A Robust CFD Methodology For Physically Realistic and Economically Feasible Results in Racing - Part II Intake Cowl

Downloaded from SAE International by Embraer S/A, Sunday, December 03, 2017
SAE TECHNICAL
PAPER SERIES 2006-01-1441
A Robust CFD Methodology for Physically

Realistic and Economically Feasible
Results in Racing – Part II: Intake Cowl
J. Logan Marshall, Yaling Ma, Adrian M. Mistreanu and James H. Leylek
Clemson University
2006 SAE World Congress

Detroit, Michigan
April 3-6, 2006
400 Commonwealth Drive, Warrendale, PA 15096-0001 U.S.A. Tel: (724) 776-4841 Fax: (724) 776-5760 Web: www.sae.org
The Engineering Meetings Board has approved this paper for publication. It has successfully completed
SAE's peer review process under the supervision of the session organizer. This process requires a
minimum of three (3) reviews by industry experts.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise,
without the prior written permission of SAE.
For permission and licensing requests contact:
SAE Permissions
400 Commonwealth Drive
Warrendale, PA 15096-0001-USA
Email: permissions@sae.org
Tel: 724-772-4028
Fax: 724-776-3036
For multiple print copies contact:
SAE Customer Service

Tel: 877-606-7323 (inside USA and Canada)
Tel: 724-776-4970 (outside USA)
Fax: 724-776-0790
Email: CustomerService@sae.org
ISSN 0148-7191
Copyright © 2006 SAE International
Positions and opinions advanced in this paper are those of the author(s) and not necessarily those of SAE.
The author is solely responsible for the content of the paper. A process is available by which discussions
will be printed with the paper if it is published in SAE Transactions.
Persons wishing to submit papers to be considered for presentation or publication by SAE should send the
manuscript or a 300 word abstract to Secretary, Engineering Meetings Board, SAE.
Printed in USA
2006-01-1441
A Robust CFD Methodology for Physically Realistic and

Economically Feasible Results in Racing – Part II: Intake Cowl
J. Logan Marshall, Yaling Ma, Adrian M. Mistreanu and James H. Leylek
Advanced Computational Research Laboratory (ACRL)
Department of Mechanical Engineering
Clemson University
Copyright © 2006 SAE International
ABSTRACT Through detailed analysis of flow physics novel

configurations were identified and analyzed to determine
Part V of the present five-part paper focuses on a intake cowl systems that produced the highest flow rate.
research project designed to uncover new and Performance was improved significantly above the
innovative means of increasing airflow to the engine baseline configuration; flow rate was improved as much
within NASCAR rules. Computational Fluid Dynamics as 222% in one case. A new phenomenon was
(CFD) offers an alternative to the current “build-and- discovered that directly coupled external aerodynamics
bust” technique reducing costs and time per design with intake cowl flow. For the first time in the open
iteration, and provides the sponsor with a physics-based literature, detailed physics involving intake cowl flow
design tool with true predictive capability. Armed with a phenomena are described in detail.
validated CFD based design system, the team could
respond quickly to rule changes by analyzing new INTRODUCTION
configurations through simulations avoiding the
fabrication and track testing that is currently necessary. In NASCAR, as in most industries, competition fuels
A robust and easy-to-use CFD methodology for this the need for research. Power or engine output is of
class of problems was developed and implemented to utmost concern. Increasing engine horsepower and
understand the flow physics and explore novel torque by less than one percent will give a team a
configurations, as described in Part I. Specifically, these significant advantage over the competition. This
problems involve external flow around a racecar increase can only come from burning more fuel. In turn,
traveling at 180 mph. One critical issue attracting increased fuel burn must have increased air for
attention in this research was the link between external combustion. Current rules in NASCAR are very strict
high speed flow and how it negotiates a hairpin turn into concerning the modifications made to the vehicle and
the intake cowl. engine. The engine air inlet is located at the base of the
This comprehensive computational methodology front windshield. This is known as cowl induction. This
was used to reduce or eliminate errors due to geometry, opening leads to an air box that encompasses the air
grid, discretization, false diffusion, and turbulence filter and empties into the carburetor. The present
modeling. The RANS equations were solved on an research project is focused on finding new and inventive
exact electronic replica of the geometry. A multi-block, means of increasing airflow to the engine within
multi-topology, unstructured, adaptive grid made up of NASCAR rules.
high-quality, high-density finite volumes was created for Computational Fluid Dynamics (CFD) can be used
use in all the simulations presented here. Fully to reduce the design cycle time by avoiding the current
converged and grid-independent solutions based on “build-and-bust” technique, saving money and time,
strict convergence criteria were obtained for all the which is extremely important in NASCAR due to the
cases without an exception. Turbulence closure was intense weekly race schedule. The race teams need a
obtained through the realizable k-İ turbulence model physics-based design tool that could provide consistent
with non-equilibrium wall functions to account for and accurate results in a timely and economical fashion.
curvature and pressure gradient effects. This would eliminate the effort put into designing and
1
testing many solid prototypes. There are, however, model and standard wall functions (SWF). Many
unique problems for CFD practitioners, as NASCAR different configurations, using approximately 3.65 million
teams typically are not well prepared for this technology. structured Cartesian cells, were explored and compared
For example, computer aided drawing (CAD) data may to experimental data with good results.
not be available, and boundary condition information Andra et al. [7] studied the effects of boundary and
may be sparse. Therefore, it is necessary to develop an geometry simplification on the numerical simulation of
especially robust and easy-to-use CFD methodology as front-end cooling. Fluent was used with its SKE
described in Part I. Beyond that, the specific goals of turbulence model and standard wall functions. The
this project were to understand the complex flow RANS equations were solved on an unstructured
physics, explore novel configurations for increased tetrahedral grid of 1.2 million cells. With cooling system
engine airflow through “what if?” type experiments, and flow as a gauge, the results showed that geometric
document the performance of readily available simplification in the under hood area was useful for quick
turbulence models. An evaluation of available near-wall evaluations of alternative designs in early design stages
treatments was performed as well for this class of when packaging issues are not finalized. Secondly, the
problems involving internal flow as it is coupled with exit plane was moved from the wind tunnel exit to
external flow around a race car traveling at 180 mph between the B and C pillars of the vehicle with no effect
(miles per hour). on the flow rate through the cooling package. This
Often, race teams report that dynamometer tests of information, in particular, is important to the current
certain engines demonstrate great performance, but study. Gillieron et al. [8] evaluated the potential of CFD
track testing of the same engine show less than to predict under hood flow phenomena. The primary
adequate performance. Therefore, a focal point of the concerns, air speed and turbulence values at the air inlet
present research was how the airflow negotiates a and in the engine compartment were solved using
hairpin turn into the intake cowl, and the effects this has Fluent’s first order, incompressible scheme with the SKE
on external aerodynamics. turbulence model and standard wall functions.
2-D and 3-D external aerodynamic simulations were
LITERATURE REVIEW performed and reported by different researchers in
reference [9] to [20]. There are numerous other papers
Little or no information exists in the open literature, describing aspects of CFD-aero simulations such as
either experimental or computational, on the intake cowl underbody flow, wheel cavity flow, and wake flow, but
aerodynamics related flow phenomena found in this little additional information pertains to the current study.
study. Similarly, the authors could not find a single
paper on high-speed external aerodynamics coupled PRESENT CONTRIBUTIONS
with a intake cowl into flow. There is also no literature
that describes the methodologies involved when only a The unique set of contributions of this research may
solid prototype is available for the creation of the be listed as follows:
electronic description of the computational domain. A
number of papers addressed the relevant issue of x Develop a complete computational methodology for
current study are discussed as follows. this class of problems while striking a balance
Laise and Bayless [1] extensively documented the between physical realism and economic feasibility;
aerodynamic development of a NASCAR Winston Cup x Investigate the physical mechanisms coupling
race car through purely experimental means. Since the external aerodynamics and internal air box flow;
date of the publication, NASCAR rule changes designate x Explore novel techniques such as geometry
a new fully enclosed air cleaner housing as seen in this modifications, energizing the boundary layer, and
current study. Other purely experimental vehicle boundary layer suction to increase engine airflow;
developments are reported, such as in references [2] x Package and transfer the complete computational
and [3]. methodology to a typical race team which has no
Katz and Dykstra [4] reported on the computational previous experience with CFD;
methods applied to the aerodynamic development of a x Explain why some engines perform well on
prototype race car. A three-dimensional inviscid solver dynamometer, but not in race conditions.
was used to perform numerous “what-if” scenarios
centered on underbody flow and wing cluster design. SIMULATION DETAILS
Flow physics and CFD methods are not discussed.
Werner et al. [5] performed an aerodynamic optimization Within any CFD simulation four important tasks must
of the Opel Calibra ITC race car using both experimental be completed in order to provide consistent and
and computational fluid mechanics. Internal and accurate results. These tasks include proper: 1)
external flows were considered using the RANS solver computational modeling; 2) geometry and grid
and RNG turbulence model of Star-CD. Williams and generation; 3) discretization schemes; and 4) turbulence
Vemaganti [6] performed a calibration study for front-end modeling. Satisfaction of these critical steps insures that
cooling airflow using a Ford in-house code. The solver the right answers are found for all the right reasons.
is based on the RANS equations for incompressible
fluids employing the standard k-H (SKE) turbulence COMPUTATIONAL MODELING
2
The flows encountered in the present simulations Once computational modeling is completed and the
are highly three-dimensional due to curvature of domain extents are set, the geometry internal and
surfaces such as the hood, windshield, and roofline. external to the cowl was created with exact detail. Since
The internal flow in the air box includes even more no CAD data was available, a new technique was
complex geometry leading to a three-dimensional needed to insure that an electronic definition of the exact
regime. vehicle geometry was captured. A laser scanning
The nature of this study leads to a steady-state system was employed to scan and digitize the needed
simulation. The actual environment being simulated is a surfaces such as hood, roof, and windshield for input
vehicle moving in a straight line at a constant speed. into CAD software. Interior air box dimensions were
This simulation involves the external aerodynamics of obtained through a similar system. Thin slices of the air
the vehicle that is upstream of the reciprocating engine box were cut and digitized laser points taken. The laser
flow processes. Also, the area of interest is situated far scanning information resulted in a “point cloud” that was
upstream of the highly unsteady flow region typically used to generate computational model, the specific
found downstream of the C-pillars. This simulation does technology can be found in Part I.
not take into account any engine components below the Figure 1 shows the computational domain for
air box exit. Steady-state simulations have been widely the two-dimensional centerline study case used to
used in the past to characterize total pressure losses generate the boundary conditions for the clipped three-
throughout the engine geometry and pressure on the dimensional domain. The reference three-dimensional
external surfaces of vehicles. Also, this simulation does computational model is found in Figure 2. A partial
not try to resolve wake structures and vortex shedding; section of the hood and roofline, and the windshield are
therefore a steady-state simulation will provide the included as well as the cowl air box. A significant
information in question. The main physics in question volume above the car is included for an accurate model
are the interactions of external aerodynamics with the of the free stream. The upper boundary follows an
amount of air that can be taken into the engine through actual streamsurface where no flow crosses.
the cowl opening at the base of the windshield. Note The quality and density of the grid will play a large
that a serious downside of unsteady simulations is that role in the accurate modeling of the flow physics. The
they are much more expensive, time-consuming, and techniques used to obtain a high density, high quality
difficult to solve. An unsteady simulation could take up grid are discussed in Part I in detail. Figure 3 shows a
to ten times longer to converge if convergence is ever sample grid along the cowl centerline.
reached at all. Such simulations are not consistent with A hybrid topology approach was used. In areas of
the stated objectives regarding the need to develop complex geometry and areas of high interest, an
robust, time effective CFD technologies. unstructured tetrahedral grid was used. Structured grids
It is very important that the correct decision about are placed only in the boundary layers on the external
the amount of space encompassed by the computational vehicle surfaces and designed so that the hexahedral
domain be made. The extents of the domain considered elements are aligned in the flow direction to reduce
are closely linked to the type and location of the numerical diffusion. A multi-block and multi-topology
boundary conditions. Inlets and exits can only be placed meshing technology was employed to control grid size.
where known information exists concerning velocity, Also, hexahedral grids was stepped and stretched on
+
pressure, temperature, and turbulence. Boundary layer the vehicle surfaces to efficiently maintain y values and
growth on the hood must be modeled to correctly navigate the 180-degree turn into the cowl air box. This
represent the conditions experienced at the cowl stair-step method allows for quick, but well-controlled
opening (history effects). changes in grid density and sharp turns. Pyramids were
To reduce the number of finite volumes used in the used for transitions between hexahedral and tetrahedral
simulation, a preliminary two-dimensional simulation was grids. For transition into the free stream a non-
performed on the centerline plane in order to resolve the conformal interface (NCI) was used.
boundary layer growth on the hood. This information The geometry and grid were designed using a
was used to clip the three-dimensional computational modular approach so that new designs can be quickly
domain and reduce its size. This allows the inlet added to the exterior geometry with minimal re-meshing.
placement closer to the area in question; therefore All of the mesh need not be generated each time the
higher-density grids can be used where needed. The cowl geometry is changed. On average, the
exterior domain exit was also moved closer to the area computational grids contained about five million finite
of interest using the same technique. This technique volume cells.
captures all the essential physics with much more
economy. The vehicle and all details are symmetric BOUNDARY CONDITIONS AND DISCRETIZATION
about its centerline allowing for a half-domain to be SCHEMES
considered further reducing computational needs.
Overall, computational resource requirements were The boundary conditions found here involve inlets,
reduced by nearly 60 %. The internal cowl exit plane is exits, walls, and symmetry planes. The inlet conditions
placed at the entrance to the carburetor where the static were extracted from a comprehensive two-dimensional
pressure can be prescribed. full-vehicle simulation of the centerline plane including
the far-field (upstream, downstream, and above the test
3
vehicle). This allowed for a smaller, more manageable, x Scoop 2 (SC2), where an arbitrary scoop was added
and accurate three-dimensional model. The two- where the flat rear wall of the REF air box along with
dimensional data captured all pertinent variables needed a more rounded forward entrance wall;
to define the inlet conditions and boundary layer x Scoop 3 (SC3), the same as SC2 without the
formation on the hood. Two exit conditions are needed, rounded forward entrance wall.
one pertaining to external aerodynamics and the other
for internal cowl flow. The same two-dimensional model Upon these geometries the effects of an energized
was used for the external exit. The internal, or cowl, exit boundary layer, boundary layer suction, and a
data was obtained from tests performed by the sponsor. combination of these techniques were performed.
The conditions here specify a static pressure across the Additionally, the performance of available turbulence
exit. This value was determined by varying the pressure models and near-wall treatments was performed and is
until the known test flow rate of 580 CFM (cubic feet per presented for the REF geometry. The results are
minute) was obtained. This is the value that the engine presented in a chronological order. Since an increase
receives at 180-mph, according to tests. The resulting in flow rate and additional physics were not found in
pressure, 0.19 PSI (pounds per square inch) was used cases 9, 10, 12, 13, and 14, they are omitted from the
for all simulations. A half model in the lateral direction discussion.
was created with symmetry planes on either side. All
surfaces are modeled as adiabatic since heat transfer is TABLE 1. TEST MATRIX
not of concern. The upper boundary, which happens to Case Geometry Variation
coincide with a streamsurface, has a prescribed free
1 Roush Standard
stream pressure, which allows flow in and out of the
2 Roush Hood BL Energized
domain to accurately model flow around the race car.
No mass flow enters through this “slip” surface. All the 3 Roush BL Suction (-1.0 PSI)
results presented in this study are declared grid- 4 Roush SKE
independent and fully converged. 5 Roush RNG
6 Roush SWF
TURBULENCE MODEL AND CONVERGENCE 7 Scoop 2 Standard
8 Scoop 2 Hood BLEnergized
The choice of turbulence model is key in the trade 9 Scoop 3 Standard
off between physical realism and economic feasibility as 10 Scoop 3 Hood BL Energized
documented in Part I. Realizable k-H turbulence model 11 Scoop 3 BL Suction (-1.0 PSI)
(RKE), with non-equilibrium wall function (NWF) near- 12 Scoop 3 BL Suction (-1.5 PSI)
wall treatment, is used for the modeling of turbulence in 13 Scoop 3 BL Suction (-0.5 PSI)
the present study. 14 Scoop 3 Combo (-0.5 PSI)
Convergence for this simulation was declared when 15 Scoop 3 Combo (-1.0 PSI)
global mass and energy imbalances fell below 0.1%. In
most cases these imbalances were driven three orders
of magnitude below 0.1% to insure full convergence. In
Contours and vectors are shown on the vehicle
summary, convergence is declared when:
centerline, planes parallel to the cowl inlet and outlet, a
x Residuals of all dependent variables fall at least plane parallel to the cowl bottom, and a cross plane
three orders of magnitude or more; normal to the flow direction. Contours of total pressure
x Velocity and turbulence kinetic energy profiles do (Pt), static pressure (Ps), and velocity magnitude (V,
not change; mph) demonstrate the physical mechanisms that play a
x Global mass and energy imbalances fall below role in overall cowl air box performance while velocity
0.1%. vectors are used to show flow patterns. All pressures
are given in PSI (gage) and all velocities are given in
RESULTS AND DISCUSSIONS mph.
All together, fifteen different simulations were REFERENCE CASE RESULTS

performed on three different geometries. The complete
test matrix can be seen in Table 1. Standard indicates The Roush air box built for the Daytona
that RKE is used with non-equilibrium wall functions. configuration of the 2000 Monte Carlo race car served
Combo indicates a combination of boundary layer as the reference case for all three-dimensional cowl
suction and an energized boundary layer. The simulations. A 3-D computational model was developed
pressures shown in the table below are gage pressures. that eliminated the need to model the entire car body
The three geometries used are as follows and can be while still maintaining a high level of modeling accuracy.
seen in Figure 4: Based on the boundary conditions provided by the
sponsor, the air flows through the cowl into the
x The reference geometry (REF), also known as the carburetor at approximately 580 CFM. The cowl outlet
Roush Daytona configuration;
4
pressure was varied in a series of numerical studies until the carburetor and will result in additional variations in
this flow rate was obtained. cylinder-to-cylinder performance.
The velocity contours and velocity vectors along the
cowl centerline plane in Figure 5 highlight that the high- REFERENCE CASE WITH ENERGIZED BOUNDARY
speed flow is not able to navigate the sharp turn into the LAYER
cowl. Thus, a large recirculation region is formed in the
inlet region so that the effective area seen by the flow In order to assess the effects of a thinner boundary
entering the cowl is only about 20 % of the actual inlet layer on cowl performance, a “proof-of-concept”
area defined by the bounding metal surfaces. The small simulation based on the REF model was developed and
amount of high-speed flow that does enter the cowl is analyzed. This simulation has the designation RBL
not diffused efficiently, so it stays close to the cowl wall. (Reduced Boundary Layer).
The small scoop region at the base of the cowl does The computational model used is an exact duplicate
decelerate the flow nearly isentropically, but the flow of the REF geometry. However, in the current
must accelerate again as it moves over the large hump simulation imposing a constant total pressure inlet has
for the carburetor. Because of the corner shaping at the decreased the boundary layer thickness from the REF
cowl outlet, the flow must accelerate as it turns into the case. In essence a brand new boundary layer is started
carburetor, apparently resulting in a separation region at at the inlet plane resting on the vehicle hood. This is
the carburetor inlet reducing the effective flow area. accomplished through a flat Pt profile at the inlet.
Nevertheless, the flow mechanisms most damaging By reducing the boundary layer thickness along the
to cowl efficiency are not due to the cowl geometry, but hood of the car, the amount of high-momentum flow
the cowl placement and location, which are mandated by entering the cowl is greatly increased. This observation
NASCAR. The contours of total pressure along the cowl represents a fundamentally important alteration in airflow
centerline plane shown in Figure 6(a) illustrate the thick management over the hood of a race car for the purpose
boundary layer (low energy fluid that is dragged down by of increasing the flow rate into the engine. It is important
the hood) that forms over the length of the hood. This to note that, even without any optimization, the RBL
boundary layer separates the highest-energy (high total case shows a 28.9 % higher CFM than the REF case.
pressure) flow from the cowl inlet by almost two inches The flow physics behind this improvement are described
and allows only a small amount of high-energy flow to below.
enter the cowl. It is primarily the low energy flow in the Contours of total pressure and static pressure in
hood boundary layer that enters the cowl. Highest Figure 11 show the effect of a thin boundary layer
losses in the Roush (REF) configuration are in the approaching the cowl. Because the high-speed, high-
boundary layer at the cowl inlet. Furthermore, the energy flow is closer to the hood surface, more is able to
stagnation region shown by contours of static pressure enter the cowl, as shown in Figure 11. The high-velocity
in Figure 6(b) is downstream of the cowl inlet at the base air is unable to turn the sharp corner completely, but
of the windshield. The highest pressure then is not at because of the thin boundary layer, the small amount of
the cowl inlet but downstream of it. Consequently, the flow that is deflected into the cowl has the highest value
pressure differential through the cowl, which is the of total pressure. This high-speed, high-energy flow
driving force behind the engine airflow, is reduced, stagnates on the rear of the cowl, forming a high-
further reducing the efficiency of this reference pressure region closer to the cowl inlet, as shown in
geometry. The flow in the cowl stagnates against the Figure 11(b). The high-energy flow that is not deflected
forward-most wall, forcing the air back towards the cowl into the cowl forms another stagnation region
outlet Figure 6(b). immediately downstream of the cowl at the base of the
Figure 7 shows the flow features at and below the windshield. The close proximity of these two regions
cowl inlet plane viewed from the top. As shown in Fig. turns them into an excellent high-pressure source to
6(a, b), the effective inlet area for high-speed, high- drive more airflow toward the low-pressure sink present
energy flow is reduced by as much as 80 %. Sidewall at the carburetor inlet plane.
effects account for about 5 % of this blockage, but Other flow features inside the cowl are similar to
compared to the inlet losses described above, this those seen in the REF case, but there are significant
feature is less significant. The recirculation region also differences in the details of the flow. The high-energy
prevents the inlet geometry from efficiently diffusing the flow entering the cowl diffuses somewhat, with the small
high-speed flow, so it remains close to the back wall of scoop at the bottom of the filter housing acting as a
the cowl. stagnation region. However, the standard Roush cowl
As shown in Figure 8, sidewall and pressure effects design is not at all an effective diffuser, and most of the
combine to create a secondary flow that moves towards high-energy flow maintains a relatively high velocity near
the cowl outlet. However, the total pressure contours in the cowl wall. In addition, as in the REF case, the RBL
this figure show the low energy of this flow compared to case shows the low-pressure regions near the
that of the free stream. This secondary flow combines carburetor inlet that is a result of flow accelerating
with the pressure differentials shown in Figure 6(b) to around the corners of the cowl outlet. However,
create the large, counter-rotating vortex structures on because the air entering the RBL cowl has a higher
the cowl outlet plane shown in Figure 9. This flow enters velocity and energy level than that entering the REF
cowl, the magnitude and extent of the low pressure
5
regions are more pronounced, indicating that the flow Figure 14 shows the velocity contours and vectors
turning into the carburetor is moving at a relatively high associated with an energized boundary layer, which was
velocity and is more susceptible to separation. Finally, generated in a separate simulation. The boundary layer
the airflow stagnates at the back wall of the cowl, as was energized as mentioned above for the REF
shown in Figure 11(b), creating a high-pressure region geometry. Flow separates immediately from the front of
that forces air back towards the cowl outlet. Clearly, the cowl due to the much higher velocities. This causes
these observations highlight the opportunities available the reformation of the recirculation region seen in the
for the creation of a substantially better cowl design. standard Roush (REF) case. The scoop at the rear of
An analysis of contours of velocity magnitude and the cowl helps to diffuse the flow better than the Roush
velocity vectors, shown in Figure 10, further highlights (REF) box. An increase in volumetric flow rate by about
the unique effects of a reduced boundary layer over the 37.7 % is seen compared to the reference Roush case.
hood of the car. Because of the larger amount of high- The new geometry creates a sharp edge at the rear of
velocity air that enters the cowl, the recirculation region the cowl. This edge causes flow separation, and a
is reduced somewhat. However, the high-velocity fluid region of recirculation grows. The velocity magnitude in
still remains close to the cowl wall and does not diffuse. the cowl is on average higher with an energized
The small scoop does decelerate the flow somewhat, as boundary layer.
shown in Figure 10(a), but the relative magnitude of the The flow mechanisms most damaging to cowl
incoming air velocity prevents this deceleration from efficiency are not due to the cowl geometry, but the cowl
becoming significant. The challenge is now centered on placement and location. The contours of total pressure
the transfer of kinetic energy of the incoming air stream along the cowl centerline plane shown in Figure 13(a)
into high pressure right over the carburetor. illustrate the thick boundary layer that forms over the
The benefit of capturing more high-energy flow in length of the hood. This boundary layer separates the
the cowl is that the total pressure of the air entering the highest-energy flow from the cowl inlet by almost two
cowl is higher than in the REF case. However, the high- inches and allows only a small amount of high-energy
energy, high-velocity flow is not well diffused, so most of flow to enter the cowl. Thus, the highest losses in the
it remains close to the cowl wall. Complex secondary Scoop 2 (SC2) configuration are in the boundary layer at
flow structures are enhanced due to the higher flow the cowl inlet. Furthermore, the stagnation region
rates entering the cowl compared with REF case shown shown by contours of static pressure in Figure 13(b) is
in Figure 9. downstream of the cowl inlet at the base of the
windshield. The highest pressure then is not at the cowl
Scoop 2 Results: Standard vs. RBL Case inlet but downstream of it. Consequently, the pressure
differential through the cowl is reduced.
The computational model, a modified scoop
Figure 15(a) shows the increase in high total
geometry (SC2), is an exact replica of that used in
pressure air that enters the cowl due to an energized
modeling the reference geometry, the standard Roush
boundary layer. Outlet flow is increased drastically by
design, with only two changes. An arbitrary scoop was
this modification. This case shows that the geometry will
added to the rear cowl wall in an attempt to better diffuse
need to be refined much more before effective use of the
the flow. In addition, the front opening was rounded
flow increase is made. A better geometry will most
(front scoop) to give the flow a smoother transition into
probably lead to higher flow rates throughout the engine.
the cowl. Boundary conditions for these two simulations
Figure 15(b) shows that the stagnation region remains
are the same as for the REF case and the REF case
downstream of the cowl inlet. Relocation of the cowl
with an energized boundary layer, respectively.
inlet closer to the base of the windshield would most
The velocity contours and velocity vectors along the
likely result in flow increases as well. Similarly, elevating
cowl centerline plane in Figure 12 highlight the nature of
the surface at the base of the windshield, while at the
the flow entering the SC2 configuration. The high-speed
same time lowering the hood at the cowl inlet, will be
flow is more able to navigate the new turn into the cowl;
very helpful regarding all aspects of what has been
thus, eliminating the large recirculation region. Only a
highlighted.
relatively small region of recirculation is located at the
Figure 14 also shows that the recirculation region
roof of the cowl. The small amount of high-speed flow
has reformed. This region and sidewall effects account
that enters the cowl is still not diffused as efficiently as
for much of the blockage at the inlet. The recirculation
possible, so it remains close to the cowl wall. The larger
region also prevents the inlet geometry from efficiently
scoop region at the base of the cowl decelerates the
diffusing the high-speed flow, so it remains at high
flow nearly isentropically, but the flow must accelerate
speed and still close to the back wall of the cowl.
again as it moves over the large hump. Because of the
Sidewall effects in both cases are much less significant
corner shaping at the cowl outlet, the flow must
in reducing outlet flow rate than the other mechanisms
accelerate as it turns into the carburetor, resulting in a
mentioned above.
separation region at the carburetor inlet. These features
Sidewall and pressure effects combine to create a
are not at all desirable as they induce losses, increased
secondary flow that moves towards the cowl outlet. This
resistance to flow, and swirl into the carburetor. Overall
secondary flow combines with the pressure differentials
the geometry modification alone increased airflow by
shown in Figures 13(b) and 15(b) to create the large,
only 2.4 %.
counter-rotating vortex structure on the cowl outlet plane
6
compared to REF case shown in Figure 9. Such a boundary layer that forms over the length of the hood is
swirling flow entering the carburetor will result in an removed allowing more high-energy air to enter the
inefficient performance and non-equal charging of the cowl. The result is a significantly higher flow rate at the
cylinders. The vortex structures increase in the exit plane of the cowl. Quantitatively, this increase
energized boundary layer case due to the much higher corresponds to about 150.2 % improvement (from 577.2
flow rate. to 1444.3 CFM) over the standard Roush design. The
stagnation region shown by contours of static pressure
REFERENCE CASE WITH BOUNDARY LAYER in Figure 18(b) is within the cowl inlet for the first time.
SUCTION This is a very desirable outcome; such a high pressure
increases the driving potential for the flow through the air
The reference case was modified by opening a box. The flow in the cowl stagnates against the back
portion of the front surface at the cowl inlet and applying wall, forcing the air back towards the cowl outlet.
a suction pressure of –1 PSI. This technique, called However, it is possible to have a clever geometry
boundary layer suction (BLS), has the effect of removing modification in the future that could take advantage of
the boundary layer and helping the high-speed flow the existing flow patterns and guide the low-speed air
navigate the turn into the air box. Removal of the toward the cowl exit plane. The remaining mechanisms
boundary layer also allows more high-energy air to enter contributing to losses are found at the lower cowl wall.
the air box. These two effects eliminate the inlet dead Flow stagnates in the small scoop at the base of the
region and increases the effective flow area. The result cowl. This flow then has to accelerate up the hump and
is about 150.2 % increase in exit flow rate. turn at the same time. The contours clearly indicate
Certain faces on the front wall of the cowl opening unnecessary losses in this region.
were rezoned into a newly created pressure-outlet. This As shown in Figures 18 (a, b), the effective inlet area
pressure outlet was set to –1 PSI after reviewing 2D full for high-speed, high-energy flow is much larger than
car simulations. So that this technique could be seen in all the previous cases. For the first time, the air
implemented with no additional mechanical hardware a flowing over the hood and approaching the cowl inlet
location of very low gage pressure was sought. It was area actually “sees” almost as much opening as
determined that this pressure could be achieved if piping suggested by the surrounding metal. Sidewall effects
was ran to just behind the chin spoiler at the nose of the account for about 5% of the blockage. This blockage,
car. The pressure sink found here was -1.5 PSI. Such which has not been addressed thus far, is now the most
an installation has a built-in control system. The sink significant mechanism creating flow blockage. Figure
pressure develops as the speed of the race car 19(c) shows that high-energy flow remains present to
increases. This is precisely the type of variation needed this point. All three plots in Figure 19 show clearly that
for this devise to be effective. Figure 16 shows an overall performance of the intake cowl is much improved
isometric cutaway view of the exact replica of the Roush over the standard Roush design REF. Moreover, the
geometry used in the simulation, along with the location specific features highlighted in this figure indicate that
of the boundary layer suction surface. boundary layer suction may prove to be an especially
The velocity contours and velocity vectors along the potent technique when used in conjunction with
cowl centerline plane in Figure 17 highlight the nature of previously documented methods, such as the RBL case.
the flow entering the Roush configuration with suction As shown in Figure 20, sidewall and pressure
(REF BLS). The high-speed flow slows by nearly 100 influences combine effectively to create a secondary
mph and turns sharply with the aid of the suction flow that moves towards the cowl outlet (as desired). It
surface; thus, the large recirculation region found in is significant to note that the total pressure contours in
previous simulations that reduced the effective area to this figure show the energy of the flow is only slightly
about 20 % of the actual inlet area is eliminated. A less than that of the free stream. This secondary flow
region of very low velocity flow arises just below the roof combines with the pressure differentials shown in Figure
of the air box. This level of suction cannot turn the flow 18(b) to create the flow structure found on the cowl
a full 180 degrees. The high-speed flow entering the outlet plane shown in Figure 21. The exit flow pattern is
cowl is not diffused efficiently, so it stays close to the more uniform than those seen in prior simulations. This
cowl wall. This is where one of the previously should reduce cylinder-to-cylinder power variations. A
documented scoop configurations may prove to be uniform flow moving vertically down into the carburetor
effective in diffusing the flow. The airflow first stagnates with a minimum swirling motion is highly desirable.
in the small scoop region at the base of the cowl, and
then accelerates to move over the hump as it enters into SC3 WITH BLS AND RBL
the carburetor. Because of the corner shaping at the
cowl outlet, the flow must accelerate as it turns into the The reference case was modified by adding an
carburetor, resulting in a separation region at the arbitrary scoop where the original back wall was located
carburetor inlet. However, the increase in volumetric and opening a portion of the front surface at the cowl
flow rate is significant enough not to worry about this inlet as described above. A suction pressure of –1.0 PSI
detail. was applied as in the previous boundary layer suction
The contours of total pressure along the cowl case. Additionally the boundary layer was energized.
centerline plane shown in Figure 19(a) illustrate that the This presents a combination of all the techniques used
7
in the current study to increase airflow to the engine. up the hump and turn at the same time. The contours
The results are surprising. Engine airflow increased to clearly indicate unnecessary losses in this region.
1306 CFM, an increase of over 126%, but still less than Figure 24 shows the flow features at and below the
that of the REF geometry with BLS. cowl inlet plane. The combination of suction and an
The same back-pressure is carried over to the energized boundary layer seems not to pull as much air
suction case with energized boundary layer. The through as suction alone. Instead the combined
suction case was developed in the same manner as techniques seem to turn very little air sharply. The result
above for the REF case with boundary layer suction. is a low sink pressure that slows a great volume of air
The most important flow feature that arose from this that is pulled into the cowl. Sidewall effects account for
simulation is the separation region above the car’s about 5% of the blockage. This blockage, which has not
roofline. This brand new phenomenon that appeared in been addressed thus far, is now the most significant
this simulation has never been reported through mechanism creating flow blockage. Figure 24(c) shows
computations or experiments in the open literature. It that moderate-energy flow remains present to this point.
can be seen in Figures 22 and 230 as blue velocity All three parts in Figure 24 clearly show that overall
magnitude and total pressure contours. This couples performance of the intake cowl is much improved over
engine intake air with aerodynamics, and is possibly the that of the standard Roush (REF) design.
reason that some engines perform well on the As shown in Figure 25, sidewall and pressure
dynamometer but not on the race track. This region influences combine effectively to create a secondary
arises when the suction surface slows the free-stream flow that moves towards the cowl outlet (as desired). It
by nearly 100 mph as seen in Figure 22(a). The flow is significant to note that the total pressure contours in
then accelerates along the windshield. The flow cannot this figure show the energy of the flow is much less than
navigate the sharp turn onto the roof due to its lack of that of the free stream. The Roush case with suction
momentum. The result is separated flow. A showed more desirable contours. This secondary flow
recirculation region appears along the roof. This region combines with the pressure differentials shown in Figure
is characterized by very low pressure and velocity. In an 23(b) to create the flow structure found on the cowl
aerodynamic sense down force is reduced and drag is outlet plane. The exit flow pattern is more uniform than
increased. those seen in prior simulations except for the Roush
Less importantly, the velocity contours and velocity design with suction.
vectors along the cowl centerline plane in Figure 22
highlight the nature of the flow entering the Scoop 3 SC3 WITH BLS
configuration with suction and energized boundary layer.
The high-speed flow turns sharply with the aid of the Computational model used here, as shown in Figure
suction surface; thus, a large amount of air enters the 48, is an isometric cutaway view of the exact replica of
engine. Low velocity flow arises in most of the air box. the Scoop 3 (SC3) geometry used in the simulation,
Suction cannot turn the flow a full 180 degrees, leading along with the location of the boundary layer suction
to areas of near-zero velocity. In turn, this leads to high surface. A suction pressure of –1.0 PSI was applied as
static pressures in the cowl, as seen in Figure 23(b). It in the previous boundary layer suction case. One last
can also be seen that the airflow first stagnates in the case presented here in order to increase engine airflow
scoop region at the base of the cowl, and then while trying to avoid roofline separation. Engine airflow
accelerates to move over the hump as it enters into the increased to 1857 CFM, an increase of over 222 %, but
carburetor. Because of the corner shaping at the cowl roofline separation still occurs. So, once again, it is not
outlet, the flow must accelerate as it turns into the recommended that this geometry be used with BLS due
carburetor, resulting in a separation region at the to the aerodynamic effects seen downstream of the cowl
carburetor inlet. This effectively shrinks the opening. opening.
The contours of total pressure along the cowl Obviously, airflow cannot continually increase
centerline plane shown in Figure 23(a) illustrate that the through an engine. There is a theoretical maximum
boundary layer that forms over the length of the hood is amount of airflow that can be obtained. For this
removed allowing more high-energy air to enter the particular engine with 358 cubic inches of displacement,
cowl. The result is a significantly higher flow rate at the the maximum flow rate was found to be 1657.4 CFM
exit plane of the cowl. Quantitatively, this increase when the engine is rotating at 8000 rpm. When possible
corresponds to about 126 % improvement over the it would be advantageous to design the entire intake
standard Roush design (REF). The stagnation region path toward this maximum flow rate.
shown by contours of static pressure in Figure 23(b) is In light of these details, the physics of the cowl flow
outside the cowl inlet as seen in all previous cases with are discussed since this is the highest flow rate thus far.
the exception of the Roush case with suction (REF The velocity contours and velocity vectors along the
BLS). The flow in the cowl stagnates against the back cowl centerline plane in Figure 26 highlight the nature of
wall, forcing the air back towards the cowl outlet. The the flow entering the Scoop 3 configuration with suction.
remaining mechanisms contributing to losses are found The high-speed flow turns sharply with the aid of the
at the lower cowl wall. Flow stagnates in the scoop at suction surface; thus, a large amount of air enters the
the base of the cowl. This flow then has to accelerate engine. Low velocity flow arises along the scoop wall
and just above the cowl outlet. This level and location of
8
suction cannot turn the flow a full 180 degrees, leading prevents the highest-energy flow from entering the cowl;
to areas of near-zero velocity. In turn, this leads to high The free stream flow stagnation point is at the base of
static pressures in the cowl, as seen in Figure 27(b). the windshield, past the cowl inlet plane, and results in a
The location of this stagnation region is highly desirable reduced driving potential due to lower pressure
as it acts to pump flow out the exit. It can also be seen differential between the cowl inlet and outlet; Because
that the airflow first stagnates in the scoop region at the the high-speed flow cannot navigate the sharp turning
base of the cowl, and then accelerates to move over the radius into the cowl, a large separation region is formed
hump as it enters into the carburetor causing losses in that both reduces the effective inlet area by 80% and
total pressure. Due to the corner shaping at the cowl pumps flow in the wrong direction away from the outlet
outlet, the flow must accelerate as it turns into the (toward the cowl inlet); The Roush geometry is a poor
carburetor, resulting in a separation region at the diffuser and fails to slow the high-speed flow entering
carburetor inlet. This effectively shrinks the opening, the cowl except in the small scoop region, where the
however, this would be of no consequence if the above- flow is decelerated nearly isentropically; Sidewall and
mentioned flow rate was introduced to the engine pressure effects result in large secondary flow vortices
without the separation region above the roofline. on the cowl outlet plane.
The contours of total pressure along the cowl Two new geometry changes as SC2 and SC3
centerline plane shown in Figure 27(a) illustrate that the combined with RBL and BLS techniques showed variety
boundary layer that forms over the length of the hood is improvements over REF case. By reducing the
removed allowing more high-energy air to enter the boundary layer thickness along the hood of the car, the
cowl. The result is the highest flow rate at the exit plane RBL case produced a 28.9 % higher CFM than the REF
of the cowl attained in this study, which is about 222 % case. Boundary layer suction allowed more high-energy
improvement over the standard Roush design (REF). air to enter the air box and resulted about 150.2 %
The stagnation region shown by contours of static increase in exit flow rate than REF case. Both
pressure in Figure 27(b) is outside the cowl inlet as seen techniques can be used to improve cowl performance.
in all previous cases with the exception of the Roush Geometry modification SC2 alone is not able to add
case with suction (REF BLS). The flow in the cowl more airflow (only 2.4% than REF case), however, SC2
stagnates against the back wall, forcing the air back combined with RBL generated about 37.7% increase
towards the cowl outlet. The remaining mechanisms compared to REF case. Geometry modification SC3
contributing to losses are found at the lower cowl wall. with BLS and RBL increased about 126% in flow rate
Flow stagnates in the scoop at the base of the cowl. over the REF case, even better, SC3 with BLS only add
This flow then has to accelerate up the hump and turn at 222% improvement in flow rate over REF case.
the same time. The contours clearly indicate Although flow rates have been increased greatly in
unnecessary losses in this region. two SC3 cases, however, this case and others, slowed
Figure 28 shows the flow features at and below the air so much at the base of the windshield that the flow
cowl inlet plane. The boundary layer suction surface was unable to navigate the roofline. The flow then
alone with this geometry provides the most high-energy, separated created a recirculation region characterized
high-velocity air for the engine. This technique seems to by low-velocity, low-pressure air. This would in turn
turn the flow and slow a great volume of air that is pulled reduce the down force and increase the drag
into the cowl. Sidewall effects account for about 5 % of experienced by the race car. A car with large amounts
the blockage. This blockage, which has not been of horsepower and torque, but no down force and
addressed thus far, is now the most significant increased drag will not be effective during a race. To the
mechanism creating flow blockage. Figure 28(c) shows authors’ knowledge this is the first documented account
that high-energy flow remains present to this point and of this new phenomenon coupling external
diffused throughout the cross-sectional area. All three aerodynamics and internal engine airflow. This also
plots in Figure 28 clearly show that overall performance offers an explanation for certain engines performing well
of the intake cowl is the best of all studied. on the dynamometer, but not in race conditions.
Sidewall and pressure influences combine Excluding cases that demonstrated this new
effectively to create a secondary flow that moves phenomenon, the flow rate was increase by 150 % in the
towards the cowl outlet (as desired). This secondary Roush REF case with BLS, an impressive amount. The
flow combines with the pressure differentials shown in research sponsor reported a 0.238 second reduction in
Figure 27(b) to create the flow structure found on the lap time using some techniques outlined in this study.
cowl outlet plane shown in Figure 29. The exit flow When this is multiplied by the number of laps driven on a
pattern is more uniform than those seen in prior super speedway (usually 250), this equates to an
simulations with the exception of the Roush (REF) impressive advantage over the competition.
design with suction.
ACKNOWLEDGMENTS
CONCLUSIONS
The authors would like to thank Team Amick
The important flow characteristics were identified MotorSports for funding this study, and Fluent, Inc.
from the computational results of REF case simulation: personnel for continued support, with special thanks to
the thick boundary layer approaching the cowl inlet Dr. Lounsbury and Dr. Koutsavdis. Thanks as well to
9
Will Robinson for his excellent support of computing 11. Williams, J., Quinlan, W. J., Hackett, J. E.,
systems. Thomson, S. A., Marinaccio, and T., Robertson, A.,
1994, “A Calibration Study of CFD Automotive
REFERENCES Shapes and CD,” SAE Paper 940323, International
Congress & Exposition, Feb. 1994, Detroit, MI, pp.
1. Laise, T. D., and Bayless, K. S., 1994, “Aerodynamic 308-327.
Development of a Successful NASCAR Winston 12. Okumura, K., and Kuriyama, T., 1995, “Practical
Cup Race Car,” SAE Paper 942521, International Aerodynamic Simulations (CD, CL, CYM) Using a
Congress & Exposition, Feb. 1994, Detroit, MI, pp. Turbulence Model and 3rd-Order Upwind Scheme,”
275-284. SAE Paper 950629, International Congress &
2. Emmelmann, H. J., Berneburg, H., and Schulze, J., Exposition, Feb. 1995, Detroit, MI, pp. 1149-1161.
1990, “the Aerodynamic development of the Opel 13. Hajiloo, A., Williams, J., Hackett, J. E., and
Calibra,” SAE Paper 900317, International Congress Thomson, S. A., 1996, “Limited Mesh Refinement
& Exposition, Feb. 1990, Detroit, MI, pp. 472-483. Study of the Aerodynamic Flow Field Around a Car-
3. Woodbridge, D. M., and Miller, R. B., 1996, “The Like Shape, Computational vs. Experimental Fluid
Aerodynamic Optimization of a Successful IMSA GT Dynamics,” SAE Paper 960321, International
Race Car,” SAE Paper 962518, International Congress & Exposition, Feb. 1996, Detroit, MI.
Congress & Exposition, Feb. 1996, Detroit, MI, pp. 14. Gaylard, A. P., Baxendale, A. J., and Howell, J. P.,
107-114. 1998, “the Use of CFD to Predict the Aerodynamic
4. Katz, J., and Dykstra, L., 1994, “Application of Characteristics of Simple Automotive Shapes,” SAE
Computational Methods to the Aerodynamic Paper 980036, International Congress & Exposition,
Development of a Prototype Race Car,” SAE Paper Feb. 1998, Detroit, MI, pp. 97-106.
942498, International Congress & Exposition, Feb. 15. Axelsson, N., Ramnefors, M., and Gustafsson, R.,
1994, Detroit, MI, pp. 161-169. 1998, “Accuracy in Computational Aerodynamics
5. Werner, F., Frik, S., and Schulze, J., 1998, Part 1: Stagnation Pressure,” SAE Paper 980037,
“Aerodynamic Optimization of the Opel Calibra ITC International Congress & Exposition, Feb. 1998,
Racing Using Experiments and Computational Fluid Detroit, MI, pp. 107-118.
Dynamics,” SAE Paper 980040, International 16. Perzon, S. Sjogren, T., and Jonson, A., 1998,
Congress & Exposition, Feb. 1998, Detroit, MI, pp. “Accuracy in Computational Aerodynamics Part 2:
149-158. Base Pressure,” SAE Paper 980038, International
6. Williams, J., and Vemaganti, G., 1998, “CFD Quality Congress & Exposition, Feb. 1998, Detroit, MI, pp.
– A Calibration Study for Front-End Cooling Airflow,” 119-131.
SAE Paper 980039, International Congress & 17. Perzon, S., Janson, J., and Hoglin, L., 1999, “On
Exposition, Feb. 1998, Detroit, MI, pp. 133-147. Comparisons Between CFD Methods and Wind
7. Andra, R., Hytopoulos, E., Kumar, K., and Sun, R. Tunnel Tests on a Bluff Body,” SAE Paper 1999-01-
L., 1998, “The Effect of Boundary and Geometry 0805, International Congress & Exposition, Mar.
Simplification on the Numerical Simulation of Front- 1999, Detroit, MI, pp. 133-148.
End Cooling,” SAE Paper 980395, International 18. Makowski, F. T., and Kim, S., 2000, “Advances in
Congress & Exposition, Feb. 1998, Detroit, MI, pp. External-Aero Simulation of Ground Vehicles Using
213-225. the Steady RANS Equations,” SAE Paper 2000-01-
8. Gillieron, P., Samuel, S., and Chometon, F., 1999, 0484, International Congress & Exposition, Mar.
“Potential of CFD in Analysis Under-Bonnet Airflow 2000, Detroit, MI, pp. 79-90.
Phenomena,” SAE Paper 1999-01-0802, 19. Martin, W., and Gillieron, P., 2001, “Integration of
International Congress & Exposition, Mar. 1999, Fluid Flow Modeling in the Vehicle Renault
Detroit, MI, pp. 99-111. Development Process,” SAE Paper 2001-01-0700,
9. Glober, S., Klimetzek, F., and Wessels, M., 1991, International Congress & Exposition, Mar. 2001,
“Simulation of Aero- and thermodynamic Problems Detroit, MI, pp. 133-140.
in Automotive Applications,” Presented at the Fourth 20. Basara, B., Przulj, V., and Tibaut, P., 2001, “On the
FIDAP Users Conference, April 14-16, 1991, pp. 1- Calculation of External Aerodynamics: Industrial
17. Benchmarks,” SAE Paper 2001-01-0701,
10. Williams, J., and Hajiloo, A. 1994, “Aerodynamic International Congress & Exposition, Mar. 2001,
Flow Field Around Two Car-Like Shapes, Detroit, MI, pp. 141-153.
Computational vs. Experimental Fluid Dynamics,”
Presented at the Seventh FIDAP Users Conference,
1994.
10
Far-field Open Boundary
Inlet
Outlet
Relative Motion
of the Ground
Figure 1. Two-dimensional domain (center plane) used to generate boundary conditions for 3D model.
Far-field Open Boundary
Vehicle Hood
Vehicle Windshield
Cowl Opening
Cowl Air Box

Cowl Outlet
Outlet
Inlet
Figure 2. Three-dimensional computational domain (with foreground symmetry plane removed).
11
Triangle Cells
Stair-stepping using quads
Zoom Area
Figure 3. Two-dimensional sample grid along symmetry plane (REF).
Cowl Inlet
(a) REF Cowl Exit (b) SC2
(c) SC3
Figure 4. Cut-away of the three geometries used in simulations: (a) REF, (b)SC2, (c)SC3.
12
(a) Velocity Magnitude

Zoom Area
External Flow
Direction Stagnation region is at the
Viewing Plane: base of the windshield, behind
Centerline the cowl inlet
Velocity High speed flow does not

diffuse well as the cowl volume
200 increases
To carburetor Flow must accelerate to Small scoop in airbox isentropically

160 turn into the carburetor decelerates the flow
High velocity flow cannot
navigate sharp turn into cowl
(b) Velocity Vectors
120
mph Additional recirculation
Flow Direction
region
80 Large recirculation region

forms large blockage at the inlet
40
Effective area for the flow
entering the cowl is ~25% of
0 the available area
Flow in recirculation region

actually pumps through the
cowl backwards!
Figure 5. The complex nature of the flow along the centerline plane of the Roush configuration is shown by (a) contours of
velocity magnitude and (b) velocity vectors colored by velocity magnitude (REF).
Flow Direction
Thick boundary layers form

along the hood, separating high-
energy flow from cowl inlet
Viewing Plane:
Centerline
Only a small amount of high-
Total Static energy flow enters cowl
0.74 0.50
High-energy flow eventually
dissipates, allowing low-energy
(a) Total Pressure flow to enter the carburetor
0.62 0.40
0.50 0.30 Stagnation region is behind the

PSI PSI White area cowl inlet, decreasing the driving
clipped to scale potential through the cowl
0.38 0.20
Flow stagnates against
the forward wall Small scoop area acts as a
0.26 0.10 stagnation region
0.14 0.00
Flow accelerating into the carburetor

(b) Static Pressure causes a localized low-pressure
region at the cowl exit
Figure 6. Contours of (a) total pressure and (b) static pressure along the cowl centerline highlight the major physical
characteristics of the flow in the critical regions of the cowl (REF).
13
Velocity (mph) 200 150 100 50 0

Recirculation region reduces
Total Press. (PSI) 0.74 0.59 0.44 0.29 0.14 effective inlet area by about 80%
Front of Car
Rear of Car
(a) Velocity Magnitude at Inlet
Side-wall effects reduce the
effective inlet area by about 5%
Viewing Plane:
Inlet
(b) Total Pressure at Inlet

Inlet geometry cannot diffuse the
high-energy, high-velocity flow
Viewing Plane: (c) Total Pressure Below Inlet

1.25” Below Inlet
Cowl Centerline
Figure 7. Contours of (a) velocity magnitude at the inlet, (b) total pressure at the inlet, and (c) total pressure 1.25” below
the inlet show the reduction in effective inlet area of the cowl (REF).
High-energy flow does not

reach the carburetor
Cowl Centerline
Viewing Plane: Total Pressure

Vertical Crossplane at
Outlet Center (a) Total Pressure at 0.74
Vertical Crossplane
Secondary flow moving
toward outlet due to a 0.62
combination of wall effects
Higher-energy flow does
and pressure gradients
not diffuse and remains 0.50
close to the cowl wall
PSI
0.38
Outlet
0.26
Viewing Plane:
0.14
Horizontal Crossplane at
1” Above Cowl Bottom
(b) Total Pressure at

Recirculation region Horizontal Crossplane
Figure 8. Contours of total pressure on (a) a vertical crossplane and (b) a horizontal crossplane show the low energy of the
flow moving towards the carburetor (REF).
14
Front of Car
Viewing Plane:
Outlet
Vortex structures increase cylinder-

to-cylinder flow variations
Rear of Car
Figure 9. Velocity vectors on the outlet plane show the complex vortex structure entering the carburetor (REF).
(a) Velocity Magnitude Zoom Area
The thinner boundary layer

allows the high-speed, high-
Flow Direction energy flow to stagnate closer
to the cowl entrance instead of
Viewing Plane: at the base of the windshield
Centerline
Since it is closer to the cowl
Velocity entrance, more high energy flow is
captured by the cowl, but the
225
geometry is a poor diffuser
Flow must accelerate to Because of the high velocity, the
180 turn into the carburetor small scoop does decelerate the
flow significantly
High velocity flow cannot (b) Velocity Vectors
135 navigate sharp turn into cowl
mph Flow Direction
90 Because a thin boundary layer

means the high-speed flow is
closer to the wall surface,
more high-speed, high energy
45
flow is captured by the cowl
0
Large recirculation region
forms large blockage at the
inlet, but its size is smaller than
in the REF case
Figure 10. The complex nature of the flow along the centerline plane of the RBL Roush configuration is shown
by (a) contours of velocity magnitude and (b) velocity vectors colored by velocity magnitude.
15
The simulation is modified to Flow Direction The thin boundary layer

allows the high-speed, high-
demonstrate the effects of a
thin boundary layer on cowl energy flow to remain close
to the surface
performance
Because the high-energy
Viewing Plane: flow is closer to the cowl
Centerline inlet, more is able to enter
the cowl
Total Static
0.75 0.75 The high energy flow diffuses

More high-energy flow somewhat, especially near the
enters the carburetor scoop
(a) Total Pressure
0.63 0.60
The stagnation region is

0.51 0.45 closer to the cowl inlet
White area because the high-energy
PSI PSI clipped to scale
flow is closer to the surface
and inlet
0.39 0.30
the forward wall
Small scoop area acts as a
0.15 0.00

Figure 11. Contours of (a) total pressure and (b) static pressure along the cowl centerline for the RBL Roush case
highlight the major effects of a thin boundary layer on the hood of the car.

Zoom Area
Flow Direction Stagnation region is still at the

Velocity High speed flow diffuses better

as the cowl volume increases in
200 comparison to the reference
geometry
Flow must accelerate to Scoop in air box better diffuses the
160 turn into the carburetor flow

120
mph High velocity flowFlow
can Direction
better
navigate turn into cowl
80
A smaller recirculation region
forms allowing larger effective
area
40
entering the cowl is much
0 greater
Figure 12. The complex nature of the flow along the centerline plane of the scoop configuration (SC2) is shown by (a)
contours of velocity magnitude and (b) velocity vectors colored by velocity magnitude.
16
Flow Direction
Thick boundary layers form over

the hood, separating high-energy
flow from cowl inlet
Viewing Plane:
Centerline
Only a small amount of high-
Total Static energy flow enters cowl
0.74 0.50 High-energy flow eventually

dissipates, allowing low-energy
flow to enter the carburetor
0.62 0.40 (a) Total Pressure
0.50 0.30 Stagnation region is still behind

White area
PSI PSI clipped to scale the cowl inlet, decreasing the
driving potential through the
0.38 0.20 cowl
the forward wall Larger scoop area reduces
0.26 0.10
stagnation region
0.14 0.00

characteristics of the flow in the critical regions of the cowl (SC2).

Zoom Area
Flow Direction Stagnation region is still at the

Velocity With boundary layer tripped,

flow velocity needs to be
200 diffused more effectively
Flow must accelerate to Scoop design does not decelerate

160 turn into the carburetor flow with tripped boundary layer

120
mph With tripped boundary
Flow Direction
layer,
more high velocity flow
reaches the cowl
80
Large recirculation region
reappears blocking the inlet
40
entering the cowl is once again
0 ~25% of the available area
Flow in recirculation region

pumps upstream toward the
cowl inlet plane again
Figure 14. The complex nature of the flow along the centerline plane of the scoop configuration with tripped boundary
layer is shown by (a) contours of velocity magnitude and (b) velocity vectors colored by velocity magnitude (SC2 RBL).
17
Flow Direction
Boundary layers are much

thinner allowing high-energy
flow to enter cowl inlet
Viewing Plane:
Centerline
More high-energy flow enters
Total Static cowl
0.74 0.50
Overall the air in the airbox has
higher energy than the case
without a tripped boundary layer
0.62 0.40 (a) Total Pressure
0.50 0.30 Stagnation region is behind the

PSI PSI cowl inlet, decreasing the driving
potential through the cowl
0.38 0.20
Flow stagnates against Stagnation region reappears with
the forward wall increased flow
0.26 0.10
Flow stagnates at base

0.14 0.00 due to sharp curvature
(b) Static Pressure Flow accelerating into the carburetor

causes a localized low-pressure
Figure 15. Contours of (a) total pressure and (b) static pressure along the cowl centerline highlight the major
physical characteristics of the tripped boundary layer flow in the critical regions of the cowl(SC2 RBL).
Note: Because of the complex nature of the 3D model, each

figure will indicate the viewing plane in relation to the geometry
Cowl Inlet
Viewing Plane
Standard Roush
Aircleaner Housing
Suction Surface
Cowl Outlet/Carburetor
Inlet
Figure 16. Cutaway of the exact replica of the Roush cowl configuration shows the interior geometry. A strategically placed
suction surface at -1 PSIG pulls the boundary layer off allowing higher energy air to enter the intake cowl (REF BLS).
18

Zoom area
shown below
Flow Direction Boundary layer suction moves

Viewing Plane: stagnation point into air box
Centerline
Velocity Flow cannot diffuse properly due to

improper geometry
200
Effective flow area is increased
Flow must accelerate to because recirculation region is
160 turn into the carburetor removed
Boundary layer suction helps

120 (b) Velocity Vectors
more high-energy flowDirection
Flow make
mph the sharp turn into air box
80
Large recirculation region seen
in other cases is eliminated
40 allowing more flow through
Dead region at top of air box

does not aid in flow
Figure 17. The extremely complex nature of the flow along the centerline plane of the Roush configuration with suction is
shown by (a) contours of velocity magnitude and (b) velocity vectors colored by velocity magnitude (REF BLS).
Flow Direction
Suction removes boundary layer

that has grown on hood allowing
high-energy air to enter the cowl
Viewing Plane:
Centerline
High-energy flow dissipates, still
Total Static allowing for improved designs
0.90 0.90
(a) Total Pressure Flow stagnates on front

0.62 0.62 walls of air box
Stagnation region is in the cowl

0.50 0.50
inlet for the first time, increasing
PSI PSI the driving potential through the
cowl
0.38 0.38
Small scoop area acts as a

0.00 0.00

Highest stagnation pressure above region at the cowl exit
cowl outlet observed thus far
characteristics of the flow in the critical regions of the cowl. Large amount of high energy flow enters the cowl (REF BLS).
19
Velocity (mph) 200 150 100 50 0

Recirculation region is
Total Press. (PSI) 0.9 0.59 0.44 0.29 0.0 eliminated allowing flow to
make use of entire opening
Front of Car
(a) Velocity Magnitude at Inlet Rear of Car
Side-wall effects reduce the

effective inlet area by about 5 %
Suction surface
Viewing Plane:
Inlet

Total pressure remains high to
this point in the geometry aiding
in higher exit flow rate
Viewing Plane: (c) Total Pressure Below Inlet

1.25” Below Inlet
Cowl Centerline
the cowl inlet show much improved performance compared to the standard Roush design (REF BLS).
More high-energy flow

reaches the carburetor
Cowl Centerline

Vertical Crossplane
combination of wall effects
and pressure gradients
0.50
PSI
0.38
Outlet
0.26
Viewing Plane:
0.00
Horizontal Crossplane at

Horizontal Crossplane
Figure 20. Contours of total pressure on a (a) vertical crossplane and (b) horizontal crossplane show the higher energy flow
moving towards the carburetor (REF BLS).
20
Front of Car
Viewing Plane:
Outlet
Flow patterns are more uniform with

minimum swirling motion reducing
cylinder-to-cylinder power variations
Rear of Car
Figure 21. Velocity vectors on the outlet plane show relatively uniform flow with minimum amount of undesirable
swirling motion entering the carburetor (REF BLS).
(a) Velocity Magnitude Zoom area

shown below
Flow Direction
Viewing Plane: Low-momentum fluid cannot
Centerline navigate roof line; separation
occurs
Velocity
Non-conformal Interface (NCI)
200
Combined suction and
energized boundary Flow accelerates as it travels up the
160 layer slows air velocity windshield
by more than 100 mph
120 Low-velocity fluid fills the (b) Velocity Vectors

Flow Direction
mph airbox leading to a high static
pressure as seen in Figure 3
80
Suction Surface
40
0 Large recirculation region

arises on roof of car increasing
drag
Point of Separation Non-conformal Interface (NCI)
Figure 22. The extremely complex nature of the flow along the centerline plane of the SC3 configuration with suction and energized
boundary layer is shown by (a) contours of velocity magnitude and (b) velocity vectors colored by velocity magnitude (SC3 BLS RBL).
21
Flow Direction
Flow separates causing
large losses in total
pressure above the roof
Viewing Plane:
Centerline
Total Static NCI
1.06 0.770 Moderate amounts of total pressure

remain in the cowl.
0.848 0.616
(a) Total Pressure
Separation causes a large
0.636 0.462 area of low-pressure air
above roof reducing
PSI PSI
down force
0.424 0.308
0.212 0.154
High static pressure regions are

0.000 0.000 caused by stagnated flow

causes a localized low-pressure
(b) Static Pressure
characteristics of the flow in the critical regions. Separated flow is clearly seen (SC3 BLS RBL).
Velocity (mph) 200 150 100 50 0
Total Press. (PSI) 1.06 0.795 0.530 0.265 0.000
Front of Car
Suction surface causing Side-wall effects reduce the

high-velocity flow effective inlet area by about 5 %
Viewing Plane:
Inlet
(b) Total Pressure at Inlet Total pressure remains relatively

constant to this point in the geometry
aiding in higher exit flow rate
(c) Total Pressure Below Inlet

Viewing Plane:
1.25” Below Inlet
Cowl Centerline
the cowl inlet show much improved performance compared to the standard Roush design (SC3 BLS RBL).
22
Flow is slowed due to high

level of suction and energized
Cowl Centerline boundary layer

Vertical Crossplane
combination of wall effects Very little total pressure
and pressure gradients drop is seen through cowl
0.636
PSI
0.424
Outlet
0.212
Viewing Plane:
Horizontal Crossplane at 0.000

Horizontal Crossplane
Figure 25. Contours of total pressure on a (a) vertical crossplane and (b) horizontal crossplane show the higher energy
flow moving towards the carburetor (SC3 BLS RBL). NOTE: Not to scale.
Flow is slowed by
nearly 100 mph
Flow Direction
Viewing Plane: Zoom area
Centerline shown below
Velocity Boundary Layer is sucked off

hood allowing high energy air
200 to enter cowl
Flow must accelerate
across bends along cowl
160 bottom
Flow accelerates and separates
as it moves into carburetor
120
Stagnation Region
mph (b) Velocity Vectors
80
Two smaller recirculation regions form
further reducing effective flow area
40
Large recirculation region is

0 eliminated allowing for nearly
all effective area to be used
Flow does not need such a large

scoop surface
Figure 26. The extremely complex nature of the flow along the centerline plane of the Scoop 3 (SC3) configuration with
BLS at -1 PSIG is shown by (a) contours of velocity magnitude and (b) velocity vectors colored by velocity magnitude.
23
Flow Direction
Suction surface
effectively aids flow in
navigating turn into cowl
Viewing Plane:
Centerline
Much high-energy fluid
Total Static
enters cowl and remains
1.30 1.20 until cowl outlet
1.04 0.96
(a) Total Pressure
0.78 0.72 Stagnation point is nearly

PSI PSI in ideal location
increasing the driving
0.52 0.48 force through the cowl
0.26 0.24
Flow stagnates in bends across
bottom of cowl
0.00 0.00
(b) Static Pressure A desirable high static pressure

Flow stagnates against forward cowl wall region above the outlet is obtained
characteristics of the flow in the critical regions (SC3 BLS).
Velocity (mph) 200 150 100 50 0
Total Press. (PSI) 1.300 0.975 0.650 0.325 0.000
Front of Car
Significantly higher velocity air Side-wall effects reduce the

entering cowl effective inlet area by about 5 %
Viewing Plane:
Inlet

More high total pressure air
compared to the free-stream Total pressure remains fairly
constant until this point
(c) Total Pressure Below Inlet

Viewing Plane:
1.25” Below Inlet Higher velocity flow diffuses
Cowl Centerline nicely
Figure 28. Contours of (a) velocity magnitude at the inlet, (b) total pressure at the inlet, and (c) total pressure 1.25”
below the cowl inlet (SC3 BLS).
24
Front of Car
Viewing Plane:
Outlet
Nearly uniform flow patterns helping to

reduce cylinder-to-cylinder variations
Rear of Car
Figure 29. Velocity vectors on the outlet plane show fairly uniform motion entering the carburetor (SC3 BLS).
25

A Robust CFD Methodology For Physically Realistic and Economically Feasible Results in Racing - Part II Intake Cowl

Uploaded by

Copyright:

Available Formats

A Robust CFD Methodology For Physically Realistic and Economically Feasible Results in Racing - Part II Intake Cowl

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Robust CFD Methodology For Physically Realistic and Economically Feasible Results in Racing - Part II Intake Cowl

Uploaded by

Copyright:

Available Formats

Downloaded from SAE International by Embraer S/A, Sunday, December 03, 2017

A Robust CFD Methodology for Physically

2006 SAE World Congress

For permission and licensing requests contact:

For multiple print copies contact:

SAE Customer Service

A Robust CFD Methodology for Physically Realistic and

Copyright © 2006 SAE International

ABSTRACT Through detailed analysis of flow physics novel

All together, fifteen different simulations were REFERENCE CASE RESULTS

Far-field Open Boundary

Far-field Open Boundary

Cowl Air Box

Figure 2. Three-dimensional computational domain (with foreground symmetry plane removed).

Stair-stepping using quads

Figure 3. Two-dimensional sample grid along symmetry plane (REF).

(a) REF Cowl Exit (b) SC2

(a) Velocity Magnitude

Velocity High speed flow does not

To carburetor Flow must accelerate to Small scoop in airbox isentropically

80 Large recirculation region

Flow in recirculation region

Thick boundary layers form

0.50 0.30 Stagnation region is behind the

Flow accelerating into the carburetor

Velocity (mph) 200 150 100 50 0

(b) Total Pressure at Inlet

Viewing Plane: (c) Total Pressure Below Inlet

High-energy flow does not

Viewing Plane: Total Pressure

(b) Total Pressure at

Vortex structures increase cylinder-

(a) Velocity Magnitude Zoom Area

The thinner boundary layer

90 Because a thin boundary layer

The simulation is modified to Flow Direction The thin boundary layer

0.75 0.75 The high energy flow diffuses

The stagnation region is

Flow accelerating into the carburetor

(a) Velocity Magnitude

Flow Direction Stagnation region is still at the

Velocity High speed flow diffuses better

(b) Velocity Vectors

Thick boundary layers form over

0.74 0.50 High-energy flow eventually

0.50 0.30 Stagnation region is still behind

Flow accelerating into the carburetor

(a) Velocity Magnitude

Flow Direction Stagnation region is still at the

Velocity With boundary layer tripped,

Flow must accelerate to Scoop design does not decelerate

(b) Velocity Vectors

Flow in recirculation region

Boundary layers are much

0.50 0.30 Stagnation region is behind the

Flow stagnates at base

(b) Static Pressure Flow accelerating into the carburetor

Note: Because of the complex nature of the 3D model, each

(a) Velocity Magnitude