D. C.H.

Yang
A Rotary iodel for Spur Gear
Assistant Professor.
Assoc. Mem. ASME
Dynamics
We develop a dynamic model for a spur gear system with backlash. This model is
Z. S. SUIT circular and is geometrically different from the rectilinear gear model ofAzar and
Visiting Scholar. Crossley. By taking advantage of involute tooth prof He, we are able to take material
compliance and energy dissipation into account. Furthermore, the complicated
University of California,
phenomenon of contact tooth pairs alternation between one and two during
Los Angeles, CA 90024 meshing is also included in the model. This model is believed to be closer to reality
than the existing model and hopefully is useful in studying gears in high-speed and
intermittent motions.

1 Introduction
Gear transmission plays an important role in modern
technology. It transfer both power and motion, and appears
in various kinds of machinery and control systems. Backlash
between gear teeth is unavoidable in gear transmission. It is
also necessary for lubricating the tooth surface and preventing
teeth from getting jammed. However, backlash also results in
additional dynamic force, reduced stability, and profound
noise and vibration. Particularly during high speed operations
as well intermittent motions, this problem becomes non-
negligible. Therefore, it is important to have a reasonable
model for analytical analysis of gear pair with backlash. 'a»ao
In 1971, Dubowsky and Freudenstein [1,2] developed a
rectilinear model called "Impact Pair" to analyze the
dynamic response of a mechanical system with clearance.
Using Dubowsky's model, Azar and Crosseley [3] later
studied the dynamic behaviors of meshing gears including the
effects of backlash, time-varying stiffness damping of the
gear teeth.
Today most gears used in industry are of involute tooth
profile. Teeth with this profile satisfy the fundamental law of Fig. 1 A pair of spur gear with backlash
gearing even with backlash. By using the kinematic charac-
teristics of involute curve, we now propose a circular model and r2), are separated. New pitch circles (with radii i?t and
for studying gear dynamics. This model differs from the R2) become tangent to each other. On these new pitch circles,
previous rectilinear model in shape, and considers of the tooth thickness is less than one half of the circular pitch so
backlash, material compliance, and energy dissipation. As a that backlash appears. Consequently, the line of contact
result, we think this model is more adequate in gear between teeth, which coincides with the common tangent of
simulation than previous ones. the two base circles, gets sharper and the new pressure angle
will be slightly greater than the original value.
2 Dynamic Model of Spur Gear If the gear teeth stay in contact during rotation, the ratio of
angular velocities can be written as
2.1 Kinematic Description. A pair of standard spur gears Rh _Ri
(pressure angle = 15 or 20 deg) is generally assembled with (1)
slightly larger center distance than the nominal one. In this R h R,
case, as shown in Fig. 1, the original pitch circles (with radii r, where Rb is the radius of the base circle, and Z is the number
of teeth. Since backlash exists, when the driving gear slows
down, stops, or reverses direction, gear teeth may lose contact
momentarily. In this case, impacts between teeth will occur,
The second author is currently with the Department of Mechanical Engineer- and the angular velocities of gears 1 and 2 will no longer
ing, Shaanxi Mechanical Engineering Institute, the People's Republic of China. satisfy the relationship given in equation (1).
Contributed by the Mechanisms Committee and presented at the Design
Engineering Technical Conference, Cincinnati, Ohio, September 10-13, 1985, Spur gear drive with involute teeth profiles is a rotary
of THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS. Manuscript received at system described by inertia, applied torques, and angular
ASME Headquarters, December 4, 1984. Paper No. 85-DET-2. motion. When gear teeth are in contact, they always touch

Journal of Mechanisms, Transmissions, and Automation in Design DECEMBER 1985, Vol. 107/529
Copyright © 1985 by ASME
Downloaded From: http://mechanicaldesign.asmedigitalcollection.asme.org/ on 05/22/2015 Terms of Use: http://asme.org/terms
 Fig. 3 Herzian contact between cylinders

Fig. 1 Rotary impact model of meshing spur gears

along the common normal line of the tooth profiles. As

shown in Fig. 1, this common normal line, which is also the
common tangent line of the two base circles, will remain fixed
during gear meshing. If we consider the friction force along
the tangent direction of contact to be negligible for a well
lubricated meshing gear pair, the contact forces between teeth
can then be considered as transfering along the common
normal line only. (In this investigation, the "common normal
line" (in terms of contact surfaces) and the "common tangent
line" (in terms of the base circles) are used interchangably.)
As previous mentioned, backlash makes a loss of contact
between gear teeth possible. This loss may provoke sub-
sequent impacts which will raise vibration. In this in-
vestigation, we simulate meshing gear system backlash by a
rotary model. As shown in Fig. 2, this model is similar to
Dubowsky [1] and Azar's [3] in representing surface com-
pliance and damping, yet different in shape. At the nominal
position given in Fig. 2, a designated tooth of gear 1 coincides
with the center line Oi, 02, and the backlashes, B, are equal at
both sides of this tooth with respect to corresponding teeth of Fig. 4 Contact of the spur teeth
gear 2. Let /, be the moment of inertia of gear ; about the
center 0,; 0, be the angular displacement of gear i from the
nominal position, and T, be the applied torque to gear ('. For N2. Let us denote this relative displacement along the line of
convenience, we define 0) and Ti{t) to be positive in the contact as S, and S is given by
counterclockwise direction, and Q2 and T2(t) as positive in the S = Rbl6\ -Rb282 (3)
clockwise direction. Backlash B is measured on the common
normal between the profiles of adjacent teeth. According to Apparently when S>B, there is contact between gears 1 and 2
properties of gears with involute profile, the displacement oh Nj. Physically, this means that the tooth of gear 1 pushes
traveles by the tooth profile along the common normal equals into the corresponding tooth of gear 2, and consequently, the
the arc length rotated on the base circle of the corresponding contact forces to gear 2 are clockwise and those to gear 1 are
gear. This means that the arc length Rbidj can be measured on counterclockwise. Let F{t) and G(t) represent the elastic and
the common normal as a straight length. damping forces during the contact, respectively. In this
situation the dynamic equations during the contact can be
2.2 Equations of Motion. Because backlashes exist written as
between teeth, the dynamic equation of the gear pair during y1e1 = r 1 (o-[F(/)+G(0]i? w
meshing depends on three different geometrical situations:
J2d2 = T2(t) + [F(t) +G{t)]Rb2
(i) Rbldl~Rb2d2>B (2)
Usually, T2{t) in equation (4) represents the load applied to
Rbl 0[ and R 1,262 are the travelled arc lengths on the base circle gear 2, and should have a negative value.
of gears one and two, respectively. The difference between
Rbl6l and Rb262 equals the relative displacement between the (ii) B>RMBx~Rb2d2>-B (5)
two meshing teeth profiles along the common tangent line, In this situation, no contact occurs between these two gears,

530 / Vol. 107, DECEMBER 1985 Transactions of the ASME

Downloaded From: http://mechanicaldesign.asmedigitalcollection.asme.org/ on 05/22/2015 Terms of Use: http://asme.org/terms

 DOUBLE TOOTH CONTACT SINGLE TOOTH CONTACT

Bi |Ai)

Base Circle 1

CONTACT REGION

Fig. 6 Alternation of the number of contact pairs

F> Compression Force

Base Circle2

Fig. 5 Geometrical relationships of teeth contact

i.e., both F{t) = G{t) = Q. Therefore the dynamic equation of

equation (4) becomes Xmax
M=ri(0 (6)
Fig. 7 Hysteresis loop
j2e2 = T2(t)
(iii) Rb262~RbiBl^B (7) where L is the thickness of these two gears, and rx and r2
represent the distance of CXA and C2A, respectively.
In this situation, the contact occurs along the common The magnitude of the interpenetration between teeth can
tangent line, Nx. Therefore, the contact forces become then be obtained (referring to Fig. 3) by
positive to gear 1 and negative to gear 2. The dynamic
equations should be written as b +r r l
J A =T{(t) + [F(t) + G(t)]Rbi
(8)
^ ^ 4 -{y)2 -r4'-& a i)
J262 = T2(t)-[F(t) + G(t)]Rb2 If we substitute the expression of b in equation (10) into
equation (11), we will have a highly nonlinear relationship
3 Derivation of Elastic Force between the applied force, F, and the complianced in-
terpenetration, 8. In other words, the stiffness function will
During the gear meshing, equations (4) and (8) have two become very complicated. Fortunately, we have found a way
unknowns: the elastic force F(t) and the damping force G(t). to avoid this obstacle. If we approximate the square-root
In this section we will develop formulas for the elastic force terms in equation (11) by their first two terms of the binomial
function, F{t), expansion, i.e.
According to Hertzian law, the elastic compression of two
isotropic elastic bodies can be approximated by two
paraboloid in the vicinity of the contact point [4, 5]. Referring
to Fig. 3, if the two contact bodies can be approximated by
j^y-
'

Equation (11) becomes

• 1 - ^
b2
0 2 )

two cylinders, then, the half-width of the surface of contact

presented by a normal force F can be expressed as 5
= r
^r2-rl(l-^)-r2(l-^)
\-v\ r2_
^rU*r vE7 ' r, + r2
(9)
4(-U-> d3)
Where b is the half width of the surface of contact; L is the 2 \rt r2/
thickness of the cylinders; and rh Eh and e,- represent the The error of this approximation is on the order of (b/ff, and
radius, Young's modules and poisson's ratio of cylinder i, will be less than 0.5 percent for steel gears. Substituting
respectively. equation (10) into equation (13), we obtain
Now consider a pair of spur gears (with involute profile) in 4(1 - v2)
8= (14)
mesh, as shown in Fig. 4. Based on the kinematic properties TTEL
of the involute profile, we know that the point of contact, Equation (14) can be rewritten as
point A, shall always be on the common tangent line. The
circle of curvature which approximates the tooth profile at the vELS
=Kb (15)
vicinity of point A is always centered at the fixed tangent 4(1 - v2)
point, point C,, on the corresponding base circle i. Consider where 5 can be calculated from relative gear displacement as
these two circles of curvature as contact cylinders, and let Fbe
the compressing force between these two teeth. The half width h=\RbXel-Rb2e1\-B=\S\-B (16)
of the contact could, therefore, be expressed as Equation (15) gives a remarkable result: the relationship
2 between input force F and the interpenetration of two
8(1 -v )F r,/-2 meshing teeth becomes linear, and the stiffness of Hertzian
(10)
TTEL n +r2 contact, denoted by K, becomes constant.

Journal of Mechanisms, Transmissions, and Automation in Design DECEMBER 1985, Vol. 107/531

Downloaded From: http://mechanicaldesign.asmedigitalcollection.asme.org/ on 05/22/2015 Terms of Use: http://asme.org/terms

 4 Variation of Stiffness due to the Number of Contact IC,tf2l=V*2^R^-ef0 (22)
Pairs The angular displacement measured from O ^ to the
Although the stiffness during the contact of a single pair of symmetrical line of the meshing tooth can be found by
teeth can be approximated by a constant shown in equation 0i=^-06 (23)
(15), we can not guarantee that this constant will hold for the
where 6b is half of the tooth thickness angle measured on the
entire gear motion. In the system, the stiffness of contact will
base circle and
change because at different times, the number of meshing
e
"-^k+invao (24)
teeth pairs will alternate between one and two. In the
currently existing dynamic model for bears, this consideration
has not been included [3]. It is one of our objectives to in-
tegrate this consideration into our model. where Sb represents the tooth thickness on the base circle of
Figure 5 shows a pair of spur gears with involute profiles in gear one. Referring to Figs. 5 and 6, the 41 angles for other
mesh. Assume that this pair of teeth comes into contact reference points (A2,AX, andB x ) on the common normal are
starting at point B2 and separates at point Bx, where both B2 2TT
and Bi are on the common normal line. Three periods exist in
terms of the number of the contact teeth pairs: 1. from B2 to
A2: in this region, the preceding pair of teeth are still in 2TT
contact, so we have two pairs of teeth in contact. Since the (25)
area of contact is doubled, it is reasonable to believe that the
stiffness of Hertzian contact also doubles. This means that for 2TT
two pair contact, K{6) is considered to be twice the K given by i>B\ = i>B2 + e '
equation (15), i.e., K{B) = TTEL/2(1 - v2). 2. From A2 to A x: in
this region, the preceding pair of teeth has already separated, If the contact occurs at the left face of a gear one tooth, the
yet the succeeding pair has not come in contact. Therefore we contact line will be switched to another common normal, N2.
have a single pair of teeth touching. In this case, the stiffness In this case, the geometric relationships and the variation of
given in equation (15) holds. 3. From ,4[ to B2: this region is contact regions become the mirror image of that given in Fig.
the same as that in situation 1, except this pair of teeth 5.
becomes the new preceding pair, and another new pair of In summary, we can express the varying of stiffness of
teeth starts contact in the region ofB2A2. Since region three is Hertzian contact as
exactly the same as region one, actually only two regions exist.
Figure 6 illustrates the reciprocating characteristics of these
two regions of meshing gears. K(6X) = \ for (26)
\2K 0 m <0,<</>„
In the next step we determine the locations of points B2,
A2, Ax, and Bx on the common normal line in terms of gear
parameters and angular displacements. Referring to Fig. 5, When S > 0 , i.e., contact occurs at the right face(s) of tooth
we find that the contact region for a pair of teeth is (teeth) on gear one, we have
\BxB2\=et0 (17) 2TT
where t0 = \B2AX I ( = 7r/M1cosa0). is the circular pitch on the 4>/ = >/<M + ( e - l ) ^ - h
Z
' .
base circle, mx is the module of gear 1, and a 0 is the pressure
angle, e in equation (17) represents the overlap coefficient 2TT
given by 4>m = ^m+ -7T ~6b (27)
z
i
2TT
e= — (Z^tana,,! - t a n a 0 ) + Z 2 (tana a2 - t a n a 0 ) ) (18)

where Z, denotes the number of teeth on gear /, and aai is the for S < 0 , i.e., contact occurs at the left face(s) of tooth (teeth)
pressure angle of the tooth profile on the corresponding on gear one, we have
addedem circle, which can be calculated by
Z C Sa 0/ = -[^B2 + e ~ - 0 A J
Ct„l = COS " 'f^)=cos-'f
V r I
' ° °)
V Zj + 2 /
(19)
ai

As mentioned previously, a contact point measured on the </>m = - [ f e > + -£- -eb] (28)
common normal line can be mapped to the corresponding
point on the base circle via the property of involute curve. The
mapping is one-to-one, e.g., in Fig. 5, points B2 and B2 are a <t>n -[^2+(e-l)|^ -6b]
pair of two corresponding points. This property is useful in
finding the angular displacement of a gear from a given S Damping Factor and Damping Function
contact position on the common normal. Referring to Fig. 5,
let the contact occur at the right face of a gear one tooth, such Another unknown in this dynamic model of meshing gears
as point B2. Let the angle \j/ be the angle between the center (Fig. 2) is the damping force G{t) given in equations (4) and
line O ^ and line 0XB2: the line connecting the center of gear (8). When two elastic bodies impact on each other, most of the
one and the corresponding contact point on the base circle. elastic strain energy is restored, but a portion of it will be
Then dissipated in heat due to random molecular vibration. This
energy loss can be represented by the area of a hysteresis loop
tm= ~(at0-(3B2) (20) as shown in Fig. 7, and can also be considered as a damping
where effect during the impact [6]. There are two boundary con-
ditions for this damping force: zero damping at 6 = 0 (no
PB2=\ClBi\/rbUor = \ClB2\/r„ (21) contact), and zero damping at <5 = <5max (zero relative velocity).
and One way to satisfy these two conditions is to assume that this

532/Vol. 107, DECEMBER 1985 Transactions of the ASME

Downloaded From: http://mechanicaldesign.asmedigitalcollection.asme.org/ on 05/22/2015 Terms of Use: http://asme.org/terms

 .surface °

TIME IN m. Fig. 9(a) Case 2: Relative displacement along the common normal
line
Fig. 8(a) Case 1: Relative displacement along the common normal
line

O 5 10 15 20
Fig. 8(b) Casel: Angular displacements T I M E IN ms

Fig. 9(b) Case 2: Angular velocities

studies [6, 7, 8]. Following the approach of Lee and Wang [7],
we choose
G(t) = D88 (29)
where D is the still unknown damping factor. To determineD,
we consider the equation of relative motion of meshing gears
without external force. The equation can now be written as
J8 + D55+K8 = 0 (30)
where / represents the effective inertia of the two meshing
gears and J=JiJ2/(J] + J2)- This equation is nonlinear.
Instead of using the polynomial approximation technique of
[7], here we solve it analytically. Equation (30) can be
rearranged into the form of

8d8 (31)
D \ D8 + K/
Fig. 8(c) Case 1: Angular velocities
An analytical solution of equation (31) can be obtained by
integrating both sides directly. The result is
damping force is a function depending on both the relative
displacement, 6, and the relative velocity, 6, of the two
J . . JK / D8 + K\
D^-&)+DHD8TTK)-
t
meshing teeth. In other words, this damping force is con- where 6, (or say V,) represents the initial relative impact
sidered viscous, and at the same time, proportional to the velocity between the teeth. The relative departing velocity Vj
magnitude of the interpenetration. To express this damping can be obtained from equation (32) by letting 6 = 0, i.e.,
function, different formulas have been suggested in previous Vf=6\s=q. When 6 = 0, there are two possible solutions for 6,

Journal of Mechanisms, Transmissions, and Automation in Design DECEMBER 1985, Vol. 107 / 533

Downloaded From: http://mechanicaldesign.asmedigitalcollection.asme.org/ on 05/22/2015 Terms of Use: http://asme.org/terms

 namely, 8 = V, and 5 = Vf. The right answer for 5 should be
the Vj, which does not equal K, and has an opposite sign to
Vj. By letting 5 = 0 and 8=Vf, equation (32) becomes

This result matches that obtained by Herbert and Mc-

Whannell [8]. Let e be the coefficient of restitution of the
impact, i.e., e= Vf/Vh then equation (33) can be rearranged
into

D(Vi - Vf) +K In ( l - y + ^ ' ) =0 (34)

From previous studies (references [6, 7]), we know that

D~Q.15aK, where, for steel, A'has a magnitude about 1.8 x
105 N/mm and a is closed to 2 x 10" 6 s/mm. Therefore, if
the initial relative velocity is less than 106 mm/s (a reasonable
assumption for meshing gears), then the second term of TIME IN ms

equation (34) can be approximated by taking the logarithmic Fig. 10(a) Case 3: Relative displacement along the common normal
expansion. The expression of the damping factor D can then line
be obtained (from reference [8]) as
6(1 -e) K
D= 2 (35)
[ ( 2 e - l ) + 3] ~V>
The coefficient of restitution in equation (35) can be obtained
approximately from a best-fit curve of Goldsmith's ex-
perimental data [5] as:
e= 1 - 0.022 VfM (36)
The maximum relative error between calculated e from
equation (36) and the original Goldsmith's data are less than
1.5 percent.
By using equations (29), (35), and (36), the damping force
G(t) in equations (4) and (8) can be calculated. Consequently,
the rotary model we propose for studying meshing gear
dynamics is complete.

6 Dynamic Simulation
For the convenience of digital simulation, the equations of
motion given in equations (4), (6), and (8)) are rewritten into a
TIME IN ma
compact form as
Fig. 10(b) Case 3: Angular velocities
jJ,=Ti(t)-RM[K(el) + D8]8*
J2e2 = T2(t)+Rb2[K(dl) + D8]8* rad/s. Figure 8(6) clearly shows the successive impact on the
alternative sides of the tooth pair. It also illustrates that the
where penetration \S\-B decreases and the period of impact in-
8* = 8 if S > 5 creases (i.e., the impact frequency decreases), as time in-
8*= -8 if S<-B creases. The angular displacements and velocities of gears one
and two are given in Figs. 8(b) and 8(c), respectively. Since
=0 if -B<S<B (37) gear one has a smaller diameter and inertia, the magnitude
Formulas for calculating 8; K and D are given in equations (3) and fluctuation of its angular displacement and velocity are
(16); (15) and (26); and (15), (26), (35), and (36), respectively. larger than those of gear two.
A program for simulating the spur gear impacts has been
developed by using the Runge-Kutta method, including a Case 2: Constant Load Operation. In this case, gears one
subroutine DVERK from the IMSL software package. The and two are loaded respectively with constant torques Tt and
program is called SIMULAT and is written in FORTRAN. T2 equal in magnitude but opposite in sign (Fig. 2). At the
Using the developed rotary model of equation (37) and this beginning, a 50 rad/s initial velocity is assigned to gear one.
program, we have investigated the dynamic behaviors of a The simulation result of time versus relative displacement is
pair of meshing spur gears. The design parameters of these given in Fig. 9(a). At time equals about 14.5 ms, there is a
gears are Young's modulus, Ex =E2 = 2.06S x 10 5 N/mm 2 ; small disturbance resulting in a higher relative displacement.
Poisson's ratio, v1 = p2=0.3; radii of gears, rx =20 mm, The disturbance is caused by the switch from two-pair contact
/•2 = 80 mm; pressure angle, <x0=20 deg; number of teeth, to one-pair contact. At 18.9 ms, another small disturbances
Z j = 2 0 , Z 2 = 80; thickness of gears, L = 10 mm; and due to the change from one-pair contact to two-pair results in
backlash, 5 = 0.05 mm. Three cases are studied. They are: lower relative displacement. We believe that a gear impact
free vibration, constant load operation, and sinusoidal ex- model which successfully includes the variation in number of
citation. contact pairs is new. Figure 9(a) shows that this change in
contact pairs also affects angular velocity during the meshing;
Case 1: Free Vibration. First, we study the case of free it periodically introduces small disturbances into the gear
vibration. Figure 8(a) given the relative displacement, S motion, and prevents the system's dynamic response from
plotted against time. The initial velocity of gear one is set at 50 reaching a real steady state.

534/Vol. 107, DECEMBER 1985 Transactions of the ASME

Downloaded From: http://mechanicaldesign.asmedigitalcollection.asme.org/ on 05/22/2015 Terms of Use: http://asme.org/terms

 Case Three: Sinusoidal Excitation. In this study, we code based on the Runge-Kutta method is developed in
applied a sinusoidal torque, T{t) = 100 sin (lOOOt) Nm, to gear FORTRAN. Three cases of simulation studies are performed.
one. The dynamic response of the relative displacement The effects of contact pair variation are clearly shown in the
contains two kinds of vibrations, as shown in Fig. 10(a). The simulation results. In conclusion, we hope that this rotary
driving sinusoidal torque contributes the one with lower model will be useful for studying gear dynamics, partaicularly
frequency and higher magnitude. The one with higher high-speed and intermittent motions.
frequency, similar to the beginning portion of Figs. 8(a) and
References
9(a), is the transient response due to the tooth impact. Figure
10(6) shows the angular velocities of both gears under this 1 Dubowsky, S., and Freudenstein, F., "Dynamic Analysis of Mechanical
sinusoidal excitation. Systems With Clearances, Part 1: Formulation of Dynamic Model," ASME
Journal of Engineering for Industry, Vol. 93, Feb. 1971, pp. 305-309.
2 Dubowsky, S., and Freudenstein, F., "Dynamic Analysis of Mechanical
7 Concluding Remarks Systems With Clearances," Part 2: Dynamic Response," ASME Journal of
Engineering for Industry, Vol. 93, Feb. 1971, pp. 310-316.
In this paper a rotary model for studying gear dynamics has 3 Azar, R. C , and Crossley, F. R. E., "Digital Simulation of Impact
been constructed. This model is quite different from the Phenomenon in Sour Gear Systems," ASME Journal of Engineering for In-
dustry, Vol. 99, Aug. 1977, pp. 792-798.
rectilinear gear impact models in shape, and much closer to 4 Timoshenko and Goodier, Theory of Elasticity, 3rd Edition, 1970.
reality. 5 Werner Goldsmith, IMPACT, The Theory and Physical Behaviour of Col-
The model includes as factors: backlash, elasticity, and liding Solids, Edward Arnord, LTD., London, 1960.
damping during impact. By taking advantages of an involute 6 Hunt, K. H., and Crossley, F. R. E., "Coefficient of Restitution Inter-
preted as Damping in Vibroiimpact," ASME Journal of Applied Mechanics,
tooth profile, we can evaluate the surface compliance of this Vol. 97, June 1975, pp. 440-445.
model. In this study, the relation between the impact force 7 Lee, T. W., and Wang, A. C , 2 "On the Dynamics of Intermittent-Motion
and the inter-penetration of two meshing gears is close to Mechanisms, Part 1: Dynamic Model and Response," ASME Paper
linear, and therefore, the stiffness of Hertzian contact may be 82-DET-64, 1982.
8 Herbert, R. G., and McWhannell, D. C , "Shape and Frequency Composi-
considered constant. We also include in our study the varying tion of Pulses From an Impact Pair," ASME Journal of Engineering for In-
number of contact tooth pair during meshing. A computer dustry, Vol. 99, Aug. 1977, pp. 513-518.

Journal of Mechanisms, Transmissions, and Automation in Design DECEMBER 1985, Vol. 107 / 535

Downloaded From: http://mechanicaldesign.asmedigitalcollection.asme.org/ on 05/22/2015 Terms of Use: http://asme.org/terms