Tire Cornering Simulation Using Finite Element Analysis

KAZUYUKI KABE, MASATAKA KOISHI

Computational Mechanics Lab. Research and Development Center, Yokohama Rubber Co., 2-1, Oiwake, Hiratsuka,
Kanagawa 254-8601, Japan

Received 10 March 2000; accepted 2 May 2000

ABSTRACT: The finite element method (FEM) has been used widely in tire engineering.
Most tire simulations with FEM are static analysis, because tire is one of most complex
structures. Recently, transient analysis has been studied with explicit finite element
analysis code. In this paper, at first, we describe characteristics of tire analysis by
means of FEM, and implicit/explicit finite element codes for analysis. And then we
conduct tire cornering simulation with implicit and explicit finite element analysis
codes, where implicit finite element analysis (FEA) describes the steady state cornering
simulation and explicit FEA describes transient cornering simulation. In the case of
implicit FEA, finite element model of tire requires the fine mesh only in the contact
region because of formulation by moving reference frame technique. On the other hand,
for explicit FEA, fine mesh is required in the circumferential direction of tire. Predicted
cornering forces of passenger car’s tire using implicit/explicit FEM are compared with
experimental results obtained from MTS Flat-Test Tire Test Systems. We can get a
good correlation between predicted ones and experimental ones. CPU time for cornering
simulation using implicit FEM is shorter than that of explicit FEM. © 2000 John Wiley
& Sons, Inc. J Appl Polym Sci 78: 1566 –1572, 2000

Key words: tire; cornering; finite element method

INTRODUCTION And then we conduct tire cornering simulation

with implicit and explicit finite element analysis
The finite element method has been used widely codes, where implicit FE analysis (FEA) describes
in tire engineering. Since a tire is one of the most the steady state cornering simulation and explicit
complex nonlinear structures, most finite element FEA describes transient cornering simulation. In
analysis of a tire has been limited to static anal- the case of the implicit FEA, the finite element
ysis. Recently, transient finite element analysis of model of tire requires a fine mesh only in the
a tire has been studied using an explicit finite contact region because of formulation by moving
element code1,2 due to the development of both reference frame technique. On the other hand, for
finite element (FE) program and computer hard- explicit FEA, fine mesh is required in the circum-
ware. Although the explicit finite element analy- ferential direction of tire. Predicted cornering
sis code is efficient for dynamic analysis of a roll- forces of passenger car’s tire using implicit/ex-
ing tire, it has a high CPU cost.3 plicit FEA are compared with experimental re-
In this paper we first describe characteristics of sults obtained from MTS Flat-Test Tire Test Sys-
tire analysis by means of FE method (FEM), and tems.
implicit/explicit finite element codes for analysis.
CHARACTERISTICS OF TIRE
Correspondence to: K. Kabe. Basic functions required of a tire are as follows:
Journal of Applied Polymer Science, Vol. 78, 1566 –1572 (2000)
© 2000 John Wiley & Sons, Inc. 1. Load-carrying capacity
1566
 TIRE CORNERING SIMULATION 1567

2. Transmission of driving and breaking

torque
3. Cushioning ability
4. Road-holding ability

In addition to the above four functions, a tire

needs to have many other functions: for example,
to provide steering response, adequate mileage,
and dimensional stability; to consume minimum
power; and to have minimum noise and vibration.
In order to satisfy these many functions, a tire
forms a thin-walled composite toroidal shell with
rubber and cord, which is both highly flexible and
relatively inextensible. Characteristics of such a
tire are shown as follows (refer Table I): Figure 1 Cross-section of passenger car’s radial tire.

● large deformation and large strain,

● incompressibility of rubber, When we apply FEM to tire analysis, these
● material nonlinearity of rubber and cord, characteristics should be taken into account for
the analysis of actual tire response.
● composite structure (FRR; Fiber Reinforced
Rubber),
● viscoelasticity of rubber, and FINITE ELEMENT METHOD FOR TIRE
● rolling contact of tire in service. ANALYSIS

FEM has been used widely in tire engineering.

Table I Characteristics of Tire Most of tire simulations with FEM are static
analyses, because tires are very complex nonlin-
ear structures. Recently, transient analysis has
Large deformation been studied with explicit finite element analysis
and large strain code. In order to compute dynamic response of a
rolling tire, two different methodologies have
been proposed. One is the implicit time integra-
tion technique,4 – 8 in which special kinematical
Incompressibility relationships are used to compute steady state
dynamic response of rolling tire. For example,
Faria8 shows a steady state formulation of the
rolling contact problem with friction that allows
analyses of free rolling, cornering, acceleration,
Material nonlinearity
and breaking. Another is the explicit time inte-
gration technique, e.g., LS/DYNA3D1 and PAM-
SHOCK,2 as commercial codes. Kamoulakos and
Kao2 studied the transient dynamic responses of
Composite structure a tire rolling on a spinning drum with cleat.
In the following section, we also show a finite
element model of tire and cornering simulation of
tire using implicit FEA code (ABAQUS/Standard)
Viscoelasticity
and explicit FEA code (ABAQUS/Explicit).

Rolling contact FINITE ELEMENT MODEL OF TIRE

Figure 1 shows a structure of the typical passen-

ger car’s radial tire. The tire is made of both
1568 KABE AND KOISHI

location in a cross-section of tire should be deter-

mined using following lift equations. Assuming a
deformation during manufacturing process is
purely due to pantographic action, the angle and
spacing of REBAR in the cured tire can be ob-
tained as follows:

␣⫽ 冋 r⬘sin ␤ ⬘
r sin ␤ 册␣ ⬘, ␤ ⫽ cos⫺1 冋 r cos ␤ ⬘
r⬘共1 ⫹ ␧兲 册 (1)

Figure 2 Coordinate system of tire.

where ␣, ␤, r, and ␧ are spacing, angle, radius,
and elongation factor of fiber reinforcements after
rubber components such as cap tread and side- lift, respectively. Also, ␣⬘, ␤⬘, and r⬘ are spacing,
wall, and fiber-reinforced rubber components angle, and radius on a tire-building drum. After a
such as steel belt and textile carcass. tire is mounted on a rim, an inflation pressure of
Figure 2 shows an axis system for this simula- 200 kPa is applied to the axisymmetric tire FE
tion. A slip angle is an angle between direction of model.
wheel travel and direction of wheel heading. A
cornering force and a self-aligning torque charac- Tire Model for Implicit FEA
terize cornering ability of a tire. Direction of the
cornering force is normal to direction of wheel Figure 4 shows a three-dimensional FE tire model
travel. The self-aligning torque is moment around that can be generated based on the inflated re-
the axis in the direction of normal to road surface. sults with the symmetric model generation, and
Figure 3 shows an axisymmetric FE tire model, the symmetric results are transferred to a three-
which is a passenger car’s radial tire of 235/ dimensional format with capability of ABAQUS/
45ZR17. Rubber components are modeled by the Standard. A pavement is modeled as a rigid ele-
continuous elements with hyperelastic material ment. The steady-state transport analysis capa-
of Mooney–Rivlin form. The hybrid axisymmetric bility can be formulated with a moving reference
continuous element with twist degree of freedom frame technique, so a mesh needs only to be re-
is useful for tire modeling. Although a tire is fined in a contact region, as shown in Figure 4.
axisymmetric geometry, the inflated tire becomes This model has 5976 nodes and 4464 elements.
twist along the axisymmetric axis due to anisot- On the other hand, a fine mesh is required in
ropy of fiber reinforcements. Fiber reinforcement circumferential direction for the explicit finite el-
is modeled by the REBAR, which is embedded in ement technique, as mentioned later.3
the continuous elements, with linear elastic ma- Before a cornering simulation of a tire, braking
terial. and driving simulations are conducted with slip
Since an angle and a spacing of fiber reinforce- angle of 0° in order to determine a free rolling
ments are changed during manufacturing pro- radius of a tire. In the steady-state rolling simu-
cess, the angle and the spacing of REBAR at each lation based on moving reference frame tech-

Figure 3 Axisymmetric FE tire.

 TIRE CORNERING SIMULATION 1569

Figure 6 Three-dimensional tire model for explicit

Figure 4 Three-dimensional tire model for implicit FEA.
FEA.

nique, angular velocity of a tire is required to

synchronize with travel velocity. The free rolling
radius of a tire can be determined from the brak-
ing and driving simulations. Free rolling is de-
fined as the state at which no longitudinal force
occurs.
The tire model is in contact with a road under
vertical load of 4 kN traveling at 10 km/h. Angu-
lar velocity of the tire is varied from 8 to 10 rad/s
in order to determine the free rolling radius. Fig-
ure 5 shows the driving force of the tire (referred
to as longitudinal force) at different angular ve-
locities. It is shown that free rolling occurs at
angular velocity of 8.87 rad/s. Therefore, we can
estimate the free rolling radius of 313.28 mm.

Figure 7 (A) Transient results of cornering force. (B)

Figure 5 Driving force vs angular velocity of tire. Transient results of self aligning torque.
1570 KABE AND KOISHI

Figure 8 Cornering force and self aligning torque.

Figure 10 Cornering power at various vertical loads.

TIRE MODEL FOR EXPLICIT FEA

Figure 6 shows a three-dimensional FE tire model

by means of the same procedure as mentioned SAT (Self-Aligning Torque) using explicit FEM.
above. In the case of explicit FEA, the FE tire From these figures, external loads and external
model must have a fine mesh (14,940 nodes, moments (torque) are applied for 0.25 s, and then
11,160 elements) all over the circumferential di- each force and moment remain constant with
rection in order to analyze a transit response of a time. We define the computed CF and SAT as a
tire. This causes more numbers of element and mean value of this stable region.
node than in case of the implicit FEA. Also, three-
dimensional FEA must be needed not only for a
rolling analysis, but also for an inflation analysis.
These cause to spend longer CPU times than in CORNERING SIMULATION
the case of the implicit FEA.
Figure 7(A) and (B) show examples of time Cornering simulation is conducted at seven dif-
history for computed CF (cornering force) and ferent slip angles such as 0°, ⫾1°, ⫾2°, and ⫾3°

Figure 9 Cornering force at various slip angles. Figure 11 Self-aligning torque at various slip angles.
 TIRE CORNERING SIMULATION 1571

with the experimental results, which are between

results of implicit and explicit FEM. The numer-
ical and experimental results shown in this figure
are averaged value between ones at negative slip
angle and at positive slip angle. The calculated
results agree well with the experimental results.
Figure 10 shows cornering power at various
vertical loads using implicit and explicit FEM,
together with the experimental results. The cor-
nering power is defined as cornering force at slip
angle of 1°. The calculated results agree well with
the experimental results.
Figure 11 shows SAT at various slip angle us-
ing implicit and explicit FEM, together with the
experimental results, which are larger than both
calculated SAT.
Figure 12 Self-aligning torque stiffness. Figure 12 shows SATS (Self-Aligning Torque
Stiffness) at various vertical loads using implicit
and explicit FEM, together with the experimental
with inflation pressure (P) of 200 MPa, vertical results. The SATS is defined as the self-aligning
load (W) of 4 kN, and velocity (V) of 10 km/h. torque at slip angle of 1°. This figure shows good
Figure 8 shows predicted CF and SAT at slip correlation between calculated results and exper-
angles of ⫺1°, 0°, and 1°, which are simulated imental ones.
using implicit FEM. In this figure, experimental Also, Figure 13 shows contact shapes and con-
results are obtained using the MTS Flat-Track tact pressure distributions at slip angle of 3° us-
Tire Test Systems. From this figure, nonzero CF ing both implicit and explicit FEM, respectively.
and SAT are obtained at slip angle of 0° on both From this figure, both contact shapes and contact
finite element analysis and experiment. These are pressure distributions have the same tendency,
caused by ply steer force due to anisotropy of steel i.e., the maximum pressure distributions occur at
belt. The calculated results agree well with the shoulder area of tire tread.
experimental results. Table 2 shows the CPU times in this simula-
Figure 9 shows predicted CF at various slip tion with four conditions (0°, 1°, 2°, 3°) with a
angles using implicit and explicit FEM, together vector computer of Fujitsu VX-2. CPU times for

Figure 13 Contact pressure distribution at slip angle.

1572 KABE AND KOISHI

Table II CPU Time sis, the CPU times of implicit finite element anal-
ysis is about 30 times shorter than that of explicit
CPU Time finite element analysis.
FEA (Fujitsu VXE)

Implicit FEA (ABAQUS/Std.) 6.5 h

Explicit FEA (ABAQUS/Exp.) 192.0 h REFERENCES

1. Kao, B. G.; Muthukrishnan, M. Tire Sci Technol

1997, 25(4), 230.
the implicit FEA and explicit FEA were 6.5 and 2. Kamoulakos, A.; Kao, B. G. Tire Sci Technol 1998,
192 h, respectively. Obviously, the CPU time of 26(2), 84.
implicit FEA is about 30 times shorter than that 3. Koishi, M.; Kabe, K.; Shiratori, M. Tire Sci Technol
of explicit FEA. 1998, 26(2), 109.
4. Padovan, J. Comput Struct 1987, 27(2), 249.
5. Kennedy, R.; Padovan, J. Comput Struct 1987,
CONCLUSION 27(2), 259.
6. Nakajima, Y.; Padovan, J. Comput Struct 1987,
27(2), 275.
Cornering simulation of a tire is shown using both 7. Oden, J. T.; Lin, T. L. Comp Methods App Mechan
implicit and explicit finite element analysis codes. Eng 1986, 57, 297.
The computed cornering forces of tire using both 8. Faria, L. O. Ph.D. dissertation, The University of
implicit and explicit finite element analysis agree Texas at Austin, 1989.
well with experimental results. In comparison be- 9. Sakai, H. Tire Engineering; Guranpuri Syupan,
tween implicit and explicit finite element analy- 1987 (in Japanese).