pressure Outlet

2.3 CFD models

A realizable k-ϵ turbulence model was used in the analysis of the flow over the Ahmed body. This was chosen over
standard k-ϵ due to the better prediction of flow separation, and capture of rotational flow at the rear of the body [5].
k-ϵ models have been tested in predicting drag coefficients and it has been found that it is superior to standard
and RNG k-ϵ models [6], with the model predicting the drag coefficient with reasonable accuracy .

Realizable 𝑘 − 𝜖 model constants: Solver settings include:

• 𝐶2 − 𝐸𝑝𝑠𝑖𝑙𝑜𝑛: 1.9 • Scheme: SIMPLEC

• 𝑇𝐾𝐸 𝑃𝑟𝑎𝑛𝑑𝑡𝑙 𝑁𝑢𝑚𝑏𝑒𝑟: 1 • Skewness Correction: 0
• 𝑇𝐷𝑅 𝑃𝑟𝑎𝑛𝑑𝑡𝑙 𝑁𝑢𝑚𝑏𝑒𝑟: 1.2 • Gradient: Least Squares Cell Based
• Pressure: Second Order
• Momentum: Second Order Upwind
• Turbulent Kinetic Energy: Second Order
Upwind
• Turbulent Dissipation Rate: Second Order
Upwind

3 Results
The lift and drag of the Ahmed body was recorded at 5° increments in the flap angle ranging from 10° to 40°. The
coefficient of lift and coefficient of drag was then found using the equations below:

1
𝐹𝐷 = 𝐶𝐷 𝜌𝑣 2 𝐴 (3.1)
2
Where 𝜌 = 1.225kg/m3, 𝑣 = 30m/s, 𝐴 = (0.34×0.29)
1
𝐹𝐿 = 𝐶𝐿 𝜌𝑣 2 𝐴 (3.2)
2

Figure 9 shows the variation of the coefficient of lift and coefficient of drag at these angles. The maximum lift and
minimum drag occurs at an angle of 15°. A general trend can be seen in which the lift decreases and drag increases as the
flap angle is increased passed 15°. Refer to Appendix B for the table of recorded results.

6
Team Number 12 CFD Simulation of Airflow Around Ahmed Bodies MECH4620

Figure 8. Plot of lift and drag against various flap angles Figure 9. Drag and Lift Coefficients plotted for various
angles

As the angle is increased, the pressure in the wake of the body is decreased. This can be seen in the pressure
contours in figure 10, which use an identical scale from -275 Pa to -110 Pa. The larger flap angle creates a wider flow
separation, therefore generating a larger wake that extends a further distance seen on the right of figure 10. In the velocity
vector plots of Figure 11, a larger recirculating flows are shown in the body to the right. As a result, there is increased
pressure drag as the flap angle is increased, as seen with the increased drag trends in the plot from Figure 8. At the larger
flap angle seen on the right of Figure 10, a high-pressure region forms about the flap due to the sudden flow obstruction
and separation. In addition to this, the reduced cross-sectional area above the Ahmed body forces more of the flow to
divert below the body. This in turn creates a lower pressure on the underside of the body, as seen in the pressure contour
plots. As a result, the generated lift is reduced with larger flap angles.

Figure 10. Pressure contours for a 10 degree flap angle (left) and a 40 degree flap angle (right)

7
Team Number 12 CFD Simulation of Airflow Around Ahmed Bodies MECH4620

Figure 11. Velocity vectors for a 10 degree flap angle (left) and a 40 degree flap angle (right)

To account for the peak in lift and minimum value of drag at 15°, the critical angle of the flap must be considered. At
angles lower than 15°, as seen on the left in Figure 12, a larger change in cross-section above the body causes early
boundary layer separation. This in turn leads to turbulent flow and induces a larger pressure drag and decreased lift.

Figure 12. Velocity vector plot about the flap for 10 degree (left) and 15 degree (right)

At angles larger than 15°, although flow separation is delayed, the steeper angle creates a larger wake. Therefore, a larger
turbulent region is formed generating a lower pressure and more drag.

4 Optimisation Results
The optimisation goal for this study was to reduce the overall drag experienced by the Ahmed body. This represents the
real-world objective of making vehicles more fuel efficient by reducing aerodynamic drag. However, there is a trade-off
between making a car streamlined whilst also having sufficient downforce to retain traction. The assumption made was
that the mass of the car’s body would be large enough that the effects of lift can be adequately controlled. The optimisation
process involved recording the drag forces when varying the flap angle by 10° increments at a second configuration, as
seen in Section 1.1. The results of the findings can be seen in Figures 13 and Figure 14. Refer to Appendix B for the table
of results.

The main effect of the inclusion of the flap was the ability to control flow separation at the rear of the Ahmed body,
particularly with the flap positioned on the top surface. Despite its small size, a significant variation in lift and drag was
observed in the first configuration, as seen in Section 3. Without the top flap, the adverse pressure gradient immediately
after the trailing edge on the top surface of the Ahmed body, as seen in Figure 15, causes early flow separation. As the
fluid velocity in the recirculating flow, particularly around the slant edge and rear face, was significantly small, it was
found that the trailing edge flap had very little effect on the drag and lift experienced by the Ahmed body. An improvement
8
Team Number 12 CFD Simulation of Airflow Around Ahmed Bodies MECH4620

of 0.67% for drag reduction and 2.09% for lift maximisation was found between the optimal flap angle at the trailing edge
to having no flap.

Figure 13. Plot of lift and drag over various angles Figure 14. Plot of coefficient of lift and coefficient of
drag across various angles

With the flap on the trailing edge, the drag remained larger or equivalent to the drag experienced with no attached
flap. This was mostly due to the slow-moving flow over the rear flap which meant that it had almost no effect on the
variation of pressure in the wake of the body. This is further illustrated in Figure 15, in which no significant change in
pressure distribution was observed between the two flap angles, as opposed to that seen in Figure 10. Therefore, it was
concluded that the flap on the top edge at an angle of 15° had the largest effect for drag reduction on the Ahmed body.
The flap in this position helped reduce the turbulence by preventing flow separation for a longer distance, while being
positioned at an angle which follows the streamline of the flow separation. This allowed for a drag force of 64.5N as
opposed to 72.3N on the unmodified body, representing a 10% decrease in overall drag. This reduction can potentially be
very important in fuel savings over a long distance. Furthermore, this optimisation provides another opportunity to reduce
drag, where instead of adding a flap to keep the flow streamlined, the rear slant angle could be changed to the same angle
to produce similar effects to that of the flap. This would also provide a similar effect to the addition of a rear flap, without
the large flow separation and turbulence experienced past its trailing edge.

Figure 15. Pressure contours of trailing edge flap at 10 degrees(left) and 40 degrees(right)

9
Team Number 12 CFD Simulation of Airflow Around Ahmed Bodies MECH4620

5 Conclusions
The study found the effectiveness of adding flaps to the standard Ahmed body model in changing the drag and lift
coefficients. In configuration 1 the flap was able to generate a 10% drag reduction at an optimum angle of 15° when
compared to the standard un-modified Ahmed Body. This was found to occur due to the flap delaying flow separation
and minimising flow re-circulation, which reduced the pressure gradient between the main flow. The flap in configuration
2 was found to have significantly less effect, achieving an optimised drag reduction of just 0.67%. This was found to
occur due to the 30° slant of the Ahmed body, which surpassed the critical angle for flow separation, such that the flow
was relatively unaffected by the flap. The lift coefficients were found to have a negative correlation to the drag
coefficients, with the greatest lift occurring at the drag optimised configuration. The results of the study demonstrate the
potential to use flaps to control the flow to generate beneficial drag reductions in the vehicle design.

Acknowledgements

The authors gratefully acknowledge the guidance and dedication of Cheng and the University of New South Wales
school of Mechanical and Manufacturing Engineering for providing the resources to complete this study.

References

[1] S. R. Ahmed, G. Ramm, and G. Faltin, "Some Salient Features Of The Time-Averaged Ground Vehicle Wake," 1984.
Available: https://doi.org/10.4271/840300
[2] I. Bayraktar, D. Landman, and O. Baysal, "Experimental and Computational Investigation of Ahmed Body for Ground
Vehicle Aerodynamics," 2001. Available: https://doi.org/10.4271/2001-01-2742
[3] J. Tian, Y. Zhang, H. Zhu, and H. Xiao, "Aerodynamic drag reduction and flow control of Ahmed body with flaps," Advances
in Mechanical Engineering, vol. 9, no. 7, p. 1687814017711390, 2017.
[4] B. Costa. (2015, 5). Studying the Airflow Over a Car Using an Ahmed Body Available:
https://www.comsol.com/blogs/studying-the-airflow-over-a-car-using-an-ahmed-body/
[5] S. Wasserman. (2016, 6). Choosing the Right Turbulence Model for Your CFD Simulation. Available:
https://www.engineering.com/DesignSoftware/DesignSoftwareArticles/ArticleID/13743/Choosing-the-Right-Turbulence-
Model-for-Your-CFD-Simulation.aspx
[6] Z. Di, "Comparison of Various Turbulence Models for Unsteady Flow around a Finite Circular Cylinder at Re =20000,"
Journal of Physics: Conference Series, vol. 910, no. 1, p. 012027, 2017.

10
Team Number 12 CFD Simulation of Airflow Around Ahmed Bodies MECH4620

Appendix A: Triangular Meshing Method

Figure A-1. Triangular mesh using edge sizing of 0.25mm around flap and 0.5mm around remaining edges

Figure A-2. Inflation layer around the flap region

Figure A-3. Pressure convergence comparison between quad-dominant mesh vs triangular mesh

11
Team Number 12 CFD Simulation of Airflow Around Ahmed Bodies MECH4620

Appendix B: Simulation Results

Table B-1. Results obtained for flap positioned on the trailing edge of top surface

Flap Angle Drag (N) Coeff. of Lift (N) Coeff. of

(°) Drag Lift
10 71.6367 1.317984 330.538 6.081292
15 68.4397 1.259165 345.614 6.358663
20 69.9338 1.286654 335.926 6.180422
25 73.217 1.347058 315.742 5.809073
30 76.1746 1.401473 301.381 5.544857
35 79.0768 1.454868 291.49 5.362881
40 82.4249 1.516467 280.09 5.153142

Table B - 2. Results obtained for flap positioned on trailing edge of rear slanting face

Flap Angle Drag (N) Coeff. of Lift (N) Coeff. of

(°) Drag Lift
0 74.70018 1.374346 312.7125 5.753335
10 74.57275 1.372002 314.9014 5.793608
20 74.4478 1.369703 316.0786 5.815266
30 74.39988 1.368821 316.6241 5.825301
40 74.42648 1.369311 317.5966 5.843193
50 74.31106 1.367187 317.9192 5.849129
60 74.19779 1.365103 318.3547 5.857141
70 74.2721 1.36647 319.3975 5.876327
80 74.22579 1.365618 318.9462 5.868024
90 74.25515 1.366158 319.3116 5.874748

12