2013/2014 – Muzio M.

Gola – Politecnico di Torino

A survey of gear strength problems

1 - Bending strength of a gear tooth - Introduction
The dominant phenomenon to be taken into account when calculating the strength of a gear
tooth is fatigue.
The so called “Lewis simplified method” , here shown in detail only for spur gears, is a cantilever
bending calculation based on the following assumptions:

αtip

b: measure of “face width”

s: measure of “segment” tooth thickness – as contrasted to “arc”
tooth thickness- at any height

α Figure on the left from G. Henriot, Traité théorique et pratique des engrenages,
Tome, 1, Dunod, Paris, 1968
• The normal load Fn is applied at the tip of the tooth
• One couple only of teeth is in contact
• Only the bending component W of the force Fn is taken into account; this component has
(in the case of spur gears) an effective value:
W=Fn cosφφ; with: Ft = Fn cos α this means W=Fn cosφ φ/cosα
α;
However, W is taken – (approximation, see the figure) - equal to the pitch-line force i.e. W ≅ Ft
After tracing the tooth profile, root and fillet included, the parabola is traced through point A
(point of application of the force on the cantilever axis) and tangent to the root profile. Such
parabola is the constant stress beam in bending according to:
M Wh σ b  2  4 σroot b  2 s
σ= 2
= 2
that is for constant σ=σroot h =  root  s =   x with x =
bs / 6 bs / 6  6W   6W  2

which is the equation of a parabola with abscissa x and ordinate y ≡ h , having h=0 in point A,
positive “down”.
It is seen that at all h < hL (where the suffix L stands for “Lewis”) the width s of the parabola is
lower than the width of the tooth, this meaning that the stress in the tooth is lower than in the
parabolic shaped cantilever.
Only at position h = hL =AE where σL= CD, then σroot = σmax .
σ=σ

1
2013/2014 – Muzio M. Gola – Politecnico di Torino

The determination of the tangency points C, D is greatly facilitated by the property of the
parabola that SA=AE . In fact:

dh  4 σ root b  s L  dh  s  4 σroot b   4σ b  2
=  2 x and SE=   = L   2 xL = 2  root  xL = 2 hL = 2 AE
dx  6 W  2  dx L 2  6W   6W 
h
Ft 6 Ft h 6 L
At position hL , with W = Ft , σroot = 2 = YL where the Lewis factor: YL = m 2
b sL b m  sL 
 
m
The Lewis approach illustrated above does not take into account load sharing between more than
one couple of teeth, and for gears of sufficient accuracy is too severe, not representing the real
working conditions.
The following figure summarizes a different – and more complex – approach, where the load is
placed at the position where the force Fn is applied at the point where the load is not effectively
shared between two teeth pairs, and at the max distance from the root.
It would be possible to develop
formulas also for this case, but this
will not be done here.
Also this method is based on
approximations, however, because
one should know how much load is
taken by a teeth pair when the load
is shared in case of multiple contact,
and hence of much is the max
bending stress.

From G. Henriot, Traité théorique et pratique des engrenages, Tome, 1, Dunod, Paris,
1968

The simplified ISO method is based on the following assumptions:

2
2013/2014 – Muzio M. Gola – Politecnico di Torino

- Only the bending component is taken into

account
- The normal load is applied at the tip, but its
φ 1 F
value is taken as : n = Fn Yε
εα
where ε α is the spur gear (front) contact-ratio
(this takes into account load sharing).
The root section is defined by the tangency
points C, D of the root fillet with two lines
inclined 30° to the tooth axis; segment AE= h30
will be the arm of bending force
From G. Henriot, Traité théorique et pratique des engrenages, Tome, 1,
Dunod, Paris, 1968
h
6 30
In this case: σroot =
Ft
Yε YF with YF = m cos φ
 s 30  cos α
2
bm
 
 m 
DIN standards, which consider helical and shifted gears, introduce further considerations, as it
will be illustrated in Section 2.
Consider that the tooth shape can vary
significantly according to base and
pitch radius, teeth number and profile
shift; the figure on the left represents
some quite extreme cases.

This is a typical tooth breakage from fatigue cracks The dominating factor in bending
starting at the root of a tooth and arising from bending failure is fatigue.
loads from the driving torque.
Tooth breakage from fatigue failures
under the cyclic bending loads applied
to teeth in normal gear operation, is
one of the main failure criteria used in
gear design. The failure starts as a crack
which is usually at the root of the tooth
and proceeds across the base of the
tooth until the complete tooth breaks
away from the gear. When failure arises
from this cause there are often other

1
When the pitch error is such that no load sharing takes place, it will be Yε = 1

3
2013/2014 – Muzio M. Gola – Politecnico di Torino

adjacent teeth showing cracks at an

earlier stage of development.
From: http://www.tribology.co.uk/services/investigate/g01-
0.htm

However, there may be also cases of failure for shock overloading.

Shock overloading can arise from some major torsional shock in the drive system, often
associated with the failure of a machine driven by the gear box. In these cases the fractured
surfaces show a single brittle break with no sign of fatigue marking, and usually one or two teeth
are affected. Apart from the broken teeth, the others are usually in good condition with no signs
of cracks. Similar effects can also occur if a foreign hard body enters the gear mesh. This is
generally obvious from the nature of the impression in the teeth and the gear blank, and the body
which has caused the problem can generally be found in the failure debris in the bottom of the
gearbox. From: http://www.tribology.co.uk/services/investigate/g01-0.htm

4
2013/2014 – Muzio M. Gola – Politecnico di Torino

2
2 – Root tooth-strength calculation according to DIN standards (fatigue)
Tooth bending strength
wFt = w ⋅ KI ⋅ K V ⋅ KFα ⋅ KFβ specific pitch-line force

Ft
- w= in N/mm, is the specific nominal pitch-line force
b
- Ft is the nominal pitch-line force, b is the tooth face width
- KI is the overload factor (from 1 to 2.25 depending on the torque non-uniformities due to
driving or the driven machine; values collected in a DIN Table not shown) ; we shall
assume in our tutorials the value 1.25.
- K V is the dynamic factor, taking into account the internal additional forces due to the
errors in gear profiles introduced by manufacturing; it is given by the formula:
 k e ⋅ fpe 
K V = 1 +  1 + 0.65N / mm  ⋅ C V ⋅ Cβ ⋅ v
 w ⋅ KI 
where:
- fpe can be obtained from Table A21.6 as a function of the pitch diameter and of
the manufacturing accuracy; the manufacturing “accuracy level” number for
standard gears ranges from 5 to 12 (5 highest accuracy) according to DIN 3961-
3967; we shall assume in our tutorials the manufacturing “accuracy level” number
equal to 5.
- ke in µm can be obtained from Table 21.10 as a function of the teeth material (in
-1

our tutorials the material is steel-steel - St/St)

- C V in s/m can be obtained from Table 21.9 as a function of the pitch line velocity
v in m/s
b ⋅ tgβ b ⋅ tgβ b
- Cβ can be obtained by Table 21.9 as a function of εβ
3
( εβ = = )
pt p t cos α t
- v is the pitch line velocity in m/s
- K Fα is the load distribution factor; it depends on the factor qL
f
- qL = 0.4 + E
w
- fE is a mating factor - dependent on the admissible base pitch deviation for
4
wheel 2 and on the coupling of wheel materials - that can be calculated from
Table 21.10
- if qL < 0.5 use qL = 0.5 ¸- if qL > 1 use qL = 1
1 1
- if qL > KFα = qL ⋅ ε α - if qL ≤ KFα = 1
εα εα

2
symbols and treatment from K.H. Decker- Maschinen-Elemente, C. Hanser Verlag München Wien, 1985; this is a highly
recommended reference book

We have named pt and αt in our lecture notes: pc and αc

3

4
index 1 is given to the small wheel (pinion), and index 2 to the large wheel

5
2013/2014 – Muzio M. Gola – Politecnico di Torino

- K Fβ is the longitudinal load distribution factor; it requires to specially determined case

by case (taking into account tooth straightness or intentionally altered longitudinal tooth
profiles, tooth longitudinal deformation, shaft or gearbox or gear-body deformation). In
ordinary construction it is assumed K Fβ = 1 for spur gears and K Fβ = 1,2 for conical gears.

However, the figure on the right shows K Fβ ,

From G. Henriot, Traité théorique et pratique des engrenages, Tome, 1,
Dunod, Paris, 1968
which depends mainly on misalignment of
teeth and shafts, in torsion and in bending, due
to all possible causes. Curve 1 holds for teeth
with longitudinally altered profiles (see
Appendix II), while curve 2 holds for non
altered profiles. The diagram comes from
Bureau Veritas, as quoted by Henriot. It is seen
that for b/d1<1 we can take KFβ = 1 , i.e., the
factor counts only for relatively long gears.

Once the specific pitch-line force wFt is known, the tooth bending stress is calculated as:
w
σ F = Ft YF ⋅ Yε ⋅ Yβ
mn
- mn is the modulus of the cutting rack - as seen in the “normal plane”
- YF is the tooth form factor that can be obtained by Table A21.11 as a function of the
z
equivalent or “virtual” number of teeth zn = and the “profile shift” coefficient x.
cos3 β
1
- Yε contact ratio coefficient, for spur gears: Yε =
εα
- Yβ Helical gear coefficient
β 0° 5° 10° 15° 20° 25° ≥30°
Yβ 1 0.96 0.92 0.88 0.84 0.79 0.75

From G. Henriot, Traité théorique et pratique des engrenages, Tome, 1,

Dunod, Paris, 1968

The safety factor SF must be calculated for both wheels, and is calculated as:
σ - σFD is the allowable fatigue stress of the tooth material from Table A21.12
SF = FD
σF - SF ≥ 1.8 for infinite life calculation

6
2013/2014 – Muzio M. Gola – Politecnico di Torino

3 - Surface strength of a gear tooth - Introduction

From : http://www.tribology.co.uk/services/investigate/g03-
0.htm

Pitch line pitting on a case hardened

pinion which has operated at high load
and low speed

The drawing on the left shows

two wheels with centres O1, O2
and base circles tangent to the
pressure line at points T1, T2 .

Contact segment is AB. When

the contact point is at M, the
profiles of the teeth pair in
contact are approximated to
the second order by two circles
of radii ρ1, ρ2 ; of course, it is
always:
ρ1 + ρ2 = T1T2
In cylindrical Hertz contacts the
max contact stress σmax is
proportional to:

 1 1 
Fn  + 
 ρ1 ρ 2 

Figure from G. Henriot, Traité théorique et

pratique des engrenages, Tome, 1, Dunod,
Paris, 1968

 1 1  T1T2
In:  +  = the value of the denominator is a parabolic function of ρ1 , zero
 ρ 1 ρ 2  ρ 1 (T1 T2 − ρ 1 )
in T1 and T2 and maximum at the mean point of segment T1T2 . Therefore, σmax is infinite at
points T1 and T2 , is minimum in the midpoint M of the segment T1T2 and symmetric about M.
Moreover, diagram 1 shows the value of max contact stress if only one teeth pair takes the full
7
2013/2014 – Muzio M. Gola – Politecnico di Torino

load Fn . The highest stress occurs at the point (in this case point B) which is more far from
midpoint; midpoint position is indicated by the dashed line crossing T1T2 very near to point M.

However, points V and W mark the ends of the segment of single contact, while segments AV and
WB are double contact (contact force is assumed as divided in two equal parts). It is then seen
that the maximum contact stress in W is very near to the one at point “I” (which in Henriot’s
terminology is what we - instead - call “C”). The ISO convention states that the contact stress
value to be used to determine surface strength is the one at point “I” alias “C” .

The value of Hertz max stress when (single)contact is at point C is calculated through the formula:
1 1
+
F ρ ρ2 Fn  1 1 
σmax = 0,59 n 1 for equal materials: σmax = 0,42 E +  which for both
b 1 + 1 b  ρ1 ρ2 
E1 E2
Fn  1 1 
teeth in steel, gives: σmax = 190  + 
b  ρ1 ρ 2 
Fn r1 + r2 F 1 u+1
In which: ρ1 = r1 sin α and ρ2 = r2 sin α then: σmax = 190 = 190 n
b r1 r2 sin α b r1 sin α u

where u = r2 / r1 (reduction ratio or gear ratio) .

DIN standards, which consider helical and shifted gears, introduce further considerations, as it
will be illustrated in Section 3.

The diagram above also shows which would be the width 2a of the contact surface at various
points on segment T1T2 under the normal contact force Fn : it is maximum at the midpoint and
zero at the two ends. Its value is calculated, in the case of steel on steel:

Fn 1 F ρ ρ
2a = 6 ,81 10 −3 = 6 ,81 10 −3 n 1 2 (units: force in N, dimensions in mm)
b 1 + 1 b ρ1 + ρ 2
ρ1 ρ2

4 - Tooth surface fatigue strength according to DIN standards5

wHt = w ⋅ KI ⋅ K V ⋅ KHα ⋅ KHβ specific pitch-line force

- w, KI , K V are the same as the gear tooth bending strength calculation.

- KHα is the spur gear load distribution factor –
 1 
K Hα = 1 + 2 ⋅ (q L − 0.5 ) ⋅  2 − 1
Z 
 ε 

5
symbols and treatment from K.H. Decker- Maschinen-Elemente, C. Hanser Verlag München Wien, 1985

8
2013/2014 – Muzio M. Gola – Politecnico di Torino

that for spur gears takes into account the non uniform distribution of teeth pairs in
simultaneous contact - it depends on the contact ratio factor Zε :

 4 − εα εβ 
Zε =  ( )
1 − ε β +  ⋅ cos β b
εα 
 3
where βb, the helix angle on the base cylinder, calculated as sin βb = sin β ⋅ cos αn
where qL is the same as the gear tooth bending strength calculation.

- KHβ is the longitudinal load distribution factor, KHβ = 1 for spur gears

wHt u + 1
The Hertz stress is calculated as: σH = ⋅ ⋅ ZH ⋅ Z M ⋅ Z ε
d1 u

- d1 is the pinion pitch diameter

- u is the gear ratio
6
- Z H is the flank form factor

1 cos βb
ZH =
cos α t tan α wt

where αt is the standard pressure angle, αwt is the working pressure angle
βb,is the helix angle on the base cylinder, calculated as sin βb = sin β ⋅ cos αn

- ZM is the material factor that can be obtained by Table A21.13

The safety factor is calculated as:

σHD
SH =
σH

- σHD is the allowable fatigue stress of the tooth material from Table A21.12

- SH ≥ 1.25 for infinite life calculation and z1 > 20

- SH ≥ 1.4 for infinite life calculation and z1 ≤ 20

We have named αt in our lecture notes: αc . Accordingly, he same may be for αWt called αWc
6

9
2013/2014 – Muzio M. Gola – Politecnico di Torino

5 – Design to wear: specific sliding and its balancing

The figure on the left shows tangential
speeds vt1 and vt2 of tooth profiles 1,
2, i.e., the speed component of each
tooth, at the contact point, orthogonal
to the contact line. Their difference is
the relative or “sliding” speed vg.
Since vg is proportional to the power
dissipated on the contact, considering
the lenght of the arc on the profile on
which this power is dissipated, for
instance on profile 1:
ds1= vt1 dt
and assuming that the power then
wear is indicated by energy dissipated
divided times the arc length on which
it is dissipated, then the ratio:
v gdt vg
ς1 = =is an indicator of
v t1dt v t1
wear. The same for ς 2 (represented
below).

Naming as usual A (access) , E (exit) the ends of the effective contact segment, it can be further
ω ρ − ω1ρ1 ω ρ − ω1ρ1
elaborated: ς1 = 2 2 ς2 = 2 2 (signs are HERE purposely chosen so that
ω1ρ1 ω2ρ2
both will be positive in access, AC, and negative in recess, CE). It is seen that ς 2 is 1 at T1 and
tends to minus infinite in T2. Vice versa, ς1 is -1 at T2 and tends to infinite at T1. It is seen that ς1
is larger than ς 2 in access, vice versa in recess.

10
2013/2014 – Muzio M. Gola – Politecnico di Torino

An appropriate profile shift might try to equalize as much as possible the (absolute values of)
specific sliding at the two ends, then, taking into account the signs:
ω ρ − ω1ρ1A ω ρ − ω1ρ1E
ς1 A = 2 2 A = − ς 2E = − 2 2E
ω1ρ1A ω2ρ2E
which brings: ω22 ρ2A ρ2E = ω12 ρ1A ρ1E

6 – Design to scuffing and its balancing

The terms scuffing and scoring are frequently interchanged.
Scuffing, also termed "scoring" (incorrect according to gear standards), is a severe type of
adhesive wear which instantly damages tooth surfaces that are in relative motion. In fact, a single
overload can lead to catastrophic failure.
Scuffing welds together unprotected surfaces in metal-to-metal contact. Metal particles detach
and transfer from one or both meshing teeth. During successive rotations, these particles can
scratch teeth flanks in the sliding direction. This type of damage generally happens in areas of
high contact pressure and sliding velocity, far from the pitch surface. Conditions there are less
favourable to form a protective lubricant layer that would prevent direct metal-to-metal contact.
This protective layer could be a thick oil film (relative to surface roughness) or an adsorbed or
chemically deposited layer established by lubricant additives.
Gear scuffing is characterized by
material transfer between sliding
tooth surfaces. Generally this
condition occurs when inadequate
lubrication film thickness permits
metal to- metal contact between
gear teeth. Without lubrication,
direct metal contact removes the
protective oxide layer on the gear
metal, and the excessive heat
generated by friction welds the
surfaces at the contact points.
From:
http://www.tribology.co.uk/services/investigate/g03-
0.htm

As the gears separate, metal is torn and transferred between the teeth. Scuffing is most likely to
occur in new gear sets during the running-in period because the gear teeth have not sufficient
operating time to develop smooth surfaces.
Scuffing is controlled by appropriate lubrication conditions. An old criterion, credited to Almen, is
known as the PV (pressure-velocity) criterion.
• The product σHv g , named Almen parameter, i.e. the max Hertz stress times the sliding
speed, is calculated and compared with a limit value which is (experimentally) characteristic
of the lubricant employed
• Almen parameter is proportional to a specific power: in fact, naming the Hertz contact
area AH , and assuming a value µ for the friction coefficient, the lost power is: σH AH µ v g

11
2013/2014 – Muzio M. Gola – Politecnico di Torino

which will produce a temperature increase in the lubricant. Usually accompanied by a

decay in viscosity, a fact that makes lubrication even more difficult. Breaking of the
lubricant layer may then result in metal contact and scuffing.
Evidently, the most dangerous values of Almen parameters are those at the ends of the
contact segment, where both σH and v g are higher (things may be complicated by load
sharing, however).
Limit values for case-hardened and ground teeth are, according to Henriot:
Almen factor σHv g (MPa m/s) Observations

2200 to 6600 normal oils, no tooth profile modification

8800 to 11000 aircraft applications, with profile modifications
and special oils with additives
13200 upper limit

Equalizing the Almen factor starts from the consideration that σH is proportional to the square
root of the relative curvature, that in points A and E:
 1 1  TT  1 1  TT
in A:  +  = 1 2 and in E:  +  = 1 2
 ρ1 A ρ 2 A  ρ1 A ρ 2 A  ρ1E ρ2E  ρ1E ρ2E
and sliding speeds:
in A: v gA = ω2ρ2A − ω1ρ1A and in E: v gE = ω2ρ2E − ω1ρ1E

then equalization:
T1T2 T1T2
(ω2ρ2A − ω1ρ1A ) = - (ω2ρ2E − ω1ρ1E )
ρ1 A ρ2 A ρ1Eρ2E

which after simplification:

ω1 ω2
= ⇒ ω12 ρ1A ρ1E = ω22 ρ2 A ρ2E
ρ2 Aρ2E ρ1Aρ1E
i.e., the same condition for the equalisation of specific sliding.

12
2013/2014 – Muzio M. Gola – Politecnico di Torino

Appendix I
Tables from K.H. Decker- Maschinen-Elemente, C. Hanser Verlag München Wien, 1985.

Admissible deviations of pitch and base pitch for gears according to DIN 3962
(Excerpt)

Contact ratio fE of gear wheel pairs

13
2013/2014 – Muzio M. Gola – Politecnico di Torino

Reference values for the Dynamic Factor Kv of gear wheel pairs

Form factor YF in function of the profile shift factor x and of the virtual teeth number
zn or zvn (compiled from DIN 3990)

14
2013/2014 – Muzio M. Gola – Politecnico di Torino

Materials for gears and their fatigue strength σFD e σHD (according to DIN 3990)

15
2013/2014 – Muzio M. Gola – Politecnico di Torino

Material factors ZM according to DIN 3990

16
2013/2014 – Muzio M. Gola – Politecnico di Torino

Appendix II

In order to avoid overloads at the ends of teeth

From G. Henriot, Traité théorique et pratique des engrenages, Tome, 1,
Dunod, Paris, 1968
due to a great variety of unfavourable events:
• tolerances in machining gearboxes
• gearbox deformations under loads
• errors of alignment in gears
• torsion and bending deformations of
pinion
it is not convenient to machine a “perfect”
tooth.
Rather, it is convenient to modify the tooth
form in the longitudinal direction, as indicated
qualitatively on the left, case (1). This is done
symmetrically because it is difficult to foresee
on which side the greater deflections will
occur. In the case of car-shift gearboxes, a
crowning max of 10 to 15 µn is usual.

On the contrary, if one can foresee that load will concentrate at a specific end, crowning will take the form
of case (2), i.e., a light taper of the tooth. This is the typical case of cantilevered pinions of electric motors,
where the load concentrates on the tooth side which is nearer to the motor. Henriot reports a taper of 10
minutes of degree (1 minute=1/60 of a degree) on the half of the pinion width.
The general idea is that these corrections will bring the tooth in uniform contact when the load is applied;
iterative computing is needed to determine the deflections under load on which the contact depends.

17