Académique Documents
Professionnel Documents
Culture Documents
Abstract: This article describes an investigation into the contact behaviour of polymeric gear
transmissions using numerical finite element (FE) and analytical techniques. A polymer gear
pair was modelled and analysed using the ABAQUS software suite and the analytical results were
calculated using the BS ISO 6336 rating standard. Before describing the results, the principles
of the strategies and methods employed in the building of the FE model have been discussed.
The FE model dynamically simulated a range of operating conditions. The simulations showed
that the kinematic behaviour of polymeric gears is substantially different from those predicted
by the classical metal gear theory. Extensions to the path of contact occur at the beginning and
end of the meshing cycle. These are caused by large tooth deflections experienced by polymer
gear teeth, as a result of much lower values of stiffness compared to metallic gears. The premature contact (occurring at the beginning of the meshing cycle) is hypothesized to be a factor in
pitch line tooth fractures, whereas the extended contact is thought to be a factor in the extreme
wear as seen in experiments. Furthermore, the increase in the path of contact also affects the
induced bending and contact stresses. Simulated values are compared against those predicted
by the international gear standard BS ISO 6336 and are shown to be substantially different. This
is particularly for the case for bending stresses, where analytically derived values are independent of contact stiffness. The extreme tooth bending and the differences between analytical and
numerical stresses observed in all the simulations suggest that any future polymeric gear-rating
standard must account for the effects of load sharing (as a result of tooth deflection) and friction
(particularly in dry-running applications).
Keywords: spur gears, polymers, steels, friction, temperature, kinematics
1
INTRODUCTION
Mechanical Engineering, The University of Birmingham, Edgbaston, Birmingham B15 2TT, UK.
email: k.d.dearn@bham.ac.uk
JMDA315
102
2 THEORY
The complexities of polymer gearing offer an ideal
application for the finite-element method. Many
researchers have used it to study the fundamental
kinematic and kinetic behaviours of polymer gears. In
recent years, as computational power has increased,
along with the sophistication of commercial finiteelement software, researchers have employed numerical techniques not only to study the behaviour of
polymeric gear transmissions, but also to modify and
optimize gear trains for specific applications. However,
these polymer gear simulations are all based on quasistatic solutions (as detailed above). The following
section discusses current rating methods for nonmetallic gears and discusses those that are deemed
most appropriate to the unique behaviour of plastic
gears. The development of a dynamic non-linear FE
model to study the kinematic behaviour of a polymer gear transmission is then described. Employing
a dynamic solution allowed the whole meshing cycle
to be continuously simulated, resulting in a better
understanding of tooth bending effects, and the ramifications of these effects for other aspects of polymer
gear performance. In addition to this, it provided an
answer as to whether gear-rating standards, developed
specifically for metallic gears, could be used to design
and rate polymer gears.
2.1
2.2
103
Contact stresses
The basis of all contact stress calculations is the classical Hertzian contact analysis. Stress levels are usually
calculated at the pitch point where the sliding friction may be assumed to be zero as well as negligible
bending deflection due to contact. The only difference between the standards is the load-sharing effect.
ANSI/AGMA 2001 assumes no load sharing, whereas
ISO 6336 does consider a load-sharing factor in the
calculation of contact stress. When considering polymeric materials, load sharing cannot be disregarded
and hence ISO 6336:1996 would seem to be the most
accurate available method. When comparisons are
made against a standard in this article, they are made
against ISO 6336:1996 method B (from this point
forward referred to as the Standard).
3
Fig. 1
to the accuracy of the solution. The size of the elements was determined based on the simulation of
metallic gear teeth from reference [12]. As the polymer
material used in the simulations has a lower stiffness than the material used by Wang and Howard, the
field variable gradients are much smaller than those of
metallic gears. Hence, the same mesh density with a
structured mesh throughout the tooth has been used.
This will be conservative in terms of mesh refinement
but should guarantee accurate results. The rest of the
teeth were modelled with a coarse mesh to increase
computational efficiency.
As the stress elements do not have rotational degrees
of freedom, an advanced surface-based tie technique
was utilized to attach the nodes at the gear hub
to a rigid-body shaft, later used to apply rotational
loads and boundary conditions (also shown in Fig. 2).
In order to minimize fluctuations between contacting surfaces, an inertia load was assigned to the
shaft to dampen unwanted vibrations. The elements
used (CPE4R) were four-node bilinear plain strain
quadrilaterals with reduced integration [13].
Solutions were obtained in three steps, namely
approach, loading, and rotation, selected to reduce
the occurrence of contact inaccuracies and vibration.
A smooth step profile was defined as custom amplitude to apply all the boundary conditions and loads
as smoothly as possible. Boundary conditions and
loads were applied at the drive shaft reference points
(centres), according to the solution step sequence.
Four separate surfaces were defined on each meshing gear, to isolate each zone of the meshing cycle. The
boundary conditions were selected as follows.
Proc. IMechE Vol. 224 Part L: J. Materials: Design and Applications
104
Fig. 2
(a) Loads and boundary conditions and (b) generated mesh representing the gear
Table 1
Simulation schedule
E (GPa) T (N m)
Test 1
Test 2
Test 3
Test 4
Test 5
3.1
3.1
3.1
3.1
2.5
5
8
11
15
8
0
0
0
0
0
Test 6 2
Test 7 1
Test 8 3.1
Test 9 3.1
8
8
0.3
0.5
Notes
polymer manufacturer Du Pont [15]. For each simulation, Poissons ratio was taken as 0.35 and the material
model used was isotropic linear elastic.
4
RESULTS
Fig. 4
Fig. 3
4.1
4.1.1
Kinematics
Path of contact
105
Load sharing
The fraction of the applied load transmitted by individual gear teeth was governed by the theoretical contact
ratio, which, in case of the benchmark geometry, was
1.65. Physically, this means that for approximately
two-thirds of the meshing cycle, two pairs of teeth
Proc. IMechE Vol. 224 Part L: J. Materials: Design and Applications
106
Fig. 5 The load share ratio against the roll angle predicted by simulation 2
Stress analysis
Fig. 6
Fig. 7
Bending stresses at various nodes against the roll angle predicted by simulation 2
Fig. 8
JMDA315
107
A similar approach was adopted to establish the position of maximum contact stress across the tooth flank.
Figure 8 shows the corresponding stress profile plotted against the roll angle along with the nodes along
the tooth flank. Once again, the maximum stress value
occurred at the pitch point. A maximum contact stress
value of 38.69 MPa was predicted. This is in close
agreement with the Standard that predicts a value of
Contact stresses at various nodes against the roll angle predicted by simulation 2
Proc. IMechE Vol. 224 Part L: J. Materials: Design and Applications
108
Fig. 9
Table 2
Test 1
Test 2
Test 3
Test 4
109
Torque
(N m)
FE
simulation
BS ISO
6336
FE
simulation
BS ISO
6336
5
7
10
15
31.57
38.70
44.20
48.75
29.71
37.58
44.06
49.71
17.13
25.94
34.36
42.31
23.80
38.08
52.36
66.64
Fig. 10
JMDA315
Fig. 11
The effect of temperature in this case is taken to represent the reduction in the material stiffness because
of elevated operating temperatures. This is a simplification of the physical manifestation of high gear
temperatures; however, in terms of the kinematic
behaviour of the gear teeth, the reduction in modulus
has a strong influence. Table 3 summarizes the reduction in stiffness with temperature rise. These values are
assigned to the materials specified in simulations 2, 5,
110
Table 3
Test 2
Test 5
Test 6
Test 7
Torque
(N m)
Equivalent
gear body
temperature
( C)
3.1
2.5
2
1
7
7
7
7
23
35
50
90
Fig. 12
Fig. 13
Comparison of contact and bending stresses versus Youngs modulus, at the pitch point
JMDA315
Fig. 14
Fig. 15
Friction
111
112
ratio and hence increases transmitted loads and subsequent contact stresses. Friction does not seem to
affect the kinematic behaviour of the gears and has
a negligible effect on the length of the path of contact.
Similar trends, shown in Fig. 15, are observed when
FE and the Standard derived stresses are compared
against increasing coefficients of friction. As expected,
Standard derived stresses show no variation with friction. The FE simulations increasingly overestimate
contact stresses compared with those calculated from
the Standard. FE contact stresses show a strong dependence on friction. Bending stresses are also shown to
increase with friction but remain below stress levels
predicted by the Standard.
Therefore, for the Standard to become a more accurate means of specifying polymeric gears, a frictional
factor should be incorporated into the rating equations to account for tangential forces, particularly in
dry-running applications. The dominance of contact
stresses (being greater than the bending stresses) may
also contribute to wear being the predominant failure
mechanism in dry-running polymeric gears.
5.
6.
7.
8.
9.
10.
5
CONCLUSIONS
The above conclusions point to the need for a specific polymer gear Standard that accounts for the
idiosyncrasies that are not based on metallic gearrating methods. The authors have evidence to show
that the lack of applicable design data and a rating
Standard, in the public domain, is preventing this
novel form of gearing from being fully exploited.
Authors 2010
REFERENCES
1 Klein Meuleman, P.,Walton, D., Dearn, K. D.,Weale, D. J.,
and Driessen, I. Minimization of transmission errors in
highly loaded plastic gear trains. Proc. IMechE, Part C:
J. Mechanical Engineering Science, 2007, 221(C9), 1117
1129. DOI: 10.1243/09544062JMES439.
2 Walton, D., Tessema, A. A., Hooke, C. J., and Shippen,
J. M. Load sharing in metallic and non-metallic gears.
Proc. IMechE, Part C: J. Mechanical Engineering Science,
1994, 208(C2), 8187. DOI: 10.1243/PIME_PROC_1994_
208_104_02.
3 Walton, D., Tessema, A. A., Hooke, C. J., and Shippen, J.
M. A note on tip relief and backlash allowances in nonmetallic gears. Proc. IMechE, Part C: J. Mechanical Engineering Science, 1995, 209(C6), 383388. DOI: 10.1243/
PIME_PROC_1995_209_169_02.
4 White, J.,Walton, D., and Weale, D. J. The beneficial effect
of tip relief on plastic spur gears. In Proceedings of the
Conference at ANTEC98, Society of Plastics Engineers,
Atlanta, USA, April 1998, vol. III, pp. 30133017.
5 Senthilvelan, S. and Gnanamoorthy, R. Effect of gear
tooth fillet radius on the performance of injection
JMDA315
9
10
11
12
13
14
15
16
17
18
19
20
APPENDIX 1
113
Subscripts
1
2
pinion
wheel
Notation
E
ra
T
coefficient of friction
theoretical pressure angle (degree)
JMDA315
Pbt = mt cos t
Length of path of contact
1
2
2
2
2
g =
da1 db1 da1 db1 a sin wt
2
(positive sign for external gears)
Proc. IMechE Vol. 224 Part L: J. Materials: Design and Applications
114
Calculation of hFe
g
Pbt
(6)
=
A2.1.4
(7)
ZD = 1
if M2 1
ZB = M1
if M1 > 1
Z D = M2
if M2 > 1
= zn sin
3
cos
mn
mn
(8)
A2.2
Ft
bmn
(9)
fp
hfp
fp
+x
mn
mn
(12)
(10)
0.5 + 2 tan n x
+ inv n inv en
zn
Fen = en e
hfe
cos n
= 0.5zn
cos
cos Fen
3
mn
fp
G
+ 0.5
mn
cos
tan wt
M2 =
2
2
(da2 /db2 ) 1 (2/z2 )
2
2
(da1
/db1
) 1 ( 1)(2/z1 )
if M1 1
dbn = dn cos n
dan = dn + da d n =
cos2 b
dan 2
dbn 2
2
2
z
2
den = 2
|z|
d cos cos n
dbn 2
(n 1) +
|z|
2
dbn
en = cos1
den
e =
tan wt
M1 =
2
2
(da1
/db1
) 1 (2/z1 )
2
2
(da2
/db2
) 1 ( 1)(2/z2 )
ZB = 1
d n = mn z n
(11)
E
2G
H=
zn 2
mn
3
JMDA315
Hence
=
2G
tan H
zn
A2.2.2
(13)
115
L=
Fp
f
2G 2
=
+
mn
mn
cos (zn cos2 2G)
JMDA315
(14)
SFn
,
hFe
qs =
SFn
2F