Evaluation of Low-Temperature Properties of HMA Mixtures

state ofwith thisgoal of the

inders andam lowthe thermaler. Anrect tensionhe TSRST

icate thattures areing is useded asphaltecessary.

Dow

nloa

ded

from

asc

elib

rary

.org

by

The

Uni

vers

ity o

f Q

ueen

slan

d L

ibra

ry o

n 05

/13/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

Evaluation of Low-Temperature Properties of HMA MixturesPeter E. Sebaaly1; Andrew Lake2; and Jon Epps3

Abstract: The deterioration of flexible pavements due to low-temperature cracking is a significant and costly problem in theNevada. The Nevada Department of Transportation initiated several research efforts aimed at exploring Nevada’s problemdistress. The research evaluated several newly developed low temperature performance tests under Nevada’s conditions. Theresearch was to determine the applicability of the tests for characterizing the low-temperature response of Nevada’s asphalt bhot-mixed asphalt~HMA ! mixtures. This paper summarizes Nevada’s experience with the Strategic Highway Research Progrtemperature tests and specifications, highlighting the effectiveness of the Superpave Performance Binder Grading System andstress restrained specimen test~TSRST!. The contribution of asphalt aging to Nevada’s cracking problem is also included in the papinvestigation of the Superpave Performance Graded Binder tests has determined that the bending beam rheometer and the ditest correlate very well and it may not be necessary to run both tests, as they are set up in the current Superpave specifications. Tappears to provide the greatest value for evaluating low-temperature properties of HMA mixtures. Findings from the research indthere are some significant correlations between the low-temperature properties of asphalt binders and HMA mixtures if the mixaged appropriately. This emphasizes the need to implement the appropriate conditioning procedure when low-temperature crackas part of the mix design and evaluation process. On the other hand, the research showed that, when using polymer-modifibinders, the low-temperature grade of the asphalt binder may be conservative enough that testing of the HMA mix may not be n

DOI: 10.1061/~ASCE!0733-947X~2002!128:6~578!

CE Database keywords: Temperature; Pavement deterioration; Nevada; Cracking.

engackial,nifi-ingemereighf

suedise oasence-

ly orMA

w-e to

s inck-sures tou-luate

hatpro-

r asined,tureity,ers,ac-

-tressturelops

of30

ave-an

7.

of

ox

usonetor.ibleape

6-

Introduction

Low-temperature cracking of hot-mixed asphalt~HMA ! pave-ments has been a serious concern to pavements/materialsneers for many years. The mechanism of low-temperature cring is very complex in nature due to the influence of materstructural, and environmental conditions on the process. Sigcant research over the years has enhanced the understandthis form of pavement distress; however, changes in binder chistry and loading conditions over the past 20 years has uncovthe need for more advanced analysis. In 1987, the Strategic Hway Research Program~SHRP! was established with the intent oproviding advanced technologies to address many of these isWhile SHRP researchers considered all forms of pavementtress, low-temperature cracking received more than its sharattention. When the products of the SHRP research were releto the public in 1992 in the form of the Superpave PerformaBinder Grading System~Superpave PBGS!, several new tests be

1Professor of Civil Engineering, Univ. of Nevada, Reno, NV 8955E-mail: [email protected]

2Formerly, Graduate Student, Dept. of Civil Engineering, Univ.Nevada, Reno, NV 89557.

3Materials Engineering Manager, Granite Construction, Inc., P.O. B2087, Sparks, NV 89432.

Note. Discussion open until April 1, 2003. Separate discussions mbe submitted for individual papers. To extend the closing date bymonth, a written request must be filed with the ASCE Managing EdiThe manuscript for this paper was submitted for review and posspublication on May 11, 2001; approved on January 18, 2002. This pis part of theJournal of Transportation Engineering, Vol. 128, No. 6,November 1, 2002. ©ASCE, ISSN 0733-947X/2002/578–586/$8.001$.50 per page.

578 / JOURNAL OF TRANSPORTATION ENGINEERING / NOVEMBER/DEC

J. Transp. Eng. 200

i--

of-d-

s.-fd

came available, allowing engineers and researchers to directindirectly evaluate the response of asphalt binders and/or Hmixtures to low temperatures.

Nevada has historically experienced excessive lotemperature cracking and other forms of pavement distress duits unique climate. The extreme cold and warm temperaturethe region lends itself well to rutting and low-temperature craing. These extreme conditions, however, place enormous preson material engineers who must use nontraditional techniquecombat both forms of distress. With the introduction of the Sperpave system, engineers are now better equipped to evapavement performance prior to construction.

Background

Low-temperature cracking of HMA pavements is a distress taffects many regions of the United States and Canada. Thecess is associated with volumetric changes in the HMA layethe pavement temperature drops. If the pavement is unrestrathe contraction associated with the drop in pavement temperawill not result in the development of tensile stresses. In realhowever, friction developed between the HMA and base layalong with the infinite length, restrain the pavement from contrtion and in return induce tensile stresses in the pavement~Vinsonet al. 1989; Jung and Vinson 1994a!. As the temperature continually drops, tensile stresses increase until a point where the sequals the tensile strength of the mixture. At the temperawhere tensile stress equals tensile strength, a crack devethrough the HMA layer to relieve the stress.

The pattern of thermal cracking is transverse to the directiontraffic and typically spaced at regular intervals between 3 andm, depending on the age, thickness, and conditions of the pment. Although initially harmless, with time these cracks c

t

r

EMBER 2002

2.128:578-586.

ent,ave

d ts toly toavre-

turthe

c-

ropperf ais

alngtesth

ess

tourettes

imest,

wasre

stnceat-rtiesaveST

A aAecsileilurres

ges itssefle

for-

ort-er.ecttruc

for-t isandtheac-

pe-

r-rsity

resests.mixwaswasilarce-hefordure.w-enseayshalt-

as

rop-P-

-ultsrsnd

gs.

gedermep-

nd

ly-

ofinged totheST

nghanse,e.p ansucethe

ifi-per-

inthens

rfor-

testST

hat it

Dow

nloa

ded

from

asc

elib

rary

.org

by

The

Uni

vers

ity o

f Q

ueen

slan

d L

ibra

ry o

n 05

/13/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

allow the transfer of water and fines in and out of the pavemleading to performance problems and the eventual loss of pment life.

Over the years, a number of test methods have been useevaluate the response of asphalt binders and HMA mixturelow-temperature cracking. Many tests have been used directmeasure failure properties at low temperatures, while others hbeen primarily used for modeling and prediction purposes. Aview of the more common test methods is presented below.

The Superpave PBGS introduced several new low-temperaasphalt-binder tests to the asphalt paving industry. Amongtests introduced were the bending beam rheometer~BBR! and thedirect tension~DT! tester for measuring the rheological and frature properties of asphalt binders at low temperatures.

The BBR is a creep test used to measure the rheological perties of asphalt binders at low temperatures. The test isformed through the application of a static load to the center osimply supported beam. A time history of load and deflectiongenerated from the test that enables stiffnessS(t) and its rate ofchange with loading time~m! to be determined through classicbeam theory. The DT test was developed to measure the streand failure strain of asphalt binders at low temperatures. Theis performed by subjecting a dogbone shaped specimen ofasphalt binder to a uniaxial tensile load while recording the strstrain properties at failure.

In 1989, Vinson et al. compiled a summary of the methodsevaluate the resistance of HMA mixtures to low-temperatcracking~Vinson et al. 1989!. The summary included the indirecdiametral tension test, the direct tension test, the tensile creepthe flexural bending test, the thermal stress restrained spectest, theC* -line integral test, the three-point bend specimen teand the coefficient of thermal contraction test. Each testevaluated over a series of criterion, with the thermal stressstrained specimen test~TSRST! determined to exhibit the greatepotential for evaluating the low-temperature cracking resistaof HMA mixtures. Although extensive research went into validing the TSRST, fundamental viscoelastic and fracture propeneeded to predict performance under environmental and pment conditions cannot be determined from the test. The TSRtest is performed by cooling a beam-shaped specimen of HMa specified rate while restraining it from contracting. As the HMis cooled, thermally induced tensile stresses develop in the spmen as a result of being restrained. When the induced tenstress equals or exceeds the tensile strength of the mixture, faoccurs with the corresponding characteristics of fracture stand fracture temperature.

Thermal cracking occurs at cold temperatures when the aasphalt binder, which is already stiff, becomes brittle and loseability to dissipate stress through viscous flow. For this purpoSHRP has adopted two asphalt-binder-aging procedures to rethe contribution of short-and long-term aging to pavement permance. The rolling thin film oven test~RTFOT!, a test first intro-duced in California in 1959, was selected to simulate the shterm volatilization and oxidative hardening of the asphalt bindThe RTFOT provides an aged-asphalt binder that closely reflthe hardening experienced during the manufacture and constion of HMA. The pressure-aging vessel~PAV! was made avail-able to account for the effects of oxidative hardening on permance at intermediate and low temperatures. The tesperformed by exposing RTFOT residue to high air pressuretemperature in the PAV. The PAV is speculated to simulateoxidative aging equivalent to 5–10 years of service with the

JOURNAL OF TRA

J. Transp. Eng. 200

-

o

e

e

--

thte/

t,n

-

-

t

i-

es

d

,ct

s-

tual time frame found to be binder, temperature, and region scific ~Bahia and Anderson 1995!.

The selection of the HMA mixture aging protocol for Supepave was based on research performed at Oregon State Univeduring the Strategic Highway Research Program~Bell 1994; Bellet al. 1994!. Several short- and long-term aging-tests proceduwere evaluated using a series of mixture-and binder-related tFor the short-term aging research, conditioning of the looseusing a forced-draft oven-aging and extended-mixing processevaluated to determine its effectiveness. Long-term aginginvestigated by conditioning compacted samples using a simforced draft-oven-aging procedure, a pressure-oxidation produre, and a triaxial cell-aging method. At the conclusion of tinvestigation, the forced-draft oven aging of the loose mixture4 h at 135°C was recommended as a short-term aging proceFor long-term aging, both forced-draft oven aging and lopressure oxidation were recommended for further use. For dgraded mixtures with stiff asphalt binders, oven aging for 5 dat 85°C was recommended, while for open graded or soft-aspbinder mixtures, a low-pressure oxidation for 5 days at 85°C wselected.

The first attempt to relate the Superpave asphalt binder perties to laboratory mixture performance was reported in SHRA-398 ~SHRP 1994!. A thermal cracking validation was performed by relating asphalt-binder properties to TSRST test resconducted on HMA mixtures. A ranking of the asphalt bindeusing the BBR limiting stiffness temperature at 300 MPa adirect tension ultimate failure strain at226°C was found to com-pare favorably with the TSRST fracture temperature rankinSimilarly, linear regression relationships relatingm-value, stiff-ness, and ultimate strain at failure using both aged and unabinders to TSRST fracture temperature at short-and long-taging conditions were satisfactory for the most part. The exction to the rule was the ultimate failure strain at226°C, whichdid not correlate well with fracture temperature of both short-along-term aged mixtures~Jung and Vinson 1994b!.

Similar studies on the low-temperature performance of pomer modified binders were performed by King et al.~1993!. Intesting of several RTFOT-aged asphalt binders with a rangepolymer concentrations, relationships between the BBR limittemperature and the TSRST fracture temperature were reportbe excellent. When similar relationships were developed withDT, correlations were comparably worse. Therefore, if the TSRis truly representative of thermal cracking, then the BBR limititemperature is certainly a better single performance indicator tthe DT limiting failure strain temperature. Even if this is the cafindings in the study still indicate that the DT provides valuWhen the BBR limiting temperature data were used to develorelationship with DT limiting temperature data, the correlatioprovided some interesting trends. The polymers tended to redthe limiting-failure-strain temperature more than they reducedlimiting-stiffness temperature.

A preliminary evaluation of the relationships between speccation properties and performance at low temperatures wasformed during the SHRP research. Five test sites locatedAlaska, Pennsylvania, and Finland, along with research atFrost Effects Research Facility of the U.S. Army Cold RegioResearch and Engineering Laboratory~USACRREL!, were usedto validate the TSRST as an accelerated low temperature pemance test for HMA pavements~Kanerva et al. 1994!. The re-search concluded that cracking behavior at four of the fivesites and the USACCREL could be explained using the TSRfracture temperature. Furthermore, the research suggested t

NSPORTATION ENGINEERING / NOVEMBER/DECEMBER 2002 / 579

2.128:578-586.

Dow

nloa

ded

from

asc

elib

rary

.org

by

The

Uni

vers

ity o

f Q

ueen

slan

d L

ibra

ry o

n 05

/13/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

Table 1. Bending Beam Rheometer and Direct Tension Limiting Temperatures

Contract AC grade PG grade Polymer modification

Limiting Temperature

Direct tension required?BBR S(t) BBR m-value Direct tension

2480 AC-20P PG58-28 Yes 220.4 218.9 218.0 No2491 AC-20P PG52-16 Yes 215.5 216.2 214.2 No2501-B AC-30 PG64-16 No 218.0 211.7 217.0 No2530 AC-20P PG58-28 Yes 220.4 219.2 222.7 No2545 AC-20P PG58-22 Yes 214.0 217.3 213.5 No2552 AC-20P PG58-28 Yes 221.7 219.9 220.8 No2558-93 AC-20P PG58-22 Yes 214.2 215.4 214.5 No2558-94 AC-20P PG52-16 Yes 214.1 215.8 214.3 No2603 AC-201TLA PG70-16 No 211.8 27.5 212.4 No

2604 AC-30 PG70-22 No 216.3 214.8 214.8 No2611-A AC-20 PG64-16 No 216.9 212.5 213.4 No2611-B AC-20P PG64-28 Yes 222.7 223.8 223.5 No2617 AC-20P PG52-22 Yes 215.7 217.2 215.6 No2622 AC-30P PG70-22 Yes 213.2 212.4 219.6 No2704 AC-20P PG58-28 Yes 219.2 219.6 217.4 No2711 AC-20P PG58-22 Yes 216.8 218.5 217.7 Yes2825 Unknown PG70-28 Yes 221.2 222.5 220.4 No2838 Unknown PG64-34 Yes 228.3 228.1 230.8 No2880-A Unknown PG64-22 No 217.1 216.2 217.6 No2880-B AC-20P PG58-22 Yes 214.1 216.5 217.2 No

on-t o

ing

x-od

ureAl-

, thof

y td b

or-ingw-

u-ve-icaanhalorb

tesThgn

atehisderigh

n bederains

tingsev-Thetest

60hese,cter-un-arlysentangegh0P,,

ed inngteria,ted

on

theT isPatem-ired

not

xist

d to

is possible to develop a model using the TSRST, field aging cditions, and local temperature data to predict the developmencracking in all climates. This statement was confirmed followthe development of models by Raad et al.~1998! to predict thelow-temperature cracking of unmodified and modified HMA mitures in Alaska. Crack spacing of Alaskan pavements was meled as a function of age, predicted minimum air temperatTSRST fracture temperature, and TSRST fracture strength.though the study was site specific and based on limited datacorrelations were very promising, indicating the effectivenessthe TSRST as a low-temperature prediction device.

In 1993, SHRP researchers reported on an extensive studevaluate the low-temperature properties of mixtures measurethe indirect tensile creep test at low temperatures~ITLT ! withfield performance and binders properties~Lytton et al. 1993!. Thestudy concluded that the ITLT is suitable to predict field perfmance of asphalt mixtures subjected to low-temperature crackand theS and m parameters are suitable to measure the lotemperature rheological properties of asphalt binders.

Low-Temperature Properties of Asphalt Binders

It is widely recognized that asphalt binder is the primary contribtor to the resistance of low-temperature cracking in HMA paments. The level of resistance is largely controlled by the physproperties of the asphalt binder, which dictates the magnitudeextent of low-temperature cracking. At low temperatures, aspbinders behave in a brittle manner and lose their ability to absenergy through viscous flow.

Presently, the Superpave PBGS uses the BBR and the DTto evaluate the low-temperature properties of asphalt binders.two tests have been specified in the Superpave PBGS in recotion that the prefailure properties of the BBR may not correlwell with true fracture properties experienced in the field. Thas been shown to occur in many polymer-based asphalt binwhere the binder may exhibit high stiffness yet still poses a h


J. Transp. Eng. 200

f

-,

e

oy

,

ldt

tei-

s,

strain tolerance. In instances where this is the case, the DT caused as a supplemental test to the BBR if the asphalt binstiffness is between 300 and 600 MPa and the creep rate remgreater than 0.300.

The data presented in this paper were collected from a tesplan conducted to evaluate the low-temperature properties oferal Nevada asphalt binders using the BBR and DT tests.limiting temperatures of each binder based on the individualparameters was used as the criterion for evaluation.

In a 6-year span between 1993 and 1999, approximatelyasphalt binders were graded using the Superpave PBGS. Of tover 20 asphalt binders have complete low-temperature charaizations. The remaining projects lack the DT results due to fdamental problems experienced with the testing equipment ein the program. Nevertheless, the 20 asphalt binders repremany core projects constructed around the state, where the rof asphalt binders is typical of those used in Nevada. Althouthe majority of the asphalt binders were classified as AC-2AC-20, AC-20 with 25% Trinidad Lake Asphalt, AC-30, AC-30Pand some performance-graded binders were also representthe study. Following testing of the asphalt binders, the limititemperatures based on each of the three low-temperature criS(t), m-value, and failure strain was compiled and are presenin Table 1.

The limiting temperatures of the 20 asphalt binders basedthe BBR creep stiffnessS(t) and creep ratem-value were firstanalyzed to determine if the DT was required to completegrading. The current Superpave specification states that the Donly required for creep stiffness values between 300 and 600 Mand creep rate values greater than 0.300 at the selected testperature. Of the 20 asphalt binders analyzed, only one requtesting using the DT test. DT testing of the one binder didintroduce any change in binder grade.

The next area of interest was to determine if relationships ebetween the two low-temperature characterization tests~i.e., BBRand DT!. A series of linear regression analyses were performe

EMBER 2002

2.128:578-586.

Dow

nloa

ded

from

asc

elib

rary

.org

by

The

Uni

vers

ity o

f Q

ueen

slan

d L

ibra

ry o

n 05

/13/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

Fig. 1. Relationship between BBRS(t) and DT limiting temperatures

de-wer

anFigifi-torbeof

liersedree

)

nonande-la-thet thecat-g

n-e

sis-in-re

ofthe

determine if relations exist between the limiting temperaturestermined based on the three design criterion. Three modelsfit relating BBRS(t) to direct tension failure strain, BBRm-valueto direct tension failure strain, and BBRS(t) to BBR m-value.The analysis was performed using the SAS macro REGRESSthe associated macro call files mldumreg.sas and mlrind.sas.1–3 present the relationships with the effects of polymer modcation included in the analysis through the use of an indicavariable. The significance of the relationships were found tostrong,R250.82 to 0.86. The analysis of variance assumptionsnormally distributed error, equal variance, and absence of outand/or influential observations were all checked and satisfiEquations relating the limiting temperatures based on the thdesign criterion are presented as follows:

DTT523.33310.813 BBR-S~ t !

for neat and polymer modified AC (1

DTT525.98410.703 BBR-m-value

for polymer modified AC (2a)

DTT522.35910.703 BBR-m-value

for neat asphalt cement (2b)

JOURNAL OF TRA

J. Transp. Eng. 200

e

ds.

.

BBR S~ t !51.80611.061 BBR-m-value

for polymer modified AC (3a)

BBR S~ t !529.23010.541 BBR-m-value

for neat asphalt cement (3b)

It is worth noting from Eq.~1! that the relationship betweeDT limiting temperature and BBR limiting temperature basedstiffness is represented by a single equation for both neatpolymer-modified binders. While performing the analysis, it bcame evident that the effect of polymer modification on the retionship was insignificant, and the term was removed frommodel. A scatter plot of the data presented in Fig. 1 shows thadata are randomly scattered close to the line of equality, indiing that the BBRS(t) and DT limiting temperatures are providinsimilar results regardless of binder composition.

In Eq. ~2!, the effects of polymer modification on the relatioship between BBRm-value and DT limiting temperatures can bnoticed through a 3.5°C improvement in low-temperature retance over the nonpolymerized materials. Additionally, Fig. 2dicates that the BBRm-value appears to be consistently moconservative than the DT at assigning the limiting temperaturenonmodified binders. A similar trend was experienced when


2.128:578-586.

Dow

nloa

ded

from

asc

elib

rary

.org

by

The

Uni

vers

ity o

f Q

ueen

slan

d L

ibra

ry o

n 05

/13/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

Fig. 2. Relationship between BBRm-value and DT limiting temperatures

it-halteatgai

uretate

limods.thetheer-for

thengix-

g-

n-

-

n-

forucedereare

theonsed-ms,ithhavenseaten ofingn-ind-ition

BBR m-value was compared with the BBR creep stiffness liming temperatures. From Fig. 3, the polymer-modified aspbinders plotted close to the line of equality, while the five nasphalt binders remained above the equality line. This once ashows that the criteria for the BBRm-value is most conservativeat assigning limiting temperatures. Additional low-temperatdata from neat asphalt binders are needed to validate this sment.

A statistical analysis was conducted to assess whether theiting temperatures determined by the three different meth~BBR S(t), BBR m-value, and DT! are statistically the sameUsing a significance level of 0.05, the analysis concluded thatlimiting temperatures determined by the three criteria are allsame. Since the majority of the projects included polymmodified binders, this conclusion should be considered validpolymer-modified binders only.

Resistance of HMA to Low-Temperature Cracking

The objective of this part of the research was to evaluatelow-temperature properties of HMA mixtures at various agiconditions. In order to achieve this objective, the following mtures were tested in the TSRST:1. LMLC PAV—lab mixed lab compacted–pressure-agin

vessel binder;


J. Transp. Eng. 200

n

-

-

2. LMLC STOA—lab mixed lab compacted–short-term oveaged mix;

3. LMLC LTOA—lab mixed lab compacted–long-term ovenaged mix;

4. FMLC—field mixed lab compacted~represents short-termaged condition!; and

5. FMLC LTOA—field mixed lab compacted–long-term oveaged mix.

It is well recognized that laboratory produced mixtures,some reason or another, often do not represent the field prodmixture. For this reason, lab and field produced mixtures wincluded in the study to see if low-temperature propertiessensitive to a mixture’s origin.

A total of 24 projects constructed throughout Nevada overpast 6 years were tested in the TSRST to determine their respto low-temperature cracking. Due to material availability, amenments to the testing plan, and unforeseeable testing probleonly a partial factorial is available for analysis. Nevertheless, wall the obstacles encountered, 13 of the 28 contracts testedcompleted matrices. All contracts consisted of a similar degraded asphalt mixture with a nominal maximum size aggregof 3/4 in. and percent passing #200 between 3 and 7%. Sixteethe 24 projects contained an AC-20P binder, while the remainprojects consisted of AC-30, AC-30P, AC-20, AC-20 plus Triidad Lake Asphalt, and several performance-graded asphalt bers. Whenever possible, three replicates of each mixture cond

EMBER 2002

2.128:578-586.

Dow

nloa

ded

from

asc

elib

rary

.org

by

The

Uni

vers

ity o

f Q

ueen

slan

d L

ibra

ry o

n 05

/13/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

Fig. 3. Relationship between BBRm-value and BBRS(t) temperatures

cturd tobleed

ting

exuresingableturamthethe

esitre

nt.Paas

ndre-m-tractsblestureLCedem-term

nalincts

were fabricated and tested. Fracture temperature and frastress were recorded directly from the TSRST and later usetest several hypotheses regarding the mixtures conditions. Tasummarizes the average values of the TSRST data generatthis research effort.

Impact of Mixture’s Condition on Low-TemperatureProperties

The objective of this part of the research was met by conducstatistical analyses to test the various hypotheses.

Ho: TSRST Response Variable for LMLC STOA and FMLCAre the SameThis hypothesis was established to determine if a differenceisted in the low temperature response of short-term aged mixtproduced in the laboratory and those produced in the field. Ufracture temperature and fracture stress as the response vari18 contracts were investigated. The hypothesis that the fractemperature of the two short-term aging conditions are the scould not be rejected. As anticipated, we can conclude thatfracture temperature of short-term aged mixtures produced inlab and field are statistically the same.

With fracture stress as the response variable, the hypothwas retested and rejected. We can conclude that the fracture s

JOURNAL OF TRA

J. Transp. Eng. 200

e

2in

-s

s,ee

sss

for the different short-term aged mixtures is statistically differeOn the average, the field produced mix exhibited a 0.33 Mlower fracture stress under the short-term aging conditionscompared with the lab produced mixtures.

Ho: TSRST Response Variable for LMLC STOA and LMLCLTOA Are the SameThis hypothesis was investigated to determine if the long- ashort-term aging conditions of laboratory produced mixturessulted in different low-temperature properties. Using fracture teperature and fracture stress as the response variables, 16 conwere investigated. The hypotheses for both response variawere rejected, concluding that fracture temperature and fracstress are not the same for short- and long-term aged LMmixtures. In the study, the long-term aging of the lab producmixtures experienced on average a 5.5°C warmer fracture tperature and a 0.32 MPa lower fracture stress than the short-aging.

Ho: TSRST Response Variable for FMLC and FMLC LTOAAre the SameA similar hypothesis was developed to determine if the additiolong-term oven aging of the field produced mixtures resulteddifferent low-temperature properties. Once again, 16 contra


2.128:578-586.

Dow

nloa

ded

from

asc

elib

rary

.org

by

The

Uni

vers

ity o

f Q

ueen

slan

d L

ibra

ry o

n 05

/13/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

Table 2. Low-Temperature Properties of HMA Mixtures Measured by TSRST

Contract

LMLC STOA LMLC LTOA FMLC FMLC LTOA LMLC PAV

Fracturestress~MPa!

Fracturetemperature

~°C!

Fracturestress~MPa!

Fracturetemperature

~°C!

Fracturestress~MPa!

Fracturetemperature

~°C!

Fracturestress~MPa!

Fracturetemperature

~°C!

Fracturestress~MPa!

Fracturetemperature

~°C!

2480 2.32 238.4 0.80 229.12491 1.79 225.1 2.07 217.6 0.92 225.62501-A 1.38 231.9 0.52 218.9 0.73 213.5 0.73 226.32501-B 2.42 234.8 0.8 236.62530 1.49 238.9 1.07 230.1 1.56 233.2 0.88 231.9 0.34 224.32545 1.91 233.3 2.11 226.4 1.91 228.0 2.08 226.4 0.73 227.02552 2.74 234.6 1.22 238.7 0.86 232.2 0.99 233.22558-93 1.79 220.9 2.64 224.8 1.11 230.32558-94 2.47 232.3 2.07 225.2 1.89 233.4 0.81 233.02603 0.30 215.0 0.12 217.6 0.53 213.1 0.70 213.3 0.20 216.32604 1.05 226.2 1.88 215.1 0.59 223.0 1.51 219.3 0.76 224.32611-A 0.80 224.5 0.47 217.52611-B 0.79 232.5 1.07 226.4 1.06 234.52617 1.26 230.9 3.17 228.2 1.12 230.52622 0.75 234.1 1.46 228.7 0.98 229.8 0.73 231.42704 2.54 235.1 2.34 228.0 2.38 236.7 3.01 231.7 2.56 230.02711 3.84 234.5 2.59 228.0 2.32 235.0 1.96 227.8 2.47 230.02742 0.80 236.6 0.76 234.1 1.36 238.8 1.80 240.4 1.58 235.62825 2.40 232.2 2.24 226.4 2.81 235.0 2.76 232.6 2.14 228.22838 4.70 244.6 3.80 237.6 2.82 244.7 2.50 240.5 2.24 238.52880-A 2.36 229.7 1.87 224.2 1.94 231.5 2.12 226.4 2.11 224.62880-B 2.23 229.8 2.06 228.0 2.35 229.1 2.30 225.9 2.85 229.02880-C 2.32 229.7 1.77 223.9 1.81 230.8 1.89 226.6 1.34 223.22880-D 2.45 231.6 2.37 227.9 1.96 230.2 2.02 226.6 1.84 227.5

reswa

ingeldmixtha

ce itedthiim-

enonsureesewape-Pa

.

f allturon

tureamever,

areex-y the

tureithC

foronsison-

actketheed

pari-dg-andix-the

d tofur-re a

were investigated, with fracture temperature and fracture stused as the criteria for testing the hypotheses. The hypothesisrejected, concluding that the addition of long-term oven agproduced a significant difference in fracture temperature of fiproduced mixtures. As anticipated, the long-term oven-agedon average experienced a 3.5°C warmer fracture temperaturethe short-term aged mix~FMLC!.

Using fracture stress as the response variable, the differenshort- and longterm aging of field produced HMA was retesand could not be rejected. It can be concluded, based onstudy, that aging of field produced mixtures has no statisticalpact on fracture stress.

Ho: TSRST Response Variable for LMLC STOA and LMLCPAV Are SameThis hypothesis was initiated to determine if the TSRST was ssitive enough to distinguish changes in asphalt binder conditiTwenty-four projects were investigated, with fracture temperatand fracture stress used as the criteria for testing the hypothThe difference in fracture temperature and fracture stressfound to be significant. On the average, the PAV condition exrienced a 2.5°C warmer fracture temperature and a 0.62 Mlower fracture stress than the short-term oven-aged condition

Ho: TSRST Response Variable for LMLC LTOA, FMLCLTOA, and LMLC PAV Are the SameThis hypothesis was developed out of interest in determining ilong-term aging processes have the same effect. Using fractemperature and fracture stress as the response variables


J. Transp. Eng. 200

ss

n

n

s

-.

s.s

ece

again, 14 contracts were investigated. The hypothesis that fractemperature of the three long-term aging systems were the scould not be rejected. The difference in fracture stress, howewas significant.

Although the findings of the five hypotheses as a wholeconsistent with engineering judgment, caution should still beercised due to some serious discrepancies in data generated bTSRST. On 10 of the 27 projects investigated, the TSRST fractemperature for the lab mixed lab compacted mixtures made wPAV-aged binder was very close to if not colder than the LMLor FMLC short-term aged mixtures. A possible explanationthis is that PAV may not adequately simulate the aging conditicritical to low-temperature cracking. In response to this, itworth noting that all the discrepancies occurred on contracts csisting of polymer-modified asphalt binders except for contr2603, which contained a neat binder with 25% Trinidad LaAsphalt. This may indicate that the PAV-aging system and/orTSRST has some limitations when used with polymer-modifiasphalt binders.

The recommendations of the hypotheses testing and comson of these recommendations with pavement engineering jument showed that the fracture temperature is a more stablereliable measure of the low-temperature properties of HMA mtures than the fracture stress. Combining these findings withfact that the fracture temperature can be more directly appliea mixture’s evaluation, while the fracture stress data requirether interpretation, makes the use of the fracture temperatumore attractive approach.

EMBER 2002

2.128:578-586.

Dow

nloa

ded

from

asc

elib

rary

.org

by

The

Uni

vers

ity o

f Q

ueen

slan

d L

ibra

ry o

n 05

/13/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

Table 3. Comparison of Binder Limiting Temperatures and HMA Mixtures Fracture Temperatures

ContractBinder limiting

temperature~°C!LMLC STOA fracture

temperature~°C!LMLC LTOA fracture

temperature~°C!FMLC fracture

temperature~°C!FMLC LTOA fracture

temperature~°C!LMLC PAV fracture

temperature~°C!

2480 218.0 238.4 229.12491 214.2 225.1 217.6 225.62530 219.2 238.9 230.1 233.2 231.9 224.32545 213.5 233.3 226.4 228.0 226.4 227.02552 219.9 234.6 238.7 232.2 233.22558-93 214.2 220.9 224.8 230.32558-94 214.1 232.3 225.2 233.4 233.02611-B 222.7 232.5 226.4 234.52617 215.6 230.9 228.2 230.52622 212.4 234.1 228.7 229.8 231.42704 217.4 235.1 228.0 236.7 231.7 230.02711 216.8 234.5 228.0 235.0 227.8 230.02825 220.4 232.2 226.4 235.0 232.6 228.22838 228.1 244.6 237.6 244.7 240.5 238.52880-B 214.1 229.8 228.0 229.1 225.9 229.0

thets omixthipeon-ndedon

thesion10°tann-eraw-

ela-

intingitingtemdeloodsgehalateinddid

innd

there-theis

n,sur.

wto orfivetheavew-. For

turesalt

indernce.ers;tate-

sts. Theilure

p-

Relationship between Asphalt Binder Parametersand Performance in TSRST

The objective of this part of the research was to comparelow-temperature characteristics obtained from measuremenbinders properties and those obtained from measurements ontures. Using data presented in the first and second parts ofpaper, single variable linear regression models were develorelating TSRST fracture temperature at the various mixture cditions to limiting temperatures determined from the BBR aDT. Since the majority of the projects included polymer-modifibinders~15 out of 20!, the regression analyses were conductedthe polymer-modified mixtures only. It should be noted thatlimiting temperatures of asphalt binders used in the regresanalysis are based on actual test temperatures, which arewarmer than the anticipated failure temperature. This consshift in the temperature will not affect the forms of the relatioships; however, a 10°C should be subtracted from the temptures generated from the models in order to obtain the lotemperature grade of the binder.

Without extensive discussion of each of the models, the rtionships ranged from fair to extremely good;R2 ranged from 53to 90. To reword this, between 53 and 90% of the variabilityTSRST fracture temperature can be accounted for in the limitemperature of the asphalt binder. Table 3 compares the limtemperatures based on binder testing and the TSRST fractureperatures of the mixtures, and Table 4 summarizes the modeveloped based on these data. All of the correlations with glevels ofR2 values~above 70%! were obtained when the binderproperties are correlated to the properties of the long-term amixtures. The BBR and DT properties are measured on aspbinders that are subjected to long-term aging. Attempting to rethe limiting temperatures determined on the long-term aged bers with the fracture temperatures of unaged HMA mixturesnot result in good correlations, due to the significant differencethe conditions of the binder. However, correlating the BBR aDT limiting temperatures with the fracture temperatures oflong-term aged HMA mixtures resulted in good–excellent corlations. This emphasizes what is already a known fact—thatresistance of the HMA mixture to low-temperature crackingmainly controlled by the conditions of the binder. In additiosuch consistency gives the TSRST a great credibility in meaing the low-temperature cracking resistance of HMA mixtures

JOURNAL OF TRA

J. Transp. Eng. 200

f-

sd

Ct

-

-s

dlt

-

-

For the majority of the contracts evaluated, the critical lotemperature assigned by the Superpave PBGS was equalcolder than the fracture temperature experienced under themixture conditions. This trend appears more consistent withpolymer-modified asphalt binders, indicating that the Superpbinder specification is conservative enough that the lotemperature binder grade can be used to ensure performancethe neat asphalt binders, however, TSRST fracture temperawere consistently warmer than the polymer-modified asphbinders, and the limiting temperatures based on Superpave bspecification appeared nonconservative in almost every instaThe findings are based on only a small numbers of neat bindtherefore, a larger sample base is needed to confirm this sment.

Conclusions and Recommendations

An investigation of the Superpave low-temperature binder tesuggests redundancies exist between the BBR and DT teststhree test parameters of creep stiffness, creep rate, and fa

Table 4. Summary of Regression Models Relating Fracture Temerature of HMA Mixture with Limiting Temperature of Binder

Mixture TSRSTfracture temperature,Y(C)

Binder limitingtemperature,

X(C) Relationship

R2

value~%!

LMLC STOA DT Y5216.79510.875X 65LMLC LTOA DT Y5215.02710.712X 85FMLC DT Y5215.74910.850X 55FMLC LTOA DT Y5213.35710.873X 88LMLC PAV DT Y5221.80810.448X 69LMLC STOA BBR S(t) Y5222.15610.636X 71LMLC LTOA BBR S(t) Y5216.33210.682X 78FMLC BBR S(t) Y5219.58110.683X 60FMLC LTOA BBR S(t) Y5215.45610.810X 90LMLC PAV BBR S(t) Y5220.40810.515X 64LMLC STOA BBR m-value Y5212.83611.085X 53

LMLC LTOA BBR m-value Y5214.60710.745X 74

FMLC BBR m-value Y5211.41011.072X 51

FMLC LTOA BBR m-value Y5215.84010.784X 71

LMLC PAV BBR m-value Y5220.49310.552X 59


2.128:578-586.

ow-sis

rran

therwase o

reng-coms toandA

xpeof

RSn ier-

ntee oea-ith

thin and-

onnggeer-

l

ash-

ix-a-

,

ena-

P.the

ndeving

,

,

alt-

rch

Dow

nloa

ded

from

asc

elib

rary

.org

by

The

Uni

vers

ity o

f Q

ueen

slan

d L

ibra

ry o

n 05

/13/

13. C

opyr

ight

ASC

E. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved.

strain were determined equally capable of establishing the ltemperature grade of polymer-modified asphalt binders. Incontencies experienced with neat asphalt binders, however, wathe inclusion of the BBR creep rate~m-value! parameter, after itappeared sensitive to binder composition. Modifications toperformance-graded binder specification are presently undeto help address these issues and hopefully provide better uthe DT test.

The continual use of the TSRST for the low-temperatuevaluation of lab and field produced mixtures subjected to loterm oven aging is suggested. The fracture temperature is remended for use in determining the resistance of HMA mixturelow-temperature cracking, since it provides a direct measureit is more stable than the fracture stress. The LTOA of the HMmixture appears to provide a better representation of aging erience in the field by accounting for the significant effectsasphalt, aggregate, and aging interactions. Although the TSshowed signs of inconsistency during the research, concerminor because the discrepancies were limited to polymmodified binders under PAV-aged conditions. The data presein this paper showed that determining the fracture temperaturthe long-term aged HMA mixtures is more appropriate than msuring the fracture temperature of HMA mixtures produced wPAV-aged binders.

The data generated in this study showed that the use oflimiting temperature measured on the asphalt binder resultsconservative low-temperature grade for polymer-modified biers, while it is unconservative for neat asphalt binders. Basedthis finding, it is recommended that both the binder’s limititemperature and the fracture temperature of the long-term aHMA mixture be determined to ensure good long-term field pformance.

References

Bahia, H. U., and Anderson, D. A.~1995!. ‘‘The pressure aging vesse~PAV!: a test to simulate reological changes due to field aging.’’Physi-


J. Transp. Eng. 200

-t

yf

-

-

Ts

df

e

d

cal properties of asphalt cement binders, J. C. Hardin, ed., ASTM,West Conshohocken, Pa., 67–88.

Bell, C. A. ~1994!. ‘‘Aging: binder validation.’’ Rep. SHRP-A-384, Stra-tegic Highway Research Program, National Research Council, Wington, D.C.

Bell, C. A., AbWahab, Y., Cristi, M. E., and Sosnovske, D.~1994!. ‘‘Se-lection of laboratory aging procedures for asphalt-aggregate mtures.’’ Rep. SHRP-A-383, Strategic Highway Research Program, Ntional Research Council, Washington, D.C.

Jung, D. H., and Vinson, T. S.~1994a!. ‘‘Low-temperature cracking: testselection.’’Rep. SHRP-A-400, Strategic Highway Research ProgramNational Research Council, Washington, D.C.

Jung, D. H., and Vinson, T. S.~1994b!. ‘‘Low-temperature cracking:binder validation.’’Rep. SHRP-A-399, Strategic Highway ResearchProgram, National Research Council, Washington, D.C.

Kanerva, H. K., Vinson, T. S., and Zeng, H.~1994!. ‘‘Low-temperaturecracking: field validation of the thermal stress restrained specimtest.’’ Rep. SHRP-A-401, Strategic Highway Research Program, Ntional Research Council, Washington, D.C.

King, G. N., King, H. W., Harders, O., Arand, W., and Planche, P.~1993!. ‘‘Influence of asphalt grade and polymer concentration onlow temperature performance of polymer modified asphalt.’’J. Asso-ciation of Asphalt Paving Technologists, 62, 1–22.

Lytton, R. L., Uzan, J., Fernando, E. G., Roque, R., Hiltunen, D., aStoffels, S. M.~1993!. ‘‘Development and validation of performancprediction models and specifications for asphalt binders and pamixes.’’ Rep. SHRP-A-357, Strategic Highway Research ProgramNational Research Council, Washington, D.C.

Raad, L., Saboundjian, S., Sebaaly, P., and Epps, J.~1998!. ‘‘Thermalcracking models for AC and modified AC mixes in Alaska.’’Trans-portation Research Record 1643, Transportation Research BoardWashington, D.C., 117–126.

Strategic Highway Research Program~SHRP!. ~1994!. ‘‘Stage 1 valida-tion of the relationship between asphalt properties and asphaggregate mix performance.’’Rep. SHRP-A-398, National ResearchCouncil, Washington, D.C.

Vinson, T. S., Janoo, V. C., and Haas, R. C. G.~1989!. ‘‘Summary reporton low temperature and thermal fatigue cracking.’’Rep. SHRP-A/IR-90-001, Strategic Highway Research Program, National ReseaCouncil, Washington, D.C.

EMBER 2002

2.128:578-586.

Documents

Evaluation of Low-Temperature Properties of HMA Mixtures