ARTICLE IN PRESS

International Journal of Machine Tools & Manufacture 47 (2007) 5362

www.elsevier.com/locate/ijmactool

An improved thermal model for machine tool bearings

Xu Mina, Jiang Shuyuna,, Cai Yingb
a

Department of Mechanical Engineering, Southeast University, Nanjing, 210096, P.R. China

b
Wuxi Machine Tool Corporation, Wuxi, 214061, P.R. China

Received 28 November 2005; received in revised form 24 February 2006; accepted 27 February 2006
Available online 19 April 2006

Abstract
Thermal model for machine tool spindle is of great importance to machine tool design. Traditionally, the thermal contact resistance
between solid joints and the change of the heat generation power with the bearing temperature are often ignored when thermal
characteristics of a machine tool spindle are analyzed. This has caused inaccuracies in the thermal model. With the heat source models
and the heat transfer models from Bossmanns and Tu [Journal of Manufacturing Science and Engineering 123 (2001) 495501,
International Journal of Machine Tools & Manufacture 39 (1995) 13451366], a model including the thermal contact resistance at solid
joints based on a fractal model and the change of the heat generation power, viz. the amount of the heat generation per second, with the
bearing temperature increases is developed. The complete thermal model is used to simulate the temperature distribution in grinding
machine housing with a conventional spindle bearing. Compared with experiment, it is shown that the completed model is much more
accurate than the traditional model which ignores the two important factors above. The thermal expansion of the housing system is
analyzed.
r 2006 Elsevier Ltd. All rights reserved.
Keywords: Machine tool spindle; Thermal characteristics; Rolling bearing; Joint; Heat generation power

1. Introduction
In recent years, with the development of the high speed
machining, the thermal characteristics of machine tools
have been held of much account by many researchers. It is
one of important factors that affect the performance of
machine tool systems. There have been a number of
thermal or thermo-mechanical models to investigate the
thermal and dynamic responses of machine tool spindles.
Bossmanns and Tu [1,2] developed a nite difference model
to characterize the heat generation, heat transfer and heat
sinks of a high-speed motorized spindle. Lin et al., [3]
presented an integrated model with experimental validation
and sensitivity analysis for studying various thermomechanical-dynamic spindle behaviors at high speeds. Li
and Shin [4] developed a more comprehensive integrated
thermo-dynamic model for high-speed spindles using nite
element method, which is coupled with the spindle dynamic
Corresponding author. Tel.: +86 25 83794920; fax: +86 25 83791414.

E-mail address: [email protected] (J. Shuyun).

0890-6955/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijmachtools.2006.02.018

model through bearing heat generation and thermal

expansion of the whole system. They [5] also investigated
the effects of bearing conguration on the thermo-dynamic
behavior of spindles using the model. In the models
mentioned above, heat transfer and heat conduction,
which are the important boundary conditions for thermal
analysis, are taken into account. But all of them ignored
the thermal contact resistance at the solid joints and the
change of the heat generation power of the bearings with
temperature increases. They assumed that the temperatures
of two contact surfaces were coupled or set an experiential
constant value on the resistance for all kinds of joints and
the heat generation power was constant.
There are many joints existing in a machine tool spindle
system, such as the interfaces between the bearing and the
shaft, the bearing and the bearing support, the housing and
the covers and so on. When two surfaces are in contact, the
presence of surface roughness produces imperfect contact
at the joint, no matter how much the pressure between the
surfaces is. The imperfect contact results in a sharp
temperature drop across the joint. Such a temperature

ARTICLE IN PRESS
54

X. Min et al. / International Journal of Machine Tools & Manufacture 47 (2007) 5362

jump plays a signicant role in the thermal characteristics

of the system. On the other hand, as the heat generation
power of bearings is in proportion to the viscosity of the
base oil of the grease and the viscosity changes greatly with
temperature, the change of the heat generation power is
notable. Sometimes, the decrease in heat generation power
can reach 50% when the temperature increases. Ignoring
these two facets, the previously developed models fail to
provide a precise analysis result for the system.
In this paper, the thermal characteristics of a given
grinding machine housing with a conventional spindle
bearing was investigated experimentally and numerically,
using nite element method. With the heat source models
and the heat transfer models from Bossmanns and Tu [1,2],
this paper develops the models for the thermal contact
resistance at the solid joints based on a fractal model and
the heat generation power as the bearing temperature
varies to complete the thermal model.
2. Grinding machine housing
The setup to be modeled is a grinding machine, housing
of a centerless grinder, as shown in Fig. 1. Its spindle
bearing system is a conventional spindle with a grinding
wheel located on the bearing span (Fig. 2). The spindle
works at 1100 rpm and its maximum speed is 1500 rpm.The
material of the shaft is steel 40Cr and that of the frame is
cast iron.
Two pairs of bearings were mounted on the spindle. The
bearing span is about 600 mm.A double-row short cylindrical roller bearing NN3018 K and a pair of angular
contact ball bearings 7018 are on the left and a double-row
short cylindrical roller bearing is on the right. From left to
right, the bearings are marked 1, 2, 3 and 4, respectively.
The bearing mounting area of the shaft was ground.
The angular contact ball bearings are mounted with
clearance tolerance to avoid the preload variation of the
bearings due to temperature variation and arranged in O
arrangement (back to back arrangement) and preloaded by

Fig. 1. Model of the grinding machine housing.

Fig. 2. The spindle system.

Table 1
The viscosity-temperature characteristic of the grease
Temperature (1C)

Kinematic viscosity (mm2/s)

25
30
40

40
23
20

two sleeves and a locknut. The double-row short cylindrical roller bearings are mounted with interference tolerance. The preload of the bearings is 600N.
The Asnic HQ72-102 grease is used to lubricate the
bearings. The viscosity-temperature characteristic of the
base oil of the grease is shown in Table 1.
3. Thermal model
3.1. Finite element model
The nite element model for the grinding machine
housing is established using the ANSYS software package,
as shown in Fig. 3. Z-axis is on the axis of the shaft. The
origin of coordinate is on the mid of the abrasion wheel.
The SOLID87 element is used to simulate the temperature
eld distribution. And the structural element SOLID92 is
adopted to calculate the displacement of the grinding
machine housing due to temperature variation.
To get good calculating precision and speed simultaneously, the elements near the bearings are meshed much
more renedly than in other regions. There are a total of
51,355 solid elements for the FEA model.
The contact elements CONTA174 and TARGE170 are
used to simulate the joints in the system. Because the joints
between the housing and the covers are far from the
bearings, the inuence of the thermal resistances of these
joints on the thermal characteristics of the whole machine
tool system can be neglected. So only the thermal contact
resistances between the bearing outer rings and the bearing
support, and the inner rings and the shaft neck have been
considered. There are eight joints in the grinding carriage,

ARTICLE IN PRESS
X. Min et al. / International Journal of Machine Tools & Manufacture 47 (2007) 5362

55

thermal contact resistance is founded based on MB fractal

model to overcome the shortages of the common thermal
contact resistance models.
3.2.1. Thermal contact resistance.
The thermal contact resistance R can be expressed as [6]
R

Fig. 3. The nite element model.

1
,
Ahc

(1)

where A is the apparent contact area of a contact region,

hc is the contact conductance coefcient. Ignoring the
radiation heat transfer, the contact conductance coefcient
follows the relation


1 Ac 2k1 k2
Av
hc
kf ,
(2)
Lg A k1 k2 A
where Lg is the thickness of the void space between two
contact surfaces, Ac is the real contact area of the joint, Av
is the void area of the joint, k1, k2, kf are the thermal
conductivities of the materials of the two parts and the
medium, respectively. If Lg, Ac and Av are known, the
thermal contact resistance can be given.

Fig. 4. Joints between bearings and frame, bearings and shaft.

shown as Fig. 4. To simulate the joints, create contact pairs

at the joints and dene the real constant TCC, i.e. thermal
contact conductance, of each contact element as the
contact conductance coefcient of relative joint, which is
to be explained below.
3.2. Thermal contact resistance
The imperfect contact produced by surface roughness
indicates that it isnt a full contact at the joint. There are
many contact spots and cavities between the surfaces. In
the cavities, it is lled with lling material. Generally, the
thermal conductivity of the lling is much lower than that
of the parts, so it gives to the thermal resistance and the
temperature jumps [6].
Conventional methods to study the thermal contact
resistance are by experiment [7,8] and the models based on
the statistical characterization of the rough surfaces [9]. But
both the experiments and the parameters of the models
depend strongly on the resolution and the precision of the
roughness measuring instruments. They fail to provide an
analysis model that is suitable for all occasions. To
establish a scale-independent model, some researchers have
introduced fractal theory into the contact resistances [10],
but they ignored the upper limit of the sample length Lu of
the contact surface, which is important to the regular
engineering surfaces [11,12]. A new fractal model for

3.2.2. Dimensionless fractional contact area

A unique property of rough surfaces is that if a surface is
repeatedly magnied, increasing details of roughness are
observed right down to nanoscales. In addition, the proles
at all magnication appear quite similar in structure. Such
a behavior can be characterized by fractal geometry. In
contrast with the statistical parameters of the rough
surfaces, the fractal parameters do not depend on the
resolution of the roughness measuring instruments or the
length scale of the sample and are scale-independent.
Majumdar and Bhushan [13] developed a fractal
contact model based on the fractal theory to describe the
characterization of the rough surfaces. However, as an
engineering surface cannot be fully characterized by a pure
fractal because the surface contains a deterministic part in
its shape, Wang and Komvopoulos proposed the concept
of a fractal-regular surface [11]. Take an engineering
plane for example, though the surface prole appears
random, multiscale and disordered on a microscale, as it is
intentionally made at, it is regular in macroscale and
different from the naturally formed fractal surfaces, such as
the surface of the Earth. Therefore, the engineering
surfaces exhibit a fractal behavior only within a nite
range of length scales. To overcome the difculty, the
upper limit of the sample length Lu is proposed. The fractal
sample length L in WM function should be less than or
equal to Lu. For the sample whose length is larger than Lu,
it should be divided into a number of fractal domains with
areas equal to L2u . As the pressure distributes evenly, the
dimensionless fractional contact area of the sample
A Ac =A is equal to the dimensionless fractional contact
area of the fractal domain Ar Arf =Lu2 , where Arf is the
real contact area of the fractal domain.

ARTICLE IN PRESS
X. Min et al. / International Journal of Machine Tools & Manufacture 47 (2007) 5362

56

The improved MB fractal model has three parameters:

the fractal dimension of the surface D (1oDo2), the
fractal roughness parameter G and the upper limit of the
sample length Lu. The three characterization parameters
can be gotten from the power spectrum of the WM
function [11].
According to the contact theory [13], Ar* is related to the
deformation style of the asperities. When an island is in
contact with plane and brought deformation, if the contact
area of the island a is less than the critical microcontact
area ac, its deformation is plastic. If a4ac, it is elastic. ac is
given by
ac

H=2E2=D1

(3)

where H is the hardness of the softer material between two

parts, E is the equivalent elastic modulus, dened as
E [(1v21)/E1+(1v22)/E2]1, E1, E2, v1 and v2 are the
elastic moduli and Poissons ratios of the two materials,
respectively.
Therefore, if the largest contact area aL of the fractal
domain is less than the critical area, aLo ac, only plastic
deformation will take place at the joint. Ar* follows the
relation
Ar

p
.
H

(4)

If aL4ac, part of asperities are plastic contact and the

other asperities are elastic contact. Calculating the integrals
of the elastic and plastic forces at the joint and neglecting
small term, Ar* follows the relation [12]
When Do1.5
Ar

p!2=3D  2D2=3D
 p 2=3D
3 2p
Lu
g4 D
,
4
G
E

3.2.3. Thickness of the void space between two contact

surfaces
For two contact planes, the thickness of the void space at
the joint Lg can be expressed as
Lg z1 z2  d1  d2 ,

(7)

where z1, z2 are the height of the asperities of two planes

respectively,d1,d2 are the largest deformation of the
asperities of the planes respectively. Many of the engineering contact surfaces have the same roughness and
machining process. So Eq. (7) can be written as
Lg 2z  dL ,

(8)

where z can be obtained by the roughness of the surfaces.

On the other hand, according to the fractal theory, the
largest deformation of the asperities on the plane is
dL G D1 a0L 2D=2

(9)

where aL is the largest truncated area of the asperities,

aL 2aL.
The statistical distribution of the truncated microcontact
area a is given by
na0

D 2D=2 0 D=2 0 D2=2

c
aL a
,
2

(10)

where a 2a, then, the integrals in Eq. (10) can be

evaluated as
Z a0
L
D
a0
Arf
na0 a da0 c2D=2
(11)
4

2D L
0
From Eqs. (8)(11), and Ar Arf =L2u ; it is can be
obtained that
"


4  2D 2D=2
Lg 2 z  cD2D2=4 GD1
D
#
Ar 2D=2 Lu2D

(5)

12

when D41.5
Ar

g3 D

2=D

 p 2=D

 D1


32D=2
4
G
p
g1 D ac
3 2p Lu
2=D

2D=2
H
g2 D ac
E

3.2.4. Thermal contact resistance of the joint

In Eq. (2), Av/A 1-Ac/A. So the thermal contact
resistant of the joint R is
R

ac

where p is the apparent pressure,

is the normalized
critical microcontact area, ac 2ac =L2u ; c is the domain
extension factor for microcontact size distribution.

Lg
,
AAr k 1  Ar kf 

(13)

where k 2k1k2/(k1+k2). Ar* and Lg are determined from

Eqs. (4)(6) and (12).

g1 D 2  D=DD=2 D=3  2D,

Table 2
Contact conductance coefcients at the joints (m2  K/W)

g2 D D=2  D2D=2 ,
g3 D 2D=2 c44DD
g4 D g1 Dg3 Dc

2 =4

D232D=4 2=D3

D=4  2D

32D=3D

No. of bearings

#1

#2

#3

#4

Inner ring/shaft neck

Outer ring/bearing support

7680
500

7680
500

10122
2000

10122
2000

ARTICLE IN PRESS
X. Min et al. / International Journal of Machine Tools & Manufacture 47 (2007) 5362

57

The values of the contact conductance coefcients at the

joints are shown in Table 2.

condition, this paper complete the thermal analysis for the

grinding machine housing.

3.3. Heat generation and heat transfer

3.3.2. Heat transfer coefficient

The heat transfer coefcient for convection h is
investigated in [2,4]. It is dened as

3.3.1. Heat generation

The major heat generation of the system is caused by the
cutting process and the friction between the balls and races
of the bearings [14]. Assumed that the majority of cutting
heat is taken away by coolant and chips, the heat generated
by bearings is the dominant cause of temperature change.
The heat generated by a bearing can be computed as
H f 1:047  104 nM,

(14)

where Hf is the heat generated power (W), n is the rotating

speed of the bearing (rpm), M is the total frictional torque
of the bearing (N mm). The total frictional torque M
consists of two parts, one is the torque M1 due to applied
load and the other one is the torque M2 due to viscosity of
lubricant.
That is
M M1 M2,

N u kfluid
,
d

(19)

where kfluid is the thermal conductivity of the ambient air,

Nu is the Nusselt number. When the convection occurs at
outer surfaces of a long cylinder, such as a shaft, d is the
diameter of the cylinder.
The Nusselt number Nu is computed from the Reynolds
number, Re, and the Prandtl number, Pr, based on
different convection conditions. For this research, the
following equation is used [15]:
N u 0:133Re2=3 Pr1=3 ,

(20)

where

(15)

Re

ufluid d
,
nfluid

(21)

(16)

Pr

cfluid mfluid
,
kfluid

(22)

where
M 1 f 1 p1 d m ,

where f1 is a factor related to the bearing type and load, p1

is the bearing preload (N), dm is the mean diameter of the
bearing (mm).
M 2 107 f 0 n0 n2=3 d 3m

if n0 n  2000,

(17)

M 2 160  107 f 0 d 3m ;

if n0 no2000,

(18)

where f0 is a factor related to bearing type and lubrication

method, v0 is the kinematic viscosity of the lubricant (mm2/
s).
From above equations, it is can be seen that the heat
generation power is dependent on the viscosity of the base
oil of the grease. As the temperature rises, the viscosity will
decrease and therefore the heat generation power will
decrease, too. According to the viscosity-temperature
characteristic of the base oil of the grease shown in Table
1, the heat generation power of each bearing is shown in
Table 3.
Taking the heat generation powers with respect to the
bearing working temperatures as body load table and
applying it to the bearing volumes as the thermal boundary

Table 3
The heat generation power of each bearing (W)

where ufluid is the velocity, vfluid is the kinematic viscosity,

cfluid is the specic heat capacitance and mfluid is the dynamic
viscosity of the air.
This equation is valid for Reo4.3  105, 0.7oPro670.
For free convection around stationary surfaces, h
9:7 W=m2 K is used [2].
3.4. Bearing stiffness
When calculating the displacement of the grinding
machine housing due to temperature variation, the radial
and the axial stiffness of the bearing must be known. The
bearing stiffness of the two pair of bearings can be
obtained by the aid of the Rolling Bearing Analysis
Software developed by Xu and Jiang [16]. The values of
the stiffness are listed in Table 4. To calculate the stiffness,
the contact pairs discussed in 3.1 are used, too. Evaluate
the real constants of the contact element FKN and FKT,
i.e. normal stiffness and tangent stiffness, with relative
stiffness of the bearing.

Table 4
The stiffness of the bearings (N/m)

No. of bearings

#1

#2

#3

#4

251C
301C
401C

44.3
30.8
28

44.3
30.8
28

60.8
44
40.6

60.8
44
40.6

Bearing

NN3018K

7018

Radial stiffness
Axial Stiffness

5.3277e9
1.7759e8

1.817e8
6.057e6

ARTICLE IN PRESS
58

X. Min et al. / International Journal of Machine Tools & Manufacture 47 (2007) 5362

Fig. 5. Steady-state temperature distribution of the entire grinding machine housing.

42

4. Thermal analysis

4.1. Temperature analysis

40
38
Temperature (C)

The analysis includes the steady state and transient state

temperatures and the thermal expansion. For steady state
analysis, the temperature distribution of the whole grinding
machine housing is presented. For transient state analysis,
temperatures of six locations are measured with infrared
thermoscope to compare with simulation.

36
34
32
30
28

Fig. 5 shows the steady-state temperature distribution of

the entire grinding machine at a rotational speed of
1100 rpm and the reference temperature is 25 1C. The
temperature at the left of the housing is much higher than
that at the right. The maximum temperature occurs at the
inner ring of the bearing #2. This is mainly because the heat
generation powers of the left bearings are much greater
than that of the right bearings and the thickness of the left
bearing support is smaller than that of the right bearing
support. This causes very different thermal expansions
between the left and right bearing supports, which will be
discussed in Section 4.2.
Fig. 6 shows the temperature variation of the bearing #2
with respect to time. It rises sharply at the beginning and
gently after a short time.
Fig. 7(a)(f) are comparisons between the measured and
predicted temperature histories of six locations. The
locations are on the side surface of the grinding carriage
as shown in Fig. 1.

26
24
0

5000

10000

15000
Time (sec)

20000

25000

30000

Fig. 6. Temperature history of the bearing #2.

From Fig. 7, it can be seen that the temperature

predictions for the locations match with the measured
values very well despite the elements far from the bearings
were coarser. In contrast to bearing #2, the temperatures of
the six locations increase slowly at the very start and
sharply at the early stage, and gradually saturate to the
nal temperature when the amount of heat generation
balances with the heat dissipation into the atmosphere.
Their rising times are much longer than those of the
bearings.

ARTICLE IN PRESS
X. Min et al. / International Journal of Machine Tools & Manufacture 47 (2007) 5362

36

Simulations
Measurements

34

Temperature (C)

Temperature (C)

36

32
30
28
26
0

5000

(a)

10000
15000
Time (sec)

30
28

5000

10000
15000
Time (sec)

20000

36

Simulations
Measurements

34

(b)

32
30
28
26

Simulations
Measurements

34
32
30
28
26

24

24
0

5000

(c)

10000
15000
Time (sec)

20000

5000

(d)

36

10000
15000
Time (sec)

20000

36
Simulations
Measurements

34

Temperature (C)

Temperature (C)

32

24

20000

Temperature (C)

Temperature (C)

36

32
30
28

Simulations
Measurements

34
32
30
28
26

26

(e)

Simulations
Measurements

34

26

24

24

59

24
0

5000

10000
15000
Time (sec)

20000

0
(f)

5000

10000
15000
Time (sec)

20000

Fig. 7. Comparison of the temperature histories, simulated, experimental: (a) Locations 1; (b) Locations 2; (c) Locations 3; (d) Locations 4; (e) Locations
5; and (f) Locations 6.

As shown in Fig. 8, the temperatures of the two parts

are not continuous and there are temperature jumps
at the bearing and shaft joints and also at the bearing
and bearing support joints. This is due to the thermal
contact resistance at the joints. As the heat ow has been
dammed up by the resistance, the temperature of the
parts in which heat is generated is higher than those of the
other parts.
Fig. 9 is a comparison of temperature histories
measured, the calculated either with thermal contact
resistance and the change of the heat generation
considered or without thermal contact resistance and
the change of the heat generation considered. As shown
in Fig. 9, the temperature without considering the
thermal contact resistance and the change of the heat
generation power is about 1.5 1C higher than the measured
values.

4.2. Thermal displacement analysis

Fig. 10 shows the thermal displacements of the grinding
machine housing, which takes the ANSYS coordinate
system as datum and are magnied 1000 times, after heat
balancing. For clarity, the covers on the shaft are not
shown in the gure. The black lines are the original shape
of the housing. As shown in Fig. 5, the temperature at the
left of the housing is much higher than that at the right.
The displacement at the left of the housing is larger than
that at the right, and the maximum displacement appears
at the left top corner of the coping, the X-, Y- and
Z-component displacements of which are 14.776, 51.982,
and 6.962e3 mm, respectively. The fact that the total heat
generation power of the left bearings is much larger than
that of the right bearings is the reason for the thermal
displacement distribution.

ARTICLE IN PRESS
60

X. Min et al. / International Journal of Machine Tools & Manufacture 47 (2007) 5362

Fig. 8. Simulated temperature distribution of the grinding machine housing.

36

Temperature (C)

34
32
30
With contact resistance
Without contact resistance
Measurements

28
26
24
0

5000

10000
Time (sec)

15000

20000

Fig. 9. Comparison of temperature histories among the three results of location 2.

To investigate the effect of the thermal expansion on the

machining precision, the displacements of six points along
the shaft are listed in Table 5. The locations of the six
points are shown in Fig. 2. Point A and F are at the mid of
the two double-row short cylindrical roller bearings
respectively. B, C, D and E are between A and F, sharing
the bearing, span approximately.
The results of both models, one considering the thermal
contact resistance and the change of the heat power, and

the other neglecting them, are listed in Table 5. As the

thermal contact resistance and the variation of the heat
generation power are been considered, the values of
displacements in result I are much less than the corresponding ones in result II.
From Table 5, it can be seen that the X- and
Y-component displacements of the six points are signicant
mainly due to the displacement of the housing, meaning
that the X- and Y-component displacements of the

ARTICLE IN PRESS
X. Min et al. / International Journal of Machine Tools & Manufacture 47 (2007) 5362

61

Fig. 10. The thermal displacement of the grinding machine housing: (a) front view; (b) side view.

abrasion wheel are considerable. These displacements

may affect the machining precision seriously. At the
same time, the Z-component displacements of the six
points are also large, as the shaft will expand along the
bearing house for the intrinsic structure design when
the thermal expansion occurs, but they have less effect on
the machining precision.

5. Conclusions
In this paper, a thermal model based on the Bossmanns
and Tus model [1,2] has been developed to characterize the
heat distribution of a grinding machine housing, in
particular the change of the heat generation power and
the thermal contact resistance at the solid joints based on

ARTICLE IN PRESS
62

X. Min et al. / International Journal of Machine Tools & Manufacture 47 (2007) 5362

Table 5
Displacements of six points (  103 mm)
Point
Result I: Considering the thermal contact
resistance and the change of the heat
power

Result II: Without considering the

thermal contact resistance and the change
of the heat power

X-component
displacement
Y-component
displacement
Z-component
displacement
Compound displacement
of X- and YTotal displacement
X-component
displacement
Y-component
displacement
Z-component
displacement
Compound displacement
of X- and YTotal displacement

fractal model. Compared with experimental results, it has

been shown that the model has much better accuracy than
those without taking the heat generation change and the
thermal resistance into account. Although this model was
developed for analysis of a given grinding machine
housing, it can be used for thermal analysis of various
machining systems without loss of generality, and the
equations of the thermal contact resistance are suitable for
all joints.

6.0793

5.2543

4.0029

2.5072

1.6402

0.9971

19.2

17.724

14.654

11.917

11.4

9.1554

2.785

13.417

19.574

20.065

22.095

32.086

20.139

18.486

15.191

12.178

11.517

9.21

20.331

22.842

24.777

23.471

24.916

33.381

9.2974

8.1949

6.0795

3.8323

2.5264

1.2326

29.577

27.409

22.698

18.472

17.441

13.766

0.3385

12.74

19.833

20.45

22.734

33.563

31.004

28.608

23.498

18.865

17.623

13.821

31.006

31.316

30.749

27.823

28.765

36.297

[5]

[6]
[7]

[8]

Acknowledgements
The authors gratefully wish to acknowledge the
supports of National Science Foundation through
grant No.50475073 and Jiangsu Province Science and
Technology Plan through grant BK2002059, BE2003071,
BE2004025, BA2005015.

[9]

[10]
[11]

References
[12]
[1] B. Bossmanns, J.F. Tu, A power ow model for high speed motorized
spindlesheat generation characterization, Transactions of the
ASME, Journal of Manufacturing Science and Engineering 123
(2001) 494505.
[2] B. Bossmanns, J.F. Tu, A thermal model for high speed motorized
spindles, International Journal of Machine Tools & Manufacture 39
(1999) 13451366.
[3] C.-W. Lin, J.F. Tu, J. Kamman, An integrated thermo-mechanicaldynamic model to characterize motorized machine tool spindles
during very high speed rotation, International Journal of Machine
Tools & Manufacture 43 (2003) 10351050.
[4] H. Li, Y.C. Shin, Integrated dynamic thermo-mechanical modeling of
high speed spindles, Part1: model development, transactions of the

[13]

[14]
[15]

[16]

ASME, Journal of Manufacturing Science and Engineering 126

(2004) 148158.
H. Li, Y.C. Shin, Analysis of bearing conguration effects on high
speed spindles using an integrated dynamic thermo-mechanical
spindle model, International Journal of Machine Tools & Manufacture 44 (2004) 347364.
J.P. Holman, Heat transfer, ninth ed, McGraw-Hill, New York,
2002.
Z. Honglin, H. Yumei, X. Jielan, Experiment research on thermal
contact resistance of normal used joints, Journal of Xian University
of Technology 15 (3) (1999) 2629.
G. Weilan, Y. Yansheng, The Inuence of Temperature on Contact
Resistance of Metallic Surfaces Temperature, Journal of Nanjing
University of Aeronautics & Astronautics 26 (3) (1994) 342350.
Z. Ming, A Study on Thermal Contact Resistance and Temperature
Field of Two Layered Complex Mediums, Department of Thermal
Science and Energy Engineering, Hefei, China (2001)
A. Majumdar, C.L. Tien, Fractal network model for contact
conductance, Journal of Heat Transfer, ASME 113 (1991) 516525.
S. Wang, K. Komvopoulos, A fractal theory of the interfacial
temperature distribution in the slow sliding regime: part multiple
domains, Elastoplastic Contacts and Applications, Journal of
thermal biology, ASME 116 (1994) 824831.
S. Wang, Real contact area of fractal-regular surfaces and its
implications in the law of friction, Journal of Tribology, ASME 126
(2004) 18.
A. Majumdar, B. Bhushan, Fractal Model of ElasticPlastic Contact
Between Rough Surfaces, Journal of Tribology ASME 113 (1991)
111.
T.A. Harris, Rolling Bearing Analysis. Wiley Sons, New York, 1991,
pp. 540560
J.K. Choi, D.G. Lee, Thermal characteristics of the spindle bearing
system with a gear located on the bearing span, International Journal
of Machine Tools & Manufacture 38 (1998) 10171030.
X. Yanzhong, J. Shuyun, Dynamic characteristics of high speed
angular contact ceramic ball bearing, Journal of Southeast University
20 (2004) 319323 (English Edition).