See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/276948147

A Rolling Resistance Simulation of Tires Using Static

Finite Element Analysis

Article in Tire Science and Technology · April 1999

DOI: 10.2346/1.2135980

CITATIONS READS

54 1,810

4 authors, including:

Masataka Koishi
The Yokohama Rubber Co.,Ltd., Japan
90 PUBLICATIONS 859 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Conceptual design of tires using multi-objective design exploration View project

All content following this page was uploaded by Masataka Koishi on 01 February 2017.

The user has requested enhancement of the downloaded file.

Z. Shida, 1 M. Koishi, 1 T. K o g u r e , l a n d K. K a b e t

A Rolling Resistance Simulation of Tires

Using Static Finite Element Analysis 2

REFERENCE: Shida, Z., Koishi, M., Kogure, T., and Kabe, K., " A Rolling Resistance
Simulation of Tires Using Static Finite Element Analysis," Tire Science and Technology,
TSTCA, Vol. 27, No. 2, April-June 1999, pp. 84-105.

ABSTRACT: A practical rolling resistance simulation method for tires using a static
finite element method is presented that fulfills three requirements: ( 1 ) easy input data
preparation, (2) shorter computation time, and (3) adequate accuracy. The method
implements a static deflection analysis first and the stress and strain thus obtained,
together with the loss factors of the materials determined separately, are used to esti-
mate the energy dissipation of a rolling tire.
First, ~he stress and strain profiles of all element groups that have the same cross-
sectional coordinates and are located along the circumferential direction are obtained.
Second, hysteresis loops are computed by introducing a viscoelastic phase lag between
the stress and strain profiles. The sum of the areas of the hysteresis loops is regarded
as the dissipation energy density of the element group. The loss factors of the rubber
materials are experimentally obtained and the effective loss tangents of the fiber-re-
inforced rubber are determined by the homogenization theory of dynamic viscoelas-
ticity. The rolling resistance simulation of a passenger radial tire using this approach
accurately captures the trends of an actual tire.

KEY WORDS: tire, finite element analysis, rolling resistance, energy dissipation, ho-
mogenization, viscoelasticity

Economical vehicle fuel consumption is rapidly increasing in importance

regarding energy savings and a cleaner environment. Because the rolling re-
sistance of tires is an essential factor influencing vehicle fuel consumption, it
must be reduced. In the past, the reduction of rolling resistance was mainly
accomplished by decreasing the loss factor and volume of viscoelastic mate-
rials, especially cap tread rubber. Although many researchers continue to study
these subjects, further reduction in rolling resistance is required by optimizing
the tire geometry and material configuration.
The rolling resistance simulation of tires using finite element analysis is
a useful tool for designing low-rolling resistance tires. For the method to
be a practical tool, the input data must be easily prepared, and an adequate

i Yokohama Rubber Company, 2-10iwake, Hiratsuka, Kanagawa, 254-860t Japan.

2 Presented at the seventeenth annual conference of The Tire Society, Akron, Ohio, April 28-29,
1998.

84
 SHIDA ET AL. ON ROLLING RESISTANCE SIMULATION 85

accuracy needs to be obtained in a short computational time. To this end,

a rolling resistance simulation method using static finite element analysis
is proposed, and parametric studies are presented.

Tire Rolling Resistance Simulation

The prediction of rolling resistance for tires or an energy dissipation
using finite elements has been studied by many researchers [1-5]. One
method evaluated the energy dissipation using the product of elastic strain
energy and the loss tangent of materials [1]. However, the anisotropy of
the loss factor in fiber-reinforced rubbers can not be considered appropriate
in this method. To overcome this difficulty, Warholic [2] proposed a
method that estimated the energy dissipation as a function of strain rate
components and the viscous damping matrix. The Kelvin-Voigt model was
chosen to represent the viscoelastic behavior of tire materials, and it was
extended to the more general case of isotropic and orthotropic materials
under triaxial state of stresses. On the other hand, other methods were also
proposed, in which the hysteretic losses are obtained from the relationship
of the variations of stresses and strains and impose phase lags between them
[3-5]. Park et al. [4] obtained the energy dissipation from the approxi-
mated variations of stresses and strains using cubic polynomials, and Allen
et aL [SJ obtained energy dissipation from approximated variations of
stresses and strains using harmonic cosine series.
To obtain adequate accuracy for the prediction of rolling resistance, an
energy dissipation model must be adopted that can handle anisotropic loss
factors of fiber-reinforced rubbers and can use the proper values of loss
factor for fiber-reinforced rubber. In this paper, a new algorithm is proposed
to estimate the energy dissipation from the relationship between the vari-
ation of stresses and strains that are approximated by Fourier series and the
phase lag due to the viscoelasticity of rubber. The equivalent loss tangent
of fiber-reinforced rubber is obtained by the homogenization method of
dynamic viscoelasticity in the frequency domain [ 7].

Tire Rolling Resistance

The rolling resistance of a tire is estimated by

RR = ED/L (1)

where ED and L are the total energy dissipation of the tire and the traveling
distance, respectively. Both Eo and L may be evaluated by finite elements for
a statically loaded tire. Since the CPU cost of static finite element analysis is
much cheaper than the dynamic analysis, the method is thought to be an ef-
fective tool for evaluating the rolling resistance of tires.
86 TIRE SCIENCE & TECHNOLOGY

A B
SS 0 ~3

Strain O9

Strain

FIG. 1 - - Stress-strain relationship of viscoelastic materials under cyclic loading." (A ) Phase lag
between stress and strain; ( B ) hysteresis loop.

Energy Dissipation
When a sinusoidal strain is imposed on viscoelastic materials such as rubber,
the stress varies sinusoidally with a phase angle (5 as shown in Fig. 1A. The
area confined by the hysteresis loop of the stress-strain plot corresponds to the
dissipated energy density caused by cyclic loading, as seen in Fig. lB. When
the stress cr and the strain e of the viscoelastic material are assumed to vary
sinusoidally, they may be expressed using the phase lag (5:

cfft) = or0 sin (a:t + (5) (2)

r = eo sin a:t (3)

The energy dissipated Ed in one cycle is

Ed = V f cr(t)dc
ycle (4)
= 7r- V" o-0"r sin (5

where V is the volume of the viscoelastic material, c~0 is the amplitude of the
stress, % is the amplitude of the strain, and (5 is the viscoelastic phase lag
between stress and strain. Therefore, the energy dissipation may be obtained
if these four values V, ~o, Co, 6 are known. In the present method, the finite
element analysis of a statically loaded tire is used to evaluate the amplitude
of the stress and strain and the volume of the material. For the phase lag
between stress and strain, experimental values corresponding to the loss factor
of the material are used. The treatment of the loss factor of the fiber-reinforced
rubber is discussed in more detail later. The estimation procedure for the en-
ergy dissipation during one rotation of the tire follows.
 SHIDA ET AL. ON ROLLING RESISTANCE SIMULATION 87

t, 3

z s 13 1

x/
FIG. 2 - - F i n i t e e l e m e n t c o o r d i n a t e s y s t e m : x, y, z - g l o b a l a x e s ; r, s, t - e l e m e n t a x e s ; 1, 2,
3 - m a t e r i a l axes.

The stress and strain evaluated at the continuous elements along the circum-
ferential direction are considered to be the time history of the stress and strain
at any point of the rolling tire. Thus, the time t which is a parameter repre-
senting the variation, may be replaced by an angle 0 which represents the
circumferential position of a tire. Then, the stress profile or(0) and strain profile
e(0) are obtained from the toroidal element groups located along the circum-
ferential direction with the same cross-sectional coordinates.
In many conventional finite elements, the stress and strain are expressed
with respect to the global coordinates. However, this treatment results in a
variation in appearance, even if they are constant, e.g., pure inflation of the
tire. Therefore, the stress and the strain in terms of the global coordinate
system need to be transformed into the stress and strain with respect to the
element coordinate system, as shown in Fig. 2. Also, the stress and strain
of the fiber-reinforced rubber materials need to be transformed into the
material coordinates which are convenient to deal with in the case of an-
isotropic loss factors in unidirectional reinforced composites.
Next, the viscoelastic phase lag/5 is imposed on the stress profile o-(0).
However, since the general profile of the stress and strain is impulsive rather
than sinusoidal as shown in Fig. 3, it is not possible to apply the phase lag
in the form of Eq 2. Relating to this, Park et al. [4] proposed a technique
to approximate the stress and strain profiles using cubic polynomials, and
a viscoelastic phase lag is imposed on the approximated stress. However,
the phase lag 6 in their method necessarily results in an n-times excess in
the nth harmonic component. Allen et al. [5] proposed a technique to fit
the stress and strain profiles numerically to a harmonic cosine series using
88 TIRE SCIENCE & TECHNOLOGY

0.200 ' ' ' ' I ' ' I I I I I i I I i i

0 (2=)

0.150
ca
v
=
0.100

09
0.050

0.000 . . . . i . . . . i . . . . i . . . .
0 1/2~ = 3/2= 2~

Circumferential position 0 (rad)

FIG. 3 -- Stress profile of the sidewall in the circumferential direction.

the least squares approximation. However, as shown in Figs. 4 and 5, the

actual stress and strain profiles referring to the local coordinates are not
necessarily symmetrical with respect to the center of the contact patch (0
= ~r) in the circumferential direction. This asymmetry is caused by the
asymmetry of the tire structure, such as the bias belt (Fig. 4) or by the
asymmetric shearing deformation (Fig. 5). Thus, it is not possible to ap-
proximate the asymmetric profile by the harmonic cosine series, because it
lacks phase information. In this paper, the stress profile c7(0) and the strain
profile e(0) are expanded by finite Fourier series of order N:

N
or(0) = ag + ~ {a~ cos (nO) + b~sin(nO)} (5)
n=l

and

N
e(0) = a~ + ~ {a~ cos (nO) + b~, sin (nO)} (6)
n=l
 SHIDA ET AL. ON ROLLING RESISTANCE SIMULATION 89

2 5 9 00 I I I I I I I I I I I I I I I I I I

o (2~)
20.00

13_
15.00
v

1
10.00
if)

5.00

0.00

500 .... I .... i .... i . . . .

0 1/2~ ~ 3/2rc 2~

Circumferential position 0 (rad)

FIG. 4 -- Stress profile of the belt edge in the longitudinal direction along the cord.

Our experience shows that the number of terms N needs to be more than
15 to adequately capture the peaks of the impulsive stress and strain pro-
files. Equations 5 and 6 may be rewritten as Eqs 7 and 8, using amplitude
A~ and phase 4~.:

or(0) = ag + ~ {A~ sin (nO + qS~)} (7)

t4=l

e(0) = a; + ~ {A~ sin (nO + ~b~)} (8)

rt=l

where

A. = ~aa] + b]

qS. = tan -L (b./a.)

90 TIRE SCIENCE & T E C H N O L O G Y

0.050 I , I , I ' I I , I I I I I I I I I I

0 (2~)

0.025
Ix.

t3~ 0.000

-0.025

-0.050
0
.... I .
1/2=
. .
U . .
3/2~ 2~

Circumferential position 0 (rad)

FIG. 5 -- Shear stress profile of the sidewall.

When the viscoelastic phase lag ~5is imposed on the stress profile c~(0), the
shifted stress profile 8(0) is expressed as

8(0) = ag + ~ {A~ sin (nO + 4)~ + 6n)} (9)

n=l

From the strain profile e(0) and the phase-shifted stress profile 8(0), the
area of the hysteresis loop is calculated in each harmonic order. The sum of
each area corresponds to the density of energy dissipation in the toroidal el-
ement group. The energy dissipation is obtained from the product of energy
dissipation density and the volume V of the circular element. Therefore, by
applying Eqs 8 and 9 to Eq 4, the energy dissipation Ed is given by Eq 10.
The coefficient n in Eq 10 is necessary, because nth order harmonics exist n
times within the fundamental period:

E a = ~ [Tc.n. VA,~ A n s i n ( 4 ~ -
9 (y .
4)~ + 6n)] (10)
n=l
 SHIDA ET AL. ON ROLLING RESISTANCE SIMULATION 91

The energy dissipation of one circular element group is obtained by in-

tegrating each energy dissipation with the six components of the stress and
strain vectors. Furthermore, the total energy dissipation of the entire tire is
calculated by integrating the energy dissipation at each toroidal element
group. Therefore, the total energy dissipation ED is obtained in the extended
form of Eq 10:
nel 6 N

ED= ~. ~. ~ [Tv.n. Vi.A~,,,n.A~mnSin(q~]mn-~9~m,~+5~,,,,,)] (11)

I--1 m--1 n = l

where l is the element number, m is the component of the stress and strain
vector, n is the harmonic order, and nel is the total number of elements in
the tire section.
In this model, the proper phase lag 5 can be imposed in each of the six
components of the vector for the fiber-reinforced rubber. Also, frequency- and
strain amplitude-dependent materials can be treated by the appropriate phase
lag.

Traveling Distance
The traveling distance L per one rotation of the tire is calculated by

L = 2~r[k. (R1 - R0) + R0] (12)

where Ro is the outer radius of the inflated tire and R1 is a statically loaded
radius; k is an experimentally determined coefficient whose value is between
0 and 1. The tire rolling resistance is calculated by substituting Eqs 11 and 12
into Eq 1.

Loss Tangent of Fiber-Reinforced Rubber Using the Homogenization

Method
The homogenization method has been studied as a powerful mathemati-
cal tool to predict the equivalent material properties and microscopic re-
sponse of composite materials [6]. Since the homogenization method is
based on a rigorous mathematical theory, it can provide reasonable solu-
tions for problems for which experimental data are not available.
The homogenization method of dynamic viscoelasticity in the frequency
domain was proposed by Koishi [ 7] and was applied to predict the equivalent
loss tangents of twisted steel cord-rubber composites and those of polyester
cord-rubber composites. We briefly derive the homogenization method of dy-
namic viscoelasticity from the principle of virtual work, and provide the equiv-
alent loss tangents for reinforcements using the homogenization method.
When the d'Alembert force is used for the inertia effect, the principle of
virtual work is
92 TIRE SCIENCE & TECHNOLOGY

(13)

where p, ii, v, o-, b, and t denote mass density, velocity, virtual displacement,
stress, body force, and traction, respectively.
We assume that the dynamic response is steady state harmonic vibration,
so that we can rewrite the principle of virtual work as

too,1
_~2~Lv,v,,j{U')d~+~ 0v' 0v"l[~'' -D"]~ 0x ~d~
LOx -b-]-xJ D, j/ou,,/
L OxJ

(14)

v J{bb}d + Lv v J{t}dF
t t

where ' and" denote the real and imaginary parts of a vector, respectively.
In the homogenization method, it is usually assumed that the composite
material is locally formed by the spatial repetition of unit cells, which are very
small microstructures in comparison with the overall macroscopic dimensions
of the structure. Let x be a macroscopic description of a structural body and
let y be a microscopic description of a unit cell. The microscopic level y is
expressed as

x
Y = -~7 (15)

where ~7is the ratio between the macroscopic level and the microscopic level.
It is reasonable to assume that the complex displacement can be expressed
as an asymptotic expansion with respect to the parameter q, namely,

x
u ~* = u ~ + ~ul*(x,y) + ... y - (16)

Substituting Eq 16 into Eq 14 and using Eq 15 gives the following two un-

coupled weak forms of the equilibrium equations: the first is on the micro-
scopic level and the second is on the macroscopic level.

f ox__L]
fy Ov' Ov"[[D' -D"] Oy~dY= f]Ov' 0v"l~_D,,)dY
D'
L Oy 7yJLD"
k 8yJ
 SHIDA ET AL. ON ROLLING RESISTANCE SIMULATION 93

- w 2 f~ p~Lv' v"J~~u~ + 1" Ov' 0< to-, -n-,,]

tu J jo Lax -bTJLo ' ' o" j

b n' ] t'

L-~-x J
where X is a characteristic function of the microstructure. Furthermore, the
following parameters,

p~= iF]
1 fr pdY

= eLI (
Irr n, - D' ~
-~y
- n" Ox"~ ] d r
Oy )J
,,

= ]YI D"- D" - ~ + D ' Oy ) dr

bn,= 1
]YI fy b'dY
and

bm, = IYI
1 fy b"dr

are homogenized mass density, homogenized storage elastic matrix, homog-

enized loss elastic matrix, real part of the homogenized body force, and imag-
inary part of the homogenized body force, respectively. [ Y[ stands for the
volume of a unit cell.
Let tan 6 H be a homogenized equivalent loss tangent, which is defined as
the ratio of the homogenized loss modulus to the homogenized storage mod-
ulus.
~((" C'Z:'
tan 6" =
e~' c ij~;
~ 'J - ~ ~J~

where E and C denote the homogenized modulus and homogenized compli-

ance matrix which is defined as the inverse of the homogenized elastic matrix.
The microscopic response calculated by the homogenization method certainly
depends on the ratio between the macroscopic level and the microscopic level.
However, the macroscopic response represented by equivalent material prop-
erties (including equivalent loss tangent) using the homogenization method is
so robust that it is not affected by the ratio.
94 TIRE SCIENCE & TECHNOLOGY

FIG. 6 -- Unidirectional reinforced rubber composite composed of two rubber phases.

The equivalent loss tangent of the fiber-reinforced rubber is computed using

the homogenization method discretized by the finite element technique. To
demonstrate the accuracy of this formulation, the equivalent loss tangent of a
rubber composite obtained by the homogenization method is compared with
the measured equivalent loss tangent. Since the experimental results can be
obtained accurately, a two-phase rubber composite specimen is considered in
this illustration. The unidirectional composite contains stiff cylindrical rubber
inclusions, whose diameters are 1.5 ram; embedded in a soft rubber matrix as
shown in Fig. 6. The measured material properties of both soft and stiff rubber
are shown in Table 1. Figure 7 shows the computed equivalent loss tangent
and the measured one for different skew angles. The equivalent loss tangents
obtained by the homogenization method agree well with the experimental re-
sults; however, those predicted by the law of mixture do not agree with the
experimental results.
At the next step, a parametric study is performed. Figure 8 shows the re-
lationship between the volume fraction of the fiber and the equivalent loss
tangent of the fiber-reinforced rubber, and Fig. 9 shows the typical finite ele-

TABLE 1 - - Dynamic viscoelasticproperties of soft and stiff rubbers

(Frequency: 20 Hz, Temperature:20~

tan E' (MPa)

Soft Rubber 0,026 1.93

Stiff Rubber 0.237 11.7
 SHIDA ET AL. ON ROLLING RESISTANCE SIMULATION 95

I ' ' ' ' ' I ' ' ' ' ' I ' ' ' ' ' I

0.20
-" Homogenization
o Experiment
t- o "',, ,,,,,,, -..... Law of mixture ]

t-
0.15

Ip
1/1
.9o
t-
0.10

._>
C3r x

LL!

0.05

I ~ ~ ~ , I , , , , , I . . . . . I

0.0 30.0 60.0 90.0

Skew angle 13[deg]

FIG. 7 - - Comparison of calculated equivalent loss tangent using homogenization and experi-
mental results.

ment model used for computing the equivalent loss tangent. The tan 6L and
tan 6r ill Fig. 8 indicate equivalent loss tangents in the longitudinal and the
transverse directions, respectively. The ratio of the fiber modulus Ej to matrix
modulus Er is also changed parametrically. When the ratio ET/Eris increased
to 10 3, it is observed that the equivalent loss tangent in the longitudinal di-
rection can be approximated by the loss tangent of the fiber, and the equivalent
loss tangent in the transverse direction can be approximated by the loss tangent
of the rubber materials.

Results and Discussion

The rolling resistance computation is performed through the deflection anal-
ysis of a statically loaded tire using a noncommercial finite element code.
According to the parametric study using the homogenization method of dy-
namic viscoelasticity, the loss factor of the fiber-reinforced rubber needs some
caution. The experimental loss factor of the cord is used for the longitudinal
direction, and the experimental loss factor of the rubber matrix is used for the
96 TIRE SCIENCE & TECHNOLOGY

I ' ' ' ' I ' " ' ~ I ' ' ' ' I ' ' ' " I

=I.-,
/ tan~T(Ef/Er=101)
i-
/ / tan~-r(Ef/Er=102)
t.- / / ~ tanST(EJEr=l03)
1.00 11--: 4 - - o .... .
~)
0
e-- :\ .. ,,
,, ,

N
= 0.50
E
Z
o \-- tan~,(E#E,=102)
tan~L(EJEr=l 03)
I , , , , l , , ~ , l , , I I ] l i i ~ l

0.0 25.0 50.0 75.0 100,0

Volume fraction of fiber [%]

FIG. 8 - - Equivalent loss tangent of fiber-reinforced rubber as a function of fiber volume fraction.

other direction, because the modulus ratio of fiber to matrix is very large ( 103-
105) in the case of the usual fiber-reinforced rubber in tires. A computation
time of 1 - 2 min per case is needed to calculate the rolling resistance on a
SUN SPARCstation 20 after the finite element analysis has been conducted.
The rolling resistance experiment is performed on an indoor drum testing
machine, with an outer diameter of 1707 mm, and the ambient temperature is
maintained at 25 + 2~ The rolling resistance is measured at a velocity of 80
km/h, after a 25-rain warm-up.
Table 2 summarizes the tire size and inflation pressure as well as load, for
the comparison between calculation and experiment. Figure 10 shows the re-
lationship between the calculated rolling resistance and experimental results.
The computational results agree well qualitatively with experimental data. The
correlation coefficient between calculation and experiment is 0.949.

Distribution of Energy Dissipation Density

Figure 11 is a half section of a finite element model for a 175/70R13 radial
tire used to calculate the distribution of energy dissipation, to which an infla-
tion pressure of 200 kPa and vertical load of 2.94 kN are applied. Figure 12
 SHIDA ET AL. ON ROLLING RESISTANCE SIMULATION 97

FIG. 9 - - Finite element model o f fiber-reinforced rubber.

shows the contour plot of the calculated energy dissipation density. The large
energy is dissipated in the tread cap region, especially at the belt edge and
shoulder regions, because the loss factor of the tread cap rubber is large com-
pared with other regions. The very large energy dissipation at the belt edge
region is due to the generation of extremely large interlaminar shear strains.

TABLE 2 - - List of tire sizes, inflation pressures, and load conditions used for calculations and experiments.
i

Inflation Pressure
Tire (kPa) Load (kN)

185/65R13 210 2.45, 3.43, 4.4l

185/65R14 (1) ~ 210 2.45, 3.43, 4.41
195/70Rl4 210 2.45, 3.43, 4.41
175/70R15 210 2.45, 3.43, 4.41
175/80R15 210 2.45, 3.43, 4.41
185/65R15 210 2.45, 3.43, 4.41
185/70R15 210 2.45, 3.43, 4.4I
195/65R15 210 2.45, 3.43, 4.41
135R13 180 2.75
185/65R14 (2) ~ 200 3.43

Different constructions.
98 TIRE SCIENCE & TECHNOLOGY

40.0 ' ' ' I ' ' ' ' I ' ' ' ' l ' ' ; '

Y=0.95X-3.45 (r=0.949) 9 185/65R13

Z o 185/65R14(1)
(I} 9 195/70R14
o
r
30.0
[] 175/70R15
._~ * 175/80R15
o~
<> 185/65R15
9 185/70R15
t- 20.0
z~ 195/65R15
135SR13
185/65R14(2)
- - Regression
"5 10.0
_o
(3

i n a J I i n n
0.0
0.0 10.0 20.0 30.0 40.0
Experimental rolling resistance [N]

FIG. 10 -- Relationship between calculated and experimental rolling resistances.

The large energy dissipation at the shoulder region indicates that the triaxial
compressive stress is generated in the region near the contact area in the cir-
cumferential, radial, and normal directions. Finally, the large energy dissipa-
tion in the upper portion of the bead apex is due to the compressive stress in
the radial direction as well as shear strains.

Effect of Inflation Pressure and Vertical Load

Parametric studies performed for a 175/70R13 radial tire are discussed in
the following. Figure 13 shows the effect of inflation pressure on rolling re-
sistance. In the figure, rolling resistance is normalized by rolling resistance at
the inflation pressure of 200 kPa. The agreement between calculated rolling
resistance and experimental results is good in the pressure range 150-250
kPa, which is a practical range for the tire, even though normalization is made
at the middle point.
Figure 14 displays the effect of vertical load on rolling resistance. Here, the
rolling resistance is normalized by the roiling resistance at the load of 2.94
kN. The calculated rolling resistance agrees well with the experiment in the
load range from 2.5 to 3.5 kN, which is a practical range for this tire size.
 SHIDA ET AL. ON ROLLING RESISTANCE SIMULATION 99

)<

FIG. 11 - - Half section of a 3-D finite element model f o r a 175/70R13 tire.

Effect of Tire Geometry

In general, rolling resistance tends to decrease when the section width in-
creases, the height of the bead apex decreases, or the tread radius increases.
Figure 15 shows the relationship between rolling resistance and section width,
which supports this variation. The calculated results agree well with an em-
pirical curve obtained by regression analysis for accumulated experimental
data. Figure 16 illustrates the calculated energy dissipation for each region.
The rolling resistance decreases as the energy dissipation in the tread cap and
belt region decreases.
Figure 17 shows the effect of bead apex height on rolling resistance. The
calculated results agree well with the empirical ones. Figure 18 indicates that
the increase in rolling resistance is caused by the increase in the energy dis-
sipation in the tread cap region as well as the lower sidewall region.
Figure 19 depicts the relationship between rolling resistance and tread ra-
100 TIRE SCIENCE & TECHNOLOGY

FIG. 12 -- Distribution of energy dissipation density.

dius. Rolling resistance decreases as tread radius increases. Figure 20 suggests

that the decrease in energy dissipation in the tread cap and belt regions makes
the rolling resistance small.
The calculated rolling resistance agrees well with the experiments, and
also the effects of operating conditions and tire geometry on rolling resis-
tance are well represented by the simulation presented. In addition, the
changes in energy dissipation in each tire component can be accurately
estimated using the approach.

Conclusion

An easy-to-use and practical method for predicting the rolling resistance of

tires based on the static finite element analysis is presented. An energy dissi-
pation model that estimates the hysteretic loss in a tire using the variations of
 2.0 , ' ~ ' I ' ' ' ' I ' ' ' ' I ' ' ' '

9 Calculation
- - Experiment
0
e'-
1.5

1--
1.0
0

0.5
o Tire: 175/70R 13
z Load=2.94 kN

I i I I I I I I I ] I I I I ] I I I I
0.0
100 200 300 400
Inflation pressure [kPa]

F I G . 13 - - Rolling resistance as a function of inflation pressure.

2.0 . . . . I . . . . I . . . . t ' ' 9 '

9 Calculation
O
r
1.5

1.0
"0
._N
m
0.5
O Tire: 175/70R 13
Z
I.P.=200 kPa

I I I I I I I i I ] I I I I ] I I I r
0.0
1.0 2.0 3.0 4.0 5.0
Vertical load [kN]

FIG. 14 - - Rolling resistance as a function of vertical load.

 1.2 ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' '

9 Calculation
O - - Empirical curve
O
t--
1.1
.on

1.0

0
.N_

0.9
O Tire:175/70R13
z I.P.=200 kPa
Load=2.94 kN
8 ,I ,I ,I ,I I ,I ,I ,I ,I I ,I ,I ,| ,I I ,I ,I ,I ,

171.0 172.0 173.0 174.0 175.0

Section width [mm]
FIG. 15 -- Rolling resistance as a function of section width.

Cap tread a

Belt

i , 9 v . | ,,,,|,

[] 174.7 mm
Carcass

Side tread
Tire: 175/70R 13
~, I.P.=200 kPa
Lower side b Load=2.94 kN
, , I I I , I , I . . . . I . . . .

0.0 10.0 20.0 30.0 40.0

Dissipative energy [J]
FIG. 16 -- Energy dissipation in each region when the section width changes, alncluding base-
tread; bexcept side-tread.
 1.2 . . . . I ' ' ' ' I ' ' ' ' I ' ' ' '

Calculation
- - Empirical curve
(..}
t-
1.1
09

C~ / / e
t,-
1.0 J
/
@
N

0.9
O Tire: 175/70 R 13 /Ht//
Z I.P.=200 kPa
Load=2,94 kN bead apex height
l , ~ , I I , , l I , t , l I , r , ,
0,8
20 30 40 50 60
Bead apex height [mm]

FIG. 17 - - Rolling resistance as a function of bead apex height.

Cap tread a

Bead-apex height
Belt
26 mm
9 38 mm
51 mm
Carcass

Side tread
--~ Tire: 175/70 R 13
k I.P.=200 kPa
Lower side b Load=2.94 kN

0.0 10.0 20.0 30.0 40.0

Dissipative energy [J]
FIG. 1 8 - - Energy dissipation in each region when the bead apex height changes. "Including
base-tread; bexcept side-tread.
 1.2 ' ' ' I ' ' ' ~ I ' ' ' ' I ' ' '

9 Calculation
(1)
(3
1.1
.(n

c-
1.0 =

CO
0.9
o Tire:175/70R13
Z I.P.=200 kPa
Load=2.94 kN
0.8 '

200 300 400 500 600

Tread radius [ram]
FIG. 19 -- Rolling resistance as a function of tread radius.

Cap tread a

Tread radias
Belt
[] 260 mm
9 380 mm
[] 580 mm
Carcass

Side tread
Tire:175/70R13
~, I.P.=200 kPa
Lower side b Load=2.94 kN

0.0 10.0 20.0 30.0 40.0

Dissipative energy (J)
FIG. 2 0 - - Energy dissipation in each region when the tread radius changes, alncluding base-
tread," bexcept side-tread.
 SHIDA ET AL. ON ROLLING RESISTANCE SIMULATION 105

stresses and strains approximated by Fourier series with a viscoelastic phase

lag is proposed. This model can accommodate the anisotropic loss factor of
fiber-reinforced rubbers. This simulation system is composed of a combination
of the energy dissipation model with a homogenization method for the equiv-
alent loss tangent of fiber-reinforced rubbers.
Application of the simulation system to several radial passenger tires reveals
that the calculated rolling resistance qualitatively simulates the change in 1"oil-
ing resistance. Thus, the effect of the change in energy dissipation in each tire
component can be estimated, giving the total change of the rolling resistance
value for the tire. This simulation tool can be used for studying the relationship
between design variables and the rolling resistance of tires.

References
[1] Kobayashi, Y., Fukuda, M., Takagi, A., Kanetsuki, M., Matsuzawa, F., and Kabe, K., "Tech-
nique for Estimating Tire Rolling Resistance Using Finite Element Method," Transactions of
the Society of Automotive Engineers of Japan, No. 35, 1987, pp. 140-146, in Japanese.
[2] Warholic, T. C., "Tire Rolling Loss Prediction form the Finite Element Analysis of a Statically
Loaded Tire," M.S.E. Thesis, University of Akron, 1987.
[3] Luchini, J. R., Peters, J. M., and Arthur, R. H., "Tire Rolling Computation with the Finite
EIement Method," Tire Science and Technology, TSTCA, Vol. 22, No. 4, October-December
1994, pp. 206-222.
[4] Park, H. C., Youn, S.-K., Song, T. S., and Kim, N.-J., "Analysis of Temperature Distribution
in a Rolling Tire Due to Strain Energy Dissipation," Tire Science and Technology, TSTCA,
Vol. 25, No. 3, July-September, 1997, pp. 214-228.
[5] McAllen, J., Cuitino, A. M., and Semas, V., "Numerical Investigation of the Deformation
Characteristics and Heat Generation in Pneumatic Aircraft Tires, Part lI. Thermal Modeling,"
Finite Elements in Analysis and Design, Vol. 23, 1996, pp. 265-290.
[6] Lions, J. L., Some Methods in the Mathematical Analysis of Systems and Their Control 1981,
Science Press.
[;7] Koishi, M., "Analysis of Rubber-Based Composite Materials Using Homogenization
Method," Doctoral dissertation, 1998, Yokohama National University, in Japanese.

View publication stats