www.ijmme.

orgInternationalJournalofMaterialandMechanicalEngineering(IJMME)Volume3Issue1,February2014
DOI:10.14355/ijmme.2014.0301.02

FiniteelementAnalysisofHorizontalaxis
WindturbineGearboxFailureviaTooth
bendingFatigue
M.Grujicic*1,R.Galgalikar,S.Ramaswami,J.S.Snipes,V.Chenna,R.Yavari
DepartmentofMechanicalEngineering,ClemsonUniversity
Clemson,SC29634,USA
*1

gmica@clemson.edu

Abstract
Wind energy is one of the most promising and the fastest
growing alternativeenergy production technologies which
have been developed in response to stricter environmental
regulations, the depletion of fossilfuel reserves, and the
worldsevergrowingenergyneeds.Thisformofalternative
energy is projected to provide 20% of the US energy needs
by 2030. For economic reasons, wind turbines (articulated
structureswhichconvertwindenergyintoelectricalenergy)
are expected to operate, with only regular maintenance, for
at least twenty years. However, some key windturbine
components (especially the gearbox) tend to wear down,
malfunction and fail in a significantly shorter time, often
three to five years after installation, causing an increase in
the windenergy cost and in the cost of ownership of the
wind turbine. Clearly, to overcome this problem, a
significantincreaseinlongtermgearboxreliabilityneedsto
be achieved. While purely empirical efforts aimed at
identifying shortcomings in the current design of the
gearboxes are of critical importance, the use of advanced
computational methods engineering analyses can also be
highlybeneficial.Thepresentworkdemonstratestheuseof
the finite element analysis in modeling and elucidating the
root cause of one of the gear failure modes (i.e. tooth
bending fatigue) under a variety of normal operating and
extremewindloadingconditions.
Keywords
Horizontal Axis Wind Turbine; Gearbox Reliability; Tooth
bendingFatigueFailure

Introduction
The main objective of the present work is to address
theproblemoflongtermreliabilityandthemodesof
failureofgearboxesusedinwind(energyharvesting)
turbines. Consequently, the concepts most relevant to
the present work are: (a) windenergy harvesting; (b)
windturbine gearbox reliability; and (c) root causes
and main modes of gear damage and failure. In the
remainder of this section, a brief description is

providedforeachoftheseconcepts.
WindEnergyHarvesting
The depletion of fossilfuel reserves, stricter environ
mental regulations and the worlds evergrowing
energy needs have led to deployment/ utilization of
various alternative/renewable energy sources, among
which windenergyis one of the most promising and
the fastest growing installed alternativeenergy
production technologies. In fact, it is anticipated that
by2030,atleast20%oftheU.S.energyneedswillbe
metbyvariousonshoreandoffshorewindfarms.
A wind turbine is essentially a converter of wind
energyintoelectricalenergy.Thisenergyconversionis
basedontheprincipleofhavingthewinddrivearotor,
thereby transferring wind power to the electrical
generator. To attain greater structural stability of the
rotor and a high value of aerodynamic efficiency, the
rotor is usually constructed as a set of three
aerodynamically shaped blades. The blades are
(typically) attached to a horizontal hub (which is
connectedtotherotoroftheelectricalgenerator,viaa
gearbox/drivetrain system, housed within the
nacelle).Therotor/hub/nacelleassemblyisplacedona
tower and the resulting wind energy converter is
referred to as the Horizontal Axis Wind Turbine
(HAWT).
To reduce the energy production cost (typically
expressed in $/kWhr), commercial wind turbines
havegrownconsiderablyinsizeoverthelast30years.
Thelargewindturbineeconomicsisbasedonthefact
that as the hubheight/windturbine rotor radius
increases, the average wind speed/wind energy
captured increases due to the so called wind shear
effect. Consequently, for the same energy production
level,lessernumberofwindturbineunitsarerequired,
which in turn leads to a reduction in the cost of

InternationalJournalofMaterialandMechanicalEngineering(IJMME)Volume3Issue1,February2014www.ijmme.org

spur (the present case) or helical gears. In this

configuration, the planetarygear carrier is driven by
the windturbine rotor, the ring gear is stationary/
reactionary, while the sun pinion shaft drives the
intermediate gearbox stage, and, in turn, the high
speed stage. Typically, both the intermediate and
highspeed stages are composed of helical gears (the
damage and fatiguefailure of which is the subject of
thepresentmanuscript).

operationofthefarm.Asthesizeofthewindturbine
rotor increases, the structural performance, durability
and dynamicstability requirements tend to become
moreandmorechallengingtomeet.
Turbinebladesandthegearboxareperhapsthemost
criticalcomponents/subsystemsinthepresentdesigns
of wind turbines. The present work deals only with
the issues related to the performance, reliability and
modesoffailureofgearboxcomponents.Inourrecent
work [Grujicic et al. (2010ab)], twolevel multi
disciplinary designoptimization methods and tools
were developed for determination of the optimal
shape and size of glassfiber reinforced epoxymatrix
compositeHAWTblades.
WindTurbineGearboxReliability
Windturbinegearboxfailureremainsoneofthemajor
problems to the windenergy industry [Musial et al.
(2007)].Therootcausesofgearboxfailureintheearlier
designsareassociatedwiththeproblemsrelatedto:(a)
fundamental design errors; (b) manufacturing
deficiencies;and(c)underestimationoftheoperating
loads. While these problems have been mainly
eliminated over the last 20 years, windturbine
gearboxesstillgenerallyfailtoachievetheirdesignlife
goal of twenty years. The combination of these high
failure rates and the high cost of gearboxes has
contributedto:(a)increasedcostofwindenergy;and
(b) higher sales price and cost of ownership of wind
turbines. Clearly, to make wind energy a more viable
renewableenergy alternative, the longterm gearbox
reliabilitymustbesignificantlyincreased.

FIG.1SCHEMATICOFAPROTOTYPICALWINDTURBINE
GEARBOX.THEMAJORCOMPONENTSANDSUBSYSTEMS
AREIDENTIFIED.

MainModesofGearDamageandFailure
Postmortem examination of the field windturbine
gearboxes revealed two main modes of gear failure
[Fernandes(1996);FernandesandMcDuling(1997)]:
(a)toothbendinghighcyclefatiguefailureThedefining
features of this geartooth failure mode can be
summarized as: (i) Fatigue cracks are typically found
tobeinitiatedattherootradiiontheengaged(loaded
in tension) side of the gear teeth; (ii) Cracks tend to
originatepreferentiallyatthelocationsassociatedwith
the largest principal stresses. Under normal loading
conditions, the highest stresses are typically found at
the tooth base, while under abnormal loading
conditions(e.g. in the case of gear misalignment), the
location of the highest stresses is related to the
characterandextentofloadingabnormality;(iii)Crack
growth is generally characterized by an Lshaped
trajectory,i.e.thecrack,nucleatedattheengagedside
of the tooth initially propagates inwards and below
thetooth,andthenmakesaturnoutwardtowardsthe
opposite side of the same tooth (leading to the tooth
breaking off); and (iv) Due to the subsequent
overloading effects, failure of one tooth is often
accompanied by failure of adjacent teeth, in rapid
succession;and

The current state of understanding of the basic

features and processes/mechanisms related to the
failureofwindturbinegearboxescanbesummarized
as follows [Musial et al. (2007)]: (a) gearbox failure
appears to be of a generic character, i.e. not strongly
related to the differences in their design; (b) gearbox
failure cannot be generally attributed to poor work
manship; (c) gear failure is frequently the result of
excessiveandunexpected(e.g.misalignment)loading
conditions. In other cases, the gearbox failure may be
initiated in overloaded bearings, and the resulting
bearing debris propagate to the gears, causing tooth
wear and gear misalignment; and (d) the essential
features and mechanisms of gearbox damage and
failure appear not to change with size of the wind
turbine.
A labeled schematic of a prototypical wind turbine
gearbox is shown in Figure1. The lowspeed stage of
the gearbox is a planetary configuration with either

(b) surface contact fatigue failure Depending on the

characterandspatialdistributionofthestressesatand

www.ijmme.org

InternationalJournalofMaterialandMechanicalEngineering(IJMME)Volume3Issue1,February2014

beneath the contact surfaces, the following three

modes of this type of failure have been identified: (i)
Formation of microscopic pits typically associated
with rollingcontact fatigue loading conditions along
thepitchlineofgearteeth.Thesepitsmainlyaffectthe
extentandnatureofthegeartoothsurfaceroughness;
(ii) Formation of largersize surface pits which are
typically attributed to a combination of sliding and
rollingcontact fatigue loading conditions (typically
away from the pitchline, especially in the regions
characterizedbynegativeslidingconditions).These
pits act as potent stress risers and can facilitate
initiation of other geartooth failure mechanisms (e.g.
toothbending fatigue failure); and (iii) Spallation, i.e.
formation and shedding of large flakes from the
contactsurfaces.Often,thespalldebrisareformedas
a result of fracture along the interface between the
casehardenedsurfacelayerandthetoothcore.

shaftandmountingreactionsoccursinanonlinearor
unpredictedmanner.
Considering the aforementioned potential benefits of
the computeraided engineering analysis, the main
objective of the present work is to carry out a
computeraided engineering analysis of the tooth
bending highcycle fatiguefailure of windturbine
helicalgearsfoundintheintermediatespeedstageof
thegearbox.Failureofthesegearsisoftenfoundtobe
thecauseofthewindturbinegearboxfailure.
Finite-Element Stress Analysis
Asmentionedearlier,helicalgeartoothbendinghigh
cycle fatiguefailure is one of the main modes of
failureofwindturbinegearboxes.Sincefatiguecracks
are generally initiated at surface locations associated
with the largest contact (as well as subsurface)
stresses, one must determine accurately spatial
distributionandtemporalevolutionofthecontact(as
well as throughthevolume) stresses, before
attemptingtoassessfatiguestrengthandservicelifeof
windturbine gears. Accurate determination of these
stresses is most conveniently carried out through a
finiteelementbased analysis. In the present section,
details regarding the helical matinggear interactions
during the transfer of windturbine torque loads and
the finiteelement modeling and analysis procedure
employedarepresented.

MainObjectives
Torespondtotheaforementionedproblemsrelatedto
the relatively frequent and costly failure of wind
turbine gearboxes, windturbine manufacturers tend
to reengineer critical components and include them
into new subsystems (gearbox designs). To demon
stratetheutilityofthenewdesigns,thereengineered
gearboxes are installed and field testing is started.
While this approach may help address the gearbox
reliabilityconcerns,itisassociatedwithlongfieldtest
times and costly postmortem failure analyses
necessarytoachievethedesiredlevelofconfidencein
thenewdesign.Inaddition,whenthefieldtestresults
become available, it is likely that new windturbine,
and hence new gearbox designs, will dominate the
market,makingthefieldtestresultslessvaluable.

ProblemFormulation
The basic problem analyzed here involves the
structuralresponseoftwomatingwindturbinehelical
gears located within the intermediate stage of the
gearbox,duringthetransferofthetorqueloadsunder
different expected andabnormal windloading condi
tions. The results to be obtained will subsequently be
used to assess toothbending highcyclefatigue
failurestrength and servicelife of the subject helical
gearsunderimposedwindloadingconditions.

Toovercometheforegoingshortcomingsofthepurely
empirical approach aimed at addressing the wind
turbine gearbox reliability, the use of advanced
computeraided engineering methods and tools is
advocated in the present work. While such a
computational approach is not a substitute for the
aforementionedreengineerandfieldtestapproach,it
can provide complementary insight into the problem
of windturbine gearbox failure and help gain insight
into the nature of the main cause of this failure. In
addition, computational engineering analyses enable
investigation of the gear failure in a relatively short
time, under: (a) a variety of windloading conditions
comprisingboththeexpecteddesignloadspectrumas
well as the unexpected extreme loading conditions;
and(b)conditionsinwhichthetransferofloads(both
primary torque loads and nontorque loads) from the

ComputationalAnalysis
The finiteelement analysis (FEA) used here is an
adaptation of our recent work [Grujicic et al. (2012a,
2013ac)]dealingwiththefrictionstirweldingprocess
model. In the remainder of this section, a brief
overview is provided of the key aspects of the
employedFEA.
1)GeometricalModel
The geometrical model/computational domain of
theproblemanalyzedinthisportionoftheworkis

InternationalJournalofMaterialandMechanicalEngineering(IJMME)Volume3Issue1,February2014www.ijmme.org

gravityinducedstresses.Then,atthebeginningof
the analysis, the rotational speed of the shaft
associated with the larger (driving) helical gear is
ramped up to its final value by ensuring that: (a)
the two gears are engaged; (b) the shaft of the
smaller (driven) helical gear is allowed to rotate
about its axis; and (c) a prescribed torque load is
transmitted through engagement, and subsequent
meshing of the gears. It should be noted that the
shafts and their respective gears are connected so
that the rotation of a shaft implies rotation of the
associatedgearandviceversa.

depicted in Figure 2(a). The model comprises two

matinghelicalgearsandtheirtwoassociatedshafts.
Under ideal gearmeshing conditions (i.e. in the
absenceofgearmisalignment),theaxesofthetwo
shafts are parallel and aligned in the global
Cartesian ydirection, as indicated in Figure 2(a).
On the other hand, under abnormal loading
conditionsinvolvinggearmisalignment,theaxesof
thetwoshaftsareassumednottobeparallel.
2)MeshedModel
Eachofthefourcomponents(i.e.twohelicalgears
and two shafts) is meshed using fournode, first
order, reducedintegration, tetrahedral continuum
elements. After conducting a meshsensitivity
analysis to ensure that further refinement in the
mesh size does not significantly affect the results
(not shown for brevity), a meshed model
containing ca. 460,000 tetrahedron elements (of
comparable size and shape) was adopted for the
analysis. A closeup of the meshed model used in
thisportionoftheworkisdepictedinFigure2(b).

4)InitialConditions
As mentioned above, the twogear/shaft assembly
is initially assumed to be stationary and only the
stresses associated with gravityloading are
assumedtobepresentwithineachcomponent.
5)BoundaryConditions
The following boundary conditions were utilized:
(a)thecenterpointsoftheshaftendfaces(treated
as rigid surfaces) are prevented from undergoing
translation;(b)rotationalspeedisprescribedtoone
of the endfaces of the driving shaft (i.e. the shaft
associated with the larger helical gear); and (c) a
constant opposing torque is applied to one of the
endfaces of the driven shaft (i.e. the shaft
associatedwiththesmallerhelicalgear).

(a)

6)ContactInteractions
The geargear normal interactions are analyzed
using a penaltycontact algorithm. Within this
algorithm, (normal) penetration of the contacting
surfacesisresistedbyasetoflinearspringswhich
producesacontactpressurethatisproportionalto
the depth of penetration. Typically, maximum
default values, which still ensure computational
stability, are assigned to the (penalty) spring
constants.Forceequilibriuminadirectioncollinear
with the contactinterface normal then causes the
penetration to acquire an equilibrium (contact
pressuredependent)value.Itshouldbenotedthat
no contact pressures are developed unless (and
until) the nodes on the slave surface
contact/penetrate the master surface. On the
other hand, the magnitude of the contact pressure
that can be developed is unlimited. As far as the
tangential geargear interactions (responsible for
transmissionoftheshearstressesacrossthecontact
interface)areconcerned,theyaremodeledusinga
modified Coulomb friction law. Within this law,

(b)

FIG.2(a)GEOMETRICALMODEL;AND(b)CLOSEUPOFTHE
MESHEDMODELCONSISTINGOFTWOHELICALGEARSAND
TWOSHAFTS,USEDINTHEPRESENTWORK.

3)ComputationalAlgorithm
All calculations are based on a transient,
displacementbased,purelyLagrangian,conditionally
stable, explicit finiteelement algorithm. Before the
analysisisinitiated,thetwogearshaftassemblyis
assumed to be stationary and subject only to the

www.ijmme.org

InternationalJournalofMaterialandMechanicalEngineering(IJMME)Volume3Issue1,February2014

care had to be taken to ensure that the time

increments during the analysis do not exceed the
criticaltimeincrement[Grujicicetal.(2007)].

the maximum value of the shear stresses that can

betransmitted(beforethecontactingsurfacesbegin
to slide) is defined by a product of the contact
pressure and a static (before sliding) and a kinetic
(during sliding) friction coefficient. In addition, to
account for the potential occurrence of a sticking
condition (sliding occurs by shear fracture of the
softerofthetwomaterials,ratherthanbyarelative
motionatthecontactinterface),amaximumvalue
of shear stress (equal to the shear strength of the
softermaterial)thatcanbetransmittedatanylevel
ofthecontactpressureisalsospecified.

Fatigue-Strength and Life-Cycle Prediction

In this section, a brief description is provided
regarding the postprocessing procedures applied to
the results yielded by the finite element analysis
(described in the previous section) in order to assess
the fatiguelife of the helical gears used within the
intermediatestageofthegearbox.
FatigueCrackInitiation

As far as the friction coefficient is concerned, it is

generally assumed that this contact parameter is
controlledbytheformationandshearingofmicro
welds(i.e.micronsizedregionsatwhichcontacting
surface asperities are bonded). Furthermore, it is
recognizedthatthefrictioncoefficientisafunction
ofanumberoffactorssuchasthecontactinterface
(mean) temperature, slip speed, contact pressure,
contactsurfacesroughness/topology,etc.Toassign
the appropriate value to the friction coefficient,
functionalrelationshipsderivedinourrecentwork
[Grujicicetal.(2012b)]wereanalyzed.

Fatiguecrack initiation is a complex process which is

greatly influenced by factors such as material
microstructure, the character and intensity of the
appliedstress,andonvariousmicroandmacroscale
geometrical parameters. Since fatiguecrack initiation
is often observed to be associated with the formation
of persistent slip bands and plasticaccommodation
zones around grain and twin boundaries, inclusions,
etc. [e.g. Kramberger et al. (2004)], it is generally
treated as a (straincontrolled) shortcycle fatigue
process.Inotherwords,formationoffatiguecracksis
assumed to be preceded by the operation of highly
localized plasticdeformation processes. Furthermore,
itisgenerallyassumedthat:(a)fatiguecrackinitiation
occurs in the region associated with the maximum
value of the largest principal stress; and (b) the
transition from the straincontrolled fatiguecrack
initiation stage to the stresscontrolled fatiguecrack
growth stage occurs at a threshold crack length, ath

7)MaterialModel
Thehelicalgearandshaftmaterialsareassumedto
be of an isotropic (linearly) elastic, and (strain
hardenable) plastic character. Due to the isotropic
nature of the material(s) used, the elastic response
isfullydefinedintermsoftwoelasticengineering
moduli (e.g. the Youngs modulus, E, and the
Poissons ratio, ). The plastic response of the
material(s) is defined by specifying the following
three functional relations: (a) a yieldcriterion; (b) a
flowrule;and(c)aconstitutivelaw.Thesefunctional
relations and their parameterization for the gear
andshaftreferencematerial,AF1410,asecondary
hardening martensitic tool steel, can be found in
Grujicicetal.(2012c,2013bc).

(typicallysettoavalueinthe0.10.2mmrange).
Duetoitsstraincontrolledcharacter,thefatiguecrack
initiationprocessismodeledherebycombining:
(a) the conventional CoffinManson equation, ' p 2
' f (2 N i )c , where ' p 2 is the equivalent plastic

strainamplitude, ' f isthefatigueductilitycoefficient,

c is the fatigue ductility exponent, Ni is the number

8)ComputationalToo

of cycles required to reach ath , and 2 N i is the

The problem of helical gear engagement, meshing

and torqueload transfer is executed using an
explicitsolutionalgorithmimplementedinABAQUS/
Explicit, a generalpurpose finite element solver
[Dassault Systemes (2011)]. This algorithm was
chosenbecauseitisassociatedwithcomparatively
low computational cost when dealing with three
dimensional problems dominated by contact (as is
thepresentcase).Sincethedynamic,explicitfinite
element formulation is only conditionally stable,

10

correspondingnumberofstressreversals;with
(b) the additive decomposition of the total equivalent
strain amplitude ' 2 into its elastic, 'e 2 , and
plasticcomponents;
(c) fatigue microyielding constitutive law, ' p 2

' f ' 2 ' f

1 n ' , where ' 2 is the equivalent

stress amplitude, n ' is the cyclic strainhardening

exponentand ' f isthefatiguestrengthcoefficient;

InternationalJournalofMaterialandMechanicalEngineering(IJMME)Volume3Issue1,February2014www.ijmme.org

(d) Hookes law, ' E 'e , where E is the

loading conditions. While closedform K vs. a

functional relations are available for the cracks of
simple geometry and for the simple loading cases,
under more complex crackgeometry/loading
scenarios,thisfunctionmustbeevaluatednumerically.
In the present work, the interactionintegral finite
element method [DassaultSystemes(2011)]is used to
determine the soughtafter K vs. a relation. Within
this method, the K vs. a relation is obtained by: (a)
introducinganinitialcrackoflength ath atthelocation

Youngsmodulusand
(e) stressbased fatiguelife relation, ' 2 'FL

' f 2 N i b , where 'FL is the material fatigue/

endurancelimitand b isamaterialparameter.
Thisprocedureyieldsthefollowingequation:
' ' 2 'FL

'f
E
2

'f

' 2 'FL

2 ' f

1 n'

(1)

yielded by the foregoing crackinitiation analysis; (b)

setting the incremental fracture surface equal to the
planeorthogonaltothemaximumprincipalstress;(c)
controlling the crack growth by successively
prescribing small crack extensions; (d) evaluating the
mixedmodestressintensityfactorasafunctionofthe
associatedmodesI,IIandIIIstressintensityfactorsas
2
2
K 2 K I2 K II2
K III

(4)
1 v
where v is the Poissons ratio; and (e) repeating the
procedure until the computed mixedmode stress
intensity factor reaches its critical (unstable crack
growth)value.

2 Ni b ' f 2 Ni c

E
Once the region within a gear associated with the
largest value of the maximum principal stress is
identified and the corresponding equivalent stress
amplitudecomputed(usingthefiniteelementresults),
Eq. (1) can be solved iteratively to get the number of
cycles to fatiguecrack initiation Ni for a given

combination of gearmaterial and cyclic loading. To

include the effect of surface condition on the fatigue
crackinitiationprocess, 'FL istypicallymultipliedby
apositivecoefficient(smallerthan1.0)whichaccounts
for the effect of initial surface roughness or contact
fatigueinducedsurfacedamage.

Results and Discussion

FatigueCrackGrowth

In this section, the main results of the finiteelement

stress analysis and the postprocessing fatiguecrack
initiation and growth analyses are presented and
discussed. While the present computational frame
work enables the generation of results under numer
ous gearmaterial/transfertorque/gearmisalignment
scenarios, due to space limitations, only a few
prototypicalresultswillbepresentedanddiscussed.

Once the crack reaches its threshold length ath , the

fracture process transitsinto the fatiguecrackgrowth
regime. Since this regime is stresscontrolled, it is
modeledhereusingthetheoryoflinearelasticfracture
mechanics (LEFM). Specifically, fatiguecrack growth
ismodeledusingtheParisequation,whichrelatesthe
rateofincreaseinthesubcriticalcracklengthwithan
increaseintheloadingcyclenumber, da dN ,withthe
(maximum minimum) applied (generally mixed
mode)stressintensitycyclingrange K K th as:

TemporalEvolution/SpatialDistributionofGear
ToothStresses

da
m
C K (a )

(2)
dN
where K th , C and m are material parameters. The

Inthissection,afewprototypicalfiniteelementresults
pertaining to the distribution of the maximum
principal stress over one tooth of the driven helical
geararepresentedanddiscussedinthecontextofthe
expected fatiguelife (in particular, the portion of the
fatigueliferelatedtothecracknucleationstage).

maximum number of fatigueloading cycles in the

crackgrowth regime, N g , is obtained by integrating
Eq.(2)as:
Np

dN

1 a
da

C a K (a ) m

1)AlignedGearCase

(3)

Typicaltemporalevolutionandspatialdistribution
ofthemaximumprincipalstressoverthesurfaceof
atoothofthedrivengear(forthecaseofperfectly
aligned shafts) are shown in Figures 3(a)(d). It is
seen that as expected, the maximum principal
stressdisplayscyclicbehavior.Thatis,asthegears
rotate, the (unengaged) tooth in question becomes

th

where ac is the critical crack length (i.e. the crack

length at which unstable fracture is initiated under
staticloadingconditions).Tocarryouttheintegration
described by Eq. (3), one must know the functional
relationship between the K (i.e. K ) and the current
crack length under the given crack configuration and

11

www.ijmme.org

InternationalJournalofMaterialandMechanicalEngineering(IJMME)Volume3Issue1,February2014

DISTRIBUTIONOFTHEMAXIMUMPRINCIPALSTRESSOVER
THESURFACEOFATOOTHOFTHEDRIVENGEAR(FORTHE
CASEOFPERFECTLYALIGNEDSHAFTS).

progressively engaged and subsequently dis

engaged. Furthermore, examination of the results
displayed in Figures 3(a)(d) reveals that during
thisprocess,thelocationassociatedwiththelargest
value of the maximum principal stress changes
withtheextentofgearrotation.Thisobservationis
importantsince,aspostulatedbythefatiguecrack
initiation model described in an earlier section,
fatiguecracks are nucleated (via the operation of
plastic microyielding phenomena) in the region
associatedwiththehighestvalueofthemaximum
principalstress.

Theeffectofthetorquetransferredbythegearpair
analyzed on the largest value of the maximum
principalstress,andonthecorrespondingvalueof
thevonMisesequivalentstress,inthesubjectgear
tooth (for the case of perfectly aligned shafts) is
showninFigure4.Itisseenthatasthetransferred
torque increases, both the highest value of the
maximum principal stress and the associated von
Misesstressonthesurfaceofthesubjectgeartooth
alsoincrease(approximatelylinearly).Thisfinding
thenimplies,andtheresultsofthefatigueservice
lifeanalysis(presentedbelow)willconfirm,thatas
the transferred torque increases, the number of
loadingcyclesrequiredforfatiguecracknucleation
decreases.

FIG.4THEEFFECTOFTHETORQUETRANSFERREDBYTHE
GEARPAIRANALYZEDONTHELARGESTVALUEOFTHE
MAXIMUMPRINCIPALSTRESSINTHESUBJECTGEARTOOTH).

2)EffectofGearMisalignment
The effect of shaft misalignment (as quantified by
the corresponding misalignment angle, ), at a
constant transferredtorque of 165 kN.m, on the
spatialdistributionandthemagnitudeofthegear
tooth maximum principal stress, at the instant
whenthesubjectgeartoothexperiencesthelargest
valueofthemaximumprincipalstress,isdepicted
in Figures 5(a)(d). Examination of the results
showninFigures5(a)(d)revealsthatasexpected,
as the extent of gear misalignment increases, the
magnitudeofthelargestprincipalstressincreases,
and its location drifts (relative to that in the
perfectlyaligned case). Since the location
associated with the largest value of the maximum
principal stress is considered to be the place of

FIG.3TYPICALTEMPORALEVOLUTIONANDSPATIAL

12

InternationalJournalofMaterialandMechanicalEngineering(IJMME)Volume3Issue1,February2014www.ijmme.org

EXPERIENCESTHELARGESTVALUEOFTHEMAXIMUM
PRINCIPALSTRESS:(a)=0;(b)=1;(c)=2;AND(d)=3.

fatiguecrack initiation, the results displayed in

Figures 5(a)(d) suggest that the location of the
fatigueinduced geartooth failure may change
withtheextentofgearmisalignment.Thisfinding
further suggests that perhaps, change in the
locationofthegeartoothfailure(relativetothatin
theperfectlyalignedgearcase)canberegardedas
anindicationofshaftmisalignmentinducedfailure.

The effect of the gearmisalignment angle at a

constant transferredtorque of 165 kN.m, on the
largestvaluesofthemaximumprincipalstressand
the corresponding von Mises equivalent stress is
replicated, as a line graph, in Figure 6. It is seen
that as the extent of gear misalignment increases,
both the largest value of the maximum principal
stressandthecorrespondingvonMisesequivalent
stressincrease(ataprogressivelyhigherrate).This
findingthenimplies,andtheresultsofthefatigue
servicelifeanalysis(presentedbelow)willconfirm,
that as the gearmisalignment angle increases, the
numberofloadingcyclesrequiredforfatiguecrack
nucleationdecrease.

FIG.6THEEFFECTOFTHEGEARMISALIGNMENTANGLE,AT
ACONSTANTLEVELOFTHETRANSFERREDTORQUE,ON
THELARGESTVALUESOFTHEMAXIMUMPRINCIPALSTRESS.

FatigueLifePrediction
Inthissection,afewprototypicalresultsarepresented
which exemplify the effect of transferredtorque and
gear misalignment on the fatiguelife of the driven
helicalgear.

1)TheEffectofTransferredTorque
It should be recalled that according to the results
displayed in Figure 4, the fatiguecontrolled
servicelifeofthedrivenhelicalgearisexpectedto
decreasewithanincreaseofthetransferredtorque
through the gearassembly. As evidenced by the
results displayed in Figure 6, this prediction is
validated through the use of the fatiguecrack
initiation and growth postprocessing method
ologies (described in Section III). The results
displayed in this figure show the effect of the
transferredtorque on the number of cycles to

FIG.5THEEFFECTOFSHAFTMISALIGNMENT(AS
QUANTIFIEDBYTHECORRESPONDINGMISALIGNMENT
ANGLE,),ATACONSTANTLEVELOFTHETRANSFERRED
TORQUE,ONTHESPATIALDISTRIBUTIONANDTHE
MAGNITUDEOFTHEGEARTOOTHMAXIMUMPRINCIPAL
STRESS,ATTHEINSTANTWHENTHESUBJECTGEARTOOTH

13

www.ijmme.org

InternationalJournalofMaterialandMechanicalEngineering(IJMME)Volume3Issue1,February2014

failure(or,alternatively,onthenumberofyearsof
service) of the driven helical gear, for the case of
perfectlyalignedgears.

analysis in modeling and investigating the root cause

of one of the gear failure modes under a variety of
normal operating and extreme windloading
conditions.Thestudyresultsareasfollows:

2)TheEffectofGearMisalignment

1. In the present work, it is argued that the purely

empiricaleffortsaimedatidentifyingshortcomingsin

The effect of the misalignment angle on the total

fatiguecontrolled servicelife of the driven helical
gear,underaconstanttransferredtorquecondition,
is depicted in Figure 8. It is seen that as predicted
bytheresultsshowninFigure7,gearmisalignment
canseverelyshortentheservicelifeofthegear(the
drivenhelicalgearintheintermediatestageofthe
windturbinegearbox,inthepresentcase).

the current design of the gearboxes should be

complemented with the appropriate advanced
computational methods and engineering analyses.
Such methods/analyses can help shorten the time of
development of new gearbox designs and help with
the identification of the root causes of failure of this
windturbinesubsystem.
2.Specifically,inthepresentwork,aparticularmode
of gearbox failure (i.e. geartooth bending fatigue) is
modeled by combining advanced finiteelement
structural/stress analysis with the computational
procedures developed for prediction of fatiguecrack
initiationandgrowthprocesses(andultimatefailure).
3. While the methodology and the procedures
developedandusedareofapreliminarycharacter,the
results obtained clearly revealed the effect of the
serviceloading conditions (as quantified by the
transferredtorque and the gearmisalignment) on the
fatigueservicelifeofthegearbox.

REFERENCES

FIG.7THEEFFECTOFTHETRANSFERREDTORQUEONTHE
TOTALSERVICELIFEOFTHEDRIVENHELICALGEAR,FOR
THECASEOFPERFECTLYALIGNEDGEARS.

Dassault Systemes, ABAQUS Version 6.10EF, User

Documentation,2011.
Fernandes, P. J. L. Tooth Bending Fatigue Failures in
Gears.EngineeringFailureAnalysis3(1996):219225.
Fernandes, P. J. L., and McDuling, C. Surface Contact
FatigueFailuresinGears.EngineeringFailureAnalysis
4(1997):99107.
Grujicic, M., Pandurangan, B., Zecevic, U., Koudela, K. L.
and Cheeseman, B. A.. Ballistic Performance of
Alumina/S2GlassReinforcedPolymerMatrixComposite
HybridLightweightArmorAgainstArmorPiercing(AP)
and NonAP Projectiles. Multidiscipline Modeling in
MaterialsandStructures,3(2007):287312.

Grujicic,M.,Arakere,G.,Sellappan,V.,Vallejo,A.,andOzen,

FIG.8THEEFFECTOFTHEMISALIGNMENTANGLEONTHE
TOTALFATIGUECONTROLLEDSERVICELIFEOFTHE
DRIVENHELICALGEAR,UNDERACONSTANT
TRANSFERREDTORQUECONDITION.

M.StructuralresponseAnalysis,FatiguelifePrediction
and Material Selection for 1MW Horizontalaxis Wind
Turbine Blades. Journal of Materials Engineering and

Conclusions

Performance19(2010a):780801.

This paper demonstrates the use of finite element

14

Grujicic, M., Arakere, G., Pandurangan, B., Sellappan, V.,

InternationalJournalofMaterialandMechanicalEngineering(IJMME)Volume3Issue1,February2014www.ijmme.org

Vallejo,

A.,

and

Ozen,

M.

Multidisciplinary

during Gas Metal Arc Butt Welding. Journal of

Optimization for FiberGlass Reinforced EpoxyMatrix

MaterialsEngineeringandPerformance22(2013a):1209

Composite 5MW Horizontalaxis Windturbine Blades.

1222.

Journal of Materials Engineering and Performance 19

Grujicic,M.,Arakere,A.,Ramaswami,S.,Snipes,J.S.,Yavari,

(2010b):11161127.

R., Yen, C.F., Cheeseman, B. A., and Montgomery, J. S.

Grujicic,M.,Arakere,G.,Hariharan,A.,andPandurangan,B.

Gas Metal Arc Welding Process Modeling and

Twolevel WeldMaterial Homogenization for Efficient

Prediction of Weld Microstructure in MIL A46100

Computational Analysis of Welded Structure Blast

ArmorGrade Martensitic Steel. Journal of Materials

Survivability. Journal of Materials Engineering and

EngineeringandPerformance22(2013b):15411557.

Performance21(2012a):786796.

Grujicic, M., Galgalikar, R., Snipes, J. S., Yavari, R., and

Grujicic, M., Arakere, A., Pandurangan, B., Yen, C.F., and

Ramaswami, S. MultiPhysics Modeling of the

Cheeseman,B.A.ProcessModelingofTi6Al4VLinear

Fabrication and Dynamic Performance of AllMetal

Friction

AuxeticHexagonalSandwichStructures.Materialsand

Welding

(LFW).

Journal

of

Materials

EngineeringandPerformance21(2012b):20112023.

Design51(2013c):113130.

Grujicic,M.,Pandurangan,B.,Yen,C.F.,andCheeseman,B.

Kramberger,J.,raml,M.,Glode,S.,Flaker,J.,andPotr,I.

A..ModificationsintheAA5083JohnsonCookMaterial

Computational model for the analysis of bending

Model for Use in Friction Stir Welding Computational

fatigue in gears. Computers and Structures 82 (2004):

Analyses. Journal of Materials Engineering and

22612269.

Performance,21(2012c):22072217.

Musial, W. D., Butterfield, S., and McNiff, B. Improving

Grujicic,M.,Ramaswami,S.,Snipes,J.S.,Yavari,R.,Arakere,

Wind Turbine Gearbox Reliability. Paper presented at

A., Yen, C.F., and Cheeseman, B. A. Computational

the European Wind Energy Conference, Milan, Italy,

Modeling of Microstructure Evolution in AISI1005 Steel

May710,2007.

15