@

Pergsmem
Mech. Mach. Theory Vol. 29. No. 5. pp. 713--723, 1994
Copyright C 1994 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
0094-114)(/94 $7.00 0,0o
A BASIS FOR SOLID MODELING OF GEAR TEETH WITH
APPLICATION IN DESIGN AND MANUFACTURE
RONALD L. HUSTON
University of Cincinnati. Cincinnati. OH 45221-0072, U.S.A.
DIMITRIOS MAVRIPLIS
EDS, Cincinnati, OH 45246, U.S.A.
FRED B. OSWALD
National Aeronautics and Space Administration, Lewis Research Center. Cleveland, OH 44135, U.S.A.
YUNG SHENG LIU
Feng-Chia University. Taichung, Taiwan, R.O.C.
(Receired 21 April 1992: in revised form 13 May 1993: receit,ed f or publication 16 July 1993)
Abst ract --Thi s paper discusses a new approach to modeling gear tooth surfaces. A computer graphics
solid modeling procedure is used to simulate the tooth fabrication processes. This procedure is based on
the principles of differential geometry that pertain to envelopes of curves and surfaces. The procedure is
illustrated with the modeling of spur. helical, bevel, spiral bevel and hypoid gear teeth. Applications in
design and manufacturing are discussed. Extensions to nonstandard tooth forms, to cams. and to rolling
element bearings are proposed.
INTRODUCTION
A difficult task facing analysts and designers ofgeared power transmission systems is understanding
and utilizing the complex geometry of the gear teeth. Even for involute spur gears where the tooth
geometry is generally well understood, the details (e.g., trochoidal geometry and profile modifi-
cations) are neither simple nor accessible to most designers. For helical bevel, spiral bevel and
hypoid gears, the geometry is even more complex so that analyses and designs are of necessity
approximate and empirical. Optimal gear design is thus elusive. Even if the optimal gear tooth
geometry is known, it is distorted under load. Hence, the geometry of meshing gear teeth is
generally less than optimal. Indeed, the geometry of meshing gear teeth under load is virtually
beyond analytical description.
While the details of gear tooth geometry may not be important in many applications, for
precision gears the tooth geometry is the single most important factor influencing transmission
kinematics, gear strength and gear wear. That is, gear tooth geometry has a greater effect upon
the performance, strength and life of gearing systems than any other factor. Accordingly,
comprehensive gearing kinematic, strength and life analyses cannot be conducted without an
accurate representation of the gear tooth geometry.
in recent years, a number of analysts [I-9] have used the procedures of differential geometry and
classical geometrical analyses to describe gear tooth surfaces. These analyses have greatly extended
understanding of surface geometry and contact kinematics. Some of these procedures have been
combined with numerical methods to obtain tooth contact analyses [10-14]. However, there is still
a need for more extensive and more accurate representations of gear tooth geometries if analysts
and designers are to achieve improvements in transmission efficiency, reliability and life. Such
improvements are especially desirable for precision gears meshing under load. Better represen-
tations of gear tooth geometries are also essential for the development of specialized, nonstandard
tooth forms.
Recently, a new approach has been proposed to simulate the fabrication of gear teeth by using
comput er graphics [15, 16]. This approach has led to a procedure for developing solid models of
713
714 R. U Ht~sTos eta/.
gears which in turn can be used for studying meshing kinematics, contact stresses, root stresses and
lubrication. Herein we discuss the analytical basis for the procedure used to develop solid models
of gears. The procedure is illustrated by developing models of spur. helical, bevel, spiral bevel and
hypoid gears.
This report is divided into five sections, the first of which provides some background material
that was needed to develop the procedure for modeling gear tooth surfaces. The succeeding sections
examine involute tooth generation, provide some graphical results, discuss the procedure and
present conclusions.
P R E L I M I N AR Y AN AL Y S I S
Envelopes of curves and surfaces; modeling of surface generation
Consider a plane curve C defined by the equation
y =f ( x , t ) (I)
where t is a parameter. Suppose that C has a typical form as in Fig. 1. Let C be such that it can
be moved (or repositioned) in the plane without being distorted. Let t be a parameter determining
this repositioning. Then as t varies, the locus of positions of C form the family of curves seen
in Fig. 2. The curve E, representing the limiting location of the family, is the "envelope" of the
family.
Suppose that a family of curves is represented in analytical form as
y ~ - f ( x , t ) or F( x, y, t ) - - - O, (2)
where F is continuous with nonvanishing partial derivatives with respect to x and y. Then it is
known that an envelope of the family may be determined by using the relation [17, 18]
OF/Or = O. (3)
That is, the elimination of the parameter t between equations (2) and (3) provides an analytical
description of E, if the envelope exists.
To illustrate the construction of an envelope, consider a family of lines such that each line of
the family is a fixed distance r from a fixed point O, as depicted in Fig. 3. Let ~ define the inclination
of a typical line L of the family and let 0 define the inclination of the line normal to L as shown.
The equation of L may be written in the standard form
Y - Ye = m( x - x, ) , (4)
Fig. I. A plane curve.
Modeling gear tooth surfaces 715
Y
X
Fig. 2. A family of plane cruves and its envelope.
where m is the slope o f L and where (xp, yp) are the coor di nat es o f a poi nt P o f L. If P is t he poi nt
of intersection of L with the nor mal line which passes t hr ough O, t hen m, 0 and ~ are rel at ed by
the expressi on
m = tan ~ = - c o t 0. (5)
Fr om Fig. 3 we see t hat the coor di nat es of P are
x p = r c o s 0 and y p =r s i n O. (6)
Hence, f r om equat i on (4), the equat i on of L may be expressed as
y - r sin 0 = ( - c o t O) ( x - r cos 0). (7)
In the f or m of equat i on (2), this becomes
y sin 0 + x cos 0 - r = F( x , y , O) -- 0. (8)
Equat i on (8) defines a fami l y of lines with 0 being the "f ami l y par amet er ". Then, f r om equat i on
(3), i f we di fferent i at e with respect t o 0, we have
(~F/c~O = y cos 0 - x sin 0 = 0. (9)
Y
I.
I
Fig. 3. A f ami l y of lines equidistant f rom a point.
X
716 R. L. HUSTON et al .
Hence, the equation of the envelope may be obtained by eliminating 0 between equations (8) and
(9). That is, by solving equations (8) and (9) for x and y we obtain
x = r c o s O and y = r si n O. (10)
Finally, by eliminating 0, we have
x2 + y 2 =r 2. ( I I )
This is the equation of the envelope of the family of lines. As expected, it is a circle with radius
r and center O.
The envelope of a cutting tool may be obtained in a similar manner. The cutting tool envelope
is determined by the infinite family of surfaces formed by the tool. Hence, using a procedure based
upon equations (2) and (3). we may simulate surface generation in a manufacturing process. More
specifically, we can simulate and model gear tooth surfaces.
I n r o l u t e o f a ci r cl e
To establish the procedures for modeling gear t oot h surfaces, it is helpful to briefly review the
properties of involute curves. Recall that the involute of a circle may be characterized as the locus
of positions of the end point of a cord being unwrapped form the circle. Figure 4 depicts an involute
I of a circle C. Let P be a typical point on the involute and let #b measure the unwrapping. Then
the position vector p locating P relative to the circle center O may be expressed as
p = x n , + y n , , = r n , + r ~ n ~ ( 1 2 )
where x and y are the horizontal and vertical coordinates of P, r is the circle radius, and n,, n,.,
n, and n, are the horizontal, vertical, radial and tangential unit vectors as shown in Fig. 4. The
unit vectors are then related by the expressions
n , =s i n q) n , +c o s ~n , , and n ~ = - c o s ~ n , + s i n ~ n y . (13)
When a substitution for n, and n~ from equation (13) is made in equation (12), x and y are seen
to be related to r and ~ by the expressions
x - - r s i n ~ - r ~ - c o s ~ and y - - r c o s ~ + r ~ s i n ( k . (14)
Equations (14) are parametric equations defining the involute. The slope of the involute at P
is then
dy/dx = ( d y / d ~ ) / ( d x / d c ~ ) = c ot q). ( I 5 )
I ! r
n x
F i g . 4 . I nvol ute geometry.
Modeling gear tooth surfaces
Fig. 5. A plastic disk rolling over a rigid triangular protrusion.
717
Use of computer graphics
The concept of parametric equations defining the envelope of a family of lines can be used as
a basis for comput er graphics modeling of involute spur gear teeth. To illustrate, consider a
perfectly plastic circular disk rolling in a straight line, as in Fig. 5. Let the disk encounter a rigid
protrusion, such as an isosceles triangle, on the rolling surface. After the disk rolls over the
protrusion, the impression (or footprint) left in the plastic disk represents the envelope of the sides
of the triangle relative to a coordinate system fixed in the disk. This envelope is an involute of a
circle whose radius is equal to that of the rolling disk less the triangle height [15].
This simulation can be used to represent the manufacture of an involute spur gear tooth: let the
rolling disk be replaced by a gear blank rolling on its base circle as seen in Fig. 6. When the rolling
gear blank encounters a cutter in the shape of an involute rack tooth, the impression on the blank
is the gap between the teeth of an involute spur gear. Furthermore (as demonstrated in Refs
[15, 16]), the simulation also defines the trochoidal geometry at the root of the gear teeth.
Figure 6 shows how this procedure was used with comput er graphics software [19] to simulate
a gear blank rolling over a rack cutter.
The analysis of Ref. [15] and the image of Fig. 6 show that comput er graphics can be used to
simulate involute gear tooth manufacture, as with a rack cutter or hob cutter. However, the success
of this simulation raises several questions: (I) Can the simulation be extended to the manufacture
of other t oot h forms, that is, to noninvolute t oot h forms? (2) Can the simulation be extended to
the manufacture of other gear forms, for example, to bevel, spiral-bevel and hypoid gears? (3) Can
a procedure be developed for simulating surface generation in general? (4) is there an analytical
basis for such simulations? We address these questions in the following sections.
ENVELOPE OF A ROLLING INVOLUTE CUTTER
To develop an analytical basis for the graphical simulation, consider first two rolling disks W,
and W 2 whose radii r~ and r2 define pitch circles of mating spur gears (Fig. 7). Let a cutter shaped
like an involute t oot h be placed on W 2 and let Gt be a gear blank placed on W,. The cutter will
leave a gear t oot h impression (or footprint) on the gear blank as WI rolls with W:.
To obtain an analytical representation of the impression, it is helpful to introduce coordinate
axes fixed in W, and IV.,, as shown in Fig. 8. Let ~" - ~ a n d X - Y be Cartesian axes systems fixed
in Wt and W 2 with origins O~ and 02. Let r, be the angle between )( and X as WI rolls on W 2.
Fi g. 6, A gear bl an k r ol l i n g over a rack cutter.
718 R. L. HUSTOr~ et aL
t l X l ~ s l
I I
kt + w, ] /
\ " J .
et s~\
'IL
/ l l \ \ \", - - ~, , , ~,
o~,-'7,1 / \ \ \ \ " " "
) I
! I
I !
Fig. 7. Rol l ing disks simulating meshing gears.
Let L be t he line connect i ng the cent ers O~ and 02. Let 0~ be the angl e bet ween 1:" and L and let
02 be the angle bet ween Y and L. Let P be a typical poi nt in space with coor di nat es ~, .P relative
t o ~" - l ~ and coor di nat es x, y relative to X - Y. Then f r om Fig. 8, these coor di nat es are related
by t he expressi ons
x - - ( r , + r 2 ) si n 02 + .~ cos + .~ si n cc
Y L
F i g . 8. C o o r d i n a t e a x e s i n t h e rolli ng di sks.
Modeling gear tooth surfaces 719
and
a n d
y = (r, + r . ) cos 02 - :~ sin a + ~ cos a
.~ = (r, + r2)sin(', - 02) + x cos - y sin a
(16)
.~ = - ( r t + rz)COS(~t - 02) + x sin ;t + y cos ,t. (17)
Observe t hat . since W t rolls on W 2, ;t and 0: are not i ndependent . Indeed, f r om Fig. 8 we see
t hat
= = 0 , + 0 2 . ( 1 8 )
The roi l i ng condi t i on requi res t hat
rt O, = r20 :. (19)
Hence, by solving for 02 and for a - 0 2 , we have
02 = rl = and = - 0 2 = R2___~ (20)
r l + rz r i + rz
Suppose that the graph of a f unction y = f ( x ) describes the cutter profil e in W 2. Then
f rom equations (16), the cutter profil e may be represented in W, (and, hence, also in gear bl ank
G , ) as
y ( : c , y , = ) = A x ( . L : , = ) ] o r F ( . L. ~ , = ) = 0 . ( 2 1 )
T h e r e p r e s e n t a t i o n o f t h e c u t t e r profi le i n G , t h u s d e p e n d s u p o n the roll a n g l e =. H e n c e , f r o m
e q u a t i o n ( 3) , t h e e n v e l o p e o f t h e c u t t e r profi le i n G , is d e t e r m i n e d f r o m e q u a t i o n ( 2 1 ) a n d the
e x p r e s s i o n
OF(:c, : . a ) I O = = 0. ( 2 2 )
B y su b sti tu ti ng f or x a n d y f r o m e q u a t i o n s ( 1 6 ) i n t o e q u a t i o n ( 2 1 ) a n d b y u s i n g e q u a t i o n ( 20) ,
we have
(r, + r2 ) c os( r i . ~ a '~ - . ~ si n = + f c o s a = J (r, + . r , ) s , n _ _ _ _ + . ~ c o s a + . ~ s i n = . ( 2 3 )
\ r , + r H \ r , r21
Then by per f or mi ng t he di fferent i at i on o f equat i on (22), we obt ai n the expressi on
- rt si n - .~ c o s a - ~ si n a = d-~x d-"~ = r, c o s - .~ si n = + : c o s = . ( 2 4 )
B y u s i n g e q u a t i o n s ( 1 6 ) a n d ( 2 0 ) , e q u a t i o n ( 2 4 ) m a y b e e x p r e s s e d i n t e r m s o f x a n d y a s
d.f
_ . + , . . J ,.. , , . + , , , , , L ', " ( 2 5 )
Suppose t hat the cut t er profile is an i nvol ut e [as in equat i on (14)] with par amet er ~. Then f r om
equat i on (15), d f / d x equal s cot 0 and equat i on (25) becomes
- x + r2 s i n ( r, , '~ == ( c o t , ) [ y - r 2 c o t ( r,. , '~I . ( 2 6 )
\ r , + r z ] \ r l + r , . / I
Mul t i pl yi ng by cos ~ and rearranging terms results in
x s i n O + y c o s O = r 2 c o s ~ r ~ r 2 / ( 2 7 )
O b s e r v e , h o w e v e r , f r o m e q u a t i o n ( 1 4 ) t h a t
x si n ~ + y c o s ~ = r. ( 2 8)
720 R. L Hcs-roN e t al.
Thus, we have
( r,~ ~ r , , (29)
cos 4 ' r ~+- r , / =l or 4 ' =r , +r 2 "
Finally, consider the expression of the involute cutter profile in Ga: by substituting for x and y
from equations (14) into (17) and by using equation (20), we obtain
. /' r i a '~
. = ( r, r , ) s l n ~ , ~ ) r: si n( 4 ' -- a ) - r , 4 ' c o s ( 4 ' -- a) ,
- ( r , + r : ) c o s l ~ ) +r z c o s ( 4 ' - a ) +r 2 sin(4' - a). (30)
~ =
Observe from equation (24) that
\r~ + r2/ \ r d
where the par amet er / / i s defined by the final equality. Hence, by substituting into equation (30),
we have
.? -- r, s i n / / - r , p cosp and .P -- - r l cos/~ - r , p sin//. (32)
by comparing equations (32) with equations (14). we see that they have the same form. Therefore,
the envelope (or footprint) of the involute cutter in G, is itself an involute.
COMPUTER GRAPHICS RESULTS
A computer graphics software system [19] was used to simulate the gear cutting process for
several types of gears. First, to simulate an involute rack cutter, a plastic disk was rolled over a
series of rigid straight-sided protrusions (Fig. 6). The disk was examined and found to have involute
teeth cut into its surface. Finally, a similar procedure was used to generate images of helical, bevel,
spiral bevel, and hypoid gears. Figures 9-12 show the simulations.
DISCUSSION
The discussion presented earlier on the envelope of a family of curves establishes the basis for
a computer graphics procedure for simulating gear tooth generation. The analysis shows, for
example, that the envelope of an involute protrusion is an involute. While this is not a new or
unexpected result (indeed, it is the basis of spur gear fabrication by hobbing), the analysis
establishes that the result is based on the principles of differential geometry.
The computer graphics procedure described in this report provides a realistic simulation of gear
generation processes. The only simplifying assumptions involved are the assumptions of a perfectly
rigid cutting tool and a perfectly plastic workpiece (gear blank).
The significance of this result is that tooth form complexity need no longer be a hindrance to
comprehensive analysis and design; that is, the comput er graphic procedure is unaffected by the
complexity of the tooth form whereas analytical procedures quickly become intractable as the
complexity increases. Hence, the procedure can be applied without modification for the analysis
of bevel gear, hypoid gear, and noninvolute, nonst andard gear tooth forms.
The computer graphics procedure can also be used to define the geometry for a finite-element
model of a tooth form. The combined graphical and finite-element analysis is thus a design tool
for studying stresses and deformations in tooth forms, particularly in the less well understood tooth
forms of spiral bevel and hypoid gears. Noninvolute and nonstandard tooth forms are also readily
accommodated. Indeed, the comput er graphics procedure can be used to modify, develop and
examine new forms of conjugate teeth. It can also be used to study fillet geometry, tip relief, dressing
operations and contact patch geometry.
The computer graphics procedure provides accurate simulations in a short time. For example,
the simulations shown in Figs 9-12, although they have only two or three teeth, are generally
Modeling g~ar t oot h surfaces 721
Fig. 9. Helical gear simulation.
Fig. I0. Bevel gear simulation.
Fig. I I. Spiral bevel gear simulalion.
Modeling gear tooth surfaces 723
sufficient f or st ress anal ysi s; noncont act i ng t eet h have little effect upon the stresses of cont act i ng
teeth. The r un t i me f or a spi ral bevel gear si mul at i on is appr ox. I h on an HP model 350 comput er .
The c omput e r gr aphi cs pr ocedur e can al so be appl i ed t o the manuf act ur e o f t oot h surfaces.
To o t h f or ms f r om ar bi t r ar y cut t er shapes can be predi ct ed. Conver sel y, cut t er shapes pr oduci ng
a desi red t oot h f or m can be det er mi ned. Specifically, t he cut t er profile f or any gi ven t oot h f or m
is the conj ugat e of t hat t oot h f or m. Thi s conj ugat e can be gener at ed usi ng c omput e r gr aphi cs by
rol l i ng a bl ank di sk over t he desi red t oot h f or m. Tha t is, the shape of the cut t er profi l e is the s ame
as t he i mpr essi on left on the bl ank by the desired t oot h f or m.
Fi nal l y, the c omput e r gr aphi cs pr ocedur e can be ext ended t o st udy surface gener at i on in general ,
and appl i cat i ons can be made t o cam design, bear i ng desi gn and surface design of ar bi t r ar y rolling
el ement s. Anal ogous pr ocedur es ma y be devel oped f or si mul at i ng the manuf act ur e of ski n surfaces
such as aut omobi l e fenders, ai rfoi l s and shi p hulls. Indeed, the pr ocedur es ma y be devel oped t o
di rect l y gover n the manuf act ur e of these surfaces.
C ONC L US I ONS
An anal ysi s of the envel opes of fami l i es of curves, based on differential geomet r y, was present ed
as the f oundat i on f or a solid model i ng procedure. Commer ci al comput er gr aphi cs sof t war e was
used t o si mul at e gear t oot h gener at i on. The concl usi ons ar e
I. An anal yt i cal basi s f or numeri cal c omput e r gr aphi c model i ng of gear t oot h
surfaces has been est abl i shed.
2. The c omput e r gr aphi c pr ocedur e can be used f or the model i ng and anal yses of
a wide vari et y of t oot h f or ms, i ncl udi ng s pur gears, helical gears, bevel gears,
spi ral bevel gears, hypoi d gears and nons t andar d gears.
3. The pr ocedur e may be ext ended t o st udy the desi gn and manuf act ur e of rolling
surfaces in general and, ul t i mat el y, skin surfaces.
Acknowledgement--Research summarized in this paper was partially supported by the National Aeronautics and Space
Administration under Grant NSG-3188 to the University of Cincinnati and is gratefully acknowledged.
RE FE RE NCE S
I. M. L. Baxter Jr, J. Mech. Des. tO0, 41-44 (1978).
2. M. L. Baxter Jr, Ind. Math. II, 19-43 (1961).
3. R. L. Huston and J. J. Coy. J. Mech. Des. 103, 127-133 (1981).
4. R. L. Huston, V. Lin and J. J. Coy, J. Mech. Transm. Automn Des. 105, 132-137 (1983).
5. F. L. Litvin, NASA TM-75130 (1977).
6. F. L. Litvin, N. N. Krylov and M. L. Erikhov, Mech. Mach. Theory 10, 365-373 (1975).
7. F. L. Litvin et al., J. Mech. Transm. Automn Des. 109, 163-170.
8. F. L. Litvin, P. Rahman and R. N. Goldrich, NASA CR-3553 0982).
9. Y. C. Tsai and P. C. Chin, J. Mech. Transm. Automn Des. 109, 443-449 (1987).
10. F. L. Litvin and J. Zhang, NASA CR-4224 0989).
II. F. L. Litvin W.-J. Tsung and J. J. Coy, NASA TM-87273 0987).
12. F. L. Litvin et al., SAE Paper 851573 0985).
13. F. L. Litvin, W..J. Tsung and H.-T. Lee, NASA CR.4088 0987).
14. J. A. Remscn and D. R. Carlson, Gleason Publication No. AD 1906, Gleason Works (1968).
15. S. H. Chang, R. L. Huston and J. J. Coy, ASME Paper 84-DET-184 0984).
16. R. L. Huston, D. Mavriplis and F. B. Oswald. Int. Power Transmission and Gearing Conf. 5th ASME, pp. 539 -545
(1989).
17. W. C. Graustein, Differential Geometry Macmillan p. 64 (1935).
18. D. J. Struik, Lectures on Classical Differential Geometry pp. 73, 167. Addison-Wesley, Reading, Mass.
19. I . DEAS (Integrated Design and Engineering Analysis Software). Structural Dynamics Research Corporation (SDRC),
Milford, Ohio 0985).