Downloaded from SAE International by University of Auckland, Sunday, August 19, 2018

2009-01-1156

Response of a Prototype Truck Hood to

Transient Aerodynamic Loading

Anurag Gupta, Joaquin I. Gargoloff and Bradley D. Duncan

Exa Corporation

Copyright © 2009 SAE International

ABSTRACT INTRODUCTION

A study was performed to determine the fluid structure Fluid structure interaction, or FSI, remains important to
interaction (FSI) for a prototype truck hood for transient many disciplines of engineering, automotive being one
aerodynamic loads. The growing need to make vehicle of the important ones. Automotive designs are
panels lighter to enhance the fuel economy of vehicles has constantly in a state of flux, driven by ever changing
made hood panels more prone to deformation and market needs and government regulations. However,
vibration from aerodynamic loads. Moreover, as global the challenges faced by the automotive industry and the
pedestrian crash standards become more stringent to need to innovate have never been greater. Rapid
provide safer front end designs to minimize injuries to head increase in fuel prices, stricter government oversight on
and leg, automotive manufacturers are being required to pollution standards, continuously evolving pedestrian
design flexible hoods that crush significantly more than the and occupant safety standards, and an increasingly
present designs to absorb the crash energy better. These demanding consumer has led to the need for innovation
flexible designs lead to potentially undesirable at an accelerated pace. The structural design of
deformations and/or vibration behavior of the hood at automobiles is at the forefront of this much needed
typical highway speeds. This paper presents a design revolution to address these various challenges.
methodology for performing fluid structure interaction For example, automotive panels need to be lighter
computations for a typical hood by detailing the process of and/or made of other non-traditional materials to lower
computing dynamic (time varying) aerodynamic loads the weight and gas consumption of vehicles. These
using CFD and related deformations of the hood using panels need to reduce the impact to pedestrians and
commercial structural solvers. The Lattice Boltzmann occupants during a crash. Automotive structures need to
Method (LBM) was employed for performing transient address crash safety, without sacrificing comfort and
Computational Fluid Dynamics analysis, and commercial aesthetics. New designs to satisfy these rapidly
finite element analysis (FEA) software was used for evolving needs have increased problems in the domain
structural calculations. The coupling between the two of FSI due to increased sensitivity of lighter-weight
codes, which is described in the paper, was achieved structural panels to aerodynamic loads.
using proprietary in-house tools. The paper identifies the
wind loading conditions that have significant potential to The vehicle hood is an important example. As
induce hood vibration by studying “disturbed” wind profiles automobile manufacturers have explored unconventional
such as those caused by high speed wakes, high speed materials and reduced panel thickness to make the hood
travel at substantial yaw angles, and wind gusts. This work lighter to meet the new fuel economy and legislative
demonstrates the benefits of the inherent unsteady flow standards, the problem of hood deformation and
field computational capability of LBM in determining the vibration under various highway driving conditions has
dynamic structural response to aerodynamic excitation. worsened. Automotive manufacturers are exploring

SAE Int. J. Commer. Veh. | Volume 2 | Issue 1 75

 Downloaded from SAE International by University of Auckland, Sunday, August 19, 2018

solutions that can be applied early in the design phase with such a simulation, for which accurate pressure
of the vehicle so that these problems can be captured loading must be provided in the low-frequency range.
before they occur. Transient CFD methods coupled with
FEA techniques can conveniently predict the impact of The domain of computational aeroacoustics of vehicles
disturbed wind profiles on hood deformation and is closely related to FSI, in that the transient pressure
vibration. This early design evaluation helps field on the surface produces a response of the structure
manufacturers gain confidence in their design and and the interior of the vehicle. The requirements for
reduce the cost of vehicle production. This paper transient CFD simulation are similar, and LBM has been
discusses the application of coupled CFD-FEA previously validated for aeroacoustics applications,
techniques for the computation of transient wind profile including the case of cross-wind.
with significant turbulence that is responsible for causing
undesirable hood vibration. It investigates the effect of In Senthooran et al. [8] a detailed sedan was simulated
time varying gradients of pressure and the periodicity of at both zero and ten degree yaw angle for both
load as the hood excitation parameters. aerodynamics and aeroacoustics analysis, and wall-
pressure fluctuation data were compared between
Because multidisciplinary test and simulation work simulation and experiment. The geometric details such
performed by the manufacturers is often proprietary in as complex grille geometry, glass offsets, A-pillar
nature, the literature on coupled physics simulation moldings, side-mirror shape, and door handle geometry
pertaining to automotive applications, such as FSI, is are evident in the model. Detailed flow structures were
rather sparse. Duncan et al. conducted one such study visualized in the flow field and on the surface side-mirror
[1] involving aerodynamics, thermal and acoustics. surface, to gain understanding into the pressure
Ramsay et al. conducted a study [2] on the static fluctuation spectra. In Senthooran et al. [9] the pressure
deformation of the hood and compared the results to the fluctuation spectra on the side glass were evaluated on
experimental tests. a vehicle with and without a side-mirror.

Other studies have been reported which examine the

COMPUTATIONAL APPROACH
effect of transient cross-wind. Demuth and Buck AERODYNAMICS
conducted numerical investigations on the unsteady
aerodynamics of road vehicles under gusty weather The commercial software, PowerFLOW was used for
conditions [3]. Dominy and Ryan [4] conducted an performing time-accurate, turbulent fluid dynamics
experimental investigation into modeling of transient cross simulations in this study. This software uses a special
wind gust influences on passenger vehicles. They also discretization of the Lattice Boltzmann method (LBM) on
looked at the aerodynamic forces in response to a a variable resolution Cartesian volumetric grid, which is
0
transient crosswind at a relative incidence angle of 30 [5] automatically generated by the software. In the LBM
method, the fundamental equation of motion for fluid
Examinations of transient aerodynamics on vehicles, in particles, the Boltzmann equation, is solved on a cubic
particular using LBM, have been reported by other lattice using a discrete set of fluid velocity states. The
authors. Albukrek et al. [6] showed the influence of discrete Boltzmann equation is coupled with the ideal
transient flow structure on the performance of a detailed gas equation of state to represent realistic transfer of
open-wheel race car, including the effect of steering mass and momentum sufficient to describe small-scale
angle. Duncan and Golsch [7] used statistical and spectral fluid motions with very low numerical dissipation. The
analysis to evaluate the unsteady flow features of various surface geometry is provided as an input in the form of a
complex-flow regions on the exterior of a stock-car race triangular surface mesh. The geometry of the surface
car. The results showed that turbulent eddies were enters the LBM discretization as boundary facets
captured in both the low frequency range up to 100 Hz, immersed in the lattice, and they provide boundary
which is dominated by fundamental and secondary conditions for the discrete velocity states by enforcing
instability mechanisms such as vortex shedding and wake zero-flux through the wall. The LBM thus provides a
oscillations, and in the mid-frequency range up to 1000 Hz mass and momentum solver that is capable of
and higher, in which the fluctuation spectra collapse to a computing the evolution of shear-layer instabilities,
theoretical cascade representative of isotropic turbulence. vortices, and pressure waves with high-order accuracy,
The low-frequency behavior of the various flow regions in the presence of complex geometric boundaries –
were used to characterize the nature of the turbulent flow aspects of disturbed fluid flow critical to accurately
structure in those regions, while the mid-frequency range predicting the wind induced vibration behavior of
for each collapsed to the more universal behavior when structures. A feature of the solver is that the method is
the simulation grid resolution was adequate. This study inherently time-dependent. A detailed account of the
shows that flow structure details needed for LBM method and the treatment of unsteady flow and
understanding vibration of vehicle panels can be obtained turbulence can be found in references [10] through [14].

76 SAE Int. J. Commer. Veh. | Volume 2 | Issue 1

 Downloaded from SAE International by University of Auckland, Sunday, August 19, 2018

STRUCTURAL MECHANICS 3) Compute the dynamic structural solution using

NASTRAN.
The commercial finite element code, MSC.NASTRAN was
used for structural computations. The structural mesh PHYSICAL MODEL
consisted primarily of shell elements. However, limited
number of beam and solid elements were also used in the GEOMETRY AND TEST CONDITIONS
model formulation. The fluid flow calculations from
PowerFLOW and the structural computations from The fluid simulations were performed on a prototype
NASTRAN were one-way coupled in this study. In other pickup to provide a complete description of the
words, the structural excitation was explicitly defined in aerodynamics under different gust conditions, including
the time domain and consisted of the time varying the interaction of the unsteady flow with the vehicle and
pressure field over each element in the structure model the unsteady loads exerted on different structural
that was computed by the fluid solver. The solution components (hood and door). The geometry of the
process concluded upon the calculation of the dynamic model was complete in terms of detail of trim, underhood
structural deformation (vibration) from the unsteady and underbody components. The test conditions for the
pressure field. One-way coupling is an accurate and aerodynamic baseline were a free-stream velocity of
0
efficient FSI formulation for scenarios in which the 41.1 m/s (92 mph) at 101325 Pa pressure and 20 C
structural deformation caused by the unsteady fluid flow is temperature. The conditions were matched to emulate a
relatively small as compared to the physical dimensions of physical wind-tunnel test in a static floor wind-tunnel.
the structure such that an iterative exchange of boundary The cooling fans were removed (as an approximation to
conditions between the fluid and the structural solver is a free-wheeling fan), and open grille with porous media
not needed. The transient structural calculations were models for the radiator and condenser were employed.
performed using the modal transient response method. The computational domain was represented as a low-
This analysis methodology is especially advantageous for blockage static wind-tunnel condition with velocity inlet
models with a large number of degrees of freedom and for and pressure outlet, both at a distance from the vehicle
the computation of a large number of time steps. This is a which would not influence the air flow. The inlet was at
common analysis technique, the details of which can be 1.5 vehicle-lengths upstream, and the outlet at 3.5
found in [15]. vehicle-lengths downstream. The floor condition was
matched to the wind-tunnel by providing boundary layer
FLUID-STRUCTURAL COUPLING growth using friction on the floor surface in the vicinity of
the vehicle. The ceiling was modeled as frictionless wall
The coupling between the fluid and the structural
and the side-walls were modeled as periodic boundary
disciplines was achieved by using an in-house tool that
conditions. Details of these boundary conditions follow in
maps the pressures from the fluid mesh (used in
the subsection titled “transient aerodynamic simulation”.
PowerFLOW calculations) on to the structural mesh. The
The setup of the physical model is shown in Figure 1.
tool allows the mapping of either the averaged pressure;
the pressure transient history; or the pressure spectra that
is obtained by performing Fast Fourier Transform of the
Outlet, P0
pressure time history. This pressure export tool makes
Flow
possible the computation of static, transient or harmonic
calculations using NASTRAN. A separate pressure map is
created on every element of the structural mesh that inlet
allows for an accurate representation of the spatial
variation of pressure. It must be noted that in the most
general case, the fluid and structural meshes are
substantially different, not only because of the mesh
fidelity which is much finer in the fluid model, but also
because the type of elements used in the two meshes can Figure 1. Computational domain to simulate an open-
be different. The pressure export tool used in this study is road condition
capable of accommodating various differences between
the fluid and the structural mesh. A summary of the Grid Resolution - The simulation parameters and
sequence of steps in the coupling loop for hood vibration geometry specify the simulated test environment. To
calculation is as follows: complete the specification of the simulation setup, the
1) Evaluate the unsteady flow field using fluid domain is divided into nested regions surrounding
PowerFLOW. the vehicle, and these regions are used to specify the
2) Transfer the transient pressure field from the grid spacing for the cubic lattice. The Lattice-Boltzmann
CFD solution to the NASTRAN model, using the Method in PowerFLOW allows the grid spacing to
pressure export tool. change by a factor of 2 between adjacent regions, and

SAE Int. J. Commer. Veh. | Volume 2 | Issue 1 77

 Downloaded from SAE International by University of Auckland, Sunday, August 19, 2018

so these regions are called Variable Resolution (VR) Figure 3 shows the location of the VR regions for each
regions. For the aerodynamics baseline the finest cell level from VR level 9 (4 mm cells) located at several
size was set to 4.0 mm, and 5 levels of refinement were critical locations on the vehicle surface such as the grille
used between the edges of the flow domain and the bars, front of the hood, cowl region including the
finest region on the vehicle. The grid spacing and windshield wipers, and in the cooling flow path from the
location of each region is summarized in Figure 2. grille to behind the radiator and condenser (not shown
here), to VR level 8 (8 mm cells) at key locations
mentioned above, and out to VR level 5 (64 mm cells)
used to cover the entire computational domain. This
particular resolution strategy has been standardized for
production vehicles, with some flexibility left to individual
judgment on the location of the finest regions, and the
exact cell size used. This standardized approach has
been validated to regularly produce drag predictions that
match very well with wind tunnel and road tests. This
(a) (b) approach results in models that take 2000-3000 CPU-
hours to solve on a 128-CPU cluster with 3.0 GHz
Pentium Xeon processors.

(c) (d)

Figure 3. Planar slices of the cubic lattice grid.

TRANSIENT AERODYNAMICS SIMULATION

With the specified grid resolution setup, the cubic lattice

grid is automatically generated during a pre-processing
step. Then the fluid domain is initialized by running a no-
gust straight flow simulation to produce a starting
condition for future runs. The set of gust cases were
then run starting from this initial condition, for an overall
(e) time window of 100,000 time steps, or 1.769 seconds of
physical real time.
Figure 2. Variable resolution regions specifying the grid
spacing for the cubic lattice grid: (a) VR9, 4 mm on key
Gust setup - Four different fluid configurations were
areas on driver-side A-pillar vortex, mirror wake, hood,
considered, in order to encompass a broad range of
windshield, cowl, and lateral of the vehicle, (b) VR8, 8
possible gust scenarios. These fluid configurations were:
mm on key areas on driver-side A-pillar vortex, mirror
wake, in region offset from upper body and enclosing the
1. Step at 20 degree yaw: This configuration simulates
underbody, (c) VR7, 16 mm in near-field flow volume
the flow over a vehicle plunging into a crosswind.
down to floor, (d) VR6, 32 mm large offset from VR7 and
2. Harmonic yaw at 5 Hz: This configuration simulates
(e) VR5, 64 mm cells in box around vehicle and
the flow over a vehicle in a pulsating crosswind,
extending through the computational domain.
such as the wake of a large vehicle with 5 Hz
frequency content.

78 SAE Int. J. Commer. Veh. | Volume 2 | Issue 1

 Downloaded from SAE International by University of Auckland, Sunday, August 19, 2018

3. Harmonic yaw at 10 Hz: Similar to harmonic at 5 Hz,

but with flow pulsating at a frequency of 10 Hz.
4. Disturbance from a grid: This configuration simulates
the flow over a vehicle behind a grid consisting of 5
horizontal slats. The design of this grid has been
adapted from an actual test grid used at Jaguar-
Land Rover facilities to study the soiling of vehicles
caused by dusty flow passing through such a grid.

The following figure shows the yaw profile for the

different gust models.

20

Figure 5. Flow configuration for 20 degree yaw step

15 gust. The images show snapshots of flow velocity at four
different time intervals as the vehicle approaches and
Yaw angle [deg]

then plunges into the crosswind. Vortex shedding due to

10
the yawed flow is quite evident in the lower set of
pictures.
5
Yaw-step [deg] Harmonic yaw at 5 Hz - The harmonic yaw at 5 Hz
Yaw-5Hz [deg] configuration models the flow over a vehicle that plunges
Yaw-10Hz [deg] onto a crosswind region that varies harmonically
0
between 0 and 20 degree yaw, the variation occurring
0 0.1 0.2 0.3 0.4
Time [sec] with a frequency of 5 Hz. Though this condition is
idealized compared to natural winds, it emulates a gusty
Figure 4. Yaw profile for different gust models used in condition in which cross-wind is variable over time. It
the present study. could also be experienced by a vehicle passing a static
array of obstacles on a windy day. The simulation is
The next section will present the details of each of these initialized with a straight flow condition until
four fluid configurations. Please note that the velocity convergence, and then a time-dependent lateral velocity
plots are presented in dimensionless magnitude, with component is added to the incoming flow such that the
freestream velocity having a magnitude of 1. In the resulting flow impinges on the vehicle at a time-
actual road tests, the vehicle speed on the test track was dependent yaw angle. The oscillating yaw front travels
about 31.2 m/s (70 mph) with 8.9 m/s (20 mph) of head downstream with the free-stream velocity. Figure 6
wind. This resulted in average anemometer readings of shows the flow configuration as the different yaw fronts
about 41.1 m/s (92 mph). Thus, 41.1 m/s (dimensionless pass over the vehicle.
magnitude of 1) was chosen as the vehicle velocity for
all simulations done for this paper.

Step at 20 degree yaw - The step at 20 degree yaw

configuration models the flow over a vehicle that plunges
into a 20 degree yaw crosswind. This condition can be
experienced by vehicles on the road due to strong
winds, especially in the presence of obstacles such as
buildings. The simulation is initialized with a straight flow
condition until convergence, and then a lateral velocity
component is added to the incoming flow such that the
resulting flow impinges on the vehicle at a 20 degree
yaw angle. The yaw front travels downstream with the
free-stream velocity. Figure 5 shows the flow
configuration as the yaw front passes over the vehicle. Figure 6. Flow configuration for the crosswind gust
After the yaw front passes over the vehicle and the pulsing at 5 Hz. The figure shows snapshots of flow
transient stage is over, the flow settles to a standard 20 velocity at four different time intervals as the pulsing
degree yaw flow. crosswind approaches and then travels over the vehicle.

SAE Int. J. Commer. Veh. | Volume 2 | Issue 1 79

 Downloaded from SAE International by University of Auckland, Sunday, August 19, 2018

Harmonic yaw at 10 Hz - The harmonic yaw at 10 Hz

configuration models the flow over a vehicle that plunges
into a crosswind region that varies harmonically between
0 and 20 degree yaw, the variation occurring with a
frequency of 10 Hz. The simulation is initialized with a
straight flow condition until convergence, and then a
time-dependent lateral velocity component is added to
the incoming flow such that the resulting flow impinges
on the vehicle at a time-dependent yaw angle. The
oscillating yaw front travels downstream with the free-
stream velocity. Figure 7 shows the flow configuration as
the different yaw fronts pass over the vehicle.

(a)

Figure 7. Flow configuration for the crosswind gust

pulsing at 10 Hz.
(b)
Flow through a grid like disturbance – The grid
configuration simulates the flow over a vehicle behind a Figure 8. Flow configuration for the grid disturbance in
grid consisting of 5 horizontal slats arranged vertically. the flow. The figure shows snapshots of flow velocity at
The turbulence pattern generated by this grid has a two different time intervals as the flow passes through
dominant fluctuation peak around 15 Hz, with broadband the grid and impinges on the vehicle. Substantial change
turbulence occurring up to much higher frequencies. in the velocity profile with time is an indication of strong
This provides a reasonable simulation of the flow behind unsteadiness in the flow.
a vehicle with a very unsteady wake. The simulation is
initialized with a coarse version of the computational DYNAMIC STRUCTURAL SIMULATION
domain, generated by removing the two finest variable
resolution regions. The turbulent wake of the grid The construction of the hood and the boundary
impacts the vehicle and generates a very unsteady constraints are shown in Figure 9. The hood geometry
pressure distribution on the surface of the vehicle. considered in this study consisted of four main
Figure 8 shows the flow configuration past the grid, and components: the top panel, the inner stiffening panel,
its impact on the vehicle. two sets of hinges to facilitate the opening and the
closing of the hood, and a latch lock at the front. All the
major components are constructed of steel. The outer
panel is 0.75 mm thick, and the inner panel, 0.66 mm.
The structural properties of the steel panels are as
3
follows: density=7850 kg/m , Young’s modulus=210
GPa, and poisson’s ratio=0.3. All panels were meshed
primarily with four noded quadrilateral shell elements.

80 SAE Int. J. Commer. Veh. | Volume 2 | Issue 1

 Downloaded from SAE International by University of Auckland, Sunday, August 19, 2018

Z
An evaluation of the structural model and the accuracy
of the boundary conditions was conducted by computing
X the natural frequencies and the mode shapes of the
Y hood structure. The first five computed natural
frequencies of the hood are shown in Table 1. These
frequencies matched well with previous experimental
tests and numerical simulations.
Mode Frequency (Hz)
Top side 1 20.5
2 22.2
3 26.4
4 33.3
5 34.9

Hinges: Table 1. The first five natural frequencies of the hood.

All DOF=0
The first mode at 20.5 Hz is primarily a torsional mode.
The second mode exhibits “oil canning” like bending,
partly aided by the fact that the levers connecting the
hinge and the hood are allowed to rotate about the Y-axis.
The third mode has a shape similar to mode 2, except that
Under side
the primary contribution to the “oil canning” motion is the
inward bending (Y-direction) of the hinge levers rather
Bumper: than their rotation about Y-axis. The first few modes are
All DOF=0 on car frame end typically the major contributors to the dynamic deformation
(vibration) behavior, which in this case are clustered
around 20-25 Hz. Any aerodynamic excitation that has a
periodicity of 20-25 Hz can be expected to excite the hood
substantially. The above mentioned mode shapes of the
Latch Ty=Tz=0
hood are shown in Figure 10. The undeformed shape is
shown in black outline.

Figure 9. Structural model of the hood – construction

and boundary constraints.
(a)
There are three primary constraint locations on the hood
– the hinges, the latch and hood bumpers. Hood
bumpers are two small flexible supports on each end of
the hood to allow the hood to seat on the vehicle body
while providing compliance against mechanical stresses.
All degrees of freedom were constrained from motion at
3 locations on the hinges, as shown in Figure 9, to (b)
simulate a rigid connection to the vehicle body. The latch
was constrained from translation in the Y and Z
direction. There were two additional constraints to
simulate the seating of the hood on two bumpers on the
vehicle body. These bumper constraints were simulated
by modeling 3-D (translational only) springs between
these two locations of the hood underside and the
(c)
vehicle frame. The ends of the springs on the vehicle
frame were constrained from motion in all directions. Figure 10. Exaggerated mode shapes of the first 3
These spring connections are also shown in Figure 9. resonant modes: (a) 20.5 Hz, (b) 22.2 Hz and (c) 26.4 Hz.

SAE Int. J. Commer. Veh. | Volume 2 | Issue 1 81

 Downloaded from SAE International by University of Auckland, Sunday, August 19, 2018

A linear modal transient response analysis method was VEHICLE PLUNGING INTO A CROSSWIND
used to compute the dynamic structural response of the
structure to the aerodynamic loads using the NASTRAN Figure 12 is a plot of the time varying composite lift force
solution sequence 112. Modes of up to 150 Hz were on the hood as the vehicle enters from a 00 yaw situation
included in the linear normal modes calculation. The to a 200 yaw situation while traveling at 41.1 m/s (92
hood was found to have a total of 25 modes up to 150 mph). The first domain represents the vehicle travelling
Hz, all of which were used in the dynamic response straight at 41.1 m/s. The second represents a rapid
calculations. 150 Hz was chosen as the modal change of pressure on the hood as the vehicle
truncation point because it is about 3 octaves higher encounters the crosswind at 200 yaw, and the third
than the frequency of the fundamental mode(s) at about domain is the time period after the pressure has
20 Hz. Since only the first few modes have significant developed and stabilized over the hood for constant 200
participation in the dynamic response, being 3 octaves yaw flow. Plot labeled “backwards” are plots of backward
larger provided significant cushion against modal averaged lift force on the hood and represent the mean
truncation errors. The structure was considered lightly composite lift force.
damped. A critical damping of 2% was employed in all
the structural runs.

RESULTS
20 0 yaw
This section discusses the fluid and the structural results
for the four different cases discussed above, namely, a) Crosswind
the vehicle plunging into a crosswind, b) the impact of plunge
sinusoidal gust at 5 Hz & 10 Hz and 200 yaw, and c) the
impact of a grid disturbance in the oncoming flow. The
dynamic deformation of the hood was mapped by 0 0 yaw
plotting a total of 20 points on the hood that span the
entire hood. Careful evaluation of multiple points on the
hood confirmed that the hood vibration behavior can
indeed be qualitatively and quantitatively represented by
these 20 points. These points will be referred by the Figure 12. Transient development of the composite lift
“line” they belong to in the discussion of results. Line 1 is force on the hood as the vehicle travels from 00 yaw to
a set of 5 points on the hood that span the length along 200 yaw.
the cowl, and line 4, the length of the front edge of the
hood. Lines 2 and 3 constitute a set of 10 points that In the structural model, the resulting aerodynamic
span the central section of the hood. This spatial pressures computed by the fluid solver are applied as a
mapping of the hood is shown in Figure 11. transient pressure field on each element. The static
deformation of the hood for the pressure distribution at
Line 4 Line 3 Line 2 Line 1
00 yaw was used as the initial condition for structural
calculations. This helped avoid introducing an artificial
step loading and ensured that any computed vibration of
Pt 1 a the hood resulted purely from the transient aerodynamic
Pt 4 a
forces on an already deformed hood structure from the
00 deg yaw travel. The resulting vibration behavior of the
hood is shown in Figure 13. Only the results for 3 points
Pt 1 b
each for line 1 and line 4 are presented to keep the
graphs uncluttered while capturing the essence of hood
Pt 1 c
vibration.

Pt 1 d

Pt 4 e
Pt 1 e

Figure 11. Spatial mapping of the hood for the

discussion of results.

82 SAE Int. J. Commer. Veh. | Volume 2 | Issue 1

 Downloaded from SAE International by University of Auckland, Sunday, August 19, 2018

1.0 step loads. The small amplitudes of dynamic motion

Point 1a observed might very well be resulting from the
1c instantaneous turbulence in the flow field.
0.5
Hood Disp (mm)

1e
CROSSWIND GUST PULSING AT 5Hz AND 10 Hz
0.0
The transient composite lift force on the hood for the
-0.5 sinusoidally pulsing crosswind gust at 5 Hz and 10 Hz is
shown in Figure 14. The yaw angle varies sinusoidally
-1.0 as well because of the pulsing nature of the inlet gust.
The amplitude of forces are higher at 5 Hz as compared
0.17 0.27 0.37 0.47 0.57 0.67 0.77 0.87 0.97
Time (sec) to 10 Hz, however, the rate of change in the hood
pressure is noticeably larger for the 10 Hz pulse. An
(a) 0
initial period of about 0.25 seconds represents a 0 yaw
travel to establish the initial conditions for the fluid and
3.0
the structural simulations.
2.5
Point 4a
Hood Disp (mm)

2.0
4c
1.5 4e
1.0
0.5
0.0
-0.5
0.17 0.27 0.37 0.47 0.57 0.67 0.77 0.87 0.97
Time (sec)
(b)

Figure 13. Hood displacement versus time for the case

of vehicle plunging into a 200 yaw crosswind while (a)
travelling at 41.1 m/s: (a) plot of the region of the hood
near the cowl, (b) plot of points at the front edge of the
hood.

It is noteworthy that hood vibration has maximum peak-

to-peak amplitude of 1.6 mm and it takes place right at
the edge close to the cowl, line 1. This result is
qualitatively and quantitatively in agreement with the real
road test data. This has special significance to vehicle
manufacturers as the region of the hood near the cowl is
typically directly in the line of sight, and hence vibration
in this region is easily noticeable. Also, the peak
vibration happens at point c (please refer to Figure 11),
i.e., the center of the hood in Y-direction. The dynamic
response of points b and d was found to be similar to
that of point c, except that the mean deformation of point
(b)
d is shifted in the positive Z axis. Another interesting
aspect to note is that the period of vibration at point c is
Figure 14. Development of the transient composite lift
slightly higher than 20 Hz. This indicates that the major
force on the hood for crosswinds pulsing at (a) 5 Hz and
contribution to the vibrating hood results from the
(b) 10 Hz
excitation of the first 2 to 3 modes, which is the classic
structural response to a step load. An examination of the
Spectral analysis was performed of the pressure history
20 point hood vibration map of Figure 11 shows that the
on the surface of the hood to determine the frequency
amplitude of hood vibration decreases progressively
content and validate the fluid results. A case that is set
from line 1 to line 4, i.e., from the region near the cowl to
up well with adequate resolution should show most of
the front edge of the hood. In fact, the results for line 4
the energy distributed in a band close to the frequency
don’t show any discernable vibration from the crosswind

SAE Int. J. Commer. Veh. | Volume 2 | Issue 1 83

 Downloaded from SAE International by University of Auckland, Sunday, August 19, 2018

of the gust pulse, and this band should be relatively 1.0

narrow. Figure 15, with narrow band peaks, indeed Point 1a
0.5 1c
confirms that the fluid calculations allowed little energy

Hood Disp (mm)

1e
dissipation or frequency smearing by preserving the gust 0.0
behavior at the inlet right up to the vehicle.
-0.5

-1.0

-1.5

-2.0
Pressure Spectrum (Pa)

0 0.25 0.5 0.75 1 1.25 1.5

Time (sec)
(a)

3.0
4b
2.5
4d

Hood Disp (mm)

2.0 4e
1.5

1.0
Frequency Hz
(a) 0.5

0.0
0 0.25 0.5 0.75 1 1.25 1.5
Time (sec)
(b)
Pressure Spectrum (Pa)

Figure 16. Hood displacement versus time for a pulsing

crosswind gust at 5 Hz and a peak yaw angle of 200 at
41.1 m/s (92 mph) road speed: (a) plot of the region of
the hood near the cowl (line 1), (b) plot of points at the
front edge of the hood (line 4).

The hood can be seen to be driven at the excitation

frequency of 5 Hz with maximum peak-to-peak
amplitude of 2.25 mm that occurs at the front edge of the
Frequency Hz
hood (line 4). The maximum vibration amplitude on the
(b) hood region near the cowl still remains substantial at
about 2 mm (point 1c). Careful examination of point 1 c
Figure 15. Spectrum of the pressure time history on the clearly shows the contribution of frequencies higher than
surface on the hood using amplitude scaling in units of the excitation frequency of 5 Hz to the dynamic
Pa: (a) 5 Hz and (b) 10 Hz. deformation, most likely due to the excitation of the first
few natural modes from the instantaneous peaks (and
valleys) in pressure that are superimposed on the 5 Hz
5 Hz gust - The static deformation of the hood for the pulse. These instantaneous peaks in pressure are a
pressure distribution at 00 yaw was used as the initial result of the complex transient aerodynamics on the
condition for structural calculations. The resulting hood due to the periodic cross-wind. The behavior at
dynamic hood displacement response for the 5 Hz pulse point b and d is very similar to that at point c, with the
is shown in Figure 16 on lines 1 and 4. exception at the front edge of the hood (line 4). A
complete reversal of phase can be observed between
point b and point d at the front edge of the hood. This
difference in phase was also observed between point a
and e. This phase reversal is indicative of the fact that
the pressure difference is asymmetric about the
0
centerline of the hood due to a yaw angle of 20 . This
effect becomes more prominent at the front edge

84 SAE Int. J. Commer. Veh. | Volume 2 | Issue 1

 Downloaded from SAE International by University of Auckland, Sunday, August 19, 2018

because the hood is constrained at only one latch that is GRID DISTURBANCE IN THE FLOW FIELD
at the center of the hood, thereby allowing it more
freedom for torsional motion at the front. Among the The transient composite lift force on the hood for the
results not shown, the amplitude of vibration is largest at flow passing through an upstream grid is shown in
point e as compared to points a through d for the entire Figure 18. The first ~ 0.4 seconds of data is the
front portion of the hood. The cause for this is also the numerical transient due to the initiation of the
pressure asymmetry as discussed above. aerodynamic solution from a set of imposed initial
conditions. Only the last 1 second of solution was used
10 Hz gust - The resulting dynamic hood displacement for all structural calculations. It is evident that the
response for the 10 Hz pulse is shown in Figure 17 on placement of a grid in the oncoming flow causes large
lines 1 and 4. The overall behavior for the 10 Hz pulse is wake turbulence, resulting in large variation of force on
very similar to that for the 5 Hz pulse. The excitation the hood with steep time gradients.
response frequency of the hood is 10 Hz, with higher
frequency contributions from the natural modes. The
maximum peak-to-peak amplitude of vibration is 1.75
mm at point c on the edge of the hood near the cowl
(line 1), which is noticeably smaller as compared to 2.25
mm for the 5 Hz gust. This is in response to the smaller
forces that develop on the hood at a higher excitation
frequency, as evident in Figure 14.

1.0
Point 1a
0.5
1c
Hood Disp (mm)

0.0 1e
-0.5

-1.0 Figure 18. Development of the transient composite lift

force on the hood for the flow passing through a grid
-1.5 disturbance placed 6 m upstream of the vehicle.
-2.0
Spectrum of the pressure history on the surface of the
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 hood is shown in Figure 19. A peak at 15 Hz can be
Time (sec)
seen in the spectrum. This 15 Hz content relates to the
(a)
Strouhal number for shedding behind a bluff body (grid)
at 41.1 m/s (92 mph). It is also noteworthy that the peak
2.5
is relatively broad, indicating that the time history is likely
4b
2.0
to contain frequency content above and below 15 Hz.
4d
Hood Disp (mm)

4e
1.5
Pressure Spectrum (Pa)

1.0

0.5

0.0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (sec)
(b)

Figure 17. Hood displacement versus time for a pulsing

crosswind gust at 10 Hz and a peak yaw angle of 200 at
41.1 m/s road speed: (a) plot of the region of the hood Frequency Hz
near the cowl (line 1), (b) plot of points at the front edge
of the hood (line 4). Figure 19. Pressure spectrum, in Pa, of the pressure time
history on the surface on the hood for the flow passing
through a grid disturbance 6 m upstream of the vehicle.

SAE Int. J. Commer. Veh. | Volume 2 | Issue 1 85

 Downloaded from SAE International by University of Auckland, Sunday, August 19, 2018

The results for hood vibration for this case are shown in first few natural modes. Maximum peak-to-peak
Figure 20. Just like all the previous structural amplitudes of 7.5 mm were observed on line 1 at points
simulations, the dynamic deformation solution was b, c, and d, i.e., at the edge near the cowl. The results
started from an initially deformed structure due to the for point b and d were very similar to those for point c;
static load. hence those points are not shown on the plot. This is
understandable as the grid was placed symmetrically
2.5 about the center of the vehicle in the Y-direction. Also,
1.8 Point 1c the results for point a were very similar to that of point c.
1.0 1e
The maximum peak-to-peak amplitude of vibration
Hood Disp (mm)

0.3 decreased on the hood from line 1 to line 4 in the

-0.5
following manner: 7.5 mm at line 1, 5 mm at line 2, 3 mm
-1.3
at line 3 and 2 mm at line 4. This can be observed by
-2.0
-2.8
comparing the plots of line 1, line 3 and line 4 in Figure
-3.5 20. All points on line 3 show comparable amplitude of
-4.3 displacement, unlike lines 1 and 2 (plot not shown for
-5.0 line 2) which had the largest amplitude at point b, c, and
0.25 0.35 0.45 0.55 0.65 0.75 d. This is because the hinged constraints on the hood
Time (sec) close to points a and e serve to reduce the deformation
(a) at those points. In contrast, points a and e show the
largest amplitude at the front edge (line 4) because of
1.5 the latched constraint in the vicinity of points b, c, and d.
Point 3c
1.0 3e
SUMMARY OF VIBRATION AMPLITUDE
Hood Disp (mm)

0.5 RESULTS
0.0
A graph summarizing the maximum peak-to-peak
-0.5 vibration amplitudes for all the 4 lines on Figure 11 and
-1.0
for all scenarios simulated for this study is presented in
Figure 21 to supplement the results presented above.
-1.5 Each line has been labeled with the point at which the
0.25 0.35 0.45 0.55 0.65 0.75 maximum vibration amplitude occurs on that particular
Time (sec) line, for example, the maximum vibration amplitude on
(b) line 1 for the 5 Hz gust case is 2 mm. It occurs at point c
1.5 (out of the 5 points, a through e) on that line.
Point 4c
1.0
4e
Hood Disp (mm)

pt e
0.5 pt e
Grid Disturbance pt c
pt c
0.0 pt e
pt e Line 4
-0.5 10 Hz gust pt c
pt c Line 3
pt e Line 2
-1.0 pt e
5 Hz gust pt e Line 1
pt c
-1.5
0.25 0.35 0.45 0.55 0.65 0.75 Step 20 deg yaw pt c
Time (sec) pt c
(c)
0 1 2 3 4 5 6 7 8
Vibration amplitude (mm)
Figure 20. Hood displacement versus time for grid
disturbance 6 m upstream of the vehicle at 41.1 m/s (92
mph) road speed: (a) plot of the region of the hood near Figure 21. Maximum dynamic displacement versus the
the cowl (line 1), (b) plot of points on line 3, (c) plot of point for each line and for all simulated conditions.
points at the front edge of the hood (line 4).
There is no discernable vibration at lines 3 and 4 at any
Figure 20 shows the hood vibrating at a frequency of the points for the 20 degree yaw step condition.
slighter higher than 20 Hz, clearly because the rapid Hence, the data is not represented on the plot. Figure 21
transient pressure gradients in the pressure time history makes it readily evident that points c and e experience
act as step loads on the structure, thereby exciting the the largest amount of displacement. This is because of

86 SAE Int. J. Commer. Veh. | Volume 2 | Issue 1

 Downloaded from SAE International by University of Auckland, Sunday, August 19, 2018

the pressure distribution asymmetry caused by the yaw 2. Ramsay, T., Fredelake, A., and Stevens, K.,
flow in each case that results in larger forces on the “Correlation of a CAE Hood Deflection Prediction
driver side of the hood. Method”, SAE Paper 2008-01-0098.
3. Demuth, R., and Buck, P., “Numerical investigations
CONCLUSIONS on the Unsteady Aerodynamics of Road Vehicles
under Gusty Weather Conditions”, Proceedings of
th
The simulation of the interaction of the transient flow the 6 MIRA International Vehicle Aerodynamics
field with the structure of a prototype hood was Conference, October 2006, pp. 48-60.
conducted in this study. The goal was to apply the 4. Dominy, R., and Ryan, A., “An Improved Wind
inherent transient flow field computational capability of Tunnel Configuration for the Investigation of
the Lattice Boltzmann solver to the prediction of the Aerodynamic Cross Wind Gust Response”, SAE
unsteady fluid flow around the hood of a vehicle, and to Paper 1999-01-0808.
predict the vibration behavior of the hood. To this end, 5. Ryan, A., and Dominy, R., “The Aerodynamic Forces
the effect of the time varying gradients of force and the Induced on a Passenger Vehicle in Response to a
periodicity of the load were investigated as excitation Transient Cross-Wind Gust at a Relative Incidence
0
parameters by imposing a wide range of aerodynamic of 30 ”, SAE Paper 980392.
loading conditions on the hood, namely, the plunge of 6. Albukrek, C., Doddegowda, P., Ivaldi, A., and
the vehicle into a crosswind that causes strong time Amodeo, J., “Unsteady Flow Analysis of a Formula
varying gradient of load; crosswind gusts pulsing at 5 Hz Type Open Wheel Race Car in Cornering”, SAE
and 10 Hz that result in strong time varying gradients as Paper 2006-01-3661, 2006.
well as periodicity of forces on the hood; and flow 7. Duncan, B.D., and Golsch, K., “Characterization of
through a regular grid disturbance that causes large Separated Turbulent Flow Regions in CFD Results
turbulence in the flow impinging on the hood of the for a Pontiac NASCAR Race Car”, SAE Paper 2004-
vehicle. 01-3556, 2004.
8. Senthooran, S., Crouse, B., Freed, D.,
Results showed that the hood is highly sensitive to the Balasubramamian, G., Noelting, S., Duncan, B. and
method of excitation in terms of the amplitudes of Powell, R., “Prediction of Wall Pressure Fluctuations
deformation as well as the nature of vibration. Step load on an Automobile Side-glass using a Lattice-
caused the hood to vibrate at frequency close to its first Boltzmann Method”, AIAA 2006-2559.
and second natural modes with maximum peak-to-peak 9. Senthooran, S., Crouse, B., Freed, D.,
amplitude of 1.6 mm. Crosswind gust pulsing at 5 Hz Balasubramamian, G. and Caridi, D., “Numerical
resulted in maximum amplitude of 2.25 mm, while the Simulation of Wind noise on the Side-glass of a
gust at 10 Hz resulted in smaller forces and smaller Production Automobile”, F2006D124, FISITA 2006.
amplitude of vibration, of 1.75 mm. The oscillatory 10. Lietz, R., Mallick, S., Kandasamy, S., and Chen, H.
signature of the pulsing crosswind was captured well in “Exterior Airflow Simulations Using a Lattice
the dynamic deformation response of the hood. The Boltzmann Approach”, SAE Paper 2002-01-0596.
turbulence generated by the grid had the effect of 11. Chen, H., Filippova, O., Hoch, J., Molvig, K., Shock,
exciting the first few natural modes of the hood and R., Teixera, C., and Zhang, R., “Grid Refinement in
resulted in 7.5 mm maximum peak-to-peak vibration. Lattice Boltzmann Methods Based on Volumetric
Formulation”, Physica A 362, pp. 157-167, 2006.
This study is the first step in the development of a 12. Fares, E., “Unsteady flow simulation of the Ahmed
methodology for automotive customers that can be reference body using a lattice Boltzmann approach”,
utilized early in the design phase of the vehicle to Computers and Fluids 35 (2006) 940-950.
capture and resolve potential panel vibration issues. 13. Chen, H., Teixeira, C., and Molvig, K., “Realization
Although applied to a hood, this solution methodology is of Fluid Boundary Conditions via Discrete Boltzmann
applicable to any automotive panel that could Dynamics,” Intl. J. Mod. Phys. C, 9 (8), 1998, p.
experience vibration. 1281.
14. Teixeira, C., “Incorporating Turbulence Models into
REFERENCES the Lattice-Boltzmann Method,” Intl. J. Mod. Phys.
C, Vol. 9 No. 8, 1998, pp. 1159-1175.
1. Duncan, B., Senthooran, S., Hendria, D., Sivakumar, 15. MSC.NASTRAN Version 68, Basic Dynamic
P., Freed, D., Gleason, M., and Hall, D., “Multi- Analysis User’s Guide, Chapter 6, section 6.3.
Disciplinary Aerodynamics Analysis for Production
Vehicles: Application of External Flow Simulations to
Aerodynamics, Aeroacoustics and Thermal
Management of a Pickup Truck”, SAE Paper 2007-
01-0100.

SAE Int. J. Commer. Veh. | Volume 2 | Issue 1 87