SERVICE LIFE PREDICTION OF RUBBER COMPONENTS USED IN ENGINEERING APPL1CATIONS - A Review Abraham Pannikottu Manager, Predictive Testing Akron Rubber Development Laboratory, Inc. January 26, 2002
SERVICELIFE PREDICTIONOF RUBBERCOMPONENTSUSEDIN ENGINEERINGAPPL1CATIONS- A Review
AbrahamPannikottuManager,PredictiveTesting
Akron RubberDevelopmentLaboratory, Inc.
January26, 2002
I. Introduction
A. DegradationAgents
B. DegradationMechanisms
1. Introduction
2. Thermo-oxidativedegradation
3. Thermaldecompositieon
4. Radiationdegradation
5. Ultraviolet light degradation
6. Ozonedegradation
C. Time DependentLimitations
1. Inductionperiod
2. Oxygendiffusion
3. Fluid transport
D. Parametersto Monitor Degradation
1. General
2. Tensilestress-strainproperties
3. Hardness
4. Stressrelaxation
5. Set
6. Dynamicstress-strainproperties
7. Volumechange
8. Otherproperties
9. Functionaltests
10. Chemicalanalysis
II. ComponentTesting
A. SimulatingService
1. General
2. Naturalenvironmentalexposure
3. Simulateddesignlife
B. Principlesof ComponentTesting
1. Introduction
2. When to testproducts
3. Designof producttests
Page1 of 14
4. Examplesof testrigs
5. Summary
III. AcceleratedTests
A. General
1. Purposeof acceleratedtests
2. Methodsof acceleration
B. DesigningAcceleratedTestProgramme
C. Effect of Temperature
1. Low temperature
2. Propertiesat servicetemperature
3. Thermalexpansion
4. Heatageing
D. Effect of Liquids
1. Generalprocedures
2. Standardliquids
3. Water
E. Effect of Gases
1. General
2. Exposureto ozone
3. Evaluationof cracking
F. Weathering
G. Fatigue
1. General
2. Heatbuild-uptests
3. Flex crackingandcut growth tests
4. Testsin tension
5. Non-standardmethods
H. Abrasion
1. General
2. Typesof abrasiontest
3. Abrasiontestconditions
4. Abrasiontestapparatus
5. Expressionof abrasiontestresults
Page2 of 14
I. OtherDegradationAgents
1. Ionisingradiation
2. Electricalstress
J. ServiceConditions
1. General
2. Temperature
3. Solarirradiation
4. Otherfactors
K. PredictionTechniques
1. General
2. Standardisedprocedures
3. Modelsfor changeof parameterwith time
4. Environmentaldegradationtests
5. Arrheniusrelationship
6. Time/temperatureshift
7. Artificial weathering
8. lonisingradiation
9. Effect of liquids
10. Effect of gases
11. Creepandstressrelaxation
12. Set
13. Fatigue
14. Abrasion
15. Dynamicconditions
L. Limitations and Pitfalls in AcceleratedTesting
1. Limitations
2. Pitfalls
IV. Appendix
V. References
VI. Abbreviations
VII. Index
Page3 of 14
ABSTRACT
INTRODUCTION
Thetopic of ServiceLife Predictionis of both practical and scientificinterest. The servicelife ofa elastomersets limit to engineeringdesign. Therefore,Life Prediction of elastomersshould bepartof theengineeringdesignprocess.Elastomerpropertiesare sensitiveto heat,moisture,light,fluids and mechanical stressFigure 1. Elastomerscan undergo changesin propertieslargeenoughto causeproductfailure. Most elastomerparts, in engineeringapplications,are intendedto be in servicefor severalyears. Hence,theengineeror thedevelopingscientistcannotwait thatlong to evaluatetheagingprocessin actualserviceconditions. Thethreemajorengineeringtasksin elastomerapplicationsare to determinethe Shelf Life. Service Life, and Remaining Useful
LifePart already inuse.
Page4 of 14
Elastomer Products
EnergyEnvironmental Factors
li/AcidSI0I
Solvent .ation
L1-ozone b [iht
WindLjJ..Corrosive * __ Stress
..____ Mechanical Stres!jdynamicMistand static
Hej
Environment DustSalt -.
Corrosion Water
Structural
Properties
Figure 1
Page5 of 14
Elastomericmaterialsare frequentlyusedunderseverethermal,chemical and mechanicalstressconditions Table 1. A major drawbackof elastomersis their tendencyto oxidize. Most ofthem becomeunstablein contactwith atmosphericoxygen. Elastomersare also affectedby lowand high temperature. The wide range of service conditions produceschangesin physical,chemical,thermodynamicand otherpropertiesof elastomersTable2. Theseserviceconditionscan changein an uncontrolledmanner.
ComprehensiveServiceLife Predictionstudiesshouldinvolve fundamentalchangesin physicalpropertiesof the material due to degradationmechanisms,inter-atomicbonding, microstructureandcrystal structures.Elastomerdegradationshould alsoinclude characteristicssuchasthe lossof plasticizer,separationof polymerfrom fillers, andsurfacefriction.
ServiceLife Prediction is clearly of great scientific interest and has attractedthe attention ofchemists,engineers,scientistsand solid statephysicists. Hence,it is not surprisingthat the topichashistorically been discussedform a varietyof different scientific approaches.Thereis a greatneedfor a comprehensiveLife Predictionmodel which incorporatesvariousapproaches.Oneofthe main challengesin developinga servicelife is that severalparametersto be consideredbothfrom the material and serviceenvironment. Many of theseparametersaredifficult to describepreciselyin a mathematicalmodel. The objectiveof this paperis to establisha practicalusablemethodfor quantitativelife prediction of elastomers. This paperdescribesmethodsof ServiceLife Predictionfor an 0-ring, a Fiber-reinforcedRubberPipeJoint anda ConveyerBelt Cover.
Table 1: Main FactorsResponsiblefor Degradation.Thermal MechanicalThermo-Oxidative HydrolyticPhoto ChemicalPhotooxidative High EnergyRadiationOzone
Table2: DominatingEventsduring Degradation.RandomChain Scission SubstitutionDepolymerization PlasticizerLossCrosslinking Filler BondingChangeSide GroupElimination
Page6 of 14
AkronRubber Development Laboratory, Inc.ARDL Methodology for Ouantitative LifePredictionof Elastomers
QuantitativeServiceLife Prediction of the elastomericcomponentis becomesan increasinglyimportantrequirementas elastomersare usedfor more critical engineeringapplications. Servicelife prediction methodology should include all processesthat may affect the function of theelastomeric component. ARDL’s approach is to: 1., select the predominant degradationprocessesand establish an appropriate accelerated aging test; 2., compare the failuremode/degradationprocessof lab samples with field sampling using chemical, physical andoptical techniquesTable 3; 3., establishthe failure rates using acceleratedlab tests; and 4.,extrapolaterates to theservicecondition to determineservicelife.
Table 3: Techniques
__________________________________
OpticalMicroscopeVideoMicroscope
Scanning ElectronMicroscopyFI-IRGCIMSDSCDMAPulseNMRTGA
CrosslinkDensity WetChemistrySpecificGravityMicroHardnessMicroHardnessDecay
DielectricConstantElectronSpinResonance
Page7 of 14
ARDL’s methodologyflowcharts are shownin Figures 2a and 2b. The first stepis to definethefunctionsof theelastomercomponent.Basedon thefunction,establisha failure criterion. Thefailure criterion may be an unacceptablechange in function and the changemay cause aparticular failure. Changesmay be stressrelaxation, creep, tear resistance,stiffness/moduluschange,swelling, dielectricproperties,dynamicproprieties,etc. Then,characterizeand identifythe underiiningmechanisminvolved in this change. Establishthe rateof changeby acceleratedlaboratorytest at different levels of severity and at different time intervals. It is important tokeepthe acceleratedtest condition similar to the servicecondition and perform the test at fourtemperatureshigherthan averageservicetemperature.
Material Characterization
‘JrFunctional Characterization of Material
4.Functional Characterization
‘I,of the Part
Identify the Predominant Failure Mode
Field Data Correlation
4.Service Life Prediction of thePartj
Figure2a
Pagesof 14
Definition of ComponentFunction
chemicalcrack
Chri0Tafon
friction fluidfailure permeation
Evaluationof geometricalaspectsFunctionalDurability Test and Finite ElementAnalysis
rmaldiffusionanalysisnalysis
Contributionof eachprocessfor contributionsto ratefailure
FieldData
Life prediction of component
Figure 2b Basedon Reference6
Aging processin the acceleratedtest should be verified with "service aging" process. Theverification can be done by chemicalor physical evaluationof both field samplesand samplesfrom different levelsof acceleratedtest conditionsandtime intervals.
After the rateof degradationprocesshasbeen determinedusing acceleratedlaboratory testing,then therateat the servicetemperaturecanbe determinedby Arrheniusextrapolationor by time-temperaturesuperpositionextrapolation. It is importantto study the mechanismsof failure inserviceand correlatewith eachacceleratedagingtest. This will help to improvereliability of theServiceLife PredictionModel.
I. ArrheniusApproach
This approach of life prediction involves determining the rate of failure for a range oftemperaturesof interestandthen plotting theserateson an Arrheniustypeplot againstreciprocal
-IDefinition of FailureCriterion
Itemizeprocessescontributingto failure
Material teststo measurerateof eachprocess
fracture dynamicmechanics testing
stressanalysis
Page9 of 14
absolutetemperature.This approachconsidersthe failure processasa chemical reactionwherethe rateof reactionwill increaseastemperatureincreases.
ServiceLife = Ae ff1
Where:Ea is the activationenergyfor thefailure process
A. A is a constantdependingon typeof failure
To be able to useArrheniusequation,acceleratedtestingshouldbe usedto identify the"time-tofailure" for a minimumof four temperaturesabovetheaverageservicetemperature.Servicelifepredictioncanbe obtainedby plotting the logarithmof the "time-to-failure" versustheinverseofabsolutetemperature‘/‘- andextrapolatingthecurve suchthat it intersectsthestraight line
=Tserviceis theaverageservicetemperature.To estimatetheexpectedtime-to
sevice
failure, it is importantto usea materialpropertywhich featuressufficient rangeto assureareliabledeterminationat anyparticularstageof thepropertyduring both acceleratedagingandfield aging.
1. Time - Temperature SuperpositionApproach
This approachconsidersthe rateof deteriorationas a predominantlyviscoelasticbehavior. Thisapproachwill allow theinvestigatorto developa mastercurvefor servicelife predictionusingtheWLF-equationWilliams, Landel andFerry thegeneralform of which is:
c,T -
loga =
c2÷Tr-TWhere:
T is referencetemperaturesusuallyT shouldbe 50K aboveTg
aT is the shift factorC1 andC2 are materialconstantsoftenC1 = 17.44 andC2 = 51.6T is the test temperature
From the presentpoint of view, it must be notedthat by rule of thumb one decadeof test timeincreaseis equivalentto an increasein temperatureof approximately6-7°C. The main problemwith this approachis that polymerdegradationmay not follow strict viscoelasticbehavior. Theequation permits the shifting of datataken over a time rangeand at a variety of temperaturesalong the time axis, to createa mastercurve which extendsthe time axis by severalorder ofmagnitude.
Page10 of 14
CASESTUDIESCase1: Life Prediction of 0-Rings by Compression Stress RelaxationTesting
The two important propertiesof elastomeric seals are Compression Stress Relaxation andDynamic Response.Dynamic Responseis a measureof the ability of the seal to retract to itsoriginal shapeafter a constrainthas been removed. Rapid recoveryof an elastomericseal isimportantin which thecontactbetweenthesealandmating surfaceis momentarilybrokenduetovibration or dynamic motion. Dynamic propertiesof elastomerscan be measuredusing forcedynamics. Theforcedecayof elastomercomponentsunderconstantcompressivestrain is knownascompressionstressrelaxation. Thetest measuresthesealingforce exertedby a sealor 0-ringundercompressionbetweentwo platesFigure 3. This property is very importantwheresealsare relatively inaccessible,and the cost of replacementis too high. In such instances,lifeprediction is highly desirable. The automotive industry has focusedon the percentof retainedsealingforceasa functiontime in a compressionstressrelaxationtest.
Shouldbe parallelwhile themeasurementismade.
Glass Plates
The apparatususedat ARDL to study compressionstress relaxation is the Wykeham-Farrancecompressionstressrelaxationequipment. Cylindrical samplesare usedin accordancewith ISO3384. Compressionstressrelaxation is usedfor the service life prediction by measuringthesealingforcedecayasa functionof time and temperature.IRM 903 is usedasthe agingfluid. Across-sectionalview of thetestjig is shown in Figure 4. Thesecurvesare obtainedat 25 percentconstantcompressivestrain at three acceleratedaging temperatures100°C, 125°C and 150°C.Sixty percentsealingforce decayis selectedas the failure point. This failure point is developedbasedon an earlier study by ARDL using Leak test, CompressionStressRelaxationtesting,DynamictestingandAutomotive OEM Engineeringdata.
Load Cell
Plunger
Specimen
Figure3
Jig Housing
Page11 of 14
Field data was collected for silicone compound A, B and C. Chemical and physicalcharacterizationof field sampleand lab sampleswas usedto verify the mode of failure anddegradationprocess.
a TestParameter
CompressionStressRelaxation
Accelerated oven aging100°C, 150°C and 150°C
b Aging Fluid: IRM 903
c Test Condition:25% compression
d Failure Mode:
e FailureCriteria:
20% increasemax80%increasemax
Pilunger
Figure 4
1 Test:
a AcceleratedAging: at
Sample were aged inside jigs with
StressRelaxation
CompressStressRelaxationForceRetained,%
i DynamicProperties
40%
tan Sdynamicmodulus
Page12 of 14
2 ExtrapolationTechniques: ArrheniusandWLF
Results: Shown in Tables 4d and 4e based on Arrheniusextrapolationtechnique. The datain Tables4a, 4b and 4cwere used to make a mastercurve to determinea shiftfactor aT value. Calculatedvalues were not consistentwith published literature values. From this one canconcludethat viscoelasticmechanismsdo not determinetherate of failure. Table 4f shows dynamic modulus andtan S valuesof acceleratedagedsamples.
Table 4a: SiliconeCompoundA SealingForceRetained,%hours 100°C 125°C 150°C
0.5 100.0 100.0 100.048 40.5 38.0 36.5
168 34.5 33.0 28.5336 31.0 29.5 24.5840 29.5 25.0 21.5
1008 28.5 23.5 19.52000 26.5 22.5 18.55000 25.0 19.5 14.5
Table4b: Silicone CompoundB SealingForceRetained,%hours 100°C 125°C 150°C
0 100.0 100.0 100.048 81.0 75.5 71.5
168 69.5 62.5 56.5336 62.5 56.0 48.0840 58.0 51.0 44.0
1008 56.0 48.0 40.02000 54.0 45.5 36.55000 48.0 40.0 26.5
Figure 4c: SiliconeCompoundC Sealing forceretained,%hours 100°C 125°C 150°C
0 100.0 100.0 100.048 61.0 56.0 53.5
168 52.5 47.5 42.5336 48.0 41.5 36.5840 42.0 38.0 32.5
1008 41.5 35.5 29.52000 40.0 34.0 27.05000 37.5 28.5 20.5
Page13 of 14
Table 4d:
Pmperty A B CHardness,ShoreA 59 55 64Ulitiniate tensile,MPa 5.8 4.7 6.9CompressionSet, %22hrs/70°C 33 16 1422hrs/100°C 73 18 1722hrs/125°C 87 26 2222hrs/150°C 89 34 35
SealingForceRetained,%168 hrs/150°C 28.5 56.5 42.51008hrs/150°C F 48.0 29.5
FieldData F* G F***in1anljately **after 1 yearservice
Table4e:Compound A B CFieldData Early 5 years+ 1 year
ExtrapolatedArrhenius 61 hours 6 years 7 monthsWLF iN
otSuccessful
Table4f: DynamicModulus by DMA tan?150°C
Compound A B C0 10.5 0.22 9.5 0.20 12.5 0.24
48 18.9 0.25 10.8 0.21 13.8 0.25168 19.2 0.26 11.2 0.21 16.8 0.27336 19.8 0.27 12.8 0.22 22.8 0.28840 20.4 0.28 14.8 0.23 24.8 0.29
1008 20.4 0.28 16.9 0.24 25.8 0.30
Page14 of 14
Case2: Remaining Useful Life Determination of Elastomeric Pipe Joints
In this study, Arrheniusextrapolationtechniquewith Laboratoryacceleratedoven aging testingwas usedto establishthe Remain Useful Life of ElastomericPipe Joints. Failuremode of thispart wasidentified asoxidativedegradation.
b TestParameter
1 AcceleratedAging Test: Accelerated oven aging at 60°C,80°C and 100°C
2 Field Samples: a Inside section b Middle section cOutsidesection
3 ServiceDuration: Five years
4 AverageServiceTemperature: 45°C
5 MeasurementTemperature: DSC induction timemeasurement
FailureModeEvaluation: Dynamictesting - Dampingcoefficient N-sec/mm
II. Differential Scanning Calorimetry DSC ASTM D-3418-88
The samplewas heatedfrom 30°C to 300°C at 25°C per minute in helium for approximately10minutes. Thesamplewas allowedto equilibratefor 6 minutes. After equilibrating,thegaspurgewas switchedfrom helium to oxygen. DSC inductiontime was calculatedbasedon time fromintroductionof theoxygento completionof decomposition.
1 InductionTime Of Field Samples: 1 Outside layer= 28 minutes
2 Middle section = 44 minutes3 Inside layer = 38 minutes
DampingCoefficient N-sec/mmof field sample: 1 Outsidelayer = 38.88
2 Middle section = 2.57
3 Inside layer = 16.43
FailurePoint: Failurepoint is calculatedbasedon failedfield sample.Inductiontime = 13 minutesandDampingcoefficient= 51.5 N-sec/mm
ServiceUfe PredictionExtrapolationa Technique: ArrheniusApproach
2 Table Sa: InductionTime by DSChours 100 °C 80 °C 60 °C
0 45 45 454 35 41 42.516 18 32 36
Page15 of 14
48 6 25 34168 0 17 301000 0 4 242000 0 2 19
*Extrapolatedservicelife - 8 years.Therefortheremaininguseful life of this partwill be 3 years.
TableSbInductionTimeminutes
DampingCoefficientN-sec/mm
Partin UseInside 38 16.43Inner Case 44 2.57Outside 28 38.88
FailedPart Outside 13 51.5
Table Sc: DampingCoefficientby DMA N-sec/mmTeperature 100°C 80°C 60°C
0 1.58 1.58 1.5848 68.1 28.4 16.5
168 98.1 52.8 24.52000 failed 108.5 48.7
Page16 of 14
Case 3:Service Life Predictionof IndustrialConveyer BeltCovers:
Heavy, steelcable-reinforcedconveyerbelts arecommonlyusedin the coal industry.The failuremode of the belt cover in theparticularapplicationwas identifiedas hydrolysis with subsequentultimatetearfailure.
2. TestParameter
234
a AcceleratedTest: Fatigue testdynamic tensiontest
Aging Fluid: WaterTestTemperature: 60°C,Test: Five million cycles
Table6b: DampingCoefficientat 60 by DMA N-sec/mmCompoundA CompoundB
0 0.852 0.6841 million 1.048 0.8452 million 1.284 1.0483 million 0.748 1.1284 million failed 1.2325 million 0.8486 million failed
MeasurementTemperature:
at 10 Hz with 10%
80°C and 100°C
MTS Dynamic characterson dumbbell sampleat Sand2
a Properties:Dampingcoefficientb AverageServiceTemperature: 40°C
Table6a: Cyclesto FailureTemp
1008060
Extrapolated
Comp A120000700000
20000004,600,000
CompB180000
10000004000000
9,200,000
Page17 of 14
CONCLUSION
The above approachescan be applied to determinelife of elastomerscomponentsusedinengineeringapplications.However,it is importantto definefailure modeand failure mechanism.It is also importantto establishverification andcorrelationbetweenfield and lab samplesusingphysical and chemical techniques. The primary rate determiningmechanismof componentfailure can be predicted using the Arrhenius methodology.The Arrhenius method provides aquantitativedeterminationof the servicelife of elastomercomponentsin a particularapplication.Furtherresearchstudiesare requiredfor eachnew application. The servicelife predictioneffortconducted on elastomeric materials provides a good materials databasefor computer-aideddesign engineerswho in turn can usethe information to effectively model part durability, thusreducingtheneedfor complexandcostlyprototypetesting.
Service life prediction as an engineering tool is still in the infancyAlthough some successeshavebeen reported,more applicationresearchcontinuing interest showedby the automotive OEM’s is likely to leadelastomerservicelife prediction.
stage of development.is needed. Recentandto further advancesin
Page18 of 14
REFERENCES
1. A. Stevenson"Rubber in Engineering" in KempesEngineeringYearbook1986-90
2. A.N. Gent , J. Appl. Polym. Sci., 6 22, 442 1962
3. C.J.Derham,J. Mater. Sci., 8 1023 1973
4. A.S. Farid,Plastic,RubberandCompositesProc.,25 911996
5. A. Pannikottu,J.J.Leyden RubberandPlastic News 18 1995
6. A.N. Gent "EngineeringwithRubber"HanserPublisherDurability Chapter7 171 1992
7. K.C. Ludema "Friction, Wear, Lubrication" CRC Press Strength and DeformationPropertiesof Solids Chapter2 9 1979
8. R.K. Eby "Durability of MacromolecularMaterials" ACS SymposiumAmericanChemicalSociety Chapter1 2 1979
9. T. Kelen "Polymer Degradation" Van ReinholdCompany Methods of StudyingPolymerDegradation Chapter2 10 1983
10. Y.A. Shlyapnikov "AntioxidativeStabilizationof Polymers" Taylor Publishers Estimatesof ReliableServiceLife of PolymericMaterials Chapter7 1996
I