Systematic

Analysis of
Gear Failures
Lester E. Alban
Metallurgical Engin eer
Fairfield Manufacturing Company, Inc.
Subsidiary o f Rexno rd, Inc.

American Society for Metals

Metals Park, Ohio 44073
Copyright © 1985
by the
AMERICAN SOCIETY FOR METALS
All rights reserved

First printing, June 1985

Second printing, August 1986
Third printing, November 1990

No part of this book may be reproduced, stored in a retrieval system,

or transmitted , in any form or by any means, electronic, mechanical,
photocopying, recording, or otherwise . without the prior written per-
mission of the publisher.

Nothing contained in this book is to be construed as a grant of any

right of manufacture. sale, or use in connection with any method. pro-
cess. apparatus, product . or composition. whether or not covered by
letters patent or registered trademark, nor as a defense against liabil-
ity for the infringement of letters patent or registered trademark.

Library of Congress Catalog Card Number: 85-70125

ISBN: 0-87170-200-2
SAN: 204-7586

PRI NTED IN TH E U NITED STATES OF AM ERI CA

Preface

From the dis covery of th e failure to the final letter of transmit-

tal, the line of communication should remain open, accurate,
and sincere. There is no other way to achieve an acceptable plan
for corrective a ction.
Not only the titl e of this book, but the entire con ten ts
strongly emphasize the s ystematic approach to an analysis of
the problem. Putting the com ple te picture in a sequential pat -
tern makes sense. Case histories are good, but they often do not
detail the methods used to determine the final answer. All too
frequently a person vi ews a gear failure , finds a similar pi cture
in a book , and beli eves it to be of the same mode and cause ; this
may not be so.
The purpose of this text is to train the reader in the art of
discipline , to establish a logi cal step-by-step system of anal y-
sis: begin at the be ginning and continue methodically to the end.
If only we can influ en ce fi eld personnel, field representatives ,
mechanical and metallurgical analysts , and responsible engi-
neers in management to be aware of the overall picture and to
appreciate the role each plays in the final analysis, this book
will be a success.
No work of thi s ex te n t is accomplished by the author alone.
Not only has 36 yea rs of first-hand experience been necess ary,
but also the experienc e of many experts in our peer g roup . Of
course, Fairfield Manufa cturing Company, now a subsidiary of
Rexnord , Inc ., has be en uppermost in supplying the environ-
ment of quality reputation ne cessary to maintain a cons is te n t ly
ethical leadership. M y first em ploy er, the late Harrold J. Bat es,
was meticulous in his concern for detail and accurac y. We
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
iv
learned to orie nt a ll our efforts to ward the need s of our cus -
t om er s . Thi s tr ain in g ha s been good, a nd thi s philo soph y
continu es .
Since I retired a year ago, Fairfi eld h a s graci ously supplied
a n office for m y use while writing. The Met allurgi cal Dep art-
ment, under th e directi on of R. L. Hugh es, has be en excep tion-
a lly pat ient a nd ext re me ly helpful in allo w ing me full us e of its
time, tal ent, past reco rds , pho tographi c abilit y, a nd laborat or y
facilities, a nd so me ti mes ju s t b y li st ening. Without the st eno-
gra p hic help of its secre ta ry , th is wo rk w oul d ha ve been a lmo st
impos si bl e. M y wi fe, Fa ye, was a lways enco u rag ing a nd reall y
didn 't beli e ve I had re tire d a ye a r ago , s ince I alwa y s w ent to
wor k at th e sa me time eve ry m orning. Thi s ha s been a work of
di scipline, but pos sible only w ith th e help a nd the faith of those
menti on ed a bove.
Contents

Basic Understanding of Gears 1

1 Purpose, Design, Function 1
Basic Applied Stresses 6
Spur Gear 7
Helical Gear 7
Straight Bevel Gear 9
Spiral Bevel Gear 9
Hypoid Gear 10
Strength 10
Gear Tooth Characteristics 12
Tooth-to-Tooth Running Pattern 12
Involute Pattern 13
Lead Pattern 15
Mating-Tooth Contact Pattern 17
Backlash 18
Composite Errors 20
Associated Parameters 20
Round Bores 21
Splined Bores 23
Shafts 24
References 24

2 Basic Understanding of Environment 27

Lubrication 28
Incorrect Lubrication 29
The Lubrication System 30
Lack of Lubricant 30
 CONTENTS
vi
Excessive Lubrication 31
Lubricant Contamination 31
Temperature 32
Lubricants 33
Ambient Temperatures 34
Gear Tooth Operating Temperature 34
Components 35
Mechanical Stability 36
Personnel -Related Activities 38
Assemblymen 38
Operators 38
Maintenance Personnel 38
Management Attitude 39
References 40
43
3 Systematic Examination
Field Examination 44
Visual Examination 47
Physical Examination 53
Magnetic-Particle Inspection 53
Tooth Characteristic Studies 54
Surface Hardness Testing 54
Ultrasonic Testing 55
Nital Et ching 57
Profilometer Measurements 61
Dimensional Che cking 61
Metallurgical Examination 62
Cross-Sectional Hardness Survey 63
Macroscopic Examination 65
Carbon Gradient Traverse 65
Chemical Analysis 71
Case Hardness Tra verse 72
Microscopic Examination 75
Scanning Electron Micros copy 82
References 83
Modes of Gear Failure 85
4 Fatigue 86
Tooth Bending Fatigue 86
Surface Contact Fatigue (Pitting) 94
 CONTENTS
vii
Rolling Contact Fatigue 100
Contact Fatigue [Spalling) 105
Thermal Fatigue 106
Fatigue of Round, Splined, and Keyed Bores 109
Shaft Fatigue 109
Impact 112
Tooth Bending Impact 112
Tooth Shear 115
Tooth Chipping 118
Case Crushing 118
Torsional Shear 120
Wear 120
Abrasive Wear 121
Adhesive Wear 123
Stress Rupture 125
Internal Rupture 125
External Rupture 127

5 Causes of Gear Failure

Basic Material 130
129

Steel 130
Forgings 136
Castings 138
Engineering 140
Design 140
Material Selection 145
Heat Treatment Specifications 147
Grinding Tolerances 147
Manufacturing 147
Tool Undercutting, Sharp Notches 149
Tooth Characteristics 151
Grinding Checks , Burns 152
Heat Treatment Changes 153
Heat Treatment 154
Case Properties 154
Core Properties 158
Case/Core Combination 159
Hardening 161
Tempering 164
Miscellaneous Operations 165
 CO NTENTS
viii
Service Applicatio n 167
Set Matc hi ng 169
Asse mb ly , Alignme nt, Deflectio n, a nd
Vibration 169
Mech anical Da mage 1 72
Lu b rica tion 173
Foreig n Material 175
Corros io n 177
Continual Overloading 177
Impact Overloading 180
Bearing Fa ilure 181
Mai nte na nce 182
Operator Error 182
Field Ap plica tion 18 3
References 184

The Final Analysis

6 Broken S haft 185
185

Bro ke n Spiral Bevel Gear 188

S piral Beve l Drive Set 194
Spur Pinion 197
Spu r Gear 200
Spiral Beve l Se t 203
Spi ral Beve l Set 207
S pur Gear 211
H ypoid Pinio n 214

7 Writing the Report

Field Report 220
219

Transmittal Report Wit h Fa iled Unit 220

Laboratory Notes of the A na lyst 220
Fail u re Analys is Report 222
Lett er of Tra ns mi ttal 22 4
Et hica l Overview 225
Referen ce 226

Index 227
 CHAPTER 1
Basic Understanding
of Gears

Analyzing a failed part without taking into account th e entire

picture of magnitude and direction of applied forces, the spacial
relationship of all component parts within the assembly, and
the environmental conditions existing around the unit would be
like examining a torn automobile tire for defects but never exam-
ining the bent axle that was causing the detrimental side thrust.
For a systematic study of a gear failure, the basic parameters of
a gear must be understood.

Purpose, Design, Function

A gear is a machined component that transmits motion and
force from one element in a working unit to another element in
the same unit or to another working unit in either the same
plane and direction or a completely different plane or direction.
The force , due to this transmission, may either increase or de-
crease in power from one element to the next. Design and func-
tion are closely associated because a gear is designed with a
specific function in mind. The question is if this gear will per-
form the function that was intended by the designer .

Spur gear and pinion. A spur gear and pinion (Fig. 1-1) is a
parallel-axis unit with tooth forces and motion at exactly 90°
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
2

Fig. 1-1. Spur gear and pinion.

from the central axis of each part. Due to the ratio of numbers of
teeth in each part, the speeds may be increased or decreased.
The transmission of power is in a straight line.

Helical gear and pinion (Fig . 1-2 J is also a parallel -axis

gearing application with transmission of power about a straight
line. However, the teeth, being helical, exert a resultant force in
an angular direction at the tooth contact interface in proportion
to the angle of helix. Along with any tendency to move the tooth
laterally by the applied contact pressure, there is also the ten -
dency for a surface sliding action in the same direction. A
double-helical gear is one in which two sets of helical teeth are

Fig. 1-2 . Helical gear and pinion.

 BASIC UNDERSTANDING OF GEARS
3
cu t around th e same periph er y, b ut w it h a n opp osi ng a ng le of
heli x. The ce n tral a rea is mach in ed out so that eac h porti on is
disco nnected . If th e ce ntral portio n is sec ure ly co n nected, a her-
ri ngbo ne gear w ill be obtai ne d . T he advantage to the double-
helical a nd /or her ring bon e gears is tha t th e s ide thr ust co mmo n
to the s ingle hel ix ha s been elim inated .

Internal gear. Bo th the sp ur gear se t a nd the heli ca l gear set

ma y be desi gned so tha t th e gear memb er w ill ha ve th e teeth on
th e in side diameter of a ring . Thi s is ca lle d a n in ternal gear (see
Fig. 1-3J.

Fig. 1-3. Internal gear and pinion. This set can be either
spur or helical.

The straight bevel gear and pinion (Fig. 1-4 J wi ll tran sm it

mo tio n in a n a ng ula r d ir ecti on (us ua lly a t 90 0 J. If th e s pee ds a re
to rem a in th e s am e with only a 90 0 cha nge of directi on , the se t is
ca lled a mit er ge a r set. An y cha nge in the numb er of teeth will
cha nge s pee d as w ell as direct ion. The co n tac t fo rces ten d to
pu sh th e op pos ing teeth a part as we ll a s to ca use a lateral slide
a long eac h tooth surface.

Spiral bevel sets. As the off-a ng le gear tee th are give n a n

a ngular d is placement, they na tu rall y ass ume a spiral (circular J
shape to conform to the rotating motio n . This gearing is called
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
4

Fig. 1-4. Straight bevel gear and pinion.

Fig. 1-5. Two types of spiral bevel sets, consisting of

spiral bevel gear and pinion.
 BASIC UNDERSTANDING OF GEARS
5
s p ira l bevel se ts. Th ere are m an y typ es of s p ira l be vel config u-
ration s; two t yp es a re sho wn in Fig. 1-5.

Hypoid sets. U suall y th e axis of th e s p iral bev el gea r a nd

pinion will int ers ec t at a co mmo n p oint in spa ce. Ho w ever,
w he n th e pinion a xi s is rai sed or lowered in relation to th e gea r
axis, th e resu lt is a h ypoid se t [Fi g. 1-6J. Wh en th e axis of th e
pinion is di spla ced a lm os t to th e cent er of th e gea r teeth a t th e
periph ery , the numb er of teeth in the pini on decrea ses to three
or les s a n d h a s the ac tio n of a wo rm pini on . Thi s se t is ca lle d th e
hi gh- rati o h yp oid set [Fi g. 1-7J.

Fig . 1-6. Hypoid gear and pinion set.

Fig. 1-7. High -ratio hypoid set.

 SYSTEMATIC ANALYSIS OF GEAR FAILURES
6
Basic Applied Stresses
T he loa d s a pplied to one tooth by the action of its mating tooth
are at a ny moment of time a line contact at the most; or, at the
lea s t , a poin t co ntact. As th e load s are increased, the line may
lengthen or eve n b roaden , or the point may expand to a rounded
area .
T he bas ic stresses applied to a gear tooth include the six
types listed in Fig. 1 -8; often, a combination of two or three
types are app lied at a time. Commonly they are tensile, com-

Tensil e

Comp ressive
+
+
TN

Slide-Shear ... .. S

<:.:»
Rol ling
---..... R

Roll-Slide
..
....------..... R-S

Torsion -~- TR

Fig. 1-8 . Basic s tresses that are applied to gea r teeth.

Often , two or three are si multane ous ly a pplied to a spe-
cific area.

pressive, shear (slide), ro lling, ro lling-slide, and torsion . Each

type of gear tooth w ill ha ve its own characteristic stress
patterns .
For specific gearing terminology and nomenclature of tooth
elements, refer to "Geometry and Theory of Gears" by Paul M ,
Dean-s a nd Standards published by American Gear Manufac-
turers Associatio n."
 BASIC UNDERSTANDING OF GEARS
7
Spur Gear
As the contacting tooth moves up the profile of the loaded
tooth , a sliding -rolling action takes place at the profile inter-
face. At the pitchline, th e stresses are pure rolling . Above the
pitchline, the rolling-sliding action again takes over, but the
sliding will be in the opposite direction. Keep in mind that the
action on the profile of the contacting tooth is exactly the same
as the loaded tooth except in reverse order (see Fig. 1-9J . The

Fig. 1-9. Diagramatic stress areas on basic spur gear

tooth.

sliding action of two surfaces, when lubricated properly , will

have no problem. However, surface disparities, insufficient
lubrication, improper surface hardness, higher temperatures ,
and abrasive or adhesive foreign particles will contribute to a
breakdown during a sliding contact. At the same time, there is a
tensile stress at the root radius of the loaded side of the tooth
and a compressive stress at the root radius of the opposite side.

Helical Gear
The helical gear tooth receives the same contact action as
the spur gear; i.e. , a rolling-sliding action from the lowest point
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
8

Fig.l-IO. Secondary stresses set up in associated parame-

ters of a helical gear due to the side thrust action of the
helix.

of active profile up to the pitchline, rolling over the pitchline,

then sliding -rolling from the pitchline over the addendum . An
additional stress is being applied to the helical tooth; a lateral
sliding action is applied at all contact levels , including the
pitchline . The force comp onen t at 90 ° to the dire ction of rota-
tion increases as the helical angle increases . Resultants of this
side thrust are often overlooked (see Fig. 1-10). The web be-
tween the center shaft hub and the outer gear rim is constantly
undergoing a cycle of bending stress; it is not un common for a
relativel y thin web to fail in bending fatigue . If the hub of the
gea r fa ces against a thrust bearing, the bearing itself is under a
constant thrust load. The shaft carrying the gear undergoes a
continual rotation bending stress. It is also not uncommon to
have such a shaft fail by rotational bending fati gue. The above
 BASIC UNDERSTANDING OF GEARS
9
secondary stresses are found only in a gear of a single-helix
pattern. A double-helical ge a r or a herringbone gear will not
have a side thrust component of stress; therefore, the entire
stress load will be absorbed by the teeth.
One additional stress that should be discussed at this time
is a stress common to all gearing because it involves rolling sur-
faces . It is a shear stress running parallel to the surface at a
distance from 0 .007 to 0.012 in. below the surface. The distance
below th e surface given above is the average depth for a normal
loading condition. The ac tual depth of maximum shear could be
deeper, depending on the radius of curvature of the mating sur-
faces and the tangential forces being applied . In one instance
there has be en evidence of rolling loads above the shear
strength as deep as 0.034 in. The subsurface shear stress is most
often the originator of initial line pitting along the pitchline of
gear teeth , line pitting low on the profile due to tooth tip inter-
ference , line pitting along the tooth tip due to the same tooth tip
interference, and subsurface rolling contact fatigue. The sub-
ject of rolling contact fatigue is discussed more fully in
Chapter 4.

Straight Bevel Gear

The straight bevel gear undergoes the same stresses as dis-
cussed above, including a very slight helical action laterally.
The larger sliding action component is parallel to the axis of the
gears and tends to push the gears apart, causing a higher profile
contact, and to exert a rotational bending stress in the web of
the part as well as in the shaft.

Spiral Bevel Gear

Aside from all the stresses applied above, a spiral bevel
gear has a resultant peculiar to itself. As the rolling-sliding
stress tends to move in a straight line laterally, the progression
of the points along the stress line moves in a bias across the
profile of the tooth. As long as at least two teeth are in contact,
the resulting load per unit area is well within reasonable limits.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
10
However, there are circumstances (and it may be only momen-
tary) when there is a 1-to-1 tooth contact. This very narrow line
contact may be accepting an extremely high load per unit area,
and a line of pitting will result early in the life of the tooth.
Careful attention should be given to the design characteristics
of these parts, such as spiral angle and pressure angle.

Hypoid Gear
The hypoid gearing has the same applied stresses as those
discussed for the spiral bevel, but sliding becomes the more
predominant factor . This predominance increases as the axis of
the pinion is placed farther from the central axis of the gear, and
is maximum when the set becomes a high-ratio hypoid .

Strength
The strength of any component is measured by the amount of
stress that can be tolerated before permanent strain [deforma-
tion) takes place.
Strain, or deflection under load, is a constant for steel re-
gardless of hardness or heat treatment. The amount of deflec-
tion under load of a thin gear web or the shank of a pinion can-
not be changed by heat treatment or by use of a stronger
material. Hooke's law is the same : A change of deflection can be
accomplished only by a change of design.
Bending strength of a gear tooth is the amount of load per
unit area acceptable at the root radius to the point of permanent
deformation. Permanent deformation of a carburized tooth is
usually accompanied by a crack at the root radius, whereas with
a noncarburized tooth, actual bending may occur. The root ra-
dius is mentioned as the point of deformation because it is the
area of greatest stress concentration in tension. Also, stress
[load per unit area] calculations assume that the load is applied
at the pitchline or the midheight of the tooth. Actually , the real-
istic stress at the root radius varies from approximately one-
half, when the load is applied low on the active profile, to dou-
ble, when the load is applied near the tooth tip. Bending
 BASIC UNDERSTANDING OF GEARS
11
strength of the root radius is a function of the surface hardness
and the physical condition of the surface, such as smoothness,
sharpness of radius, and /or corrosive pitting .
The strength of the core material-Le., the basic material
under the carburized case-is generally to be considered as
compressive strength rather than tensile strength. It measures
the ability to withstand surface pressures that may crush
through the case and/or brinell the surface.
Torsional strength of a pinion shank or of a shaft is a bit
more complex. The maximum tensile stress is at the surface in a
direction 45 0 from the central axis or longitudinal direction.
The maximum shear stress , also at the surface, is longitudinal
(parallel to the central axis] and transverse (90 0 across the cen-
tral axis] (see Fig. 1-11]. The strength at the surface is a func-
tion of surface hardness; therefore, surface-originated torsional
Torsion

Tmax

Elastic-stress
"m ax distribution

0' 3= compressive stress

0' ,= tensile stress
J m ax = maximum shear stress

Fig. 1-11. Free-body diagram of maximum tensile and

shear stress orientation on a surface element of a shaft in
a torsional mode. Both maximums are at the surface.
Stress is considered to be zero at the central axis.'
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
12
tensile failures of carburized parts are rare unless a specific
type of stress raiser is present at the surface. This is not so with
through hardened or non-heat treated parts since the strength is
uniform throughout the part. Under this condition, torsional
tensile failure is expected to originate at the surface. In most
instances of torsional failure of carburized or induction hard -
ened shafts and pinion shanks, the initial fracture is along the
shear plane [i.e., longitudinal or transverse), and not at the 45°
angle. This means that the shear strength of the subsurface
material is the controlling factor . Shear strength is considered
to be only about 60 % of the tensile strength. The area most
vulnerable to the origin of torsional shear failure of a shaft is
the transition zone between the case and the core of either a
carburized or an induction hardened part. The maximum ap-
plied stress often exceeds the shear strength of the material at
this area and initiates a start of subsurface failure.

Gear Tooth Characteristics

Tooth characteristics are designed into the gear as an integral
part and are usually included in the dimensional specifications.
Each characteristic contributes significantly to the tooth load
contact pattern, which greatly influences the unit area accept-
ing the maximum loads.

Tooth-to-Tooth Running Pattern

Tooth-to-tooth running pattern is plotted on a chart as a
closely controlled master gear runs with the production gear.
The accuracy of the master gear is such that any discrepancy
noted in the chart can be attributed to the lack of accuracy in the
production gear. The characteristics noted in this test are gear
tooth runout and tooth-to-tooth jump. The ideal, straight line
pattern, shown in Fig . 1-12[a), can be attained only from a heli-
cal gear tha t has more than one tooth in contact and has no run-
out. A spur gear is expected to have some amount of "jump" as
the contact goes from tooth to tooth. Of course , the more accu-
rately the teeth are cut, the less jump. The pattern shown in Fig.
 BASIC UNDERSTANDING OF GEARS
13

(a) (b) (c)

[a] Idea l straight li ne . n o ru no u t: very lit tle toot h - to- tooth jump. d ue to he li-
cal tooth o ve rlap .Tb ] Al m ost st raig h t li ne . D.DDl -in . overal l ru no u t : pe rcep ti-
bl e too th-t o-tooth j um p . typ ica l of s p u r ge a r ing . N o le Ihe o ne pro t ub eran ce
d ue to s ur face ni c k . lei Sp ur ge a r w i th D.DD6 - in . ove ra ll runo u!.

Fig. 1-12. Tooth-to-tooth running pa Item.

l -12[ b ] is a n exa m ple of a s pur gea r co ns ide re d to be exce lle n t.

w it h onl y O.OOl-in. runout. The chart show n in Fig. l-1 2[ c] is of
a spur gea r that has a fairl y good tooth-to-tooth patt ern, but a
runout of 0 .006 in . is indi cated.
It becom es ev ide nt, a lso, th at th es e test s are fun ctional in
ways ot he r th an tho se specifi ed. For in st an ce, the chart in Fig.
l-12[ b] shows one a b ru pt mo vem ent be yo nd the n ormal pat-
te rn, which indi cat es a di scr epancy th at in mo s t inst an ces
w ould be du e to a slig h t ex te rn al ni ck or bump.

Involute Pattern
As an in d icat or foll ow s the t ooth profile fr om the lo w est
po int of th e ac tive profile to the tip of the tooth , the in volute is
scribed onto a cha r t. A true involut e for a to oth would show up
a s a straight line on the chart [Fig. l -13J .
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
14
.- -~ .-
. I~
- 1-<'

If \

...
tt- I-
,
I
L.l I
_v l--
10...

Fig. 1-13. Charted involute pa tt ern . The straight line indi-

cates true involute of the active tooth profile.

INVOLUTE
Fig. 1-14 . Charted involute pattern of two teeth (180 0
a part) from a gear in service. Unused profile has an excel-
lent involute; the loaded profile shows pitchline and tip
wear.

Again , this c har ti ng al so ca n be used as a n in sp ection tool

for reading what ha s happ en ed in servi ce. As a n exa m ple, Fig.
1 -14 shows a n in volu te patt ern of two teeth of a gear re turned
fr om serv ice. One s ide profile has no t been wo r n; the gra ph is a
s traig h t lin e. The pat tern of the op pos ite si de s hows a d efinite
 BASIC UNDERSTANDING OF GEARS
15
wearing of the midprofile area and slight wear over the adden-
dum. This pattern also reveals that the gear ran in one direction
only . The involute charts should be read carefully to distin-
guish differences of thicknesses low on the profile (positive for
interference), high on the profile (positi ve for tip interference or
negative for excessive falling away), and at mid profile (a posi-
tive or negative barreling effect that would concentrate the load
pattern at the midpoint or at either extremity).

Lead Pattern
The lead pattern is the graphic line inscribed on a chart
when an indicator traverses from one end of the tooth to the
other end along a line parallel to the central axis, most com-
monly along the pitchline. A true lead should be a straight line
for both the spur gear tooth and the helical gear tooth. A tra-
verse with very little error, as shown in Fig. 1-15(a], is consid-
ered a quality lead. Teeth often are crowned for the purpose of
keeping contact loads off the ends of the teeth. This is accom-
plished by cutting a taper toward the ends of the teeth and leav-
ing the central areas somewhat thicker. Figure 1-15(b) shows a
crowned condition of approximately 0.0015 in. on a side. Note
that the ends of each traverse line are at the same position on the
chart, indicating a true lead. Lead charts also can be used for
diagnosis of problems or conditions that may exist. Figure
1-15 (c) shows a gear tooth tha t has a true lead bu t shows pi tting
along the pitchline where the indicator had traversed. Note that
the pitting is central on one side of the tooth but is off-center on
the opposite side, which would indicate that the lead of the mat -
ing tooth may have a tapered condition on one side or that it had
been deflected under load . Figure 1-15 (d) shows a tooth tha t has
considerable lead error that could be due to either a tapered
condition or an actual lead error in cutting. Note also that the
tooth had been crowned on both sides, but because of the errors
in lead, the crown has shifted toward the ends of the tooth and
each side at an opposite end. Figure 1-15(e) shows the amount
and position of abrasive wear on two teeth in opposite positions
on a gear but on the same side of the teeth. Figure 1-15(f) shows
a gear tooth that is crowned and well centered , but 0.006 in. on a
side seems to be abnormally high.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
16
~\
----.,
I

. 1 I i ,.
\ .

l\ /
)
j
I

(a) (b) (c) (d)

(e) (f)

(a ] True lead a long both profil es . [b) True lead . tooth crow ne d 0.0015 in . per
s ide . [c ) True lead with c row n. Ind icato r s ho ws pi tting a lo ng th e pit ch line. (d )
A cro w ne d tooth with co ns idera ble lea d er ro r. sh ifting th e crow n toward op-
po si te ends of ea ch pr ofil e. [e ] Lead pa t tern of load ed side of two ra nd om teeth
of sa me gear; result du e to a bra s ive wea r. If ) T ru e lead . to oth crow ne d 0 .006
in . per side. w hic h see ms abno rm all y high.
Fig. 1-15. Charted lead patterns.

What can result from the above lead characteristics is de-

scribed here. Figure 1-16 is a photograph of a "perfect" gear; i.e.,
the involute is correct, the lead is true, and there is no runout or
taper. But every tooth has failed at one end in one direction by
heavy tooth contact at that end . The reason? It had been mated
 BASIC UNDERSTANDING OF GEARS
17

Fig. 1-16. Helical gear. All tooth characteristics were

"perfect." Field failure had been matched with pinion
showing lead pattern of Fig. l-lS(d) .

with the pinion teeth shown in Fig. 1-15(d) . Refer now to Fig .
1-17. which shows a crack at the center of the tooth contact area
characteristic of tooth crushing through the case. The lead was
true. the case depth was within specification. and the core
hardness was normal for the material. The reason for the crack?
The tooth had been matched with the pinion teeth shown in Fig.
1-15(f).
It is time to mention a very important consideration of fail-
ure analysis . The actual part that failed will show th e mode of
failure, but very often the cause of failure is to be found in the
mating or matching part. One should always inspect both parts
very closely for the solution to the problem. In fact. if there are
several components to an assembly. each component must be
suspect until eliminated by the examiner.

Mating-Tooth Contact Pattern

It is difficult to determine any graphic method of checking
characteristics of a spiral bevel gear tooth . Therefore. a mating-

Fig. 1-17. Spur gear. Tooth crushing at mid profile contact

area. Reason: mated with pinion showing lead pattern of
Fil!. I-15m.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
18
tooth contact pattern method has been established. The sup-
plier and customer should agree on a contact pattern acceptable
for the application.
The tooth contact pattern method is as follows: A prepared
substance is brushed onto the tooth profile of both matching
parts; the parts are then run on a tester at a predetermined posi-
tion; the parts are first run at no-load and the pattern is noted ;
and finally, the parts are run under a predetermined load and
the pattern is noted. The pattern is the position on the profile of
each tooth that runs tightly against its mating tooth, thus push-
ing away the applied substance. The substance used was com-
monlya mixture of litharge (lead-oxide) and oil. However, the
use of lead products is now highly discouraged, and two pro -
prietary compounds [one white>and one yellows) are available.
In general, the normal tooth contact pattern under no-load
will be central profile toward the toe end. Ideally, as the load is
increased to full-load, the deflection of the tooth causes a shift
of the contact to a full-length pattern [shown in Fig. 1-18a
and b).7
There are also a large number of usually undesired varia-
tions found in tooth-to -tooth contact patterns. They are usually
found during the development stages of gear tooth cutting and
can be eliminated by readjustment of cutting. These are shown
in Fig. 1-18[c) through [pl .
This method of checking tooth contact patterns is very val-
uable in determining actual field loading applications. The test
machine can be set on the specified positions to see if the parts
had actually run as expected. Then the tester can be readjusted
to show the prevailing conditions [such as deflection or mis-
alignment) under which the parts had been running in the field.

Backlash
Backlash is the rotational arc of clearance between mating
gear teeth when the gears are set at the proper mounting dis-
tance. Ideally, when a very accurate involute is generated on
both parts, they will rotate smoothly and easily with zero back-
lash. However, this condition precludes the use of any lubri-
cant. An intentional backlash is engineered into the design to
compensate for many factors, including lubrication, differences
 BASIC UNDERSTANDING OF GEARS
19

~
(a) Cen tral toe bea ring
~
(b) Des ired bea ring und er full load

~
.....
.:;;~:t

(c) Toe bear ing

a s
(d) Heel beari ng
.;:::;::::'

(e) Cross bear ing

~ ~ ~
(f) Low bea ring (g) High bea ring (h) La me bear ing

~ ~ ~
(i) Wide bear ing
..............:...

(j) Narrow (k) Bridged

(pitch line ) bea ring (profile) be ar ing

~ ~ ~
(I) Long
(m) Short bearing
(n) Bridged
(full le ngth) bearing (lengthwis e) bear ing

~
(0) Bias in
~
(p) Bias out

Ske tch es illustra te too th bearings on the pini on toot h . Alth ough a left -h and
pinion is us ed t hroug h out . th e beari ngs are representa tive of those on a rig ht -
hand pin ion or a s tr a ig ht bevel pinio n as well .
[a] Ce n tra l toe beari ng. Note tha t t he beari ng ex te nd s a long a p p rox ima te ly
one -half the toot h len gth a nd th at it is nea rer th e toe of the toot h th an the heel.
In a d d it io n, th e bearin g is re lieved s lig h tly a long th e face a nd flank of th e
too th . U nder light load s th e tooth bea r in g s hou ld be in th is pos it ion on th e
tooth.
[b] Same tooth as in [a] with a be aring a s it s hould be under fu ll loa d . It
shou ld show s lig h t relief a t the ends a nd a long th e fa ce a nd flank of the tee th .
T here s ho uld be no load co nce n tr a tion a t the ex tre me edges of th e teeth .
[c] thro ug h [e ] s ho w d iffe ren ce s in s pira l a ng le be t wee n th e ge a rs test ed . If )
th ro ugh (h) s how d iffer e nces in p res su re a ng le bet w een the ge ars tes ted . [i ]
throug h [k] illu st ra te w idth of to oth be ari ng. II) thro ug h [rn] illustrate len gth
of toot h beari ng.
(0) a nd [p] illu s t rate b ia s bea rings . Rega rd less of the hand of spi ral o n th e
pin ion , " bia s in" w il l alw a ys ru n from the fla nk a t the to e to the top a t the heel
on the con ve x si de a nd fr om th e top at the toe to the fla nk at the hee l on the
co nca ve s ide .
Fig. 1-18. Tooth bearings .
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
20
of involute profile, tooth tip interferences , tooth runout, tooth
deflections under load, dimensional changes of teeth due to heat
treatments, and size changes of gears due to a rise in tempera-
ture during operating conditions.
A designed backlash allows the teeth to be cut slightly
thinner than the theoretical size , with a resulting involute pro-
file that will give optimum contact under load. When an opti-
mum backlash has been established for a gear/pinion set, it is
important that this amount be maintained at the initial assem-
bly operation. The life of a gear set can be prolonged by a slight
reduction of backlash from the amount specified for initial as -
sembly, as long as none of the deterrent factors prevails. A ten-
dency to hit a zero backlash at any time or position will imme -
diately become destructive. On the other hand, as backlash may
be increased at the time of assembly, the expected life of the
gear set is greatly reduced. This condition accelerates improper
contact and brinelling.

Composite Errors
When all errors of dimensional characteristics and distor-
tion are added together, the result is the overall, or composite,
error. Many times they are additive and many times they are
subtractive; thus, the resulting composite error may vary from
zero to a major magnitude from gear to gear, even within the
same manufactured and processed lot.
L. D. Martine did an extensive study of the effect of compos-
ite error and found that its influence on gearing was affected
most by the contact ratio of the gears . Composite errors were
not affected by a 1.0 contact ratio but tended to integrate rap-
idly toward zero effect as the contact ratio increased-graph-
ically represented in Fig. 1-19 .

Associated Parameters
The importance of the function of the mating part has been em-
phasized . There are also components in the design and struc-
ture of each gear and/or gear train that must be considered in
conjunction with the teeth.
 BASIC UNDERSTANDING OF GEARS
21
0 .0016

I .1 .I I I .1
Integrat ing effect of conta ct rauo _
:;: 0 .00 14 on tooth- to-tooth co mposite e rrors
"c
~ 0 .0012
Q;
s
·iii \
0 .0 01 0

\
0
c.
E
0
" 0 .0008

\
s:
15
E Tooth -to-tooth co mposite
8 0 .0006 e rror inte gration curve
-
i:.
15
0
t- 0 .0004
1\\ /
0 .0002
r-,
~
/
1.0 1.2 1.4 1.6
"-
1.8 2.0
'"-
2 .2 2.4 2.6 2.8 3.0
Conta ct ratio

As con tac t r a tio in c rea ses. increasingl y la rge t ooth- to-t ooth com posit e e r-
rors a re s mo o thed ou t. T h is curve is based on empirica l d at a ob ta ine d in a
series of tes ts . II is pred icat ed on rea so n ab ly we ll -c u t gea r s a nd is ind epe n-
de nt of pit ch a nd pressure a ng le.
Fig. 1-19. Integrating effect (smoothing) of contact ratio.

Round Bores
A round bore , with close toleran ces and ground . may rotate
fr eely around a ground shaft diameter. may fit ti ghtl y against a
gro und diamet er b y ha ving a press fit . or may be the outer race
of a ne edle bearing that giv es fr eedom of rotation. Each applica-
tion has its own unique problems . A ground bore is always sub-
ject to tempering, burning, and ch ecking during the grinding
operation. A freely rotating bore requires a good lubricating
film, or seizing and galling may result. The bore used as a bear-
ing outer race must be as hard as any standard roller bearing
surface and is subject to all conditions of rolling contact, includ-
ing fati gue, pitting . spalling, and ga lling . The bore that is press
fit onto a sha ft is subject to a definite amount of initial tensile
stress. Also , an y tendenc y for th e bore to slip under the applied
rotational forces will set up a unique corrosi ve action betw een
 SYSTEMATICA NALVSIS OF GEAR FAILURES
22
the two surfaces, which leads to a condition recognizable as
either stress corrosion or fretting corrosion-the two having
subtle differences, even though they are very similar in nature.
A spur or helical round bore gear acting as an idler (or re-
versal) gear between the input member and the output member
of a gear train has an extremely complicated pattern of stresses
to contend with. The most common application is the planet
pinion group in wheel reduction assemblies or in planetary-
type speed reducers. The photoelastic study shown in Fig. 1-20
reveals certain facts:
[a) The tensile and compressive stresses in the bore are
caused by bending stresses of the gear being loaded as
a ring.
[b) The maximum tensile and compressive stresses in the
bore increase as the load on the teeth increases.
(e) The maximum tensile and compressi ve stresses in the
bore increase as the clearance between the bore and the
shaft is increased.
[d) The maximum tensile and compressi ve stresses in the
bore increase as the ratio of the size of the bore to the
root diameter of the teeth is increased.
( e) During one revolution of the gear under load, the teeth
go through one cycle of complete reversal of stresses,
whereas each element of the bore experiences two cy-
cles of reversals.
Three modifications of the round bore alter the stress pat-
terns considerably:
[a) Oil holes that extend into the bore are intended to lu-
bricate the rotating surfaces. Each hole may be a stress
raiser that could be the source of a fatigue crack.
[b) Tapered bores are usually "shrunk" fit onto a shaft.
This sets up a very high concentration of stresses not
only along the ends of the bore but also at the juncture
on the shaft.
( e) A keyway in the bore also initiates a high stress-
concentration area. A keyway is also required to with-
stand a very high, continuous load. In fact, the applied
load to the side of the keyway is directly proportional
 BASIC UNDERSTANDING OF GEARS
23

I T I
5.5 ,
.- - /'

Fig. 1-20. Photoelastic pattern of idler gear between load

input pinion and load output gear.

to the r ati o of tooth pi tch line radi us to t he radius of th e

k e y w a y p o s it ion. Fa t ig ue fa i l u re a t thi s p oint is
co m mo n .

Splined Bores
The load s a p p lie d to th e splined b ore a re a a re al so directl y
p rop o rti on al to th e ra tio of pit chline radiu s of th e teeth to the
pit chline ra di us of th e sp li nes . H o w e ve r , th e loa d is d is trib uted
eq ua lly on to eac h spline , so th e stre ss pe r s p line is not us uall y
e xces s i ve . It is po ssibl e for a n ou t-o f- ro u nd condition to ex is t
that w o uld co nc e n t ra te th e load s o n s lig h tly more th an tw o
s pli nes . A lso , a tape r ed co nd it io n wo uld pl a ce a ll loadings a t
 SYSTEMATIC A NALYSIS OF GEAR FAILURES
24
on e end of th e splines . w h ic h would be detriment al both to th e
gea r s plines and to th e shaft. Heat tr eating of th e splined area
mu st also be w at ch ed clo sely for quench cra cks . Grinding of th e
fac e against the end of th e s p li ne s can also cause grind ing
checks to radiate from th e co rne rs of the root fill ets .

Shafts
Th e shafts within a gea r train [as well as th e shank of a
pinion , which consti tut es a sh aft in function] are of import an ce
to load -carrying capa cit y a nd load distribution. The y a re con-
tinuall y ex p os ed to torsi on al loads, both unidirecti on al a nd re-
versing. The less ob vious st re sse d cond it ion that is eq ua lly im -
portant is bending. Bending stresses ca n be id entifi ed a s
unidi rectional, bidirectional, or rotational. Wh en th e type of
s tre ss is identified, th e ca use s fo r s uc h a stress can be explore d .
A number of stresses a p p lie d to a shaft can be imposed by
parts riding on it. For inst ance, heli cal gears will transpos e a
bending s t re s s , as will straight a nd spiral bevel ge a r s ; round
bores ma y be tight e noug h to ca use scoring and galling; a
s pline d bore may ca use hi gh stress conce n tra tion at it s end fa ce;
runout in gea r s ma y ca use rep eat ed defl ections in bending of
th e s ha ft; a nd loose bearings ma y ca use exc essive end pla y and
mor e bending.

References
1. P aul M. Dean , [r. , Geomet r y a nd Theor y of Ge a rs, in Gea r Manu -
facture a nd Performan ce. Am erican Soc ie ty for Metals, 1974. p
1-24.
2. Paul M. Dean. [r .. Geometr y a nd Theor y of Gea rs . in So urce Book
on Gea r Des ign. T echnology a nd Performan ce. Ame rican So cie t y
for Met als. 1980. p 1-24.
3. Stand ard s pu bli shed b y the Am eri can Gear Manufa cturers As so -
cia tio n, Suit e 1000 . 1900 North Fort M yer Dri ve, Arlington. Vir-
ginia, 22209.
AGMA 112.04 "Gear Nomenclature," A ug us t 1965.
AGMA 115 .01 " Basic Gear Geom etry ," Jul y 1959.
A GMA 116 .01 "G los sa ry . T erms Use d in Gearing ," October 19 72.
 BASIC UNDERSTANDING OF GEARS
25
4. Don a ld'. Wulp i, T orsi on al Frac ture of S hafts. Meta l Progress. Vol
120 (No .3 j, Aug us t 1981. P 26.
5. #35 W Gear Marking Com po und. Way ne Ch emical Products Com -
pany. 9470 Copla nd Street. Detroit . Mic higa n 48209.
6. #10 Yellow Gear Marking Compound. P res co tt & Company. Ltd ..
2625 Rue Pare Street, Montreal. Q uebec H4P-1S 1.
7. Gleason Works P ub lica ti on S D3025 B. 1955. in Source Book on
Gear Design . T echnology and Performa nce. American Society for
Metals. 1980. p 341.
8. L. D. Martin. Large Contact Ra tios Mi nimize Effects of Gear Er-
ro rs. Ma chinery. Vol 73 (No .6). February 1967. p 97 .
 CHAPTER 2
Basic Understanding
of Environment

Environment is a set of conditions that exists in, around, and

about an object, and has had, is having. or can ha ve an influence
on that object, either for betterment or for detriment during its
lifetime or its present acti vity.
It is very important, when studying a gear failure, that the
examiner obtain an understanding of all possible environmen-
tal factors. In order to accomplish this, he must determine the
things to look for; and. more important. he must realize that it is
indeed "things" he is looking for (Fig . 2-1). Yet these things are
so intricately intertwined that it is very difficult to assess a
quantitative value for anyone of them. An attempt should be
made, however; but how all-i nclusive is environment when re-
lated to things? For instance, in a survey conducted by the au-
thor, a question-what is environment?-was asked, and the
answers given were lubrication, temperature. and mechanical
stability. It is a lso interesting that invariably the negative re-
spo nse was give n; for examp le, lack of lu b rica n t, or dirt in oil;
parts running extremely hot; or a lot of deflections in housing,
causing misalignment. Then there was a shift in the respo nses
to the inclusion of subtler details, such as operation of the
equipment, maintenance personnel and procedures. and finally,
management attitude. Are all of these answers interrelated?
They certainly appear to be.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
28

the
vibrat ion
thing

Fig. 2-1. Environmental factors consist of mostly "things":

lubrication, temperature , and mechanical stability.

lubrication
Alth ou gh t he ma jorit y of persons-b oth p rof es s ion al a nd n on -
p rofes sional-t end s to blame lub ri cati on for most ge ar failures ,
th ere is no doubt in the ir minds that lu bri cati on is a must. The
bl ame app ears to re st on not enough of, t oo mu ch of, n ot th e
rig h t kind of. breakd own of. or con ta m inatio n of th e lubri cant.
Looking a t these causes, one immedi atel y recog nizes the m to be
fa ults not of th e lub ricant , but of the a p p lica tio n of th at lu bri -
ca n t, or of a number of ex te rn a l for ces a nd co n di tio ns wo r k ing
ag a ins t that lubri cant.
If one were to co ns ide r what a specifi c lubricant is a n d
w ha t its performance is to be, it would b e ea sy to underst and
th e reasons for it not to function under certain condi ti on s. Lu-
brication is a ccomplished on gear teeth b y th e formation of two
types of oil films : the reaction film, also known a s the " bo un d -
a ry lubri cant ," produ ced b y ph ysical abs orpti on and/or che mi-
ca l rea ction to form a desired film that is soft a nd ea s ily sheared
but diffi cult to pen et rat e or remo ve from th e s urface : and the
ela st oh yd r od yn ami c film that forms h ydrost aticall y on th e
gear tooth su rf a ce as a func tion of th e s urface speed . Thi s sec-
ond film is very thin , has a ve ry hi gh s hear s tre ngt h , a n d is n ot
 BASIC UNDERSTANDING OF ENVIRONMENT 29
affected by compressive loads as long as constant temperature
is maintained.
There are certain natural rules about a lubricant that
should be remembered when gearing is designed, as well as
when it is examined for failure: 1
(a) Load from a gear tooth to its mating tooth is transferred
through a pressurized oil film . If not, metal-to-metal
contact may be detrimental.
(b) By increasing viscosity, a thicker oil film will be devel-
oped (keeping load, speed, and temperature constant).
(c) Heat generation cannot be controlled above a certain
maximum viscosity (for a given oil).
(d) Breakdown of the oil film will occur when the gear
tooth surface equilibrium temperature has reached a
specific value .
(e) Scuffing load limit of mating tooth surfaces is speed-
dependent. With increasing speed, the load required to
be supported by the reaction film decreases, whereas
the load required to be supported by the increasing
elastohydrodynamic film increases. The result is a de-
creasing scuffing load limit to a certain speed as the
reaction film decreases ; then, as the speed picks up to
the point at which the elastohydrodynamic film in-
creases, the scuffing load limit increases , allowing an
increase in the overall load-carrying capacity (assum-
ing no change in temperature that would change
viscosity).
(fl At constant speed , surface equilibrium temperature
increases as load increases, lowering the scuffing load
limit of the reaction film. NOTE: Surface equilibrium
temperature is attained when the heat dissipated from
the oil is eq ual to the hea t extracted by the oil.
Damage to and failure of gears can and do occur as a dire ct
or indirect association with the lubricant. There are several
such occasions.

Incorrect lubrication means either the wrong specification

was given for the application, or the right specification was not
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
30
followed by the user. This means that type, quality, or viscosity
was incorrect, and damage to the gears was accomplished in a
short period of time . For instance, as one type of oil can be used
for spur and helical gears, another type is necessary for spiral
bevel or hypoid gearing. Keep in mind also tha t the bearings
may be running in the same lubricant as the gears, and they, too,
require that certain characteristics are present. A constant jug-
gle of several characteristics of an oil is necessary in order to
satisfy all moving components within a system.

The lubrication system cannot be overlooked or ignored . If

a splash lubrication is working satisfactorily for a certain sys-
tem, it may be completely inadequate if higher speeds are re-
quired . One very important function of proper lubrication is
heat dissipation. Any lubrication system must take into ac-
count that the oil is supplied to the gear teeth at a greater rate
than that necessary for lubrication, in order to remove fric-
tional heat as quickly as possible. If not, the oil films will be
broken down by continuing higher temperatures. After study-
ing many works on the subject, T. I. Fowle- of Shell Interna-
tional Petroleum Company , Ltd., concludes: "Heat is best re-
moved from the gear teeth by spraying the oil onto them when
they are the hottest; i.e., as they come out of mesh. Only in the
case of very high speed gears is there a danger that the amount
of oil left on the teeth on reentry into mesh will be insufficient to
form a lubricating film and, therefore, to need a supplementary
supply at the ingoing side of mesh . And, only in the case ofheav-
ily loaded gears with large numbers of teeth and running at high
speeds, do these authors conclude that better cooling than by
spraying onto the gear teeth is needed." Some lubrication sys-
tems become very complicated; but the more complex the sys-
tem, the more easily it becomes contaminated , and the more dif-
ficult it is to inspect and clean.

Lack of lubricant is perhaps the most devastating condition

that can exist within a unit of moving parts [Fig. 2-2J. There
must be an amount of lubricant sufficient to form the two nec-
essary oil films; but, of utmost importance, there must also be
an amount sufficient to dissipate the heat away from those
 BASIC UNDERSTANDING OF ENVIRONMENT 31

Fig. 2-2 . Comparison of gear sets showing that lack of

lubrication can be most devastating.

films to maintain th eir stabilit y. Leakage is th e most co m m on

mea ns of oil loss , but occas io na lly a n a sse mb ly w ill b e pla ced
int o se r vi ce wit ho u t eve r h a ving th e init ial oil ad d ed. Oil can
a lso be lost by va po riza tio n or so lid ificat io n und e r hi gh
te m perat ures a n d p ressu res.

Excessive lubrication p resent s some un us ua l p roble ms in a

gea r train . Th e churning effe c t resulting fr om exce s s ive lub ri-
ca ti on ca n rai se ambi ent temp er atures. be caus e the en erg y in-
vol ved in cu t ting th e shear plan e of th e oil is reab sorb ed a s h eat.
A co n ti n ua l ch urn ing of th e oil, a long w it h in crea sing tempera-
tu res , w ill tend to so lidify the oil. In so me in s tances , high va por-
izati on pressure ha s brok en seal s, evo lv ing a leakage probl em.

Lubricant contamination is prob abl y th e most co mmo n

lubri cant-a s s oci at ed p robl em a nd is al s o the most co m p lex .
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
32
Atmosphere contamination consisting of oxidizing vapors and
water vapors tends to oxidize freshly worn surfaces, to break
down the oil chemistry, or to set up cells of corrosion on metal
surfaces . Liquid contamination may consist of water [which
may also be contaminated with gases, acids , salt, or mud) , anti-
freeze compounds, liquid coolant compounds, or even other
types of grease or oil added inadvertently. These contaminants
will ca use corrosion and general deteriora tion of both the Iubri-
cant and th e metallic surfaces. Sand and dust particles intro-
duced by atmosphere or water are common solid contaminants.
There is another very common solid contaminant that is
internal to the gear system and is self-perpetuating. As gea r s
are "running-in," an adjustment of the mating surfaces occurs
with high rates of local wear, producing a fine metallic debris
that is small, flaky, magnetic , and abrasive. As contacting sur-
faces become smooth, the rate of this debris production be -
comes nil. For this reason, many gear box assemblies have a
running-in period at the assembly department, or the customer
is instructed to completely flush the system and replace the en-
tire supply of lubricant after 50 hours of operation. Other ex-
amples of solid contaminant are molding sand remaining in cast
iron housings, welding rod splatter from repaired housings,
drillings and turnings not cleaned out of machined parts, chips
from nicks and bumps during assembly, carbon buildup from
"burned" oil, shot and grit carryover from the gear-cleaning de-
partment, lapping compounds, grinding-wheel breakdown par-
ticles, and floor-cleaning compounds. Then, of course , when-
ever there is an initial stage of failure, those particles become
contaminants that can easily propel further progression in as-
tounding proportions (Fig. 2-3).

Temperature
Well over 90 % of all lubrication has one specific function-to
remove heat. It is difficult to separate this discussion into two
distinct categories, such as causes of temperature increase and
effects of temperat ure, beca use the causes and effects may be so
closely related as to be simultaneous. For instance, two asperi-
 BASIC UNDERSTANDING O F ENVIRONMENT 33

ties come in direct co ntact with each ot her and one is sheared by
t he o ther. T he e ne rgy of s hearing is co nve rted imm edi at el y to
a n extreme ly h igh tem pe rat u re at the s hear p oin t that is above
the flash point of the lu b rican t. and a lso above the critical
tem perature of s tee l tra nsformation . Th e quenc hi ng of this in -
s ta n taneo us p oint of h eat n ot only causes a slig h t tem per atu re
rise in t he lu brican t. b ut burns a pa r ticl e of o il th at fo rms a min -
ute amou nt of free carbo n blac k a nd t rans for m s the steel spot to
a n u nte m pere d m a r ten sit e. Th e ma rt en sit e ma y b e s usce p ti b le
to c rac king o r to becomin g a n a bras ive point d uri ng co n ti n ue d
serv ice . In o the r wo rds. th e m ost co m mo n ca use of tem per ature
rise is th e co nve rs io n of mech ani cal ene rgy to heat. w hic h is
mea sured b y temp erature. A fe w of th e more ob vious so urce s of
tempe ra tu re vari ant s w ill b e di scu s s ed in thi s cha pte r . thou gh
th e d iscu ssi on wi ll not be a ll- inc lus ive.

Lubricants ha ve a ce rta in tena cit y o r resist ance to s hea r.

The hi gh er th e viscos ity. the h igh e r the s hear s tre ng th . As th e
gears a nd o t her movi ng parts cut th r ou gh the oil. t he energy
invo lved cha nges to heat a nd t he temperatu r e rises. It w il l co n-
ti n ue to ri se until th e viscos ity is redu ced. th e s hea r s tre ngt h is
reduced . and a s urface equilibrium tem pera ture is r ea ch ed . A
surface equi li bri um tem pe ra t ure is at tai ned w he n the hea t dis-
s ipated fro m th e oil is e q ua l to the hea t ex trac te d by t he oi l. An y
s u bse q ue n t cha nge in speed or load w ill cha nge th e s urface
eq uil ib rium temper ature.

Fig. 2-3. Abrasive wear progresses rapidly and uniformly

when an abrasive contaminant saturates the lubricant.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
34
Ambient temperatures of th e surro undi ng atmosphere wi ll
d irect ly infl uen ce the tem pera ture a nd viscosity of th e lu bri-
cant. Ex tremely low temperatures may cause a lub rican t to be -
come so viscous it cannot function. In these ins tances, the oil
may be heated externally and used in a reverse function, trans-
mitting heat back into the gears . More frequent , though, is the
occurrence of higher atmosp heric temperatures . As tempera-
tures in c rea se . viscosity decre ases, the oil film on gea r teeth
becomes thi n ner, a nd th e fr ic tio n loa d in crea ses , caus ing sc uff -
in g a nd sco ring a nd hi gh er temperat ures. T he op ti ma l co ndit io n
exis ts w he n a s urface eq uil ib r ium te m pera t ure is maint a ined
wit h a s ufficie nt oil fil m remaining.
T w o id entical unit s were ope ra ting in th e same localit y in a
h ot , ari d climat e. Both w ere h igh- speed gea ri ng un it s ex posed
to direct s un lig ht. On e use r lik ed bright , s hi ni ng eq uip me n t, s o
his uni t was covered wit h a bright , met all ic, a l um in um fi n is h.
T he o ther unit w a s covere d w ith a dull , bl a ck fi nis h . The s un
heat abso rbe ncy of th e bl a ck un it was so mu ch grea te r tha n th at
of th e met alli c un it th a t it a llowed prem a t ure sc uffing of the
gear teeth in the b lac k unit.
Ris in g temperatures will expa nd the volume of bot h th e
lubricant and the entrapped vapor in a closed unit. As volume
in c rea ses , pressure increases, a nd sea ls may break loose . H ow -
ever, th e h ousing of th e closed unit is usually made of steel, cast
iron , or cas t a lumi n um-a ll three w i th a hi ghe r ther ma l coeffi-
cie nt of expa ns io n th an th e li qui d s a nd gases ins ide, w hic h w ill
te nd to a lleviate the in tern al press ures. Agai n, hig her te mp era-
tures w ill increase the vaporizatio n of the lub rica n t , a nd the
vapor pressure w il l increase . If th e va po r volume is large co m-
pare d to th e l ub ri cant vo lume, th e va por p ress ures w ill rise
s low ly a nd ma y not be a fac tor. H ow e ve r, if th e va por vo lume is
s ma ll co m pare d to th e lub r icant vn lume , the va po r pres sures
may in crea se ra pidly a nd b ecome d angerou s.

Gear tooth operating temperature is a s ig nifica n t so urce of

heat becau se loa d s tha t ca use fric tio n a nd deform a ti on of th e
su rface are changed fro m mec ha nical energy to hea t ene rgy . For
insta nce, under norma l loading co ndi tio ns of 3300 lb . per in ch of
fa ce , a nd a speed of 7500 rp m, a s pur gea r toot h was calc ulated
to ha ve a rise in surface eq uilibri um tem pera tu re of 27 of [15°C)
 BASIC UNDERSTANDING OF ENVIRONMENT 35
a t th e point of ma xi m um co m p ress ive a nd s li d ing s t res s. which
was ju st b el ow th e p it chlin e o n th e ded endum as th e teeth were
com ing out of me sh ."

Components , su ch as gears, bearin g s , brak es , s h ifte r s , and

sea ls, a ll tend to h a ve rub bing co n tac t or fr ic tio n . The frictional
energy is co nve rted to heat a nd s ho uld b e a b so rbe d b y th e lu-
bri cant. As lon g as there re mai ns a n oil film b etween two metal-
li c surfa ces, fri ction w ill be a t a min imum a nd metallic se izu re
or s cuffing w ill no t oc cur . A s th e s urface eq uil ib ri um tempera-
ture rises . the prote c ti ve film over the gea r teeth decreas es. At
so me point , th e firs t d anger poi n t is reach ed ; i.e., th e film ca n no
lon ger w i ths ta nd th e load , a nd met al-t o-m et al co n tac t occ urs .
Rubbing , sc uffing, scoring , seizing , a nd ga lling of t he ge a r te eth
ens ue, w h ic h ra pi d ly in crea se th e a m oun t of ge ne ra te d hea t.
The heat ge ne ra tion m a y p rog re s s fas te r th an th e lubri cant will
a b s orb. In th at ca s e, t he gear it s elf will rea b s orb th e heat, it s
tempe ra t ure w ill ri se, a nd th e seco nd d ange r point beco me s ev i-
den t: th e tem pe ri ng of th e gea r.
The tempe rin g of th e gea r, w he n certai n temperatures a re
r ea ch ed , re d uces th e hardnes s a nd a ll ows a fa st er pr ogressi on
of fric tio na l wear. For ins ta nce . a ca r b ur ize d gear is pla ced in
se rvice a t a hardnes s of 61 - 62 on th e Rockw ell "C" sca le [HRCJ .
Th e n orm al tempering temper ature during produ cti on w as
325 OF. An y se r vice- in d uce d tempera tures up to tha t p oint will
ha ve n o effec t upo n th e orig ina l ha rdn es s . How eve r, s u ppose
hi gh er te m pe rat u res are s us pected . Wh at are th e clues ? N ot
on l y is la c k of h a rdn es s a cl ue , but [if a ir is p res en t J th e surfa ce
of th e part s wi ll cha nge col or. T h is is ca lle d " te m pe r co lor":
Temperature Ha rdn ess ex pec te d
(OF) IHR CI Temp er color

350 60 -62 No ne
400 57 -59 Ligh t st raw
450 55-57 Da rk straw
500 52- 55 Dar k p ur p le
600 47- 52 Me d ium blue
700 35-47 Very lig h t b lue
Th e thi rd d anger point s oo n becom es a p pa re n t. No t on ly d oes
th e lub ri cant cease to exis t in a ny s tate of us efulnes s, but th e
gea r teeth melt a nd becom e ex ti nct (Fig. 2-4 J.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
36

Fig. 2-4. With continuing rise of temperatures, the melting

point is ultimately reached.

Mechanical Stability
All gears within a composite system depend not only on their
interaction, but on the action and reaction of every other com-
ponent within the system. However, before any action or reac-
tion occurs within the unit, each component must be perfectly
aligned with its functioning counterparts [Fig. 2-5J. Reviewing
this important step, in the first place, each part must be ma-
chined correctly with the awareness that each dimension and
surface finish is crucial not only to itself, but to its functioning
counterpart. Any mistake in the quality of manufacturing may
be a multiplying factor in the breakdown of the entire unit.
Size change in gears that occurs due to change in tempera-
ture might have been discussed under the "temperature" head-
ing; but size change is also mechanical. Nevertheless, the di-
ameter of the pitch circle of a gear will increase in size in direct
proportion to an increase in temperature. The linear coefficient
of thermal expansion for the average gear steel is 6.36 X 10-6 in.
per inch per degree Fahrenheit. Why is this important? Many
gear sets running at high speed with minimal backlash will
soon reach a zero or negative backlash with an unexpected rise
 BASIC UNDERSTANDING OF ENVIRONMENT 37

in temperature. The ensuing pressure will break down the lubri-

cation film, resulting in high frictional wear.
It is imperative when inspecting a failed gear to ask from
what direction this failure occurred and what conditions ex-
isted to cause this directionality. These questions invariably
will lead the analyst to check alignment, deflections, stability,
tooth characteristics, and dimensional accuracies .
These accuracies are so very important that any deviation
from the norm will tend to throw a tooth contact pattern from
its normal central area to perhaps an end-face contact, or from a
broad contact to a point contact. Each condition will drastically
increase the applied load per unit area, which tends to increase
immediately the surface equilibrium temperature and to break
down the lubrication film.
In any transmission or power train equipment there are vi-
brations. Even the smoothest source of power will generate an
output in pulses. Roller bearings vibrate; gear teeth vibrate; ro-
tating shafts vibrate. Each part vibrates in its own peculiar am-
plitude and frequency. There is usually no problem unless one
or more add together proportionately to form a high noise level,
or to form a sequence of extremely high load peaks. Some ana-

Fig. 2-5. Perfect alignment is necessary for accurate

results.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
38
lysts claim that surface fatigue is caused by torsional vibration
in approximately 85 % of the cases. This figure may be high, but
torsional vibrations, or chatter, will often reduce the fatigue life
of a unit drastically.
Interestingly, under the heading "Life Adjustment Factors
for Environmental Conditions," the Timken Companylists the
following factors , relative to tapered roller bearings: load zone
(as a function of end play, preload, geometry, clearance, tem-
peratures, and housing deformations); misalignment ; and
Iubrication.'

Personnel-Related Activities
This subject is as much an environmental factor as any of the
above and should be recognized as such.

Assemblymen putting parts together in one unit or in plac-

ing several units into a composite machine must be scrupulous
regarding cleanliness and accuracy of alignment. There is no
allowance for error at this point of the process. Contamination
at this point is instantly destructive. Misalignment will imme-
diately place one or several parts under stress.

Operators of the equipment are the first line of defense. A

good , conscientious operator will have his equipment outlast
any other in the field. The author has said many times: "When a
catastrophic failure occurs, don't look for a microscopi c cause.
Find the cowboy at the wheel or the one that threw the monkey
wrench into the works!"

Maintenance personnel, equipment, and procedures will

make or break a company with several pieces of equipment that
must be kept in good working order. Not only must good re-
placement parts be used, but care should be taken that all used
parts are cleaned and thoroughly inspected before being placed
back into service. No gear should be reused unless a magnetic-
particle inspection method has determined a freedom from
 BASIC UNDERSTANDING OF ENVIRONMENT 39

crac k p rop aga tion. No mated gear or pinion shou ld be run un-
mated or with a random part. Again, cleanliness and correct
alignment must be maintained during the reassembly of re-
paired ite ms. T here are man y use rs of eq uipme nt w ho ha ve p ro-
gra ms of p reventi ve maint en ance, w hic h are in va rious deg rees
of so p histicat ion fro m we ll-t houg ht -out. sys temat ic chec king
of all compo ne nts to a yea rly hit -or-miss checkup of only the
major ope rating pa r ts . It appears that the slogan of some pro-
grams is " If it b reak s d own . fi x it !" The utilizati on of th ese p ro-
grams lead s directly in to th e next ca tegor y of per sonn el -rel a ted
ac tiv ities- ma nageme nt a tt itude .

Management attitude. Th is is a n as pect re lated t o pe rso n-

nel th at is ofte n ove rloo ke d . T hose in management need to de -
term ine h ow the ir a tt it ude s affec t th e work hab its of those
worki ng for them . Very few assembly . mai ntenance, or opera-
ti on al men w ill co nsiste n tly do a job bet ter tha n tha t w hic h is
ex pec te d of th em. Ex pec t th e bes t from yo ur wo rkers, a nd you'll
ge t th e best from yo ur eq uipme n t. Give t he m goo d tool s a nd
good equi pme nt alo ng wit h your good attit ude , and they will do
a good job .
Seve ral years ago. a ro utine in sp ec ti on of a new plane ta r y
syste m w as s po nso re d by a maj or user a nd th e gea r manufac-
turer . T he in spec tion was d on e on th e wor ki ng sites, w hic h
were ope n-pit mines located over the full b rea dt h of bot h Can-
ada a nd the United States. In on ly four of 143 insta nces were
there fou nd nega tive enviro nme n ta l co nditio ns :
• A truck th at pulled up for inspec tion was making a lot of
rea r- end n ois es a nd ha d bee n fo r q uite a w hile. It was
discovered tha t a la rge idler gear ha d broken away from
the hold in g- ca p screws a nd was ru nning free ly . The gear
was being held in pl ac e partially b y the side of th e hou s-
ing, whi ch b y the time of discove ry had w orn a large area
1f4-in. deep in to th e h ou si ng.
• T he cover plate of a w hee l unit was removed, a nd out
dro pped a blob of bl ac k grease wit h the co nsistency of
lard. Th e rea s on ? The bott om drain plug had been bat-
tered b y a la rge boulder a nd co uld not be re moved . Con-
 SYSTEMATIC A NALYSIS OF GEAR FA ILURES
40
se q ue n tly , th e unit h ad ne ver been d r ain ed or re fille d .
Th e tran sform ati on of oil to la rd result ed from a co ns ta n t
churning for o ver 43,000 mil es . The gea rs w er e in ver y
good cond itio n .
• Wh en th e cove r pl at e of a no the r w hee l unit had been re-
m o ved , it w a s di s co vered th e gea rs we re run ni ng in a
b ath of wa ter w it h only a s ma ll a m o un t of oil. An ex p la -
nation wa s forth coming fr om th e maint enan ce forem an:
" I g uess wh en I told th at ne w b o y to fill up the un it w i th
oil from a 5-ga ll on ca n whi ch was across t he roo m, h e
must h a ve p icked up th e op en can b y th e drinking foun-
ta in b y mi st ake." The gea rs w ere s ti ll in very goo d
conditi on .
• In o ne so u thwe s te rn st a te, tw o d ri vers we re in const ant
a rg ume n t a s to th e s tre ng th of th e tru ck eac h w a s d ri v-
in g. On e t ru ck had a co nve n tio na l d rive co n s is ting of
dri ve shaft , di ffe rential, red uc tio n ge ars , a nd ax le s h a fts.
Th e s eco nd tru ck had elec t ric m otor-d r iv en w hee ls th at
dro ve thr ou gh a pl ane tar y gea r set. Th e cri terion for se t-
tling t he a rg u me n t? Nos e both load ed tru ck s aga in s t t he
ve r ti ca l w all of the pit a nd g un the e ng ine. Whi ch e ver
tru ck st alled the eng ine lost. The re s ult? The tru ck with
th e conve n tio na l ge a ring st alled o u t fi rst. It lost. It a lso
lo st tw o twi st ed a nd split a xl e sh aft s, tw o brok en a nd
m an gl ed wh eel -r edu cti on as se m b lies, a n d on e m a sti -
ca te d differen tial. The e ng i ne of the second tru ck ne ver
eve n sl owed d own! It w on th e pri vil ege of ha ving t w o
melted 350-h p , DC m otors repl a ced , a s w ell a s a ll th e
res is ta nce panels a n d w iring . Tw o dri ve rs w ere out look -
in g for new jobs.

References
1. W. J. Bart z, The Influen ce of Lubri cant s on Failures of Bea rin gs
a nd Gea rs , in So urce Book o n Ge a r Desi g n. T ec h no logy a n d Pe r -
fo rma nce, A me rica n S ociet y for Metal s , 1980 , p 172-1 83.
2. T . 1. Fowl e, Gear Lub ri ca tion: Relat in g Theor y to Practi ce, in
So urce Book on Gear De s ign. T ech n ol og y a nd Pe rform an ce,
A merica n Soc ie ty for Met al s, 1980 , p 205 -219 .
 BASIC UNDERSTANDING OF ENVIRONMENT 41

3. A la n Hichco x , Computer-Aided Gear Lubrication. in Pow er

Tran smi s si on De si gn. Pent onllPC. Ma y 1983. Vol. 25 (No . 5 1.
P 19-20.
4 . T h e T imk en Comp an y. Bearin g S e le cti on Handb oo K. 1983.
p 20 -26 .
 CHA~R 3
Systematic Examination

Th e actual part that fail ed will show th e

mode of failur e: but ve ry often th e ca us e of
failure is to be found in th e matin g o r mat ching
part. O ne s ho uld alwa y s in sp ect both parts
ver y closel y for the co r rec t an sw er to the prob-
lem . In fact. if there are seve ral com pone nts to
an asse m bly . each com pone nt mu st be s us pec t
until elim ina ted by the e xamine r.

S ystemati c e xa m ina ti on implies the us e of a sp ecific s ystem or

procedure to finalize a conclusi ve answer. It ma y sound labori-
ous, and it ma y e ve n be boring; but s ystematic means methodi-
ca l in proc edure or plan, and th is means to begin at the
beginning.
There a re three types of approach es that hav e al wa ys be en
frustratin g bec ause so much information is lacking.
Approa ch No.1. Sev eral ver y good books and man y articles
have been written on failures . Inv a ri abl y, eac h is a case
histo ry ; that is, a pi cture is shown, and th e ca use and ef-
fect are explained. The assumption is th at any part that
looks lik e the picture should fail in the same way for the
sa me reason. This is not so, for ev ery failure is an indi -
vidual ca se and must be treated as su ch. The unfortunate
fa ct a bou t a case history is th a t the a u thor oft en d oes not
exp lain th e steps he took to re ach his concl us ions .
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
44
Approa ch No.2. Often a failed gear or a small segment of a
failed gear is set on a failure analyst's desk with the sin-
cere question, "What kind of a failure is this, and what
caused it?" Most of the time, an immediate answer is re-
quired, though often the background information sup-
plied is grossly inadequate to enable an analyst,to find a
solution quickly. It is a difficult situation for an analyst
to be in, but a tentative answer must be given, based
solely on past experience and a whole bunch of "ifs ."
Approach No.3 . This approach is illustrated by an exam-
ple: A man was using his electric shaver one morning
when it stopped cutting and started pulling . The change
in tactics was painful. He had difficulty removing the
cutters from the head because they had not been cleaned
for a long time and were severely caked with whiskers
and grease. When all the parts were laid out, he found a
broken actuating spring. He was about to brush all the
parts and the gunky mess into a white envelope and take
them to the shaver supplier, but instead, he did what any
good engineer would do. First, he washed all the parts
under the hot water faucet to remove the grime. Then he
put the clean parts in a white envelope and took them to
the supplier. "Look what happened to this spring," he
said. "Shouldn't I ha ve a new sha ver?"
This is not meant to advocate the return of all gears in a
grimy condition, but it would certainly be helpful to the exam-
iner for the environmental conditions to be carefully docu-
mented . The examiner should visit the field application to see
first -hand the failed parts as they are being disassembled and
to observe the conditions under which they had operated. Since
the metallurgical examiner rarely has this opportunity, field
service personnel assume this respunsibil it y. The following in-
structions are specifically for them.

Field Examination
In the examination of failed systems, documentation is very
important. It is not a good practice to rely on memory or oral
 SYSTEMATIC EXAMINATION
45
transcription. P h o tog ra ph s and sketches are excellent for doc-
ume ntat io n, a lo ng wit h writte n hi st or y. Record th e t yp e of
eq uipme n t, m od el number , se r ia l number, date of p urch a se,
date of fa il ure, number of opera ti ng h ours, numb er of mil es, a nd
t yp e of ser vice (c ontinual, dail y, yea r ly , se aso nal ). Not e the
loading cycle in vol ved. (Is it in onl y one direc tion , or is it in b oth
d irections? Is it stea dy under ma xi mum load , or is it intermit-
te n t? Is it u phill , d ownhill, ove r roa ds , or ove r rou gh terrain ? Is
it a gradual loadi ng, or is it a n abru p t im pact type?) Copy the
ma inten ance hi s tor y re lati ng to th e unit s in q uestio n. (Wha t is
the ty pe of lub r ican t a nd its co nditio n? An y recorded te m pera-
tu res? Had a ny pa rt s been pre vi ou sl y repla ced ? Had th e gear
assem b lies been di sas sembl ed ? If so, had they b een s ub m i tte d
to magnetic-p article in sp ection b ef ore being rea sse m b led? )
Thi s is a ll pa st hist or y , b ut it mi gh t co n ta in th e ke y to p resent
problem s .
T he next ste p is to re move the asse m b ly fro m the eq uip-
me nt. Note the co nditio n of th e reta ini ng bolt s-tight or loose.
Has the asse mb led un it bee n leak in g oil or is i t dry? Does the
assem bly re move eas ily fro m the equ ip me n t or is the re obvio us
d iff ic ulty? H a s th e lub ricant been dra ine d fro m the asse mb ly?
What is th e co nd i tio n of the d ra ine d oil? A re th ere a ny vis ua l
co n ta mina n ts in th e oil? If so, a re th e co n ta m ina n ts ma gnetic or
not? Not e th e asse m b ly numb er a nd a ll other stam pi ngs or id en -
tifica tio n ma rk s o n th e asse m b ly . If helpful, tak e phot ogra ph s.
The cover of the asse mb ly may now be ope ne d . Note the
co ndit io n of the b ol ts . As the cover is b eing re moved, no te the
condit io n of the seal , espec ia lly im p o rtan t if there h a s bee n oil
see page. Wh at is the a p pea ra nce of th e re main ing oil on the
parts? Does it ap pear that lu b ri ca ti on was sufficie nt throug h-
out th e asse m b ly, or are th ere dr y a reas? Is th ere ev ide nce of
co rrosio n [ru s t ) a nyw he re? A lways keep in min d th e s pacia l
relationship of a ll pa rts w it hi n th e asse mb ly . T he top of th e as -
se m b ly ma y be dr y , or it ma y be r us ty du e to co nde nsatio n. Th e
bot to m of th e unit w i ll ac t as a s um p a nd w ill co llec t a ll typ es of
metallic frag ments, as we ll as water. Gears a nd bearings low in
the asse mb ly w ill co nt in ua lly g ri nd up met alli c pa rts a nd, if
there is mo isture p resen t , m ig ht corrode easily. Could photo-
gra phs hel p at th is p oint ?
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
46
Each part is now to be removed from the assembly. The
operation must be methodical. Starting at the power input, take
out each part in order, record all identification marks, and note
the condition of the part. Keep parts in some sort of order. Many
times, a person will spot the failed part, take it out of the assem-
bly first, and walk away with it to get the answer, assuming
that its failure was both cause and effect. This is often an inac-
curate assumption. Rather than removing it, take a photograph
showing the relationship of all parts involved.
Now that all parts of the assembly have been reviewed, de-
termine which fracture is primary and which are secondary. A
primary fracture is one that occurs initially; other fractures oc-
curring subsequent to the initial fracture are secondary frac-
tures. Determine if the primary fracture is random [only one) or
one of many. Ifthere are many, is there a pattern to the failures?
For instance, are the gear teeth failing in no particular sequence,
or is every sixth tooth failing? If there is a pattern, determine
what has influenced that type of pattern. Count the number of
teeth in both gear and pinion. If the pinion has an odd number of
teeth [hunting tooth), perhaps every gear tooth will be affected.
Take a good look at the tooth contact patterns. Are they well
centered or are they at one end? [As a reference, study Fig.
1-15.) Is there evidence of runout or taper in any of the parts?
Runout may have been in the original cutting of the gear teeth,
but it may be caused by a bent shaft. If the failure is surface
deterioration, are the affected areas consistent, variable, or
random? Try to find out why.
When the primary failure has been determined, a decision
should be made regarding the mode of the fail ure: then it should
be decided if the failed part or parts are to be sent to the gear
failure analyst. If the primary failure is in one of two mating
gears, by all means send both parts : If there are several gears in
an assembly and there is a question of cause and effect, send all
parts. In any situation, send all documented information
available.
The field service man should now make some very impor-
tant observations . He should, if at all possible, determine the
cause of the primary failure. If the cause is in the assembled unit
itself and is not externally controlled, the failure may be beyond
 SYSTEMATIC EXAMINATION 47

the scope of his control. It could be a wise move at this time to

send the entire subassembly back to the manufacturer, who
should be in a better position to make a detailed analysis. On
the other hand , if he determines that the cause is external to the
assembly and that a remedy or adjustment can be made in the
field to keep other machines in operation, he should make those
recommendations to his management.
Cooperation between field service personnel and the gear
failure anal yst is extremely important and cannot be overem-
phasized. The analyst sees usually one or two failed gears and
must base his findings on the story they tell. Field service per-
sonnel usually have access to all other parts and to the back-
ground knowledge. The analyst may state that a set of gears is
grossly overloaded. The field service person may then , for ex-
ample, take the input shaft and find torsional fatigue cracks by
the use of magnetic-parti ile inspection. Verification of facts by
the sharing of know ledge should al ways be common practice.
So much is left unsaid about field examination, such an im-
portant aspect of gear failure analysis . Perhaps it will suffice to
say that many bits of knowledge about a failure will point di-
rectly back to the field application and can in the future be used
by those responsible for field examination.
When the failed parts and documented information are in
the hands of the failure analyst, he too must be systematic in the
study that is now before him. He must also "begin at the begin-
ning" from his perspecti ve.

Visual Examination
By far the most important phase of visual examination is that
which is done by the unaided eye. In most instances, all of the
field examination has been accomplished by this method; and
now, the most important decisions of procedure must be made
by this same method.
First, document the part that is being examined. Write
down the part name, part number, and all other markings or
hieroglyphics found stamped, etched, or embossed on the part.
Next, describe in detail what you actually see; i.e ., the physical
 SYSTEMATIC ANALYSIS O F GEAR FAILURES
48
co nd itio n of th e s urface . such as rubbing. sc ra tc hi ng, a nd ma rk s
of a b us i ve ness [h ammer mar k s, c u tti ng- torc h mark s , threa d
stri pp ing, etc .). T he n d esc rib e th e toot h contac t patterns fro m
b oth directi on s a nd th e a mo un t of normal o r a b no rma l wea r on
these patt ern s. Fin all y, des cribe in d e tail th e a ppearance of the
failure . a nd catego rize it accord ing to mo de [see Ch ap te r 4 ),
sp ecific all y looking for th e ori gin of th e fr a cture.
Loo k a t th e fail ure. S t udy it clo s el y . Wh at d oes it re veal ?
T he fo llowing exa m p le s hows h ow a n a na lys t d et e rm in es th e
ca use of a fa ilure.
On e s pira l be vel d ri ve gea r was returned from a coa l min-
in g operatio n . Eve ry too th was crus hed a t and over the toe end ,
midprof'ile , co n vex [l oaded) s id e [Fi g. 3 -1). Th e ove ra ll tooth
co n tac t area was we ll ce n tere d and ex te nde d a lmos t full length.
Ho w e ve r, there had bee n a rece n t s hift. s ince the load pa tt e rn
wen t direct ly to th e toe end a t th e a rea of failure. T he custo mer
fi eld se rv ice w a s as ke d to che ck for pini on a lig nme n t a nd to
s ub m i t th e p ini on for exa m ina tio n . It rep ort ed tha t th ere was
" no mi s alignm ent a nd th e p in io n is okay ." Ho w e ver . three m ore
gears w ere returned w i th th e toe end in e xac tly th e s a me
crus he d co ndition . Aga in . the re port sta ted th a t th e " p in io ns are
oka y ." By th is time a n a na l ys is of th e firs t gea r h ad be e n ma de
an d a ll tooth ch aracteris tics we re as s pec ifie d . When th ree ad-

Fig. 3-1. Spiral bevel gear, 2.5 D.P. SAE4820H, case depth
0.068 in ., 58 HRC. Operation: coal mining. Every tooth
crushed and subsequently broken at toe end, mid profile ,
convex (loaded) side.
 SYSTEMATIC EXAMINATION
49
ditional gears arri ved with the same failure and with no mating
pinions, the decision was made to visit the mine. The mainte-
nance department was waiting with nearly a dozen gears on the
floor, in the same damaged condition, but with no pinions. The
first question asked was "Where are the mating pinions?" The
reply: "They're all okay, but you'll find them over there inthe
corner." The pinions were then inspected visually. Every tooth
of every pinion had a visible crack at the root radius, heel end,
concave [drive) side that extended not only down the back face
angle, but along the tooth root radius toward the toe end, to
about three-quarters of the tooth length. This pinion condition
caused full load to be applied from the toe end of each pinion
tooth to the toe end of the gear teeth, which were crushed by the
extremely high load per unit area. The pinion teeth were the
primary failure; the gear teeth were the secondary failure . The
entire story was told by the parts themselves , but someone had
to listen.
This example points out four basic lessons of observation
and logic that a failure analyst must know:
• Only the gear was submitted to the analyst for consider-
ation. The analyst observed that every tooth was crushed
at the toe end of the driven side. and that the normal tooth
contact had been properly centered for the greater por-
tion of its service, The next step was to ask two ques-
tions: What exerted the extremely high pressure contact
at the toe end of every tooth? And why did the contact
load shift to the toe end? Logically. the only answer to the
first question is that the toe end of the pinion tooth is the
only object available to impress the toe end of each
gear tooth. The answer to the second question must be
found by examining the pinion.
• The pinion was examined . Every tooth showed evidence
of bending fatigue at the root radius of the drive side.
from the heel end to three-quarters of the length toward
the toe end . Each tooth should show a normal central
contact pattern against the gear tooth until the bending
fatigue crack originated. The progression of this fatigue
crack automatically started the sequence of events as fol-
lows: the tooth deflection increases each time it is loaded;
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
50
the deflection allows a premature loading of the next ad-
jacent tooth, which becomes suddenly overloaded until it
originates a bending fatigue crack; and the deflection ac-
tually relieves the load at the point of contact and shifts
the heavier contact load toward the toe end. Thus, with
every pinion tooth failed at three-quarters of 'the heel
portion, the entire load was concentrated at one-quarter
of the toe end portion. This answers the questions about
the gear's failure .
• The preceding discussion of observation and logic could
have been handled by knowledgeable field service per-
sonnel, if only they had recognized that the fault was
within the function of the pinion and had submitted both
parts to the analyst. Time, always an important element ,
could have been saved not only in the final analysis. but
also in a number of down-time hours used for replacing
parts . This shows the very close association that should
exist between field examination and failure examina-
tion. In this instance. the field service personnel did not
read the failure correctly, nor did they submit enough ev-
idence (the mating pinion] to the analyst. The analyst
read the failure that was before him. recognized the prob-
lem, but did not have the part that provided the cause of
the failure. Only when the whole story was together was
something constructive accomplished .
• This example was used primarily to stress the impor-
tance of visual examination; but two conclusions stated
were not definitive. The conditions of "crushed" and
"tooth bending fatigue," although generally recognized
by an experienced analyst. cannot be conclusively de-
termined without microscopic examination. Also. with-
out a doubt. even the experienced analyst would have
required that the parts undergo microexamination be-
fore recommending corrective action.

Before leaving the above example there are other questions

and comments that are pertinent to the present discussion. It
was obvious to the analyst that the gear was a secondary fail-
ure; but the analyst could not be absolutely certain that the pin-
 SYSTEMATIC EXAMINATION
51
io ns h a d fa iled by toot h be nd ing fa tigue a t th e heel porti on be-
for e he had ever s een t he pinion . He was s eeki ng a reason for t he
shift in load pa tt e rn that caus ed the extremel y heavy toe co n-
ta c t. H e found t he s pec ific rea son . Oth er reaso ns coul d h a ve
been th e breakdown of the p ilot bearings. unstab le assembl y
a llow ing for d eflec ti on s unde r loa d. improp e r ad jus tme n t or
rea dju stm ent in th e a lig nme n t. or imp rop er tooth c u tti ng or dis -
tor ti on during manufa ct u r in g .
Alt hou gh t he primary failure was establis hed within the
pini on. it co u ld h a ve bee n pos sibl e th a t th e crus hing of th e gear
teeth was th e pri m a r y fa ilu re. If th e microexa m ina tio n of th e
gea r tee th ha d estab lis hed a metallurgical condit ion that cou ld
ne ver ha ve w it hs too d a n o rm al loa d a p p lica Lio n; if th e gear
tee t h h ad bee n cu t or di s tort ed eno ug h to h a ve a llowed hea v y
toe co ntact ; or if severe de flec tio ns or mi s align men t ha d oc -
curre d ; the crus h ing of th e gea r teeth co uld ha ve bee n the pri -
mar y failure. Keep i n min d tha t th ere had been n orm al tooth
con ta ct for a n es tablis hed leng th of ti me. a nd that the gear teet h
had been a na lyzed a n d were found to meet th e nece ssary met al -
lurgi cal requirements fo r norm al loading.
It is ge ne ra lly und erst ood tha t occ urre nces of too th be nd -
ing fa tigue are ca use d by ov er load . Is thi s a lways the case . a nd
was it th e case here? T ooth bending faLi gu e is a ten sil e o verl oad .
but s pecifica ll y . of the ac tual d esign a nd meta ll urg ica l co nd i-
ti ons ex is ti ng. Th eor e ticall y . th e pi nio n in questio n had n o t
bee n overloa de d. Actu all y . it was sev erel y overl oad ed du e to
s harp unde rcu tt ing a t th e roo t ra d ii. s ha llow case dept h, and
in s uffic ie n t su rf a ce h ardn es s .
As suming th at a ll manufa cturing a nd process cha rac te ris-
ti cs fo r th e pin ion a nd th e gea r had been w it hin acce p ta b le lim-
its . th e sa me m od e of fail ure co uld ha ve ha p pened . bu t p roba bl y
n ot as ex te ns i ve ly a s to includ e eve ry s et in o pera tio n. The
ca use wo uld neces s aril y h a ve been a n occasio na l hi gh p eak
ove rloa d th at wo uld ha ve resulted in a ra ndom too t h fa ilure
ra ther th an a patt erned fa ilure .
Vi sual exa m inat io n d oes n ot p recl ud e th e us e of lo w -pow er
microsco pes u p to 20X th a t ca n a nd s hould b e u s ed t o s t udy
fr a cture s urfaces. But th e exam ine r s ho uld be care ful n ot to
ove r loo k th e lower m a gnifi cati on. Ver y oft en , th e b est resolu-
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
52
tion is lX (eye contact) for an overall or larger view, then 3X,
lOX, and 20X, in that order. Findings should always be reported
under the magnification that clearly shows what has happened.
Many times the story will best be shown with photography by a
combination of two different magnifications: i.e., l X or lower
for position of a failed area , and 20X for details.
Since a visual pattern is the result of reflected light, much
can be accomplished or hindered by the light source. When ex-
amining a fracture, it is best to use a rather small spotlight
source that can be easily manipulated to give the light beam a
directional pattern. Fatigue striations are more easily seen
when the light beam is almost parallel with the fracture surface .
This flat-angle beam will highlight the peaks and darken the
valleys, allowing an excellent resolution for a photograph.
Changing direction in lighting is often more important to pho-
tography than is intensity or magnification.
There are many pieces of equipment that can be utilized for
visual examinations. The equipment alluded to in the above
two paragraphs is described in detail by R. C. Anderson in
Chapter 2 of his book Inspection of M etals, Volume I: Visual
Exurninnt ion.! This book is highly recommended as a supple-
ment to any visual study of materials .
In the first paragraph of this section, it is stated that impor-
tant decisions of procedure are to be made. The procedures that
must now be determined come after the visual examination and
the documentation of that examination. The decisions are to be
made concerning which physical tests and which metallurgical
tests are to be conducted. Since these procedures are to be ac-
complished after visual examination, there are some very im-
portant points to be remembered:
Do not jam two broken parts together to see how they fit.
Delicate fracture surfaces may be altered.
Do not indiscriminately cut through a fracture surface for
microscopic examination before completing visual exam-
ination. There must be a valid reason for sectioning . Fi-
nally, cut only in the area that is most likely to give a
conclusive answer. It may be wise to take several photo-
graphs of the area before cutting .
 SYSTEMATIC EXAMINATION
53
Do not grind fractured areas flat to take hardness readings
unt il after visual exam ina tio n -and t hen only if neces s a r y.
Do n o t clean fracture s urfaces wit h the pa lm of yo ur h an d
or wit h greasy ra gs, or remove a n y rust with a n acid .
Cl eaning is b es t accom plished with a spra y so lve nt and
ge n tle a ir p res sure.

Physical Examination
Ph ys ica l exam ina tion in vol ves a ny proced ure of nond es t ru c-
tive testing , in cl ud in g m ag ne tic-pa r ti cle inspectio n , toot h char-
ac teristic st udies , s urface ha rd nes s tes ting , ultra s on ic testi ng,
nital e tc h ing, p ro filo me te r mea surement s , a nd di men si onal
c hec ki ng.

Magnetic-particle inspection is perh ap s th e mos t useful

an d importa n t p roced ure to use ; as suc h , it is th e fir s t s te p in our
sys tematic p h ys ica l examina tio n . The re are se veral typ es of
testing equip me nt tha t may be us ed in specific places or in-
s ta nces, bu t the we t m eth od, us in g flu o res cent ma gne tic parti -
cl e s , is th e m os t common a nd the m ost revea li ng . In fa ct, ph ot o-
gra phs are a ccom p li s hed more easil y b y th is method (Fi g . 3 -2) .
T he firs t p u rp ose of magnetic-partic le inspection is to d eter-
min e th e ex te nt of da mage . Perh ap s onl y a porti on of one gea r

Fig. 3-2. Unidirectional torsion overload in a l-in.-diam

jack shaft, as revealed by magnetic-particle inspection b y
"Magna -Glo" method. Photographed us in g a standard
"black" light and Polaroid Polacolor-2 T yp e 58 , 4X5 Land
film . ASA75 ,f/ll , for one minute.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
54
tooth appears fractured visually when, in reality, several teeth
have cracked in the same area. More experience with the analy-
sis of fractures brings recognition that the magnetic indications
can clearly show the progression, or sequence of events, that
leads to final failure. In fact, the indications may also reveal the
point of origin on other teeth. If this is so , one should section for
microexamination the area showing the beginnings of the fail-
ure mode that has not yet been destroyed or mutilated by addi-
tional damage. Other surface discontinuities discovered by
magnetic-particle inspection may be surface or slightly subsur-
face inclusions , bar stock seams, forging laps , machine tool
tears, grinding checks (Fig. 3 -5), and quenching cracks.

Tooth characteristic studies give significant analysis con-

cerning what has happened to the gear teeth. These studies are
detailed in Chapter 1 and should be reviewed at this time to
determine which are appropriate.
If there is any question about the tooth contact pattern on a
spiral bevel gear/pinion set, the set should be placed in a spiral
gear tester to determine actual field loading position. The tester
should be adjusted first at the specified position to see if the
parts had run as expected. Then the tester should be readjusted
to the existing pattern. This will show the conditions of deflec-
tions or misalignment prevailing at the time of field service.
(See "Mating-Tooth Contact Pattern" in Chapter 1.]

Surface hardness testing is always necessary but must be

used with caution. Techniques of testing should be in accor-
dance with the practices outlined by the manufacturer of the
test. In failure analysis, surface hardness readings should be
taken primarily in areas pertinent to the analysis. Most likely, a
microhardness survey (discussed below under "Metallurgical
Examination"] will be advisable. Since a large percentage of
gears have been carburized (or surface hardened], the case
depth is significant in the interpretation of surface hardness
results . For instance, a Rockwell "C" indenter and load might
penetrate a case ofless than O.025-in. depth and gi ve an errone-
ous reading. For lighter case depths, it would be wiser to use a
hardness scale with lighter penetration loads. Table 3-1 is a
 SYSTEMATIC EXAMINATION
55
Table 3-1. Minimum work-metal hardness values for testing various
thicknesses of metals with regular and superficial Rockwell
hardness testers (a)2
Mini mum hardn es s for Minimum ha rdn ess for
s uperficia l ha rdn es s testing. reg ul a r ha rd nes s test ing.
Metal di am ond "N" Brale inde nter di amond Bra le ind ent er
thickness . 15N 30N 45N A 0 C
in. 115 kg ) (30 kg ) (45 kg ) 160 kg ) (100 kg ] [150 kg )

0.006 92
0 .00 8 90
0.010 88
0 .012 83 82 77
0 .014 76 80 74
0.016 68 74 72 86
0. 0 18 (b ] 66 68 84
0.0 20 (b ] 57 63 82 77
0 .0 22 (b] 47 58 78 75 69
0 .0 24 (b] [b] 51 76 72 67
0 .026 (b ) (b) 37 71 68 65
0 .028 [b ] [b ] 20 67 63 62
0 .030 [ b] (bl (b) 60 58 57
0 .032 [b] (bl [b ] [b] 51 52
0 .034 [b] [b] [b] [b ] 43 45
0 .0 36 (b] (b) [b ] (b ) [b] 37
0 .038 (b ] [b ] [b] [b] (b) 28
0 .04 0 [b ] [b] [b] [b ] [b) 20
(a j T hese values are a pp rox ima te only a nd are int end ed primaril y as a guide . (b I No
min imum hardn es s for met al of eq ua l or gr ea ter thi ckn ess.

"H ardness vs Minimum Thickness" cha r t fo r chec king hard-

ness of thin materi al .! It is a ls o a fairl y reliable p ractice to sub-
s tit ute "c ase depth " for "me ta l thickn ess" w hen che cki ng th e
surface hardness of a ca rburized part.
Hardnes s is th e only crit eri on normall y us ed to int erpre t
th e stren gth of the part being exa mine d and ca n be us ed inter-
cha nge a bly w it h confide nce . Table 3- 2 lis ts th e a p prox ima te
equivalent hardne ss numbers for st eel as tabulated b y th e So -
ciet y of Automotive Engineers Hnndbook .!

Ultrasonic testing is not uni ver sa ll y req ui red , bu t it is used

to a n ad vantage for re vealing subsu rfa ce d is continuities . Ul-
trasonic testing can be utilized to locate subsurface casting de -
fects such as porosit y, shrinkage , or blow hol es; subsurface
voids or cracks at or near welded zones; large inclusions or
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
56
Table 3-2 . Approx imate eq uivalent hardness numbers for steeP
Rockwell
s uge rficia l
Brinell har ness No..
hard nes s Rock well Ha rdn ess No. s uperfic ia l
Rockwell No.. l0-mm A-scal e. Bvsca le . Brale Tensile
C-scal e s ta nda rd 60-k g load. 100-kg load. pen etral or : s tre ng th
hard ness ball. Bral e lI l 6-in. l 5N scal e. [a p p ro x ].
No . 3000- kg load penet rator di am ball l5 -kg load 1000 psi

60 81.2 9 0 .2
59 80 .7 89 .8
58 80. 1 89.3
57 79 .6 88 .9
56 79.0 88 .3
55 78 .5 87.9 301
54 78.0 87.4 292
53 77.4 86 .9 283
52 50 0 76.8 86 .4 273
51 487 76 .3 85 .9 264
50 4 75 75 .9 85 .5 255
49 464 75 .2 85.0 246
48 451 74 .7 84 .5 237
47 442 74 .1 83 .9 229
46 432 73 .6 83 .5 222
45 42 1 73 .1 83 .0 2 15
44 40 9 72.5 8 2 .5 208
43 40 0 72.0 82 .0 20 1
42 390 71.5 8 1.5 194
41 381 70. 9 80 .9 188
40 37 1 70.4 80.4 181
39 362 69 .9 79. 9 1 76
38 3 53 69.4 79.4 171
37 34 4 68 .9 78 .8 168
36 336 6 8 .4 78 .3 162
35 327 67.9 77.7 157
34 319 67.4 77 .2 153
33 311 66 .8 76 .6 149
32 30 1 66 .3 76 .1 145
31 294 65 .8 75 .6 142
30 286 6 5 .3 75 .0 138
29 279 64 .7 74 .5 135
28 27 1 64.3 73 .9 132
27 264 63 .8 73.3 128
26 258 63 .3 72 .8 125
25 253 62.8 72.2 122
24 247 6 2.4 71 .6 120
23 24 3 62 .0 100 .0 71.0 117
22 237 61.5 99 .0 70 .5 114
21 231 61.0 98 .5 69 .9 11 2
 SYSTEMATIC EXAMINATION
57
Ta ble 3-2 (continued)
Rockwell
s uperficial
Brin ell hardness No ..
ha rdness Rockwell Hardness No. superficial
Rockwell No.. l0-mm A-scal e. B-scal e. Brale Tensile
C-scale s ta nda rd 60- kg load. 100-kg load. penetrator: strength
hardness ball. Bral e 1I16-in. 15N sca le. [approx) .
No. 3000-kg load penetral or d iam ball l 5- kg load 1000 psi

20 226 60.5 97 .8 69.4 110

219 96 .7 10 6
2 12 95 .5 10 2
203 93.9 98
194 92.3 94
187 90.7 90
179 89 .5 87
171 87 .1 84
165 85 .5 80
158 83 .5 77
152 81.7 75

"pipe" in steel parts [Fig . 3-3a); internal rupture at the ca se and

core transition zone in gear teeth [Fig. 3-3b); and extreme varia-
tions in material composition.

Nita} et chi ng is used to locate tempered areas on hardened

steel surfaces . This method should not be confus ed with the ni-
tal etch used for microstructure examination; the solution,
procedure, and purpose are not the same. The most common us-
ag e for the subject etch is to loca te grinding burns-spots or
areas on a ground surface overheated by the grinding wheel.
Figure 3-4 shows distinct chatter burns extending the full
length of the profile of helical gear teeth; the tempered spots are
softe r than the remaining carburized surface and more suscep-
tible to surface breakdown. As Fig. 3-5 shows , b y nital-etching
a failed gea r , a darkened surface reveals burning of the entire
end face by severe grinding , and ma gnetic-particle inspection
rev eals that grind ing ch ecks of the same pattern as the fra cture
origin also occurred.
Nital etching is also used extensively to show an indu ction
hard ened pattern. The induction heated zone will show a ve r y
clear pattern on the end fa ce of gear te eth . Figure 3-6 shows a n
end fac e pattern of a consecutive -tooth-space hardening tech-
 SYSTEMATICANALYSIS OF GEAR FAILURES
58

.-

Fig. 3-3(a). A massive flat inclusion parallel to the top

face of a spur gear tooth at a distance of 0.228 in . from the
surface. This inclusion gave a sharp sonic indication for a
distance of 1/2 in. along the tooth length. 3% nital etch,
200X.

Fig. 3-3(b). An internal rupture in a gear tooth at the easel

core transition zone which does not reach the surface.
This condition can be discovered b y ultrasonic tes ting.
 SYSTEMATIC EXAMINATION
59

Fig . 3-4. Helical gear carburized and hardened to 60-63

HRC. A chattering tooth grinding wheel has "burned"
small consecutive spots the full length of the profile on
several teeth. Discovered by nital etching.

Fig. 3-5 . Spur gear, end face ground . Grinding severity

tempered the entire face (revealed by nital etching) and
checked the surface (revealed by magnetic-particle in-
spection) .
 SYSTEMATIC ANALYSIS O F GEAR FAILURES
60

Fig. 3-6. End face nita I-etched pattern of a consecutive-

tooth-space induction hardening technique. Pattern is
exaggerated and distorted due to the extra heat picked up
by the square edges. The true pattern should be estab-
lished below the surface.

Fig. 3-7. End face nita l-etched pattern of an induction

hardened internal gear, showing tw o tooth spaces that
had not been hardened.

ni qu e used on a la rge in ternal w hee l re d uct io n gear. The pat-

te rn s how n is exagge ra ted du e to th e e nd effec t of th e sq uare
edges pic king up a g rea te r a mo un t of flu x . S ince toot h end s ur-
fa ce pa t tern s of ind uc ti on h ardening proces ses ca n be so me-
w hat dis to r te d, i t is reco m me nde d th a t th ey be u sed only as a
g uide, a nd that th e t rue p att ern , w h ic h m a y be 114 in . from th e
end face, be exa mine d. Also , as a n ins pec tio n tool, nit al e tc hi ng
 SYSTEMATIC EXAMINATION
61
Table 3-3. Nital etch production test for ground tooth gears

Solution 1 - 5% con centrated nitric acid in denatured a lcohol.

Solution 2 - 10% conce ntra ted hydrochloric a cid in denatured
alcohol.
Solution 3 - 15% concentrated ammoni a in denatured al coh ol.
(All solutions are to be us ed at room temperature.)

Etching procedure
1. Clean gear of a ll grease and oil (trichlorethylen e).
2. Place part in Solution 1 for 20 se conds .
3. Remove and rinse in clean cold water. dry with air pre ssure.
4. Place part in Solution 2 for 20 seconds.
5. Remove and rinse in clean cold water, dry with air pres sure.
6. Immerse part in Solution 3.
7. Remove and rinse in clean cold water and then dry with a ir pres sure.
8. Imm ers e the part in rust-preventive oil.
9. Tempered areas du e to grinding burn s will be dark again st th e
ligh t-gra y background.

will quickly point out discrepancies such as the two nonhard-

ened tooth spac es shown in Fig . 3-7 .
There are several very similar nital etching processes being
used by industry as a production check for all ground tooth
gears. One procedure is outlined in Table 3-3. However, if a
qui ck check of only a few parts is to be made either in the field or
in the lab, induction hardened patterns can be etched b y swab -
bing the areas in question with a 5% nital solution [5 ml concen-
trated nitric acid with 95 ml denatured alcohol] and b y rinsing
with warm wa ter.

Profilometer measurements show the varia tions in average

roughness height that occur on the surface being tested. This
ma y be important information where surface deterioration is
e vide n t, or at radii a nd other points of high stress concentra-
tion.

Dimensional checking has not been mentioned in Chapt er 1

as a ch aract eristic , but its importance is self-evident. Wrong
dimensions may often pla ce a high stress raiser in an area of
high stress conc en tra tio n. If this cond iti on had been in the de-
sign, proof of th e results should demand a change in that design .
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
62
Metallurgical Examination
In this section will be discussed the tests that necessitate the
destruction of some portion of the part for further study. The
decision to cut into a part must be done with forethought and
knowledge ; and the examiner should ask whether it is neces-
sary to section the part to gain additional data; what the data
are that must be known; and where the cut must be made in
order to make the da ta meaningful.
It is often the case tha t cu tting merely for the sake of cu tting
turns out to be an exercise in futility. However. when it has
been established that sectioning a part is necessary. one or more
of the following procedures will be applicable:
Cross-sectional hardness survey
Macroscopic examination
Carbon gradient traverse
Chemical analysis
Case hardness tra verse [microhardness 1
Microscopic examination
Scanning electron miscroscopy
The cutting or sectioning of a part must be done very care-
fully to prevent the heating of any structure. Lower heating may
temper a surface. and higher heating may radically alter the mi -
crostructure. Parts of lower hardness [less than 38 HRCl should
be cut with a mechanical hacksaw, using a continuous stream of
cutting/cooling fluid . Harder parts, including carburized parts,
should be cut with a relatively soft, powered, abrasive wheel
that will break down readily to ensure a nonglazed surface. A
glazed surface will tend to "burn" the sample. At the same time,
a heavy spray of coolant should play on the part at the cutting
edge . In fact, two sprays are recommended, one at the cutter
inlet and one at the cu tt er outlet.
Electrostatic methods are also very accurate means of cut-
ting sections . But these methods metallurgically alter the cut
surface for a depth of a few thousandths of an inch. and the
surface then must be ground sufficiently to eliminate this
condition.
 SYSTEMATIC EXAMINATION
63
Cross-sectional hardness survey is used for two purp os es:
to determine th e a ctual strength of a fail ed part , and to d eter-
mine if the he at tr eatment w a s optimum for th e material us ed.
This procedure, which is ge nera lly us ed on pinion shanks
or shafts that have fail ed , is done as follow s: cu t laterally across
th e part, making a cross secti on a bout 3/4 to 1 in . thick, as clo se
to th e fra cture a re a a s possible without including an y alter ed
stru cture. Pl ac e th e sample in a s ur fa ce g rinde r a nd grind [cool)
both sides so th ey a re parallel. Take surfac e hardness readings
at four positions a nd re cord . T ak e reading s across the sampled
se ction at 1I16- in . int erv als, s ta rting 1116 in. from the surface at
th e establish ed four positions a nd continuing to th e cen te r [Fig.
3 -8 J. Record eac h reading. Average the readings taken at eac h

Fig. 3-8. Cross-sectional hardness survey through the

splined section of a spiral bevel pinion shank. First read-
ings at the surface (O.D. of the sample). Continuous read-
ings at 1116-in. intervals starting 1/16 in . from the sur-
face .
 SYSTEMATIC A NALYSIS OF GEAR FAILURES
64

60
i\ 1\
\ \ SAE 8620H HI. 517L4 226

C= 0.20% Jl = 45 HRC .
Mn = 0.78 2 = 44
50 Ni = 0 .50 - 3 =38 -
Cr = 0.42 4 = 32
Mo = 0 .17 5 = 28
6 = 26
u 7 = 24
l
-'"
40 8
10
= 24
= 22
-
o
o

\
a:
<Ii
Vl
Q)
c
"E 30
~
I
I '"

20
Q;
a;
~
Q;
a;
l- - I
I

I
E E ~
.s '"
"0
I
"0
(; (;
10 c
'"E ·E I

Q)
c
Q)
c
c.
I
oS
C.
Ul Ul
I I I 1 II
1 2 3 4 5 6 7 8 10 12 14 16
S 1 2 3 4 5 6 7 9 11 13

Dist ance from su rf ace. 16 t hs in.

Fig. 3-9 . Cross-sectional hardness s urvey chart of splined

section shown in Fig. 3-8.

p ositi on and inco r po rate th em onto a g ra ph s im ilar to the g rap h

in Fig. 3-9 .
Th e informati on on th e g ra p h ca n now be used to interpret
th e s tre ng th of th e part b y s u b s ti tu ti ng s tre ng th for hardnes s .
a nd to interpret th e re s ults of th e heat treating p roces s b y u ti -
lizing quenc h rates fo r the hard ena b il it y data. The part a s
represe n te d b y thi s procedure had fa ile d b y to rs iona l sh ea r.
so obvio us ly, the torsional stress appli ed w a s g rea te r tha n the
stre ng th of th e par t. If th e a p p lied s tress had bee n a b no rma ll y
 SYSTEMATIC EXAMINATION
65
high and could have been controlled, nothing more needed to be
done. If the stress is to be applied regularly, the material and/or
heat treating process must be changed to compensate.

Macroscopic examination is the visual inspection of any

sectioned area by use of low magnification, aided by an etching
process that intensifies differences within a material.
The "macroetch" procedure is accomplished by placing the
desired surface in a solution of 50 % hydrochloric acid and 50%
water, boiling for 20 minutes, rinsing well in hot water, and
drying immediately. Samples prepared specifically to study
material flow lines are shown in Fig. 3-10 and 3-11. The mac-
roetching of two sections 180 0 apart [Fig. 3-12) illustrates
the differences in material flow that can occur, even in one forg-
ing . Nonmetallic inclusions are vividly exhibited in Fig. 3-13
and 3-14. Forging defects [laps) that sometimes show up very
lightly on the surface will open up vividly on surface and sub-
surface when sectioned and macroetched [Fig. 3-15) . Sound-
ness of material will range from clean, to slight porosity and
pinholes, to a center pipe condition as exhibited in Fig . 3-16.
A 5% nital etching solution [5 ml nitric acid to 95 ml dena-
tured alcohol) is also used effectively to show induction hard-
ened areas [Fig . 3-17) and differences in basic material [Fig.
3-18).

A carbon gradient traverse on a failed part is taken only

when the carbon content in the carburized case is questioned.
There are also severe restrictions in this procedure . There must
be a smooth, round surface of sufficient length to allow for the
turning of the outside diameter, in order to obtain enough steel
chips from each cut to get an analysis. When the proper section
has been taken from the part, it must then be tempered to a
machinable hardness in an atmosphere that will not deplete
carbon from the surface. A submersion in molten lead at 1200 of
is recommended. Cool and clean the surface, and prepare the
ends for accura te centering in the la the. An example of a carbon
gradient traverse is shown in Fig. 3-19 . To obtain the points on
the graph [as shown in the example), take the first cut across
the surface at a depth of 0.003 in. that measures 0.006 in. from
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
66

Fig. 3-10. Internal gear blank, 1/4X . Macroetched to show

material flow lines after forging and machining.

Fig. 3-11. Longitudinal section through the center of a

spur gear tooth, 1/2 X. Macroetched to show material flow
lines after forging and machin ing.
 SYSTEMATIC EXAMINATION
67

Fig. 3-12. Spiral bevel gear, 2/3 X. Macroetching of two

surfaces 180 0 apart from this gear reveals a major differ-
ence in material flow lines during the forging operation.
The dark voids had contained several inclusions.

, "

. . .. .
"

Fig . 3-13. Macroetched section near a bearing diameter

showing nonmetallic inclusions following the flow lines
near the surface, lO X,
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
68

Fig. 3-14. Spiral bevel gear tooth, 3/4X. Macroetched to

show a large inclusion, which is aluminum silicate from
the wall of the steel-producing furnace . This inclusion
was hard enough to break the gear tooth cutter edge,
which then scored the rest of the tooth profile.

Fig. 3-15. Steel-forged gear carriers, 3/4X. Macroetched

on the machined surfaces to show the extent of forging
laps.
 SYSTEMATIC EXAMINATION
69

(a)

(b)

(e)

Fig. 3-16. Macrostructure, 100X. Macroetched to show

center soundness of a steel part: (a) fairly sound, small
pinholes; (b) slight porosity; (c) open pipe.
 SYSTEMATICANALYSIS OF GEAR FAILURES
70

(a) (b)

(e) (d)

[a ] 1I 2X . Longit udi na l sectio n th r o ugh tooth roo t. Ha rd en ed by tra versing

o ne tooth s pace a t a ti me. (b J 1I 2X . Cross sec tio n of tooth s pace near ce n ter of
sa me loo th a s in [a ], [c] 2X. Longi tudin a l pa t te rn in cludes en tire toot h and well
be low th e roots . Ent ire gear har dened a t on ce by a ci rc ula r inductor. [d ) 1I2X.
S p rock et tooth space hard en ed by a sta tio na ry tooth space in du cto r.
Fig. 3-17. Nital (5%J etching to show induction hardened
patterns.

Fig. 3-18. Spindle shank, 1X. 5% nita I etching reveals a

welded structure.
 SYSTEMATIC EXAMINATION
71
1.00

0.90
I rr-,
0 .80
~~
0 .70

'"~'"
cO
..8 0 .60
«;
u
0 .50

0 .40
'", 1/ ,0 .059 in .

0 .30

0 .20
'" i'o..
"'

0 .010 0.020 0 .030 0 .040 0 .050 0 .060 0.070 0 .080 0 .090

Depth below surfa ce. in .

Fig. 3-19. Carbon gradient traverse taken from a shank

section of a hypoid pinion. Note that the 0.40% carbon
level at 0.059 in. is very close to the effective case depth.

the diameter. Take each succeeding cut at a depth of 0.006 in. or

a measurement of 0 .012 in. from the diameter. Analyze the turn-
ings from each cut for percentage of carbon, and record the re-
sults on the graph . Document the necessary information. Note
that the depth at 0.40% carbon is identified . This carbon level in
a carburized case usually indicates closely the effe ctive ca se
depth, which is defined below under "Case Hardness Traverse."

Chemical analysis of the failed gear is very often nec essary

to establish or confirm the grade of steel. In most instances , the
part must be sectioned in an area where a drilling for sample
chips can be taken safely away from the high surface carbon.
Since carbon is very important to the interpretation of the ph ys-
ical properties of the gear, it must, above all the alloying ele-
ments, be accurately measured.
 SYSTEMATICANALYSIS OF GEAR FAILURES
72
Case hardness traverse b y th e use of micr oh ardnes s test ing
is a neces sary procedure in examinin g a n y failed ca rbur ized
part. Be very careful in se lec ting th e are a fr om w hic h thi s sec -
tio n is ta ke n. The sample for th is procedure w ill, in a ll prob abil-
ity, be used a lso for microscopi c examinatio n . In fact, th e po l-
is hi ng technique is th e sa me for both procedures. Select th e a rea
th at w ill provide the most meaningful in forma ti on . It ma y be
th roug h th e frac t ure d area, or nea r the fractured a rea, or
through a no ther tooth in the same rela ti ve positio n.
If a carb urize d gear toot h is to be sec tio ne d , it mu st be cut at
a n a ng le th at is normal to the tooth face a t th a t poin t. Cut slowly
w it h m uch coo ling . T rim the sam ple to fit a mo un ti ng pres s, b ut
do not mar or d is tort th e a rea to be examined . Mount th e sam ple
according to th e moun tin g press ins tructi on s , us ing a ther mo-
setti ng or a th e rm opl a s tic co mpou nd as a hold ing ve hicle . Pre-
pare th e surface of th e sam ple b y gri nd ing a nd poli shing in
graduated s te ps un til th e fina l polish is give n by using a com-
po und of five micro ns or les s. Place the po lished samp le in th e
micro hard ness tes te r a nd ta ke a series of ha rdn es s rea di ngs

60
..olIl.....l..
...
~ ~ ........ /
/ 1/'",
__ 0 .045 in . at midp oint

u 50
~
a;
:;: <,
-'"
o
a:
0
40 /
0.032 in. at root radius
~ , <,

'"
en
C
OJ
30
~
co
I 20

10

0 .0 10 0 .020 0 .030 0 .040 0 .050 0.060 0 .070 0.080 0.090

Depth below surface. in.

Fig. 3-20. Case hardnes s traverse of an 8620H steel s pur

gea r tooth using a Tukon microhardness tester w ith an
indenter load of 500 g. Eff ecti ve case depth s are indicated
by arrows.
 SYSTEMATIC EXAMINATION
73

s:
~
'E
Q)

15 C.
Q)
I'
C. C. -c
:"2 Q)
Q)
"0
II
::1: .2: s:
U
<I: :5:
II
I
I

Take core hardn ess he re

Fig. 3-21. Defined pos it ion s on a gea r tooth for taking a

hardness traverse at midprofile and root radi us. Ac-
ce pted core hardness is ta ken at root center and mid-
height.

through the case struct ure. Record the results on a chart similar
to that in Fig. 3-20 and document the information . The accepted
positions for taking a ca se hardness traverse are at the outside
diameter of a shaft or ground bearing of a pinion; at the center of
a space [root) between two splines; or, for a gea r tooth , at both
the midprofile and the root radius [Fig. 3-21).
The root radius tra verse generally starts at that point nor-
mal to the surfa ce on a direct line that will inte rse ct th e radial
line through the ce n te r of the tooth at 45° . If a failure occurs at a
radius betw een two different diameters , b y all means take a
microhardness survey at 45 ° through that radius . In man y in-
I ,.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
74
sta nces . th e cas e at a radius ma y be underc ut by grin ding. and a
traverse on the outside diame ter wi ll not indicate the proper
answer.
Case dep th is generally under s tood to mea n "effective case
dept h ," Effec tive case dept h is the dis tance from the surface
where the case hardness traverse curve in te rsec ts the h a rd nes s
le vel of 50 HRC. Although the effect ive case dept h of the root
rad i us is les s tha n that of the midprof'ile, it is an accepted prac -
tice for it not to be less than 60% of the specified mini mum effec-
tive case dept h. Sufficient tes ts have bee n conducted to indica te
that 50% is acceptable .
Total case depth is the perpendic ular distance from the sur-
face to the po int where the hardness of the core has been
reac hed. or w he re the carbon content is the same as the core .
Th e case h a rd nes s traverse for an ind uction ha rdened case
is accomp lished by the same procedure as above. b ut the shape
of the curves is differe n t [Fig, 3 -22J . Note the consisten t
straight -line ha rdnes s un til the start of the transitio n zone. a nd
the quick drop of hardness through the transition zo ne. T his

60 - ..0.. _

-
a
- ,,~
~

i"'"--!r.
~
u 50

~, A~
Qi
s
""'o0 40 ,,
a:
l/l
en
Q)
30
0.06 7 in . at midp oi nt A -
0.069 in. at midpoi nt B - -Y ~~ .~
II '\
c ,~ b
~~
~
'"
I
20
II
0 .08 7 in. at root radius B -
0.089 in . at root radi us A - ~
10

0.010 0 .020 0 .030 0.040 0 .050 0 .060 0.0 70 0.080 0.090

Dept h below su rfa ce, in .
Fig. 3-22. Case hardnes s tr averse of an induction hard-
ened 4145H s tee l gea r too th us ing a T ukon microhardness
te ster with an indenter load of 500 g. Effective ca se
de pth is mea sured to 45 HR C as ind icated by arrows.
 SYSTEMATIC EXAMINATION
75
sectioned gear tooth was ver y similar to th e one shown in Fig.
3-17[b] . The differen ce in case depth betw een profil es A and B
is du e to the diffi cult y of accurately positioning the tooth space
inductor in the exact center of the space, es pecia ll y s ince there
is only a 0.007- to 0.012-in . a ir ga p between the tooth surface
and the indu ctor surface . Noti ce also that the effe ctive case
depth is measured to 45 HRC. SAE Recommended Practice
J423a states: " In instances involving flame and induction hard-
ene d cases, it is desirable to use a lower hardness criterion."4In
J423a suggest ed hardness lev els are tabulated as follows for
va rious nominal levels of carbon content:
Effective
case dep th
Car bon con tent. % ha rdn ess. HRC
0 .28-0.32 35
0 .3 3-0.4 2 40
0 .43-0 .52 45
0 .53 a nd ove r . . . . . . . . . . . .. 50

Microscopic examination is the inspection of a sectioned

area using th e hi gher ma gnifica tion of 100 X through 1000 X,
aided by an et ching proc ess that enhances the different micro-
stru ctures. In most inst ances, th e sample prepared for micros-
copy will be the same sample used for ca se hardness traverse.
In fact, it is desirable to use this sample, because th e hardnesses
taken across the sampl e can now be as sociated with a specifi c
microstructure [Fig. 3-23) . The sample illustrated in Fig. 3-23
was taken from the midprofile of a gea r tooth that had not yet
been in servi ce; therefore, th e surface is unaltered. There has
been no "cold working" of the first few thousand th s of an inch,
which would tend to transform the retained aust enite into non -
tempered martensite .
The surface and near-surface microstructure of a gear tooth
will very often tell the story of the mode of failure . In fact, sev-
era l modes of surface pitting may appear to be identical when
vis ually inspe cted, but th e y ca n be co rrectl y diagnos ed only b y
studying the microstructure. These are detailed in Chapter 4.
Also, the mi crostructure is a factor in an y mode of pitting and
should be re cognized as such.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
76

50 f---/------j---=~..,_____+--+__1

40 f---/------j- - + - - _ +- -+__1

0 .010 0 .020 0 .030 0 .040 0 .050

Fig. 3-23. Case hardness traverse with Tukon 500 g at

0.003-in. interval (100 X) charted to show Rockwell "C"
hardness at each distance from surface. Material is SAE
4320H, carburized at 1700 of , diffused at 1550 of , direct
quenched in oil at 300 of , and tempered at 350 of . Micro-
structure, 500 X. 3% nita I etch. (A) Tempered acicular mar-
tensite retaining approximately 15% austenite, 58 HRC;
(B) Medium-carbon (0.40% C) martensite, 52 HRC; (C)
Low-carbon (0.21% C) martensite, 44 HRC.
 SYSTEMATIC EXAMINATION
77
The surface and near-surface mi crostructures are gene rall y
as so ciated with heat treatment and ma y con tain one or more of
t he following: mart en site, austenite, ca r bid es , decarburization,
and intergranular oxidation. Martensite may be eit he r acicular
[Fig. 3-23a ) or massive [Fig. 3-24). Au stenite is usuall y no t ob-
jectionable in m od est am ounts [Fig. 3-23) and ma y be welcome
for man y applications . Ho we ver , th e amount in Fig. 3-25 [45%J
ma y be somewhat immodest. Do not be a la rmed b y a dispersal

Fig. 3-24. Microstructure, 500X. 3% nital etch. Dense tem-

pered martensite (0.75%carbon). Carburized SAE4320H.
Retained austenite, nil.

Fig . 3-25. Microstructure, 500X. 3% nital etch. Tempered

acicular martensite in matrix of austenite (45%). Carbu-
rized SAE 4620H at 0.95% C.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
78

Fig. 3-26. Microstructure. 500X. 3% nital etch. Fine sphe-

roidal carbides scattered throughout a tempered marten-
sitic case.

Fig. 3-27. Microstructure. 200 X. 3% nital etch. Massive

globular carbides in matrix of tempered martensite.

of fine spheroidal carbides throughout th e case [Fig . 3-26); they

are welcome. Massive carbides [Fig. 3-27) may be acceptable in
some instances, but if a network appears [Fig. 3-28), it is most
unwelcome.
Decarburiza tion may be either gross or slight. Gross decar-
burization is an almost complete change to ferrite at the surface
(Fig. 3-29) that is very soft, with no strength or resistance to
 SYSTEMATIC EXAMINATION
79

Fig. 3-28. Microstructure, 200 X. 3% nital etch. Massive

carbides in network around grain boundaries. Matrix of
tempered martensite and bainite.

Fig. 3-29. Microstructure, 250 X. 4% nital etch. Gross de-

carburization of carburized SAE 4118 steel above matrix
of tempered martensite.
 SYSTE MATIC ANALYSIS OF GEAR FAILURES
80

Fig. 3-30. Microstructure, 100 X. Picral etch. Scattered

bainite (dark spots) along surface of a tempered marten-
sitic case.

Fig. 3-31. Microstructure, 100 X. 3% nital etch. Bainite

(dark structure) extending from the case/core tra ns ition
through the tempered martensitic case to the surface.
 SYSTEMATIC EXAMINATION
81
adhesive wear. Slight decarburization is more subtle. replacing
surface martensite with a bainitic structure (Fig . 3-30) that
may be only slightly lower in hardness and strength. Depending
on the amount. this mayor may not be detrimental. Bainite.
perhaps due to certain quenching problems . may also be
found to extend from the case/core transition zone to the sur-
face (Fig. 3-31). When this occurs at the root radius. the tooth
bending strength is somewhat reduced. Intergranular oxidation
(Fig. 3-32) at or near the surface is always present and usually
ignored. However. this can be practically eliminated today by
mechanical or chemical means. if necessary . If the mechanical
means is grinding. caution should be taken to avoid grinding
burns and checks. Electrochemical etching to remove inter-
granular oxidation may change the tooth pattern characteris-
tics and should be used with caution.

Fig. 3-32. Microstructure. 500X. As polished. Black grain

oxide network extending into the case from the surface to
approximately 0.0004 in.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
82

(c)
(a)

(b) (d) (e)

Fig. 3-33. (a) Spiral bevel pinion showing tooth spalling of

two adjacent teeth; 1.7X. (b) Spalling fatigue originating
subsurface. nucleated by a nonmetallic inclusion (arrow);
2.5X. (c) Scanning electron micrograph of the fatigue
origin at the inclusion (arrow); 27 X. (d) SEM closeup of
the central area of the inclusion; 237X. (e) SEM qualita-
tive analytical picture of the amount of aluminum within
the inclusion and the surrounding area.

Scanning electron microscopy (SEM) is a recent. highl y so-

phisticated tool in the study of fracture analysis; it is an exam-
ining method with a very great range of magnification used on
the fracture surface per se. The sample must not be altered or
abused prior to examination. In general. SEM is not a procedure
 SYSTEMATIC EXAMINATION
83
that is required for every fractured part, but it is highly valua-
ble when positive identification of a fracture mode is required
or when confirmation is needed for documentation. Its extreme
depth of field coverage, which allows for beautiful three-
dimensional photographs, is an important contribution. It also
has the capability of making a selective qualitative chemical
analysis of a specific structure. Occasionally, an opportunity
will arise when a unique structure within a fracture pattern
must be identified. Figure 3-33 is an example. It should be re-
membered that the scanning electron microscope is a tool to be
used to complement the other procedures and is not the tool for
a complete analysis.

References
1. Robert Clark Anderson, Insp ection of Metals. Volum e 1: Visual
Examination. American Society for Metals, 1983 .
2. Metals Handbook De sk Edition, American Society for Metals,
1985. p 34·7 .
3 . Society of Automotive Engineers, Iron & St eel Handbook Sup-
plement HS-30, 1981, Table 4. p 4·07.
4. Society of Automotive Engineers. Iron and St eel Handbook Sup-
plem ent HS-30. 1981, P 4·43-4·44.
 CHAPTER 4
Modes of Gear Failure

A ge a r has fail ed wh en it can no longer do

efficiently th e job for wh ich it w as designed.

Al though this defini tion generall y has been acc e pted , th ere
are inst ances when a cu stomer might claim a gear fail ure when
the supplier beli eves that the ge a r has been merel y "worn-in"
prope rl y and should not have b een remo ved from servic e. Also,
a gear may ha ve been placed in service under circ um s ta nces
compl et ely different from those for which it was designed. Al -
though it is very diffi cult to separate mod e from cause in many
insta nc es , the distinction should be mad e.
A mode of ge a r fail ur e is a particular type of failure tha t has
its own descripti ve id entification. One mode ma yor may not be
unique to a specific failure, since the ori gin may be of on e mode,
the progression of a second mode, and the fin al fracture of a
third mode . Generall y, the dominant mode in eac h instance will
be discussed.
Certain failure modes occur more fr equently than others.
Several failur e anal ysts in various fields of expertise have
compiled th eir findings regarding th e frequency of failure
mod es , and th ey have agreed that the most frequent is fatigue,
followed by impact (tensile or shear), and wear (abrasive or
adhesive). In a n anal ysis of over 1500 studies, the three most
frequent gear failur e modes were tooth bending fatigue (32%),
tooth bending impact (12 lf2%), and abrasive tooth wear (10%).
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
86
Fatigue
Fatigue Failure t-e-cracking under repeated stresses much lower
than the ultimate tensile strength-ordinarily depends on the
number of repetitions of a given stress range rather than the
total time under load, and it does not occur below the stress
amplitude called the "fatigue limit." The likelihood of fatigue
failure is increased greatly by the presence of notches , grooves,
surface discontinuities. and subsurface imperfections, all of
which will decrease the stress amplitude that can be withstood
for a fixed number of stress cycles. The likelihood of fatigue
failure is also increased significantly by increasing the average
tensile stress of the loading cycle.
There are three stages wi thin a fa tigue fail ure to be studied
closely: the origin of the fracture, the progression under succes-
sive cycles of loading, and final rupture of the part when the
spreading crack has sufficiently weakened the section. Most at-
tention is devoted to the first. to determine why the crack
started at a particular point. The second stage is observed to
determine the direction of the progression. A fa tigue crack will
follow the path of least resistance through the metal. The final
area of fracture may result from shear or tension. but, in either
case, examining this area may help determine the apparent
magnitude of stress that had been applied to the part.

Tooth Bending Fatigue

Tooth bending fatigue is the most common mode of fatigue
failure in gearing. There are many variations of this failure
mode . but the way to understand better the variations is first to
understand the classic failure.
The classic tooth bending fatigue failure is one that occurs
and progresses in the area designed to fail. Figure 4-1 shows an
example. Five conditions indicate why this is the classic tooth
bending fatigue failure:
[a] The origin is at the surface of the root radi us of the
loaded [concave] side of the tooth.
[b] The origin is at the midpoint between the ends of the
tooth where the normal load is expected to be.
 MODES O F GEAR FAILURE 87

Fig. 4-1. Spiral bevel pinion showing classic tooth bend-

ing fatigue with origin at midlength of the root radius on
the concave (loaded) sid e.O .6X.

[ c) One tooth failed first; the fracture progressed slowly

toward the zero -stress poin t at th e root, whic h shif ted
d uri ng the progress io n to a p oint un der the oppos ite
root radius and then proceeded outward to that radius.
[d) As the fracture p rogressed, the tooth deflected at each
cycle un ti l the loa d was pi ck ed up si multaneo us ly by
the top co rner of th e nex t too th ; beca use of th e ove r-
loa d , a too th be ndi ng fatig ue fa il ure sta rted in the same
area. [Note that the fracture of the second toot h ap-
pears to have an origin more recent than the first
fracture. )
[ e) T he ma te ri al a nd met allurgical cha racteris tics are
wit hin specifications, a nd the pinio n should ha ve
withstood nor mal operating cond itions .
If any of the five conditions is changed, there will be a deviation
from the classic examp le of tooth be ndi ng fatig ue .
Foll o wing is a g uide to th e ru di me nts of class ic toot h b en d-
in g fatigue with a study of p hotoelastic mode ls . Figure 4 -2
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
88

Fig. 4-2. Photoelastic study of two mating pinion teeth re-

ceiving full load. Note the high concentration of compres-
sive stress at the point of contact, the tensile stress at the
root radius. and the zero-stress point at the tooth center-
line below the root circle.

Fig . 4-3. Photoelastic study of mating teeth indicates the

shift of the zero-stress point during crack propagation un-
til final fracture reaches the opposite root radius. At the
same time , deflection of the tooth allowed the adjacent
tooth to pick up the load.
 MODES OF GEAR FAILURE 89

shows two mating teeth in contact at or near the pitchline. Note

the high concentration of compressive stress at the pitchline , or
point of contact; the concentration of tensile stress at the root
radius of the loaded side; the concentration of compressive
stress at the root radius of the opposite or nonloaded side; and
the zero-stress point below the root circle at or near the tooth
centerline. The progression of any fatigue crack initiated at the
root radius will be in the direction of this zero-stress point,
since this is the path of least resistance.
Once the crack a t the root radi us is initia ted and is moving
toward the zero-stress point (see Fig. 4-3 J, the point moves
away from the leading edge of the crack and shifts laterally,
until it reaches a position under the opposite root. At that lime,
the shortest remaining distance is outward to the root, where it ,

terminates. In the meantime, as the failing tooth is deflected, the i

top of the adjacent tooth picks up the load. The load on the first II
tooth is relieved, allowing a slower progression due to lower
cyclic stress. It is only a matter of time until the next tooth in-
itiates a crack in the same position.
Variations of the classic tooth bending fatigue occur for
many reasons. A few are described in the paragraphs that
follow.
Tooth bending fatigue of a spur gear tooth with the origin
along the surface of the root radius of the loaded side, but at a
point one-third the distance from the open end (Fig. 4-4), can be ~I
considered a classic failure . The maximum crown of the teeth
had been placed intentionally at the area where the failure oc-
curred, and thus, the maximum applied loads were in the same
area. This, then, is the area that was designed for the point of
initiation of normal tooth bending fatigue. Note that the pro-
gression was directly to the bore-the shortest direction and the
path of least resistance.
Tooth bending fatigue of a spur gear tooth, meeting all pre-
viously listed conditions for a classic failure except (b), is
shown in Fig. 4-5. The origin was a t one end of the tooth and not
a t the midlength. There migh t be several reasons for this poin t
of initiation; but in this instance, a severe shock load had
twisted the parts until a momentary overload was applied at
the end of the tooth. This caused the formation of a crack that
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
90

'I
II
il
,I
:i
i

Fig. 4-4. Spur tooth pinion at 1X (top) and 2.5X (bottom).

Tooth bending fatigue originating at the root radius (ar-
rows) . loaded side. one-third the distance from th e open
end. Pro gre ssion was to the bore.

b ecam e a hi gh stress-co nce n tration po in t. from w h ic h fati gu e

p rogres si on co n ti n ue d .
A not her exa m p le of a n a lmos t class ic fail ure , th ou gh la ck -
in g co ndi tio ns [ b] [off -cent er] a nd [c] [continuing under tooth to
o p pos ite roo t ra d iu s] . is s how n in Fig . 4-6 . T w o reaso ns for th is
s pec ific variat io n a lte re d th e cond i tion s . Co ndit io n [b] was
change d becau se of a n off -ce n ter w eb und er th e gea r teeth. T he
di st orti on of the tee th d uring q ue nc hing crea ted a se lf- im posed
 MODES OF GEAR FAILURE
91

Fig. 4-5. Spur pinion , 7/8 X. Tooth bending fatigue with

origin a I root radius of loaded side alone end of Ihe looth.

Fig. 4-6 . Helical gear, 1.12 X. Tooth bending faligue fol-

lowed by looth bending impact. Origin is off-center of the
tooth midpoint but is directly above the center of the web.
 SYSTEMATIC ANALYSIS OF GEAR FAiLURES
92

Fig.4-7. Spur pinion, 100 X. Tooth bending fatigue at mid-

length of tooth at root radius, but origin is at an inclusion
located in the case/core transition.

cro w n directl y above the ce n te r of th e w eb, which b ecame a

p ost-d esi gned area of initiati on for normal tooth bending fa -
tig ue. And the propagation of th e fat igu e crac k did not ex te nd
ve ry far, until th e stress appli ed b y th e normal loads excee de d
th e tensile strength of th e remaining mat erial. In this inst ance,
th e initial mode of failu re w as tooth b ending fati gue , foll ow ed
cl osel y b y tooth bending im pac t.
An exa m p le of tooth bending fa tig ue that did not co nfo rm
to c la ss ic con d it ion (0 J is s ho w n in Fig.4-7. Although th e ini ti a-
tion was at th e root radius a t th e midlength it did not originat e
a t th e s urface. The origin w a s a t a nonmetallic inclu sion at the
case/ core interface . Subs equent progression oc curred throu gh
th e ca se to th e s urfa ce, a s w ell a s through the core toward the
zero-st ress point.
Another example of a tooth b ending fatigue that did not
confo r m to condition [n] is d epi ct ed in Fig. 4-8. The fa il ure w as
at th e root radius and at the midl ength, but th e origin was n ot at
 MODES OF GEAR FAILURE
93

Fig. 4-8. Spiral bevel gear tooth, O.7X. Tooth bending fa-
tigue with origin at the apex of the drilled bolt hole , which
terminated just below the root radius.

Fig. 4-9 . Spur pinion tooth, 2X. Bidirectional tooth bend-

ing fatigue , with applied loads at a high spot at the top
edge on both sides. Origin is at the case/core interface,
near the crown of the tooth.
the surface. The origin was at the apex of a tapped bolt hole that
had been drilled from the back face of the ge a r and had termi-
nated less th an 1/4 in . from the root radius . Th e fati gue crack
progressed to the surfa ce at the root radius, a s well as into the
core toward th e zero -stress point.
Figure 4-9 also shows tooth bending fati gue, but it is of a
different nature . In fact, from appearance only, it might be
called "spalling." However, due to the reasons of origin and
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
94

Fig. 4-10. Sketch of forces , shear and tensile stress planes,

and directions present in the failure depicted in Fig. 4-9.

progression, it must be placed in the tooth bending category.

Referring to the sketch in Fig. 4-10, picture a spur tooth with a
high spot at the top of the active profile on both sides, which is
taking the full pressure and deflection (P) as the tooth goes into
and out of mesh. The maximum shear plane (S) is in line with
the pressure point and passes through the case/core transition
zone, which is of lower tensile strength than the case. The tensile
stress (T], which is maximum along a plane 45° from the shear
plane , is greater than the strength of the core material, and an
internal stress rupture originates at the case/core interface
where the two shear planes intersect. The fatigue progression is
along the planes of maximum tensile stress, until its direction is
changed by the fulcrum effect of the opposite face, causing it to
follow the path of least resistance tv the surface . Since this in-
stance is bidirectional, and the two events occur simultane-
ously, a classic bidirectional fatigue pattern is evident.

Surface Contact Fatigue (Pitting)

Pi tting, as a surface or near-s urface fail ure, is recognized as
a material-fatigue mode of failure . As elements of the structure
 MODES OF GEAR FAILURE
95
at and near the surface are subjected to alternating compressive
stresses, plastic deformation occurs in some microstructures
and elastic deformation in others. The differences of plasticity
between grains will encourage fatigue cracks to initiate under
pulsating and alternating stresses. Also, each grain is randomly
oriented to the direction of its shear plane. Under a specific
compressive load, some grains will tend to break apart in ten-
sion, while others will tend to shear, and still others will be
unaffected. It is these internal stress raisers that will allow a
crack to form at a specific point when the resultant strain ex-
ceeds the elastic limit of that point. The cracks flow together
and accumulate to form a plane of progression; then the path of
least resistance is followed and a particle falls away from the
surface. This is the start of pitting.
All too often , there may be present at or near the surface an
embedded inclusion that will act as a nucleus for a crack pro-
gression. The point of inclusion is usually very weak and most
susceptible to crack initiation [Fig. 4-11].

Fig. 4-11. Spur gear tooth, 100X. Nital etch. An internal

crack originating at an oxide type of inclusion below the
surface at the pitchline. A pit is being formed.
T

SYSTEMATIC ANALYSIS OF GEAR FAILURES

96

-
Sh ear st re ss

! /
I

Di st an ce
,./
bel ow I Pur e sliding ;~
su rf ace
V ;/
/ /"" ;~ Combin ed rolling
I / and sliding

Fig. 4 -12 . Stress distribution in contacting surfaces due to

rolling, sliding, and combined effect.

Th e initiation of pitting is confined mostly to three areas

along the profile of a gear tooth: [a] the pitchline, [b] the area
immediately above or below the pitchline, and [e] the lowest
point of single tooth contact.
The pitchline is the only area of line contact that receives
pure rolling pressure. The pits forming directly at th e pitchline
are usually ve ry small and may not progress beyond the point of
origin. Some analysts claim that this type of pitting "heals" it-
self and is not detrimental. That may be true when th e pitting is
formed b y the breakdown of a rough surface. It should also
be noted that lubrication will not deter the origin of pitting
along the pitchline. Oil is noncornpr'e ssible and will not "cush-
ion" the pressures exerted in the area of pure rolling.
The area immediately above or below the pitchline is very
susceptible to pitting. Not only is the rolling pressure great at
this point, but sliding is now a crucial factor. The mechanics of
surface/subsurface pitting can best be understood by looking at
the resultant applied stresses illus tra ted in Fig. 4-12. 2 The ext ra
shear stress of the sliding component, when added to that of th e
 MODES OF GEAR FA ILURE
97

Fig. 4-13 . Helical gear teeth, 2X. Pitting initiated along the
pitchline and just above the pitchline. In some areas , the
progres sion ha s been continuous.

Fig. 4-14 . Spiral pinion tooth , 200X. Nea r-pit chline pitting
fatigue. Origin is s ub surfac e at plane of ma ximum shear.

rolling co mpo ne nt, ofte n res ults in surface fatig ue a t the poin t
of ma ximum shear bel ow th e s ur face. In man y ins ta nces , it is
difficu lt to determi ne whet he r some pitting cracks actually ini -
tiate at or below the surface. Figure 4 -13 s hows initiation of
pitt ing fatigue both at the pitchli ne of a helical gear tooth and
d ire ctl y above the pit ch li ne. Progress ion u p the ad de nd um in
so me areas make s it difficult to diffe re ntia te betwee n the two . A
surface -pitted a rea near th e pit chline is illu s tra ted in Fig. 4-14.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
98

Fig. 4-15. Spiral bevel tooth, 100X. Nital etch. Pitting at

lowest point of single tooth contact, illustrating contact
path of the tip of the mating tooth.

Although the surface was in a rolling-sliding area, the imme-

diate surface shows no catastrophic movement; so it is most
probable that shearing fatigue took place as the initiating force
below the surface. Also , the freedom from surface movement is
an argument for the lubrication of the gear teeth.
The lowest point of single tooth contact is the point that
receives the tip of the mating tooth as it makes first contact low
on the active profile. Tip contact will produce high pressures
per unit area, even if a small fraction of the actual load is trans-
ferred by the tip. Also, the maximum sliding speed, both in ap-
proach and recess, occurs at the tip contact. Two types of pits
prevail at this area. One is shown in Fig. 4-15; this appears to be
a swift shear and lift-out type . Perhaps it can be questioned
 MODES OF GEAR FAILURE
99
whether this is fatigue or impact; however, if it is a shearing
fatigue, it is certainly rapid. Causes are not being assessed in
this chapter, but if they were, this example would certainly call
for a study of tip interference and lubrication. The second type
of pit is the same as that illustrated in Fig. 4-14, showing failure
only by compressive loads. This is not uncommon in spur gear
teeth or helical gear teeth; but when found, it is puzzling, since
one does not expect to find tip interference pits other than those
previously discussed. However, it must be remembered that
dynamic effects are to be taken into account. Gear vibration is a
fact but is seldom ever recognized as such. Studies indicate that
gear vibration at high speeds, excited by static transmission
error, may cause corner contact , even though the tip relief is
sufficient to prevent corner contact at low speed. " Also, with
sufficient lubrication, the pressures of tip interference will re-
sult in compressive-type pitting, such as in Fig. 4-14, rather
than the adhesive type , illustrated in Fig. 4-15.
Besides the three specific and most common areas of sur-
face fatigue pitting just discussed, there are random areas that
demand recognition, such as pitting only at one end of the face,
or at opposite ends of opposite faces, or along the top edge of the
addendum, or at one area of one tooth only. The random inci-
dences are without number and can be most baffling ; but each
one occurs because of a reason, and that reason must be found,
if at all possible.
One patterned type of surface fatigue pitting is found only
on spiral bevel teeth. To understand this unusual pattern of pit-
ting, which can lead to total destruction, one must recognize the
direction of the loaded patterns of a spiral bevel tooth . An excel-
lent, detailed study reproduced in the Source Book on Gear De-
sign . T echnology and Perforrnn nce- explains, in effect, that the
load pattern is a line contact lying at a bias across the tooth
profile, moving from one end of the contact area to the other end.
Under load, this line assumes an elliptical shape as its pattern
and th us distri bu tes the stress over a larger area . Also, the pur-
pose of spiral bevel gearing is to relieve the stress concentra -
tions by ha ving more than one tooth enmeshed a t all times.
There are instances , however, when a certain designed spiral
angle will allow a one -to -one tooth ratio for a very short dura-
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
100

Fig. 4-16. Surface fatigue pitting initiated in a short con-

centrated area of a spiral bevel tooth, when that tooth was
momentarily assuming full load with no help of overlap
from adjacent teeth.

tion of time . During this moment, the load more than doubles on
an area that is not much more than a line contact (Fig. 4-16 J. The
resulting stress will be high enough under full-loaded condi-
tions that pitting will quickly initiate in this area.

.1 , Rolling Contact Fatigue

Rolling contact fatigue can be described only by using the
illustration in Fig. 4-12 . Under any conditions of rolling, the
maximum stress applied at or very near the contact area is the
shear stress, parallel to the rolled surface at some point below
the surface . For normally loaded gear teeth, this distance is
from 0.007 to 0 .012 in. below the surface, just ahead of the roll-
ing point of contact. If sliding occurs in the same direction, the
shear stress increases at the same point. If the shear plane is
close to the surface, light pitting will probably occur. If the
shear plane is deep due to a heavy rolling load contact, the crack
propagation tends to turn inward (Fig. 4 -17J. The cracks con-
tinue under repea ted stress until hea vy pitting or spalling takes
place (Fig. 4 -18J .
 MODES OF GEAR FAILURE
101

.
\"
)
.1'
t'
~;{J.
rJ.
Fig. 4-17. Gear tooth section, 100 X. Unetched. Rolling
contact fatigue. Crack origin subsurface. Progression
parallel to surface and inward away from surface.

J
.
'
/ -

..,.
I
f .' '
1 ( ... . \
. ..,. .,...,O--~
V~
~ . r« ,
,_
- . -
..,-.

Fig. 4-18. Gear tooth section, lOOX. Unetched. Rolling

contact fatigue. Crack origin subsurface. Progression
parallel with surface, inward, and finally to surface to
form a large pit or spall.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
102

Fig. 4-19. Gear tooth section, 200 X. Unetched. Rolling

contact fa tigue distinguished by subsurface shear parallel
to surface. Note the undisturbed black grain oxides at the
surface, indicating no surface material movement.

There is always one and often two characteristics of rolling

contact fatigue that distinguish it from other modes of surface
contact fatigue. Both characteristics can be observed only by
examination of the microstructure, which is a metallurgical ex-
amination [Chapter 3). In rolling contact, the surface will not
show a catastrophic movement; it will remain as the original
structure . For instance, an unetched, polished sample taken
near the origin of a subsurface fatigue crack [Fig. 4-19) very
clearly shows undisturbed black grain oxides at the surface.
The subsurface cracking could not have been caused by either
abrasive or adhesive contact-it had to be caused by rolling .
This is the first evidence to look for in determining the type of
applied stress.
The second characteristic of railing contact fatigue is com-
mon only in a martensitic case that contains very little or no
austenite and is found at, along, or in line with the shear plane.
This is a submicrostructure that has been named "butterfly
wings." Take the sample of Fig. 4-19 and etch it properly with a
3% nital solution, and a classic wing shows up, as in Fig. 4-20(a).
Increasing the magnification to 500 X results in a bit more detail,
as in Fig. 4-20(b) . The original dissertation was presented by
 MODES OF GEAR FAILURE
103

(a)

(b)

Fig. 4-20. Same sample as Fig. 4-19 , 3% nital etch; at 200 X

(a) and 500 X (b), showing details of submicrostructure
called "b u tt er fl y wings."

the ro ller bearing m anufa ct urers. The y had no e x p la na tio n fo r

th e or ig in of this stru cture , exce p t th at it "was al w a ys in co n-
juncti on w it h a nd radiat ed fro m a n incl usio n ." It is true th at
man y times a n in clu si on is p resent , but not a lways . Ho w e ver, i t
is argu able th at the fr a cture is a pl ane a nd not a line s uc h as in a
polished sample ; th erefore , a n inclusion co uld be pre sent
s omew here else along th e pl ane w here it is n ot ob s e r ved. Th e
ne x t gro u p of studi es w as tw ofold-one ex te n s ive s t udy b y
Arthu r D. Littl e Compan y of Bost on, M ass a ch usett s , a nd a very
d et ailed s t udy of rolling cont a ct fat ig ue b y Illinoi s In stitut e of
Technolo g y Res earch Institut e [IITRI] in coo pe ra tio n w it h Di -
vi sion 33 [ISTC] of th e So ciety of Automotive En gineers . In
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
104
their reports, the gray substructure was referred to as "white
bands" of altered martensite .
During the time of these intensive studies, many such al-
tered structures were observed in gear teeth where there was
evidence of very heavy rolling contract. For example, observe
Fig. 4-17 and 4-18 as etched, which are now Fig. 4 -21 and 4-22,
respectively. Note the altered martensite substructure along
the sheared plane . These substructures seem to be caused
when, under an extreme shearing stress, movement is needed ,
but is restricted and contained to such an extent that the energy
absorbed institutes a metamorphic change in the microstruc-
ture ahead of the progressing crack. The substructures are
never observed when significant amounts of austenite are pres-
ent, because austenite will quickly absorb the energy and be
converted to untempered martensite. The substructures are
also in an area that has not been deformed but has definitely
been transformed. Each area has distinct boundaries, and the

..
.it
. .
J> , • •

. ., ..

, l

,.f
" .
' ..
,
"
.' v·
Fig. 4-21. Same sample as in Fig. 4-17. Nital etched. Note
the "butterfly wings," or altered martensite, along the
shear planes.
 MODES OF GEAR FAILURE
105

,I

Fig. 4-22. Same sample as Fig. 4-18. Nital etched. Note the
"butterfly wings," or altered martensite, along the shear
planes.
oncoming cracks appear to follow those boundaries. Many aca-
demic studies refer to these same structures as transformed
shear band products formed by "adiabatic shear."s

Contact Fatigue (Spalling)

Spalling generally is not considered an initial mode of fail-
ure. but rather a continuation or propagation of pitting and roll-
ing contact fatigue. It is very common to refer to this failure
mode as "pitting and spalling." As an example. Fig. 4-23 shows
a spiral gear tooth with pi tting low on the profile, which subse-
q uently progressed un til spalling occurred over the top face and
back side profile. This apparently rapid and extensive progres-
sion is often referred to as the "cyclone effect."
Idler gears were removed from three planetary drive as-
semblies. The teeth on each gear showed a remarkable example
of pitting and spalling progression. Figure 4-24[a) shows lines
of pitting just below the pitchline; [b) shows the pitting pro-
gressed to light spalling up and over the addendum; and (c)
shows complete spalling of all teeth .
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
106

Fig. 4-23. Spiral bevel gear teeth, 1.5X. Original pitting

low on the active profile gives initiation to a fast and ex-
tensive progression of spalling over the top face and
down the back profile. This is often called the "cyclone
effect."

Spalling, in the true sense of the word, is a distinctive

mode of fatigue failure unique in or-igin." It originates subsur-
face, usually at or near the case/core transition zone . As illus-
tra ted in Fig. 4-25, the origin is at the poin t where the sum of all
applied and mi sapplied stresses intersects the net strength of
the part. The applied stresses are substantially in shear, and the
point of intercept is most likely under a carburized case . Fatigue
fra ctures generally progress under the case as shown in Fig.
4-26, and will eventually spall away from the gear tooth .

Thermal Fatigue
Thermal fatigue in gearing is most often considered (al-
though not ne cessarily correctly) to be synonymous with fric-
tional heat. This is virtually the only kind of alternating heating
and cooling that is applied in the field . Then, too, thermal fa-
tigue is not commonly found on the active profile of the gear
 MODES OF GEAR FAILURE
107

(a)

(b)

(e)

Fig. 4-24. Spur gear, O.5X. Pitting fatigue progressing to

spaIling. (a) Lines of pitting just below the pitchline; (b)
light spalling up and over the addendum; (c) complete
spaIling of all teeth.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
108

60 -- <, Net
(

u
Qj
::
50
f'"
i', \ ").
/ stre ngth

-"
o 40
0
a:: / "~ _ Failure
App lied str ess '" origin
ui
I/l
30
r-,
Cll

I ~ ~<,
C
""0
:0
J:
20
"
10
"
0 .010 0 .020 0 .030 0 .040 0 .050 0 .060
Depth below surfa ce. in.

Fig. 4-25. Ap plied stress vs case de pt h (net s treng th ).

Fig. 4-26 . Spalling-a subsurface fatigue failu re ori ginat -

ing at the ca se /core inte rface , s ub seq uen tly progressing
under the case. .

teeth b u t on a rot a tin g fa ce exerti ng a hi gh amou n t of thrus t.

Fig ure 4 -27(a) is an examp le of hig h end thrust th at resulted in
frictional heat. The th ermal expansio n and co n tra ctio n initiated
a large n umber of radial cr a cks that progressed as a fatigue
fra ctu re (Fi g. 4-27 b 1 bo th into th e bore of the pa rt a nd a lo ng t he
root s of th e teeth.
 MODES OF GEAR FAILURE
109

(a) (b)

Fig. 4-27. (a) Spur gear showing radial cracking due

to frictional heat against the thrust face; O.5X. (b) Pro-
gression of thermal fatigue produced by the frictional
heat; 2X.

There is a remote chance that grinding checks and quench-

ing cracks could be construed as very rapid thermal fatigue .
Both are discussed as manufacturing causes in Chapter 5.

Fatigue of Round, Splined,

and Keyed Bores
Fatigue fractures moving from the bore outward, with very
few exceptions, originate at a high stress-concentration point.
The subject of these applied stresses has been covered thor-
oughly in Chapter 1, in the section titled "Associated Para-
meters."

Shaft Fatigue
Torsional fatigue as a failure in the tensile plane follows a
direction of 45° to the central axis [see Fig. 3-2). However, the
weakest plane of a shaft or pinion shank is the shear plane in the
longitudinal direction. Under continual reversal of torsional
loads that exceed the shear strength of the material, the crack
will initiate at the surface, if the hardness is consistent
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
110
Carb uri zed shaft (4 3 BV 12), torsio nal f at igue

Progression of failur e bel ow case

Fig. 4-28 . Schematic showing subcase failure of bidirec-

tionaltorsional shear fatigue followed by torsional tensile
failu re of the case.

Fig . 4-29. Splined section of a pinion sha nk . Torsional

tensile fa ligu e in one direction s how ing the 45 ° tensile
failure lines and evidence of lon gitudinal sh ear cracking.
 MODES OF GEAR FAILURE
111
throughout the cross section. However, if the surface is harder,
either by carburizing or by induction hardening, the crack will
most likely initiate subsurface at the case/core interface. The
schematic shown in Fig. 4-28 is typical of a longitudinal fatigue
failure in a carburized shaft. The initial failure is shear fatigue
at the case/core interface, which continues to progress through
the shaft and longitudinally under the case . The second failure
occurs as the case ruptures in a 45° cross-hatched pattern typi-
cal of a tension failure in torsion, since the case is weaker in
tension than in shear.
When splines are present, such as in the pinion shank
shown in Fig. 4-29, the area most susceptible to failure is be-
tween the juncture of the mating part and the change of section
beyond that juncture. Note three fractures involving the failure
in Fig. 4-29, First, the 45° cracks show torsional tensile failure
in one direction only; also, the entire short area has been
"stretched" in the displacement; and finally, there are longitud-
inal cracks along the root radii of the displaced splines. A typi-
cal three-directional fatigue failure leads ultimately to the clas-
sic "rosette" fracture [Fig, 4-30J often observed in splined shaft
failures .

Fig. 4-30. Splined section of a shaft. The typical three-

directional reversing torsional fatigue "rosette."
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
112
Table 4-1. Fracture appearances of fatigue failures in bending?

~
No s tress Mild str ess High s tres s (
con centration con centra tion concentration
Stress Low High Low High Low High
Case condition over stress overstress overstress overstress overstress ov er stres s

On e-way
bending load
~ c
• -e It
Two-way
bending load
e•0 8 - e
Reverse bending
and rotation f_ - • • (!)
"
load

Bending fatigue in shafts and pinion shanks is seen in many

different patterns depending on the direction of loading and the
continued load application. Charles Lipson ? devised a table
[Ta ble 4-1 J comprising the q uali ta ti ve effect of loading method,
loading magnitude, and stress concentration on the appearance
of the fracture . This table can be used very effectively to judge
types of bending fatigue. For instance , observe the similarit y
between illustrations in the table and Fig. 4 -31 , 4-32 , 4-33, and
4-34 .

Impact
Tooth Bending Impact
When a tooth is removed from a gear within very few cycles
[usually one or two J, the resulting fracture is uniform in struc-
ture and does not show the fatigue striations common to the
fatigue mode of failure . The failures are usually random, due
to a sudden load, in either a forward or a reverse direction, and
do not necessarily originate at the root radius. In fact, if the
fracture originated at the root radius , it would follow a rather
 MODES OF GEAR FAILURE
113

Fig. 4-31. An arm of a differential spider. Two-way bend-

ing load; mild stress concentration; very low overstress.

Fig. 4-32 . A tt/a-in. shaft next to thread relief. Slight re-

versed bending under rota tionalload; high stress concen-
tration; very low overstress.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
114

Fig. 4-33. A 4-in.-diameter keyed shaft. Reversed bend-

ing ; rotational load; high stress concentration; high over-
stress.

Fig. 4-34. A 5-in.-diameter taper splined shaft. Reversed

bending; rotational load; very high stress concentration;
high overstress.
 MODES OF GEAR FAILURE
115

Fig. 4-35. Tooth bending impact with stress-coat overlay,

showing fracture path short-circuiting the usual stress
flow-lines.
flat path across to the opposite root radius , rather than travel
downward toward the zero-stress point. Figure 4-35 shows a
stress-coated overlay above a one -shot broken tooth. The frac -
ture did not have time to follow the usual stress pattern. A field
failure example is shown in Fig. 4 -36.

Tooth Shear
When the impact load is very high and the time of contact
very short, and if the ductility of the material will allow it, the
resultant tooth-failure mode will be shear. The fractured area
appears to be highly glazed, and the direction of the fracture is
from straight across the tooth to a convex shape. For instance, a
loaded gear and pinion set were operating at a high rate of speed
/ when the pinion stopped instantaneously (Fig. 4-37J. The mo-
mentum of the gear was great enough to shear the contacting
pinion teeth from the reverse direction, leaving the remaining
teeth in excellent condition . The gear teeth were partially
sheared and were all "scrubbed" over the top face from the re-
verse direction.
 SYSTEMATIC ANALYSIS O F GEAR FAILURES
116

Fig. 4-36. Spiral gear teeth, I X. Tooth bending impact

with peak loads being applied high on the profile over the
top corner of the heel end of the convex (loaded] side.

(a) (b)

Fig . 4-37 . Spiral be vel gear and pinion set, 1 .5X. Sheared
in reverse direction. The pinion came to a sudden and
complete stop at the instant of a primary failure of the
unit, allowing the gear to shear the contacting teeth and to
continue rotating over the failed area.
 MODES OF GEAR FAILURE
117

Fig. 4-38 . Spur gea r tooth . Combination modes: Tooth

bending impact (top) ; tooth sh ear (bottom) . Arrows
indicate direction of applied force .

Mos t commo nly. be ndi ng impac t a nd shear are fo und in

co m bi natio n. Fig ure 4-37(a] shows thi s to be the case wit h th e
gear teeth fractures . Also clearly defined is a sp ur gea r toot h
shown in Fig. 4-38 . T ooth bending impact fractures s upersede
the shearing fractures. a nd thus. the directio n of im pact ca n be
determi ne d .
 SYSTEMATIC ANALVSIS OF GEAR FAILURES
118
Toot h Ch ipp ing
Tooth chipping is certainly a type of impact failure but is
not generally considered to be tooth-to-tooth impact. It is usu -
ally accomplished by an external force, such as a foreign object
within the unit, a loose bolt backing out into the tooth area, or
even a failed tooth from another gear. Figure 4-39 show's a chip-
ping condition thatI
existed during the operation of the part.
Most chipping, however, has occurred through mishandling
either before, during, or after the fi nished parts have been
shipped from the ma nufacturer, and is not a "fie ld" failure. T his
is one mode that must be ide ntified by a cause .

Fig. 4-39. T ooth ch ipping as a fie ld fa ilure generally

show s a "pa tt ern"; i.e ., some obj ect of impact was withi n
th e assemb led un it.

Case Crushing
Case crushing wi ll occur when an extreme overload is ap-
plied to a~~rburized case. Four factors combine to cause a case
to be crushed. Consider Fig. 4-40 : Case crushing depends on the
stress applied at the point of contact, the radius of curvature of
the contacting surfaces, the thickness of the case, and the ha rd-
ness of the core material. The resulting failure is caused by
 MODES OF GEAR FAILURE
119

Fig. 4-40. Case crushing depends on stress applied, radius

of curvature, case depth , and core hardness.

II
I
I

Fig. 4-41. Case crushing at mid profile of a spiral be vel

gea r tooth. Progression is from th e subcase area into the
core and outward to the surface .

com press ive load s per unit area tha t are excessive for the ex ist -
ing conditions. T he fracture starts at the case/core inte rface and
continues to shea r in to the core and outward to the surface . Fig-
ure 4 -41 shows the subsurface propagation of the crushi ng ef-
fect. [Refer also to Fig. 3 -1 for a view of a typ ical surface a p-
pearance of case cr us hi ng.) For an examp le quite the opposite of
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
120

Fig. 4-42. Extensive case crushing. Heavy loads, thin

case, soft core.

the one in Fig . 4-41, see Fig. 4-42: very heavy loads, light case
depth , and low core hardness; it displayed excellent ductility
and never lost a tooth.

Torsional Shear
Again, our attention is turned to shafts or to pinion shanks
directly subject to torsional stress. In most instances, torsional
shear is a secondary mode following the mode of torsional fa-
tigue, as shown earlier in Fig. 4-30 . But sudden shear is readily
identified; its appearance (Fig. 4-43) is unique in straightness
and texture.

Wear
Surface deterioration of the active profile of the gear teeth is
called "wear." There are two distinct modes of wear that will be
discussed-abrasive and adhesive .
 M O DES OF GEAR FAILURE
121

r
»
<.
-r
~
~.

-~: ...
~ "'>~'
, \"

f'I·~~(~t .
-,

""" ...
,\'
.
l
~
y.
, ~ .-, :)t
'". "~ \
. '

s- , ... ,'I

I ,"

... e , _ . 'r :

Fig. 4-43 . Torsional shear. Suddenly applied extreme tor-

sional overload.

Fig. 4-44. Spiral bevel gear teeth showing contact wear.

Insert A is a tooth area non worn. Insert B shows abrasive
wear clearly cutting away 118 in. of the surface without
damage to underlying material.

Abrasive Wear
Abrasive wear occurs as the surface is being cut away by
hard abrasi ve particles. It can happen only as two surfaces are
in sliding contact. The dissociated material must continually be
washed away and not be allowed to build up on the sliding sur-
faces, or adhesions may take place. The first evidence of abra-
sive wear is the appearance of light scratches on the surface,
followed b y scuffing. As the scuffing deepens, scoring results.
Figure 4-44 is an excellent example of pure abrasive wear,
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
122

Fig. 4-45. Pinion tooth profile,100X. Glazed surface show-

ing the start of catastrophic movement of surface mate-
rial. Frictional heat has already started to temper the
surface.

showing the entire tooth profile cleanly cut away with no mi-
crostructural damage to the underlying material.
Abrasive wear cannot be deterred by lubrication, because
the lubricant is often the vehicle that contains and continually
supplies the abrasive material as a contaminant. When contam-
ination occurs, all moving parts . within the assembly are af-
fected, including seals, spacers, bearings, pumps, and mating
gears (see Fig. 2-3).
When abrasive wear is isolated to only one part [i.e., either
gear or pinion), it is imperative to examine closely the surface of
the mating part. For instance, a very heavy amount of massive
carbides (refer to Fig. 3-27 and 3-28) that impinged upon the
surface may easily cut into a softer mating surface.
 MODES OF GEAR FAILURE
123
Adhesive Wear
Adhesive wear occurs on sliding surfaces when the pres-
sure between the contacting asperities is sufficient to cause lo-
cal plastic deformation and adhesion. " Whenever plastic de-
formation occurs, energy is absorbed as heat-frictional heat.
The first indication of trouble is a glazed surface, followed by
galling, then seizure. A glazed surface may not undergo any di-
mensional changes, but examination of the microstructure re-
veals a catastrophic movement of surface material (Fig. 4-45).
As frictional heat increases, the surface softens, and adhesive
ability becomes greater; further plastic deformation occurs; the
heat becomes high enough locally to change the microconstitu-
ent at the surface completely (Fig. 4-46); and now galling oc-

Fig. 4-46 . Pinion tooth profile,100X. Plastically deformed

by frictional heat and sliding pressures. Surface layer has
locally rehardened and galling is evident.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
124

Fig. 4-47. Spiral bevel pinion, 3/4X. Glazing, galling, and

adhesure over the active tooth profile.

Fig. 4-48. Spiral bevel gear tooth. Internal rupture at top

corner of case/core interface.
 MODES OF GEAR FAILURE
125
curs. As galling continues , the sliding surfaces begin welding
together, and adhesures take place. Large areas may be pulled
away from the surface (Fig . 4-47] . Some of the welded particles
are so hard. due to change of microstructure. that they become
sharp points that cut into the mating parts. As adhesive wear
continues with the parts still in service, there is the distinct
possibility that the part will ultimately be worn beyond repair
(see Fig. 2-4J.
Adhesi ve wear on gear teeth may not always point to the
set of gear teeth as the culprit. There are instances when severe
adhesures are secondary; particles of material from a primary
failure of another component in the assembly may impinge
upon the gear tooth surfaces . starting the sequence of events
tha t leads to seizure and destruction .

Stress Rupture
When the internal residual stresses build up to a magnitude
beyond the strength of the material, the part will rupture. The
rupture occurs at the point at which this critical value is ex-
ceeded. either internally or externally.
Internal rupture. The point within a gear most likely to at-
tract a buildup of residual stresses is the case/core interface
near the top face or corner of a tooth. Figure 3-3 (b J shows an
excellent example of case /core separation due to internal resid-
ual stresses exceeding the strength of material at the case /core
transition zone. Figure 4-48 shows the same mechanism at the
corner of a tooth. In severe cases, the entire top of the tooth
might "pop off" (Fig. 4-49).

Fig. 4-49. Spiral bevel gear tooth . Internal rupture lifting

the entire top of a tooth.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
126

Fig. 4-50. Spur gear. External rupture (assembled. but not

in service) with origin at the end face . (See Fig. 3-5 and
4-51.)

, .'

Fig . 4-51. Spur gear. External rupture originating from a

grinding check. (See Fig. 3-5 and 4-50 .)
 MODES OF GEAR FAILURE
127
External rupture is much easier to understand because it
almost always originates with a prearranged stress raiser.
Refer to Fig . 3-5. This part (not yet in service) ruptured starting
at the end face (Fig. 4-50) from a grinding crack shown in Fig.
3-5 and confirmed by the fractured face (Fig. 4-51) . The grind -
ing checks were not the entire cause of the failure, but they
supplied the "notch" in the highly stressed area.

References
1. H. J. Grover, S. A. Gordon, L. R. Jackson, "Fatigue of Metals and
Structures," Battelle Memorial Institute, prepared for Depart-
ment of Navy, Bureau of Aeronautics, NAVAER 00-25-534,1954,
P 15-19.
2. Charles Lipson and L. V. Cowell. Handbook of Mechani cal Wear,
Ann Arbor: Universit y of Michigan Press , 1961 , p 135.
3. D. L. Seager, ASLE, Separation of Gear Teeth, in Approach and
Recess, and the Likelihood of Corner Contact, in S ource Book on
Gear Design. Technology and Performance, Ameri can Society for
Metals , 1980. p 25 -29.
4. Appendix IX of Gl ea son Works Publication SD4052A, Ma y 1966,
in Source Book on Gear Design, Technology and Performan ce,
Ameri can Society for Metals, 1980. p 366-411.
5. Harry C. Rogers. Adiabatic Plastic Deformation . in Annual Re-
view of Mat erials S cience. Vol 9,1979, P 283-311.
6. D. H. Breen. Fundamental Aspects of Gear Strength Require-
ments, in Source Book on Gear Design . T echnology and Perfor-
man ce. American Society for Metals , 1980. p 63-65.
7. Charles Lipson. Why Ma chine Parts Fail. The Penton Publishing
Company . Cleveland . Ohio, 1951, p 25.
8. T. S . Eyrie, An Introduction to Wear (Section 1). in Source Book on
Wear Control T echnology, American Societ y for Metals . 1978.
p 1-10.
po
 CHAPTER 5
Causes of Gear Failure

The ca u se of a ge a r failu re ma y be bl at ant e no ug h to be recog-

nized b y an yone, or it m a y be subtle e no ug h to d ef y recogniti on
by th e most experienced anal yst. At times , a n exa m ine r a lm os t
b elieves tha t the number of ca uses ma y eq ua l th e numb er of
failur es . This i s clos e e noug h to being correct th at eac h failure
must be treat ed indi viduall y, as a compl et e case requiring its
ow n att ention . As stated ea rl ie r , it is ve ry di ffi cult in man y in-
s ta n ce s to separate ca use from mode. A s ing le ca use ca n initiate
sev eral different modes, depending on th e forc es a p plie d . Con-
ve rsely , a specifi c mode of failure ca n be initi a ted b y one of sev-
e ra l ca use s. Again , it must b e emp ha s ized th at a fr a cture pat-
tern of a fail ed ge a r should not be matched w ith a pi cture in this
text and th e co ncl us ion made that th e ca use and effec t a re th e
sam e. The y ma y be , but not n ecessaril y so . U se th e pi ct ur es of
this text a s a g uide; th en , s yst emati call y find th e parts of th e
pu zzl e a nd pi ece them tog eth er .
Caus es are di scussed und er fi ve majo r h eadings a n d sev-
e ra l s ub hea d ing s . Undoubtedly th ere ma y b e mo re , a nd th e
read er should feel fr ee to a d d to this li st. Th e maj or divi si on s
a re as follow s :
• Basic materi al
• Engineerin g
• Manufacturing
• Heat treatment
• Service applicati on
 SYSTEMATIC ANAlYSISOF GEAR FAILURES
130
Basic Material
With full understanding that gears are also manufactured from
nonferrous materials such as copper and aluminum alloys , and
from powdered metal products, both ferrous and nonferrous,
this text will discuss only steel gears .

Steel
Steel , a basic product utilized in bar stock, forgings, and
weldments, has inherent characteristics and defects that carry
over into its final products, and thus to gears. Those character-
istics applicable are discussed;'

Pipe is the shrinkage cavity located in the upper central

portion of an ingot. All escaping gases and slag pass through
this area and some may be entrapped. As the ingot is rolled, the
pipe area is cropped or sheared and discarded. Vestiges of this
pipe sometimes remain in the finished bar product and appear
in the gear or pinion as exceptionally large inclusions or a
grouping of inclusions near the central portion of the rolled or
forged product. Figure 3-16(cJ shows a clearly defined pipe at
the center of a 4-in. square billet to be used for forging. Beyond
the area of the pipe, there may be a condition of "sponginess"
consisting of slight porosity and pin holes, as shown in Fig.
3-16(a) and (b). Unless the rolled product is cropped suffi-
ciently, the condition of the ingot center will be rolled into its
final form as a machinable bar or into a billet for the forging
manufacturer. The steel mills check for this condition by
macroetching specific samples. Likewise , the forging com-
panies and the users of machinable bar stock usually make rou-
tine checks for the same purpose : (Macroetching methods are
explained in Chapter 3.J A pinion will certainly retain all of the
central area of the bar, but a gear from a center-punched forging
may have its central core removed. If not, the remaining pipe
material found in a ring-type gear may be in a restricted and
concentrated area of the bore.

Blowholes are gas-produced cavi ties , cylindrical or spheri-

cal in shape, found randomly throughout the ingot. They are
 CAUSES OF GEAR FAILURE
131
us ua lly small, a n d if not ox id ized , we ld tigh tl y as th e y are
roll ed. How e ver, tho se th at a re s lig h tly o xidized do not we ld
together a n d remain as flak es. Th e presen ce of oxyge n a nd h y-
drogen in th e s tee l-ma king proc e ss is th e main c ulp ri t in th e
fo rmation of flaking ; so vac uum dega ssing is now us ed ex te n-
s ive ly .

Segregation is a de pa r t ure fro m th e average che mica l co m-

pos i tio n. Ge nerally the rna terial ric hes t in the a lloying e le me n ts
w ill b e in the zo ne of la st cooli ng, th e ingot ce nter. T he a mo unt
of seg rega tio n, o r che mica l differen ce , is ve ry se ldo m of mu ch
co n se q uence, exce p t th at it ma y influ en ce heat tr eatmen t. As a n
exa m p le , in a ca se of ex tre me se gregati on, th ere ma y be a 0 .04%
difference in a ve rage ca r bon con te n t from th e a rea outsid e th e
in got to th e ce n tra l area. Th e central a re a co n tai n s th e hi gh er
a mo un t of ca r b on . Th is differen ce in ca r b on co n te n t is main-
tained in th e same proporti on , ev en in th e roll ed bar stock to be
used later as forgin g s or as ma chined pa rt s. An y gear manufa c-
t ure d fr om th is heat of s tee l has a ma rk ed differen ce of ha rd en-
a bi li ty from th e o uter materi al to th e inne r ma te ri a l.

Ingotism is th e remaining evid en ce of th e co lum na r s tr uc -

ture of th e in got as it has b een roll ed into us able bar s toc k s izes.
A mi crostru ctural type of se gregation , it s lig h tly a ffects tooth
charact e ri sti cs du ring heat treatment in ex tre me inst ances.

Nonmetallic inclusions are c on ta ine d in a ll st eel in gots a nd

thu s , in all steel produ ct s . The y co ns is t of oxides and s ulfid es in
va rio us co m b ina tio ns a n d a re d eri ved chie fly fro m th e oxid iz-
in g reacti ons of the refining processes, and from th e de oxid izing
materials added to th e molten steel. Some of th e large inclusions
ma y be silicates or aluminates resulting from the erosion of the
ladle or other refra ctories during pouring. This latter type of
inclusi on (illustrated in Fig. 3-3) is usuall y an isolated incident ,
random in its positi on, and not co ns is te n t throughout other
parts of th e same lot.
In the allo y steels us ed fo r hea vy dut y ge a ri ng, most non-
metallic inclusions are very small. we ll sc attered, and random.
The y ca use n o problem a t all . unless th ey occur at p oints of cri t-
ical stress. Figure 4-11 shows a subsurface crack originating
-
SYSTEMATIC ANALYSIS OF GEAR FAILURES
132

Fig. 5-1. Helical gear tooth, I X. Pitting/spalling mode

originating at a pit caused by subsurface fatigue around a
small nonmetallic inclusion. Only one tooth affected. An
isolated, random case.

from an inclusion at the pitchline in the area of maximum shear

stress, with continued progression outward in all directions
toward the surface to form a pit.
Observe Fig. 5-1 and compare with Fig . 4 -23. At first glance
the fract ures may look the same. In fact, this is graphic evidence
that two failures that look alike are not necessarily alike. Both
are pitting/spalling starting low on the tooth profile and pro-
gressing over the top of the tooth; but here the similarity stops,
because as the one failure (Fig. 4-23) originates from a definite
line of pitting low on the profile, the other (Fig. 5-1) originates
from one pit below the pitchline. Also, every tooth in the gear in
Fig. 4-23 has been affected by tooth tip interference, whereas
only one tooth of the gear in Fig . 5-1 has been affected by one
nonmetallic inclusion at a critical stress point, causing a reac-
tion identical to that shown in Fig . 4-11.
Nonmetallic inclusions at a crack initiation point are uni-
versally found in a fatigue mode of failure. They must be in the
path of high stress concentration, and they must influence that
specific area to be the point of least resistance to stress fatigue.
(Refer to Fig. 4-7 and the related text discussion.) Note the sub-
surface circle of striations around the inclusion. This encircling
crack progressed rather slowly until it broke through the case
to the surface, at which time it rapidly followed the normal
 CAUSES OF GEAR FAILURE
133

Fig. 5-2. Top corner of a spur gear tooth, lOX. A "bull's-

eye" fatigue fracture centering around a nonmetallic
inclusion at the case/core interface. Only one tooth
spalled. An isolated, random case.
tooth bending fatigue pattern. An example of the effect of a re-
motely placed inclusion is illustrated in Fig. 5-2. The inclusion is
located near the top corner of one spur pinion tooth at the case/
core interface. It would never have been instrumental in this
tooth failure if the applied load had not shifted from the central
area of the tooth to the high outside edge of the tooth profile. At
this position, the inclusion is at the applied shear plane and be-
comes the nucleus for the crack propagation. The fracture starts
slowly with a "bull's-e ye" pattern, progresses subsurface for a
period of time, and finally, drops off as a spall. The cause of
failure was the inclusion since the strength of the tooth was
sufficient to function under the applied loads, even though they
had shifted posi tion.
Figure 5 -3 shows an example of a shaft failure in torsional
fatigue along the longitudinal shear plane, with its origin cen-
tered along an elongated oxide stringer [inclusion] in the core
near the case/core interface.
Note: Many fractures show one or more inclusions some-
where on their surfaces. All fractures cannot be caused by an
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
134

Fig. 5 -3. Torsion fatigue in the longitudinal shear plane

centering along an elongated inclusion below the easel
core transition zone.
inclusion, even if it is close to the origin (see Fig . 5-29a). For a n
inclusion to be the ca use, it must be evident that the fracture did
indeed originate at the inclusion . Sometimes extensive work
must be done to establish definite proof (refer to Fig. 3-33).

Flow lines occur in all rolled products; they are the elonga -
tion of the crystalline structure along the principal dire ction in
which movement of the material has taken place 2 - a micro-
structural type of segregation through mechanical means. Flow
lines in the bar stock are utilized by the forging industry to
strengthen highl y stre ssed areas. For instance , Fig . 5-4 illus-
trates how the forger has forced the natural flow lines to follow
the contoured pattern of a pinion forging blank , thus increasing
the bending strength of the pinion head radius .
In the low-alloy steels , the condition of flow lines is more
prominent and is usually termed "ba nd ing." referring to the ob-
served alternating layers of ferrite and pearlite. Two conditions
are noted by these bands: nonmetallic inclusions generally fol-
low within the ferrite band (Fig . 5-5 and Fig . 3-13); and the pearl-
ite band will generally respond to martensitic transformation
during heat treatment more easily than will the ferrite band.
This is probably caused by a difference of hardenability within
the bands, the result of which is not erased by the heat treat
 CAUSES OF GEAR FAILURE
135

Fig. 5-4. Normal hot rolled steel bar flow lines have been
pressed into a forging-die cavity to strengthen the ulti-
mate product.

Fig. 5-5. Microstructural banding shows alternating lay-

ers of ferrite and pearlite. Nonmetallic inclusions tend to
follow the ferrite bands.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
136

Fig. 5-6. Banding influences the diffusivity of carbon dur-

ing carburizing. Note the resulting banded areas of re-
tained austenite (white structure) .

op eration but ma y also ex te nd into th e case (Fi g. 3-31]. Ev en in

hi gh-allo y steel s, the diffu sion of carbon int o th e s urfa ce b y
ca r b ur iza tio n is a ffe c te d b y b anding. Figu r e 5-6 illustrat es the
bands of a us teni te in a matrix of martensit e surround ed by a
cementite net w ork .

S eams are formed during the rolling or drawing operati on

of bar st ock and a re a s ur face con d ition . The y appea r as open
cra ck s and ma y not be v ery d eep on th e rolled st ock . but the
fo rging op eration ma y op en th em into d eep c rac ks . Also , the
op en s eam s are found on eac h surfa ce of th e forged part regard-
les s of h ow man y redu ced s te p s m a y ha v e been tak en.

Forgings
A s th e d ef ects and cha ra c te ris ti cs of basi c st eel products
affect the forging o pe ra tio n, th e forging d ef ect s and cha rac te r-
istics affect all subse quent op erations. Again, th ere is repeti-
ti on, but repetiti on w it h refinem ent. During h ot forging op era-
tions , it is com mo n for s ur face defects to occur.>Larger def ects
 CAUSES OF GEAR FAILURE
137

Fig. 5-7. Forging lap around a forged hub. Crack followed

the normal flow-line pattern and retained oxidized scale
along its surface.

are visible and can be removed at the forge source; but small
imperfections may not be visible, so most of these imperfec-
tions are removed during subsequent ma chining operations.

Flow lines. The flow lines and banding discussed above

have been reoriented to follow the direction of the material into
the forging die . The forging press is designed to direct this flow
evenly and consistently throughout the volume of the die [Fig.
3-10,3 -11, and 5-4 J. When this is not the case (Fig. 3-12J nonuni-
form gea r tooth characteristics ma y occur during subsequent
heat treatment.

Forging laps are formed at or near the surface of a forged

part when an overlapping piece of material is pressed back into
the part , yet is separa ted by an oxidized film or scale (Fig. 3-15 J.
At times, a hardening quench crack can be confused with a forg-
ing lap; but a microscopic examination reveals the difference,
because the crack formed by a forging lap will show a flow-line
pattern and will have a scaled surface [Fig . 5-7) . A special type
of forging lap occurs when the cooled corners of a billet are
pressed into the gear blank . It is easily recognized by the occur-
rence of one to four quarter-moon-shaped laps, each at a
quadrant.

Internal rupture may be caused by uneven heating of the 'I

billet that allows a differential stress during deforming. Gener-
I
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
138
all y such ruptures are formed by an op ening of residual internal
cracks and flakes from the bar product. Various types are called
bursts , mechanical pipe, clinks , or fire cracks.

Shear tears are evident when the material has been torn
apart during the shearing operation. This occurs under only
two circumstances: when a continuous bar is being sheared into
short multiples, the end face of the short bars may tear if not
cleanly sheared ; or when a cent er punch is shearing the cen tra l
plug from a forged blank, it may tear the inside bore-face of the
blank.

Overheating or burning during the heating operation of the

billet for forging will most frequently occur in a localized area
due to flame impingement, poor furnace design, or poor opera-
tion . This causes large-grain growth structures in the steel and
ultimately instills oxides, voids, segregation, porosity, or cracks
within the grain boundaries. When burning is suspected, it is
easily detected by magnetic-particle inspection; however, when
it is unsuspected, the discovery can be as dramatic as that
shown in Fig. 5-8.

Castings
The castings associated most with or having an influence
on gears are the housings used in the gear box assembly. This
may be a transmission box, a differential carrier. a wheel reduc-
tion unit, or a speed reducer assembly. The purpose of the cast-
ing is to house the gear assembly and the lubricant.
The first casting characteristic to note is the accuracy of the
alignment of the locating holes in the housing. Misalignment
within the assembly housing causes many problems for both
the gears and the bearings. A misaligned gear tends to have a
load pattern shifted toward one end of the teeth, but the pattern
will be uniform and consistent on all teeth. The shaft will be in a
continuous state of bending in one spatial direction. but with
rotational bending around its own surface.
The second characteristic to note is the rigidity of the hous-
 CAUSES OF GEAR FAILURE
139

Fig. 5-8. Burned forging. A localized area of exceptionally

large-grain structure with oxides, voids, segrega tion, and
cracks within the grain boundaries.

ing assembly. Lack of proper stability leads to deflections that

cause man y unusual problems for the gears. If cyclic loads are
applied, a cy clic problem ma y result; if a random peak load is
applied, a random type of failure may result. Lack of rigidity
invariably increases the possibility of active vibrations.
The third characteristic to note is the material soundness of
th e casting. Shrinkage cavities, blow holes, and porosity gener-
ally are very detrimental, not only to the strength, but to the
ability to retain lubricant. Leakage of oil from a cast housing is
not an uncommon occurrence, and lack of lubricant in the gear
train is a ca ta s trop he. [See Fig. 2-2 and 2-4.)
Before leaving the subject of basic material, it must be em -
phasized that relati vely few failur es in the field occur due to the
fa ctors list ed in these three categ ories. The reason for their
scarcity in field failures of the final product is the extensive
quality inspection procedures carried out by the manufacturers
of the raw products . The steel mills have excellent inspection
facilities, as do the forging and casting companies. The reputa-
ble gear manufacturer also has sophisticated testing and in-
spection equipment and personnel capable of locating most of
the basic material discrepancies. An y of these discrepancies ac-
tually causing a failure in the field is usually an isolated case.
..
SYSTEMATIC ANALYSIS OF GEAR FAILURES
140
Engineering
Engineering of gears is not only a matter of design. The entire
finished gear is accomplished by the mechanical (design] engi-
neer, with close cooperation from the materials engineer, who
in turn must consult the metallurgical engineer, who must work
closely with the industrial engineer. This places heavy respon-
sibility on several persons, with the ultimate goal of a good
product by the integration of design, material selection, heat
trea t specifica tions, and allowable tolerances for final grinding
or finished sizes.

Design
The designer's world is uniq ue, one that staggers the irnag-
ina tion with the possibilities of achievement. The following
words were written many years ago by Ken Lane of Lynn,
Massachusetts:
The designer ben t across his board,
Wonderful things in his head were stored.
And he said as he rubbed his throbbing bean,
"How can I make this thing tough to machine?
If this part here were only straigh t
I'm sure the thing would work first rate.
But 't would be so easy to turn and bore
It never would make the machinists sore.
I better put in a right angle there
Then watch those babies tear their hair.
Now I'll put the holes that hold the cap
Way down in here where they're hard to tap.
Now this piece won't work, I'll bet a buck,
For it can't be held in a shoe or chuck.
It can't be drilled or it can't be ground
In fact, the design is exceedingly sound."
He looked again and cried-"At last-
Success is mine , it can't even be cast!"!

Although the days of purely empirical design appear, for

the most part, to be over, some design characteristics remain
that should be noted.
 CAUSES OF GEAR FAILURE
141

Fig. 5-9 . Spur idler gear, I X. Fatigue crack origin in the

bore at each of the six oil holes.

One of th e s im p le s t gears to d esi gn is a s p ur gea r w ith a

round bor e, rotat in g o n a gro un d s ha ft , a n d op erating a s a n idler
between th e power input a nd th e po w er ou tp u t gea rs . These
id ler gears are commo n ly fo und in pl ane ta r y-typ e s peed reduc -
ers a n d w hee l redu cti on asse m bli es . T he co m p lex s tres s pa t-
terns for th is type of gea r ing [a s di sc usse d in Chapt er 1, a nd
illust ra ted in Fig .1-20J ca use th e bore to be hi ghly vulne ra b le to
fa tigue. Con sequentl y, a ny oil h ol es ex te nd ing fro m th e b ore
outward a ct as s tress ra ise rs [Fi g. 5-9J . Thi s is n ot always the
case , h ow e ver , s ince occasio na lly th e fat igue ori gi na tes at the
oil hole en d th at terminates at th e roo t s of th e teeth [Fig . 5-10 J. It
ha s alwa ys b een a qu esti on whether this t yp e of oil h ol e is as
b en efi cial a s it is d et riment al. It see ms a s th ou gh ce n tr if uga l
force s ho uld a lwa ys throw th e oil o ut a nd away from the ce ntra l
b ore area . The s p ider-a rm s tra ig h t beve l pini on s in a d iff eren -
ti al undergo th e same typ e of a b u s ive s tre s s a s do th e spur gea rs
d iscus sed ab o ve. Obs er vati on of a large numb er of failu res
point s to fati gue from th e b or e a t a n oil h ol e a s th e ca us e in
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
142

Fig. 5-10. Spur idler gear, 2.7X. Fatigue crack origin in

tooth root at oil holes. Final fracture was rapid.

nearly every instance. The remaining causes would be fatigue

(rapid) at the toe end of the tooth root, due to the absence of
sufficient material between the root and the bore . Take a look at
Fig. 5-11. Looking at a case as hopeless as this differential, the
only response is to attempt to avoid similar catastrophes in fu-
ture designs.
Spiral bevel drive gears have many methods of being fas-
tened to the driven members. They may be bolted through an
inner flange or fastened from the back face by a cap screw, to
name two. In either case, there is the danger of a designed stress
raiser. Figure 5-12 is an illustration of two weaknesses: the spi-
ral tooth cutters were allowed to cut directly into the bolt hole ,
forming a sharp acute angle at a highly stressed area; and the
cantilever effect of the design caused the gear to bend away
from the bolts under heavy loads, adding more stress to the
same bolted areas . A back-face cap screw hole may actually be
designed too closely to a tooth root'(Fig . 4-8), causing a fatigue
failure to originate at the apex of the drilled hole.
Keyways in the bore of pinions and gears, as well as along
pinion shanks and shafts, are very high-risk stress raisers (see
Fig. 4-33). Any sharp corner where there is a change of dimen-
sion should be closely observed as a designed stress point.
Designers and field engineers are generally aware of the
dangers of designed stress raisers in shafts , gears , bolts, and
 CAUSES OF GEAR FAILURE
143

Fig. 5-11. Differential internal spider and pinions. Fatigue

cracking origin at toe end of the root to bore section,
which is rather thin.

Crock (I of 3)

---.- - 'D
7.0 0 diorn - i6 ,

Fig. 5-12. Carburized spiral bevel gear fractured from fa-

tigue originating at the acute angle intersection of the root
fillet to the bolt hule,s
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
144

CD D=J [tJ
-- - __. . - - =_:::~- . ~
.-- .-- - --- -< - =:.o:.~-=- DtJ
c:o~~
==:::;: - - - "'-- ~

s
"
(I) (2) (3) (4)

~~ T C

(11) (10)
CLJ
(b) T C

(II)
.o=J
ee) T C

(12)

Fig. 5-13. Stress distribution at changes of sectlon,s

associated parts , but a reminder of the stress distribution lines
may be helpful." Figure 5-13 shows many of the possibiliti es .
Sharp corners are especially susceptible to fatigue unless modi-
fied by secondar y notches or nearb y corners. For instance, item
3 in Fig. 5-13 shows a high concentration of stress at the corner
radius, whereas in item 4, a modified groove away from the
shoulder reduced the stress concentration at the sam e radius .
Item 7, with two close radii , shows much less conc entration of
s tress than do the same radii farther apart, as in item 6. A notch
[i tern g) has a very high st ress concen tra tion , ye t the notches at
the base of threads [item 10) show ver y little concentration.
There is one exc eption, however-the last thread maintains the
highest amount of stress concentration in a line of threads.
Failures originating from d esigned stress raisers are generally
of the fatigue mode.
A very subtl e design characteristic is discussed in Chapter
4 , in the text illustrated in Fig. 4-16 . This very narrow yet hi gh
 CAUSES O F GEAR FAILURE
145

Fig. 5-14. Spiral bevel tooth , 2X. Pitting and s pa lli ng du e

to rolling contact fati gue in a concentrated area (see Fig.
4-16) as a de signed failure.

loa d conce ntratio n ba nd, w hi ch ca n eas ily be ove rlook ed by the

d esi gn eng inee r, is th e s ite of ro lling co ntac t fa tigu e , as s how n
in Fig. 5-14 , and is a desi gned failu re.

Material Selection
Materia l se lect io n can be acco mp lished only w hen th e ac -
tua l loadi ng co nditions are know n . Ofte n it occurs that loa d s
applied to a set of gears far exceed the des igned load charact er-
is ti cs of those gears . T he refore , th e ma te ria l a nd heat treat me n t
(or eve n th e de s ig n) wo uld ha ve bee n co r re ct if th e fie ld opera -
tio n were what the desig ne rs thought it would be .
In case of a n over load failu re, che ck firs t the specifica tio ns
to determine if they h a ve been m e t; but exercise cautio n: th e
specifications were a rrived a t arbitrarily , based on ass umed
kn owl ed ge . The specifications ma y not ha ve bee n ac ceptable.
Check to determ in e if the loa d s actuall y applied are no rm al for
th e ope ra tio n in que s ti on or if they ha ve bee n ab no rma lly o ut of
line . If the part has been und erdesig ned, wrong in for m a tion
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
146
may have been given to the engineers by the customer, or wrong
assumptions made by the engineers. Needless to say, the re-
verse is also true. A part may be grossly overdesigned and a
high-priced alloy steel used unnecessarily . This generally im-
plies that a gear may have been failing due to an obscure pri-
mary cause. If the primary cause had been found and corrected.
the gear would not have failed. It is the same as the old story of
losing a horse because of a broken horseshoe nail. A stronger
horse is not the answer. Material selection is important. but the
material need not be any costlier than what is necessary to op-
erate satisfactorily .
Mixed steel is the most common cause of material selection
failure. Steel mills ship the wrong grades , forging companies
pick up and use material from the wrong stack of billets , and
gear manufacturers use forgings from the wrong lot-all unin-
tentional, but careless and costly.
Probably the most discouraging instance, and yet very
common practice, is for a user to repair his own broken gear.
New teeth are built up by welding and then hand -ground to
shape. Usually a piece of steel-any piece of steel-is picked up
and given to a gear shop to cu t teeth. As an example (Fig. 5-15) a
C1019 annealed stock was used to replace an SAE 8620 carbu-
rized alloy steel gear. The teeth did not break out of the re-

Fig. 5-15. A gear made from C1019 carbon steel (annealed)

used to replace an SAE 8620 alloy steel [carhurized].
 CAUSES OF GEAR FAILURE
147
placement gear [as did the teeth of the alloy steel gear); they
s im ply wo re o u t in a ve ry s ho rt time.

Heat Treatment Specifications

T he heat treating ope ra tio ns a re ge ne ra lly th e so le res po n-
s ib ili ty of th e gear manufa c ture r, as lon g a s the end resu lt mee ts
th e me tallu rgi cal re quireme n ts ag reed to by th e c us to mer a nd
th e manufa cturer. Th e heat treat s pecific a tio ns [i.e., th e s peci fi -
cations to th e hea t treat de partme nt) m us t be estab lis hed to
meet tho se requirements . T he es ta b lis he d s pec ifica tio ns a re
based on th e m a teri a l used a nd th e ca pability of the hea t treat -
ing eq uipme n t. If a specificati on to th e heat tr eat dep artment is
in cor rect, it is due of te n to po orly re layed i nfo rma tio n or to poor
ju dgm ent on th e part of th ose making th e s peci fica tio n. Not fol-
low in g th e s pec if ica tio n is not th e fault of th e spec ificat io n.

Grinding Tolerances
Th e ge nera l mean s of ob ta in ing a fi nal dimen si on of a p ar t,
a fte r ca rb uriz ing a nd ha rd ening, is th rou gh grind ing . G ri ndi ng
re moves th e s urfac e of a carburized area, w hic h red uce s the
dept h of a hard en ed case . Th e judgm ent of th e indus tri a l e ng i-
neer con cern in g the allowa bl e a mo un t of grind ing s toc k to p ro-
vide , a nd th e judgmen t of th e metallurgi st to com pe nsa te with
a n a d d i ti on al a mou n t of carburized case, are a very cri tical rna t-
ter . This is t rue es pec ia lly a t a s ho ulder rad i us [as in Fig. 5-16)
where bending fati gue ma y or ig ina te due to ins uffici ent case.

Manufacturing
Afte r th e desi gn s tage , w hen a ll the s peci fi cati on s h a ve bee n
es ta b lis he d, comes th e m a nufa cture of the parts . The qu alit y of
ea ch part depends on the s kill of each pe rson parti cipa ting in its
manufa c ture a nd th e ca pab il it ie s of th e eq uip me n t used to pro-
d uce the part.
If th e read er observes that the inspec tion d ep a rtmen t has
b een left ou t, he is ri ght. Q u a li t y is nev er inspect ed int o a part.
In specti on ge ts th e "k lun ke rs'' out. T h ere is a n e xc ellent pro-
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
148

Fig . 5-16. Shoulder radius next to a bearing diameter.

Carburized case had been removed by grinding, allowing
a fatigue failure in bending due to low strength of surface
material.

Fig. 5 -17. Rapid fatigue at the base thread of a differential

holding bolt (grade 8).
 CAUSES OF GEAR FAILURE
149
gram, however, under the jurisdiction of the inspection [quality
assurance) departmenf that should be utilized to its fullest po-
tential. Called Statistical Process Control, it is a plan to develop
a statistical means of continually monitoring results of a manu-
facturing unit, to recognize when the results are getting out of
control, and to correct the unit before it starts to turn out the
klunkers. This would be a positive approach for inspectors.
People and machines manufacture innumerable high-qual-
ity gears capable of operating for many years. Yet, concerning
field failures, there are several characteristics that point back
to a manufacturing procedure.

Tool Undercutting, Sharp Notches

Although a shoulder radius at a change of section is a de-
signed function [see Fig . 5-13) many fatigue failures at a radius
cannot be attributed to the design. All too often a cutting tool
had not been properly formed to cut the designed radius and as
a result, a sharp corner, a smaller radius, or an undercut was
formed. This is a manufactured stress raiser that may readily
lead to bending fatigue.
Figure 5-17 shows rapid fatigue originating around the
base thread of a differential holding bolt. Figure 4-32 illustrates
bending fatigue that started at a sharp shoulder radius, and Fig.
4-33 shows what can happen at a sharp corner of a keyway.
As a graphic example of machining incapabilities, Fig . 5-18
should be observed. Each spline had been individually cut with
a small vertical milling cutter that was moved from the open
end to the shoulder. In order to "square" the rounded corners , a
hammer and chisel were used as a swage. This resulted in the
squared corners being cracked, and field fatigue failure took
place soon after.
From a visual observation, it would seem that all tooling
cutter marks and all tooling undercuts are detrimental. While
this is not necessarily so, generally it is good manufacturing
I
I

practice to produce a part free from cutter marks, with all radii II
surfaces smooth and continuous .
Perhaps the one gear tooth characteristic that has been held I,
responsible for many gear failures is that of roughing cutter
I
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
150

Fig. 5-18. Each individual spline had been formed with a

vertical milling cutter. The rounded area at the shoulder
was swaged into square corners with a hammer and
chisel. Each corner was precracked, which resulted in a
fatigue failure.

marks in the center of the root fillet. Oddly enough, tooth bend-
ing fatigue failures do not occur at the roughing cutter marks
but at the radius just below the active profile . A photoelastic
study was made to determine the actual effect of the cutter
mark upon the tooth bending strength. Figure 5-19 shows the
photoelastic stress pattern on a tooth with a 0.022-in.-deep cut-
ter mark in the root center being loaded by a tooth with no cutter
mark. The radii of both teeth are the same and were loaded to a
capacity equivalent to 6.5 stress fringes . At the same time, the
small radius of the cutter mark had a stress equivalent of 6.0
fringes. The fringe pattern was the same when the mark cut to a
depth of 0.017 and 0.010 in. In each instance. the cutter mark
radius was retaining a stress concentration amounting to 92% of
what was retained by the active root fillet. However, whenever
a deep cutter mark coincides with the radius, a notch effect is
expected.
 CAUSES OF GEAR FAILURE
151

Fig . 5-19 . Photoelastic s tudy of th e effect of a roughing

cutter mark on tooth bend ing s treng th . The rounded
fillets we re loaded to an eq uiva lent of 6.5 stres s fringes .
The cutter mark (0.022-in . deep ) measured 6.0 st re ss
fr inges at its radius. A s tr ess conc entra tion eq uivale nt to
92% of the applied load.

Tooth Characteristics
Tooth cha ra c te ristics [as dis cussed in Chapt er 1) ha ve a
marked in flue nce on th e concentration of load per unit area,
w hic h ma y be acce n t ua ted if a cha rac te ristic cha nges fro m th e
norm. No te the fo llowi ng examples, usi ng ill ust ra ti ons fro m
previous chapters:
[a) Fig ure 1-12 [c ) ill ust ra tes a part wit h severe ru no u!.
T ra nspose that seve re r uno u t w it h out boar d taper to
a n act ual part , a nd g ross fr ictio na l hea t may occur, as
in Fig. 5-2 0 .
[b] Refer to Fig. 1-16. T h is w a s a gea r that had good lead
a nd no ta per. Howe ver, it was ma ted wit h the pini on
tha t had a seve re ta per , as s how n in Fig. 1-1 5[d ].
 SYSTE MATIC ANALYSIS OF GEAR FAILURES
152

Fig. 5-20. Double helical gear. Severe frictional heat de-

veloped due to runout and taper in the gear. The heated
section had less than zero backlash.

(e) Refer to Fig. 1-1 7. Thi s int ern al ge a r s how s ind icatio ns
of case -crus hing, becau se the rad ius of curvature of the
im ping ing tooth ca used a co nce n tra te d ce nterlo ad, as
s how n in Fig. 1-15(f) .
(d) Looki ng es pecia lly for s urfa ce wear patterns on s pira l
bevel or s tra ig h t be vel teeth , ob se rve very close ly th e
position a nd ma g nit ude of these pa tte rn s. Use Fig. 1-18
exte ns ive ly as a g ui de to w hat ma y be occurri ng.
(e) Wh en th e in volu te patte rn s hows a swe lling of eit her
th e low er profil e edge or th e to p co rner, tooth ti p inte r-
fere nce ma y be caus ing s mall in -li ne pitti ng low on the
ac tive profile. Thi s ma y be a n indica tion of s om eth ing
catastrophi c (as demonstrated b y the cyclone effe ct in
Fig . 4 -23 a nd Fig. 5-21).

Grinding Checks, Burns

Grinding chec ks characteristica lly are a lig ned per pendi cu-
la r to the grinding directio n and are sometimes joined to form
 CAUSES OF GEAR FAILURE
153
networks. They are generally in groupings. are close together.
and are not very deep. Severe grinding may also highly temper
or even completely anneal a portion of the surface. (The tech-
nique of examining for burns is discussed in Chapter 3 in the
section titled "Nital Etching"; the technique for examining
checks is discussed in the section titled "Magnetic-Particle In -
spection" in the same chapter.) Grinding generally leaves a sur-
face in a state of residual tensile stress. If the magnitude of this
residual stress is high. an additional applied stress may cause
external rupture (as in Fig. 4-50 and 4 -51). Profile grinding of
gear teeth is a very critical operation. It has generally been ac -
cepted by the gearing industry that gear tooth surfaces should
be in a state of compressive residual stresses. Grinding will
either reduce the compressive stresses toward zero or reverse
them to a tensile stress. Grinding may also leave soft spots on
the teeth (Fig. 3 -4).

Heat Treatment Changes

Dimensional change that occurs due to the heat treating
process can be expected. and yet can be compensated for in

II
Fig. 5-21. Tooth tip interference causing in-line pitting
low on the active profile. A spalling action has resulted.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
154
prior machining practice or in the heat treating process. Four
factors contribute heavily to the dimensional changes: the heat
trea ting process, the design of the part, the mass of the part, and
the material. The first three factors are consistent from lot to
lot, but the material usually changes with every lot. Each grade
reacts differently, and each heat within a grade may react dif-
ferently. Therefore, changes can be expected with each new
heat of material being processed; and if necessary, sample parts
should precede the production lot.
The most harmful change that occurs is the change of tooth
characteristics. A self-imposed crown is discussed in Chapter
4, in the text illustrated in Fig. 4-6. Especially susceptible are
spiral bevel teeth that may lead to any of the patterns shown by
Fig. 1-18. These problematic changes of tooth characteristics
are not insurmountable, since a few parts, as a sample lot
through the heat treating processes , can be used to determine
the final cutting of the teeth as compensation.

Heat Treatment
Although heat treatment is a manufacturing process , it drasti-
cally changes the chemical and metallurgical characteristics to
the extent that a separate discussion is necessary.

Case Properties
The carburized case of a gear is specified to effecti ve depth,
surface hardness, and (sometimes) percent of carbon at the sur-
face . Unless otherwise stated, this specification is applicable to
the midprofile of the teeth midway from the ends. (See Fig.
3-21 ).
The properties of the case are important both at the midpro-
file and at the root radius of the gear tooth, but for different
reasons. John Halgren of International Harvester Company
made an in-depth study of subsurface stresses and the hardness
required to prevent subsurface failures. This work was pub-
lished in the SAE Journal of March 1954, and is reproduced in
Fig. 5 -22. The curves are designed for a 7 D.P. spur gear, but the
 CAUSES OF GEAR FAILURE
155
100 10 0 ,---- , -- ---,-- - - r ----,- -----,

(/)
(/)

e 80
~" '\ u
80
Based on
57 HRC m in. surfa ce ha rdness
304. 000 psi min. tens ile stre ngt h
1ii
E
:J
E 60
\ \
Shear str ess at pitch line

'{,
V .;,<
OJ
~
o
at point of max . stress
60 I--- - f -- - I -- -t-----t- ----i
·x
ro
\ o
a:

-.
E ui
40 ~~ (/)
40 f----'~+_ ..-t7"--f----'-+--- --1
'0 .... ...
, c
OJ

C
OJ
o r-; ~ ......... - "E
I
ro
Q;
a,
20
/
Bending stress at root f ill et
I I
r--
I
- 20 1--- -!-'Ir--I---3,.-t--f----t- ----i

0.020 0 .0 40 0 .0 60 0 .080 0 .0 20 0 .04 0 0.060 0 .080

Distance below su rface. in. Distance below surface. in.

Fig. 5-22. Subsurface stresses (left) and the hardness gra -
dients necessary to prevent subsurface failure (right);
shown for a 7-pitch spur gear.

a p p lica tio n ca n be made to an y other tooth t yp e. Not e step b y

s tep the interpretation of the illustration:
(a ] Th e shear stress is do m ina nt at the pit chline, is 100%
[or ma ximum] a t 0.0 10 in. from th e s ur face, and is
about 80% at th e surfa ce .
(b] Th e bending stress is d ominant at th e root radius , is
tensil e, is 100% (maxi m um ] at th e surfa ce, and fall s off
rather rapidl y a wa y from the surfa ce.
[c] Based on 304 ,000 psi tensile strength required to pre -
ve n t failure, th e hardness con version is 57 HRC.
(d] Using 5 7 HRC a s a mea surem ent of th e ma ximum
s tre s s point , th e hardness gra d ie n t re q uire me n t of th e
pitchline starts at about 48 HRC at the surface , reaches
57 HRC at 0.010 in ., and passes through the effec ti ve
d epth of 50 HRC at 0 .020 in . from the su rface.
(e] At the root radius, 57 HRC must begin at the surface,
but drops off abruptly to 50 HRC at about 0 .005 in.
Wi th this informa tio n, ca se prop erties ca n be specified as
follow s:
(a] Surface hardness : The hardness must be 58 HRC or
higher.
(b] Cas e depth : Th e effec tiv e depth (to 50 HRC] must be at
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
156
least 0.020 in. from the surface. It must als o be deep
enough to ensure that 58 HRC is continued beyond the
0.010 -in. maximum stress point. A range of 0.010 in.
effective case depth is allowable for this size of a tooth ;
so a case depth of 0.030-0 .040 in . is justified.
[c] The depth specified above is for the pitchline or mid-
profile. An examiner should determine if this will be
sufficient for the root radius. In most cases, gear manu-
facturers expect the root radius case depth to be 60% of
the pitchline case depth or more. Based upon the curves
of Fig. 5-22, one customer will accept a root radius case
depth of 50% of the minimum pitchline requirement. If
this is the case, would a specified root radius depth of
0.015-0.040 in. be acceptable? According to the illus-
trated curve, the minimum effective depth required for
strength is onl y 0.005 in., and th erefore, it is acceptable.
The above dissertation was for a 7 D.P. spur gear operating
under a required strength of 304,000 psi maximum. Table 5-1 is
a listing of general case depth standards for ea ch diametral
pitch. Any published standard for case depth is arbitrary; that
is, based upon common knowledge of normal loading applica-
tions. Modifications of this standard and of any other standard
must be made after considering the design of the part, the mate-
rial used, the load applications, the areas receiving maximum
stres s, and the t ypes of stresses a p plied . In examining a field

Table 5-1. Recommended effective case depth standards (a)

Diamet ra l Ca se de pth st a nda rd s . in . Diam e tra l Ca se dep th st a ndards. in .
pit ch S p ur-he lica l Bevel s pil ch Sp ur-heli cal Bevels

20 .... 0 .015-D.020 (b) 7 0 .035 -0 .04 5 0 .0 2 5 -0.03 5

18 . , .. 0.01 5 -0 .0 20 (b) 6 0 .040-D.050 0.030 -0.040
16 0 .015-0 .0 22 (b) 5 0 .045-D .05 5 0.03 5 -0 .04 5
14 0.018 -0 .0 25 (b) 4 0 .0 50-0.065 0 .04 5 -0 .060
12 0 .020 -0.028 [b ] 3 0 .060 -0.0 75 0.050 -0 .0 65
10 0 .0 25 -0 .03 3 0 .01 5-0 .0 20 2 0.0 7 5-D .09 5 0 .060 -0 .08 0
9 ' " . 0 .0 2 7 -D .0 3 5 0 .0 15 -0 .020 1 0 .09 0 - 0 .11 0 0.0 70 -0 .090
8 . .. . 0 .030-D .040 0 .0 20 -0 .030
tal Th is is on ly a g uide . s ub jec t to modificati on dep ending on desi gn a nd appl icat ion .
[b 1 Cons ult met a ll urgical de pa r tment.
 CAUSES OF GEAR FAILURE
157
failure, an analyst should be as specific as possible regarding
the relationship be tween the cause and the case depth specifica-
tion. He must also be aware of additional factors that may be
involved. For example, a shallow case might not have failed if it
had been within the range of the specification. Possibilities
such as this must not be overlooked.
Properties of the case also include other than those already
specified, such as the microstructure. A fully martensitic case
is the goal for any heat treat. The tensile properties and wearing
ability are very good. An amount of retained austenite that does
not reduce the surface hardness is not detrimental; in fact, it
may be helpful, because it is more ductile than martensite and
will absorb some energy of movement. Enough austenite to re-
duce the surface hardness may cause weakness under a bending
stress, as well as reduce rolling -conta ct resistance. However,
austenite has the ability to transform to martensite under ex-
erted pressures called cold working. This is a favorable charac-
teristic since it generates a harder wear surface while in use;
and, since martensite occupies a larger volume than did the
original austenite, the surface stresses become more compres-
sive. A fine line of differentiation exists at this point because
surface compressive stress is ne cessary to resist tensile bend-
ing or tensile surface fatigue; but a high compressive stress can
also set up conditions favorable to compressive surface fatigue
and pitting.
Cementite (or iron carbides J close to the surface of the case
can range from very fine spheroids widely scattered, to massive
globules, to a heavy network. The scattered spheroids are usu-
ally welcome because they enhance the wearing resistance of
the surface . However, the massive and network types may
cause problems under certain condi tions. For instance, Fig. 5-23
shows a carbide network near the surface that caused an inter-
granular weakness , which failed under load by crumbling. A
bainitic structure at or near th e surface may not cause undue
wear or surface fatigue problems, but it does lower the bending
strength of th e root radius. Decarburizalion of the surface is not
favorable at any time. Black grain oxides along the surface are
present at all times in carburized steel parts and are, for all
practical purposes, ignored.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
158

Fig. 5-23. Carburized surface that crumbled under load

along the carbide network boundaries, 200X.

It is difficult to point to any of the microconstituents of the

carburized case as the primary cause of a failure . Each un-
doubtedly contributes to the timing of a fracture or to the mode
of failure; but when a catastrophic failure takes place. an exam-
iner does not usually look for a microscopic cause.

Core Properties
The core of a gear tooth is the basic material under the car-
burized case that maintains the initial percentage of carbon. It
may have been quenched and tempered. or it may remain as
annealed. depending on the heat treating processes used.
The required properties of the core are few. The core must
be ductile enough to absorb shock loads without fracturing. and
hard enough to be resilient under impact and to withstand the
compressive applied loads without permanent deformation.
The structure may be ferrite and pearlite. as rolled. or annealed;
fine pearlite as heated and slow cooled; or low-carbon marten-
site as quenched and tempered. The ultimate strength of any
core is measured by its hardness (see Table 2 in Chapter 3).
 CAUSES OF GEAR FAILURE
159
Case/Core Combination
Failures in the core of a gear tooth a r e not common ; there -
fore, th e case a nd core s ho uld be co nside red toge ther in a ny
a na lysis . Two m aj or mod e s of fa il ure in gear teeth po in t di-
rect l y to th e ca se/ co re combinatio n .
The case c rush ing illustrated in Fig. 4-40 vividly de picts
th e dep th of the ca r b ur ize d case a nd the s tre ng th of th e co re as
t w o co n tr ib uti ng fa ct ors . T ypi cal result s a re s how n in Fig . 4 -41
a nd 4-4 2.
Int ernal r u p t ure of a carb u rized toot h is a di rec t resu lt of
th e ac cum ula ted res idual te nsi le stresses exceedi ng the stre ngt h
a nd ductility of the material at the case/core inte rface, nea r t he
top of th e to oth or near a to p co rner at one e nd. [S ee Fig. 4-48 a nd
5 -2 4 . J T h is co nd it ion is aggrava te d b y fo ur fa ct or s. Each fa ct or
by itself ma y be be neficial, but , w he n these fa ctors exis t to -
get her, they ca n be det r ime n ta l:

Fig. 5-24. Internal stress rupture. Origin at case/core in-

terface near top of tooth. In extreme cases, the entire tooth
top may pop off.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
160
Deep case depth. Since a martensitic case assumes a larger
volume than the precarburized material, the case gets larger
than the original surface and tends to pull away from the core .

High core hardness. If the core hardenability is low , the

core may be ductile enough to "stretch" with the case and thus
absorb the energy of the expanding case. However, a high-hard-
enability core with high strength does not move. Therefore, a
large amount of tensile stress is added to the first amount at the
case/core interface.

Subzero treatment. If austenite were retained within the

case , it would be apt to absorb much of the residual stress be-
cause of its high amount of ductility. However, subzero treat -
ment will transform all of the austenite within the case to mar-
tensite. More volume is required by this transformation . and
only one place remains for the added residual energy to go: on
top of the other stress at the case/core interface .

Shot peening is an operation used specifically to increase

compressive stress at the surface [usually of the root radius),
which enhances resistance to bending fatigue; but, at the same
time, shot peening hits all other surfaces. The tooth surfaces
already high in compression are hammered into greater com-
pression, which is counteract ed by increased tension at the
case/core interface.
Thus with all four factors applied, the tops of the teeth have
been known to explode like popcorn while merely sitting on a
shelf. It is quite likely, however, that the heat treating manage-
ment, recognizing the possibilities suggested above, will cer-
tainly specify a tempering operation; the operation, performed
as quickly as possible after the hardening and the subzero
treatments . may relieve these residual stresses.
Another mode of failure often associated with a ca se/ core
combination is th e longitudinal torsional shear of a shaft. In
general the shear strength of the core is lower than the tensile
strength of the case. As a result, the shear-fatigue crack will
originate at or near the case/core transition zone and may pro-
gress quite far before the case fails in tension [see Fig. 4-28).
 CAUSES OF GEAR FAILURE
161
Hardening
Hardening (as opposed to carburizing and hardening) as-
sumes the use of a medium- or high-carbon plain or alloy steel
that will harden consistently throughout its cross section, and
the strength generated at the surface will be maintained through-
out. Concerning gears, the application may be limited. Tooth
hardness may range from 40 HRC to 60 HRC, but teeth at lower
hardnesses may be subject to wear and those at higher hard-
nesses may be somewhat brittle. Perhaps the dominant
through-hardening cause of failure is quench cracking. Usually
quench cracking occurs in such magnitude that it is noted im-
media tel y, bu t, if it is suspected in a finished part , a metall urgi-
cal examination of a cross section through the crack may be
expedient. A quench crack is distinguishable from a forging lap
because the fracture surface of a quench crack is homogenous
and free from scale or oxides.
Selective hardening [i.e., producing a case by induction
hardening) is a very successful way of utilizing the hardenabil-
ity of a medium-carbon steel to produce core strength , as well as
producing a fairly hard case to withstand wear and maintain
bending strength. The procedure is to uniformly heat, quench,
and temper a forging or a gear blank to a machinable hardness
of 30-35 HRC; finish-machine and cut the teeth; and induction
harden the tooth surface to 48-55 HRC. Two basic methods of
induction hardening gear teeth are accomplished by placing a
circular coil around the major diameter of the gear and heating
all teeth simultaneously, and by traversing one tooth space at a
time . The first method hardens the entire tooth and should
penetra te below the root. The second method hardens only the
tooth and root profile and has the effect of a hardened case (see
Fig . 3-17).
For the induction hardened case to be a cause of failure,
quality must be deficient. For instance, a spur pinion had a
tooth bending fatigue failure originating at the root fillet at the
end face (Fig . 5-25). Etching of the end face revealed the origin
to be at the terminus of the induction pattern. The teeth had
been through hardened but had not been hardened beyond the
tooth root. As a second example , a large spur gear was failing by
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
162

Fig. 5-25. Spur pinion induction hardened by using a sin-

gle round coil. Hardened area terminus at the root fillet
end face set up a stress notch that invited failure by tooth
bending fatigue.

Fig. 5-26. Spur tooth induction hardened by scanning one

tooth space at a lime. Inductor scanning speed did not
allow lime for heat to penetrate the root radius. A stress
notch resulted which invited failure by tooth bending
fatigue .
 CAUSES OF GEAR FAILURE
163

Fig. 5-27. Internal spur gear induction hardened by scan-

ning one tooth space at a time. The inductor was mis-
aligned, allowing for a non hardened addendum on one
tooth profile. This soft area was easily worn.
tooth bending fatigu e at the root radius. Etching showed a sur-
fa ce hardness pattern a long th e en ti re profile , but it had termi-
nated at th e root fillet (Fig . 5-26) .
Some analysts claim that induction hardening is for tooth
w ear onl y and that hard ening of the root is not necessary. Other
analysts differ ver y strongly. Observations verify that most
failures. not only of bending fati gue. but of tooth surface fa -
tigue. are associated with the hard ened zon e terminu s. With
this observation in mind, a university study was conducted to
determine the residu al stress pattern at th e induction hardened
ca se/core transition zone . The result was th e establishment of it
the fact that a high peak of residual tensile stress occurs at the I
hardened case terminus. This cond it ion becomes a self-induced
metallurgical notch , potentially a point of fatigue ori gin.
Improper indu ction hard ening ma y be th e cause for other
problem s, such as excessi ve wear along a nonhardened adden-
dum (Fig. 5 -27), or tooth bending fatigue from a quench crac k at
the open end of a tooth root radius or from inductor burns at the
tooth root (Fig. 5-28).
Selecti ve hardening also incl udes flame hardening. The
problems involved with this technique are more flagrant than
with induction hardening, since flame hardening of teeth is
more difficult to control. Nonuniform hardened areas and over-
heated corners may be the most common types of problems to
occur when flame hardening is utilized.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
164

Fig. 5-2 8. A tooth space inductor set too close to the root
caused severe burning and melted material at the root
centers of several teeth.

Tempering
T empering is th e process of subj ecting h ard en ed ge a rs to a
low -t empe ratu re reh ea t , desig ned to reduce the s urface ha rd-
ness s lig h tly and to reduce th e residual s tresse s ca use d b y
quen ching. T empering of gea rs, a fte r hard ening or s ubze ro
treatment , pl a ces the ca uses mostl y in one of tw o ca tego ries , to o
much or to o little. If a part remains too h ard (to o littl e temp er-
in g), it ma y b e s ub jec t to brittl e frac t ure s . If a pa rt is to o soft
(t oo much temp ering] , it m a y be s u b je c t to surfa ce w earing
problems or tooth bending fatigue.
When ce r ta in a ll oy st eel s a re heat ed to a temperatu re b e-
t wee n 500 OF and 700 OF, th ere is a n increa se in s tre ng th , bu t a
 CAUSES OF GEAR FAILURE
165
very marked decrease in ductility and impact strength. This
embrittlement is called "blue brittleness" because it occurs in
the tempering range that produces a blue color at the surface.
(There is an excellent discussion of embrittlement in the new
Metals Handbook .?] Not all alloy steels are susceptible to this
phenomenon, but those that are should not be subjected to this
temperature range when impact loads are expected. For infor-
mation on types of steels that are susceptible, consult a good
materials handbook on impact strength vs tempering tempera-
ture. "
Two practical aspects of the occurrence of temper brittle-
ness must be considered as the causes of a failure . The first as-
pect is field oriented and has nothing to do with the heat treat-
ment process. It will occur when a part or a portion of a part is
heated to approximately 600 of either by frictional heat or by
close proximity to an area being welded. Whenever an area , re-
gardless of how small, is heated to a very high temperature, a
transition zone between the heated spot and the cold material
actually reaches a tempering temperature of 600 of. In some
steels, this zone is susceptible to impact fracture. The second
aspect definitely relates to the heat treat department and may
be caused by faulty heat treat specifications or faulty temper-
ing. This embrittlement is found only when a portion of a gear
or pinion (namely, the threaded end of a shank] is tempered at
about 1250 of for the purpose of reducing the hardness of the
threaded area. Next to the threaded area, there is usually a
shoulder leading to a larger diameter and a splined section. If
the tempering medium (perhaps molten lead) is not applied
beyond this shoulder, the transition temperature of approxi-
mately 600 of might possibly temper the shoulder radius. For
some alloy steels, this shoulder radius area would now exhibit
low impact strength. This could cause the threaded end to break
off with a brittle type fracture if the thrust or bending load were
great enough .

Miscellaneous Operations

Copperplating is used ex tensi vel y in the ind us try as a s top-

off for areas not to be carburized. If a spot of copperplate were
 SYSTEMATIC ANAlYSIS OF GEAR FAILURES
166

(a)

(b)

[a ) 1.6 X . T he proximity of a lo ng incl usio n th a t wa s no t a fact or in th is

fai lur e. [b ) 3.5X. T he colum na r s truct ure is th e res ul t of ra m pr es sure fro m a
s t ra ig h te n ing p res s . Th e crus hing effec t of th e case in to th e co re form ed a
wea ke ne d case/core in terface w hich beca me the or igin for s hee r fat igu e.
Fig. 5-29. Torsional fatigue of a shaft in longitudinal
shear. Origin was at the case/core interface beneath the
concentrated columnar structure of the case.
 CAUSES OF GEAR FAILURE
167
inad vertently allowed to remain in a cri tical stressed area [such
as the root fillet of a gear tooth], a soft spot exhibiting low fa-
tigue strength would result.

Copper stripping solutions used to remove copper plating

after the heat treat processes can be corrosive if not maintained
properly, or if the parts are left in too long.

Threaded plugs placed in bolt holes before carburizing may

be placed improperly, allowing some type of stress in the area
around the bolt hole.

Straightening of pinion shanks and shafts has several

unique hazards. Surface cracking may be the common attribut-
ing cause; however, Fig. 5-29 illustrates another type of stress
raiser. The fractured area of the shaft shows a longitudinal
shear with its origin below an odd-looking columnar case struc-
ture and very near to a long inclusion. As expected, the custom-
er claimed the inclusion was the cause. However, the real
source of the problem was the columnar case structure caused
by crushing and vertical shearing from the ram of a straighten-
ing press, which determined an excellent starting place at the
weakened case/core interface for torsional shear. Often a bend-
ing fatigue failure of a shank or shaft occurs, with no apparent
reason for its peculiar point of origin. To accomplish a straight-
ening operation, the part must be plastically bent. [See Fig.
5-30a.] Also, upon the release of the bending load, a pattern of
residual tensile stresses is measurable at the top and residual
compressive stresses measurable at the bottom, as illustrated
in Fig. 5-30 [b ].? Therefore, if bending stresses are applied in the
field operation, the top of the straightened area will receive the
maximum allowable stress before any other spot on the shaft,
and bending fatigue will probably start at that area.

Service Application
Dr. Charles Lipson was quoted as saying: "The fact is that the
vast majority of service failures cannot be traced to the metal-
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
168
Load Str ess
T-----.-,"-r>""TJ- -- -

B J....£...<....L.L.L-_

o +

Load rem oved

Str ess

c- i T----"""'~--

B--~~L-..----
o +
Com pression Tensi on

Fig . 5-30. Residual stresses in a shaft due to plastic bend-

ing duri ng a straightening operation. (a) Stress dist ribu-
tion during plastic bending. (b) Residual stress measured
after load remova\. 9

lu rgist. Inste ad , th ey are the resp onsibiliti es of th e design er, th e

installati on man , a nd th e servi ce opera tor."!" Rich ard Ga ydo s of
Repu bli c St eel, in hi s paper " Fai l ures in Hea v y Mac hine ry,"
mad e thi s com me nt: "A hist oric al re view of man y failures in a
s ing le pl ant ove r a s pa n of m an y ye ars illu strat es th e foll o wing
reaso ns: o pe rat io na l, 36 % [overloa d, 28 %J; fa ulty des ig n, 34%
[sh arp radii, 11 %J; shop m alpra cti ce , 17%; a nd met allurgi cal,
13%."
Now, w hy, s ince service a p p lic a tio ns a ppear to be th e g rea t-
es t ca use for gear fa il ures, ha s so -m uc h energy bee n s pe n t ex -
pl aining a ll of th e o the r rea s ons first ? P erhap s Rob ert P . Havi-
lan d ha s the a nswer: " A fa ilure po in t in a per fec t objec t occurs
w he n th e energy s to re d b y a g ive n mech ani sm excee ds so me
cri tica l va l ue ." He th en ex pla in s , "T h is means th e break is going
to occ ur a t the wea kes t lin k." !' Th e disc ussio n up to thi s point
has bee n se t ting th e s tage for a disc ussio n abo ut wea k link s .
An y s ervice a pp lic a ti on failure is going to occur a t th e w eak est
link . It is n eces s ar y fir st to ha ve a n un d er st anding of w ha t
 CAUSES OF GEAR FAILURE
169
caused those areas of weakness. Also, any failure at the weak-
est point within the gear train is not caused by that weakest
point, if the load applied had been greater than the designed
load-carrying capacity of the assembly or of the failed part.
Conversely, if failure at a specific area continues as a pattern
and, if the applied loads are normal for that application, the
area of failure must be strengthened.
The causes of failure relating to service application will be
discussed, not in order of importance or magnitude, but in order
of operational sequence.

Set Matching
Spiral bevel and straight bevel gear and pinion sets are
matched by tooth contact pattern and serialized as sets . Al-
though the tooth contact pattern is generally arbitrated by the
customer and vendor, there is a customarily acceptable no-load
pattern [see Fig. 1-18a]. Any deviation from this pattern may be
caused by misassembly, misalignment, or deflections. The de-
viated patterns [illustrated by Fig .1-18 c through p] can be rec-
ognized in the field by the wear pattern of the gear teeth on both
sides . Another very common cause of a deviated pattern is a
mismatching of sets. Very often a set is returned with different
serial numbers on the pinion and gear. Mismated parts should
never be in operation together.

Assembly, Alignment, Deflection,

and Vibration
In most instances, the characteristics in an assembly that
ca use misalignment and deflections are relati ve to dimension
and strength. However, the engineering of the entire assembly
may have some shortcomings. For example, a spiral bevel drive
set was placed in a differential assembly engineered for a right-
hand drive. The manufacturer had tested and matched the set
for a right-hand load but had never paid attention to the loading
pattern on the reverse side, which is normal procedure for stan-
dard shelf items purchased from parts suppliers. The customer
did not know the importance of this fact and placed two units in
operation on a special double-drive machine he had designed.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
170

Fig. 5-31. Spiral bevel pinion (O.75X) and gear (O.70X)

from a stock item differential manufactured properly for
and adjusted to a right-hand drive. It was installed in a
left-hand drive situation. The reverse-side pattern had
not been corrected for this application.

As it happened. the left-hand unit. driving in the reverse direc-

tion. had a pinion in its assembly that had never been adjusted
to this type of operation. The result is apparent in Fig. 5-31.
A fast-growing industry of renovating and reconditioning
exists in the present economy. In fact, it is possible to purchase
a reconditioned gear box for nearly every type of original box
that has been manufactured. In order to compete with this new
industry. many manufacturers of original equipment have been
forced to renovate and to resell used assemblies. This is a scary
practice and could have disastrous results unless every internal
part has been closely inspected dimensionally. observed by
magnetic-particle examination, and kept meticulously clean.
The correct alignment of parts is very important. A misalign-
ment can cause tooth loading to shift from the central profile
contact to an end contact. Whenever shifting of load pattern oc-
curs through misalignment. it is recognizable because the pat-
terns are consistent. continuous. and uniform. But this is not the
case with shifting contact patterns caused by deflections.
 CAUSES OF GEAR FAILURE
171
These patterns constantly shift back and forth as loads are in-
creased. decrease d . or reversed . In one exam p le (s how n in Fig.
4 -5]. a s ud de n ove rload ca used by a n ex treme wi ndup of a ll in-
ternal parts p laced an immedia te e nd contact loa d on one tooth
of a pinio n. whic h cracked at the root radius.
A m at ch ed gea r a nd pini on se t ha s in s crib ed on a s urface of
eac h part a setti ng or m ou n ting dis tance. in a la ng uage agreed
upon by the ve ndor and the customer . If th is sett i ng ha s n o t
bee n adjus ted p roperly in th e assemb ly. th e norma l loa d co ntac t
area w ill be di spla ced. Fig ure 5-3 2 ill us t ra tes a tooth bending
fa tigue orig ina ting nea re r th e heel e nd . By p laci ng th is pinio n on
a tes te r and observi ng the pattern. it was determined that the
part had bee n pres et 0 .0 20 in . ou t fro m t he gear a nd 0.0 20 in .
b ack tow ard th e h eel . away fr om th e es ta b lis hed se tt i ng
informat io n .
Any tra nsmissio n or power trai n equ ipme nt ha s vibrations.
Ev en the s m oo t hes t s ource of po w er will pul s at e. Roller bear-

Fig. 5-32. Spiral bevel pinion, I.OX. Tooth bending fatigue

toward the heel end caused b y misalignment in the as-
sembly 0.020 in. out and 0.020 in. back from the gear.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
172

Fig. 5-33. A hub spline that permanently fitted into a

wheel gear. Vibrations advancing to chatter, along with
foreign particles. set up a destructive wear condition.

ings vibrate; gear teeth vibrate; rotating surfaces vibrate. Each

part vibrates with its own peculiar amplitude and frequency.
but this usually creates no problem. unless two or more parts
synchronize to form a high noise level or to form a sequence of
extremely high load peaks. Perhaps 85 % of all surface fatigue is
caused by torsional vibrations. This may be a high estimate. but
torsional vibrations [chatter) reduce the fatigue life of a unit
drastically. Figure 5 -33 shows an external splined hub. fitting
permanently into an internal splined wheel gear. disintegrating
rather rapidly due to chatter and foreign material.

Mechanical Damage
Mechanical damage is damage to the gear inflicted by a
force of another object during the manufacture. packing and
shipping , unpacking. or assembly. This may be the same type of
damage as that inflicted by a foreign object. but from a different
context of environment and time. For instance. a cutting tool
had broken and gouged a tooth surface [Fig. 5-34). not causing
the trouble it might have; a bump on the top face of a gear tooth
subsequently caused bidirectional tooth bending fatigue [see
Fig. 4-9 and 4-10); one finished gear, dropping on the ends of the
 CAUSES OF GEAR FAILURE
173
teeth of a second gear (Fig. 5-35), caused impact failure or crack-
ing for subsequent fatigue. Also , many instances of ni cks and
bumps-too numerous not to be noticeable-have set up stress
raisers for several modes of failure.

Lubrication
The causes blamed on lubrication (discussed in Chapter 2J
concerning gea r failures are usually due not to the lubricant it-
self, but to the application of the lubricant or to the external
forces working against the lubricant. The gear failures that can
be attributed directly or indirectly to the lubricant are the result
of incorrect lubricant, a faulty lubrication system, insufficient
or excessive lubricant. or lubricant contamination. A correct
lubricant not only applies the proper surface film characteris-
tics under the applied loads, but readily absorbs and dissipates
heat. An incorrect lubricant may break down rapidly and retain
ver y little lub ricity. A breakdown of lubricit y causes a reaction
similar to that caused by lack of lubrication; namely. a metal -

Fig. 5-34. Sprial bevel gear tooth, 3X, showing surface

gouge due to the breaking of a cutter blade.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
174

Fig. 5-35. Spur gear damaged by the impact of a dropped

gear.

to-metal contact that quickly allows frictional heat to build up

and adhesive wear to result. If a failure appears (as in Fig. 4-15
and 4-27), the type or amount of lubricant should be questioned.
The lubrication system may be inadequate in promoting
rapid heat absorption and dissipation. resulting in high temper-
atures and adhesive wear on the moving parts.
Excessive lubricant is not common, but it tends to build up
internal pressures and temperatures detrimental to the hous-
ing's carrying capability.
Lubricant contamination is a very complex problem. The
subtle contaminants are the gas and liquids absorbed by or dis-
solved into the lubricant. They may change the chemistry
enough to cause a breakdown of the lubricant, or they may set
up chemical cells corrosive to the gear surface. The less subtle
 CAUSES OF GEAR FAILURE
175
co ntamina nt is the solid sa nd a nd dust ty pe th a t is a n a brasive
mat erial using the lubri cant as a ve hic le to tran sp ort th e a b ra d-
a n t a nd as a stora ge fo r a dd it io na l a b ra de d materi al. Th e re-
s ult s a re clearly illust rated in Fig. 5 -36.

Foreign Material
Alth ou gh th e co n ta mina n ts in th e lubri cant a re es se n tially
fore ig n materia ls, they are co ns ide red contami na n ts on ly as
long as they a re sma ll eno ug h to be tra ns ported by the lub ri can t.
Wh en pa rticles become la rge or hea vy e no ug h to set tl e out of the
lubri cant a nd becom e enme s he d w ith in th e gea r teeth or a ny
oth er moving part s of th e assembl y, they are clas sified as for-
eig n material. The failure resulting fr om for ei gn material is usu-
a lly a seco nda ry failure, s ince th e for eign ob ject in vol ved is
ge ne ra lly a b rok en par ti cl e or object res ulting from the pri ma r y
fa il ure. Se veral exa m p les foll ow. A bolt beca me loose a nd

Fig. 5-36. Differential spider arms worn deeply by abra-

sive material as a contaminant in the lubricant.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
176

Fig. 5-37. Spur gear, 3X.Tooth bending fatigue with origin

from a crack at both root radii caused by the enmeshing of
a portion of a needle bearing roller.

Fig. 5-38. A foreign object (top of adjacent tooth) became

enmeshed in spiral gear (O.5X) teeth causing a full section
break. The top of the adjacent tooth also broke by impact
from some object.
 CAUSES OF GEAR FAILURE
177

.. .- . --
. _~,,-

- ... ' _ .., ........-

Fig. 5-39. Spur gear used in a seasonal operation was at
the top of an assembly, out of the oil, and subject to corro-
sion by cold weather condensa tion during the idle season.

backed in to the tee th of a spiral pinion (Fig. 4-39 J; a portion of a

needle bearing roller became enmeshed between the gear teeth
(Fig. 5 -37J, causing a crack at the root radius that continued as
tooth bending fatigue; and an entire gear section was broken by
pressures from an enmeshed tip of an adjacent tooth (Fig. 5-38 J,
which also had been broken by impact from some object.

Corrosion
Corrosion is chemical deterioration of a surface (as shown
in Fig. 5-39J. The corroded surface may not be considered a fail -
ure in a power transmission gear, for example, since it can still
function for its intended purpose, though it may not look good
and may be slightly noisy. However, if it were a pump gear for
liquid transportation, it would certainly be a failure, since it
could not function efficiently for its purpose. The most danger-
ous aspect of a corroded surface is its ability to become a stress
raiser and enhance pitting, spalling, tooth bending fatigue, and
brittle fracturing, all of which can happen without notice and
cause tremendous overall damage .

Continual Overloading
The result of consistency is uniformity and predictability.
Consistent, continual overloading of a gear will result in a uni-
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
178

Fig. 5-40. Spiral bevel pinion, 0.6X. Two teeth out by

tooth bending fatigue, but all teeth are fractured almost
completely through. Continual overload.

form load pattern and the failure is predictable . Overload does

not necessarily mean "beyond the designed characteristics as
engineered for the product." An applied load may be well within
the designed characteristics but may be a severe overload for
the existing conditions of the gear.
Spiral bevel gears are interesting to study. With no load,
the contact pattern is near the toe end of the profile; under full
load, the contact pattern is well centered; and under continual
overload, the contact pattern may shift by deflection toward the
heel and higher on the profile. Buf this is not necessarily the
case. since some spiral designs tend to keep the overload pat-
tern at the central area and tend to go lower on the profile. Fig-
ure 5-40 illustrates an overloaded spiral pinion in which every
tooth is cracked and ready to fall out. Two teeth fell out first,
but the load pattern stayed well centered and deep, perhaps be-
cause the mating gear teeth also deflected enough to compensate
for the load shift. The gear teeth shown in Fig. 3-1 were over-
 CAUSES OF GEAR FAILURE
179
load ed for th e co ndit io n tha t exis ted a nd not for its desi gned
stre ngt h . No gear is desig ned for fu ll load to be app lied at th e
ends of the tee th . Fig ur e 4 -42 ill us tr a tes gear teet h tha t were not
fun ctioning, a lt houg h onl y no rm al load s w er e being a p plie d .
The fa ilure was co ns is te n t a nd uniform. The co nditio n of s ha l-
low case de p th a nd low core hard ness was bei ng ove rloaded .
The shaft fa ilu re illus trated in Fig . 3 -2 w as ca use d b y con-
tin ual overloadi ng . T he fail ure pattern was un ifo rm a nd p redi ct -
a ble. The part w as d esign ed to ha ve with st ood normal load s .
Figure 5-41 shows a s pur gea r tha t had been co nti n ua lly over-
loa ded. The ad hes ive wear pattern is uniform on both sides (not
th e case w it h gea rs r un ning in only one d irecti on ) a nd on a ll
teet h. The loa d s we re hea vy eno ugh to break th rough th e lub ri-
ca ting film and a llow seve re s urface fr ic tio n. O verloading do es

Fig . 5-41 . Spur gear, O.9X. High continual overloads

broke down the lubricant barrier, causing deep adhesive
wear.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
180
not cause only breakage ; it can also cause wear, pitting, and
spalling . A continual overload due to improper design is illus-
trated in Fig. 5-14 .
In general. if the mode of failure is rolling contact fatigue
distinguished by "butterfly wings," continual overloading is
almost certainly the cause .
Fatigue failures originating at sharp corners and undercuts
are caused by continual overloading for the condition existing,
but may not be overloaded for the designed application. An ana-
lyst has to be careful about making an incriminating statement
about a condition. For instance, a shaft was designed with a
keyway and the engineers took that into account and said that
the designed load for continuous operation was X-amount.
However, the part operated continually at a load of 1.25X. It
failed in fatigue as expected, at the keyway, but the failure was
the fault of overload, not of the keyway. Without the keyway,
the shaft could have easily withstood loads of 4X; but that evi-
dence had nothing to do with this application. However, had the
designing engineer ignored the keyway and stated that the de-
signed load was 4X, the keyway failure at an applied load of
1.25X would have been the fault of improper engineering de-
sign. From another angle, the designing engineer calculated that
a certain keyway, if carburized and hardened at all radii the
same as all other surfaces of the shaft, would have an operating
strength of 3X. However, manufacturing decided that dimen-
sions were the most important characteristic; so the shaft was
routed to carburize, slow cool, cut keyway, and reharden. The
closer dimensions were held; but the low hardness of the key-
way radii allowed the shaft to fail in fatigue at 1.25X of load.
The cause of failure was manufacturing misjudgment.
The obviousness of a particular mode of failure does not in
any way change the analyst's method. It should be deliberate
and thorough, and the analyst should be certain of the accuracy
of his results.

Impact Overloading
A term equivalent to impact overloading is "sudden peak
overload." Both imply that a gear may be operating at a consis-
 CAUSES OF GEAR FAILURE
181
te nt ly loaded co ndition w hen , for a brief m om en t of time, the
loa d pea ks o ut co ns iderab ly ove r t he des ig ned ma ximum. The
term "s ho c k load in g" ha s a lso b een us ed in so me re por ts . What
h appen s to the gea rs? So me times nothing. So me tim es a s ma ll
crac k a t a s ing le roo t ra d i us m a y start. And so me tim es t he e n-
tire too th brea ks away .
S om e e xa m p le s of th e e ffe cts o f impact o ver loa d ing are il-
lu st rat ed in Cha p ter 4:
ra j T oo th be nd i ng fatigue wit h origi n at a roo t radius d ue
to sudde n ove r load is c hara c ter ized b y a p pea ri ng to be
a ra ndo m too th a nd no t co ns is te n tl y t he sa me on a ll
tee th. Figure 4 - 1 ma y ha ve bee n s uc h a case a nd Figure
4-5 wa s d efinit el y s o.
(bJ A s ud de n o ve rload ma y qu ickl y s hif t a co n tac t patt ern
of a load e d sp iral gear towa rd the to p of the h eel e n d ,
ca us ing a n im pac t fr a ctu re of t w o adjace nt tee th [Fi g -
ure 4 -36 J.
[ e] Teeth that ma y ha ve start ed b y fati gue du e to ot her
causes m a y sud de nly be bro ke n o u t by imp a ct (Figure
4 -6 J.
[d] Impa c t fa il u re d ue to int ru s ion o f a for ei gn ob jec t ma y
ca us e a co ns is te n t d ama ge [F igure 4 -39 J.
(e] A s ud de n st op m a y ca use a b rittl e fr a ctu re , foll owed b y
s hea r [Fi gure 4-38J ; or, if th e st op is vio le n t eno ug h ,
sev era l teeth m a y be s heared si mu lta neo us ly [F igure
4 -3 7 ].
(tl In some ins ta nces, toot h crus hi ng ca n be ca used b y im-
pac t overload [F igure 4 -41] .
[g J Torsio na l ten si le failure ma y be ev ide n t [Fig ure 4 -2 9 J
as sud de n ove rloa d .
[h) T o r si on al t ran s vers e s hear [Figure 4 -43 J is a d efinit e
case of s hoc k loading.

Bearing Failure
Ju st as m isalign m ent a n d d efl ec tion s ca n s hif t load con tac t
areas on ge a r tee th . so ca n bea ri ng fail ures . In fa ct , a fa iled bear-
in g ac t ua lly m is a lign s the e n tire associa ted gear train. Th e
m od es a nd ca use s of bearing failu r es a re ju st as man y a nd va r-
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
182
ied as for gear failures; but all that will be said here is that a
failing bearing will affect the life of the gears. Conversely, the
bearing analyst will admit that a failing gear will affect the life
of the bearings. It is in the best interest of all concerned that
both analysts recognize the interrelationship, not only of gears
and bearings , but of all moving and nonmoving components of
an assembly.

Maintenance
Maintenance personnel, equipment, and procedures can
make or break a company . Not only must good replacement
parts be used, but care must be taken that all used parts are
cleaned and thoroughly inspected before being placed back into
service . No gear should be reused unless a magnetic-particle
inspection method has determined a freedom from crack initia-
tion. No mated gear or pinion should be run mated with a ran-
dom part. Again, cleanliness and correct alignment must be
maintained during the reassembly of repaired items. Many us-
ers of equipment have programs of "preventive maintenance,"
in various degrees of sophistication, from well-thought-out
systematic checks of all components, to perhaps yearly hit-or-
miss checkups of only the major operating parts. Other equip-
ment users wait for parts to break down before fixing them.
Probably the most common and most abusive evidence of lack
of proper maintenance is insufficient or improper lubrication.

operator Error
The operators of equipment are blamed for more than their
share of failures; but it is often the result of operational error
that failures occur. In one situation, the operator of an airport
snow-removal vehicle was pushing the large bucket at 25 miles
per hour through the snow and slamming into the large bank of
snow at the end of the field without slowing down. Each time,
the rear end of the vehicle lifted up about six feet and slammed
back down on the concrete. This procedure was not at all neces-
sary, nor was the equipment (especially the gears] designed to
take this abuse time after time. (The pinion of Fig. 5-40 and the
 CAUSES OF GEAR FAILURE
183
acco un t in Cha p te r 6- the s tudy of a s pi ra l dri ve se t- we re
ac t ually tak en from thi s m od el fro nt-e nd loa d e r. ) Rea s on s s uc h
as th is pro mpt the repetition of a previous stateme nt : Wh en a
ca tas t roph ic failu re occ urs, d on 't look for a microscopi c ca u se .
Find the cow boy at th e w hee l o r th e one th at thre w th e monk ey
wre nc h int o the works .

Field Application
It is possibl e th at field a p plica tio n is th e most imp ortant
fa c tor or, as so me be lieve, the cause of the grea test n um be r of
gea r fail ure s. It is cer ta in ly tr ue th at the fail ures occ ur dur ing
se rv ice ope ra tio ns; but unles s it ca n be prov en th at th e a p plica -
tio n is defi nite ly overs tressi ng th e gears , the a pp licat io n may
n ot be the primary cause. What e ve r the mode of failure, the
s tre ng th a t th e or igi n mu st be det e rmined, to verify w he the r th e
fi eld appli cat ion w a s beyond th e sc ope of th e d esi gn ed ca pa b il i-
ties. If th e a p p lie d loa d th a t origi na ted the fa ilure was ran d om
a n d an is olat ed case, th ere is not mu ch tha t sh oul d be d on e. If,
on th e o the r h and, th e field appli cation is to b e co ns iste n t w i th
th e loads ca us ing fa ilure , b y a ll mean s , th e app lica tio n for th e
asse mb ly be ing used s ho u ld be cha nged, or the asse m b ly rede-
s igne d to acco m m oda te reali sti c load s .
One more im por ta n t s ub jec t m us t b e d iscu ss ed ; i.e. , a
cha nge of a pp lica tio n in th e field . Ag ain , a n illu stra ti on: The
a u th or and hi s fa the r bou ght a H z-ton fl atbed truck for use to
haul wood, h a y , li ves tock , a nd w ha t-have-yo u. The y made dif-
ferent ra ck s a nd s ides to acco m mo da te the typ e of load s . Th e
father w as a lso a mining man; s o the y des igned a dumping bed
for hauling o re, put on tw o se ts of ove rload s pri ngs, a nd h aul ed
bet w een 3 a nd 4 to n s pe r load. T hey c ha nged the fie ld a p p lica -
ti on but did n ot ch ange th e differential or th e two dri ve sh afts.
An yone ca n g ues s th e re sults , w h ic h were rather co s t ly to re-
pai r. The o ther a spect of cha nge of a p p lica tio n co mes abo ut
w he n th e manufac turer of th e orig ina l equ ip men t d ecides that
g rea te r horsep ower is req uire d to me et th e cha lle nge of compe-
ti ti on. It is s urp ris ing ho w man y times inc rea se d h orsep owe r,
as we ll as ove rl oad spring s, a re the on ly items th at seem t o
make a differen ce in a job a pp lic a tion.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
184
References
1 . Ha rold E. McG ann on . Th e Mak ing . S ha p ing . and Treatin g of
St eel. 9th ed ., U n i te d Stat es St ee l. 197 1, p 588-5 89 .
2 . A. D. Merrim an. A Concise Enc ycl oped ia of Metall urgy. Ne w
York: Ame rican El se vi er Publishing Co. , In c.. 19 65 . p 284 .
3. Manua l of Open Die Forgin gs. Op en Die Forging In d us tr y , New
York , Ne w York . 1949 . p 13 .
4 . Reprint ed with perm ission fr om Am e rican S ocie ty for M et als.
5. Metal s Han d boo k 8th ed .. Vol 10 . Failure Anal y sis a nd Pre ven-
tion. American So ci et y for Met a ls. 197 5. p 52 0.
6. So u rce Book in Fa il ure A nalys is. Am eri can So ci et y for M et al s .
1974 . p 129.
7. "Embrittlement of Steels." Metal s Handb ook Ninth Editi on. Vol t ,
Propertie s and S ele ctiqn: Irons a n d Steels . Americ an Soci et y for
Metals , 19 78 , p 683-688.
8 . Metal s Prop erti e s. Handbook of Th e Am eri can S oci et y of M e-
chani ca l En gineers . 1st ed .. New York: M cGraw -Hil l. 19 54 .
9 . T . J. Dol an. Residu al Stress. Strain Hard ening a nd Fatigu e. in In -
ternal St re s s es and Fatigu e in Metal s. S ymposium on Int ern al
Stresses and Fati gue in M eta ls [held in Detroit a n d Warren . Mich -
igan. 1958] . Els evier Pub lishing Co ., New York. 1959, p 304-305 .
10. Charles Lipson . Wh y Ma chin e Part s Fail. Th e Penton Publish ing
Co mpan y . Cleve land . Ohio. 1951 . p 5 .
11 . Robert P. H a vil and. Vall e y Forge Spac e T echno log y Center. and
G enera l Ele ctric Compa n y , Engin eerin g Reliabilit y and Lon g Life
Design . P rin ce ton: D. Van N os trand Co.. In c.. 1964. p 6-13 .
The Final Analysis
CHAPTER 6
No failure ex a m ina tion has been completed until an e va lua ti on
of th e resul ts is made; an r' it is only w he n th e failur e mech anism
is und erst ood that effecti v e correc tiv e measures ca n be devised.
The preceding cha p te rs ha ve att empt ed to give th e reader
a n understanding of failure mechanisms in a logi cal sequenc e.
The purpose of this chapt er is to e xpla in how to take th e known
fa cts of a sp ecific failure, pl a ce them in a s ystematic context to
a rrive at a logical conclusion. a nd point to a correc tive me asure.
Becaus e, understandabl y, the hundreds of a vailable ex a mina -
tions ca nn ot be detail ed here, a chosen few will be present ed for
th e purpos e of showing how a systemati c approa ch to failur e
exa mina tion actually works.

Broken Shaft
One end of a broken a xle shaft w as re turn ed to th e manufac-
turer and subsequently submitted for analysis .

Background Information
Field appli cation: The a xle shaft was used in a heav y-dut y
op en -pit mining tru ck hauling overburden. The shaft had faile d
afte r op erating for 27 ,000 hours . Previous failures had resulted
from longitudinal shear, but this had not ; the c us tome r thought
the material was defective.
 SYSTEMATIC ANAlYSISOF GEAR FAILURES
186
Visual Examination
Th e returned shaft section did not have the identification
markings [they were on the other end). The part number was
listed in the accompanying report. Photographs were taken of
the fracture surface, centering on the point of origin [Fig. 6-1).
The mode of failure was torsional fatigue in the tensile
plane , with fatigue striations progressing over a large number
of cycles before final fracture: "torsion in the tensile plane" be-
cause the fracture was at 45° across the shaft surface [see Fig .
1-11) ; "fatigue " because of the radiating striations, or beach
marks , over the fracture surface; and "large number of cycles"
because the beach marks in general were very close together.
Each beach mark may represent many cycles at low stress,
since only one cycle at higher stress will form an additional
mark .
The origin of the fatigue fracture was at the surface along a
sharp, gouged indentation at the juncture of the radius. To find
the origin, follow the area of striations back to the center of the
half-circle or focal point. The focal point is at the surface. Look
at the surface to find the deep gouge mark .
At this point, the cause may seem obvious, and it may seem
unnecessary to do any further examination. But although the
part number had been submitted, it is as yet necessary to estab-
lish if the material and heat treatment verify this part as being
correct or part of the problem.

Metallurgical Examination
Chemical analysis verified that the part was SAE4340 and
thus met material specification. The part was heat treated by
through hardening, then quenched and tempered to 42-47 HRC.
The surface hardness was 46 HRC, and the core hardness (1Jz
radius) 46 HRC.
In conclusion, the shaft met the metallurgical requirements
and should have withstood normal operating conditions. The
mode of failure was torsional tensile fatigue; and the cause of
failure was mechanical damage to the surface of an area of high
stress concentration.
 THE FINALANALYSIS
187

~I!!!i--- Go ug e m ark

Fatigu e cr ack or igin

Fig. 6-1. Failure of this axle shaft resulted from torsional

fatigue in the tensile plane, originating from one of sev-
eral gouge marks observed around the shaft at the splined
radius. The fatigue crack progressed for a large number of
cy cles before final fracture.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
188
This is not the final analysis , because two questions have
yet to be answered: Where did the gouges come from, and how
could they have been prevented? The spacing of the gouge
marks was observed to be coincident with the spacing of the
splines. The spacing was then traced to the operation where
the shaft had been assembled into the splines of the mating
wheel gear. This operation had not been accomplished very
smoothly.

Broken Spira' Bevel Gear

A spiral gear and pinion set were submitted for analysis. The
pinion was intact , but the gear had broken into two sections
that resulted when two fractured areas went through the body
of the gear.

Background Information and

Customer's Evaluation
Field application: The set was a high-speed electric traction
motor gear unit driving a rapid transit train. The gears drive
about equall y in both directions; i.e., forward going north and
reverse going south . Although acceleration and deceleration are
generally smooth, they are rapid.
The assembly was installed in mid-1975. The wheel mile-
age of the assembly was 34,000 miles at the time of failure. The
axial movement of the pinion and bearing stackup was zero.
The customer observed that two complete fractures were
closely associated with two major sections of gear tooth fa-
tigue: fracture area "A" [Fig. 6-2a] developed from a bolt-hole
opening closely related to the root ef a severely fatigued tooth,
progressed along the bearing surface of the adjacent tooth and
through the 0.0.; fracture area "B" [Fig. 6-2b] developed from
the origin of a bolt -hole opening and progressed in a parallel
alignment to the root of the tooth. in a very similar pattern to
"A. "
The customer thought the failure resulted from deep root
stress fatigue, developing from the final depth of the two
mounting-bolt holes.
 THE FINAL ANAlYSIS
189

Progr essi on is normall y dow nwa rd. tow ard th e neutral poi n t bel o w the ce n-
ter of the tooth . a nd th en upward to th e opposite root. In this instance . a bolt
hole wa s in te rce p te d nea r its termi nal end . A seco ndary fatigue IS ) start ed a t
th is in ter cept. a nd s low ly con ti nue d b ot h acros s and through the en tire cro s s
sec tion of th e gear un til fi na l se para tion occu rred .
Fig. 6-2(a). Fracture " A" in a spiral gear. The origin (0) of
tooth bending fatigue is near the toe end of the concave
(reverse) root radius. See also Fig. 6-2(b) and 6-2(c).

Fig. 6-2(b). Fracture "8 " in the gear shown in Fig. 6-2(a).
This fracture is an exact duplicate of fracture "A ." See
also Fig. 6-2(c).
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
190
Visual Examination
Identifying markings are to be in terpreted first. [No part
numbers are listed or shown in photographs here, since part
numbers are the property of a customer and in some in s ta nces
might identify that customer.)
Markings on pinion : PART F-3 G3-26 OLU B
PART Part number [not disclosed].
F-3 Manufactured in June 1973.
G3-26 Mated with gear as set No . 26 in July 1973 .
OLU The material code assigned to Republic Steel
Heat No . 3113989, grade SAE 4320H.
B The second lot of this part going through the heat
treating process .
Markings on gear: PART 0 -3 G3-26 UWJ A 57-22
PART Part number [not disclosed] .
0-3 Manufactured in April 1973.
G3 -26 Mated with pinion as set No. 26 in July 1973 .
UWJ The material code assigned to Republic Steel
Heat No . 6078524, grade SAE 4320H.
A The first lot of this part going through the heat
treating process .
57-22 Set ratio; gear 57 teeth, pinion 22 teeth .

Physical appearance of pinion. Two adjacent teeth were

crushed and spa lIed severely. and all teeth were rounded over
the top face ; this physical damage is of a secondary nature .
There had been hea vy tooth contact loading high near the heel
end of the concave side of all teeth. Also. magnetic-particle in-
spection revealed a fatigue crack originating at the root radius
of each tooth, at the heel end of the concave [forward) side . The
convex [reverse] sides of all teeth were deeply worn and pitted
low on the profile toward the toe end .

Physical appearance of gear. [Study closely Fig. 6-2a. 6-2b,

and 6-2c]. Fifteen teeth broke out at the toe end with a fatigue
fracture whose origin was at the root radius, approximately 1
in. from the toe end on the conca ve (reverse] side . The conca ve
r
THE FINALANALYSIS
191

0----
0- - - A

o --:........_

Fig. 6-2(c). Of the 15 teeth that failed by tooth bending

fatigue, two fractures intersected a bolt hole (see Fig .
6-2a and 6-2b). Four other fractures touched the apex (A)
of a bolt hole. Nine teeth failed by tooth bending fatigue
completely away from a bolt hole. AllIS have an origin
(0) in the same reative position.

sides of the remaining 42 teeth showed extremely hea vy contact

at the toe end and over the top corner and face, causing a
rounded condition of the corner. Magnetic-particle inspection
indicated that several of the remaining teeth had cracks origi-
nating in the same area as the fractures. The tooth contact area
on the convex [forward) sides was very heavy, low on the
profile near the heel end .
Fractures "A" [Fig . 6-2a) and "8" [Fig . 6-2b) show the origin
[0) of tooth bending fatigue 1 in. from the toe end along the
concave root radius . Progression is normally downward to-
ward the neutral point below the center of the tooth , and then
upward to the opposite root. In these two instances, the pro-
gression intercepted a bolt hole near its terminal point. A sec -
ondary fatigue fracture then started at this intercept [point S in
Fig . 6-2b) and slowly continued both across and through the
entire cross section of the part until final separation occurred.
Of the 15 teeth that fail ed by tooth bending fatigue , two
fractures intersected a bolt hole as discussed above; nine teeth
failed in a normal pattern completely away from a bolt hole ; and
four teeth failed normally but touched the apex of a bolt hole
(shown in Fig. 6-2c as point A).

'I,
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
192
Physical Examination
Magnetic-particle inspection w as us ed to d et e rmin e a n y
e x te nde d da mage du e to th e prim ar y failure of tooth bending
fatigue of th e gea r . It in d ica te d seve ra l o the r gea r teeth prog ess -
in g in th e sa me fa il ure patt ern as th e 15 fail ed tee th. It a lso indi-
ca te d t ha t eve ry pini on tooth had s ta r ted to fail b y tooth bend-
in g fa tig ue i n th e forward dire cti on a nd a t th e h eel e n d. Both
dis c ov e ri es s u b s ta n t ia te d th e cro s s ed tooth patt ern disc usse d
ea rl ier .
Th ere was al s o a seco n d re a s on for ma gneti c-p a rticl e in-
s pec t io n . Th e teeth o f bo t h p art s h a d bee n g ro un d b y a s p ira l
be v el tooth g ri nder. so it was n eces s a r y to in spect for g rind ing
chec ks th at mi ght h av e been a fac tor . N on e w ere found .

Nital etching w a s used to d e te rmin e if th e re we re a ny

g rin di ng burn s th at co uld ca use tem pered s po ts in c ri tica l
a reas. Ag a in , n on e were found .

Metallurgical Examination
A d ec is ion mu st n ow be m ad e as to whi ch . if a ny, of th e
met allu rgi cal test s a re to be mad e . A lt ho ug h th e ca us e of failur e
ma y be a p pa re n t a t thi s time . so me do ub t w o uld re ma i n unl es s
a ll rem ainin g p os si biliti es w ere c hec ke d .
Th e p rim ar y m od e w a s tooth b en ding fatigue . bo th of th e
gear a nd of th e pini on . Wa s th e ca rb urize d case ha rd e no ug h a nd
d eep e no ug h a t th e roo t radi us to h a ve wi th st ood no rm al load s ?
A ca se h ardne s s tra ve r s e mu st be tak en to d et ermine case prop -
e rt ies . Sin ce th e failure w a s a t th e root rad iu s tow ard th e toe.
se lec t a tooth fr om eac h part th at re ma ins but is c rac ke d in th e
sa me a rea . and s ecti on th at to oth a t.th e point of th e c rac k or igi n .
Prepare the sam p le for th e mi croh a rdn ess test e r a nd tak e th e
h ardn es s re adin g s as cl osel y as reaso na b le to th e c rac k .
Th e teeth of b oth p a rt s h ad b een g ro un d . Co u ld th is m ak e a
difference in case c ha racte r is tics fr om o ne s ide to th e o the r?
Ind eed it c o uld, a n d of te n d oes . Therefo re . a case ha rdn es s
tra vers e at both r adii is n ece s s ar y.
Is it nec essa r y t o run a ca se hardness tra ver se a t th e mid-
p r ofil e o f both si des? Wh en d et e rmining fa ct ors for tooth b end-
 THE FINAL ANALYSIS
193
ing fatigue. it is no t neces sa r y. How e ver. after co m ple ting all
th e prep ara ti on up to thi s point , a n a na lys t wo uld be a mis s not
to. In fa ct, thi s is th e on ly way to es ta b lis h if th e ca se d epth a nd
ha rdness s pec ifica tio n had b een maint ained.
Is it necessary to tak e a che mical a na lys is? N ot a t thi s time.
T he mat eri al a nd hea t treatment ca n usuall y be recognized b y
the micros tr uct ure. If the micros tructure is not cons istent with
ex pec ta tions, a n anal ys is w ill be necessa ry .
Microscop ic exami natio n is a lways acco mp lis hed w he n a
sam p le ha s b een p repa red for case h a rdn es s traverse . This
p rocedu re s ho uld be auto ma tic, si nce it is the on ly vis ual m eans
of tying all o the r ph ys ica l a nd me tall urgi cal test s toget he r .
Aft er it ha s b een de te rmine d w hic h procedu res are to b e
followed a nd wh y, th e res ult s are observe d :

Pinion Gea r

Su rfa ce hardness. . . . . . . . . . . . .. 61 HRC 58/59 H RC

Eff ective ca se depth, roo l radius . 0.040 in . 0.044 in .
0.042 in . 0.048 in .
Effecti ve ca se d ep th . midprofil e. 0.061 in . 0 .077 in .
0.061 in. 0.077 in .
Ret a ined a uste ni te Less th an 5% Ni l
Core ha rdnes s [mid profi le -root ).32 -25 H RC 41 -32 HRC

Final Analysis
A ll ph y si cal a nd me ta ll urg ical cha ra cteris tics were we ll
wit hi n spec ified sta ndards. a nd bot h parts s hou ld h a ve w it h-
stood norm al loa d in g co nd itio ns.
The prima r y mode of fa ilure was too th bendi ng fatig ue of
th e gea r fro m the reve rse directi on near th e toe e nd . Occurrin g
perhaps s im ulta neo us ly was to o th be nding fati gu e of th e pini on
teeth fr om th e forw a rd directi on near th e h eel e nd . The bolt
hol es in th e b a ck fa ce of th e gea r h ap pen ed to be in th e path of
frac ture prog res si on : but they did provi de a pat h of lea st resis -
tance fr om the point of interce pt to the back face, wh ich a llowed
for th e fu ll bod y b rea k .
The ca use of fa ilure was a crossed -o ver toot h beari ng co n-
d iti on th at pl a ced loa d s a t th e h eel e nd w he n goi ng forwa rd a nd
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
194
a t th e toe e nd w he n go ing in re ve rse . T he co ndi ti on was too co n-
s iste nt to be a defl ecti on un d er load ; th eref ore. it most lik el y
was perm anent mis alignm ent w i th in th e asse m b ly.

Spiral Bevel Drive Set

A s p ira l beve l set tha t failed in service was s ubmit ted for a n
eva l ua tio n.

Background Information
Field a pp lic a tion: Thi s w as th e main dr iving set in th e dif-
fe re n tia l h ou sing of a large fr ont-end loader mo ving coa l in a
s to rage area. Th e ma ch in e ha d opera te d for a p proxima te ly 1500
hours.

Visual Examination

Identifying markings:
Marking s on pinion : PA RT FO-1 29 48 17 41 -8
PART Part number [n ot di scl os ed].
FO- 129 Mat ed w it h ge a r as se t No. 129 in Jun e 1970 .
4817 G ra de of s te el us ed [S AE].
41 -8 Set rati o; gea r 41 teeth , pinion 8 teeth.
Markings on ge a r: PA RT F-O FO-1 29 4820 4 1-8
PART Pa rt number [not di scl os ed] .
F-O Manufa ctured in June 197 0 .
FO -129 Mat ed with pini on as se t N o. 129 in June 19 70 .
48 20 G ra de of s tee l us ed [SAE].
4 1-8 Se t rati o; gea r 41 teeth , pini on 8 te eth .
Physical appearance of pinion [Fi g. 6-3 ). Two ad jace nt
teeth had brok en out. The nex t a d jace n t to oth had brok en lo os e
in the same manner and w as h anging . All fr a ctures ind icate a
tooth bending fa tig ue failure originating at th e root radius of
th e co nca ve [dri ve ] s ide a t exactl y mid way bet w een th e e nd s of
th e teeth . Th e remaining five te eth a re hi ghl y p olished on both
s ide s and th e co n tac t pattern is w ell ce n te re d.
 THE FINALANALYSIS
195

Fig. 6-3. Spir al bevel pinion tooth, 2X. T ooth bend ing fa-
t igu e w it h or igin (a rrow) at the root radius. ex ac tly mid-
wa y between th e ends of th e teeth.

Physical appearance of gear. Most of the teet h ha d bee n

chipped at the back angle, but evidently this was subsequentia l
damage to the pinio n failure. The tooth co ntact pattern of all
teet h was we ll ce n te red a nd uniforml y po lis hed on bot h sides .
No wear was evident.

Physical Examination

Magnetic-particle inspection was performed on each part.

The pi nio n showed no indication of additio nal cracki ng, nor did
the gear.

Metallurgical Examination
Fro m th e preced ing exam inatio n, the a na lyst a rrived at ~ I

severa l conclusions. T he primary failure invo lved the pinio n

tee th only ; the gear need not be exami ned . Toot h be ndi ng fa-
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
196
tigue demands a case hardness survey of the root radius and, of
course, the mid profile. Also, a microscopic examination is nec-
essary.
There appears to be no difference in the wearing of any
tooth; therefore, select anyone of the remaining teeth, section it
midway from both ends, and prepare a sample for a case hard-
ness tra verse and microscopic examina tion.
From the case hardness survey and the microscopic exami-
nation, the following results were observed:
Core structure: An equal admixture of lamellar pearlite and
low-carbon martensite. This indicates a well quenched
and tempered structure .
Case structure: Acicular martensite retaining approximate-
ly 10% austenite near the surface. Excellent structure for
material and heat treatment.
Case depth: (specified 0.075-0.090 in.)
0.085 in. effective depth at midprofile
0.054 in . effective depth at root radius
Case hardness: (58-63 HRC specified)
60 HRC at midprofile (by Tukon)
59 HRC at root radius (by Tukon)
Core ha rd ness: (no specifica ti on)
36 HRC, tooth centerline at mid profile
28 HRC, tooth centerline at root
The material and heat treatment were as specified.

Final Analysis
The mode of failure was tooth bending fatigue with the
origin at the designed position: root radius at midsection of
tooth. The load was well centered, and progression occurred for
a long period of time. The cause of failure was suddenly applied
peak overload, which initiated a crack at the root radius. Pro-
gression continued by relatively low overstress from the crack ,
which was now a high stress-concentration point. This was a
classic tooth bending fatigue failure (as explained in Chapter4) .
 THE FINALANALYSIS
197
Spur Pinion
One end of a n axle shaft containing th e integral s p ur pini on was
s ub mitt ed for e xa m ina tion, along with the report of a tooth pit-
ting failur e (Fig. 6-4).

Background Information
Th e s p ur pinion , in tegral to th e axle s ha ft, ope ra ted in a
med ium-size, off-high w a y truck at a n ope n-p it mine. The pin-
io n w a s th e input sun pini on of a gea r-red uc tion unit in th e rear

, .
,~ ~ . '1"# 0 • • 4 ' ' - - -

Fig. 6-4. Spur pinion, I X, showing low-profile pitting

along heavy line contact that is characteristic of tip inter-
ference from mating part.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
198
w hee l asse mb ly , w here it mat ed w i th four sp ur too th idl er
gears. The o nly record ed length of se rv ice was the cus to me r's
s ta te me n t- "fo r a rel a ti vel y s hort time. "

Visual Examination
Identifying markings. Onl y th e pin ion h ea d h a d bee n re-
turned . Th e s haf t por ti on h ad been torch -cut away . Th ere we re
no markings o n th e porti on re t urne d. Th e acco m pa nying rep ort
did menti on th e pa rt numb er , w h ic h is a ll th at ca n be used at th e
s tart.

Physical appearance. All th e teeth we re in ta ct a nd s howed

very little w ea r . O ne side s howe d a lin e of fin e pit s, mostl y full -
length , ver y low on the profil e. Th e y in crea sed in in tensity to-
ward th e ope n e nd of th e teeth , as ind ica ted b y heav y pitting
a nd s palli ng . The opposite s ides of th e teeth a lso s howed full-
length co n tac t ve ry low on th e profile b ut d id n ot s how th e same
exte ns ive pitting d amage.

Physical Examination

Magnetic-particle inspection revea led n o indi ca ti on of

crac k in it iatio n .

Tooth characteristic study of th e invo lu te was tak en to de-

te rmine if th ere w a s a positive er ror d evi a ti on fr om tru e invo-
lut e a t the low est p oint of th e a cti ve profil e. The result was th at
th e in volu te cha r t s howe d a n a lmost stra ig h t lin e ove ra ll, w i th a
s lig h t nega ti ve d evi ation a t th e s ta r t of th e ac tive profil e.

Surface hardness checks we re ta ke n a lo ng th e ma ximum

di amet er of th e teeth a nd a lo ng th e root ce nte r. The resulting
hardnes s was 5 7 HRC, which is w i th in th e s pe cifica tio n of
5 7-63 HRG.

Metallurgical Examination
The pitting occurred a lo ng th e low est p oint of th e ac tive
p rofil e. A ca se hardness s u rvev a nd a mic ros copic exa m ina tio n
 THE FINA L ANALYSIS
199
should be made of this area. Also. the tooth s ection s hould be
take n nea r the ope n end of the pinion tooth through a m edium
amo un t of p itt in g . in order to study t he mech ani s m of t he pi t-
ting. Since th ere ha s b een n o p osi ti ve id entifi cat ion of m at eri al .
a che mica l a na lys is s ho uld be perform ed .

Case hardness traverse

Surface hardness (b y T uk on , 500g) : 58 H RC
Effective case d epth: 0.060 in .
Co re hard ness: 35 H RC at midpoint, and at root . 33 HR C

Microstructure
Core s truct ure: Low-carbo n ma rt en sit e w i th fi ne b lac k
bound ari e s. a nd a n occa si on al c rys ta l of tit anium ni-
t rid e (TiNJ . T he s tr uc t ure is norm al for goo d heat
tr eatm ent a nd th e fi ne bl a ck b oun d a r ies a nd T iN
crys ta ls are co mmo n in bo ron-treated a lloy s tee ls.
Case stru cture: Rath er coa rse acicular tem pere d m ar-
te nsite retaini ng approximatel y 20% auste nite nea r the
surface .
Surfa ce stru cture: At t he pitted area low on the active
profile. evide nce of ca tas tro phic move me nt of m ate ria l
away fro m th e pit chline a nd tow ard the ro ot. (An
id entical exa m p le is sh own in Fig. 4-15 .J

Chemical analysis . Carbon . 0 .12%; m an g an e s e , 0. 8 2%;

ni ck el, 1.8 6%; ch ro m i um. 0 .56%. a nd m ol yb d enum , 0 .28%. Th is
a na lys is. a long w it h the mic rostr ucture. co nfirms the s pecified
ma terial to be SAE 43 BV12 .

Final Analysis
Th e m od e of fa il ure was surfa ce co n tac t fatigu e through th e
s hear pl ane s u bs urface. at th e low est point of s ing le tooth co n-
ta ct. Th e ca use of fa il ure wa s too th ti p in terf erence fr om the
ma ting gear teet h . Sin ce th e m ating pa rt s w it h in th e asse m b ly
had not bee n ret u r ne d or exam ine d. t here a re s ome unans w e red
que sti on s fa cing th e cus to mer:
• Whi ch of th e idler gears is s how ing hea v y load co ntact
over t he top corner?
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
200
• Do the teeth of the mating part or parts have a positive
involute error at the top of the active profile?
• Are all four idler gears from the same lot and mated; or is
one from another lot that may have different tooth char-
acteristics?
• What is happening within the assembly or with the four
idler gears that is causing a consistent and uniform load
a t the very end of the sun pinion?

Spur Gear
Three spur gears that had formed a straight-line train in a speed
reducer were brought in for examination and evaluation, with
no report accompanying them.

Background Information
No background information was submitted. The analyst
was told only that the gears were to be returned intact. There-
fore. he could not section them.

Visual Examination

Identifying markings:
PAR T. 8622, D-2. on each gear
PART Part number [not disclosed).
8622 SAE grade of material.
D-2 Manufactured in April 1972 .
[It was a well-mated set of three.)

Physical appearance. No evidence of wear appeared on any

tooth of any gear, and the contact pattern was well centered and
normal. The pitting, as shown in Fig . 6-5, was on only two teeth
of the input gear. Notice the heavy line of pits low on the profile
of one tooth and high on the profile of the adjacent tooth. The
next gear in line (idler gear] had two teeth with exactly the same
pitting pattern but in the reverse direction. matching perfe ctly
 THE FINAL ANALYSIS
201
the firs t gear. On the op posite side of th e se co nd gear [180 ° )
we re three ad jace n t teeth w ith a line of pit s- one hi gh, one me-
di urn, a nd one low. T he third gea r in line [out put gear) had three
ad jacent teet h w it h line pit s exac tly a s in the second gear, but in
the reve rse d irec tion, match in g per fectl y the seco nd gea r.

Physical and Metallurgical Examinations

Neithe r a physica l exami nat io n nor a me tall urgical exami-
nati on was pe rform ed. I !
I

Final Analysis
I
A tel eph on e ca ll was th en mad e to the user . A li st of q ues - I I

tio ns a nd res po nses fro m th is con versa ti on follows:

Fig. 6-5 . Spur gear, O.9X. Only two teeth pitted, one low on
profile and th e adjacent tooth high on profile. Mating gear
had two teeth as mirror image. This could onl y occur with
th e gea rs in a s ta tic posi tion under a reverbera ling typ e of
load.
-- SYSTEMATIC ANALYSIS OF GEAR FAILURES
202

Fig. 6-6(a). Spiral bevel pinion, O.9X. Seven of nine teeth

failed by heavy rolling contact fatigue with the origin at a
bias across the profile in a confined area.

Q. What type of equipment did this speed reducer operate,

and under what conditions?
A. It was the power drive unit for an overslung lumber car-
rier running back and forth on rails .
Q. What type of power input was used- electric motor,
gasoline engine , or hydraulic motor?
A. Hydraulic motor driven by a gasoline engine .
(The analyst knew from past experience that other
components driven by hydraulic motors would shud-
der violentl y when held against an end stop. His next
question reflected this knowledge .)
Q. Is it possible for this traveling cra ne to be locked in the
same position time after time with th e ga s oline engine
running?
A. Yes; in fact, every morning at one end of the track, th e
operator, who is 20 feet above the gro und , starts the
gasoline engine and lets it run until warmed up . In th e
meantime, the hydraulic motor and the speed reducer
both vibrate constantly until the brake is released.
It was quite evident that this mode of failure could only
occur with the gears in a static position under a reverberating
type of load.
 THE FINALANALYSIS 203
Spiral Bevel Set
A spiral bevel gear a nd pinio n set was submitted for exami na-
tion and e va l ua ti on because it had "worn out. "

Background Information
This was a spiral bevel dri ve set with th e g ea r a ttached to a
differential. The assembl ed unit was driving a new, large ,
experimental farm tractor in th e norma l p lowing and tilling op-
erations .

Visual Examination

Identifying markings:
Markings on pinion: PART 0-4 E4-1425
PART Pa r t number (not disclosed].
0 -4 Manufa ctured in April 1974.
E4 -1425 Mated with the gear as set No . 1425 in Ma y
1974 .
Markin gs on gear: PART E4-1425
PART Part number (not disclosed)
E4-1425 Mated with pinion as set No . 1425 in May 1974.

Physical appearance of pinion (See Fig. 6-6a). Seven of the

nine teeth failed from heavy pitchline contact on the concave
(drive) side, with a progressive fatigue fracture through the
case and core to the opposite side and down, parallel to the pro-
file , to the too th root on the co nvex side. The two intact teet h
s how t he start of line crushing a t a bias ac ross th e concave pro -
file. M a tin g too th co ntac t was hea vies t a t center profile. O f
three co n tig uous teeth that ha d been removed by co nti n ua l
crus hing a nd wear, t he ce nter tooth sti ll h a d a portio n of the toe
end remai ni ng, w hic h was p laci ng a n extre me ly hea vy poi nt
pressure on the toe e nd of every third gear toot h.

, Physical appearance of gear. 48 tee th . Every thi rd toot h

was broke n a t t he toe end on ly ; otherwise, all tee t h were in ta c t.
--
SYSTEMATIC ANALYSIS OF GEAR FAILURES
204
All teeth s how ed uniform h ea vy con tac t, w ell ce n te re d o n th e
profile of th e convex (dri ven] side.

Physical Examination

Magnetic-particle inspection showed a fati gue crac k p ro-

g res s ing through on e of th e two inta ct pinion teeth a lo ng th e
s ame path as that obse rve d in th e fail ed teeth . No indicati on s
w ere obs erv ed on th e ge a r teeth.
Hardness checks on th e s urface of th e pinion teeth ra ted th e
hardness a t 61 HRC.

Metallurgical Examination
From th e e xa m ina tio n , a numb er of conclu si on s w ere
drawn. The primary failure is definit el y associat ed with the
pini on; th e gea r need not be exa mine d . The mode offa il ure mu st
be d et ermined through th e co nce n t ra te d load area . T o acco m-
pli sh thi s , a ca se h ardnes s tra verse must be mad e throu gh a
represent ati ve a rea a nd th e m icrostructure s t ud ied. Th erefore,
se ct io n th e one remaining pin ion to oth through th e mid secti on .
Mount a nd prepare the sa m p le fo r a case h ardne ss tra verse a nd
a microscopi c ex a mina tio n.
From th e d at e of manufa ctu re, th e d ocument ed in form ati on
on han d s howe d the pini on to be one of 10 forging s mad e from
Republi c Heat 6070929, g ra de 4820H. A ni ck el s po t chec k co n-
firm ed th e grade . No fur the r che m ica l a na lys is wa s neces sar y.
Results of examination
Case d epth : (specifi ed 0.055-0 .070 in. effec tive ]
0.0 70 in . effe c tiv e d epth a t .mid p rofil e
0.0 58 in . effec tive de p th a t roo t radius
Case hardn es s: (specifi ed 59 -63 HRC]
59 HRC surfa ce at midprofil e (b y Tukon]
58 HRC surfa ce a t roo t ra d i us (b y Tukon]
Co re ha rdn css:
45 HRC tooth centerline a t midheight
44 HRC tooth cent erline a t roo t ci rcle
 THE FINAL ANALYSIS
205
Cor e stru cture:
Rather blocky structure of low-carbon martensite with a
small amount of pearlite
Ca se s t r uc t ure:
Tempered acicular martensite retaining approximately
10 % austenite near the surfa ce; the first transition prod-
uct appears below 0 .020 in. fr om the surface. The frac-
ture area (originating high on the profile) indicates roll-
ing contact fatigue , with subsurface shearing progress-
ing toward the top of the tooth. In the high compres-
sive area there are areas of "butterfly wings" extending
from 0.010 to 0.035 in. below th e surface, as shown in Fig .
6-6 (b). The surface had not been smeared by any sliding
action .

Fig. 6-6(b). Case microstructure (100X) of a high-compres-

sion area of one of the two intact teeth in the spiral bevel
pinion shown in Fig . 6(a). Note the altered martensite in
the subsurface shear plane. These "butterfly wings" ex-
tended from 0.010 to 0.035 in. below the surface.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
206
Final Analysis
The mode of failure is rolling contact fatigu e, and th e ca use
of failure improper engineering design. The pattern of con tin u-
al overload has been restricted to a specific concentrated area
situated diagonally across the profile of the loaded side, which
is consistent on every tooth . [The classic example of this failure
mode is discussed in Chapter4 and illustrated in Fig . 4-16 . The
cause of this mode of failure is discussed in Chapter 5 and illus-
trated in Fig . 5-14 .J Corrective action could be to reduce the ap-
plied stress [not practi cal], or to change th e design to allow a
greater overlap of tooth contact.
Emphasis must be made that all failures displaying this
pattern do not nec essarily indicate a design problem. Keep in
mind that the normal loading contact line of a spiral bevel tooth
travels along the profil e in a biased direction . Many failures
look like this, even though there is greater than a one-to-one
ratio . Before concluding that the condition shown in Fig. 4-16 is
applicable, a complete engineering analysis must be made. The
obvious may not always be obvious .
Damage to the gear teeth could have been reduced by
changing the ratio slightly. The set ratio is 48 to 9. By changing
to either 49 to 9, or 47 to 9, there would always be a hunting
tooth situation and all teeth would receive the same loading
characteristics .
The set ratio is important when noting the pattern of a fail-
ure. For instance, two gears with a 1-to-l ratio [i.e., the same
number of teeth in both parts] will always have the same tooth
in contact. With a set ratio of 2 to 1 [gear 20 teeth, pinion 10
teeth], each pinion tooth will contact two gear teeth only , and
those two tee th will be 180 0 apart. The same logic is used for a
ratio of 3 to 1 or 4 to 1. The 48-to-9 ratio discussed earlier has a
common denominator of three; therefore, a single tooth of the
pinion will , during three rotations of the gear, contact every
third tooth of the gear. The result was that every third tooth of
the gear had been damaged by only one pinion tooth .
 THE FINAL ANALYSIS 207

Spiral Bevel Set

A spiral bevel gear and pinion set that showed "excessi ve wear
on the pinion teeth" was submiLLed for analysis .

Background Information
This spiral bevel set was the primary drive unit for the dif-
ferential and axle shafts of an exceptionall y large front-end
loader in the experimental stages of development.

Visual Examination

Identifying markings
Pinion markings: PART L-9 L9-211 #1 6-38
PART Part number [not disclosed].
L-9 Manufactured in December 1969 .
L9-211 Mated with gear as set no . 211 in December
1969.
#1 The first set mated and tested for the order be-
ing processed.
6-38 Set ratio; pinion 6 teeth, gear 38 teeth.
Gear markings: PART L-9 L9-211 #1 38-6
The same information as for the pinion above.

Physical appearance of pinion. The tooth contact area ap-

pears to ha ve been hea vy a t the toe end and low on the profile of
the concave [drive) side. The entire active profile shows surface
rippling for over half the length of the tooth from the toe end, as
shown in Fig. 6-7[a) and [b]. Pitting has originated at the lower
edge of the profile, 2 in . from the toe end of the concave side [Fig.
6-7a]; it has subsequently progressed in both directions along
the lower edge of the active profile and upwards toward the top
of the tooth, changing the mode from pitting to spalling [Fig.
6-7b]. Both photographs show evidence of a double contact: the
original setting, and a secondary pattern with the pinion moved
out and away from the center of the gear. The convex (reverse]
sides are slightly worn and scuffed high over the profile.
-- SYSTEMATIC ANALYSIS OF GEAR FAILURES
208

(a)

(b)

[a] A rip p led surface for 'I. lengt h of th e too th fr om th e toe e nd . Pit ting has
or igi na ted low on th e a ctiv e profil e. 2 in . fr om the to e e nd . [b I Pit ti ng area ha s
e x te nded in both directions a nd has bro ad e ned . Th e ce n tra l profile ha s a n area
of s pa ll ing tha t a p pea r s to be con tig uo us wit h the pi tti ng. T he micr os t ru ct ure
in di cat es th e tw o modes a re occ urri ng inde pen dentl y .
Fig. 6-7. Spiral bevel pinion of 4820H steel, O.6X.
 THE FINAL ANALYSIS
209
Physical appearance of gear. All the teeth are intact and
highly polished along the top of the profile toward the toe end of
the convex (forward) side. There is some scuffing low on the
profile of the concave sides.

Physical Examination

Magnetic-particle inspection. There is no evidence of tooth

bending fatigue on either part. The several cracks indicated are
associated with the spalling surfaces on the concave sides of the
pinion teeth. The gear teeth show no indication of fatigue.
Tooth characteristics. The original setting test charts taken
on this set in December 1969 are on file and document a very
good and normal tooth contact pattern. The secondary tooth
pattern that prevails at this time could not be duplicated on the
gear tester. The pinion appears to have moved away from the
center of the gear, thus activating a very heavy low-profile con-
tact toward the toe end.
Surface hardness tests. The surface hardness of both parts
is 58 -59 HRC, which is within the specification of 58-63 HRC.

Metallurgical Examination
The rippling [Fig . 6-7a and b) is extensi ve and covers the
entire area of the initial contact pattern. Rippling reflects
movement or a tendency to move. The movement can be super-
ficial adjustment of a surface heavily rolled while sliding; or it
can be an adjustment of the surface during absorption of energy
by retained austenite . Rippling is not always associated with a
failure, nor is it detrimental in itself. It does usually indicate,
however, a rolling/sliding condition that is highly compressive .
The pitting is originating in a localized area, and the spall-
ing appears to be a contiguous event, although this type of spall-
ing could easily have been found as an independent mode with -
out the association of surface pitting. It is therefore necessary
to section the pinion tooth normal to the active profile, near the
spalled area shown in Fig. 6-7[b), and to prepare this section for
a microscopic examination and a case hardness traverse.
.......

SYSTEMATIC ANAlYSIS OF GEAR FAILURES

210

Note: th e surface s how s no ca ta strophic mov em ent; th e "b utt erfly w ings"
are gen erall y parallel to the s urface but ex tend 0.027 in. below th e surface; and
microstru cture is ver y fin e acic ula r mart ensite retaining less than 5%
aust enite.
Fig. 6-7(c). Section normal to surface of tooth profile
taken very near the spa lied area shown in Fig. 6-7(b),
200X.

Microscopic examination. The core structure contains an

equal admixture of low-carbon martensite and lamellar pearl-
ite. The case structure displays a very fine acicular martensite
retaining less than 5% austenite . The pitted area low on the ac-
tive profile shows evidence of a heavy metal-to-metal sliding
action tending to be adhesive. The central profile area, sub-
jected to spalling, has a surface that is relatively unaltered (Fig.
6-7c) . In the same area , there is a subsurface display of "but-
terfly wings" reaching to a depth of 0.027 in. from the surface.

Case hardness traverse. Figure 6-7(d) is self-explanatory

as to the process specifications and the results obtained.

Final Analysis
The material and process specifications have been met sat-
isfactorily. The primary mode of failure was rolling contact fa-
 THE FINALANALYSIS
211

60 /\/\ A
M
J. I POin t -

--
0 .082 in.

u 50
~ ~......., r-- r----.. /
r-..
a;
;:
-'"
o
0
40
R . /
oat radius
0 .05 3 in .
~ .,
cr:
<Ii
'"
OJ
c:
30
"0
m
J:
20

10

0.010 0 .0 20 0 .03 0 0 .040 0.050 0 .06 0 0 .0 70 0 .08 0 0 .09 0

Dept h below surface. in .

Fig. 6-7(d). Case hardnes s traverse of section used for Fig.

6-7(c) taken from tooth pictured in Fig. 6-7(b) . Results are
self-explanatory.

tigue of th e con ca ve (d ri ve] ac tive too th profile. The sp alled

area was a co nseq ue nce of thi s actio n . The pitting low on th e
profile a p pea rs to ha ve been origina ting after the shift of the
pinion tooth away fr om th e gear ce nte r. The s hift of th e pini on is
most oft e n due t o a bea ri ng di spla cem ent or malfunction. The
ca use of this failure a ppears to be co nt in uo us high ove rl oad th a t
ma y a lso ha ve co nt rib ute d to th e bearing di splacem ent.

Spur Gear
A porti on of t w o la rge s pur tooth bull gears t ha t had spalling
teeth w as submitted for eva luatio n (Fig. 6-8a ).

Background Information
The gea rs we re taken fro m a fi na l dri ve w hee l re d uc tio n
unit of a very large ope n -p it mining tru ck . T he dri ving pinion s
were not s ubmitte d ; th e usu al ass ura nce w a s give n tha t th e y
w ere in goo d condit io n a nd were still in ope ra tion. Apparentl y,
--
SYSTEMATIC ANALYSIS OF GEAR FAILURES
212

(a) (b)

Fig. 6-8. Spur gear tooth, SAE 4147H, quenched and tem-
pered to 311 HB, machined completely, induction hard-
ened with a tooth space inductor by traversing one tooth
space at a time. (a) Surface spalling along one tooth flank.
(b) No hardened case on active profile of one side of the
teeth.

these two gears had been operating for relatively few hours.
The loads were considered to be maximum for the opera tion.
These parts were to have been SAE 4147H steel, quenched
and tempered to 262-311 HB, finish machined, induction hard-
ened [both the spline and the spur teeth). and tempered at 350 of
to a specified hardness of 55 HRC minimum on the hardened
surface.

Visual Examination

Identification markings. None on either part; the markings

were on the portion not returned. The part number had been
submitted on the shipping papers.

Physical appearance. The two' parts were identical. Four

contiguous teeth had a spalled area on one side only at mid pro-
file but for varying lengths [Fig. 6-8a) . They were not consis-
tent. The nons palled teeth showed a very good surface on the
same side, and all of the teeth were in good condition on the
opposite side. There was no evidence of adhesive or abrasive
wear.
 THE FINAL ANALYSIS 213

Physical Examination

Magnetic-particle inspec tion. There were no indications

other tha n t hose assoc iated w ith the s palled areas.

Sur face hardness testing. The ends of the teeth were at 311
HB (specified , 262-311 HB). The surface at the root was 58 HRC
(specified, 55 H RC mi n).

Nital etching. The end faces of the gear teeth were polished
and nital etched for a view of the hardened area. (This viewing
area is the same area examined by the induction hardening op-
erator on eac h p roducti on pa r t to maint a in consiste ncy .) T he
end face of a ll s ubmi tted teet h s howed a cons is te n t ha rde ned
pattern a long the full profile and around the root fillets . From
this vantage point, they would have been acceptable .

Metallurgical Examination
Up to this point. no incriminating evidence has been found;
but there is a need to find out what has happened in the spalled
areas. T he consiste ncy in the spalled pattern is that every
spalled tooth has a spa lled area 1 11z in . from one ope n end. T here-
fore. both eight-toot h sections shou ld be cut para lle l to and at a
plane l lfz in . from the end face . The freshly cut plane will thenbe
nital etched for a macroscopic examination.

Macroscopic examination [Fig . 6-8bJ of th e nita l-etc hed

surface revealed that the spalled areas retained very little, if
an y. induction hardened case. The compressive loads origi-
nated the failures at the case/core interface and at the terminat-
ing junctio n of the surface harde ned zo ne . A lso . a lo ng the pro-
file of the teeth tha t ha d not failed, the h a rd en ed case was as
sha llow as 0 .020 in .

Chemical analysis confirmed the rna terial to be SAE

4147 H. as specified.
--
SYSTEMATIC ANALYSIS OF GEAR FAILURES
214
Final Analysis
The parts had met the material and initial heat treat hard-
ening specifications. The mode of failure was tooth profile spall-
ing . By definition, spalling originates at a case/core interface or
at the juncture of a hardened/nonhardened area.
The cause of this failure was either insufficient or no induc-
tion hardened case along the active profile. The cause was acti-
vated by a nonfunctioning induction hardening coil that did not
or was not allowed to harden the mid profile of several teeth.
The teeth had been induction hardened by the single-tooth
method, wherein the adjacent profiles and included root, de-
fined as a tooth space, are heated by a tooth space inductor tra-
versing the length of the tooth. Continuous cooling and quench
flow must be maintained very closely; any interruption will re-
sult in an improperly hardened pattern. Since the hardening
pattern was uniform on most of the teeth, the problem appears
to have been related to the flow of cooling water. When noticed
by the operator, the condition was corrected; but the induction
machine operator's error was failing to ha ve the gears checked
closely for a lack of a hardening pa ttern on those few teeth. Next
in line was the nital etch procedure set up by the inspection
department, which also failed to discover the condi tion . Though
the problematic condition was overlooked during two other
stages, the cause ultimately goes back to the heat treatment
process. This failure was determined to be an isolated case and
did not warrant a recall of all the gears in operation.

Hypoid Pinion
The hypoid pinion shown in Fig. 6-9(a) was placed on the fail-
ure analyst's desk with the usual 'p lea for assistance. To the
analyst's question of where the mating part was, the response
was that there was nothing wrong with it, so it was still in ser-
vice. Please note: Those who disassemble units must recognize
the importance of observing every drive gear set to determine if
it had been mated properly. Too many times, new parts are
 THE FINAL ANALYSIS
215
Rockwell " C"
33 43 45 48

Fig. 6-9(a) . Hypoid pinion , 3/4 X. Mode of failure was a deep

w ear pattern to w ard the toe end. Note the surface hard-
ness pattern that was the result of an im proper quench.

pla ced in operation with apparently "good" used parts; but the
toot h charac teristics of one may be abso lute ly wrong for the
other. S ubse q ue n t fa ilure of the seco nd set ma y be exceptio n-
a ll y rapid . If mism a ti ng occ urs , it is ve ry diffic ult to determi ne
respons ib ility; b u t no gear ma n ufa cturer should be respo ns ible
to warra n t a ny serialized pa rt th at ha s bee n misma ted.

Background Information
T he h ypoid pini on was the d ri ving memb er of a pow er unit
opera ti ng a rapid tran sit car. The pini on had been re moved fro m
service a t the end of th e initial test peri od becau se it showed
u ndue wear. (Kee p in m ind th at tran sit ge a ri ng ru ns under f ull
load in both directi on s. )
-
SYSTEMATIC ANALYSIS OF GEAR FAILURES
216
Visual Examination

Identify ing markings:

PART 4820 6-43 J-7 K7-97
PART Pa rt nu m ber [no t d isclosed ).
4820 SAE grade of materia l.
6 -43 Set ra tio; pinion 6 teet h . gear 43 teeth .
J-7 Manufactured in Oc tob er 1967 .
K7-97 Ma te d w it h a h yp oid gea r as set No. 97 in No-
vem ber 1967.

Physical appearance. T he co n tac t patt ern on th e co ncave

(f or w a rd] sides s hows n o wear a t the heel end . but th e wea r
in crea ses un if orml y to a deep ly wo rn a rea n ear th e toe end . It
a p pears to be d ue to h ea vy fr icti on . Th e convex (reverse] s ides
of th e tee th s how a more uniform co n tact area tha t is sc uffed the
full length of the pattern . The en tire pi nio n head is b lac k d ue to
a hea vy coati ng of iron -m angane se-phos pha te.

Physical Examination

Magnetic-particle inspection re veal ed n o in di ca ti on of

cracking .

Surface hardness testing wit h a te s t fil e revea led a varia -

tion in ha rdnes s from "file ha rd " on th e hea d beari ng to "very
soft" on the p ilo t bearing . The vario us HR C chec ks are indicated
in Fig. 6-9 (a ].

Metallurgical Examination
T h is exa mi natio n is us ed to de termine w ha t type of wear
h a s occ urred, a nd w hat h a s caused the exis ti ng h a rdnes s pat -
tern . A sectio n of a too th a t midl en gth is se lec ted a nd prepared
fo r a case ha rdnes s tra verse and mi croscopi c exa minatio n.

A case hardness traverse (Fig . 6-9b ] re vea led a def ini te

tem pe ring of the surface. a lo ng wit h a decrease in ha rdn es s
 THE FINALANALYSIS
217
thro ug ho ut the case, w h ic h influ enced th e e ffec tive depth . Al s o,
the mid pr ofile had be en affected mor e seve re ly th an th e r oot ra-
diu s . And th e core ha rdnes s, 38 HRC , is typi cal of a good
quench in g a nd temp ering proces s for th e tooth root but is a bo u t
three hardne ss point s lo w for w ha t wo uld have been ex pe c te d
a t th e midp oint.

Microscopic examination of t he tooth sec tio n revea le d th at

th e co re s truc t ure [at th e midp oint) co n ta ine d ve ry fin e tem -
pered low- carb on marten sit e a nd pea rlit e . The case s tr uc t ure
di spla yed very fin e tempered mart en site thr ou gh out, a n d no re-
tained a us te nite. And th e s urface s tr uc t ure s ho we d no evide nce
of fr icti on al di s tres s; it had been clea n ly cu t by a b rasio n .

Final Analysis
Th e m od e of failu re was seve re abrasive wear. The ca use of
fa ilure was in suffi ci en t s urface ha r dnes s . A qu esti on a rises
co nce rni ng w ha t ca use d th e low ha rdn es s .

60
l oot ra dius
""'-
o 50
a; e:
Fr ~ r0-
~

~u 40
o t
c:: Midprofil e
<Ii
~ 30
c
"0
rn
J: 20

10

0 .010 0 .020 0.030 0.040 0.050 0 .060 0.070 0 .080 0 .090

Dept h below surface . in.

Fig. 6-9(b). Case hardness tra verse used to diagnose th e

problem of th e h ypoid pinion shown in Fig. 6-9(a).
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
218
The basic quenching was acceptable, because the c ore
hardness and microstructure at the root are indicative of a well
quenched and tempered product. Could grinding of the teeth
have caused this drop in surface hardness? It might have on the
surface of the active profile, but not on the surface of the root
radius that had not been ground. Also, grinding would not have
influenced the entire depth of case or the core hardness at mid-
point. There appears to be only one logical answer: A delayed
quench would ha ve allowed the surface and smaller sections to
cool somewhat below the critical quenching temperature yet
still ha ve been quenched soon enough to have allowed the core
at the tooth root and the more massive portions to be above the
critical quenching temperature.
Further investigation revealed the possibility that a tray
loaded with pinions had been stuck midway into the quench for
a time until the mechanism had been manually dislodged. Thus,
the primary cause of failure was improper heat treatment.
The documented records showed the total number of pin-
ions placed on each furnace tray . A service recall for the remain-
ing pinions was immediately initiated. The instructions were
simple: A common file could be used to quickly identify each
soft pinion without removing or disassembling any uni 1. The re-
call was successful and all the pinions were accounted for.
 CHAPTER 7
Writing the Report

No single format for a report can serve as a working model for

all writers or all readers. With certain or ganizations, one may
be required to follow a specific format when writing or submit-
ting a report, though generally this is not the case . It is usually
through gi ve-and-take tha t a forma tis concei ved tha t is accept-
able to both writer and reader.
A re po rt re flee ts the pe rsonali ty of the wri ter as we 11 as his
objecti ve. A report may be factual and bare, written by someone
who hates to write. It may be detailed to the " n th " de gree , flow-
ery with nonessentials, written by someone who enjoys his own
writing a bit too much. Or it ca n be concise and complete, mak-
ing its points with clarity, written by someone who has a gen-
uine ability to communicate.
To communicate properly is an art, but one that can be
learned. For instruction in learning to write a report, the guide-
lines discussed by Harry E. Chandler in Chapter 10 of the Te ch-
nical Writ er's Handbook] are highly recommended.
Since a systematic approach to analysis has been continu-
ally advocated in the preceding chapters, it is no less important
tha t report wri ting be kept in the same perspecti ve. From the
onset of a failure in the field operation, there are five occasions
that require written accounts . Each is very important to the fi-
nalization of the incident and each has its unique purpose. In
order of their usual occurrence, they are: field report, transmit-
tal report with failed unit , laboratory notes of the analyst, fail-
ure analysis report, and letter of transmittal of the report to the
cus tnrner-.
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
220
Field Report
Before proceeding with the field report. review very closely
the instructions given in Chapter 3 concerning field examina-
tion . The field report should document all facets of the examina-
tion and answer all questions raised in that section.
Following a particular format in the field report' is not as
important as making a record of all information pertinent to the
field failure . Of course. it would be most helpful to have a well-
written. well-documented field report. and this should be the
II goal of the field representative. This leads directly into the sec-
ond document.

Transmittal Report With Failed Unit

The transmittal report is an opportunity for the field repre-
sentative of either the user of the original equipment or the
supplier of the equipment (who also might be the original
equipment manufacturer) to interpret the field report. to place
it in coherent order. and to add detailed information that may
not have been available to the field service personnel. Such in-
formation might be the model and serial numbers of the equip-
ment. the purchase date. the hours or mileage of operation. the
maintenance history of the failed unit . and other related facts
not documented in the field report. If necessary. the field repre-
sentative should consult with the field service personnel (or the
maintenance personnel) who submitted the field report. in
order to establish all facts pertaining to the specific incident
and to submit a complete report. This report will also inform the
vendor as to the reference numbers to the incident. what the fail-
ure appears to be. what (if any) is the probable cause. what is
expected in return [i.e.. a report. an evaluation. a warranty
claim . credit. etc.). and the name ~f the person to receive the
correspondence pertaining to the incident.

laboratory Notes of the Analyst

The laboratory notes are the accumulation point for all
previous information and for all forthcoming observations that
may have any bearing on the failure incident. The first recom-
 WRITING THE REPORT
221
mendation is to use a permanent notebook; loose leaves of paper
are of no value if misplaced or lost.
Start the entry with the date of receipt and quickly add the
examination reference number, the customer's name, and the
part numbers involved. Continue with the background infor-
mation and cros s -refe rence numbers. From then on, record all
observations made while following closely the procedures out-
lined in Chapter 3, under "Visual Examination," "Ph ys ica l Ex-
amination ," and "Metallurgical Examination ."
Do not hesitate to note any condition or cha racteris tic that
is observed. At the time of the observation, such an item may
not seem pertinent , but before the final analysis it may have a
profound rela tionship to the fail ure. And even if an observa tion
proves to be incidental, no harm has been done by including it in
the notes.
The examiner may be a labora tory technician or the analyst
himself. The examiner's notes may be profuse or abbreviated,
but they must be an accurate record of what was observed. Ad-
mittedly, two examiners may interpret an observation dif-
ferently, but the physi cal observation should be recorded in a
reasonabl y recogniza ble manner. The notebook should be com-
piled and written such that another analyst can determine from
the data what was done and how the information was obtained .
It should be a clear and complete record for use by any expe-
rienced analyst.
Although the analyst 's actions of procedure should be sim-
ilar to those gi ven in Chapter 3 , he must not los e sight of two
ver y important mental inputs. The field report and the trans-
mittal report may have initiated a thought of a possible ca use
and effect. He should not dismiss this thought immediatel y but
cautiously proceed to establish facts which may, in themselves,
either strengthen or eliminate the preconceived input. In addi-
tion, as the analyst proceeds with the routine of visual, physi-
cal, and metallurgical examination, he may think of possibili -
ties for cau se and effect. Each one of these thoughts should be a
motivation for continu ed observation until a final conclusion
can be made.
There is an importance attached to th e examiner's labora-
tory notes that has not often been recognized; namely, their le-
gality. Laboratory notes are often used as court evidence; but to
-
SYSTEMATIC ANAlYSIS OF GEAR FAILURES
222
be admissibl e, th ey must be documented, and to be correctly
documented, each page must be dated and initialed by the ex-
aminer at the time the page has been completed.

Failure Analysis Report

The failure analysis report is the crucial report. The analyst
must place all observed data in context and present the data in a
scholarly manner that will be understandable, helpful to the
customer, and conclusive about mode and cause of the failure .
Although there are many formats used for these reports, the fail-
ure analysis report must incorporate certain basic elements.
Therefore , the following format is suggested: a heading, fol-
lowed by general background; examination procedures and re-
sults; and a closing.

The heading should be conci se, giving only the information

necessary for qui ck and efficient reference ; namely, report
number, date , customer's name, and part or parts involved.

General background should briefly state only the informa-

tion known to have had a bearing on the incident. This includes
the cus tomer's reference numbers, the operating data from the
transmittal report, the c us tome r 's reasons for returning th e
failure, and th e feedba ck required.

Examination procedures and results include the require-

ments established in Chapter 3 under the headings of visual,
physical, and metallurgical examination. The visual examina-
tion coverage may be the most detailed due to the nature of th e
observation; whereas, only the results of the physical and met -
all urgical examina tion may be reported. Since this is the main
body of the report, it is essential that enough informati on be
given to allow the reader to follow a logical path through the
procedures to arrive , with the analyst, at the same result. The
entire body of this report should be in a step-b y-step, logi cal
sequence of data presentation, written concisely , clearly , and
accurately. It should require a minimal amount of writing and a
minimal amount of reading .
 WRITINGTHE REPORT
223
The closing of th e rep or t is n ot a routine a p pe nd age , but it
most d efinitel y fulfill s a requirem ent. The cl os in g ca n ta ke th e
for m of a s umma ry, d is cu s si on , co ncl usio n. recomm endati on ,
or eva l uatio n. It ma y be on ly one of th es e or a co mb ina tio n of
t w o or more. The au th or mu s t keep in mind the ob jec tive of the
re port a nd us e th e closing th a t fulfills th e need of th at ob jec tive .
He s ho uld also kno w w hic h of the clo s ing form s w ill be co vered
b y th e letter of transmittal.
A s um ma ry is a sh ort en ed ve rsion of th e important
thoughts from the bod y of th e report , such as th e obj ective, the
most sig nif ica n t findings, and th e main id eas pres ent ed. A
summary used alone as a closing should appear when no con -
clusion ca n be drawn from th e results, or when the purpose is
only to present data . A summary is generally us ed to precede
one of th e other four closings, to set the stage in a logical order
for a proper finale .
The purpose of the discussi on is to g ive th e author an op-
portunit y to anal yze the si gnifi cance of the result s. Keeping in
mind th e obj ecti ve of th e re por t and the int erest of the reader,
th e a u th or ma y expand on th e fa ct s of the incident , th e observa-
ti ons made during the procedures , and the va ria tio ns within the
d ata . A com pa ris on w it h o ther in cidents m a y be made at this
time . as well as a directi ve for possible futur e observations or
tests.
Conclusions are th e rea soned judgments made on the basis
of th e results and findin gs of the ove ra ll examination. A concl u-
sion must always be supported b y th e data in th e report. It must
be written clearl y and concisel y, and if there are several conclu -
sions, they must be given in a logi cal order. The analyst should
keep in mind that a negative conclusion is still a valid con-
clusion.
Recommendations should be based on the conclusions.
Again, recommenda tions should be short and to the point. They
should be clearly understood by the reader and contain answers
to his questions . In other words , the objective of the report is
answered by the recommendation . It is also under this heading
that corrective measures are suggested.
An evaluation is a type of di scussion that specifically de-
termines the si gnificance or net worth of the results obtained
-
SYSTEMATIC ANALYSIS OF GEAR FAILURES
224
during the examination. It discusses the effectiveness of an-
swering the "why" questions brought up by the customer in
relation to the incident; and it discusses the relative merit of var-
ious corrective measures that could be recommended. In gen-
eral, the evaluation is not the closing portion of the report , but it
is the main context for the letter of transmittal of the report to
the customer.

Letter of Transmittal
The shortest letter of transmittal will be, for example:" Ac-
companying this letter is our report No. 750, which is self-
explanatory." The assumption of this letter is that report No.
750 is complete regarding conclusions, recommendations, and/
or evaluation. However, if the report is not complete in these
respects, the letter of transmit tal should be expanded to incl ude
the necessary closing. The author of the report and the author of
the letter of transmittal should be in communication before the
report is concluded, to determine who is to do what about the
closing.
It is essential to be realistic, first concerning the responsi-
bility of the analyst who is also the author of the report. If the
analyst also authors the letter of transmittal, the report should
close in a logical manner and the letter used to expand or em-
phasize. If the analyst is not the head of the department, but the
letter of transmittal is to be sent by the department head , the
analyst may close his report to answer the objective and allow
the letter to range from one simple sentence to a lengthy evalua-
tion , with recommendations as necessary. If the letter of trans-
mittal is to be sent by a member of the sales department, in most
instances the report must be complete in every detail.
Also, the author of the letter of transmittal must read the
report carefully. Ha ve the objecti ves been answered? Check for
conclusions and recommenda tions. Are they clear and concise?
Are they in logical order? Are they clearly supported by the
results in the body of the report? Are there any other conclu-
sions or recommendations that come to mind as you study the
report? If there need to be addi tional emphases or expanded rec-
ommendations, the letter of transmittal is the only means
available to bring the incident to a final conclusion.
 WRITING THE REPORT 225
T he lett e r of tr an sm itt al se rves a very im p ort ant f u nc t io n
un iqu e a mo ng the fi ve d ocument s . Th e four pre vious acco unts
co ns ide r onl y th e fa ct s of t he in ci d en t a n d are n ot u su all y cog -
ni zant of th e ind iv id ua l read er . N ot so w i th th e lett e r of t ran s-
mitt al. w h ic h is wr i tte n t o a n in d i vid ua l a nd is , in m os t in-
sta nces, a pe r s on al lett e r. It is th erefore neces s ar y fo r th e
a u th or of the let te r of t ran sm it tal to b e aware of th e n ee d s of the
reci pie n t a n d ve ry ca re fu lly su p p ly a ll info r ma tio n neces s a r y to
sa tis fy th at indi vidu al. Fo r thi s rea s on , th e lett er s ho uld be au-
th ored b y th e pe rs on [m et allurgi st , e ng ineer, or sa les perso n ]
m ost cl ose ly assoc ia ted w it h th e reci p ie n t of th e rep or t.

Ethical Overview
An obse r ve r ma y of te n be in fl ue nced b y p re vi ou s obse rva -
ti on s or b y a ce r tai n des i re to see thing s th at s im p ly are not
there . For thi s reas o n, it see ms ex pe d ie n t to d isc uss so me p oin ts
of e thi cs [or ju st g ood e t iq ue tte ] n ot as c ri ticis m, but t o em p ha -.
s ize c ha racte r . In th e ac t of s t udy ing a failed part a n d in p a s sing
in format ion on to o the rs , it is of utmost imp ort an ce tha t th ree
ch a ra c te ris tics a re k ept in pl a in vie w : acc ura cy , cl a rit y, a n d
la ck of b ia s .
A cc ura c y is in th e eyes of th e behold e r. Wh at e ve r is see n
s ho uld be ex p la ined in clea r a n d si m p le d et a il. und e r st and a bl e
to th e reader. A ca tas tro p h ic fa ilure in th e fi el d of o pe ra tio n
req uires a firs t- ha nd rep ort fr om th e o pe ra to r. T he fie ld pe r-
so n ne l exam i ni ng a fa ile d un it need s to accura te ly p ort ra y w ha t
h e sees . Th e field re pres ent at i ve mu st be exac t in s ub m i tt ing
de ta i ls to th e a na lys t. Th e a na lys t mu st m ak e a si ncere effor t to
piece th e sto ry togethe r a nd run e no ug h test s to de ter m ine a
clea r-c u t a nswer.
N o ac tio n is m ore e m b a rrassi ng th an giv ing a c usto mer a
firs t-s ig h t o p in io n, a llowi ng h im to acce p t it as fa c t, a nd lat e r
fin ding th at th e "obvio us" was al l w ro ng. Th e exam iner s ho u ld
be ce r ta in tha t e noug h in fo rma tio n is re a dil y ava ilab le to be
co nvinc ing both to him self a n d to hi s c us tome r .
Clarity is th e q ua li ty of b eing cl ear , free fro m co nf us io n,
a nd s im p ly und e r s tan d a bl e. Th e ai m of th e a u thor of a n y of th e
p rint ed do c ume n ts d isc ussed ea rlier s ho uld be cla ri ty, n o l on ly
for th e a ut hor's benefi t, but fo r th e benefi t of those pers on s ye t
 SYSTEMATIC ANALYSIS OF GEAR FAILURES
226
to read the account. In field reports. the language might be gross
and the grammar atrocious, but there must be no doubt about
the clarity of information. A failure analysis report may by ne-
cessity contain metallurgical or engineering terms not wholly
understood by the layman; but the discussion section of the re-
port should make them understood . It is detrimental to the clar-
ity of a report for its author to attempt to "blind people with
science ."
As is mentioned in Chapter 5, when a gear set failed by
tooth bending fatigue due to sudden overload, the analyst ad-
mitted to having thoughts somewhat biased since he had re-
cently watched this same model front -end loader being used
abusively. Another type of bias occurs when an analyst wants
to find a cause in one direction or, perhaps, does not want to
admit the direction of an apparent cause. Any biased opinion is
an immediate barrier to true evaluation and corrective action.
An analyst must be aware of the incident before him and
careful not to introduce irrelevant information that will confuse
the issue at hand. A dimensional or process characteristic that
may be slightly out of specification, having nothing to do with
the incident, should not be stressed beyond its immediate im-
portance. A specification is established by people using their
best intentions to comply with the existing requirements.
A cause of failure should never be that it does not meet the
specifications; but rather, the cause should be discussed in light
of these questions: Would the failure also have occurred if the
specifica tion had been met? Should the specifica tion be changed
to accommodate the existing situation, or should the situation
be changed?
From the discovery of the failure to the final letter of
transmi ttal, the line of comm unica tion should remain open, ac-
curate, and sincere. There is no other way to achieve an accept-
able plan for corrective action.

Reference
1. Harry E. Chandler, T echni cal Writer's Handbook, American So-
ciety for Metals . 1983 , p 111-133.
Index

NOTE. Th e pa re n the tica l le tt ers [F) a nd [T ] indicat e

th at informa tion is prese n ted in a figure or a table.

A C
Abrasive wear [F). 121 -122 Ca r bo n grad ien t tr av erse [F). 65 . 71
mod e of failure in h ypoid pin ion . 217 Case/co re co mbi nation
Adh esiv e wea r IF). 123 -125 high core ha rdness. 160
Ali gnmen t [Fl . 36-37 la rge case depth . 160
of parts [F). 170-1 71 sho t peening . 160
Assem bly s ubzero trea tme nt. 160
in corr ect [F). 169 -17 0 tors iona l s hear of shaft, 160
mi salignment. cause of fa ilu re in spi- tw o mod es of fa ilur e (Fl. 119- 120. 124.
ra l bevel gear. 193 159
reno vation a nd reco nd itio ni ng of. 170 Ca se crus hing (Fl. 48 . 118-1 20
Ass embl ym en. con ta mi na ti on. 38 fac tors need ed (F). 118-11 9
Atmospheri c co nta mi na tio n. 32 Case depth. 74-75
case proper ty. 155-157
B
la rge in case/co re comb inatio n fail ur e.
Background info rm at ion. of fa il ure a na l- 160
ysi s rep ort. 222 recomm ended s ta nda rds [T). 156
Backlash. 18. 20 Case ha rdnes s tr averse . 72-7 5
Ban ding of flow lin es (Fl . 134-136 for carb urize d gear tooth (Fl . 72-7 4
Bearing failure. 181 for inductio n hard en ed gear too th (Fl .
Bending fati gu e 74-75
fracture a ppeara nces of fa tig ue fail - Case propert ies [see olso Case/core co m-
ure s [T) . 112 bin ati on failu re)
in shafts a nd pin ion s ha nks (F). case depth [T}, 155-1 57
112-114 microst ru cture (Fl. 157- 158
mode of fa ilure in spi ral bevel gea r. s ubs urface s tresses a nd ha rdnes s gra-
193 di ent s neces sa r y to prevent s ub-
mode of tooth fail ure in s pira l be v el surface fa il ure (F). 154-155
dr ive set. 196 s ur face ha rdness. 155
Bending stren gth . 10-11 Cau ses of failu re
Blowholes . in s tee l, 130-1 31 ba sic mat eri al, 130-139
Blue brittlen ess, 165 engi nee ri ng. 140-147
Bull 's-e ye fracture (Fl. 133 heat tr eatment. 154-167
Burning. See O verheating man ufa ctu rin g. 147-154
Butt erfl y wings [Fl. 102-105 service a pplica tion. 167-183
 INDEX
228
Chatter. See Vibrations En vironmental fa ct or s
Cl osing of fa il u re anal ys is re po rt lu b rica tio n. 28- 32
co nclusio ns . 223 mech an ical s tability . 36-3 8
d isc us sio n. 223 per sonnel-rel at ed activ it ies . 38 -40
eva lua tio n. 223 -22 4 tempera ture. 32-3 5
reco m me ndatio ns . 223 Ethi ca l ove rview. in failure a nalys is re-
sum ma ry . 223 port. 225-226
Co m posi te errors [F). 20 -21 Eva lua tio n. in closi ng of fail u re a na lysis
Co ncl us ions . in cl os in g of fa ilure a na ly - re po rt. 223 -22 4 .
sis re port. 223 Exa mination pr ocedu res and res ult s . of
Co ntact fati gue. See S palli ng fa ilu re a nalysis report . 222
Co n ti n ual overloadi ng . See Co n ti nuo us Externa l ru pt ur e [F ). 59. 126-1 27
ove rl oa d ing
F
Co n ti n uo us overloa d ing
a t co rne rs a nd underc u ts. 180 Failure a na lysis
ca us e of fail ure in s pira l bev el se t. 211 of broken s ha f t. 185-1 88
in rolling co ntac t fat igu e. 180 of h ypoid pi nion. 214 -218
in s ha ft s [F ). 53 . 179 of spira l bevel drive set, 194 -196
in s pi ra l be vel ge a rs [F ]. 48.1 20. of spiral bev e l gea r. 188-194
178-1 79 of s pira l beve l se t. 203-211
in s p ur gears [F). 179-1 80 of spur gea r. 200 -202, 211 -214
Copper s tr ip ping solutio ns . cor roaioi- of', of spur pinion. 197-200
167 Failure a na lys is rep ort. 222-224
Co pperp la ting. 165. 167 ba ck ground inf ormati on . 222
Core hardnes s . in case/ co re combi na tio n closing. 223 -22 4
failure . 160 e xam in at ion pr ocedures a nd result s .
Core propert ies [sec also Case/care com- 222
bi nation fa ilu re ). 158 heading. 222
Corners. co ntinuous ov e rloadi ng . 180 Fatig ue. 86
Cor ros io n (F ). 1 77 in s ha fts [F.T). 109 -11 4
Cut te r mark s [F). 150 -1 51 of round sp lined a nd ke yed bores . 109
pilling IF). 94-100
D rolli ng con ta c t fa tig ue ( Fl. 96 .100 -105
s palli ng IF). 105-108
Deflec tions . 170-1 71
th ermal fa tigu e (Fl . 106 . 108-109
Des ign
too th bendi ng fa tigu e [F). 86 -94
fail ure in . 140 -145
Field appli cation. 183
in ke yw a ys [F). 114 . 142
Field examina tio n . 44- 47
in s piral bevel gears [F). 93.142-1 43
primar y a nd seco nda ry fra c t ure . 46
in spur id ler gears [F). 141 -143
remo val of ass embly. 45
Dim en si on al chec k ing. 61
re vie w of parts. 46
Dis cussi on. in closi ng of failu re a na ly s is
us e of docu me n ta tio n, 44- 45
repo r t. 223
Field report . 220
Documentati on
Flame hard en in g. 163
in fie ld exami nat io n. 44 -45
Flow lin es
in vis ua l exa mina tio n. 4 7
banding [F) . 134 -136
{n s teel [F). 134 -136
E
in s tee l forg ing [F]. 66-67 . 135. 137
En gin eering For ei gn mat erial. in gear tee th or mo vin g
ca uses of failure . 140 -14 7 pa r ts of asse m bly [F), 175 -176
des ig n [F) , 140 -14 5 Forg ing la ps . in stee l forg ings (F) . 68 .
d esi gn failure ca use in s pira l be ve l se t. 137
206 Fo rg ings . Sl ~ ( ~ S te el for gi ngs
g rindi ng tol er a nces [F), 147 -148 Fri ct ion . 35
hea t t rea tme nt s pecifi catio ns. 147 Fri cti on a l e ne rg y. ef fec t on tem pe ra tu re.
mat erial se lec tio n [F). 145- 147 35
 INDEX
229
Frict ional hea l. See Ther ma l fati gu e Im pact
case crus hing [F) . 48 .118-120
G s tress ruptu re (F), 125 -127
Ge a r looth ope rati ng te m pe ra lure . 34 -35 too th ben d in g im pact [F). 112. 115-116
G rind ing. tole ran ces . 147-148 too th chipp ing (F). 118
G rind ing chec ks [F) . 59 . 126. 152-153 tooth s he a r (F) , 115-1 17
G round tooth gears . ni tal e tc h prod uc- tor sion al s hea r [F), 120-121
tio n test (T) . 61 w ear [F ), 120- 125
Im pact over loa d ing [F). 8 7. 91 , 1 10. 116,
H 118-119,1 21 , 180 -181
Induct ion ha rden in g, 16 1
Harden in g..161 cause of too th profile fa ilure in spur
na me ha rden in g. 163 gea r. 214
induc ti on hardening. 16 1 fail ur e in , 161-164
q uenc h crac ki ng. 161 Ingo tis m , in s tee l. 131
Ha rd ness Intern a l gea rs (F). 3
cross-sec t iona l survey [F) . 63 -65 In tern a l ruptu re [F). 58, 124- 125
eq uiva len t num be rs for s tee l (T ). 56 -57 in s tee l fo rgi ngs, 137-138
Head in g. of failure a na lys is rep ort . 222 In volute patt e rn (F), 13-15
Heat treatment
case/ co re com bina ti on [F ). 119 -120 . K
124.1 59 -160 Ke yed bo res. fr ac t ur e. 109
ca se prope rti es [F). 154- 158 Ke y w a ys , des ign fa ilu re [F), 114. 142
co p per st ri ppi ng so lutions . 167
co ppe rpla ti ng.165 -167 L
core prope r ties. 158 La borat or y not es of an a lyst. 220-222
follo w ing s pec ifica tions to pre vent lega li t y of. 221 -2 22
fail ure . 147 Lea d pa tt ern [F), 15- 17
ha rd ening. 161-164 Letter of tra nsm itt a l. in fa il ur e a nalysis
imp rope rly placed th readed plugs. 167 re por t, 224 -225
s t raig h te ning of pin ion sh ank s a nd Liq uid co n tami na ti on . 32
s ha fts (F). 166 -168 Lubri catio n, 28 -29
tempe rin g. 164-1 65 ca use s of fa ilure. 29 -32 ,1 73 -175
Heli cal gea rs (F ). 2-3 co n ta mina tio n a nd con ta m ina n ts [F].
a pp lied s t resses [F) . 7-9 31-32. 174 -175
int erna l gea rs [F) . 3 effec t on tempe ra ture. 33
H yp oid gea rs ex ces sive , 31, 174
a p plied s tresses. 10 ge ne ra l rul es . 29
se ts (F) . 5 incor rect , 29-30, 174
H yp oid pinio n. fa il ur e a nalysis lack of [F) , 30-31
bac kgrou nd inf ormatio n . 215 s yste m, 30
ca se hardnes s trave rse (F) . 216 -217
fi na l a na lysis. 217-21 8 M
ide n ti fy ing ma rks . 216 Machining fa ilur es [F ), 149 -15 1
magn et ic-p a rti cle inspecti on . 216 Ma cr oet ching [F). 65-69
met a llurgical exa mina tion . 216-217 Mag netic-p a rt icl e insp ec tion [ F). 53 -54.
micr oscop ic e xa mi na tion. 217 59
ph ysi cal a ppea ra nce . 216 Maint en an ce, 182
physi ca l exa m ina ti on . 216 per so nn el. eq uip me n t a nd proced ures ,
surface ha rd nes s [F ). 215 -216 38 -39
vis ua l exa minatio n. 216 Ma na ge ment a ttitu de , 39- 40
H ypoid se ts [F). 5 Manufa cturi ng
causes of fa il ure , 147 -154
gri nd ing c hec ks , bu rns (F), 59, 126.
Idler gears 152 -153
ph otoel a s tic pa tte rn [F). 22-23 heat treatme nt cha nges , 153 -154
~
I

INDEX
230
Manufacturing [co nt .] surface hardnes s te st ing , 55
tool undercutting", sh arp notches [F), tooth characteristics , 54
149-151 ultrasonic tes ti ng [F), 55, 57 -58
tooth characteristics [F), 151-1 52 Pinion sh afts
Material, causes of failure in steel, cross-sectional hardness su rvey, 63-6 5
130-139 Pinio n shan ks
Material sel ection, ca us es of failure , bending fati gu e [F,T) , 112-114 ,
14 5-147 167-168
Mating-tooth contact pattern, 17-18 cro ss- sect ional ha rdne ss s urvey [F),
Mech ani cal damage [F), 172-173 63-6 5
cause of failur e in broken shaft, 186 st raightening of, 167
Mechanical st ability, 36-38 surface crack ing, 167
a lig nme n t [F), 36-37 torsional fa tig ue [F), 120 -121 . 166-167
size cha nge, 36 -37 torsional shear (F), 120-1 21
vibrati ons, 37-38 Pipe, in st eel (F), 69 ,130
Metallurgical examination, 62 Pitch line, pitting [F), 96-98
ca rbon gr adient traverse (F), 65 , 71 Pitting (F), 94-100
ca se hardness traverse (F), 72 -7 5 in pit chline [F) , 96-98
ch em ical an alysis, 71 in spiral bevel teeth gears (F), 99-100
cro s s-sec tiona l hardness survey (F), initiation of [F), 94-9 5
6 3-65 tooth conta ct [F), 98-99
macroscopi c examination (F), 65- 70 Plugs , th read ed. improperly placed, 16 7
microscopic examination (F), 75-81 Profilometer measurements, 61
scanning ele ctron microscop y [F),
82-83 Q
Microstructure (F), 75-81 Quench crac king, 161
case property [F), 157-158
R
Modes of failu re
fatigue , 86- 112 Recommendations , in closing of failure
impact, 112-120 analy sis report, 22 3
stress rup t ure, 125-127 Report writing
wear, 120 -125 ethical over view , 22 5 -226
Mount ing distance . See Settling fa il ur e a na ly sis report, 222-224
field report. 220
N laborat or y notes of anal yst , 22 0 -222
Nit al etching [F), 57-61, 65 , 70 letter of transmittal, 224-225
production test for ground tooth gears systemati c reporting, 219
[F) ,61 transm ittal report with fail ed unit , 220
Nonmetallic inclusions, in steel [F), 58 , Rolling conta ct fatigue [F). 96 , 100-105
95 ,106 ,1 31 -134 butterfly wings [F), 102
Notches, sharp, 149-151 cha rac teris tics of [F), 102-105
continuous overloading , 180
o mode of failure in spiral bevel se t. 206 ,
Op erators of equipment, 38 , 182-183 210-211
Overheating , 152-153 mod e of failure in spur pinion, 199
in steel for gings [F). 138-139 Rosette fr act ure [F), 111
Overloading Round bores , 21-23
ca use of tooth failure in spiral bevel fr acture, 109
drive set , 196 mod ifica tions to after stress. 22-23
p Rubbing cont act. See Fri ction

Personnel-related activities , environ- S

mental factors. 38-40 Scanning electron microscope [F), 82-83
Physical examination, 53-61 Seams. in ste el, 136
dimensional checking, 61 Segregation, in steel, 131
magnetic-particle inspection (F), Sel ect ive hardening. See Induction
53-54 , 59 hardening
nital etch ing [F). 57 . 59-6 1 Service applications , 167-169
 INDEX
231
vibration, 169-173 ph ysica l exa mina ti on , 192
bea ring failure, 181-18 2 visual exa mi na ti on , 190-191
con ti n uous ove rl oa d ing [F}, 177-180 Spiral bev el gears [F). 3-5
cor ro sion (F). 177 a p plie d stresses . 9-10
field a pplic a ti on , 183 conti n uous overloading (F). 48, 120 ,
foreign material (F). 175-177 178-179
imp act overloading [F}, 8 7, 91,110 , design failu re [F). 93,142-14 3
116 ,118-119 ,1 21 ,180-1 81 mating-tooth con tact pattern , 17-18
lubricati on. 173-175 pittin g on tee t h [F], 99-100
mainten an ce , 182 Spiral bev el set . failure a na lysis of
me ch anical damage (F). 172-174 ba ckground information , 202, 207
operator er ror, 182-18 3 ca se depth, 204
set mat ch in g (F). 19 , 169 ca se ha rd ne s s, 204
Set mat ching, mismat ch ing, 169 case hardne ss tr av erse [F], 210
Settling, 171 cas e structure (F). 205
Sh aft , broken , fa il ur e analysis of core hardness . 204
ba ck ground informa tion, 185 cor e st ruc t ur e, 205
field a pplica tion, 185 final anal ysi s, 206, 210-211
met allurgic al exa mina t ion, 186- 187 hardness , 204
visual examination [F] , 186-187 ident ifying mark s . 203, 207
Shafts, 24 magn eti c-parti cle in sp ect ion , 204, 209
bending fatigue (F .Tl , 112-114 met allurgical exa mi na ti on, 204-205,
co n tin uous overloading (Fl. 53, 179 209- 211
tors ional fatigue (F). 109-111 micro scopic examination, 209- 210
torsiona l shear [F] , 120-121 ph ysi cal a ppea ra nce of gea r, 203-204,
tor sional she ar in ca se /cor e combina- 209
ti on fail ure , 160 physical a ppea ra nce of pini on (F).
Sh ear tears, in steel forgings , 138 202 -203. 207-208
Shot peening, in case/c ore comb ina tion physic al exa mina tion, 204, 207
failure, 160 surface hardnes s, 209
Size cha nge , 36-37 to oth cha ra cte ris tics , 209
Solid con ta mina nts (F). 32-33 visu al ex ami nation , 203-204, 207-209
Spall ing (Fl , 105 -108 Spiral bevel sets [F] , 3-5
mode of too th profile fa il ur e in spu r Splined bores, 23 -24
ge ar , 214 fract ure , 109
Sp ir al bevel drive set, failure an aly sis of Spur bev el gea rs , des ign failure [F], 93,
back g round info rmat ion , 194 142-143
field appli cation, 194 Sp ur gea r, failu re ana ly sis of
final anal ysis , 196 background informati on , 200, 211 -212
identifying marks , 194 chemica l a na lysis , 213
magn eti c-p article in sp ection, 195 fi na l a na lys is. 201-202 , 214
met allurgic al exa m ina tion, 195-196 id entifying marks , 200 , 212
ph ysical appearance of gea r, 195 ma cr os cop ic examin ation, 213
ph ysi cal a ppeara nce of pinion , 194 magneti c-p article inspecti on , 213
ph ysi cal exa mi na t ion , 195 met allurg ical exa mina t ion, 201, 213
visual exami na ti on, 194-195 nit al e tching, 213
Spiral bev el gear, broken, failu re a na ly - ph ys ical a ppeara nce (F). 200-201 , 212
sis of ph ysi cal ex a minat ion , 201 , 213
background in form ati on a nd custo- surface hardness , 213
mer's eva lua ti on [F), 188-189 visual examina t ion . 200- 201 , 212
field application , 188 Spur gea rs [F}, 1-2
final anal ysis, 193-194 a p plied s tr esses (F). 7
id entifying ma rks , 190 in ternal gea rs (F}, 3
ma gn eti c-par ticl e in spection, 192 Spur id ler gea rs
metallurgical exa m ina t ion [T] , 192-193 des ign fail ure [F). 141-143
nit al etc hing, 192 Spur pinion, failure a na lysis of
ph ysi cal appea rance of ge a r [F}, ba ck ground informa ti on [Fl . 197-198
189-191 ca se hardness tr a verse, 199
physical appeara nce of pinion, 190 che mica l a na lys is , 199
 INDEX
232
Sp ur pini on , fa il ur e a na lys is of [con L) T
fina l a na lys is , 199 -200 Temper color, 35
ide n tify ing mark s , 198 Temper ature, 32-33
magn et ic-p a rti cl e in spec tion , 198 eff ect of ambi ent temper atures of s ur -
me tallurgical exa mi na tio n , 198 -199 round ing at mos phere, 34
mi cr os tru cture, 199 eff ect of fri ctional ene rgy , 35
phys ica l a ppea ra nc e, 198 effect of lubri cati on, 33
ph ysi cal exa mi na tion, 198 effec t of risi ng te mp er ature s [F] , 34 -36
s urface ha rdness , 198 gea r tooth ope ra ting tem pe ratures ,
tooth charac teri s tics , 198 34-35
vis ua l exa mina tio n, 198 T emp ering , 35, 164 -165
S tati stical P ro cess Contro l, 149 britt len ess, 165
S tee l T he rm al fa tigu e [Fl, 106, 108- 109
bl owholes, 130 -131 To ol und er cutting [F) , 149-1 50
ca uses of fa ilure in, 130-139 Too th bearings [F) , 18-19
flow li nes [F), 134 To oth bend in g fati gu e [F ], 86 -89
ha rd ness numb e rs [T ), 56-57 va ria tions in [F) , 89 -94
in goti sm , 13 1 Too th be nding impa ct [F), 112, 115- 116
nonmet a lli c inclu sions [F) , 58 , 95, 106 , T ooth charac te r is tics , 12-20
131-1 34
ca uses of manufacturing fa ilure (F),
pip e [F), 69 , 130 13,1 6-17 ,106 ,151-1 53
sea ms , 136
stud y of in phys ical exa mina tion, 54
seg rega tio n , 131 Toot h chi ppi ng [F ), 118
S te el cas tings. ca uses of failure in (F) , To oth co n tac t, pitt in g [F), 98 -99
31,36 ,138-139 Tooth s hea r [F) , 115-117
S teel forgin gs T ooth tip int erfer en ce, ca use of fa ilure in
ca uses of failure in , 136-139 s p ur pini on , 199
fl ow li nes [F) , 66 -67 ,135, 137 Tooth-to- tooth runn ing patt ern [F),
forging laps [F], 68,1 37 12 c 13
inte rna l ruptu re, 137 -138
To r si onal fatigu e
ove r hea ting [F) , 138 -139
in s ha ft a nd pinio n sh ank s [F) , 109- 11 1
sh ear tea r s, 138 mod e of fa ilure in b rok en s ha ft, 186
S t ra ig ht be vel gea rs (F ), 3 -4
T ors io na l s hea r
a pplie d s tresses , 9
of s ha ft in ca se / co re combina ti on fail-
St ress di s tr ibut ion lines [F] , 144 ure, 160
Stress r up t ure of s ha fts a nd pinion sh ank s [F),
e x te rna l rup ture [F), 126-127 . 120-1 21
inte rna l rupture [F ), 58 , 124-1 25 T o rsiona l s t re ng th (F) , 11-1 2
St resses, 6- 10 T or si on al vib ra tio ns . Se e Vibra tions
Su bz ero treatmen t, in ca se / core co mbi- Tran smitt al le tt e r. See Lett e r of tra ns-
nati on failure , 160 mitt al
S ummary, in cl osing of failure a na lys is Tran smittal report wit h fail ed uni t, 220
rep ort , 223
S urface con tac t fa tigu e. See Pi tti ng u- w
Surface hardness
case p rope r ty, 155 U ltraso ni c test in g (Fl, 55, 57-58
ca use of fa ilu re in h yp oid pinion , 217 U nde rc u ts , cont in uo us overl oading, 180
testing of h ardness vs minimum Vibration s [F), 37 -38 , 171-172
th ick ness [T) , 55 mod e of fa ilure in s p ur gea r, 200- 202
S ystemati c exam in ati on Visu al e xami nati on , 47-5 3
a p proac he s , 43- 44 use of low-powe r mic roscop es , 51
field e xa mina tion, 44- 47 use of obs e r va tio n a nd logi c, 49- 50
met allurgi ca l e xa mina ti on. 62-83 Wear
physi cal exa m ina tio n , 53 -61 ab ra sive wear [F), 121-122
visu al e xa m ina tion , 47-53 a d hes ive wea r [Fl, 123 -125