Contact stress analysis of concave conical involute gear pairs with

non-parallel axes
Yi-Cheng Chen
a
, Chia-Chang Liu
b,n
a
Department of Mechanical Engineering, National Central University, 300, Jhongda Rd., Jhong-Li 320, Taiwan, ROC
b
Department of Mechanical Engineering, Ching Yun University, 229, Chien-Hsin Rd., Jhong-Li 320, Taiwan, ROC
a r t i c l e i n f o
Article history:
Received 13 August 2009
Received in revised form
10 October 2010
Accepted 10 December 2010
Available online 12 January 2011
Keywords:
Conical involute gears
Finite element analysis
Hertzian contact stress
a b s t r a c t
This study investigates the contact stress of concave conical involute gear pairs with non-parallel axes by
means of nite element analysis. The complete mathematical models of the gear pair, including the
working tooth surfaces andthe llets have been derived basedonthe theory of gearing andthe generation
mechanism. A mesh-generation program is also developed, which gives a dense mesh in the contact
regions and a coarse mesh in the rest of the teeth. The commercial software, ABAQUS, capable of contact
analysis of two 3-D deformable bodies is applied to evaluate the stress distribution on the tooth surfaces.
Several numerical examples are presented to demonstrate the stress analysis of the gear pairs with
various gear design parameters. The results show that the contact stress of non-parallel-axes concave
conical involute gear pair can be reduced signicantly by choosing grinding wheel with smaller radius.
Therefore, concave involute gear pairs relieve the high contact stress problem of non-parallel-axes
conventional conical involute gear pairs.
& 2010 Elsevier B.V. All rights reserved.
1. Introduction
Conical involute gears, also known as beveloid gears [1], are
involute gears with tapered tooth thicknesses, roots and outside
diameters. Conical involute gear represent the most general type of
the involute gear, which can mesh conjugately with most involute
type gears of spur gears, helical gears, conical involute gears,
worms and racks to serve the motion transmission between
parallel, intersected and crossed axes in any relative position. An
outstanding feature of the conical involute gear is its ability to
backlash adjustments. Owing to its tapered tooth thickness, the
backlash of a conical involute gear pair can be easily eliminated by
axial adjustments without affecting its center distance. Theoreti-
cally, the bearing contacts of conical involute gear pairs under non-
parallel axes meshing are point contacts. The contact ellipses are
relativelysmall [2], andthe toothsurface durabilityis generallylow
owing to its high contact stress. Recently, Ohmachi et al. [3,4]
investigated the fatigue strength of an intersected-axes conical
involute gear pair experimentally. The tooth surface life has been
evaluated by the pitting area rate, while the critical value of the
circulating torque has been obtained as well.
The low-load capacity thus limits the application of this kind of
gear pairs to power transmission. To overcome this drawback,
Mitome et al. [57] proposed the idea of infeed grinding for the
concave conical involute gear generation, and indicated that the
gear pairs ground by this method have larger contact ellipses
compared with those of the conventional conical involute gear
pairs. According to the concept of infeed grinding, the author [8]
derived the mathematical models of concave conical involute gear
pairs. Tooth contact analysis (TCA) and contact ellipses simulations
are performed and the results indicated that the concave conical
involute gear pairs solve the problems associated with low-load
capacity by enlarging the contact ellipses. In this study, the contact
stress of this kindof gear pair is evaluatedbynite element analysis
(FEA) for a more detail information about the stress distribution on
the tooth surfaces.
Although the equations for stress calculations of some familiar
gears are available in gear handbooks and standards, the calcula-
tion formulae for concave conical involute gears are not available.
Fortunately, the development of FEA has made approximate
calculations of stress distribution on gear tooth surfaces with
complicated geometry possible. Due to the progress of computer
technology and the computational techniques, the FEA becomes a
popular and powerful analysis tool to determine the formation of
bearing contacts and stress distributions of gear drives. To date,
many researchers have applied FEA to study the tooth deections,
contact stresses and llet stresses of various types of gearing, such
as spur gears, helical gears, spiral bevel gears, hypoid gears and
worm gears [915]. Recently, Brauer [16] proposed the mathema-
tical descriptions and FE models of conical involute gears, in which
the intervals for the parameters and the boundary point between
the involute and llet curve have been discussed in detail. Brauer
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/nel
Finite Elements in Analysis and Design
0168-874X/$ - see front matter & 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.nel.2010.12.005
n
Corresponding author. Tel.: +886 3 4581196x5535; fax: +886 3 4683301.
E-mail addresses: ethan@ncu.edu.tw (Y.-C. Chen), ccliu@cyu.edu.tw (C.-C. Liu).
Finite Elements in Analysis and Design 47 (2011) 443452
[17] also investigated the transmission error of anti-backlash
conical involute gear by FEA, in which a globallocal FE meshing
methodwas adoptedtoyielda dense FEmeshinthe contact regions
and a coarse mesh in the rest of teeth. Otherwise, Wu and Tsai [18]
studied the contact stress of skew conical involute gear drives in
approximate line contact.
In this study, the general-purpose FEA software, ABAQUS, is
adopted to evaluate the contact stress of the proposed concave
conical involute gear pairs. Firstly the complete mathematical
models for the concave involute gear pairs, including the working
tooth surfaces and the llets, have been developed based on the
theory of gearing and the generation mechanism. Then, a computer
program for the nite element (FE) discretization of three-dimen-
sional (3-D) tooth models is developed. A very dense mesh in the
possible contact regions ensures the accuracy of FEA, while a coarse
meshinthe rest of the teethreduces the computationtimeneededto
solve the non-linear contact problems. ABAQUS, capable of contact
analysis for two 3-D deformable bodies, is employed to determine
the stress distribution of a pair of contact gear teeth. Finally, some
numerical examples are presented to demonstrate the results of FE
stress analysis. The simulation results showthat the contact stresses
of non-parallel-axes concave conical involute gear pairs decrease
signicantly when they are ground by grinding wheels with smaller
radii. The results of the FEA in this study and the results of contact
ellipse simulations in previous research works [8] are consistent.
2. Mathematical model of the concave conical involute gear
pair
2.1. Infeed grinding method
The concave conical involute gears described below are man-
ufactured based on the infeed grinding mechanism [5] shown in
Fig. 1. A corresponding imaginary rack cutter for the conventional
conical involute gear generation is presented here for reference.
The pitch plane of the corresponding imaginary rack cutter is set to
forman inclined angle d with respect to the gear axis of revolution.
The pitch circle of the gear and the pitch plane of the corresponding
imaginary rack cutter are in tangency at pitch point P
0
, and r
denotes the pitch radius of the gear. Infeed grinding is a kind of
generating grinding using a cone-type grinding wheel, and is
termed after infeed hobbing in the worm gear manufacturing.
The grinding wheel feeds along the infeed direction, which is
perpendicular to the pitch plane of the corresponding imaginary
rack cutter, and does not travel along the lengthwise direction of
the gear tooth as in spur and helical gear grindings. Notably, the
grinding wheel can be regarded as the imaginary rack cutter when
its pitch radius r
w
approaches innity (i.e. r
w
N).
Fig. 1(b) presents a view perpendicular to the pitch plane of the
corresponding imaginary rack cutter, where the grinding wheel is
set a helix angle cwith respect to the gear rotational axis to satisfy
the grinding requirements for helical concave conical involute gear.
During the grinding process, the gear blank rotates with angular
velocity o, and the grinding wheel translates along the direction of
its axis of revolution, which is perpendicular to the tooth trace
direction, with the velocity rocos c. Notably, when c01, the
grinding wheel translates with the velocity ro for the grinding of
straight concave conical involute gear.
The gear pair for the contact stress simulation that follows
comprises the pinion S
1
and the gear S
2
, which are generated by
grinding wheels S
F
and S
G
, respectively. Since the generation
processes of the pinion and the gear are identical, the subscripts i
(i 1and2), represent the pinionS
1
andgear S
2
, while j (j F andG)
represent their corresponding grinding wheels S
F
and S
G
, in the
following derivation.
2.2. Mathematical model of grinding wheel
The axial section of the grinding wheel S
j
illustrated in
Fig. 2(a) contains four major regions: two straight-edges (regions
1 and 3) and two round tips (regions 2 and 4). Regions 1 and
3 generate the working tooth surfaces, while regions 2 and 4 gen-
erate the llets of the gear. Herein, a
n
denotes the normal pressure
angle, r
j
denotes the radius of llet, while m
n
and p
n
represent the
normal module and circular pitch, respectively. Owing to the
symmetry, only the right side of the grinding wheel (i.e. regions
1 and 2) is considered in the following derivation. According to the
coordinate system illustrated in Fig. 2(b), the grinding wheel
surface can be traced out in coordinate system S
j
w
X
j
w
,Y
j
w
,Z
j
w
by
rotating the above-mentioned axial section, which is attached to
plane X
j
n
Y
j
n
, by an angle t
j
along a circular arc with radius r
j
w
and
center at O
j
c
, with respect to the Y
j
c
-axis. Herein, r
j
w
represents the
Imaginary
rack cutter
Grinding
wheel
P
0
Pitch circle of
grinding wheel
Work piece
(Gear)
Pitch plane of
imaginary rack
cutter
Infeed
direction
Tooth trace
direction
Pitch plane of
imaginary rack
cutter
Fig. 1. Infeed grinding mechanism for concave conical involute gears.
Y.-C. Chen, C.-C. Liu / Finite Elements in Analysis and Design 47 (2011) 443452 444
pitch radius of the grinding wheel. The surfaces and unit normals of
the straight edges and the round tips on the grinding wheel S
j
can
be obtained in coordinate system S
j
w
as follows:
Region 1: Straight edges
R
S,j
w

x
S,j
w
y
S,j
w
z
S,j
w
_

_
_

cost
j

j
cosa
n
ar
j
w
r
j
w
cost
j

j
sina
n
atana
n
b
sint
j

j
cosa
n
ar
j
w
sint
j
_

_
_

_
, 1
n
S,j
w

n
S,j
xw
n
S,j
yw
n
S,j
zw
_

_
_

sina
n
cost
j
cosa
n
sina
n
sint
j
_

_
_

_: 2
Region 2: Round tips
R
T,j
w

x
T,j
w
y
T,j
w
z
T,j
w
_

_
_

cost
j
r
j
cosy
j
r
j
sina
n
ar
j
w
r
j
w
cost
j

r
j
siny
j
r
j
cosa
n
atana
n
b
sint
j
r
j
cosy
j
r
j
sina
n
a r
j
w
sint
j
_

_
_

_
,
3
n
T,j
w

n
T,j
xw
n
T,j
yw
n
T,j
zw
_

_
_

cosy
j
cost
j
siny
j
cosy
j
sint
j
_

_
_

_: 4
2.3. Gear ground by grinding wheel
Fig. 3 displays the coordinate systems of the grinding wheel S
j
and the generated gear S
i
during the generation process. According
to Fig. 3(a), the plane Y
j
r
Z
j
r
of coordinate system S
j
r
X
j
r
,Y
j
r
,Z
j
r
,
which represents the pitch plane of the imaginary rack cutter for
conventional conical involute gear, is set to form an inclination
angle d
i
with respect to the plane Y
j
a
Z
j
a
, which represents the
plane axode of the cylindrical gear. Coordinate systemS
b
(X
b
, Y
b
, Z
b
)
is the reference coordinate system, while coordinate system S
i
(X
i
,
Y
i
, Z
i
) is attached to the generated gear S
i
. Fig. 3(b) shows the
translation of the grinding wheel S
j
on the pitch plane of the
corresponding imaginary rack cutter, plane Y
j
r
Z
j
r
, during grind-
ing process. Coordinate system S
j
w0
X
j
w0
,Y
j
w0
,Z
j
w0
, which forms a
helix angle c
i
with respect to coordinate system S
j
r
, can be
regardedas the initial positionof grinding wheel coordinate system
S
j
w
when the generated gear rotation angle f
i
01. When the gene-
rated gear

i
rotates through an angle f
i
, the position of the
grinding wheel

j
moves fromcoordinate systemS
j
w0
to S
j
w
and the
translational distance of the grinding wheel is O
j
w0
O
j
w
!

r
i
f
i
cosc
i
.
Based on the theory of gearing proposed by Litvin [19,20], the
generated gear surface S
i
can be obtained by simultaneously
considering the locus of the grinding wheel surface S
j
, represented
in gear coordinate systemS
i
, together with the equation of meshing
between the grinding wheel and the generated gear. The locus of
the grinding wheel represented in coordinate system S
i
can be
obtained as follows:
x
i
cosf
i
cosd
i
x
j
w
sinf
i
cosc
i
cosf
i
sind
i
sinc
i
y
j
w
sinf
i
sinc
i
cosf
i
sind
i
cosc
i
z
j
w
r
i
f
i
sinf
i
cosc
i
cosf
i
sind
i
sinc
i
cosc
i
r
i
cosf
i
,
y
i
sinf
i
cosd
i
x
j
w
cosf
i
cosc
i
sinf
i
sind
i
sinc
i
y
j
w
cosf
i
sinc
i
sinf
i
sind
i
cosc
i
z
j
w
r
i
f
i
cosf
i
cosc
i
_
sinf
i
sind
i
sinc
i
_
cosc
i
r
i
sinf
i
,
Fig. 2. Formation of the grinding wheel surface S
j
in coordinate system S
j
w
.
Fig. 3. Coordinate systems of the grinding wheel S
j
and the generated gear S
i
.
Y.-C. Chen, C.-C. Liu / Finite Elements in Analysis and Design 47 (2011) 443452 445
and
z
i
sind
i
x
j
w
cosd
i
sinc
i
y
j
w
cosd
i
cosc
i
z
j
w
r
i
f
i
cosd
i
sinc
i
cosc
i
: 5
The equation of meshing can be obtained according to the
orthogonality of the relative velocity V
ji
b
and surface common
normal n
j
b
of the grinding wheel surface S
j
and the generated gear
surface S
i
at their common contact point represented in reference
coordinate systemS
b
. The following equationcanthus be observed:
n
j
b
V
ji
b
n
j
b
V
j
b
V
i
b
0: 6
Considering the coordinate systems displayed in Fig. 3, n
j
b
can
be obtained by
n
j
b

cosd
i
n
j
xw
sind
i
sinc
i
n
j
yw
sind
i
cosc
i
n
j
zw
cosc
i
n
j
yw
sinc
i
n
j
zw
sind
i
n
j
xw
cosd
i
sinc
i
n
j
yw
cosd
i
cosc
i
n
j
zw
_

_
_

_
: 7
The angular velocity of the generated gear S
i
can be acquired in
coordinate system S
b
by
x
i
b

df
i
dt

0
0
o
i
_

_
_

_ 8
Furthermore, the velocities of the grinding wheel surface S
j
and
the generated gear surface S
i
at their common contact point can be
expressed by V
j
b
and V
i
b
, respectively, as follows:
V
j
b
r
i
o
i
sind
i
sinc
i
cosc
i
cos
2
c
i
cosd
i
sinc
i
cosc
i
_

_
_

_, 9
and
V
i
b
x
i
b
R
j
b
o
i
cosc
i
y
j
w
sinc
i
z
j
w
r
i
f
i
cos
2
c
i
cosd
i
x
j
w
sind
i
sinc
i
y
j
w
sind
i
cosc
i
z
j
w
r
i
f
i
sind
i
sinc
i
cosc
i
r
i
0
_

_
_

_
10
Substituting Eqs. (7), (9) and (10) into Eq. (6) enables us to solve
the equation of meshing. The tooth surface of the generated gear S
i
can be expressed by Eqs. (5) and (6). Hence, the complete
mathematical models of concave conical involute gear including
the working tooth surfaces and llets have been derived.
2.4. Tooth prole on the transverse plane
As mentioned above, the tooth surface of the generated gear S
i
can be expressed by Eqs. (5) and (6). By assigning z
i
as a constant
value, the prole of the generated gear on specic transverse plane
(i.e. x
i
y
i
plane) can be acquired as illustrated in Fig. 4. Section AB
and BC are the llet and working tooth regions generated by the
round tips (region 2) and straight edge (region1) of the grinding
wheel shown in Fig. 2(a), and they are parametric curves which can
be represented by position vector functions R
f
i
y
j
and R
w
i

j
,
respectively. The intervals for the parameters y
j
and
j
can be
determined in a similar manner represented by Brauer [16]. Point A
is the starting position of llet with y
j
0, while point B is the
boundary point between the llet and working tooth regions.
When the transverse tooth prole is not undercut as shown in
Fig. 4(a), point B can be regarded as the end position of llet
which coincides with the beginning position of the working tooth
(i.e. R
f
i
90
3
a
n
R
w
i
0). If undercutting occurs on the transverse
tooth prole, a singular point may appear on the working tooth
surface as illustrated in Fig. 4(b). After the generating process, part
of the working tooth may be cut again by the generation line of the
llet region while point B and the theoretical singular point no
longer exist. The boundary point D, which can be solved by a
numerical method, is the intersecting point of the generation lines
of working tooth and llet (i.e. R
f
i
y
D
j
R
w
i

D
j
). Otherwise, point
C is the end position of the working tooth, which can be calculated
by solving the intersection of the working tooth and the addendum
circle with radius r
ai
which can be acquired by
r
ai
r
i
z
i
tand
i
m
n
=cosd
i
: 11
Consequently, the intervals of the parameters y
j
and
j
can be
determined and the tooth prole on every transverse plane can be
acquired.
2.5. Assembly of the gear pair
Adopting the engagement model proposed by Mitome [21],
Fig. 5 illustrates the schematic meshing model of the conical
involute gear pair with non-parallel axes, including the pinion
Fig. 4. Tooth prole on specic transverse plane.
Fig. 5. Meshing model of the conical involute gear pair with non-parallel axes.
Y.-C. Chen, C.-C. Liu / Finite Elements in Analysis and Design 47 (2011) 443452 446
S
1
, the gear S
2
, and the pitch plane of the imaginary engaging rack.
The pinion S
1
and gear S
2
can be considered to be two imaginary
cones with cone angles d
1
and d
2
, lying on opposite sides of the
pitch plane of the imaginary engaging rack. Points O
1
and O
2
are the
center of pitchcircles of the pinionS
1
andgear S
2
withpitchradii of
r
1
and r
2
, respectively; A
1
and A
2
are the apexes of the imaginary
cones; A
1
O
1
andA
2
O
2
are the axes of the imaginary cones. The pitch
circles of the pinionS
1
andgear S
2
are in tangency at pitchpoint P
0
.
In addition, the tangent lines, A
1
P
0
and A
2
P
0
, of the two imaginary
cones with respect to the pitch plane of the imaginary engaging
rack forman angle Gc
1
+c
2
. Fig. 6 depicts in detail the coordinate
systems for the assembly of gear pair. Coordinate systems S
f
(X
f
,Y
f
,Z
f
)
and S
g
(X
g
,Y
g
,Z
g
) are the reference coordinate systems for the pinion
coordinate system S
1
(X
1
,Y
1
,Z
1
) and gear coordinate system
S
2
(X
2
,Y
2
,Z
2
), respectively. Meanwhile, f
1
u and f
2
u denote the rota-
tion angles of the pinion S
1
and gear S
2
during meshing. Applying
the assembly relationship together with the proposed mathema-
tical model of the concave conical involute pinion S
1
and gear S
2
,
the following two models of concave conical involute gear pairs
with their major design parameters listed in Table 1 can be
attained.
Model A: Straight concave conical involute gear pair with inter-
sected-axes: In this model, the gear pair is composed of straight
conical involute pinion S
1
and gear S
2
(c
1
c
2
01) with cone
angles d
1
d
2
301, mounted with an intersected angle of 601, as
displayed in Fig. 7(a).
Model B: Helical concave conical involute gear pair with crossed
axes: Fig. 7(b) illustrates a crossed axes conical involute gear pair
consisting of helical conical involute pinion S
1
and gear S
2
with
cone angles d
1
d
2
201. The helix angles on the pitch plane of the
imaginaryrackcutter for the pinionS
1
andgear S
2
are c
1
c
2
151
(right handed). Applying the algorithms proposed by Mitome [21],
the shortest axial distance between two mating gear axes is
111.731 mm, while the crossed angle is 49.6281.
3. Finite element contact stress analysis
In this study, a single pair of contact teeth is constructed to
performthe stress analysis. Applicationof FEcontact stress analysis
requires the development of the FEmodel formedbyFE meshes, the
denition of contacting surfaces, the specication of surface
interaction model, and the establishment of boundary conditions.
The following assumptions have been made: (1) the stress is in the
elastic range of the material; (2) the material is isotropic and
(3) heat generation and thermal stress are ignored.
3.1. Finite element model
The three-dimensional linear brick solid element, C3D8 [22],
having eight nodes and six faces, is employed to discretize the
geometric models of the pinion and the gear tooth surfaces. A
mesh-generation program has been developed to divide the gear
tooth into elements as well as to generate nodal points. Fig. 8
illustrates a schematic FE discretization model of a single tooth.
Several regions are assigned on the edges along the tooth prole
direction, axial direction (i.e. z
i
-axis direction) and tooth thickness
direction. In tooth prole direction, the working tooth is divided
into three regions, WT1, WT2 and WT3, while BA and FI denote the
base region and llet region, respectively. In the axial direction, the
tooth is divided into three regions, AX1, AX2 and AX3. Moreover, TH
represents the region along the tooth thickness directions. The
owchart of the developed mesh-generation program has been
presented in Fig. 9. Firstly, the design parameters of the gear pair
are given, and the proportion among regions along the tooth prole
direction (i.e. WT1, WT2 and WT3) and axial direction (i.e. AX1, AX2
and AX3.) are assigned. Then, the number of elements along the
edge of each region is designated which allows the mesh density to
be adjusted for specic requirements. According to the proportion
Fig. 6. Coordinate systems for the assembly of gear pair.
Table 1
Major design parameters of the gear pairs.
Pinion S
1
Gear S
2
Number of teeth N
1
30 N
2
40
Face width (thickness of the gear blank) W
1
16 mm W
2
16 mm
Normal pressure angle a
n
201
Normal module m
n
5 mm/teeth
Model A: Straight conical involute
gear pair with intersected axes
Model B: Helical conical involute
gear pair with crossed axes
Gear
2
Pinion
1
Pinion
1
Gear
2
Fig. 7. Gear pairs with intersected axes and crossed axes.
Y.-C. Chen, C.-C. Liu / Finite Elements in Analysis and Design 47 (2011) 443452 447
and element numbers of regions AX1, AX2 and AX3, a group of
specic transverse planes, which determines the number of ele-
ment layers in axial direction, can be specied in this FE model.
Applying the methodologies considering the intervals of para-
meters y
j
and
j
discussed previously, the tooth proles on these
specic transverse planes can be obtained, and the location of the
nodes can be determined by partitioning the intervals of para-
meters y
j
and
j
according to the desired element number. Finally,
all the elements in this FE model are created by specifying their
nodes. Nodes of the FE model lying on the tooth surfaces are
guaranteed to be points of the real tooth surfaces of the pinion and
the gear. Therefore, the loss of accuracy due to the development of
FE model using CAD computer programs is avoided.
An FE model of one pair of contact teeth for the gear pair is
established and illustrated in Fig. 10. In general, an FE model with a
larger number of elements may lead to more accurate results.
However, increasing the mesh densities for the whole FE model is
not necessary from the computational viewpoint, especially consid-
ering the limit of computer memories and the computational
efciency. With a similar concept of mesh-generation in [17], we
construct a very dense mesh in the possible contact regions to ensure
the accuracy of FEA, and a coarse mesh in the rest of the teeth to
reduce the computationtime needed. The total number of elements is
18,000with22,681nodes for eachFEmodel. Averygoodconvergence
of the calculated results has been obtained with this FE model.
3.2. Surface denition and interaction properties
Nowadays, several commercial nite element packages, such as
ANSYS, ABAQUS andMARCare capable of contact analysis. ABAQUS
Tooth thickness
direction
T
o
o
t
h

p
r
o
f
i
l
e

d
i
r
e
c
t
i
o
n
A
x
ia
l d
ire
c
tio
n
( z d
ire
c
tio
n
)
i
Fig. 8. Schematic FE discretization model of a single tooth.
Fig. 9. Flowchart of the developed mesh-generation program.
Gear
2
Pinion
1
Gear
2
Pinion
1
Fig. 10. Finite element model of the gear pair. (a) Model A: Straight concave conical
involute gear pair with intersected axes. (b) Model B: Helical concave conical
involute gear pair with crossed axes.
Y.-C. Chen, C.-C. Liu / Finite Elements in Analysis and Design 47 (2011) 443452 448
is utilized in this study to determine the contact stress. In ABAQUS,
the contact surfaces are dened as master and slave surfaces,
respectively. Generally, the master surface should be chosen as the
surface of the stiffer body or as the surface with the coarser mesh if
the two surfaces are with comparable stiffness. During the analysis
processes, the nodes of the master surface can penetrate into the
slave surface; however, the nodes on slave surfaces are constrained
not to penetrate into the master surface. According to the rotating
directions of the gear pair dened in Fig. 6, the left-side tooth
surface of the gear S
2
has been chosen as the master surface, while
the left-side toothsurface of the pinionS
1
is considered as the slave
surface. Two options, small sliding and no friction, are specied
to dene the interaction between the contact pair. In this study, the
frictional forces are neglectedsince theyare lowcomparedwiththe
normal forces [17].
3.3. Boundary conditions
For the linear brickC3D8element, eachnodehas three degree-of-
freedom(DOF), i.e., translations inthenodal x-, y- andz-directions. In
this study, all the three DOFs of the nodes located on the two lateral
sides of the base of the gear S
2
are assumed xed, as depicted in
Fig. 10. On the other hand, rigid beam elements are applied to
connect the nodes on two lateral sides of the base of the pinion S
1
with those on pinions rotational axis. In addition, the nodes on
pinions rotational axis are constrained in a way such that the pinion
S
1
canrotateabout its rotational axis only. Consequently, thegear S
2
is statically xed and a torque of 50 Nm is applied directly to the
remaining DOF at pinions rotational axis to make the tooth surfaces
of the pinion S
1
and gear S
2
contact with each other.
4. Simulation results and discussion
In the following examples, the medium carbon steel AISI 1045
(with Youngs modulus E205 GPa, and Poissons ratio n0.292)
has been chosen for the gear material for the FEA.
Example 1. Contact stress distribution of straight concave conical
involute gear pair with intersected axes
To investigate the effects of the grinding wheel radius, r
j
w
, onthe
stress distribution of the straight concave conical involute gear
pairs with intersected axes, model A in Fig. 7(a) is chosen for the FE
stress analysis in the following cases:
Case 1: r
F
w
r
G
w
1(i.e. conventional straight conical involute
gear pair).
Case 2: r
F
w
r
G
w
50mm.
Case 3: r
F
w
r
G
w
45mm.
Notably, the grinding wheel can be regarded as the imagi-
nary rack cutter when its pitch radius approaches to innity
Case 1
( = =
) ( ) ( G
w
F
w
r r )
Case 2
( mm r r
G
w
F
w
50
) ( ) (
= = )
Case 3
( mm r r
G
w
F
w
45
) ( ) (
= = )
Fig. 11. Contact stress distribution on pinions tooth surface of Model A with
different r
j
w
.
Case 4
( = =
) ( ) ( G
w
F
w
r r )
Case 5
( r r
G
w
F
w
60 mm)
) ( ) (
= =
Case 6
( r r
G
w
F
w
45 mm)
) ( ) (
= =
Fig. 12. Contact stress distribution on pinions tooth surface of Model B with
different r
j
w
.
Y.-C. Chen, C.-C. Liu / Finite Elements in Analysis and Design 47 (2011) 443452 449
(i.e. r
j
w
1), which makes the concave conical involute gear iden-
tical to the conventional conical involute gear. Fig. 11illustrates the
distributions of von-Mises stress on the tooth surface of the pinion
S
1
when the contact occurs on the pitch point of the gear pair.
The maximum contact stress occurs at the contact position on the
central region of the tooth surface. According to the FEA results,
with the decrease of r
j
w
, the contact area increases and the contact
stress reduces signicantly.
Example 2. Contact stress distribution of helical concave conical
involute gear pair with crossed axes
In this example, the helical concave conical involute gear pair of
model B, as illustrated in Fig. 7(b), is adopted for the FE stress
analysis as well. Cases 46 simulate the contact of the gear pairs
with different pitch radii r
j
w
of the grinding wheel.
Case 4 r
F
w
r
G
w
1 (i.e. conventional helical conical involute
gear pair ).
Case 5: r
F
w
r
G
w
60mm.
Case 6: r
F
w
r
G
w
45mm.
Fig. 12 illustrates the distributions of von-Mises stress on
pinions tooth surface. Similarly, the contact area enlarges and
the contact stress reduces signicantly as r
j
w
decreases.
Example 3. Comparison between normal contact stress and Hertzian
contact stress
The aim of this example is to evaluate the Hertzian contact
stress and compare the results with the normal contact stress
obtained fromFEA. In order to calculate the Hertzian contact stress,
the information about the curvatures of mating tooth surfaces is
required. According to Hertzian contact stress formulae [23], the
maximumHertziancontact stress s
H
canbe evaluatedaccording to
Eq. (A.19) in Appendix A.
Since the Hertzian contact stress is evaluated in the direction
along the line of action of the mating tooth surfaces, the normal
contact stress obtained from the FEA results should be considered
herein for the reasonable comparison. According to the TCA results
[8], the element where the theoretical contact point is located can
be determined. The stress components of s
x
, s
y
, s
z
, t
xy
, t
xz
and t
yz
on integration points in this element can be obtained from the
output data of ABAQUS. Consequently, the normal contact stress
along the line of actioncan be calculatedby the following equation:
s
n
s
x
l
2
x
s
y
l
2
y
s
z
l
2
z
2t
xy
l
x
l
y
2t
xz
l
x
l
z
2t
yz
l
y
l
z
, 12
where l
x
, l
y
and l
z
are the direction cosines of the line of action
represented in coordinate system S
f
.
Tables 2 and 3 summarize the normal contact stress obtained
from FEA results and the calculated Hertzian contact stress for the
straight and helical concave conical involute gear pairs, discussed in
Examples 1 and 2 , ground by grinding wheels with various radii of
r
j
w
. The results have shown that the differences between the normal
contact stress obtainedfromFEAandthe calculatedHertziancontact
stress are less than 5%. For both the straight and the helical concave
conical involute gear pairs inthis example, the reduction of Hertzian
contact stresses are about 60% when the grinding wheels pitch
radius r
j
w
decreases fromNto 45 mm. The results indicated that the
contact stress of non-parallel-axes concaveconical involutegear pair
can be reduced signicantly by choosing grinding wheels with
smaller radii. Hopefully, concave conical involute gear pairs will
improve the performance and range of applications of these gears.
5. Conclusions
In this study, FE stress analysis is performed to investigate the
contact stress of concave conical involute gear pairs with non-
parallel axes. The complete tooth geometry has been derived,
including the working tooth surfaces and the llets. On the basis
of the developed mathematical models, a mesh-generation program
has been developed to discretize the geometric models of the pinion
and the gear using eight-node solid elements. The commercial FEA
software, ABAQUS, capable of contact analysis is applied to evaluate
Table 2
Normal contact stress obtained by FEA and the Hertzian contact stress of the straight concave conical involute gear pair (Model A).
Grinding wheel
radius r
j
w
(mm)
Normal stress
by FEA s
n
(MPa)
Contact stress by Hertzian formulations
Be=Ae c
b
c
s
s
H
(MPa)
N (Case 1) 1192.430 3.528 0.625 0.860 1272.739
200 1146.480 4.407 0.580 0.895 1192.313
100 1078.110 5.880 0.550 0.910 1113.423
80 1037.200 7.061 0.525 0.930 1068.315
70 989.258 8.246 0.500 0.940 1016.016
60 904.791 10.625 0.475 0.960 969.639
50 (Case 2) 828.803 17.837 0.420 0.990 863.381
45 (Case 3) 717.157 32.607 0.370 1.000 755.831
Table 3
Normal contact stress obtained by FEA and the Hertzian contact stress of the helical concave conical involute gear pair (Model B).
Grinding wheel
radius r
j
w
(mm)
Normal stress
by FEA s
n
(MPa)
Contact stress by Hertzian formulations
Be=Ae c
b
c
s
s
H
(MPa)
N(Case 4) 965.174 5.190 0.570 0.900 975.330
200 906.410 6.375 0.535 0.920 901.960
100 817.300 8.260 0.500 0.940 829.974
80 782.711 9.693 0.480 0.950 790.408
70 777.052 11.064 0.470 0.965 775.748
60 (Case 5) 724.931 13.636 0.450 0.975 737.163
50 680.095 20.212 0.405 0.990 656.946
45 (Case 6) 599.281 29.786 0.375 1.000 604.168
Y.-C. Chen, C.-C. Liu / Finite Elements in Analysis and Design 47 (2011) 443452 450
the stress distribution on the tooth surfaces. Based on the analysis
results, the following conclusions can be drawn:
(1) Non-parallel-axes concave conical involute gear pairs, which
are generated by grinding wheels with appropriate grinding
wheel radii r
j
w
, exhibit point contact.
(2) The contact stresses of bothstraight andhelical concave conical
involute gear pairs withnon-parallel axes are investigatedwith
the variation of grinding wheel radius r
j
w
. The simulation
results indicate that the contact stress can be reduced sig-
nicantly by choosing a smaller grinding wheel radius r
j
w
for
the generation of concave conical involute gear pairs. The
simulated results of contact stress are consistent with those of
contact ellipses estimation [8].
(3) The normal contact stress has beenobtainedby FEAsimulation,
and the results are compared with the Hertzian contact stress
calculated from the Hertzian stress formulae and curvature
analysis.
Acknowledgement
The authors would like to thank the National Science Council of
the ROC for nancially supporting this research under Contract
no. NSC 96-2221-E-231-013-MY3.
Appendix A. Curvature analysis and Hertzian contact stress
Methodologies proposed by Litvin [19,20] considering the prin-
cipal directions and curvatures of mating tooth surfaces are adopted
toobtainthe contact ellipses. This approachprovides anopportunity
to determine the principal directions and curvatures of the gener-
ated gear tooth surfaces directly instead of using the complicated
equations of the generated gear tooth surfaces. Since the generation
processes for the pinion S
1
and the gear S
2
of this gear pair are the
same, only the relationships between the principal directions and
curvatures of grinding wheel surface S
F
and the corresponding
generated pinion surfaces S
1
are shown hereinafter for simplicity.
A.1. Principal directions and curvatures of grinding wheel surface S
F
Since t
F
and
F
are the surface parameters of the grinding wheel
surface S
F
, the principal directions and curvatures of the grinding
wheel S
F
can be expressed in the coordinate system S
b
as follows:
(1) When d
F
=dt 0, the rst principal direction i
F
I
and curvature
k
F
I
are derived as follows:
i
F
I

V
F
rI
V
F
rI

cosd
1
sint
F
sind
1
cosc
1
cost
F
sinc
1
cost
F
sind
1
sint
F
cosd
1
cosc
1
cost
F
_

_
_

_, A:1
and
k
F
I

sina
n

F
cosa
n
ar
F
w

: A:2
(2) When dt
F
/dt0, the secondary principal direction i
F
II
and
curvature k
F
II
are derived as follows:
i
F
II

V
F
rII
V
F
rII

cosd
1
cost
F
cosa
n
sind
1
sinc
1
cosa
n
cosc
1
sint
F
cosa
n

cosc
1
sina
n
sinc
1
sint
F
cosa
n
sind
1
cost
F
cosa
n
cosd
1
sinc
1
cosa
n
cosc
1
sint
F
cosa
n

_
_

_,
A:3
and
k
F
II
0: A:4
A.2. Principal directions and curvatures of generated pinion tooth
surface S
1
The principal directions and curvatures of the generated pinion
tooth surface S
1
can be determined by applying the following
equations [19,20]:
tan2s
F1

2F
1
k
F
I
k
F
II
G
1
, A:5
k
1
I
k
1
II
k
F
I
k
F
II
S
1
, A:6
and
k
1
I
k
1
II

k
F
I
k
F
II
G
1
cos2s
F1
, A:7
where
F
1

a
1
31
a
1
32
b
1
3
V
F1
i
F
I
a
1
31
V
F1
i
F
II
a
1
32
, A:8
G
1

a
1
31

2
a
1
32

2
b
1
3
V
F1
i
F
I
a
1
31
V
F1
i
F
II
a
1
32
, A:9
S
1

a
1
31

2
a
1
32

2
b
1
3
V
F1
i
F
I
a
1
31
V
F1
i
F
II
a
1
32
, A:10
a
1
31
n
F
x
F1
i
F
I
k
F
I
V
F1
i
F
I
, A:11
a
1
32
n
F
x
F1
i
F
II
k
F
II
V
F1
i
F
II
, A:12
and
b
1
3
n
F
x
1
V
F
tr
n
F
x
F
V
1
tr
A:13
All the vectors expressed in Eqs. (A.5)(A.13) are represented in
the coordinate systemS
b
. Herein, i
1
I
and i
1
II
are the rst and second
principal directions of the generated pinion tooth surface S
1
; k
1
I
and k
1
II
represent the rst and second principal curvatures of S
1
;
and s
(F1)
is the angle formed by the rst principal directions of the
cutter surface i
F
I
and the generated pinion surface.
A.3. Contact ellipses and Hertzian contact stress
The principal directions of the pinion S
1
and the gear S
2
at their
instantaneous contact point can be represented in coordinate
system S
f
by applying the coordinate transformation matrix
equation. Obviously, at any instantaneous contact point, unit
vectors i
1
I
and i
2
I
are located on the common tangent plane of
the mating tooth surfaces S
1
and S
2
, as shown in Fig. 13. The half
lengths of the major and minor axes of the contact ellipse are
represented by a
e
and b
e
, respectively. The orientation of the
contact ellipse is determined by the angle g which is measured
fromthe minor axis to the rst principal direction of the pinion S
1
.
The angle g can be calculated by the following equations [19,20]:
tan2g
g
2
sin2s
g
1
g
2
cos2s
, A:14
Where
g
1
k
1
I
k
1
II
, A:15
Y.-C. Chen, C.-C. Liu / Finite Elements in Analysis and Design 47 (2011) 443452 451
and
g
2
k
2
I
k
2
II
: A:16
Meanwhile, angle s is formed by the rst principal directions
of the pinion and gear tooth surfaces i
1
I
and i
2
I
, and it can be
expressed by
s cos
1
i
1
I
i
2
I
: A:17
According to Hertzian contact stress formulae [23], the semi-
minor axis b
e
of the contact ellipse, and the maximum Hertzian
contact stress s
H
can be evaluated by
b
e
c
b

FD
3
p
, A:18
and
s
H
c
s
b
e
D
_ _
, A:19
where the auxiliary parameter D is dened as
D
21n
2

A
e
B
e
E
: A:20
Herein, n is Poissons ratio while E denotes Youngs modulus,
and the values of A
e
and B
e
are determined as follows [19,20]:
A
e

1
4
k
1
S
k
2
S
g
2
1
2g
1
g
2
cos2sg
2
2

1=2
_ _
, A:21
B
e

1
4
k
1
S
k
2
S
g
2
1
2g
1
g
2
cos2sg
2
2

1=2
_ _
, A:22
where
k
1
S
k
1
I
k
1
II
, A:23
and
k
2
S
k
2
I
k
2
II
: A:24
Meanwhile, coefcients c
b
and c
s
can be attained by [23]
c
b

3kEku=2p
3
_
A:25
c
s

1k
2
ku
2
Eku
_ _
A:26
Where
k b
e
=a
e
cosy
e
, A:27
ku

1k
2
_
siny
e
, A:28
and
Eku
_
p=2
0

1ku
2
sin
2
y
e
_
dy
e
: A:29
For convenience, coefcients c
b
and c
s
can be determined
according to Fig. 18.10 and 18.11 in the Ref. [23]. Otherwise, F
denotes the contact force acting on the contact point along the line
of action and can be determined by [9]
F
T
f
1
R
1
f
n
1
f
a
1
f
, A:30
where T
f
(1)
is the torque applied to the pinions rotational axis; R
1
f
and n
1
f
denotes the position vector and the surface unit normal
vector of the contact point, and a
f
(1)
represents the unit vector of the
pinions axis of rotation represented in coordinate system S
f
.
References
[1] A.S. Beam, Beveloid gearing, Machine Design 26 (1954) 220238.
[2] C.-C. Liu, C.-B. Tsay, Contact characteristics of beveloid gears, Mechanism and
Machine Theory 37 (2002) 333350.
[3] T. Ohmachi, K. Lizuka, H. Komatsubara, K. Mitome, Tooth surface fatigue
strengthof normalized steel conical involute gears. In: Proceedings of the DETC
2000 (ASME), Baltimore, MD, USA, 1013 September 2000.
[4] T. Ohmachi, J. Sato, K. Mitome, Allowable contact strength of normalized steel
conical involute gears. In: Proceedings of the JSME International Conference on
Motion and Power Transmission (MPT2001-Fukuoka), Fukuoka, Japan, 1517
November 2001.
[5] K. Mitome, Infeed grinding of straight conical involute gear, JSME International
Journal Series C 36 (4) (1993) 537542.
[6] H. Komatsubara, K. Mitome, T. Ohmachi, Development of concave conical gear
used for marine transmissions (1st report, principle of generating helical
concave conical gear), JSMEInternational Journal Series C45(1) (2002) 371377.
[7] H. Komatsubara, K. Mitome, T. Ohmachi, Development of concave conical gear
used for marine transmissions (2nd report: Principle normal radii of concave
conical gear and design of a pair of gears), JSME International Journal Series C
45 (2) (2002) 543550.
[8] C.-C. Liu, C.-B. Tsay, Mathematical models and contact simulations of concave
beveloid gears, ASME Journal of Mechanical Design 124 (2002) 753760.
[9] F.L. Litvin, J.S. Chen, J. Lu, R.F. Handschuh, Application of nite element analysis
for determination of load share, real contact ratio, precision of motion, and
stress analysis, ASME Journal of Mechanical Design 118 (1996) 561567.
[10] F.L. Litvin, Q. Lian, A.L. Kapelevich, Asymmetric modiedspur gear drives: reduction
of noise, localization of contact, simulation of meshing and stress analysis,
Computer Method in Applied Mechanics and Engineering 188 (2000) 363390.
[11] V. Simon, FEMstress analysis in hypoid gears, Mechanismand Machine Theory
35 (2000) 11971220.
[12] C.-H. Li, H.-S. Chiou, C. Hung, Y.-Y. Chang, C.-C. Yen, Integration of nite
element analysis and optimum design on gear systems, Finite Elements in
Analysis and Design 38 (2002) 179192.
[13] Y.-C. Chen, C.-B. Tsay, Stress analysis of a helical gear set with localized bearing
contact, Finite Elements in Analysis and Design 38 (2002) 707723.
[14] J. Argyris, A. Fuentes, F.L. Litvin, Computerized integrated approach for design
and stress analysis of spiral bevel gears, Computer Methods in Applied
Mechanics and Engineering 191 (2002) 10571095.
[15] F.L. Litvin, I. Gonzalez-Perez, K. Yukishima, A. Fuentes, K. Hayasaka, Design,
simulation of meshing, and contact stresses for an improved wormgear drive,
Mechanism and Machine Theory 42 (2007) 940959.
[16] J. Brauer, A general nite element model of involute gears, Finite Elements in
Analysis and Design 40 (2004) 18571872.
[17] J. Brauer, Transmission error in anti-backlash conical involute gear transmis-
sions: a global-local FE approach, Finite Elements in Analysis and Design 41
(2005) 431457.
[18] S.-H. Wu, S.-J. Tsai, Contact stress analysis of skew conical involute gear drives in
approximatelinecontact, MechanismandMachineTheory 44(2009) 16581676.
[19] F.L. Litvin, Theory of Gearing, NASAPublication RP-1212, Washington, DC, 1989.
[20] F.L. Litvin, Gear Geometry and Applied Theory, Prentice-Hall, Englewood Cliffs,
NJ, 1994.
[21] K. Mitome, Conical involute gear (design of nonintersecting-nonparallel-axis
conical involute gear), JSME International Journal Series III 34 (2) (1991) 265270.
[22] Hibbitt, Karlsson, Sorensen, ABAQUS/Standard 6.5, Users Manual, Pawtucket,
Rhode Island, 2004.
[23] A.P. Boresi, R.J. Schmidt, O.M. Sidebottom, Advanced Mechanics of Materials,
5th Ed., John Wiley and Sons, New York, 1993.
i
I
(1)
i
(2)
I
2a
2b
e
e
Fig. 13. Orientation and dimensions of contact ellipse.
Y.-C. Chen, C.-C. Liu / Finite Elements in Analysis and Design 47 (2011) 443452 452