Professional Documents
Culture Documents
Validation of A FEA Tire Model For Vehicle Dynamic Analysis and Full Vehicle Real Time Proving Ground Simulations
Validation of A FEA Tire Model For Vehicle Dynamic Analysis and Full Vehicle Real Time Proving Ground Simulations
John O. Hallquist
(Livermore Software Technology Corporation)
ABSTRACT
1. support and transfer vertical loads, absorb and reduce
A tire model and its interface performance with road ground impact and the consequent vehicle vibration
surface plays a major role in vehicle dynamics analysis and 2. provide longitudinal forces for acceleration and braking
full vehicle real time proving ground simulations. 3. provide lateral forces for cornering and steering
The successful tire model must be able to support the
vehicle weight, provide vehicle control and stability, transfer The complexity of modern pneumatic tires has
various forces and torques from road/tire interaction to a limited the application of analytical methods. Over the past
vehicle chassis/ suspension system. The dynamic effects in ten years or so, the Finite Element Analysis (FEA) method has
terms of tire stiffness and internal damping characteristics in been integrated into the tire design/testing process with
impact loading conditions must also be accounted for in the increasing pace [2]. Finite element tire model analysis has
model. been used to decrease the tire development cycle and even to
A Finite Element Analysis (FEA) tire model is replace certain tire tests. Noor and Tanner [3] have reviewed
established and its performance is validated using FEA tire modeling used by other researchers. While many
LS/DYNA3D * analysis code simulating the radial and lateral different FEA tire models have been developed to study the
static stiffness test conditions, the one-meter dynamic free- general tire behavior such as stress and deformation due to
drop test condition and the rolling cornering stiffness. The inflation, natural frequencies, footprint shape and rolling
analysis results are compared with available test data and a contact models [4-8], rolling resistance [9, 10], tire/rim
generic empirical formula. interaction [11, 12], or even to simulate some destructive tire
testing [13], very few tire models were developed to be
INTRODUCTION directly used in vehicle dynamics analysis [14] and none, to
the authors knowledge, were used in real time Proving
The tire-wheel system is one of the most important Ground simulation.
subsystems of a ground vehicle. Different control, drive and Virtual Proving Ground *** (VPG) is an analysis methodology
resistance forces created from the tire-ground interaction are developed at Engineering Technology Associates, Inc. for full
carried and transferred to the vehicle by tire. A modern tire- vehicle real time proving ground simulations. VPG
wheel system plays a key role in vehicle load carrying ability, applications require a totally different FEA tire model than
handling and steering stability, drivability and comfort. those used in tire design/testing studies. To investigate
Viewing the tire-wheel system as a load carrying and transfer general tire behavior or even to study tire dynamics [14], it
device, it provides the following three basic functions [1] ** : was common and usually sufficient to model a small portion
of the tire in fine finite element mesh and leave a very coarse
*
LS/DYNA3D is a trademark of Livermore Software Technology
***
Corporation Virtual Proving Ground (VPG) is a trademark of Engineering
**
Numbers in brackets refer to references at the end of this paper Technology Associates, Inc.
mesh for the rest part of the tire model. In many cases, only a The tire-wheel cross-sectional profile was defined in
two-dimensional model of the tire cross-sectional profile is the global y-z plane where the y-axis is in the tire width
necessary to fulfill the analysis [14]. Also in most of the direction and the z-axis is in the radial direction. The tire
previous studies, no wheel model was used. For a real time center is defined as the origin. Two kinds of tire cross-
VPG analysis, the tire model has to be a three-dimensional sectional profiles are used and are shown in Figure 1. The
axisymmetric one because it will roll on the ground in full positions of nodes in the profiles are created by ETA/FEMB
cycles. The wheel also has to be modeled together with the and then the coordinates of these nodes are output as a
tire model in order to attach this subsystem to a NASTRAN file and read into the input file. Also included in
chassis/suspension FEA model or, as in most of the VPG the input geometry data are the thickness of the plies.
applications, to a full vehicle FEA model. Asymmetric cross-sectional profiles can also be used.
The tire model used in vehicle dynamics analysis and
real time VPG applications is also differentiated from the
design/testing oriented tire models in several other aspects.
For example, most of the previously developed tire models
each had a portion with very fine finite element mesh. The
tire model would have had tens of thousands of shell/solid
elements if this fine element mesh had been extended
axisymmetrically to the entire model [15]. It would need
much larger and faster computers and much more CPU power
to use four such tire models, plus the entire full vehicle model,
in a real time analysis.
Another difference between the two types of tire
models is that while it is necessary for a design/testing FEA
tire model to focus on the detailed material properties of every
component for a particular type of tire, it is not necessary for a
VPG tire model to use the exact material properties for tires
from different manufacturers and types. The characteristics of
a vehicle dynamics or VPG tire model is determined by the
fact that the analysis is focused on the vehicle instead of the Fig. 1 Models for tire cross-sectional.
tire itself. In these applications, tire functions mainly as a
device which carries the weight of the vehicle and transfers
2) MATERIAL PROPERTIES
the various loads produced during tire-ground interaction to The plies, treads, sidewall, and chafer were put into
the vehicle chassis/suspension system. Keeping this in mind, different parts and material properties were assigned
a simple, efficient and yet accurate enough tire model is then separately for each part to simulate the inhomogeneities in tire
possible and, maybe, more practical. materials. For plies, either isotropic or anisotropic elastic
In the following sections, a FEA tire model for the
materials can be used. Layered composite plies can also be
vehicle dynamics analysis and VPG application purposes was
used if detailed material properties are provided. Elastic plies
developed. The tire model was also validated through several
are used presently because they provide the same global
simulations using LS/DYNA3D. These validations were
characteristics we are currently interested in and save
based on the available test data as well as the main features a
computing CPU time.
VPG tire should have. As discussed above, the interested
The explicit nonlinear three-dimensional dynamic
characteristics of a VPG tire model are some of the global
analysis code LS/DYNA3D has been used for the entire study.
properties. The main features included in this paper are the
Rubber materials (chafer, lower and upper treads) are modeled
static radial (vertical) stiffness, the static lateral stiffness, the
by solid elements and are of Mooney-Rivlin type [16].
dynamic response in a free drop condition, and the (steady
Shell/plate elements are used to model the plies and the wheel
rolling) cornering stiffness. Correlation of the analysis results
and are of Belytschko-Lin-Tsay type [17-19].
with the available test data and a generic empirical formula are
3) TIRE PRESSURE
also given. ETA/FEMB * were utilized to pre- / post-process
Tire inflation pressure can be set to be equal to the
all the simulation results.
exact value (in MPa), as required by user. Tire pressure will
change during the course, when the vehicle is running on the
FINITE ELEMENT MODEL road. The pressure change is mainly due to the tire operating
temperature variation and to the tire deformation (air volume
The construction of the FEA tire model includes two steps:
change). For an ideal gas like air, it is well-known that:
preparation of the input data file and generation of the FEA
model.
The input data file provides the following
(1) pV / T = const .
information:
1) TIRE GEOMETRY In vehicle dynamics analysis and VPG applications,
the assumption is that temperature change is small for a short
*
time period and then ignored. Equation (1) is then used with
ETA/FEMB is a trademark of Engineering Technology Associates, constant temperature.
Inc.
The tire model generation proceeds by utilizing a
FORTRAN program developed at Engineering Technology
Associates, Inc., to rotate the tire-wheel cross-sectional profile
about the y-axis and generate a complete axisymmetric mesh.
The plate elements in the input profile are extruded into solid
elements and the bar elements into shell elements. The holes
in the wheel disk are also axisymmetric and could be of any
reasonable number.
In general the tire-wheel model consists about 2100
to 2500 elements (depending on which type of cross-sectional
profile being selected), among which about half are solid
elements and the other half are shell/plate elements. There are
960 shell/plate elements onto which inflation pressure is
assigned.
Shown in Figure 2 are the different views of a
P215/60 R16 low-profile FEA tire model used in this study. Z
This model utilizes tire profile type 1 in Figure 1. It consists
Y
of 1280 solid elements and 2500 shell/plate elements. Figure
3 shows the FEA model of a P195/70 R14 tire which has also X Fig. 3 FEA model for P195/70 R14 tire.
been used in the present study. This model consists of 960
solid elements and 2150 shell/plate elements, and is The quasi-static analysis described in this section is
constructed from tire profile type 2 in Figure 1. designed to validate the overall radial (vertical) stiffness of the
Elements of the wheel center part are defined as rigid present FEA tire model. In the present simulation, instead of
plate elements. This is necessary when the tire model is gradually increasing the load at the wheel center, we
mounted to the vehicle drive axle via a revolute or cylindrical equivalently constrain the rigid wheel center part while move
joint in the LS/DYNA3D code for a vehicle dynamics analysis the test ground vertically up at a low, constant speed. The
and VPG simulation. load-time history is recorded and translated into a load-
In all the validation test simulations, a flat test ground deformation curve and the stiffness calculated.
surface modeled as rigid plate elements was used. Utilizing LS/DYNA3D system dynamic relaxation is used to
LS/DYNA3D, a contact interface is defined between this minimize the dynamic effect during the simulation process.
surface and the outer surface of the tire upper tread. Since the kinetic energy calculated during the entire loading
period is less than one percent of the total energy, this
QUASI-STATIC RADIAL STIFFNESS ANALYSIS simulation is considered a quasi-static one and so does the
resulting stiffness.
The overall radial (vertical) stiffness is one of most For the low profile tire P215/60 R16, as shown in
important tire properties which characterizes the load carrying Figure 2, the designed (vertical) static load is about 5400 N
Z Z
Z
Y
Y
X
X
Fig. 2 FEA model for P215/60 R16.
180
5000 170
4000 160
3000 150
2000 0.15 0.17 0.21 0.25 0.30
1000 Tire Pressure (MPa)
0
0 4 9 13 18 22 27 31 36 40 Fig. 6 Analysis results of the effects of inflation on the radial
stiffness of the P195/70 R14 tire model.
Displacem ent (m m )
The influence of tire inflation pressure upon tire The quasi-static lateral stiffness of a tire is closely
radial stiffness is shown in Figure 6 for the P195/70 R14 FEA related to vehicle cornering and steering ability. In quasi-
tire model shown previously in Figure 3. One can see from static lateral stiffness analysis conducted here, a certain
this plot that tire radial stiffness increases almost linearly with constant vertical load is first applied to the wheel center and a
Original Deformed
lateral load is then applied gradually at a constant speed. The
lateral load-displacement curve is one of the simulation
Fig. 5 Original and deformed cross-sectional profiles for P215/60 R16 tire model in radial stiffness simulation.
results. Tire lateral stiffness is then calculated from this Also listed in Table 3 are the lateral to radial stiffness
curve. (266.33 N/mm for this model) ratios. Experience indicates
that the static lateral stiffness of a tire is around 40% of its
Z
Y
Y
Fig. 7 Deformation of P195/70 R14 tire model in radial
stiffness simulation.
Fig. 8 Deformation of P195/70 R14 tire model in
lateral stiffness simulation.
Table 2: Simulated Radial Stiffness of P195/70 R14 Tire
Model at Different Loading Speed static radial stiffness value [20]. Based on this experience, the
Loading Speed Stiffness validation of the present tire model as shown in Table 3 can be
(mm/s) (N/mm) assumed to be excellent.
Simulation # 1 750.000 226.33
Simulation # 2 375.000 225.69 Table 3: Tire P195/70 R14 Model Quasi-Static Lateral
Simulation # 3 187.500 227.15 Stiffness Simulation Results
Simulation # 4 93.750 226.86 Wheel Center Lateral Stiffness Lateral/Radial
Loading (N) (N/mm) Stiffness Ratio
Simulation # 5 46.875 225.83
3078 82.76 36.57%
Average 226.37
5363 90.36 39.32%
7578 91.74 40.53%
Shown in Figure 8 is the deformed shape of the
P195/70 R14 tire model during lateral stiffness simulation. 10051 89.43 39.51%
The wheel disk center part is defined as a rigid body and kept
vertical (no rotation about x-axis) during the simulation
process. ONE-METER FREE-DROP SIMULATION
Tire lateral stiffness is a function of the applied
steady vertical load. It is common knowledge that the static The free-drop test is one of the standard dynamic
lateral stiffness of a tire will be of maximum value at a vertical tests of tires. The one meter free-drop simulation conducted
load near its desired load capacity. Shown in Table 3 are the in this study is designed to validate the dynamic behavior and
analyses results of the quasi-static lateral stiffness of the properties of the FEA tire model. From the equation of
P195/70 R14 tire model at different static vertical loading. It motion in dynamics, one can readily obtain:
can be seen that this tire model does have a maximum
stiffness at a vertical loading near 7500 N. v = 2 gh (2)
600 (3)
500 ∂α α =0
400
300 where α is the slip angle. The unit of the cornering stiffness is
200 conventionally chosen to be N/rad.
100 Previous FEA studies of tires under cornering forces
0 were focused on the transmission of loads from the tire
-100 footprint to the tire bead area and the wheel rim, and thus
0.00 0.10 0.20 0.30 0.40 0.50 0.60 static analysis was used under certain assumptions [21, 11].
Tim e (s) With LS/DYNA3D, we are able to simulate the side slip
phenomenon of a steadily rolling tire and calculate its
Fig. 9 Vertical displacement of P215/60 R16 tire model in cornering stiffness through nonlinear, transient dynamic
one-meter free-drop simulation. analysis.
6000 In the simulation, a constant vertical load is first
applied to the wheel center and then the tire is set in rotation
4000 at a constant angular velocity, driving the tire in the x-
Velocity (mm/s)
7500 (N)
4000 80000
3000 60000
5000 (N)
2000 40000
0 0
10000
15000
20000
25000
5000
0
1 4 7 10 13 16 19 22 25 28 31 34 37 40
Slip Angle (degree)
Vertical Load (N)
Fig. 11 Cornering force vs. slip angle as a function of vertical Fig. 12 Cornering stiffness vs. vertical loading.
loading: Simulation results of P195/70 R14 tire model. Solid line-“Magic Formula”; Dots-simulation results.