2013
André Ri ardo
Correia dos Santos
Válega
Estudo termo-reológi o e estrutural de peças
2013
André Ri ardo
Correia dos Santos
Válega
Estudo termo-reológi o e estrutural de peças
obtidas por mi roinjeção
DissertaçãoapresentadaàUniversidadedeAveiropara umprimentodos
req-uisitosne essários àobtenção do graude mestre emEngenharia Me âni a,
realizada sob orientação ientí a da Professor Doutor Móni a Sandra
Abrantes de Oliveira Correia, Professora Auxiliar do Departamento de
En-genharia Me âni a da Universidade de Aveiro e do Professor Doutor Rui
António da Silva Moreira, Professor Auxiliar do Departamento de
Presidente/ President Professor Doutor JoséPaulo Oliveira Santos
ProfessorAuxiliardaUniversidadedeAveiro
Vogais /Committee Professor Doutor Rui Jorge Sousa Costa de Miranda Guedes
ProfessorAuxiliardaFa uldadedeEngenhariadaUniversidadedoPorto
Professor Doutor Móni aSandra Abrantes de OliveiraCorreia
Professora AuxiliardaUniversidadedeAveiro(orientador)
Professor Doutor Rui António da SilvaMoreira
A knowledgements entes, amigos e olegas queestiveram onstatemente presentes durante o
meu per urso a adémi o. Porisso, quero deixar-lheso meu agrade imento
pessoal, eem espe iala:
Aos meuspais e avós,pelo in ansável einquestionável apoio durante toda
a minhavida e emespe ialnesta fase.
Àminha madrinhapelossábios onselhosdurante oper urso a adémi o.
A todos os ompanheirosde urso, e emparti ularao Bruno Barroqueiro,
Carlos Oliveira, Rui Bártolo, Rui Pais e Tiago Godinho por todo o apoio e
entre-ajuda durante o urso.
ÀProfessora Doutora Móni a Oliveira,pormeterdado aoportunidadede
trabalhar nesta área, por nun a ter duvidado de mim e tersido uma fonte
de motivação e in entivo onstante.
Ao ProfessorDoutorRuiMoreirapela sua disponibilidade, onselhos ertos
e ajuda na análisedosdados.
À EngªTatiana por toda aajuda naparteexperimentale nas simulações.
Aos olegasdelaboratórioquemeintegraramnumgrupodinâmi oea tivo.
A todosos que ontribuiram deforma dire taou indire taneste trabalho.
prin ípio de sobreposiçãotempo temperatura;TTSP
Resumo Nas últimasdé adas, aminiaturizaçãodosequipamentoseletróni ose
apar-elhos me âni ostemsidouma tendên iaem onstante desenvolvimento. A
mi romoldaçãoporinjeçãoéumadasté ni asmaisutilizadasparaobtenção
de mi ro omponents plásti os para os mais variados propósitos. Através
desta té ni a é possível obter mi ro omponentes em grandes quantidades
e a baixo usto, usufruindo ainda do alto nível da automação inerente ao
pro esso. Contudo, durante o pro esso de moldação são originadas
ten-sões residuaisinternas nas peças moldadas, devido ao es oamento do
fun-dido e ao arrefe imento e solidi ação do plásti o. Neste trabalho
propõe-se uma metodologia que ontempla a simulação numéri a do pro esso de
obtenção de peças plásti as por injeção juntamente om uma análise
es-trutural. Esta última visa levar em onsideração o históri o do
pro es-samento bem omo o omportamento típi o de um material poliméri o
aquando em serviço. Desse modo, o trabalho in lui o estudo e denição
do omportamento vis oelásti o do material. Para tal, são abordadas
tam-bémmetodologiasdeanálisedo omportamentovis oelásti odosmateriais.
O trabalho divide-se em duas partes prin ipais: a primeira onde são feitos
ensaiosDMTApara ara terizaçãodo omportamentome âni odomaterial
e propostas metodologiasde análise do mesmo atravésdo usodo prin ípio
desobreposiçãotempo-temperatura;asegundapartein idenaobtençãodas
tensõesresiduaisatravésdasimulaçãonuméri adopro essodeinje çãoeda
simulaçãonuméri a deumensaioderelaxaçãodetensõesnoqual omodelo
vis oelásti opreviamente riadoé inserido. Osresultadosobtidospermitem
on luiraadequabilidade dametodologiadesenvolvida,permitindo postular
asuaviabilidadenasimulaçãodo omportamentodemi ropeçasquandoem
time temperaturesuperpositionprin iple;TTSP
Abstra t In re ent de ades, the miniaturization of ele troni and me hani al
equip-ment is a tenden y in onstant development. Mi ro-inje tionis one of the
most used te hniques for obtaining plasti s mi ro- omponents for various
purposes. Throughthiste hniqueitispossibletoobtainmi ro- omponents
inlargequantitiesandatlow osts,yettakingadvantageofthehighlevelof
automationinherentto thepro ess. However, duringthe mouldingpro ess,
residualstressesstarttodeveloponthepart,duetothemeltowanddueto
the oolingandsolidi ationoftheplasti . Inthisstudyamethodologythat
addresses the numeri al simulation of the inje tion moulding oupled with
a stru tural analysis is proposed. The latter takes into a ount the ee t
of residual stresses on the part and the fa t that polymers are vis oelasti
materials. Forthispurpose,methodologiestoanalysevis oelasti behaviour
of polymers are also proposed. This workis omposed by two main parts:
the rst, where DMTAtests are performed to hara terize the me hani al
behaviour of the material and where analysis methods to handle resultant
data are proposed, namely by using the prin iple of time-temperature
su-perposition; the se ond part relates to obtaining the part residual stresses
throughthe numeri alsimulationoftheinje tion pro essandthe numeri al
simulation ofa stressrelaxation test,inwhi hthe previously developed
vis- oelasti modelisa ounted for. Theobtainedresultsprovidean insighton
the adequa y ofthe methodologyhere developed toassess thedurability of
1 Introdu tion 1
2 Literature review 3
2.1 Vis oelasti ity. . . 3
2.2 Linear onstitutive equations (simple shear) . . . 4
2.2.1 Equations of hange . . . 4
2.2.2 Innitesimal straintensor . . . 5
2.2.3 Stresstensor . . . 5
2.2.4 Constitutive equation for linearvis oelasti ity(simple shear) . . . 5
2.3 Me hani al models for vis oelasti behaviour . . . 7
2.3.1 Maxwell Model . . . 7
2.3.2 Voigt-Kelvin Model. . . 8
2.3.3 ZenerModel . . . 9
2.3.4 Vis oelasti spe tra. . . 11
2.4 Polymeri materials. . . 12
2.4.1 Temperature inuen e onregionsof vis oelasti behaviour . . . 13
2.4.2 Thermoplasti s and thermosets . . . 15
2.4.3 Amorphousand rystallinethermoplasti s . . . 15
2.5 Linear time-dependent experiments (shear) . . . 17
2.5.1 Stressrelaxation . . . 17
2.5.2 Creep . . . 17
2.5.3 Sinusoidal soli itationsand dynami experiments . . . 18
2.6 Dynami and transient experiments. . . 20
2.7 Boltzmann Superpositionprin iple . . . 22
2.8 Time temperature superposition prin iple (TTSP) . . . 24
2.8.1 Mole ular interpretations oftime-temperature equivalen e . . . 25
2.8.2 Time temperature supeposition prin iplefailure . . . 27
3 Polymer vis oelasti behaviourassessment 31
4 Polymer mi ropart - numeri al analysis 51
2.1 Testing modesandspe imengeometriesfor DMTA[17℄ . . . 22
3.1 Mi ropartpro essing onditions . . . 33
3.2 Log
α
T
(T )
obtained through rstand se ondmethods . . . 464.1 Inuen eof initial stresseson thestressed zone . . . 55
4.2 Realand Imaginary partvaluesfor alibrating vis oelasti model . . . 56
4.3 Vis oelasti modelinuen e on stressde rease, withinitial stresses . . . . 60
2.1 Componentsof thestresstensor [3 ℄ . . . 6
2.2 Maxwell model- adaptedfrom [23℄ . . . 7
2.3 Maxwell's modelbehaviour . . . 8
2.4 Voigt-Kelvinmodel- adaptedfrom[23 ℄. . . 8
2.5 Voigt-Kelvin'smodelbehaviour . . . 9
2.6 Zenermodel- adaptedfrom [6 ℄. . . 9
2.7 Zener'smodel behaviour . . . 10
2.8 ImprovedZener's models . . . 11
2.9 Transitions inpolymer-adapted from[24 ℄ . . . 14
2.10 Regions of vis oelasti behaviour- adaptedfrom [21℄ . . . 14
2.11 Symboli representation oflongandshortrangerelationships ina exible polymermole ule- adaptedfrom [10℄ . . . 15
2.12 Typi al polymer stru ture - adaptedfrom[4℄ . . . 16
2.13 Phaselag between stress and strain- adaptedfrom[10 ℄ . . . 18
2.14 Ve torial representation of omplexmodulus- adaptedfrom [10℄ . . . 19
2.15 Range of frequen iesfor dierent te hniques [24℄ . . . 21
2.16 Boltzmann superposition prin iple- adapted from[6℄ . . . 23
2.17 Two-sitemodel - adaptedfrom[24℄ . . . 25
2.18 Spe i volume versus temperature for a typi al amorphous polymer-adaptedfrom [24℄ . . . 27 3.1 Moldow ® ll timepredi tion. . . 32 3.2 Moldow ® temperature at owfront predi tion . . . 32
3.3 Moldow ® mi ropart warpagepredi tion . . . 33
3.4 Mi ropart . . . 34
3.5 Triton Te hnology TTDMADynami Me hani al Analyser. . . 35
3.6 Sample mounting . . . 35
3.7 Matlabappli ation ow hart . . . 36
3.8 Cole-Cole plot for200
µm
ABSmi roparts. . . 373.9 Cole-Cole plot for300
µm
ABSmi roparts. . . 373.10 vanGurp Palmen plotfor 200
µm
ABSmi roparts . . . 383.11 vanGurp Palmen plotfor 300
µm
ABSmi roparts . . . 383.12 Densityversus temperature plot for ABS. . . 39
3.13 Lossmodulussuperpositionfor200
µm
ABSmi ropartswithverti alshifting 39 3.14 Storage modulus superposition for 200µm
ABSmi roparts withverti al shifting . . . 403.16 Storage modulus superposition for 300
µm
ABSmi roparts withverti alshifting . . . 41
3.17 Mi ropartgeometry inuen e on tests . . . 41
3.18 Se ondmi ropart . . . 42
3.19 Cole-Cole plot for600
µm
ABSmi roparts. . . 423.20 vanGurp Palmen plotfor 600
µm
ABSmi roparts . . . 433.21 Loss fa torexperimentaldatafor 600
µm
ABSmi roparts . . . 443.22 Overlappingwindows for TTSP . . . 44
3.23 Flow hartfor TTSP rst method . . . 45
3.24 Master urve for lossfa tor withmean value method, Tref=
30
◦
C
for 600µm
ABSmi roparts . . . 463.25 Flow hartfor TTSP se ond method . . . 47
3.26 Master urveforlossfa torwithiterativemethod,Tref=
30
◦
C
for600µm
ABSmi roparts . . . 483.27 Master urve for storagemodulus for 600
µm
ABSmi roparts . . . 483.28 Master urve for lossmodulus for 600
µm
ABSmi roparts . . . 493.29 Linear regressionfor a tivationenergy . . . 49
3.30 Comparison of Arrhenius and WLF equations urve tting with experi-mental datafor ABSat severaltemperatures . . . 50
4.1 Mi ropartinanalysis . . . 52
4.2 Initialstate of stress,reading fromle . . . 53
4.3 Partdeformation andshrinkage, s ale fa torof 50 . . . 54
4.4 Field of stresses when a 20
µm
strain is imposed (true deformation s ale fa tor) . . . 554.5 Vis oelasti model alibration . . . 57
4.6 Element 63657stress evolution . . . 58
4.7 Element 60460stress evolution . . . 59
4.8 Element 61575stress evolution . . . 60
4.9
1%
strain withoutinitial stresses . . . 614.10
1%
strain withinitial stresses . . . 62 5.1 Methodoly for inje tionmoulded parts rheologi aland stru tural analysis 64Introdu tion
Nowadays one of the most ommon te hniques to produ e parts is inje tion moulding.
A ording toCefamol-NationalMoldIndustryAsso iationinPortugal,thiste hnique in
parti ular,enables theprodu tionofparts atahighrate, in reasingenterprises
ompet-itivity. Thisimportant industry an supply omponents to world leading ompanies in
time, thanksto the Portuguese engineersexperien eandknow-how [1 ℄.
Inthelastde ades omponentsminiaturizationhasbeen agrowingtrendasa
onse-quen e of industry development. Mi roinje tion allows for produ ing parts invery high
quantities with a very low ost, thanks to the pro ess automation level. Produ ts
ob-tained bymi roinje tion mayhave a very distin t purpose. Automotive industry is one
of thelargest onsumers of these mi ro omponents and uses it in many imper eptible
ways. Mi ro omponentsarealsowidelyusedinele tri al/ele troni produ ts,inmedi al
omponents or even insimplethings su hasgears for wristwat hes.
Parts produ ed withthe mi roinje tion pro ess should also ensure good me hani al
hara teristi s su h as durability, physi al stability at working temperatures and even
small strain, sin e any deformation in su h a small pie e an make the whole part to
ollapse.
Inordertofulllthese riteria,itisne essaryaspe ial arewhen hoosingthe
poly-mer whi h will be inje ted. Polymers thatfulllthe riteria areusually moreexpensive
when ompared to ma ro inje tion polymers but,on ethe quantity usedis smaller, the
total pri e of the part is not mu h ae ted. In mi roinje tion theused material is
usu-allynon-re y led, whi h impliesthatthemold shouldbevery well proje ted inorderto
minimize wastes.
Highpre isionandnishingdemands auses themi roinje tionmoldstobedierent
from ma ro inje tion[2℄.
Similarlyto what happens on the onventional inje tion molding pro edures, mi ro
inje tiongeneratesinternal residualstressesinthe omponentsduetonumerousreasons,
in luding: stressesindu edbythemeltoworevenstressesdueto thepressureimposed
during thepolymer's ooling andsolidifying. Thelatterareresponsiblefor thewarpage
and deformation of the part if not properly ontrolled and may lead to a more fragile
omponent after its extra tion. In mi ro inje tion molding this ee t assumes an even
larger importan e be ause of the high pressure and inje tion velo ity needed to avoid
premature polymer solidifying due to thepart high surfa e/volume ratios. Shearstress
resultant from these extreme pro essing onditions may lead to breakage of mole ular
in-tegrityofthemi ropart. Itisthennoti eabletheimportan eofthepro essing onditions
on the omponent future me hani al behaviour. In fa t, dierent pro essing onditions
may indu e dierent part me hani al properties due to the material distin t
thermo-rheologi alhistories. Thelattermaybetaken intoa ount through ommer ialsoftware
pa kages su h as Autodesk Moldow ®
whi h allow the user to simulate and assess the
omplete partprodu tion,from llingto ooling.
Asso iatedto thethermo-rheologi alhistory,itsindu edresidualstressesduring
pro- essing, it isdetrimental to a knowledge the material me hani al behaviour inorder to
postulate its durability during servi e. It is ommon knowledge that polymersare
vis- oelasti , nevertheless its me hani al behaviour is a onsequen e of its pro essing and
thereforeits thermo-rheologi history andrequiresfurther assessment.
An adequate vis oelasti model needs, therefore, to be provided and a suitable
methodology needsto bedeveloped inorderto postulatethepart inservi ebehaviour.
The present work aims at the predi tion of a mi ro part behaviour during its duty
servi e. Toa omplishthe latteranhybrid methodology isproposed. Firstly,the
mi ro-part thermo-rheologi al histori will be obtained through pro ess simulation by using
Autodesk Moldow ®
tostudypartlling, pa king andwarpage. Thelatterinformation
willbeusedaskeyinputtoAbaqus ®
,wherethemi ropartdutyservi ewillbestudied.
In order to obtain an assertive predi tion of the mi ro part duty servi e the adequate
vis oelasti onstitutivemodelmustalsobepromptedtoAbaqus ®
. Toadequatelymodel
themi ro-partvis oelasti ity,Dynami alMe hani al TemperatureAnalysis(DMTA)will
be arriedout to establish the mi ro-part history at dierent pro essing onditions and
hen e provide insight in what on erns its me hani al behaviour under ertain duty
y les.
In this resear h work, it is intended to develop methodologies and pro edures for
me hani al tests at mi ro s ale, ombining a omplete stru tural analysis based on an
inje tion simulationwithdatafromme hani al tests.
Themain obje tivesare:
literature reviewon polymers onstitutive models,polymertesting, vis oelasti ity
and polymersme hani al behavior;
to runDMTAteststoobtaindata,establishing theme hani albehaviourof mi ro
parts at spe i pro essing onditions;
to develop a polymer onstitutive model from the me hani al tests data using
te hniques su h asthetime temperature superposition prin iple;
toassessme hani alperforman eofami ro omponentwhensubje tedtoa ertain
Literature review
2.1 Vis oelasti ity
Classi altheoryofelasti itydealswithme hani alpropertiesofelasti solids,for whi h,
ina ordan e withHooke's law, stress is always dire tlyproportional to strain insmall
deformations but independent of the rate of strain. Classi altheory of hydrodynami s
deals withpropertiesofvis ous liquids,for whi h,ina ordan e withNewton'slaw, the
stress is always dire tly proportionalto therate ofstrain but independent of the strain
itself.
Although Hooke's law approa hes the behaviour of many solids for innitesimal
strains,andthatNewton'slawapproa hesmanyliquidsbehaviourforinnitesimalrates
of strains, some deviations may be observed under ertain situations. Thesedeviations
may be distinguishedintwo:
stress-strain relations are more ompli ated when nite strains are imposed on
solids that andeform substantiallywithoutbreaking -theso allednon-Hookean
deformation;
polymeri solutions and undilutedun rosslinked polymersmaydeviatefrom
New-ton's lawwhen insteadyowwithnite strainrates - Non-Newtonianow.
Polymeri materials may present both vis ous resistan e to deformation and
elas-ti ity. Polymers' linear vis oelasti domain is limited to relatively small stresses. This
kind ofmaterials has hara teristi s between thoseof idealelasti s andthose fromideal
Newtonianuids. Anot ompletelysolidmaterialdeformsthroughtimewhensubje ted
to a onstant stress,insteadof having a onstant deformation,in otherword it reeps.
When su ha bodyis onstrained at onstant deformation, thestress required toholdit
diminishes gradually, or relaxes. When an external onstant stress is applied to a not
ompletely liquid material it stores some of the input energy, instead of dissipating it
all as heatduring its owand it mayre overpart of its deformation when thestress is
removed (thisis alledelasti re oil) [8℄.
Idealelasti solidsstoreallsupplieddeformationenergywhilstidealNewtonianuids
dissipate all supplied energy (ex ept hydrostati state of stress) [6℄. When sinusoidally
os illating stresses are applied to su h bodies, the resulting strain is neither exa tly in
phasewiththestress(asitwouldbeforaperfe tlyelasti solid)or
90
◦
would be for aperfe tly vis ous liquid). It isthenpossibleto build uniaxial vis oelasti
behaviourmodels ombining solidsbasi physi almodels-springs andNewtonianuids
models - dampers[6 ℄.
In ertain ases,su hasvul anized( ross-linked)rubber,owisnotpossibleandthe
material has a unique onguration assumed inthe absen e of deforming stresses. The
vis ousresistan etodeformationisnoti edduetotherubber'sdelayona hangeinstress
[8℄. Insu hmaterialssomeoftheenergyinputisstoredandre overedinea h y le,and
some isdissipated intheformof heat. Materials whose behaviourexhibits su h
hara -teristi sare alledvis oelasti . Ifbothstrainandrateofstrainareinnitesimalandthe
time-dependent stress-strain relations an be des ribed by linear dierential equations
with onstant oe ients,linear vis oelasti behaviouro urs. Inanexperiment with
these materials the ratio ofstress tostrain isa fun tionof time(orfrequen y).
A " onstitutive equation" or "rheologi al equation of state" relates stress, strain
and their time dependen e. Finite strain/strain rates produ e ompli ated onstitutive
equations. Onthe other hand,innitesimal strain/strain rates ( orresponding to linear
vis oelasti behaviour) have relatively simpler onstitutive equations. Vis oelasti ity's
onstitutive laws maybeexpressedthroughthe Boltzmann Superposition prin iple [6℄.
2.2 Linear onstitutive equations (simple shear)
Polymersvis oelasti ityisdependantofmole ularstru ture,modesofmole ularmotion,
mole ular weight and its distribution, temperature, on entration, hemi al stru ture,
among others.
2.2.1 Equations of hange
Experimental measurements of me hani al properties relate stress and strain applied
to a body. Assuming that the pro ess is isothermal (therefore despising the equation
of onservation of energy), this relation depends not only on stress-strain onstitutive
equation but also on equation of ontinuity and the equation of motion. The earlier
expresses onservationof mass(Equation2.1)
∂
∂t
ρ = −
3
X
i=1
∂
∂x
i
(ρυ
i
)
(2.1)andthe latter expressesthe onservation ofmomentum (Equation2.2
ρ
∂
∂t
υ
j
+
3
X
i=1
υ
i
∂
∂x
i
υ
j
!
=
3
X
i=1
∂
∂x
i
σ
ij
+ ρg
j
(2.2)Intheseequations
ρ
isdensity,t
istime,x
i
arethreeCartesian oordinatesandυ
i
are the omponents of velo ity in the respe tive dire tions of the previous oordinates. InEquation2.2,the index
j
mayassumesu essivelythevalues1, 2, 3
,g
j
isthe omponent ofgravitationala elerationinthej
dire tionandσ
ij
aretheappropriate omponentsof thestress tensor. Mostof theexperimentsare madein onditions that make both sidesarenegligible. Here,internalstatesofstressandstrain anbe al ulatedfromobservable
quantities bythe onstitutive equationalone [10℄.
2.2.2 Innitesimal strain tensor
Like in a perfe tly elasti body, vis oelasti bodies exhibit a state of deformation at a
given point whi h an be spe ied by a strain tensor, representing relative hanges in
dimensions and angles of a small ubi al element in that position. The state of stress
is spe ied by a stress tensor whi h represents the for es a ting on dierent fa es of a
ubi alelement. Therateofstraintensor des ribesthetimederivativesof theserelative
dimensions andangles.
For an innitesimal deformation, the omponents of the innitesimal strain tensor,
inre tangular oordinates withthethree artesian dire tions
(1, 2, 3)
are:γ
ij
=
2
∂u
1
∂x
1
∂u
2
∂x
1
+
∂u
1
∂x
2
∂u
3
∂x
1
+
∂u
1
∂x
3
∂u
2
∂x
1
+
∂u
1
∂x
2
2
∂u
2
∂x
2
∂u
2
∂x
3
+
∂u
3
∂x
2
∂u
3
∂x
1
+
∂u
1
∂x
3
∂u
2
∂x
3
+
∂u
3
∂x
2
2
∂u
3
∂x
3
(2.3)Here,
x
i
andu
i
arethe point oordinates and its displa ement in thestrained state (u
i
= x
i
− x
0
i
), respe tively. The rateof strain tensor,˙γ
issimilar to strain tensor withu
i
repla edbyυ
i
,thevelo ityofdispla ement [10 ℄.2.2.3 Stress tensor
Stress omponentsof
σ
ij
thatappearin2.2 an berepresentedby:σ
ij
=
σ
11
σ
12
σ
13
σ
21
σ
22
σ
23
σ
31
σ
32
σ
33
(2.4)For a stress omponent
σ
ij
the rst indexi
indi ates the plane that the stress is perpendi ulartoandthese ondindexj
denotestheparalleldire tioninwhi hthestress a ts, as seen in Figure 2.1. Normal stressesσ
ii
are usually positive for tension and negative for ompression [10 ℄.2.2.4 Constitutive equation for linear vis oelasti ity (simple shear)
When deformation is homogeneous/uniform, stress and strain omponents do not vary
with position and are independent of
x
i
. In ase of simple shear, where two opposite fa es aredispla ed by sliding, both tensors assume more simple forms. Assuming thattheplane
13
slides indire tion 1,straintensor isgiven by:γ
ij
=
0
γ
12
0
γ
21
0
0
0
0
0
(2.5)Figure2.1: Componentsof thestresstensor [3℄
where
γ
12
= γ
21
=
∂u
1
∂x
2
= tan(α) ≈ α
and stresstensor isgiven by:
σ
ij
=
−P
σ
12
0
σ
21
−P
0
0
0
−P
(2.6)where Pisanisotropi pressure. Strain
γ
12
andstressσ
12
arefun tionsoftimeandthey arerelated bya onstitutive equationfor linearvis oelasti ity:σ
21
(t) =
Z
t
−∞
G(t − t
′
) ˙γ
21
(t
′
)dt
′
(2.7)where
˙γ
21
= ∂γ
21
/∂t
isthe shear rate,G(t)
isthe relaxation modulus.Strain an beexpressedinterms ofthe history of thetimederivative of thestress:
γ
21
(t) =
Z
t
−∞
J(t − t
′
) ˙σ
21
(t
′
)dt
′
(2.8)where
˙σ
21
= ∂σ
21
/∂t
andJ(t)
is the reep omplian e.The equation that expresses stress in terms of the history of thestrain rather than
theits timederivative is:
σ
21
(t) = −
Z
t
−∞
m(t − t
′
)γ
21
(t, t
′
)dt
′
(2.9)wherem(t)isthememoryfun tionanditisequalto
−dG(t)/dt
andγ
21
(t, t
′
)
isareferen e state.Knowingtheshearrelaxationmodulus,thememoryfun tion,orthe reep omplian e
fun tionofamaterial,enables topredi titsstress-strainrelations aslongasmotionsare
2.3 Me hani al models for vis oelasti behaviour
Toperform al ulationsitisusefultohavethedesiredvis oelasti propertyasanequation
instead of a graph or table of data. It is then onvenient to have a general form that
ontains su ient parameters to t experimental data for a wide range of polymers.
This form an emulate the time or frequen y dependen e of the vis oelasti properties
byusingame hani al modelwithasu ient numberofelasti elements(representedby
springs) andvis ous elements (represented bydashpots) [10 ℄.
2.3.1 Maxwell Model
Oneofthesimplestmodelsthat anemulateavis oelasti systemistheMaxwellmodel,
depi ted in Figure 2.2. In this model, whi h is omposed by a series asso iation of
one spring and one dashpot, the total deformation is equal to the sum of ea h of the
deformations, i.e.,
ε = ε
s
+ ε
d
.Figure2.2: Maxwellmodel- adaptedfrom[23℄
Here, if
k
represents spring's stiness andµ
uid's vis osity the following equation an bewritten:dε
dt
=
dε
s
dt
+
dε
d
dt
=
1
k
dσ
dt
+
σ
µ
(2.10)A new on ept, the relaxation time of the element that an be understood as a
measure at the time required for stress relaxation an be written as
τ
i
= µ
i
/k
i
[10℄. Assuming thata onstant stressσ
0
a ts,integrating Equation 2.10 results in the reep behaviour,obtained as:ε(t) =
σ
0
µ
t + C1
(2.11)C1
an be obtained from the initial onditionε(0)
asσ
0
/k
, and dividing strain by stress, reep omplian e an be obtained as:J(t) =
t
µ
+
1
k
(2.12)Equation2.12suggeststhat reep omplian ein reaseslinearlyby
t/µ
as anbeseen inFigure 2.3.(a)Maxwell's reep omplian e urve
-adaptedfrom[6℄
(b) Maxwell's relaxation modulus
urve-adaptedfrom[6℄
Figure2.3: Maxwell's model behaviour
However, this is not noti ed experimentally and what happens is the opposite, a
de rease over time, so this simple model is not good enough to represent omplian e.
Still, making
ε = ε
0
and integrating 2.10this model an predi tstressas:ln(σ) = −
k
µ
t + C2
(2.13)Here, assumingthat
σ(0) = σ
0
one an writethat relaxationisdepi ted by:G(t) = k exp(−
kt
µ
)
(2.14)Thisis on omitant withthebehaviourexperimentally expe ted [6℄.
2.3.2 Voigt-Kelvin Model
Anothersimplemodeltorepresentavis oelasti materialistheVoigt-Kelvinmodel. The
latter is also omposed byone spring and one dashpot, but this timethe asso iationis
made inparallel.
Here, the ratio between thespring anddashpot isdened astheretardation time,a
measure of the time required for the extension of the spring to a hieve its equilibrium
length whileretarded bythe dashpot. Inthis model, reep omplian e isgiven by:
J(t) =
1
k
h
1 − e
−kt
µ
i
(2.15)Still,relaxation an be written as:
G(t) = k
(2.16)Similarlytothepreviousmodel,thisonefailsinpredi tingthe reep omplian esin e
it la ks the initial elasti response. Oppositelyto theprevious model, this one doesnot
predi t stress relaxation sin e
G(t)
isgiven by a onstant whi h is not time dependent and sodoesnot hange overtime, asit an be seeninFigure 2.5[6 ℄.(a) Voigt-Kelvin's reep omplian e
urve-adapted from[6℄
(b)Voigt-Kelvin's relaxationmodulus
urve-adaptedfrom[6℄
Figure 2.5: Voigt-Kelvin's modelbehaviour
2.3.3 Zener Model
There are more omplex models that an emulate a vis oelasti material's behaviour.
One of them isthe Zener model. This model lls some aws of previous models but it
stillis omposedbythem. It anberepresentedasone simpleMaxwellmodelinparallel
withone spring.
Thismodelstatesthat reep omplian e and relaxationaregiven by:
J(t) =
1
k
0
1 −
k
1
k
0
+ k
1
· e
−t
ρ1
(2.17)G(t) = k
0
+ k
1
e
−t
λ1
(2.18)where
ρ
1
andλ
1
aretime onstants, given by:ρ
1
=
µ
1
k
0
k
1
(k
0
+ k
1
)
(2.19)λ
1
=
µ
1
k
1
(2.20)Although the shapes of the urves produ ed bythis model are similar to those
ob-tained experimentally (Figure 2.7), it doesn't allow for a rigorous adjustment of the
behaviour of polymers sin e time onstants
ρ
1
andλ
1
are insu ient. On e polymers often have a mi rostru tural omplex stru ture they present a ontinuous spe trum oftime onstantsandthereis aneed ofusingmodels withenoughspringsand dashpotsto
a hieve agoodapproximation tothe polymer realbehaviour[6℄.
(a) Zener's reep omplian e urve
-adaptedfrom[6℄
(b)Zener'srelaxationmodulus urve
-adaptedfrom[6℄
Figure2.7: Zener's modelbehaviour
Improved models
To a hieve a good aproximation, more omplex models whi h in lude paired or series
groups ofsimpler models weredeveloped,su hasimproved Zenermodels or generalized
Maxwell models. The main dieren e between these is the addition of one element
omposedonly bya spring asit an be seeninZener'smodel[6℄.
Improved Zenermodels reep omplian e andrelaxation anbeobtained as:
J(t) =
1
k
0
+
n
X
i=1
1
k
i
h
1 − e
−t
ρi
i
where
ρ
i
=
µ
i
k
i
(2.21)G(t) = k
0
+
n
X
i=1
k
i
· e
−t
λi
where
λ
i
=
µ
i
k
i
(2.22)Due to this large amount of parameters, rigorous representation of vis oelasti
be-haviour demands an enormous experimental eort whi h may be ome time- onsuming
(a)Parallelasso iation-adaptedfrom [6℄
(b)Seriesasso iation-adapted from[6℄
Figure2.8: Improved Zener'smodels
2.3.4 Vis oelasti spe tra
Maxwell elements in series have the properties of a single element with
J =
P J
i
and1/µ
0
=
P 1/µ
i
. Voigt elementsinparallel have thesame properties asa single element withG =
P G
i
andµ
0
=
P µ
i
. However, whenmore ompli ated modelssu hasthose presentedbeforeareused, theseproperties relations arenot sodire t.A group of parallel Maxwell elements represents a dis rete spe trum of relaxation
times, ea h one
τ
i
being asso iated with a spe tral strengthk
i
. In a parallel arrange-ment thefor es areadditive and vis oelasti fun tionssu hasG(t)maybeobtained bysumming termsequal to those inEquation2.14, obtaining:
G(t) =
n
X
i=1
k
i
e
−t
τi
(2.23)AgroupofVoigtelementsinseriesrepresentsadis retespe trumofretardationtimes,
ea h beingasso iated witha spe tral omplian e magnitude
J
. In a seriesarrangement thestrainsareadditive,the omplian efun tionJ(t)
maybeobtainedbysumming2.15 overall theserieselements:J(t) =
n
X
i=1
µ
i
1 − e
−t
τi
(2.24)Theseequations antrelaxationand reepdatawithanydesireddegreeofa ura y
if
n
islargeenough bydeterminingthedis retespe trum of urves, ea hwithalo ationτ
i
andintensityk
i
. Experimentallyitisdi ulttodeterminetheparametersτ
i
ork
i
and its arbitrary hoi e may be enough to predi t ma ros opi behavior but wouldn't haveenoughvaluefortheoreti alinterpretation. This anbeavoidedbyanalysingvis oelasti
Ifthenumber of elements ina generalized Maxwell modelin reases indenitely, the
resultis a ontinuousspe trum, where ea hinnitesimal ontribution to systemrigidity
is
F dτ
and is asso iated with relaxation times betweenτ
andτ + dτ
. This ontinuous relaxationspe trumisdenedasH dln(τ )
,whereH = F τ
. Inthese onditions,Equation 2.23 be omes:G(t) = G
e
+
Z
∞
−∞
He
−t
τ
dln(τ )
(2.25)where
G
e
is a onstant added to allow for a dis rete ontribution of the spe trum withτ = ∞
for vis oelasti solidsand for vis oelasti liquidsitassumes value0.Thesame analogy anbe establishedfor theVoigt-Kelvin model. Here ifthemodel
is indenitely extent it will represent a ontinuous spe trum of retardation times
L
. Equation 2.24maybewritten as:J(t) = J
g
+
Z
∞
−∞
L
1 − e
−t
τi
dln(τ ) +
t
µ
0
(2.26)Here, an instantaneous omplian e
J
g
is added to exist the possibility of a dis rete ontribution whenτ = 0
. This parameter an not be determined experimentally but should be takenina ount [10 ℄.2.4 Polymeri materials
Industrial polymers onsist of large mole ules whi h are primarily ovalently bonded.
These mole ules may arrange themselves in main hains and may present side hains,
ir ular mole ulesand inter onne tions amongthem bymany me hanisms.
Atoms inside mole ulesmayintera t withatoms froma dierent mole ule. Thereby
largermole ules attra tea h othermore strongly than shorterones. Thisis ree ted in
polymersmelting points, whi h generally in rease within reasing hain length. Hen e,
melting points may be a good indi ator of the bond strength among mole ules of the
ompound.
Thesepolymer long hainsare omposed bysmall basi units alledmers,whi h are
repeated to generate the hain. The bigger the number of repeating mers, the larger
the degree of polymerization of the polymer, whi h an be evaluated by an average
number sin e dierent mole ules may have dierent hain lengths. As the number of
intermole ular attra tions between mers in reases, the total for e of attra tion and the
me hani al strength of the polymer generally in reases too, sin e me hani al failure is
asso iatedwithbreakingtheattra tivefor es between theatoms inmole ules. However,
in reasingthedegreeofpolimerizationindenitelydoesnotensureanin reaseinstrength
be ause the hain may be ome so long thatsome parts may behave independently and
an even break from the main hain. The large dimensions of these mole ules are also
responsiblefor theunique propertiesof polymersrelatively toother materials.
Thermalenergymay ausevibration oftheatomsandin reaseatomi andmole ular
motions. Thesemotions an be:
atomi vibrations in the arbon ba kbone hain, whi h may in rease the average
vibrationsperpendi ulartothebond dire tionwhi h an ause smallvibrationsin
bondangles;
rotation ofthe arbon atoms around thebonds ausing twisting ofthe hain.
Thesevibrations an be understoodasameasureofthekineti energy whi h
ontin-uously hanges with time. Therefore, the average interatomi distan e between atoms
in reases withtemperature.
Be ausebond within the hain ba kbone are ovalent,they areof highstabilityand
strength, and their rupture requires large amounts of energy. For es involved in
inter-a tion with neighbouring mole ules are signi antly weaker and failure of a polymer is
morelikelytoo urbybreakingthese onne tionsratherthanbyrupturingthemole ule
itself [18 ℄.
There anbestill onne tionpointsbetweenlinearpolymers hainsalongtheirlength
makinga ross-linkedstru ture. Whenthe ross-linkingagentinthepolymerisa tivated
by temperatures, hemi al ross links are obtained and therefore a thermoset polymer
may be obtained. Physi al ross-links may alsoo ur temporarily when longmole ules
inlinear polymersget entangle. Anotherphenomena that an o uris hain-bran hing,
where ase ondary haininitiates from a point on themain hain. These bran h points
leadto onsiderabledieren esinme hani al behaviourwhen omparingthesame
poly-mer withand withoutbran hingpoints.
Polymers hains may be of various lengths and therefore their mole ular mass will
also vary. Mass distribution is of importan e in properties of the polymer sin e it will
inuen iate the polymernalstate.
The manner in whi h polymers mers arrange themselves in the ba kbone lassies
thepolymer,makingita opolymerorahomopolymer. Ahomopolymerisasequen eof
identi almers inthe hainba kbone,forminglongequal groups. A opolymerisformed
when hemi al ombinationexistsinthemain hainbetweentwoor moredierentmers.
When ombination is made by groups a blo k opolymer is obtained. When they are
disposedinanyordera random opolymer isobtained.
Thesete hniques ofmodifying homopolymersmay alsoform grafts,long side hains
of a se ond polymer hemi ally atta hed to the base polymer, or blends. By applying
them,du tilityortoughnessofbrittlehomopolymersareenhan edaswellasthestiness
of some rubbery polymers. A rylonitrile-Butadiene-Styrene opolymer (ABS) is an
ex-ample ofablend. InABStheintrodu tion ofthe butadiene,therubberphase,improves
thepolymer impa tresistan e.Beyond enhan ing polymersme hani al properties, these
additivesmayalso improve polymer pro essabilityand resistan eto degradation[24 ℄.
2.4.1 Temperature inuen e on regions of vis oelasti behaviour
Thermoplasti s and elastomersshow verylarge hangesinpropertieswith temperature.
Thermoplasti s temperaturedependen ehastwopointsofmajor interest. Usually
poly-mers hange fromaglassystate to arubberystateaseitherthetemperatureisraisedor
the time-s ale of the experiment is in reased. At lowtemperatures, in theglassy state,
polymersare hard and brittle. Itsstiness relateswith hanges instored elasti energy
ondeformationwhi hareasso iatedwithsmalldispla ementsofthemole ulesfrom
equi-librium positions. Here,thermal energyis insu ient to over omethepotential barriers
rises, theamplitude ofvibrationalmotion be omesgreater,andthermal energybe omes
omparable to the potential energy barriers. Thereby, mole ules an adopt
onforma-tions where they rea h a minimum of free energy. These rubber-like deformations are
asso iated with hanges inthe mole ular onformations. Thissuggestthatthere isonly
one vis oelasti transition, from theglassy state to rubber. In pra ti e, there are more
relaxation transitions,asit an be seeninFigure 2.9[24 ℄.
Figure2.9: Transitionsinpolymer-adaptedfrom [24℄
The primary transition where the greatest hange inmodulus o urs is denoted
T
g
, and orresponds to a relaxation alledα
. Here the modulus de reases abruptly in the glass transition region. Glass transition should be seen as a region instead of a xedtemperature and itsextent an getupto
20
◦
C
[24 ℄.Beyond these regions of major importan e there are other regions requiring further
analysis. As the temperature isfurther in reased themodulus rea hesagain a plateau.
Thisnew plateu
E
2
an beseenin Figure2.10.Figure 2.10: Regions of vis oelasti behaviour- adapted from[21 ℄
Here short-range motions of thepolymers o ur very fast unlike thelong-range
mo-tions of hains that arestill restri ted by the presen eof strong lo al intera tions with
neighbouring hains,asFigure2.11exemplies.
In aseof ross-linkedpolymerstheseintera tionsareprimary hemi albonds,known
as entanglements in linear polymers. In the rubbery plateau segments of hains an
Figure 2.11: Symboli representation of longand short range relationships ina exible
polymer mole ule-adaptedfrom [10 ℄
o ur.
In the rubbery region, linear and ross-linked polymers behave approximately but
as the temperature rises this behaviour starts to dier. Be ause of the primary
hem-i al bondsthe hanges in modulus relatively to those of the glasstransition are small.
Modulusonlystartstode reaseagainsigni antlywhentemperatureapproa hestothose
where hemi aldegradationbegins. TheabruptdrophighlightedinFigure2.10isrea hed
due to mole ular motions ausedby thein rease of temperature until a state where all
themole ulesareabletobeginto translate. Theselarge temperaturesareresponsibleto
over ome the potential energy barriersor thebreakof lo al hainintera tions whi h are
no longer ableto prevent themole ular ow[24 ℄.
2.4.2 Thermoplasti s and thermosets
Polymers are usually divided inthree main ategories: thermoplasti s, thermosets and
elastomers.
When onsidering thermoplasti s, the main hain mole ules are bonded by primary
ovalentbondswithinthe hain. Se ondaryfor esbetweentheindu edorpresentdipoles
are responsible for the attra tion between hains . Therefore, mole ules are apable of
individualmotionandwhen asolidpolymerissubmitted toheatit an melt,beingable
to owand llamould. This apabilityofbeingpro essedbyinje tionmouldingmakes
thermoplasti s themost ommonlyusedtype ofpolymer.
Inthermosets, ovalent bonds mayexist between dierent hains. By existingthese
primary bondsbetween dierent mole ules, intermole ular for es are strengthened and
the identi ation of any long single hain is very di ult if not impossible, reating a
single large network instead of existing many single mole ules. Sin e ovalent bonds
annot be broken without degrading the properties of thepolymer, these materials are
not abletobemoltenbyheating,but they har andde ompose. Be auseofthereferred
ovalentbondsthesepolymersarestrengthenedbutthe apabilityofmouldllingislost
[8℄.
2.4.3 Amorphous and rystalline thermoplasti s
Likesolidmaterials, polymersmaybe lassieda ordingtotheirmole ular stru turein
Many polymers may form regions of ristalinity if ooled slowly from melt regions
so thatthe longmole ules an be ome ordered in thesolid state asseen in Figure2.12
a). Thedegree of rystalinitymayvarybe ause oolingfroma moltenstatemayindu e
imperfe tions ausedbytanglingandtwistingoflinear hains. Also,inorderto rystalize,
themole ulesmusthaveregularstru ture. Threedimensionaljuxtapositionofmole ular
hains may also lead to formation of rystallites, though for this to happen they have
to be oriented in the same way. This way solid polymers that present a periodi al
tridimensionalorganizationare alled rystallinepolymersandthosewhopresentregions
of rystallinity are alled semi- rystalline polymers [24 ℄. These rystallites may a t as
rosslinks, making the rubbery plateau more pronoun ed in polymers with a ertain
degree of rystallinity than in the amorphous ones. When a polymer has a ertain
degree of ristalinity,regionsthat an be easilyidentiedinamorphouspolymersareno
longer so well dened. Usually these kind of polymers manifest a signi antly less fall
in modulus over theglass transition temperature. It an rea h up to one or two orders
of magnitude when ompared to amorphouspolymers. Also, the hange in modulus or
loss fa tor with temperature is more gradual, what indi ates a larger relaxation time
spe trum. At higher temperatures rystalline regionsredu e mole ular mobility and so
the polymer stops behaving as a rubber-like material [24 ℄. Also, rystalline polymers
exhibitadierent maintransitionthantheamorphousones. Theymeltfromanordered
rystal dire tly to a liquid. This usually happens at a region of temperatures alled
temperature ofmelting
T
m
. Here, the modulusdrops sharplyto therubberyplateau.(a)Crystallinepolymerstru ture
(b)Amorphouspolymerstru ture
Figure 2.12: Typi alpolymerstru ture - adaptedfrom [4℄
Amorphous polymers are materials that do not have a long range periodi al
orga-nization, as Figure 2.12 b) suggests. These materials may present an elasti modulus
around
10
9
Pa at higher frequen ies, where the vis ous omponent is not sonoti eable.
Whenthematerialhashighmole ular mass,thereisaplateauinmodulusandmole ular
ow o urs at frequen ies below
10
−
5
Hz. This plateau in the rubbery region an be explainedduetotheentanglementofmole ulesandthephenomenaofphysi alrosslink-ing that tend to restri t mole ular ow through the formation of temporary networks.
At longtimes these entanglements tendto indu e irreversible owunlike what happens
with permanent hemi al ross-links (for example when a rubber is vul anized). The
modulus valueinthisplateau isdire tly proportionalto theamount of these ross-links
per unit of volume. In fa t, lower mass samples tend to not show a rubbery plateau,
mole ular weight this plateau is more noti eable sin e it will take longer time for the
hains to disentangle, in reasing its length along the timeaxes [9 ℄. In the glassy state,
at lower temperatures, the stiness of polymer is related to hanges in stored elasti
energy aused bydeformation, whi h areexplained bysmall displa ements ofmole ules
fromtheirequilibriumpositions. Athighertemperatures,intherubberystate,mole ular
hainshavegreaterexibilityand anassumeformswheretheminimumfreeenergystate
anberea hed. Therefore,elasti deformationsareasso iatedwith hangesinmole ular
onformations. Infa t, thereareseveraltransitions inanamorphouspolymer. Atlower
temperaturesthere are usuallymany se ondary transitions, alled
β
transitions, involv-ing small hanges inmodulus. These transitions an be related to side-group motions.The largest hange in modulus in this kind of polymers o urs in the glass transition
range oftemperature [24℄.
2.5 Linear time-dependent experiments (shear)
Constitutive equations des ribe theresponseof a linearvis oelasti material to various
patternsof stress andstrainin simpleshear.
2.5.1 Stress relaxation
Ifashearstrain
γ
isimposedinabrieftimeperiodξ
bya onstantrateofstrain˙γ = γ/ξ
, Equation 2.7be omesσ(t) =
Z
t
0
t
0
−
ξ
G(t − t
′
)
γ
ξ
dt
′
(2.27)where
t
0
is thetimewhen the strainimposition is omplete, sin e the rate of strain is zero before and after the dened interval. After some mathemati al simpli ations,where
t
0
is assumed to be zero and the timet
is assumed to be mu h larger than the intervalof appli ation ofthestrainξ
,Equation 2.27 an be written asσ(t) = γG(t)
(2.28)where
G(t)
istherelaxationmodulus. Generally aratio ofa stresstostrain is alled amodulusandforaperfe tlyelasti solidtheshearmodulusG
isσ/γ
. Intheparti ular ase G(t)isthe shear modulus time-dependent equivalent[10 ℄.2.5.2 Creep
Whenashearstress
σ
isappliedwithinabriefperiodandisheld onstant,thedependen e ofthestrainγ
ontime anbeobtained fromEquation2.8onthesamewayasEquation 2.28 wasobtained fromEquation2.27 andis givenbyHere,
J(t)
isthe reep omplian e,whi hfor aperfe tly elasti solidisequalto1/G
. However, for a vis oelasti materialthis isnot trueandJ(t) 6=
1
G(t)
[10 ℄.2.5.3 Sinusoidal soli itationsand dynami experiments
Stress may be varied periodi ally with sinusoidal soli itations at a frequen y
f
in y- les/se (Hz) orw = 2πf
in radians/se . A periodi experiment at frequen yw
is qualitatively equivalent to an experiment at timet = w
−
1
. Regarding the following
onstitutive equationfor a sinusoidal strain:
γ = γ
0
sin(wt)
(2.30)where
γ
0
isthemaximumamplitudestrain. Cal ulatingthederivativeone anobtain:
˙γ = wγ
0
cos(wt)
(2.31)Substituting in Equation 2.7 and making the variable hange
s = t − t
′
it an be obtained:σ(t) = γ
0
w
Z
∞
0
G(s)sin(ws)ds
sin(wt) + γ
0
w
Z
∞
0
G(s)cos(ws)ds
cos(wt)
(2.32)Figure2.13: Phaselagbetween stress and strain-adapted from[10 ℄
The two omponents of material's stress response may be distinguished as one in
phase (
sin(wt)
) an one other out of phase (cos(wt)
) with the imposed strain. This emphasises the ideathat vis oelasti materials have an intermediatebehaviourbetweenideal Newtonian uidsand ideal solids [6 ℄. Sin e the quantities inbra kets inEquation
2.32 are only fun tions of frequen y and not the elapsed, Equation 2.32 an be written
as:
σ = γ
0
(G
′
sin(wt) + G
′′
Here,two frequen y dependant fun tionsaredened:
G
′
(w)
,theshear storage mod-ulus andG
′′
(w)
,the shear lossmodulus. Asitis asinusoidal soli itationitisofinterest to formulatethe stressinan alternativeform, displayingthestressamplitudeσ
0
(w)
and
thephaseangle
δ(w)
among stressand strain:σ = σ
0
sin(wt + δ) = σ
0
cos(δ)sin(wt) + σ
0
sin(δ)cos(wt)
(2.34)Comparing equations 2.33and 2.34itis noti eablethat:
G
′
=
σ
0
γ
0
cos(δ)
(2.35)G
′′
=
σ
0
γ
0
sin(δ)
(2.36)anda newrelation an thenbe established, thelossfa tor:
tan(δ) =
G
′′
G
′
(2.37)Whenthe stressand strain vary sinusoidally, they an also be des ribed as omplex
numbers (refered withexponent
∗
) and sothe modulus isalso a omplex number given
by
G
∗
=
σ
∗
γ
∗
= G
′
+ iG
′′
(2.38)where the real part is the elasti /storage part and the imaginary partis the energy
dissipation partasilustratedingure2.14 [6℄:
Figure2.14: Ve torial representation of omplex modulus - adaptedfrom[10℄
Hereis noti eablethat
G
′
is the omponent of
G∗
inphasewiththe strainandthatG
′′
|G
∗
| =
σ
0
γ
0
=
p
G
′
2
+ G
′′
2
(2.39)Datafromsinusoidal an also be expressedintermsof a omplex omplian e
J∗ =
γ∗
σ∗
=
1
G∗
= J
′
− iJ
′′
(2.40) WhereJ
′
,the storage omplian e, is theinphase omponent of thestress to strain
ratio to stress,whilst
J
′′
,the loss omplian e isthe
90
out of phase omponent.G
′
and
J
′
aredire tlyproportionalto theaverageenergy storedina y leofdeformation whilst
G
′′
and
J
′′
are dire tlyproportional to theaverage dissipation or lossof energy as heat
inthesame y le.
Even if
J∗ = 1/G∗
, individual omponents of moduleand omplian e arenot re ip-ro ally related,thus an berelatedby[10 ℄:J
′
=
G
′
(G
′
2
+ G
′′
2
)
=
1
G
′
1 + tan
2
(δ)
(2.41)J
′′
=
G
′′
(G
′
2
+ G
′′
2
)
=
1
G
′′
1 + tan
2
(δ)
−
1
(2.42)G
′
=
J
′
(J
′
2
+ J
′′
2
)
=
1
J
′
1 + tan
2
(δ)
(2.43)G
′′
=
J
′′
(J
′
2
+ J
′′
2
)
=
1
J
′′
1 + tan
2
(δ)
−
1
(2.44)The omplexmodule anbedeterminedexperimentallyforawiderangeoffrequen ies
w
throughDynami Me hani alAnalysis(DMA)or,sin eitmayvarywithtemperature, throughDynami Me hani alThermalAnalysis(DMTA).Tipi ally,frequen ies anvaryfrom
10
−
5
to10
8
Hz
but below10
−
2
theexperiment is time onsuming and above10
2
resonan emay o ur depending onthetype of essayand thetest spe imen[6 ℄.
2.6 Dynami and transient experiments
For understanding thevis oelasti behaviourof polymers, data arerequired overa wide
rangeoftime(orfrequen y)andtemperature. Thisrangeoftimemayberedu edthrough
theappli ation of the time-temperature superpositionprin iple aswill be dis ussed
fur-ther. Also the number of experiments may be redu ed through the relations between
properties. In order to over a wide range of both timeand temperature there are ve
main lasses ofexperiments:
Transient measurements: reep and stressrelaxation
Low-frequen y vibrations: freeos illating methods
High-frequen y vibrations: resonan emethods
Wave propagationmethods
Ea h te hnique hasits own rangeoffrequen y appli ability as an beseen inFigure
2.15.
Figure2.15: Range offrequen ies for dierent te hniques[24℄
Measurements of polymers' vis oelasti behaviour are sensitive to small hanges in
temperatureandso,a ontrolledtemperatureenvironmentisessential. Insomeparti ular
ases, su h asnylon, humidity must also be ontrolled sin e they are very hygros opi .
These measurements must be done in values of strain that does not ex eed one per
ent sin e that is taken as the limit value for linear vis oelasti ity [24℄. Within this
per entage the hange in ross-se tion is small and hen e stress ause by strain is also
small. When preparing an experiment, stresses indu ed by the lamping for es at the
end of the test spe imenwill indu e a dierent state of stress from theother zones. To
avoid the inuen e of this stress the overall length of the spe imen must be 10 times
its diameter or when dealing with oriented samples
10pE/G
whereE
is the Young's modulus inthe bre dire tion andG
the shearmodulus.Many dynami te hniques areused to overwide ranges of frequen ies, su h asfree
os illating pendulum methods, for ed vibration te hniques or DMTA. Free os illating
pendulumaresimplebut anonlygofrom
10
−
1
to10Hz
anditsresultsdoesnotpresent mu hrepeatability. Onthe otherhand,for edvibration methodsaremore omplexandmayyieldhigherreprodu ibilityand anextentthe frequen yrangeinmoreonede ade.
Nevertheless,the frequen yofvibrationsdependsonthestinessofthespe imen,whi h
varieswithtemperature [24℄.
However, another widely used te hnique is DMTA (Dynami me hani al thermal
analysis). Here, a small deformation is applied to a sample in a y li manner. This
allowstostudythematerialsresponsetostress,temperature,frequen yandothervalues.
DMTAdiersfromThermome hani alAnalysis(TMA)sin ethelatterappliesa onstant
stati for etoamaterialandevaluatesthematerial hangewithtemperature orintime,
reporting dimentional hanges, the oe ient of thermalexpansion. Ontheohter hand,
inDMTAanos illatoryfor eisapplied andmaterialsresponseinstiness anddamping
is obtained. Re ent DMTA ma hines are also able to measure the same properties as
Theamountofdeformationthatthesampleexhibitsisrelatedtoitsstiness. Thanks
to high pre ision sensors in DMTA equipments, su hs as LVDT sensors, damping may
bemeasuredwithhighpre isionandapplyingthistothemeasuredmodulus,thestorage
modulus andlossmodulusmaybe obtained.
TestsinDMTAmaybedoneinvarious manners,su hasdual antilever, three-point
bending, tension, ompression or shear sandwi h. Ea h testing mode and the geometry
of the sample has its own appli ability. It is di tated by thesample's physi al state at
thebeginning of the experiment and its di ulty inapplyingtheload. Some guidelines
to preferen ialgeometries and typesoftestsaregiven inTable 2.1.
Table2.1: Testingmodesand spe imengeometries for DMTA [17 ℄
Sample modulus Preferred Geometry Sample Thi kness (mm) Free Length (mm) Ideal Heating Rate
(
◦
C/min)
10
5
to10
10
Tension 0.02to 1 2to 10 510
6
to10
10
Single Cantilever 1 to 2 5to 10 310
6
to10
10
Single Cantilever 2 to 4 10 to 15 210
6
to10
10
Dual Cantilever 2 to 4 10 to 15 210
8
to10
12
Three point bending 1 to 3 10 to 20 310
7
to10
11
Three point bending>
4 15 to 20 210
2
to10
7
Simple Shear 0.5to 2 5to 10≤
210
2
to10
7
Compression 0.5to 10 15 to 20≤
2DMTA latest developments allows tests to be made from temperature ranges from
−150
◦
C
to600
◦
C
andfrequen y ranges from10
−
6
to200Hz
.2.7 Boltzmann Superposition prin iple
Boltzmann SuperpositionPrin iple statesthattheresponseofamaterialto agiven load
is independent of the response of the material to any pre-imposed load so the strain
response due to omplex loading is the sum of the strains due to ea h step, as an be
seen inFigure2.16. Thisisonly validinlinearvis oelasti region.
Considering a bar subje ted to uniaxial stress variable in time, reep stress an be
approximated by a series of in rements
∆σ
i
applied in times t'i
. The total resulting strainint>t'n
is given by Equation2.45.Figure 2.16: Boltzmann superposition prin iple- adapted from[6℄
ε(t) = ∆σ
1
J(t − t
′
1
) + ∆σ
2
J(t − t
′
2
) + ... + ∆σ
n
J(t − t
′
n
)
(2.45)where
J(t)
isthe Complian e fun tion.Whenthe number ofin rementstends to innity,Boltzmann SuperpositionIntegral
is obtained:
ε(t) =
Z
t
−∞
J(t − t
′
)
dσ(t
′
)
dt
′
dt
′
where
J(t − t
′
) = 0
if
t < t
′
(2.46)σ
i
(t) =
Z
t
−∞
G(t − t
′
)
dε(t
′
)
dt
′
dt
′
where
G(t − t
′
) = 0
if
t < t
′
(2.47) At timet
′
1
stressσ
1
isapplied and thestrainindu ed an be givenbyε
1
(t) = σ
1
J(t)
(2.48)A ording to linearvis oelasti ity
J(t)
remainsequal for allstresses ina given time. If stressesin reases byσ
2
− σ
1
at timet
′
2
thestrainin rease isgiven byε
2
(t) = J(t − t
′
2
)(σ
2
− σ
1
)
(2.49)Also,ifstressin rement
σ
3
− σ
2
is applied,the strainin rease isε
3
(t) = J(t − t
′
3
)(σ
3
− σ
2
)
(2.50)Furtherstrains ausedbystressin rements,whi h an beeitherpositiveornegative,
2.8 Time temperature superposition prin iple (TTSP)
The temperature dependen e of polymer properties is ofmajor importan e sin e either
polymers and elastomers present hanges in their me hani al properties with hanging
of temperature. Polymeri parts should be proje ted so that reep and relaxation
phe-nomena are slow, preferentially taking several years, albeit they an not be negle ted.
Hen e,experimentalpro edures toobtain reep omplian eandrelaxationmodulusdata
should also be long enough, what is not a eptable sin e it would be time onsuming.
Short time testsare thenmade and the results are extrapolated to working onditions.
Using the time temperature superposition prin iple it ispossible to redu e testing time
by repla ing it with shorter experiments at higher temperatures, sin e an in rease in
temperaturea eleratesmole ular motions. Thisprin iplealsoallowstoobtaina
mole -ular interpretation of the vis oelasti behaviour sin e generally polymers hange from
glass-like torubber-like behaviourasthetemperaturerisesor thetimes ale isin reased
[8℄,[18 ℄.
TTSP prin iple isbased onthehypothesesthat vis oelasti materials have a simple
rheologi al behavior, i.e. thatevery propertyhasthesame dependen y oftime[6℄.
If
P
designates any vis oelasti property, time temperature superposition prin iple states that:P (t, T ) = P (t
r
, T
r
)
(2.51)where
T
r
is the referen e temperature at whi h the properties are desired andt
r
is theredu ed time,given by:t
r
=
t
α
T
(T )
(2.52)
inwhi hintervenestheshiftingfa tor
α
T
(T )
. Thesame anbewritteninfrequen y:P (w, T ) = P (w
r
, T
r
)
where
w
r
= wα
T
(T )
(2.53)The implementation of this prin iple requires hoosing a referen e temperature as
wellasobtaining urvesforthedesiredpropertyto bemeasuredat severaltemperatures.
It is then possible to obtain a master urve through shifting measured properties at
ea h temperature by their respe tive shifting fa tors, in the time s ale. This master
urve may then serve asthe basis to the predi tion of the polymer behaviour in other
situations, sin e by shifting the isothermals, the time s ale is expanded and one an
extrapolate the behaviour through longer times. Experimental al ulation of thefa tor
α
T
(T )
isessentialfortheappli ationoftheTTSPsin ethesearethevaluestobeapplied to shift theisothermals horizontally.Whilemakingmaster urves,variableshavingunitsoftime(orre ipro al)aresubje t
to horizontal shifts
α
T
while units of stress (or re ipro al) are subje t to verti al shiftsb
T
,both temperature dependent.Whenall retardation andrelaxation me hanisms have thesame temperature
depen-den easwellasstressmagnitudesatalltimes,thematerial anbesaidas
thermorheolog-i allysimple. This anbenotedwhenbuildingamaster urveif,onadoublelogarithmi
plot ofvis oelasti fun tions,these an besuperimposed by shiftingthem only
temperature is the same and if the material is onsidered thermorheologi ally simple
[16 ℄. Someauthorsassumethatstressmagnitudesareproportionaltotheprodu tofthe
densityand temperature. So theverti alshift fa tor maybe al ulated by[7 ℄
b
T
=
T
0
ρ
0
T ρ
(2.54)Anotherwaytodeterminethene essitytoin ludeaverti alshift(independentoftime
or frequen y shift), isto nd thehorizontal shift ofloss anglene essaryto superimpose
on alossangle versus
log(|G
∗
|)
plot, alledavan Gurp-Palmen plot[14 ℄.2.8.1 Mole ular interpretations of time-temperature equivalen e
Site model theory
This theory states that there are two regions, separated by an equilibrium free energy
dieren e
∆G
1
− ∆G
2
,whereea hofthetermsarethebarrierheightspermoleasFigure 2.17 shows. Also,this modelis based on thetransition probabilty for ajump from site1 tosite 2whenastress isappliedand a hange ofpopulations inbothsiteso urs [24℄.
Figure2.17: Two-site model- adaptedfrom [24 ℄
Aftersome assumptionsand mathemati al dedu tions it an be stated that the
fre-quen yofmole ular onformational hangesdependson thebarrierheight
∆G
2
and not on thedieren e between sites,asit anbe seeninEquation2.55 . [24 ℄ν =
A
′
2π
e
−∆G2
RT
(2.55) Here,A
′
is a onstant. This equation an be rewritten as:
ν = ν
0
e
−∆H
RT
where
ν
0
=
A”
2π
e
∆S
R
(2.56)Here,