master curve properties interrelationships with mix and ... etg.pdf · master curve properties...
TRANSCRIPT
Master curve properties interrelationships with mix and binder data
Dr. Geoffrey M. RoweAbatech
Presented at the
RAP ETG MeetingRAP ETG MeetingPhoenix, AZ
October 28-29, 2008
Objective
To discuss basic properties of the master curves for binder and mixesTo demonstrate how all properties for MEPDG input can be derived from the mix master curve
Why?
MEPDG based on viscosity aging profilesViscosity is estimated from G* and phase angle
It is possible to derive all this information from the mixture mIt is possible to derive all this information from the mixture master curveaster curve
New concepts
Relationship between slope of log E* (or G*) vs. ω and δGeneralized versus standard logistic functionsKaelble shift factor relationshipNeed for more frequenciesAdditional utility of data from E* master curve
How we have developed ideas
Phase angle data can be deduced from stiffness vs. frequency relationship – we don’t need to measureImprovements to help with determinations
Testing – add a few more frequenciesMaster curve functional – add a parameter to describe non-symmetrical shape of master curveShifting – use Kaelble modification to WLF
Mix vs. binder master curve
1.0E-04
1.0E-03
1.0E-02
1.0E-01
1.0E+00
1.0E+01
1.0E+02
1.0E+03
1.0E+04
1.0E+05
1.0E-03 1.0E-02 1.0E-01 1.0E+00 1.0E+01 1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E+06 1.0E+07
Frequency, Hz
Mix
and
Bin
der S
tiffn
ess,
MPa
Binder
Mix
Can use functional forms to describe mix and binder master curves.Alternate – can fit models to describe shape characteristic –for example PMB.
Increase in stiffness due to aggregate volumetrics
Mix to binder properties
Things we needHirsch modelRelationship between phase and G*
Others have shown that we can use Hirsch model to assess quality of RAP dispersion in HMA blends
Mix E* to binder G* - Hirsch
Works well for large range of mixture stiffness valuesPrevious slide is consistent with this information
Binder G* to binder δ
Christensen-Anderson proposed relationship linking G* to δRelationship is based on underlying relationship
ωωδ
log*log90)(
dGd
×=
Log-log relationshipFundamental relationshipApplies to a wide variety of materialsWe have looked at with polymers, asphalt, mixes, etc. etc.CA
SHRP A369SHRP A369
Binder
CA equationIt can be shown that the log-log relationship is related to the phase angle
Dickson and Witt (1974) and used in the development of the CA model.
Polystyrene
Very good fit with measured vs. calculated
0
10
20
30
40
50
60
70
80
90
0 10 20 30 40 50 60 70 80 90
Measured δ, degrees
Cal
cula
ted
δ, d
egre
es
Approx slopeDS Fit
Polystyrene
Estimated phase angle fits real data very well from the log-log slope information
0
1
2
3
4
5
6
7
8
9
-8 -6 -4 -2 0 2 4
Log Reduced Frequency, rads/sec (Tref = 132C)
G*,
Pa
0
10
20
30
40
50
60
70
80
90
δ, d
egre
es
Log G* Measured phase Est. phase
Phase angleShown – for a wide variety of materials – that –δ=90(dlogG*/dlogω)Analysis is consistent with that produced by discrete spectra analysis of G* or G’G” (or E equivalents)Technique can help with analysis
[ ]
[ ]2)(log
)(log
190
log*log90)(
ωγβ
ωγβ
αγω
ωδ+
+
+−=×=
ee
dEd
[ ]
[ ] )/11()(log
)(log
190
log*log90)( λωγβ
ωγβ
λαγ
ωωδ ++
+
+−=×=
ee
dEd
Standard
Generalized
RAP binder propertiesRAP binder properties1. Obtain mixture E*2. Use E* data to back-calculate binder G* using
Hirsch model3. Estimate log-log slope of G* vs. ω plot
a. Method 1 – fit CA modelb. Method 2 – obtain approximate slope from alternate
numerical method4. Use G* and d with 2.2.13 and 2.2.14 to estimate
binder viscositya. Other methods could be used to estimate viscosity
from G* data from analysis of frequency sweep data5. Apply aging and MEPDG parameters to
relationships obtained
Hirsch
( )
1
*3000,200,4100
11
000,10*3
100100,200,4*
−
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
××+
−−+⎥
⎦
⎤⎢⎣
⎡⎟⎠
⎞⎜⎝
⎛ ×+⎟
⎠⎞
⎜⎝⎛ −=
bbm GVFA
VMAVMA
PcVMAVFAGVMAPcE
58.0
58.0
*3650
*320
⎟⎠⎞
⎜⎝⎛ ×
+
⎟⎠⎞
⎜⎝⎛ ×
+=
VMAGVFA
VMAGVFA
Pcb
b
Hirsch model relates volumetrics and stiffness of binder to stiffness of mixtureCan calculate for single points or isotherms
Issues and problems
Various items in current scheme are problematic
Binder is used to dictate shift parametersSymmetric sigmoid
We would like to use better shifting techniques
Consequence – need more data points in isotherms to get better shiftingModification to shift factor relationship
Data quality
More recent testing on master curves for mixes enables more data points to be collected and with better data quality further assessment of models can be consideredNumber of test points/isotherm in present MEPDG scheme is limited resulting in numerical problems in some shifting schemesNeed in many cases to assume model as part of shift development
1.0E+01
1.0E+02
1.0E+03
1.0E+04
1.0E+05
1.0E-01 1.0E+00 1.0E+01 1.0E+02
Frequency, Hz
E' o
r E",
MPa
E' E"
Objective of better modelsLeads to better calculations
Spectra calculations and interconversionsBetter definition of low stiffness and high stiffness properties are critical if considering pavement performanceWork looking at obtaining binder properties from mix dataPhase angle interrelationshipsConsiderable evidence that we should be using a non-symmetrical sigmoid function
Sigmoid
Standard logistic
Generalized logistic
log( *) (log )Ee
= ++ +δ
αβ γ1 ω
λωγβλαδ
/1)log(1[*)log( ++
+=e
E [
Why generalized logistic
Allows non-symmetric sigmoid format consistent with asphalt material behaviorBinder CA equation also based on non-symmetric behavior
1
1.5
2
2.5
3
3.5
4
4.5
5
-15 -14 -13 -12 -11 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10
x
y
Generalized logistic example
Lower asymptoteEquilibrium modulus = 98 MPa
Upper asymptote = Equilibrium modulus = 22.3 GPa
Note T= λ
Generalized logistic example
1
1.5
2
2.5
3
3.5
4
4.5
5
-15 -14 -13 -12 -11 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10
x
y
Generalized logisticGeneralized logistic curve (Richard’s) allows use of non-symmetrical slopesIntroduction of additional parameter λ
When λ = 1 equation becomes standard logisticWhen λ tends to 0 – then equation becomes Gompertzλ must be positive for analysis of mixtures since negative values will not have asymptote and produces unsatisfactory inflection in curveMinimum value of inflection occurs at 1/e –or 36.8% of relative height
Minimum inflection
Standard logistic inflection
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
-6 -4 -2 0 2 4 6
x
y λ=-0.5λ=0.0 Gompertzλ=0.6λ=1.0 Logisticλ=2.0
Typical range in inflection values
Kaelble shift factors
Kaelble shift factors
Working with materials from MEPDG E* database – observed that shifting works best with Kaelble modification to WLF equation (Arrhenius and WLF – do not work)
||)(
a log2
1
g
gT TTC
TTC−+
−−=
Data from MEPDG database
1.0E+02
1.0E+03
1.0E+04
1.0E+05
1.0E-07 1.0E-05 1.0E-03 1.0E-01 1.0E+01 1.0E+03
Reduced Frequency, T ref = -21.1 oC
E*, M
Pa
TLA Modified - E*
AC-20 - E*
Kaelble shift factors
WLF, Arrhenius, polynomial fits to shift factors are unstable as data is extrapolated to extreme conditionsKaelble provides a sigmoid shift factor relationship
-10
-5
0
5
10
15
20
25
-60 -40 -20 0 20 40 60 80
Temperature oC
log
a T
Data obtained in this range
Pavements temperatures in this range
WLFArrheniusKaelble
Tg = -20oC
-10
-5
0
5
10
15
20
25
-60 -40 -20 0 20 40 60 80
Temperature oC
log
a T
Data obtained in this range
Pavements temperatures in this range
WLFArrheniusKaelble
Tg = -20oC
Kaelble shift factors
-4
-3
-2
-1
0
1
2
3
4
5
-20 -10 0 10 20 30 40 50 60
Temperature C
Log
a T
Gordon and and Shaw shiftsKaelble fit
Typical fit – have analyzed all MPA mixes in MEPDG data base and all give similar fits.
NeedsPhase angle important in some MEPDG work – used to derive viscosity – can obtain from back-calculation of Gb* from mix data and then use log-log slope or dy/dx of CA model to obtain phaseCan use method to assess data quality –often measurement of phase is poorReduces need to always measure phase –can be easily deducedCan go back to old historical data and obtain phase information
Example RAP – mix to binder
1.0E-04
1.0E-03
1.0E-02
1.0E-01
1.0E+00
1.0E+01
1.0E+02
1.0E+03
1.0E+04
1.0E+05
1.0E-03 1.0E-02 1.0E-01 1.0E+00 1.0E+01 1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E+06 1.0E+07
Frequency, Hz
Mix
and
Bin
der S
tiffn
ess,
MPa
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
90.0
Phas
e A
ngle
, deg
rees
E*, psi (Calc.) E*, psi (Measured)E*, Binder (Hirsch) Phase (Hirsch)Phase (Slope) Phase (Bonnaure et al.)Phase (SHRP A-404)
Ref: Jo Daniel
??
Example RAP – binder G* & δ
1.0E+04
1.0E+05
1.0E+06
1.0E+07
1.0E+08
1.0E+09
1.0E-01 1.0E+00 1.0E+01 1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E+06 1.0E+07 1.0E+08
freq (rad/s)
G* (
Pa)
0.0
15.0
30.0
45.0
60.0
75.0
90.0
δ, d
egre
es
G* Back-calculated G* CA FitPhase - from dy/dx CA fit Phase - from approx slope, CA FitPhase - DS on Back-calculated Phase - from approx slope, Back-calculated
Ref: Jo Daniel
Problems with older data
1.0E+02
1.0E+03
1.0E+04
1.0E+05
1.0E+06
1.0E+07
1.0E+08
1.0E+09
1.0E+10
1.0E+11
-8.00 -6.00 -4.00 -2.00 0.00 2.00 4.00 6.00 8.00
Log Reduced Frequency (Hz)
Log
E* (m
ix) o
r G* (
bind
er),
Pa
0
10
20
30
40
50
60
70
80
90
Phas
e A
ngle
(deg
rees
)
Mix E*, Measured Binder G*, measuredE*, Hirsh Pa Gb est, PaCA fit to Gb Hirsch Mix PhaseBinder phase, measured Binder phase, Pred., Hirsch, CA fit, fixed GgBinder Phase, Approx Slope from best fit area Binder Phase, CA Method
ALF5, AC10ALF5, AC10
SummaryE* vs. ω mixture data provides significantly more information than currently assumed
Mixture phase angleBinder G* and δTemperature shift factorsCan determine for individual isotherms if needed
RecommendationsIncrease test frequencies - does not significantly increase preparation/testing timeUse free shifting and generalized logistic sigmoidInvestigate relationships further, Hirsch, CA models etc.