Pressure Aging Vessel and Low-Temperature Properties of Asphalt Binders

MOTIVATION

A comprehensive investigation of the low-temperature crackingmechanism in asphalt pavements was performed as part of a nationalpooled-fund study (6). The analysis focused on comparing theparameters obtained from different test procedures and identifyingcorrelations between asphalt binder and mixture parameters andfield performance.

Field samples were cored or sawed from Minnesota’s cold-weatherroad research facility (MnROAD) and highways in Minnesota,Wisconsin, and Illinois. Performance information was readily avail-able for these samples (Table 1). They were evaluated using currentstandard methods, such as indirect tensile test (IDT) creep stiffnessand strength for asphalt mixtures; bending beam rheometer (BBR)and direct tension test (DTT) for asphalt binders; and newly devel-oped testing protocols, such as the semicircular bending (SCB) test,disc-shaped compact tension (DCT) test, the single-edge notched-beam (SEB) test for asphalt mixtures, and the double-edge notchedtension (DENT) test for asphalt binders.

Field cores were cut into SCB, IDT, and DCT mixture specimens,and field beams were used for SEB testing. The asphalt binders usedfor BBR, DTT, and DENT testing were extracted from the mixturespecimens tested in IDT and SCB according to AASHTO T164 asmodified by the Minnesota Department of Transportation (DOT).The modification consists in the use of toluene as a solvent to pre-vent any interaction with the polymer present in PG 58-34 and PG58-40 binders, as suggested in SHRP A-370 (7 ). The recovery tech-nique consisted in washing and filtering the loosened asphalt mixturewith the mentioned solvent. Each filtrate was then distilled under vac-uum in a rotary evaporator, and the binder was recovered and thencentrifuged to remove fine aggregates. Finally, the decanted solutionwas distilled along with nitrogen gas to remove any remaining traceof the solvent.

Test results were correlated with total length of transversecracking, which was assumed to represent a good estimator of low-temperature cracking performance. Comparisons were made at tem-peratures representative for each site’s local climate conditions.Low pavement temperatures (50% reliability) were obtained foreach site from the Long-Term Pavement Performance (LTPP) data-base and LTPP BIND software. Since laboratory parameters werenot obtained at those calculated temperatures, linear interpolationwas used to estimate the appropriate values.

Table 2 summarizes the correlation coefficients between labora-tory parameters and transverse cracking. With a 10% significancelevel and Spearman’s correlation coefficient, only fracture parameters

Pressure Aging Vessel and Low-Temperature Properties of Asphalt Binders

Eyoab Zegeye Teshale, Ki-Hoon Moon, Mugurel Turos, and Mihai Marasteanu

117

The oxidative aging during the service life in asphalt binders used inconstruction of asphalt pavements significantly affects the performanceof these pavements. A study investigating the effect of the pressureaging vessel (PAV) laboratory aging procedure on low-temperatureproperties of asphalt binders is presented. Bending beam rheometer creeptests and direct tension fracture tests were performed on laboratory-aged binders as well as extracted binders. Significant differencesexisted between extracted binder and PAV binder behavior at low tem-peratures. PAV aging was not always detrimental for asphalt binderfracture properties compared with the aging from rolling thin-filmoven testing.

One of the factors that significantly affects the performance ofasphalt materials during their service life is oxidative aging ofasphalt binders used in the construction of pavements. Effortsfrom SHRP have resulted in two proposed methods for simulatingaging during mixing and compaction and during service life:rolling thin-film oven (RTFO) testing (AASHTO T240-06) andpressure aging vessel (PAV) testing (AASHTO R28-06), respec-tively. In spite of considerable research performed in the past yearsin understanding the evolution of the aging process and in simu-lating field aging in laboratory conditions, many issues are still notwell understood (1–5), and the ability of laboratory aging methodsto simulate field aging continues to generate strong debate in theresearch community.

In this paper, the effect of the PAV aging procedure on low-temperature creep stiffness and fracture properties of asphalt bindersis investigated. This was prompted by the fact that most researchstudies on oxidative aging have focused on high-temperaturerheological properties. The low-temperature properties, which arethe most negatively affected by oxidative aging, have receivedmuch less attention. This trend is surprising, since increased pro-portions of reclaimed asphalt pavement materials are used inasphalt pavement construction.

Department of Civil Engineering, University of Minnesota, Twin Cities, 500 PillsburyDrive, SE, Minneapolis, MN 55455. Corresponding author: M. Marasteanu,[email protected].

Transportation Research Record: Journal of the Transportation Research Board,No. 2207, Transportation Research Board of the National Academies, Washington,D.C., 2011, pp. 117–124.DOI: 10.3141/2207-15

are significant. The fracture parameters with the highest correlationwith field performance are mixture fracture energy and toughness andextracted binder direct tension strain at failure. These results stronglysupport the critical need for an accurate evaluation of fracture proper-ties of aged binders in asphalt mixtures and raise the question of howwell laboratory aging procedures simulate field aging.

RESEARCH APPROACH

In order to investigate how well laboratory aging procedures simu-late field aging relative to asphalt binder fracture properties, whichhad the best correlation to field performance, the following studywas performed. Cores from MnROAD Cells 33, 34, and 35 weretaken approximately 5 years after construction. The Superpave® mixdesign was identical but three binders were used: PG 58-28 for Cell33, PG 58-34 for Cell 34, and PG 58-40 for Cell 35. To investigatethe effect of aging with depth, the cores were cut into three 25-mm

118 Transportation Research Record 2207

slices from top to bottom, and the remaining part was discarded. Theslices were marked with T (top), M (middle), and B (bottom),respectively. After the mixture testing was performed, the asphaltbinders were extracted from the cores according to the MinnesotaDOT–modified AASHTO T164 method. Several cores were com-bined for each mixture to extract enough binder for binder testing.Nine binders were recovered from the sliced cores (Table 3).

The properties of the extracted binders were compared to the orig-inal binders used to construct these cells. The original binders wereaged using RTFO and then PAV methods. Dynamic shear rheometer(DSR) (AASHTO T315-06) frequency sweeps from 1 to 100 rad/swere performed from 4°C to 76°C, and master curves of the absolutevalue of complex modulus and phase angle were generated using theChristensen–Anderson–Marasteanu model (8). BBR testing was per-formed at two temperatures, and creep stiffness curves were obtained(AASHTO T313-06). DTT was performed at the same two temper-atures, and stress–strain curves were obtained (AASHTO T314-07).The investigation also included DSR and BBR testing to fully char-acterize the aging effects on binder properties over a wider range oftemperatures.

Visual inspection of �G*� master curves in Figure 1 indicates thatPAV binders match extracted binders reasonably well, and the dif-ferences among the three layers of the core were relatively small.The finding that PAV aging matches field aging reasonably wellconfirms the results obtained in the original work that led to thedevelopment of PAV aging based on �G*� data.

Analysis of phase angle master curves reveals, however, significantdifferences between field- and laboratory-aged binders (Figure 2). Forthe two modified binders in Cells 34 and 35, the plateau observed inthe PAV condition disappears almost completely in the extractedbinders, which appears to indicate that either field aging or the extrac-tion process, in spite of the measures taken, has affected the polymerphase in these binders.

Analysis of BBR creep stiffness curves indicates that for all threecells, the extracted binders are always stiffer than the PAV-agedbinders. There is also a clear reduction in the slope of the creepstiffness curves (lower m-values) for the extracted binders com-pared to the PAV-aged binders, an indication of increased relax-ation times at low temperatures. These results appear to indicatethat the extracted binders have aged more than the PAV binders.This result is not entirely surprising since low-temperature test-ing was not used in the original development of the PAV agingprocedure. Examples are shown in Figure 3.

The DTT stress–strain curves for the extracted binders (top layer)and the PAV-aged binders are shown in Figure 4. For all three cells,

TABLE 2 Correlation Coefficients BetweenLaboratory Parameters and Field Data

Correlation Coefficients

Laboratory Parameter Pearson Spearman

Mixture Parameters

SCB, fracture energy −0.708 −0.718*

IDT, at 60 s −0.713 −0.405

IDT, at 500 s −0.590 −0.071

SCB, fracture toughness −0.639 −0.736*

IDT, strength −0.325 −0.571*

DCT, fracture energy −0.265 −0.500*

SEB, energy −0.291 −0.500*

Binder Parameters

BBR, S at 60 s 0.105 0.248

BBR m-value, m at 60 s −0.252 0.152

DTT strain at failure −0.694 −0.673*

DENT stress at failure −0.045 0.217

DENT strain at failure −0.239 −0.250

NOTE: S = creep stiffness.*Significant at 10% level.

TABLE 1 Field Sections Location and Binder

Identification State Asphalt Binder

IL I74 Illinois AC-20

MN75 2 Minnesota PG 58-28

MN75 4 Minnesota PG 58-34

MnROAD 03 Minnesota PG 58-28





US20 6 Illinois AC-10

US20 7 Illinois AC-20

WI STH 73 Wisconsin PG 58-28

TABLE 3 Recovered Asphalt Binders

Sample Original Binder BinderName Cell Grade Location in Core Content (%)

33T 33 PG 58-28 Top layer 5.78

33M 33 PG 58-28 Middle layer 5.52

33B 33 PG 58-28 Bottom layer 5.65

34T 34 PG 58-34 Top layer 5.76



35T 35 PG 58-40 Top layer 4.84



the extracted binders are not only stiffer but also more brittle thanthe PAV binders, with much lower fracture strains.

This situation suggests that, similar to the BBR results, theextracted binders have aged more than the PAV binders. The dif-ferences are particularly significant for the PG 58-40 binder; theextracted stress–strain curve suggests that it is less crack resistantthan the PG 58-34. This observation is in agreement with fieldperformance data from MnROAD, which showed that Cell 35 had significantly more cracks than Cell 34, in which a PG-34binder was used. This appears to indicate that the PG lower-temperature limit, based on BBR creep stiffness data, may notalways result in the best material selection for low-temperaturecracking performance.

These results prompted a revisit of the SHRP A-367 report (9) inwhich the PAV procedure was proposed. The authors acknowledgethat low-temperature creep stiffness and fracture data (BBR andDTT, respectively) were not used in the validation process becauseof a lack of resources:

A detailed experiment was developed to validate that the chemistry andrheology of the residue from the PAV test replicate the chemistry andrheology of long-term field exposure (see figure 2.25). Unfortunately,resources needed to complete the experiment were directed to othertasks, and much of the data specified in figure 2.25 were not obtained.The results that were obtained do verify that similar rheological behav-ior is obtained in the laboratory and field, as illustrated by figure 2.26.Values of G* and size exclusion chromatography (SEC) fraction I thatwere obtained from several original laboratory-aged and field-agedbinders are compared in figures 2.27 and 2.28. Overall, the results val-idate the hypothesis that the PAV test successfully mimics field aging.(9, p. 38)

Teshale, Moon, Turos, and Marasteanu 119

RTFO–PAV COMPARISON

The authors performed an additional experiment to better understandthe effect of PAV aging compared to RTFO aging. BBR and DTTbinder data obtained as part of the same pooled-fund study (6) wereused in the analysis. Table 4 summarizes the set of binders used.

A similar approach was used by Migliori and Corte, who comparedprolonged PAV testing to the coupled aging procedure of RTFO andPAV (10). The authors compared four unmodified asphalt binders foreach aging condition by means of conventional consistency test (pen-etration at 25°C, ring-and-ball softening temperature), BBR, DSR,and asphalt content test. For the low-temperature properties, thebinders were BBR-tested at four temperatures ranging from −10°C to−18°C. The experimental results indicated very high similarities forstiffness modulus at 60 s and m-values at 60 s obtained from RTFOand 5 h of PAV, respectively, and between RTFO + 20 h of PAV and25 h of PAV, respectively.

Analysis of BBR data indicates, as expected, that PAV bindersare stiffer and have longer relaxation times (lower m-values) thanRTFO binders. This trend was observed for all binders and all testtemperatures. As temperatures approached the PG lower-limit tem-perature (shorter times), the difference between RTFO and PAVdiminishes considerably, which appears to suggest the existence ofa common asymptotic value. Examples are shown in Figure 5.

Analysis of the DTT stress–strain curves confirms the trendsobserved in the BBR data as temperature approaches the PG limit.For all binders tested, at the lowest test temperature with a strain rateof 3% per min (Figures 6b through 10b), very small differenceswere observed between the RTFO and PAV conditions, and for

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

1.E+07

1.E-04 1.E-02 1.E+00 1.E+02 1.E+04

|G*|

, Pa

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

1.E+07

|G*|

, Pa

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

1.E+07

|G*|

, Pa

Frequency, rad/s

1.E-04 1.E-02 1.E+00 1.E+02 1.E+04

Frequency, rad/s

1.E-04 1.E-02 1.E+00 1.E+02 1.E+04

Frequency, rad/s

(a)

(c)

(b)

PAV33T33M33B

PAV

34T

34M

34B

PAV35T35M35B

FIGURE 1 �G*� master curves for Cells (a) 33, (b) 34, and (c) 35 (Tref � 34�C).


40

50

60

70

80

90

Ph

ase

An

gle,

deg

rees

40

50

60

70

80

90

Ph

ase

An

gle,

deg

rees

40

50

60

70

80

90

Ph

ase

An

gle,

deg

rees

PAV

33T

33M

33B

PAV

34T

34M

34B

PAV

35T

35M

35B

1.E-04 1.E-02 1.E+00 1.E+02 1.E+04

Frequency, rad/s

1.E-04 1.E-02 1.E+00 1.E+02 1.E+04

Frequency, rad/s

1.E-04 1.E-02 1.E+00 1.E+02 1.E+04

Frequency, rad/s(a)

(c)

(b)

FIGURE 2 Phase angle master curves for Cells (a) 33, (b) 34, and (c) 35 (Tref � 34�C).

10

100

1000

0.1 101 100 1000Time [sec]

Stif

fnes

s, M

Pa

10

100

1000

Stif

fnes

s, M

Pa

33B33T33MPAV

0.1 101 100 1000Time [sec]

(a) (b)

34B34M34TPAV

FIGURE 3 BBR creep stiffness comparison for Cells (a) 33 at �18�C and (b) 34 at �24�C.


0

2

4

6

8

0 2 4 6 8 10 12

Str

ess

[MP

a]

0

2

4

6

8

Str

ess

[MP

a]0

2

4

6

8

Str

ess

[MP

a]

Strain [ % ]

Extracted

PAV

0 2 4 6 8 10 12

Strain [ % ]

Extracted

PAV

0 2 4 6 8 10 12

Strain [ % ]

(a) (b)

(c)

Extracted

PAV

FIGURE 4 DTT stress–strain comparison for Cells (a) 33 at �18�C, (b) 34 at �30�C, and (c) 35 at �30�C (strainrate � 3%/min).

TABLE 4 Set of Binders Used for RTFO Versus PAV Comparison

Binder Grade Binder Identification Modifier or Plain

PG 58-40 SBS, Flint Hills Res. Modifier 1

PG 58-34 Elvaloy, Murphy Modifier 1

PG 58-34 SBS, Flint Hills Modifier 2

PG 58-28 Neat, Seneca Petroleum Plain 1

PG 58-28 Neat, Payne & Dolan Plain 2

PG 64-34 Elvaloy, Murphy Modifier 1

PG 64-34 Black Max, Husky Modifier 2

PG 64-28 Neat, Seneca Plain 1

PG 64-28 SBS, Seneca Modifier 1

PG 64-22 Neat, Seneca Plain 1

some binders, such as PG 54-34 Modifier 1, PG 54-34 plain, and PG64-28 Modifier 1, the failure stress and failure strain were identicalfor the two aging conditions. This observation indicates that thechanges in fracture properties are not consistent with the expectationthat PAV binders are more brittle than RTFO binders.

The stress–strain curves obtained at 10°C above the PG lower-limit temperature indicate that for most of the asphalt binders tested,PAV aging increases the brittleness of the binder, in particular inrelation to the failure strain. This trend was not observed in the PG58-28 Plain 2 binder, for which the RTFO and PAV stress–straincurves were similar.

SUMMARY AND CONCLUSIONS

The effect of the PAV laboratory aging procedure on low-temperaturecreep stiffness and fracture properties of asphalt binders wasinvestigated in this study.

Three binders extracted from three cells at MnROAD were com-pared to the original binders used in the mixture preparations thatwere aged in the laboratory by using RTFO and PAV methods. DSRhigh-temperature data indicate very small differences between the�G*� master curves of the extracted and PAV binders. However, sig-nificant differences were observed in the phase angle master curvesfor the modified binders: the plateau seen in the PAV binders dis-appears for the extracted binders, which indicates a clear differencein the presence of the polymer phase for the two conditions. Analy-sis of BBR creep stiffness indicates the extracted binders werealways stiffer than the PAV binders. There is also a clear reductionin the slope of the creep stiffness curves for the extracted binderscompared to the PAV binders. These results appear to indicate thatthe extracted binders have aged more than the PAV binders. Theanalysis of the DTT stress–strain curves reveals higher fracturestresses and lower fracture strains for all extracted binders comparedto PAV binders.

An additional experiment was performed to better understand theeffect of PAV aging compared to RTFO aging. It was found thatPAV binders had higher creep stiffness and lower m-values thanRTFO binders. However, as temperatures approached the PG lower-limit temperature (shorter times), the difference between RTFO andPAV diminished considerably, which appears to suggest the exis-tence of a common asymptotic creep stiffness value. The analysis of


1

10

100

1,000

0.1 1.0 10.0 100.0 1,000.0

Stif

fnes

s [M

Pa]

1

10

100

1,000

Stif

fnes

s [M

Pa]

Time [sec]

PAV

RTFO

0.1 1.0 10.0 100.0 1,000.0 Time [sec]

(a) (b)

PAV

RTFO

FIGURE 5 BBR stiffness comparison for (a) PG 58-34 Modifier 1 at �24�C and (b) PG 58-28 Plain 2 at �18�C.

0

2

4

6

8

0 2 4 6 8 10 12

Str

ess

[MP

a]

Strain [ % ]

PAV

RTFO

0 2 4 6 8 10 12

Strain [ % ]

(a) (b)

0

2

4

6

8S

tres

s [M

Pa]

PAV

RTFO

FIGURE 6 Stress–strain plots for PG 58-34 Modifier 1 at (a) �24�C and (b) �30�C.

0

2

4

6

8

0 2 4 6 8 10 12

Str

ess

[MP

a]

Strain [ % ]

0 2 4 6 8 10 12

Strain [ % ]

(a) (b)

PAV

RTFO

0

2

4

6

8

Str

ess

[MP

a]

PAV

RTFO

FIGURE 7 Stress–strain plots for PG 58-28 Plain 2 at (a) �18�C and (b) �24�C.


PAV

RTFO

PAV

RTFO

0

2

4

6

8

0 2 4 6 8 10 12

Str

ess

[MP

a]

Strain [ % ]

0 2 4 6 8 10 12

Strain [ % ]

(a) (b)

0

2

4

6

8

Str

ess

[MP

a]


PAV

RTFO

PAV

RTFO

0

2

4

6

8

0 2 4 6 8 10 12

Str

ess

[MP

a]

Strain [ % ]0 2 4 6 8 10 12

Strain [ % ](a) (b)

0

2

4

6

8

Str

ess

[MP

a]


PAV

RTFO

PAV

RTFO

0

2

4

6

8

0 2 4 6 8 10 12

Str

ess

[MP

a]

Strain [ % ]

0 2 4 6 8 10 12

Strain [ % ]

(a) (b)

0

2

4

6

8

Str

ess

[MP

a]


the DTT stress–strain curves confirmed the trend observed in the BBRdata as temperatures approached the PG limit. At the lowest test tem-perature very small differences were observed between the RTFO andPAV conditions, and for some binders, the failure stress and failurestrain were identical for the two aging conditions. Therefore, theassumption that PAV binders are more brittle than RTFO binders isnot valid at temperatures close to the PG limit.

A number of important issues need to be further investigated. In thecurrent PG specification, PAV-aged binders are used to determinethe low-temperature specification limit, although PAV was originallydeveloped on the basis of intermediate- and high-temperature G*experimental data. The current low-temperature specification is basedon BBR data, and most of the time fracture testing is not considered,although research studies have shown that fracture properties corre-late best with cracking occurrence. The stiffness–brittleness equiva-lence used in the original work done during SHRP was based on plainbinders. In recent years, more modified asphalt binders and increasedamounts of recycled materials are used in asphalt pavement construc-tion. For these materials, this equivalence is most likely not valid anddeserves additional research.

ACKNOWLEDGMENT

This research was sponsored by a national pooled-fund study. Thesupport is gratefully acknowledged.

REFERENCES

1. Choquet, F., and A. Verhasselt. Natural and Accelerated Aging of Bitu-mens: Effects on the Asphaltenes. In Asphaltene Particles in Fossil Fuel


Exploration, Recovery, Refining, and Production Process: Proc., Inter-national Symposium and 23rd Annual Meeting of Fine Particles Society(M. K. Sharma and T. F. Yen, eds.), Las Vegas, Nev., 1992, pp. 13–29.

2. Lu, X., and U. Issacson. Influence of Styrene–Butadiene–StyrenePolymer Modification on Bitumen Viscosity. Fuel, Vol. 76, 1997, pp. 1353–1359.

3. Verhasselt, A. Field Aging of Bituminous Binders. Proc., 5th InternationalRILEM Symposium: Mechanical Test for Bituminous Materials (H. DiBenedetto and L. Francken, eds.), Lyon, France, 1997, pp. 121–128.

4. Glover, C., R. Davison, C. Domke, Y. Ruan, P. Juristyarini, and D. Knorr. Development of a New Method for Assessing Asphalt BinderPerformance Durability. Report No. FHWA/TX-02/1872-1. TexasTransportation Institute, Arlington, 2001.

5. Al-Azri, N. A., S. H. Jung, K. M. Lunsford, A. Ferry, J. A. Bullin,R. R. Davison, and C. J. Glover. Binder Oxidative Aging in TexasPavements: Hardening Rates, Hardening Susceptibilities, and Impactof Pavement Depth. In Transportation Research Record: Journal ofthe Transportation Research Board, No. 1962, TRB, National ResearchCouncil, Washington, D.C., 2006, pp. 12–20.

6. Marasteanu, M., A. Zofka, M. Turos, X. Li, R. Velasquez, and X. Li.National Pooled Fund Study 776: Investigations of Low-TemperatureCracking in Asphalt Pavements. Research report 2007-43. MinnesotaDepartment of Transportation, Saint Paul, 2007.

7. SHRP Report A-370: Testing, Binder Characterization, and Evaluation,Vol. 4, Appendix E. National Research Council, Washington, D.C., 1994.

8. Marasteanu, M., and D. Anderson. Improved Model for BitumenRheological Characterization. Presented at Eurobitume Workshop onPerformance-Related Properties for Bituminous Binders, Luxembourg,Belgium, 1999.

9. Petersen, J. C. SHRP Report A-367: Binder Characterization and Evalu-ation, Vol. 1. National Research Council, Washington, D.C., 1994.

10. Migliori, F., and J.-F. Corte. Comparative Study of RTFOT and PAVAging Simulation Laboratory Tests. In Transportation ResearchRecord 1638, TRB, National Research Council, Washington, D.C.,1998, pp. 56–63.

The Characteristics of Asphalt Materials Committee peer-reviewed this paper.

Documents

Pressure Aging Vessel and Low-Temperature Properties of Asphalt Binders