Академический Документы
Профессиональный Документы
Культура Документы
State Key Laboratory of Digital Manufacturing Equipment and Technology, School of Mechanical Science and Engineering, Huazhong University of Science and Technology,
Wuhan 430074, PR China
b
School of Naval Architecture and Ocean Engineering, Huazhong University of Science and Technology, Wuhan 430074, PR China
a r t i c l e in f o
a b s t r a c t
Article history:
Received 8 May 2009
Received in revised form
29 October 2009
Accepted 30 October 2009
Available online 10 November 2009
A universal dynamic model of xed joints is built through considering the relative motion between the
sub-structures of the xed joints and the coupling among various degrees of freedom. The dynamic
model may accurately reect the dynamic characteristics of the joints. Based on the inverse relationship
between the frequency response function matrix and the dynamic stiffness matrix of a Multi-Degree-OfFreedom system, a high-accuracy parameter identication method is proposed to recognize the dynamic
model parameters of the joints using the dynamic test data of the whole structure including the joints.
The error between the theoretical and experimental results of the model is less than 10%, while the error
of the Yoshimura model is three times bigger than that of the model. The effectiveness and accuracy of
the dynamic model and its parameter identication have been validated. The establishment of the model
will provide a theoretical foundation for the precisely dynamic modeling of the CNC Machine Tool.
Crown Copyright & 2009 Published by Elsevier Ltd. All rights reserved.
Keywords:
Fixed joints
Dynamic modeling
Modal analysis
Parameter identication
Finite element model
1. Introduction
On the nite element analysis of machine tools, it is inevitable
to set up accurate dynamic models and reduce the huge amount
of calculation. Giving that, the machine tools are mostly
simplied as lumped mass models or distributed mass models.
In the 1960s, Taylor and Tobias developed a lumped mass model
of the rocker drilling machine [1]. Although the calculated loworder natural frequencies of the whole machine tool were close to
the experimental values, its dynamic response could not be
calculated because of not considering the damping feature of the
joints and the system. Hijink built the computation model of the
horizontal lift milling machine by means of the distributed mass
beam [2]. According to the effect of structural elastic deformation
on the overall structure, the beam was divided into a exible and
a rigid part, of which a rigid beam was assumed to only have rigid
body movement without elastic deformation. So the characteristics of the actual structure of machine tools might be simulated
more precisely. However, its resonance frequency had still about
15 percent deviation from the experimental value and its dynamic
exibility even deviated more than doubled. It is because the
dynamic characteristics of the joints were not considered in the
process. By thinking about the features of joints, Yoshihara set up
nn
0890-6955/$ - see front matter Crown Copyright & 2009 Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijmachtools.2009.10.017
ARTICLE IN PRESS
K. Mao et al. / International Journal of Machine Tools & Manufacture 50 (2010) 156164
157
ARTICLE IN PRESS
K. Mao et al. / International Journal of Machine Tools & Manufacture 50 (2010) 156164
q1
158
p
Fig. 3. (a) Beam-Column Connection, (b) Slide Block-Slider Connection and (c) Beam-Rail Connection.
Fig. 4. (a)Linear Connection joint nite element and (b)Array Connection joint
nite element.
ij
K1n
x1n x5n
n1
3
X
ij
K2n
x2n x6n
n1
3
X
ij
K4n
x4n
x8n fij
3
X
ij
K3n
x3n x7n
n1
n1
ij
Where Kmn
is the stiffness inuential coefcient, i,m= 1,2,3,4
represent the node and j,n =1,2,3 represent the direction. The
ij
is the corresponding necessary force
physical meaning of Kmn
ARTICLE IN PRESS
K. Mao et al. / International Journal of Machine Tools & Manufacture 50 (2010) 156164
159
where
fXg x11 ; x12 ; x13 ; ; x81 ; x82 ; x83
f1j f5j ;
f2j f6j ;
j 1; 2; 3:
fFg f11 ; f12 ; f13 ; ; f81 ; f82 ; f83
K
K0
K0
K0
K0
2424
where
fYg x11 x51 ; x12 x52 ; x13 x53 ; ; x41 x81 ; x42 x82 ;
x43 x83
fQ g f11 ; f12 ; f13 ; ; f41 ; f42 ; f43 :
Furthermore, Eq. (3) can be expressed as
Fig. 6. Structure of test specimen
SfQ g fYg
ARTICLE IN PRESS
160
K. Mao et al. / International Journal of Machine Tools & Manufacture 50 (2010) 156164
That is
2
h11 o
6
6 h21 o
6
6 ^
4
hn1 o
h12 o
h22 o
hn2 o
3
32
3 2
x 1 o
f1 o
h1n o
7
76
7 6
h2n o 76 f2 o 7 6 x2 o 7
7
76
76
7
7
6
6
^ 54 ^ 5 4 ^ 7
5
xn o
hnn o
fn o
10
Fig. 8. (a1) Theoretical result of rolling and (b1) Experimental result of rolling,
(a2) Theoretical result of yawing and (b2) Experimental result of yawing,
(a3) Theoretical result of pitching and (b3) Experimental result of pitching,
(a4) Theoretical result of left-right translation and (b4) Experimental result of
left-right translation, (a5) Theoretical result of front-back translation and
(b5) Experimental result of front-back translation, (a6) Theoretical result of
up-down translation and (b6) Experimental result of up-down translation.
ARTICLE IN PRESS
K. Mao et al. / International Journal of Machine Tools & Manufacture 50 (2010) 156164
2:5e 9
6
6 5:0e 8
6
6 1:0e 9
6
6
6 1:5e 9
6
6 1:0e 8
6
6
6 2:0e 8
K0 6
6 8:6e 8
6
6
6 5:0e 8
6
6 1:0e 8
6
6
6 1:1e 9
6
6 1:0e 8
4
5:0e 8
5:0e 8
1:5e 9
161
1:0e 9
5:0e 8
1:5e 9
1:0e 8
1:0e 8
3:3e 7
2:0e 8
2:5e 8
8:6e 8
5:0e 8
5:0e 8
3:6e 8
1:0e 8
5:0e 7
1:1e 9
1:0e 8
1:0e 8
6:6e 7
1:0e 8
5:0e 8
2:5e 9
2:0e 8
2:5e 8
4:6e 8
1:0e 8
5:0e 7
1:3e 8
5:0e 8
1:0e 8
2:0e 8
2:5e 9
5:0e 8
1:0e 9
1:1e 9
1:0e 8
5:0e 8
8:6e 8
5:0e 8
3:3e 7
2:5e 8
5:0e 8
1:5e 9
5:0e 8
1:0e 8
6:6e 7
1:0e 8
5:0e 8
3:6e 8
2:5e 8
5:0e 8
4:6e 8
1:0e 8
1:0e 9
1:0e 9
5:0e 8
1:0e 8
2:5e 9
5:0e 8
5:0e 8
2:5e 9
1:0e 8
5:0e 8
1:1e 9
1:0e 9
1:0e 8
1:5e 9
5:0e 7
1:0e 8
3:6e 8
5:0e 7
1:0e 8
6:6e 7
1:0e 8
5:0e 8
1:5e 9
5:0e 8
1:0e 8
3:3e 7
5:0e 7
1:3e 8
5:0e 8
1:0e 8
1:1e 9
1:0e 9
5:0e 8
2:5e 9
2:0e 8
2:5e 8
1:0e 8
5:0e 8
8:6e 8
5:0e 8
1:0e 8
1:5e 9
1:0e 8
2:0e 8
2:5e 9
5:0e 8
6:6e 7
1:0e 8
1:0e 8
1:1e 9
5:0e 8
1:0e 8
3:6e 8
5:0e 7
5:0e 7
1:3e 8
1:0e 8
2:0e 8
3:3e 7
2:5e 8
2:5e 8
4:6e 8
5:0e 8
1:0e 9
1:5e 9
5:0e 8
joints in Fig. 6 is not fully excited into the modal shapes that
characterize the relative movement of the joints. However, the
test specimen used in the form of the structural joint in Fig. 5 can
be easily excited into the modal shapes that characterize the
relative movement of joints. Therefore, the form of the structural
joint in Fig. 5 is adopted to implement the dynamic experiment of
the xed joints.
4.2. Dynamic experiment setup
As shown in Figs. 7(a) and (b), a dynamic testing device is set
up, which includes the LMS Test.lab vibration testing and analysis
system. A modal testing was performed on a test specimen. The
test specimen is composed of two sub-structures with up-andbottom blocks, which are connected as a whole by two screws.
The specimen material is 45 # steel. The dimension of each substructure is 190 mm 190 mm 110 mm. The size of the joint
surface is 60 mm 31.5 mm. The boss height is 29 mm and the
bolt pre-tightening torque is 45 Nm.
In the process of the modal testing, the used impact hammer
is 086C04 piezoelectric impact hammer produced by U.S. PCB.
Its head is a kind of nylon materials. The testing accelerometer
3
5:0e 8
7
1:0e 8 7
7
1:1e 9 7
7
7
1:0e 8 7
7
5:0e 7 7
7
7
1:3e 8 7
7
2:0e 8 7
7
7
2:5e 8 7
7
4:6e 8 7
7
7
1:0e 9 7
7
5:0e 8 7
5
2:5e 9
T
45
14062:5N
0:2d
0:2 0:016
12
Pn
F0
14062:5
7:4404762 MPa
pq
0:06 0:0315
13
Table 1
The rst six-order modal shapes and modal frequencies.
Mode shape
Theoretical result
from the model
in this paper (Hz)
Yoshimura model in
literature[10] (Hz)
Bolts-connected spring-damper
model in literature[19] (Hz)
1
2
3
4
5
6
409
471
871
1673
2130
2568
409
451
853
1765
2278
2647
110
385
755
391
394
1852
87
456
886
309
310
2430
yawing
rolling
pitching
left-right translation
front-back translation
up-down translation
ARTICLE IN PRESS
162
K. Mao et al. / International Journal of Machine Tools & Manufacture 50 (2010) 156164
l aPnm
14
dPn
1 1m
:
P
dl
am n
15
For the test specimen in Fig. 7(b), it is dened that a= 0.65 and
m =0.5. So the normal stiffness Kn is equal to 4 1.929e9 N/m.
The tangential stiffness of the joint is
Kt kt s
16
5:7e 9
6
6 1:3e 9
6
6 2:0e 9
6
6
6 3:8e 9
6
6 2:5e 8
6
6
6 4:0e 8
K0 6
6 2:2e 9
6
6
6 1:3e 9
6
6 2:0e 8
6
6
6 2:3e 9
6
6 2:5e 8
4
1:0e 9
(1) The dynamic model of the long beam structure with the
beam-rail connection
The Beam-Rail Connection in the MC2000 Plano machining
center produced by Beijing Machinery and Electricity Institute
is implemented by using bolts. The width and height of the
rail is respectively 63 mm and 58 mm, the center-to-center
distance between the bolts is 120 mm, and the bolted pretightening torque is 90 Nm. Thus, a joint specimen similar to
Fig. 7(b) is created. The size of its joint surface is 120 mm 63
mm, the dimension of its sub-structure is 300 mm 300
mm 120 mm, and the boss height is 50 mm.
A modal testing similar to Fig. 7(a) is executed on the
designed joint specimen. So the FRFs characteristics of the
structure including joint of the specimen are obtained. And
then, the stiffness matrix is identied as follows:
1:3e 9
2:0e 9
3:8e 9
2:5e 8
4:0e 8
2:2e 9
1:3e 9
2:0e 8
2:3e 9
2:5e 8
3:4e 9
1:1e 9
1:1e 9
4:8e 9
2:5e 8
4:0e 8
2:1e 8
5:3e 8
5:3e 8
4:7e 8
1:3e 9
2:0e 8
1:0e 9
1:1e 8
1:1e 8
2:6e 7
2:5e 8
1:0e 9
1:4e 8
2:1e 8
2:5e 8
4:0e 8
5:7e 9
1:3e 9
2:0e 9
2:3e 9
2:5e 8
1:0e 9
2:2e 9
1:3e 9
2:1e 8
5:3e 8
1:3e 9
3:4e 9
1:1e 9
2:5e 8
1:4e 8
2:1e 8
1:3e 9
1:0e 9
5:3e 8
4:7e 8
2:0e 9
1:1e 9
4:8e 9
1:0e 9
2:1e 8
1:9e 9
2:0e 8
1:1e 8
1:3e 9
1:0e 9
2:0e 8
1:1e 8
2:3e 9
2:5e 8
2:5e 8
1:4e 8
1:0e 9
2:1e 8
5:7e 9
1:3e 9
1:3e 9
3:4e 9
2:0e 9
1:1e 9
3:8e 9
2:5e 8
2:5e 8
2:1e 8
1:1e 8
2:6e 7
1:0e 9
2:1e 8
1:9e 9
2:0e 9
1:1e 9
4:8e 9
4:0e 8
5:3e 8
2:5e 8
1:0e 9
2:2e 9
1:3e 9
2:0e 8
3:8e 9
2:5e 8
4:0e 8
5:7e 9
1:3e 9
1:4e 8
2:1e 8
1:3e 9
1:0e 9
1:1e 8
2:5e 8
2:1e 8
5:3e 8
1:3e 9
3:4e 9
2:1e 8
1:9e 9
2:0e 8
1:1e 8
2:6e 7
4:0e 8
5:3e 8
4:7e 8
2:0e 9
1:1e 9
Table 2
Modal frequencies and errors of the joint of the long beam specimen.
Mode shape
Experimental
result (Hz)
Theoretical
result (Hz)
Error
1
2
3
4
5
394.287
409.289
783.722
1403.532
1681.608
405
409.3
777.8
1463.4
1626.5
2.717%
0%
0.75%
4.266%
3.277%
1758.438
1872.7
6.498%
rolling
yawing
pitching
left-right translation
front-back
translation
6 up-down translation
1:0e 9
7
7
7
7
7
7
2:0e 8 7
7
1:1e 8 7
7
7
2:6e 7 7
7
4:0e 8 7
7
7
5:3e 8 7
7
4:7e 8 7
7
7
2:0e 9 7
7
1:1e 9 7
5
4:8e 9
2:1e 8
1:9e 9
ARTICLE IN PRESS
K. Mao et al. / International Journal of Machine Tools & Manufacture 50 (2010) 156164
Table 3
Modal freqencies and errors of long beam specimen.
Modal shape
Experimental
result (Hz)
Theoretical
result (Hz)
Error
X rst-order bending
Y rst-order bending
X second-order
bending
X third-order bending
148
243
402
147
267
406
0.7%
9.8%
1.0%
774
798
3.1%
163
Fig. 10. (a1) Theoretical result of X rst-order bending and (b1) Experimental result of X rst-order bending, (a2) Theoretical result of Y rst-order bending and (b2)
Experimental result of Y rst-order bending, (a3) Theoretical result of X second-order bending and (b3) Experimental result of X second-order bending, (a4) Theoretical
result of X third-order bending and (b4) Experimental result of X third-order bending.
ARTICLE IN PRESS
164
K. Mao et al. / International Journal of Machine Tools & Manufacture 50 (2010) 156164
(1) The nite element model of the xed joints is built through
considering the relative motion between the sub-structures
composed of the joints. Compared with other current researches
of the joints, it is found that the dynamic model proposed in this
paper can reect the dynamic characteristics of the joints more
accurately. The reason is that it has considered about the coupling
among various degrees of freedom.
(2) A high-precision parameter identication method for the
dynamic model of the xed joints is introduced. The
advantage of the parameter identication method is that
the dynamic test data of the whole structure including the
joint can be applied to identify the model parameters, which
improve the accuracy of parameter identication.
(3) The effectiveness and accuracy of the dynamic modeling and the
parameter identication method proposed in this paper have
been veried by experiments. The error between the theoretical
and experimental results of the our model is less than 10%, while
the error of the Yoshimura model often used in the existing
literature is three times bigger than our model.
Acknowledgements
This work was supported by National Basic Research Program
of China, Grant No. 2005CB724101 and National Natural Science
Foundation of China, Grant No. 50975104. The authors are
grateful to other participants of the projects for their cooperation.
References
[1] S. Taylor, S.A. Tobias, Lumped-constants method for the prediction of the
vibration characteristics of machine tool structures, in: Proceedings of fth
International MTDR Conference, Pergamon, 1964 3742.
[2] J.A.W. Hijink, A.C.H. Van Der Wolf, Analysis of a milling machine: computed
results versus experimental data, in: Proceedings of 14th International MTDR
Conference, Manchester, UK/Ed. J.M. Alexander (1973) 553558.
[3] M. Yoshihara, Computer-aided design improvement of machine tool states
incorporating joint dynamics data, Annals of the CIRP 28 (1) (1979) 241246.
[4] M. Dalenbring, Damping function estimation based on measured vibration
frequency responses and nite-element displacement modes, Mechanical
Systems and Signal Processing 13 (1999) 547569.
[5] T. Inamura, T. Sata, Stiffness and damping identication of the elements of a
machine tool structure, CIRP Annuals 28 (1979) 235239.
[6] J.A. Greenwood, J.H. Tripp, The elastic contact of rough sphere, ASME Journal
of Applied Mechanics 34 (1967) 153159.
[7] J.A. Greenwood, J.P.B. Williamson, Contact of nominally at surface,
Proceedings of the Royal Society of London. Series A, Mathematical and
Physical Sciences 295 (1442) (1966) 300319.
[8] E.F. Crawley, A.C. Aubert, Identication of nonlinear structural element by
force-state mapping, AIAA Journal 24 (1) (1986) 155162.
[9] Xi Shi, Measurement and modeling of normal contact stiffness and contact
damping at the meso scale, Journal of Vibration and Acoustics 127 (2005) 5260.
[10] M. Yoshimura, Making use of CAD technology based on the dynamic
characteristics data of joints to improve the structural rigidity of machine
tools, Machine Tools 1 (1979) 142146.
[11] Z.F. Tong, J. Zhang, Research on the dynamic characteristic and its
identication of the joint between column and bed of a machining center,
Journal of Vibration and Shock 43 (3) (1992) 1319.
[12] C.F. Beads, Damping in structural joints, The Shock and Vibration Digest 6
(1982) 563570.
[13] J. Tlusty, F. Ismail, Dynamic structural identication task and methods,
Annals of CIRP 29 (1980).
[14] J.H. Wang, C.M. Liou, Identication of parameters of structural joints by use of
noise-contaminated FRFs, Journal of Sound and Vibration 142 (2) (1990) 261277.
[15] Y. Ren, C.F. Beards, Identication of effective linear joints using coupling and
joint identication techniques, Journal of Vibration and Acoustics 120 (2)
(1998) 331338.
[16] J.X. Yuan, X.M. Wu, Identication of the joint structural parameters of machine
tool by DDS and FEM, Journal of Engineering for Industry 107 (1985) 6469.
[17] Y.M. Huang, W.P. Fu, L.X. Dong, Research on the dynamic normal
characteristic parameters of joint surface, Chinese Journal of Mechanical
Engineering 29 (3) (1993) 7478.
[18] Y.M. Huang, W.P. Fu, J.X. Tong, A method of acquiring applied tangential
damping parameters of joint surfaces, Journal of Xian University of
Technology 12 (1) (1996) 15.
[19] B.Y. Liao, X.M. Zhou, Z.H. Yin, in: Modern mechanical dynamics and its
application: modeling, analysis, simulation, modication, control, optimization, China Machine Press, Beijing, 2004.