You are on page 1of 14

Mechanism and Machine Theory 45 (2010) 1707–1720

Contents lists available at ScienceDirect

Mechanism and Machine Theory


j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / m e c h m t

Improving bending stress in spur gears using asymmetric gears and


shape optimization
Niels L. Pedersen ⁎
Dept. of Mechanical Engineering, Solid Mechanics, Technical University of Denmark, Nils Koppels Allé, Building 404, DK-2800 Kgs. Lyngby, Denmark

a r t i c l e i n f o a b s t r a c t

Article history: Bending stress plays a significant role in gear design wherein its magnitude is controlled by the
Received 17 July 2009 nominal bending stress and the stress concentration due to the geometrical shape. The bending
Received in revised form 1 June 2010 stress is indirectly related to shape changes made to the cutting tool. This work shows that the
Accepted 8 June 2010
bending stress can be reduced significantly by using asymmetric gear teeth and by shape
Available online 23 July 2010
optimizing the gear through changes made to the tool geometry. However, to obtain the largest
possible stress reduction a custom tool must be designed depending on the number of teeth,
Keywords: but the stress reductions found are not very sensitive to small design changes. This observation
Gear
suggests the use of two new standard cutting tools.
Asymmetry
Bending stress
© 2010 Elsevier Ltd. All rights reserved.
Tool design
External spur
FEM

1. Introduction

Gear strength is influenced by geometry, the subject of this paper, as well as by material selection and production processes.
Two primary fatigue related failure modes determine gear strength; failure due to bending stress and failure due to contact
pressure. The latter failure is primarily due to pitting while the former is due to tooth breakage. The focus of this work is on
reducing bending stress levels whereby improving gear strength.
Gear design is in most cases conservative and specified by different standards. Almost all gears exhibit involute shape because
the contact forces act along a straight line and a center distance variation due to, e.g. manufacturing tolerances or loadings, does
not influence this fact. A center variation will neither influence the gear ratio. The only design variable that controls the involute
shape is the pressure angle α; it is typically assigned the value α = 20°. Only the gear region that is in contact with the other gear in
the mesh is described by the involute shape. The root geometry or bottom land region, that connects two neighboring teeth, can be
designed rather freely. The task of this paper, also done in Ref. [1], is to improve the gear strength by changing the gear geometry in
a way that retains the involute shape.
Design changes of the gears are achieved indirectly by redesigning the cutting tool. Cutting tool parameterization includes the
possibility of an asymmetric tooth; it is simple as it only requires four design parameters. Resulting optimized designs show that a
significant reduction in the bending stress is possible. Furthermore, the cutting tool shape is described analytically and hence so is
the cut teeth shape Thus, as indicated in Ref. [1], the maximum stress calculations can be trusted.
The following Section 2 discusses the general aspects of gear design with a special focus on the asymmetric design and relates
the present work to existing works. A discussion of the herein used finite element analysis is also provided. Geometric
parameterization of the cutting tool and the analytical shape of a gear cut with this are resented in Sections 3 and 4. Section 5

⁎ Tel.: + 45 45255667; fax: + 45 45254250.


E-mail address: nlp@mek.dtu.dk.

0094-114X/$ – see front matter © 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.mechmachtheory.2010.06.004
1708 N.L. Pedersen / Mechanism and Machine Theory 45 (2010) 1707–1720

presents optimized designs of spur gears with different number of teeth. These results lead to the two new suggested standard
cutting tools presented in Section 6.

2. General aspects of gear design and analysis

Contact zone geometry is totally controlled by the involute shape and the only way to reduce the contact stress (Hertzian) is to
increase the pressure angle, when keeping the width constant. Pressure angle increase will increase the force on the surface under an
assumption of constant transmitted torque, however, the curvature radius is increased so altogether a reduction in the surface stress is
the result. A disadvantage of increasing the pressure angle is a reduction in the contact ratio. With standard ISO teeth the bending
stress is also reduced with increasing pressure angle. Again the load on the teeth is increased but at the same time the root width of the
teeth is increased, so in total a reduction in the bending stress is the result. With a pressure angle decrease, the primary advantage is the
increase of the contact ratio and a possible increase of the tooth height that will have a positive influence on the noise level. A
disadvantage is an increase in the maximum teeth sliding speed that has a negative influence on the lubrication. A further listing of
advantages and disadvantages of increasing or decreasing the pressure angle can be found in Ref. [2].
Almost all gears are symmetric and defined according to the standard cutting tool. The cutting tool definition used in the
present paper is based on the ISO profile and seen in Fig. 1.
The shown profile has, as the ISO profile, an added top with the height of M/4, where M is the gear module which controls the
gear teeth size and subsequent also the gear size (the pitch diameter dp is given by dp = Mz where z is the number of teeth on the
gear). Top radius ρ is chosen such that there is no jump in the slope. The bottom of the true cutting tooth profile is not identical to
the bottom of the shown cutting profile based on the ISO profile. For the real cutting profile, the top of the cut teeth is assumed
given by the initial steel blank diameter, which is equal to the addendum diameter. The shown profile has as envelope the full cut
tooth, i.e., the envelope of the bottom part of the profile is the finished cut tooth top. Teeth cut with the ISO profile and teeth cut
with the profile shown in Fig. 1 are therefore identical.
Gears become symmetric when the cutting tool tooth is symmetric with respect to the y-axis as defined in Fig. 1. Symmetry of
the two involutes of a tooth follows a choice of identical pressure angles, αd = αc. Subscript d is used for drive side and subscript c is
used for coast side. Two identical pressures angles imply that the two straight lines have opposite gradient, and that they go
through the points (−πM/4, 0) and (πM/4, 0) respectively (the envelope of the straight side are the tooth involute).
If the object of gear design is to minimize the stresses, it follows from the listed advantages and disadvantages of changing the
pressure angle that the pressure angle should be as large as possible. The limiting factor is the needed minimum tooth top thickness.
In many applications the gears only work in a unidirectional way, or not with the same loading conditions in the two opposite
directions. From this follows that it might be an advantage to use different pressure angle values, i.e., an asymmetric gear design.
This idea is not new and can be found in, e.g. Refs. [3,4], the two papers also mention earlier work done in the Soviet Union. The two
papers are short (2 pages) and the information primarily relates to basic geometry. In the recent years the subject has gained an
increasing interest this can be seen from the number of papers published. The papers, Refs. [5–9] show that the use of asymmetric
gear teeth has a big potential.
More information with regard to design of asymmetric gear teeth can be found in e.g., Refs. [10–20]. The present paper differs
from these papers by the way the shape optimization is performed. It can be difficult to see directly what is gained by using
asymmetric gears. Primarily this is due to the nature of gear design; it has many variables and these can be varied in many ways. In
Ref. [19] the results show that the best design with respect to bending stress is the symmetric design but this paper kept the

Fig. 1. Cutting profile geometric definition and the basic profile based on the ISO profile, M is the gear module that defines the teeth size in the gear. The two sides
of the tool are termed drive and coast side respectively. Pressure angles αd and αc are shown here with the same value. The coordinate system used is shown.
N.L. Pedersen / Mechanism and Machine Theory 45 (2010) 1707–1720 1709

pressure angle at the input or drive side to the standard αd = 20° and limits the pressure angle on the opposite or coast side to be
αc≦20°. If we allow for a larger coast side pressure angle the bending stress becomes smaller as shown in the present paper.
In many papers on asymmetric tooth design, the cutting tool top is a circle, see e.g. Ref. [18]. With this choice of design, the full
optimization with respect to bending stress is not found. The positive effect on the bending stress is then solely related to a root
thickness increase of the tooth, but the stress can be further reduced using shape optimization to reduce the stress concentration.
In the papers on asymmetric tooth design there seems to be two choices: either αd b αc, i.e., the drive side pressure angle is
smaller than the coast side pressure angle or αd N αc. In Ref. [10] the choice made is αd N αc. It is stated that this choice is made
because it reduces the mesh stiffness that has a positive influence on noise and vibration levels. In Ref. [14] the choice made is
αd b αc and here it is shown that this only has a little influence on the mesh stiffness. The present paper does not discuss mesh
stiffness and both choices of relative pressure angles are shown in the examples in order to explore the possible advantages.
When increasing pressure angles the negative result is that the top land thickness becomes smaller and in the limit it becomes
pointed, with a further increase in the pressure angle the teeth become shorter. A shorter teeth decrease the contact ratio, which is
undesirable. In Ref. [21] the minimum top land thickness limit is sa ≧ 0.25 M, for carburized gears the limit is reported to be higher
sa ≧ 0.4 M.
Shape optimization of gear teeth can be done in two principal different ways. The direct and most simple way is to optimize the
tooth root directly as it is done in e.g. Refs. [6,22]. Another way of optimizing is changing the shape of the tool that cuts the shape.
The latter method is used here, the result being that the tool becomes a custom tool for the specific gear (number of teeth). This
method of optimization was also used in Ref. [1], but the method is in the present paper extended to include asymmetric teeth
design. In the optimization the focus is on the bending stress because this is the only limiting stress that is affected by the design
changes made in the bottom land. Surface stresses are only affected, as stated previously, indirectly through the choice of pressure
angle. In the different optimizations presented in the present paper, the pressure angles are selected within limits controlled by
the top land thickness and as such, the surface stresses are not the optimization objective.
From shape optimization, see e.g. Ref. [23], we know that the shape parameterization is of great importance and the node
positions in the finite element parameterization should not be used as design parameters directly. Instead the cutting tool tip is
analytically parameterized using a super elliptical shape. It is also known from shape optimization for minimum stress
concentration that the primary first design criterion is that the slope is continuous, i.e., a corner may imply infinite stress.

2.1. FE modeling

An assumption of plane stress is made in the present paper since the optimizations are made for external spur gears. The
Poisson's ratio used is ν = 0.3 and linear elasticity is assumed. The geometry, the loads and the supports need to be specified for the
FE modeling. In Fig. 2 a schematic drawing of a rack tooth is presented.
The geometry is given by the cutting tool design, see Section 3. The load is acting perpendicular to the surface. In reality the load
does not act as a single load as indicated in Fig. 2, but is spread out as given by the Hertzian pressure distribution and could be used
in the FE simulations. We apply a constant line load that is perpendicular to the surface because this study is related to the root
stress and not to the contact stress. The difference in the maximum bending stress from applying a constant line force or a force
that varies according to Hertz is negligible. Actual load size is not important because of the assumption of linear elasticity.
Support placements are a compromise. The tooth is fixed at a depth of size M and at two symmetry lines to the two adjacent
teeth, see Fig. 2. If the support is moved to the tooth root it would comply with the Lewis formula for bending stress calculation, see
e.g. Ref. [24]. By moving the support closer to the gear center the tooth becomes more flexible. Fixing the tooth at a depth larger
than M has a negligible influence on the stress we want to minimize, i.e., the maximum bending stress at the root.

Fig. 2. Schematic drawing of one rack tooth.


1710 N.L. Pedersen / Mechanism and Machine Theory 45 (2010) 1707–1720

3. Geometry parameterization

The basic ISO profile has an added top with a height of M/4 as seen in Fig. 1. The top on the basic profile is added to make a
clearance for the lubricating oil. It is however also the top that controls the bottom shape of the gear teeth. This is the case which
can be seen e.g. for the rack where the teeth are the counter part of the profile as shown in Fig. 1. In Ref. [1] the tool tip is the design
domain because the symmetric design was kept and the focus was on the stress concentrations at the tooth root. Root stress is
directly controlled by the tool tip shape.
In the present paper the parameterization used in Ref. [1] is extended to include asymmetric cutting teeth that give asymmetric gear
teeth. The straight side of the cutting tool is no longer fixed to have the pressure angle α but the idea is to change this value without
interfering with a lower constraint on the tooth top thickness(top land). This leaves most of the parameterization to the tool top.
In order to find the limits to the pressure angles a relation between the pressure angles and the top land thickness is needed. A
derivation is given in Ref. [10] and an alternative derivation is given here. Since the lengths are different on the two sides we
distinguish between drive side and coast side. Tooth thickness at the pitch diameter is given by

s = spd + spc ð1Þ


where spd and spc are the tooth pitch thickness at the pitch diameter for the drive and coast side, respectively. These are given by
 
1
spd = π + ps tanðαd Þ M ð2Þ
4
 
1
spc = π + ps tanðαc Þ M ð3Þ
4

where ps is a possible cutting tool shift. This shift is in many cases not identical on the two mating gears. The lengths related to the drive
side are indicated in Fig. 3 where we have used the known involute function definition given by inv(α) = tan(α) − α.
In Fig. 3 only the part that relates to the coast side is shown for clarity. A similar figure can be made for the drive side, the only
difference is that the base diameter is different, in the shown case the base radius for the drive side rbd is smaller than the base
radius for the coast side rbc.
At the top diameter dt = 2rt the tooth thickness for the drive and coast side is given by
!
dt s
std = + invðαd Þ−invðαtd Þ ð4Þ
2 dp

!
d s
stc = t + invðαc Þ−invðαtc ÞÞ ð5Þ
2 dp

Fig. 3. Illustration of the tooth top thickness for an asymmetric tooth related to only one side, here the coast side. The coast side pressure angle is here αc = 20° and
the drive side pressure angle is αd = 30° for a gear with 17 teeth.
N.L. Pedersen / Mechanism and Machine Theory 45 (2010) 1707–1720 1711

It should be noted that one of the length std or stc might be negative, indicating that all of the tooth top lies to one side relative to the
center line, see Fig. 3. Total tooth top length is given directly by

st = std + stc ð6Þ

It remains to determine the drive top angle, αtd, and the coast top angle, αtc, these are given by

dp
cosðαtd Þ = cosðαd Þ ð7Þ
dt

dp
cosðαtc Þ = cosðαc Þ ð8Þ
dt

From the input:


1. Module M.
2. Shift values for the two gears ps1 and ps2.
3. Number of teeth on the two gears z1 and z2.
4. A given pressure angle, αd or αc.
5. A lower limit on the tooth top thickness, e.g., st = 0.25 M.
The limiting value of the other pressure angle can be found by the use of Eqs. (1)–(8) and, e.g., Newton–Raphson iterations. In
the examples shown in the present paper either the drive pressure angle or the coast pressure angle is set to the standard value 20°
while the other pressure angle is changed.
The parameterization that remains is the tool top, the number of possible tool top design parameterizations is infinite. Here a
variation of the parameterization used in Ref. [1] is applied, the central part is to use a variation of the super ellipse. Focus is on
simplicity, although the optimization result should still be near to the optimal design. That a given parameterization is sufficiently
flexible, i.e. that it can return optimal designs, can only be checked or verified after an actual optimization procedure. If the stress is
constant along major parts of the surface then the shape is assumed optimal, see e.g. Ref. [25].
The parameterization presented fulfills the following constraints:
• The added tool tip height is fixed at M/4.
• The involute part of the tooth must not be penetrated on the drive side.
That the tool tip height is kept fixed is applied in order to allow for the same clearance in the optimized gears as is the case for
the ISO gears. The involute part should be kept unchanged to allow the optimized gears to have the same functional qualities as the
original involute gears.
A distinction is made between the tool top part that cuts the tooth root of the drive side (drive top) and the other part that cuts
the tooth root of the coast side (coast top). As indicated in Fig. 4 the coast side top is a simple circle (part of a full circle).
The radius of the circle is given as

4μ + π−5tanðαc Þ
ρc = κM = M ð9Þ
4ðcosðαc Þ−ðsinðαc Þ−1Þtanðαc ÞÞ

and might be greater than or smaller than the ISO standard ρ ≈ 0.38 M (see Fig. 1). This also means that the involute on the costs
side might not be as long as it would have been using the ISO cutting tooth, but this is ignored because of the unidirectional loading
assumption.

Fig. 4. The design domain for the optimization shown as the hatched part. The coast side pressure angle and drive side pressure angle are shown together with the
circle radius on the cutting tool coast side.
1712 N.L. Pedersen / Mechanism and Machine Theory 45 (2010) 1707–1720

Fig. 5. The tooth top parameterization that cuts the drive side tooth root, here shown for a positive value of μ. a) The parameterization of including a super ellipse is
shown to be also outside the design domain. b) The super elliptical shape is forced back into the design domain by a distortion.

Final part to be parameterized is the drive top, this is done by a modified super elliptic shape. The design domain is shown as
the hatched part in Fig. 4 and enlarged in Fig. 5. As seen in Fig. 4 the design domain size is variable and controlled through the
parameter μ, with the restrictions from the boundaries this parameter must fulfill.

π 5 π 5
μ min = − + tanðαc Þ≦ μ ≦ − tanðαd Þ = μ max ð10Þ
4 4 4 4

From the optimization presented in Ref. [1] it was found that in order to minimize the stress concentration it is important that
the parameterization includes a straight part before entering the elliptical shape, but in that paper the tooth were symmetric. The
idea used in the present paper is instead that the design domain can change size through the design parameter μ .
The remaining top part is as indicated in Fig. 5a parameterized by a super elliptical shape, only the first quarter of the super
ellipse is used. Parametric form of the super ellipse is
 
ð2 = ηÞ ∘
x = a0 + a1 cosðtÞ M; t∈½0 : 90  ð11Þ
 
ð2 = ηÞ ∘
y = b0 + b1 sinðtÞ M; t∈½0 : 90  ð12Þ

where the constants are given by

π  1
a0 = μ; a1 = −tanðαd Þ−μ ; b0 = 1; b1 =
4 4

As indicated in Fig. 5a the super ellipse might potentially come outside the design domain, which is not wanted since this has
an influence on the length of the involute of the cut tooth. To move the super ellipse back a distortion is added to the x position
parameterization. The distortion is indicated in Fig. 5b by rotating the dashed line. The quarter distorted super ellipse
parameterization is given by
  
ð2 = ηÞ b ð2 = ηÞ ∘
x= a0 + a1 cosðtÞ 1− 1 tanðαd ÞsinðtÞ M; t∈½0 : 90  ð13Þ
a1
 ð2 = ηÞ


y = b0 + b1 sinðtÞ M; t∈½0 : 90  ð14Þ

Using the parameterization given by Eqs. (13) and (14) it is possible both to achieve the design space upper limit by letting
η → ∞ or the lower boundary by letting η → 0. The given parameterization fulfills that the gradient/slope is continuous, i.e., no
jumps in the slope if η≧1.
The presented total cutting tool tooth parameterization is in principle controlled by four parameters; the two pressure angles
αd and αc, the length parameter μ and the super elliptic power η. As previously stated one of the pressure angles is assumed given
so the optimizations presented in Sections 4 and 5 are parameter studies with only three parameters, μ, η and either αd or αc. It is
shown that this simple parameterization is sufficiently flexible to achieve constant stress along a major part of the root.
N.L. Pedersen / Mechanism and Machine Theory 45 (2010) 1707–1720 1713

4. Analytical description of the teeth shape

In shape optimization it is important to have a detailed or preferably analytical shape description. Analytical description also
makes verification and comparison possible for other designs. Another reason is that it is known from shape optimization (see e.g.
Ref. [23] and references therein) that we cannot use the nodes of the FE model as design parameters. In the present paper, we have
made an analytical cutting tool parameterization and it is possible to find analytical descriptions for the envelope of the
parameterizations in case of a gear with a finite number of teeth. This might not be as easy when using e.g. splines to parameterize
the tool tip.
In Ref. [1] it is shown as to how the envelope can be found for two different cutting tooth parameterizations. The
parameterization in the present paper is not identical and for this reason and for easy reference a method for finding the envelope
analytically is presented.

4.1. Envelope of cutting tool

The cutting tooth is made by connecting curve segments. There are three different curves; distorted super ellipse, circle and
straight line. The envelope of these three curves is found separately although the circle is a special case of the distorted super
ellipse.
Gear teeth are cut by a simultaneous cutting tool movement and gear blank rotation, the same geometry is achieved by a
combined rotation and translation of the cutting tool. Rotation and translation are controlled by the angle (variable) θ.
All curves are described in parametric form, { f(t)·M, g(t)·M}T, rotating and translating this around the gear blank give the
following parameterization
    
xðt;θÞ cosðθÞ −sinðθÞ f ðtÞ−θ⋅z = 2 ∘
=M ; t∈½0 : 90 : ð15Þ
yðt;θÞ sinðθÞ cosðθÞ gðtÞ−z = 2−ps

where ps is a possible profile shift factor. Parameterization (18) is a whole set of curves controlled by the two parameters, t which
describes the position along the curve and θ that describes the curve translation and rotation.
The envelope of Eq. (18) is determined by differentiating Eq. (18) with respect to t and θ and demanding that these two vectors
are aligned, i.e., that

dxðt;θÞ dyðt;θÞ dyðt;θÞ dxðt;θÞ


− =0 ð16Þ
dθ dt dθ dt

By solving (19), θ(t) is determined as a function of t. This function can then be put back into Eq. (18) to give the analytical
description of the envelope in parametric form.
A straight segment between two points (β1M, γ1M) and (β2M, γ2M) has the parametric form.
    
xðt;θÞ cosðθÞ −sinðθÞ ð1−tÞβ1 + tβ2 −θ⋅z = 2
=M ; t∈½0 : 1 ð17Þ
yðt;θÞ sinðθÞ cosðθÞ ð1−tÞγ1 + tγ2 −z = 2−ps

Applying Eq. (19) for the straight segment we find

ðβ1 −β2 Þ2 + ðγ1 −γ2 Þ2 β ðβ −β2 Þ + ðγ1 −ps Þðγ1 −γ2 Þ


θðtÞ = 2 t+2 1 1 ; t∈½0 : 1 ð18Þ
ðβ2 −β1 Þz ðβ1 −β2 Þz

The circle on the coast side has the parametric form.


    
xðt;θÞ cosðθÞ −sinðθÞ μ + κcosðtÞ−θ · z = 2 ∘ ∘
=M ; t∈½90 : 180 −αc  ð19Þ
yðt;θÞ sinðθÞ cosðθÞ 5 =4−κ + κsinðtÞ−z = 2−ps

Applying Eq. (19) for the circle we find

4μ + ð4ðκ + ps Þ−5ÞcotðtÞ ∘ ∘
θðtÞ = ; t∈½90 : 180 −αc  ð20Þ
2z

Parametric form for the distorted super ellipse can be given as


   ( )
xðt; θ cosðθÞ −sinðθÞ a0 + C ð2 = ηÞ K−θ⋅z = 2 ∘
=M ; t∈½0 : 90  ð21Þ
yðt; θÞ sinðθÞ cosðθÞ b0 + b1 Sð2 = ηÞ −z = 2−ps

where K = a1 − b1 sin(t)(2/η)tan(αd), C = cos(t) and S = sin(t).


1714 N.L. Pedersen / Mechanism and Machine Theory 45 (2010) 1707–1720

Fig. 6. The envelope of the different curve segments and the trajectory of intersecting points. The final tooth is indicated by the gray area.

Applying Eq. (16) the result is not as simple as the previous examples but can be determined analytical as
 
θðtÞ = 2 b1 C 2 Sð2 = ηÞ ðps −b0 −b1 Sð2 = ηÞ Þ + a0 b1 C ð2 + 2 = ηÞ Sð2 = ηÞ tanðαd Þ + a0 C ð2 = ηÞ S2 K + C ð4 = ηÞ S2 K 2 + b1 C ð2 + 4 = ηÞ Sð2 = ηÞ tanðαd ÞK
 
zC=ð2 = ηÞ 2
a1 S + b1 S
ð2 = ηÞ 2 2
tanðαd ÞðC −S Þ ; t∈½0 : 90 

ð22Þ

Finally if η b 1 then the slope is no longer continuous and in this case we need the trajectory of a point. For a general point, (β1M,
γ1M), the trajectory is given directly by
    
xðtÞ cosðtÞ −sinðtÞ β1 −t · z = 2
=M ð23Þ
yðtÞ sinðtÞ cosðtÞ γ1 −z = 2−ps

The envelopes of the different cutting tool segments are now determined. Then bookkeeping must be applied to find the overall
envelope of all the envelope parts. An example is given in Fig. 6.
The curves given in Fig. 6 are given in the specific case where η = 2, μ = 0, αd = 30° and αc = 20° and the number of teeth on the
gear is z = 17. The final tooth after the bookkeeping is presented in Fig. 7, also shown are the boundary conditions for the FE
analysis. The curves in Fig. 7 are all analytically determined but divided in segments as is the cutting tool that created them.

5. Optimization of general spur gear

As discussed in Ref. [1] the design that minimizes the maximum bending stress in the tooth will depend on the position of the
external loading on the tooth. Therefore in order to compare different designs the load must be applied in the same position. In
gear design an important quantity is the contact ratio, i.e., the average number of teeth in contact between two mating gears.
Normally there are either one or two gears in contact at the same time. In the present paper nothing is done to change the contact
ratio other than the influence that is related to the pressure angle at the drive side, i.e., all teeth are cut with a rack tooth that have a
height of 2.25 M. Loading on the tooth is at its highest value when there is only one tooth transmitting the torque. At the same time

Fig. 7. The combined tooth from the curves in Fig. 6 with the used supports in the FE modeling indicated.
N.L. Pedersen / Mechanism and Machine Theory 45 (2010) 1707–1720 1715

Fig. 8. The ISO profile of a tooth for a gear with 17 teeth. Iso-lines and gray scale of the largest principal stress. In the plot the gray scale shows tension only while
compressive larger principal stresses are white.

the stress will be highest when the load is closest to the tip, the worst point is often referred to as the outer limit of single tooth
contact. The position of this point varies depending on both gears in the mesh. For consistency and for easy reference the choice
made here is, as in Ref. [1], that the tooth is loaded at the pitch point where we have rolling contact. The load is not put at a single
node but applied as a constant line load symmetrically around the pitch point.
In all examples presented in this section the starting point is the ISO tooth with αd = αc = 20°, the load size is scaled so that the
maximum of the largest principal stress is unity. To compare the bending stress of the optimized asymmetric teeth to the ISO tooth
the transferred torque is kept constant, i.e., the load size on the optimized tooth is scaled relative to the ISO tooth load.
First example is with 17 teeth, i.e., z = 17, and we are at the limit of under-cutting. In Fig. 8 a plot with iso-lines and gray scale of
the largest principal stress is shown. The plot only shows the stress at points where the numerical largest principal stress is
positive, i.e., where there is predominating tension. If the numerical largest principal stress is negative, i.e., there is predominating
compression, the color is white. In Fig. 8 the external loading on the tooth is also shown together with the reaction forces at the
clamped boundaries.
The load is scaled such that the maximum bending stress is unity, which is illustrated, by the scale in Fig. 8. Fig. 9 shows a close-
up of the stress concentration zone of the ISO tooth. Fig. 9b shows the iso-lines of the largest positive principal stress as in Fig. 8 but
now without the gray scale. Fig. 9a shows the size of the largest principal stress along the part of the boundary where the stress
concentration is present. The stress size is indicated by the gray area, the perpendicular thickness of the gray area corresponds to
the stress level. A tensile stress is plotted under the boundary for illustrative purposes.
From the stress plot in Fig. 9a it can be seen that there is a potential for improving the stress. However, the ISO tooth does have a
rather nice stress distribution along the boundary, so the room for improvement through only shape optimization is limited. This
was done in Ref. [1] where the best design for a gear with 17 teeth gave a stress reduction of 12.2% compared to the ISO tooth. With
the asymmetric design we can also improve the stress by increasing the tooth root thickness.
Fig. 10 is similar to Fig. 8. The design is optimized through a parameter study. Fixed value of this optimization is the coast
pressure angle αc = 20°.

Fig. 9. Close-up of the stress concentration zone for the ISO gear.
1716 N.L. Pedersen / Mechanism and Machine Theory 45 (2010) 1707–1720

Fig. 10. Full view of optimized tooth. The designed tooth is for a gear with 17 teeth, the optimized design variables are αc = 20°, αd = 35°, η = 1.87 and μ = μmin. The
stress is reduced with 39.2% as compared to the ISO profile.

Fig. 11. Close-up of the stress concentration zone for the optimized gear (αc = 20°, αd = 35°, η = 1.87 and μ = μ min). The stress is reduced with 39.2% as compared to
the ISO profile.

The optimized design in Fig. 10 has the following design parameters αc = 20°, αd = 35°, η = 1.87 and μ = μ min. Stress is reduced
with 39.2% as compared to the ISO profile. In Fig. 11 a close-up of the interesting tooth part is given. Since this optimization is a
parameter study with only three design parameters there might be room for improvement with a more complicated boundary
parameterization. From Fig. 11a it is, however seen that the stress is constant over a long part of the boundary so any optimization
relative to this design must give minor changes. The stress scale in Figs. 9 and 11 is the same so the reduction in the stress level is
directly visualized.
The price we pay in this design relative to the original ISO tooth is a smaller contact ratio but this can be fixed through a possible
longer tooth since the tooth top thickness is not near the limit. Alternatively to this we fix the drive side pressure angle at αd = 20°
resulting in the same contact ratio as the ISO gear. Result of this optimization is presented in Figs. 12 and 13.
The optimized design in Fig. 12 has the following design parameters αc = 34°, αd = 20°, η = 1.57 and μ = 0.07. Stress is reduced
with 23.2% as compared to the ISO profile. In Fig. 13 a close-up of the interesting tooth part is given.
Overall the improvements in the bending stresses are large. The largest stress improvement is possible with αd N αc, here we find
almost twice the improvement found when αd b αc. The improvement of 39.2% and 23.2% should be compared to the result in Ref. [1]
where the best design gave a stress reduction of 12.2%, so the influence from the enlarged tooth root thickness is clear. The choice of αd N αc
is similar to what is found in e.g. Ref. [10] while αd b αc corresponds to the choice in e.g. Ref. [14]. With the latter choice the contact ratio is
constant but some articles report that the teeth becomes too stiff while with the first choice the contact ratio goes down. It is not the
present paper's intent to comment on these matters but instead focus directly on the possible bending stress improvements.
In Ref. [1] it was found that the size of the possible stress improvement depends on the number of teeth. The largest
improvements were found when the number of teeth is low. To examine if this is also the case for an asymmetric gear and for a
direct comparison with the results reported in Ref. [14] the next examples are with a gear with z = 34. Optimization results are
presented in Fig. 14.
In Fig. 14a the optimized result for fixed coast side pressure angle αc = 20° shows that the design parameters are αc = 20°, αd = 36°,
η = 1.8 and μ = μ min. Improvement in the bending stress as compared to the ISO profile is 41.2%. In Fig. 14b the result for fixed drive side
N.L. Pedersen / Mechanism and Machine Theory 45 (2010) 1707–1720 1717

Fig. 12. Full view of optimized tooth. The designed tooth is for a gear with 17 teeth, the design parameters are αc = 34°, αd = 20°, η = 1.57 and μ = 0.07. The stress is
reduced with 23.2% as compared to the ISO profile.

Fig. 13. Close-up of the stress concentration zone for the optimized gear (αc = 34°, αd = 20°, η = 1.57 and μ = 0.07). The stress is reduced with 23.2% as compared
to the ISO profile.

pressure angle αc = 20° shows that the design parameters are αc = 34°, αd = 20°, η = 1.7 and μ = 0.09. Improvement in the bending
stress as compared to the ISO profile is 20.1%. In both examples the stress along the boundaries of the stress concentration is relatively
constant indicating that an optimal or close to optimal result is obtained with this simple cutting tool parameterization. In Ref. [14] the
reported improvement in the bending stress is 17% which can be compared directly with the 20.1% found here. Close agreement is
found although the results in [14] were found without using shape optimization. It is also noticed that in contradiction to the results
from optimizing only the root shape of the standard tooth the results are better for a gear with more teeth when αd N αc. However, the

Fig. 14. a) Close-up of the stress concentration zone for the optimized gear with 34 teeth, the design parameters shown are αc = 20°, αd = 36°, η = 1.8 and μ = μ min.
The stress is reduced with 41.2% as compared to the ISO profile. b) Close-up of the stress concentration zone for the optimized gear with 34 teeth, the design
parameters shown are αc = 34°, αd = 20°, η = 1.7 and μ = 0.09. The stress is reduced with 20.1% as compared to the ISO profile.
1718 N.L. Pedersen / Mechanism and Machine Theory 45 (2010) 1707–1720

Fig. 15. a) Close-up of the stress concentration zone for the optimized gear with 68 teeth, the design parameters shown are αc = 20°, αd = 36°, η = 1.94 and
μ = μmin. The stress is reduced with 44.3% as compared to the ISO profile. b) Close-up of the stress concentration zone for the optimized gear with 68 teeth, the
design parameters shown are αc = 35°, αd = 20°, η = 1.81 and μ = 0.1. The stress is reduced with 19.4% as compared to the ISO profile.

improvement in the bending stress is of the same order as with z = 17. To verify this we finally optimize a gear with z = 68.
Optimization results are presented in Fig. 15.
Results are of the same order for this gear. Improvements in the bending stress found are 44.3% and 19.4%, respectively.
From the performed parameter studies it is found that the reduction in the bending stress is not very sensitive to small changes
in the design parameters. This leads to the idea of a standard or two standard cutting racks, these are presented in the next section.

6. Suggested new standard asymmetric rack cutter

From the optimized designs presented in the previous section, specifically the design parameter values, it seems that it is
possible to make two standard rack cutters, one where the drive side pressure angle is fixed at αd = 20° and another where the
coast side pressure angle is fixed at αc = 20°. The suggested design variables, for the cutters, are given in Table 1.
Cutter designs are shown in Figs. 16 and 17.
Using a standard cutter for all gears, i.e. a non custom cutter, does result in designs that are not fully optimized. The differences
are, however, not large as seen in Fig. 18.
From the results in Fig. 18 the following conclusions can be made.

• The largest reduction in the bending stress can be found with αd N αc.
• With a drive side pressure angle, αd = 36° (AE1), the bending stress reduction compared to the standard ISO tooth is about 40%
independent of the number of teeth on the gear. Maximum difference as compared to an optimized tooth is 3%.
• With a coast side pressure angle, αc = 34° (AE2), the bending stress reduction compared to the standard ISO tooth is about 18%
independent of the number of teeth on the gear. Maximum difference as compared to an optimized tooth is 5%.

As expected the performance of the suggested two new standard rack cutters is not as excellent as the specific optimized cutters for
a gear with a given number of teeth. However, the difference is not large and the reduction in the bending stress compared to the ISO
tooth is still significant. A decision whether to use a standard cutter or a custom cutter must be problem dependent.

7. Conclusion

Results presented in the present paper show that large improvements in the bending stress for gears can be found by the use of
asymmetric gears. Bending stress reduction is achieved by two contributions, a thicker tooth root and a root shape change where
we have the stress concentration. The factor that has the largest influence here is the enlargement of the root thickness.
Optimization has been exemplified by three gears with the number of teeth being z = 17, z = 34 and z = 68, respectively.
The cutting tool is designed so the root shape optimization of the gear tooth is achieved in an indirect way. However the
changes made to the cutting tool are directly related to the actual gear tooth. The design parameter choice of the optimization has
been that either the coast side pressure angle αc or the drive side pressure angle αd is fixed at 20°. Maximum reported reduction in
the bending stress is 44.3%, in this case we have a custom cutting tool specifically for a gear with 68 teeth. The paper proposes the

Table 1
Design variables for suggested standard asymmetric elliptic cutters.

Asymmetric elliptic cutter αd αc η μ


π 5
AE 1 36° 20° 2 μ min = − + tanð20∘ Þ
4 4
π 5
AE 2 20° 34° 2 μ min = − + tanð34∘ Þ
4 4
N.L. Pedersen / Mechanism and Machine Theory 45 (2010) 1707–1720 1719

Fig. 16. New standard asymmetric elliptic cuter AE1.

Fig. 17. New standard asymmetric elliptic cuter AE2.

Fig. 18. The reduction in percent of the bending stress, σb, for the new standard cutter and the optimized cutters relative to the standard ISO teeth. For the number
of teeth equal to z = 17, z = 34 and z = 68 respectively.

use of a standard rack or two standard racks. Reduction in the bending stress is lower than that reported for specific optimizations,
but the difference is not significant, especially for gears with a higher number of teeth.
Increasing the pressure angle on the drive side also leads to some negative effects that must be considered as discussed in
Section 2, and in the literature. The focus of the present paper is entirely on the bending stress. In a practical application these
effects should however be included.
Overall the paper has demonstrated that with a simple rack cutter parameterization with only four parameters (only three are
used actively in the presented optimizations) we can reduce the bending stress rather significantly. With a high drive side pressure
angle the bending stress improvements is in the order of 40% independent of the number of teeth on the gear, with a high coast
side pressure angle the improvement is roughly half the size. This also holds for the two new suggested standard rack cutters.

Acknowledgment

For the discussions and suggestions I wish to thank Prof. Pauli Pedersen.
1720 N.L. Pedersen / Mechanism and Machine Theory 45 (2010) 1707–1720

References

[1] N.L. Pedersen, Reducing bending stress in external spur gears by redesign of the standard cutting tool, Structural and Multidisciplinary Optimization 38 (3)
(2009) 215–227.
[2] C. Brecher, J. Schäfer, Potentials of asymmetric tooth geometries for the optimization of involute cylindrical gears, VDI Berichte 1904 (I) (2005) 705–720.
[3] I.A. Bolotovskii, Involute gears with asymmetric teeth, Soviet Engineering Research 4 (4) (1984) 9–10.
[4] E.B. Vulgakov, A.L. Kapelevich, Non-symmetrical gear transmissions: possible developments, Soviet Engineering Research 6 (4) (1986) 2–3.
[5] G. Di Francesco, S. Marini, Structural analysis of asymmetrical teeth-part Ii: reducing size and weight, Gear Technology 14 (5) (1997) 47–52.
[6] A.L. Kapelevich, Y.V. Shekhtman, Direct gear design: bending stress minimization, Gear Technology 20 (5) (2003) 44–47.
[7] G. Di Francesco, S. Marini, Asymmetrical gear wheels: automatized procedure for the design, VDI Berichte 1904 (II) (2005) 1735–1742.
[8] G. Di Francesco, S. Marini, Asymmetric teeth: Bending stress calculation — using a modified iso c method to calculate bending stress in asymmetric gear teeth,
Gear Technology 24 (2) (2007) 52–55.
[9] A. Kapelevich, Gear design: breaking the status quo, Machine Design International 79 (9) (2007) 89–93.
[10] A. Kapelevich, Geometry and design of involute spur gears with asymmetric teeth, Mechanism and Machine Theory 35 (1) (2000) 117–130.
[11] F.L. Litvin, Q. Lian, A.L. Kapelvish, Asymmetric modified spur gear drives: reduction of noise, localization of contact, simulation of meshing and stress analysis,
Computer Methods in Applied Mechanics and Engineering 188 (1–3) (2000) 363–390.
[12] F.L. Litvin, A. Fuentes, M. Howkins, Design, generation and tca of new type of asymmetric face-gear drive with modified geometry, Computer Methods in
Applied Mechanics and Engineering 190 (43–44) (2001) 5837–5865.
[13] D. Gang, N. Tsutomu, Bending tooth root stress decrease using an asymmetric tooth profile, Proceedings of the International Conference on Mechanical
Transmissions (ICMT 2001), 2001, pp. 404–407.
[14] G. Deng, T. Nakanishi, K. Inoue, Bending load capacity enhancement using an asymmetric tooth profile (1st report, influences of pressure angle on tooth root
stress and bending stiffness), JSME International Journal, Series C: Mechanical Systems, Machine Elements and Manufacturing 46 (3) (2003) 1171–1177.
[15] S.C. Yang, Mathematical model of a helical gear with asymmetric involute teeth and its analysis, The International Journal of Advanced Manufacturing
Technology 26 (5–6) (2005) 448–456.
[16] K. Cavdar, F. Karpat, F.C. Babalik, Computer aided analysis of bending strength of involute spur gears with asymmetric profile, Journal of Mechanical Design
127 (3) (2005) 477–484.
[17] S.C. Yang, Study on an internal gear with asymmetric involute teeth, Mechanism and Machine Theory 42 (8) (2007) 977–994.
[18] V.S. Kumar, D.V. Muni, G. Muthuveerappan, Optimization of asymmetric spur gear drives to improve the bending load capacity, Mechanism and Machine
Theory 43 (7) (2008) 829–858.
[19] C. Fetvaci, E. Imrak, Mathematical model of a spur gear with asymmetric involute teeth and its cutting simulation, Mechanics Based Design of Structures and
Machines 36 (1) (2008) 34–46.
[20] F. Karpat, S. Ekwaro-Osire, K. Cavdar, F.C. Babalik, Dynamic analysis of involute spur gears with asymmetric teeth, International Journal of Mechanical
Sciences 50 (12) (2008) 1598–1610.
[21] DIN 867. Basic rack tooth profiles for involute teeth of cylindrical gears for general engineering and heavy engineering (German standard). 1986.
[22] H. Xiao, W. Zaton, J.W. Zu, Fillet shape optimization for gear teeth, Proceedings of the ASME International Design Engineering Technical Conferences and
Computers and Information in Engineering Conference — DETC2005, 5, 2005, pp. 815–820.
[23] Y. Ding, Shape optimization of structures: a literature survey, Computers and Structures 24 (6) (1986) 985–1004.
[24] R.L. Norton, Machine Design: an Integrated Approach, 2ed., Prentice-Hall Inc, Upper Saddle River, N.J. 07458, 2000 1078 pages.
[25] N.L. Pedersen, P. Pedersen, Design of notches and grooves by means of elliptical shapes, Journal of Strain Analysis for Engineering Design 43 (1) (2008) 1–14.

You might also like