Understanding The Performance of Modified Asphalt Binders in Mixtures: High-Temperature Characterization

Phase 2--Evaluation of High-Temperature Asphalt Binder Tests
Using the RSCH and French PRT Mixture Tests at 70°C

A. Background

The polymer-modified asphalt binders used in this study had continuous high-temperature PG's ranging from 71 to 77. The mixtures were tested by the French PRT at 70°C . This was the highest temperature that can be applied by this tester. The mixtures were tested for cumulative permanent shear strain at 50°C because of testing problems encountered at higher temperatures in previous studies. However, most of the problems were provided by the Frequency Sweep at Constant Height (FSCH) mode of loading. RSCH and FSCH are both applied by the Superpave Shear Tester. It was recommended that the diabase mixtures be retested using RSCH at a temperature closer to the PG's of the asphalt binders.

B. Objective

The objective of phase 2 was to retest the diabase mixtures at 70°C using RSCH to determine which asphalt binders provide high-temperature properties that do not agree with mixture rutting resistance.

C. Materials

All 11 asphalt binders were used. Information on the asphalt binders, aggregates, and mixture design are given elsewhere.^(1-2) The reduced asphalt binder content of 4.55 percent by mixture mass was used so that the mixture met the 4.0-percent design air-void level as recommended by Superpave. Prior problems with testing mixtures at temperatures above 50°C contributed to the decision to lower the asphalt binder content. However, it confounded the analysis of the data because the mixtures tested by the French PRT had a 4.85-percent asphalt binder content.

D. Tests

High-temperature asphalt binder properties were measured by a DSR after RTFO aging.⁽³⁾ Mixture rutting resistance was based on the cumulative permanent shear strains from RSCH and the rut depths from the French PRT.^(4-5) All mixtures were subjected to 2 h of STOA at 135°C . Specimens were tested 48 h after compaction.

E. Cumulative Permanent Shear Strain

Cumulative permanent shear strain from RSCH was measured at 7.0-percent air voids, 70°C , and 5,000 cycles. The applied shear stress was 69 ±5 kPa. The loading time was 0.1 s and the rest time was 0.6 s. A minimum of three replicate specimens were tested per mixture. Lower cumulative permanent shear strains indicate more resistance to rutting.

The data at both 50°C and 70°C are given in table 6 and figure 4. The two regression lines suggest that mixtures containing asphalt binders with the highest PG's tended to be highly resistant to rutting at both test temperatures. While this seems reasonable, the difference in asphalt binder content probably decreased the vertical differences between the two lines.

Table 6 shows that the rankings are not the same at 50°C and 70°C . The only certain difference in ranking is that the mixture with PG 70-22 performed relatively worse at 70°C compared to 50°C . Figure 5 shows that the data point for this mixture was furthest from the regression line. Being above the line, this mixture performed worse than expected at 70°C , or better than expected at 50°C . Because of the change in asphalt binder content, no firm conclusions could be made for the other asphalt binders.

The cumulative permanent shear strains in table 6 indicate that grafting did not improve the rutting resistance of EVA at either temperature at a 5 percent level of significance. Grafting and geometry had no significant effect on the rutting resistance of SBS at 50°C . All three SBS mixtures fell into group D. SBS Radial Grafted did perform significantly better than SBS Linear at 70°C . The shear strain for SBS Linear Grafted at 70°C was not significantly different from the shear strains for SBS Radial Grafted or SBS Linear at 70°C .

The coefficients of variation (CV) for cumulative permanent shear strain at 5,000 cycles ranged from 1.7 percent to 36.7 percent. (See table 7.) Several data points were found to be outliers. These outliers were not used when evaluating asphalt binder properties. However, all of the replicate data are included in table 7 because the variability of the data in table 7 is a good representation of the variability typically provided by RSCH. The averages and CV's in the parentheses include the outliers. At 50°C , the CV's ranged from 8.2 to 24.1.⁽¹⁾ Variability appeared to be greater at 70°C , but a paired t-test indicated that it was not greater.

For the tests at 50°C , the cumulative permanent shear strains were correlated to the G*/sind 's of the asphalt binders at three DSR frequencies: 10.0, 2.0, and 0.125 rad/s.⁽¹⁾ The G*/sind 's are given in table 8. These three frequencies provided r²'s of 0.06, 0.55, and 0.89, respectively, using log-log transformations. The correlation depended on DSR frequency. The correlation to high-temperature PG was 0.68 without transformation. (Note: A log-log transformation was used if it provided a higher r². Relationships between asphalt binder and mixture high-temperature properties are normally curvilinear. Thus, a log-log or power law transformation usually provides a higher r².)

Table 6. DSR data and RSCH data at 70°C and 50°C .

Figure 4. RSCH cumulative permanent shear strain vs.
high-temperature PG of the asphalt binder.

Figure 5. RSCH cumulative permanent shear strain at 70°C
vs. RSCH cumulative permanent shear strain at 50°C.

Table 7. Replicate cumulative permanent shear strains at 70°C .

Table 8. G*/sind 's of the asphalt binders at 10.0, 2.0, and 0.125 rad/s with the
asphalt binders listed from highest to lowest G*/sind based on 0.125 rad/s.

Table 6 gives the G*/sind 's at 70°C . When correlated against cumulative permanent shear strain, frequencies of 10.0, 2.0, and 0.125 rad/s provided r²'s of 0.22, 0.37, and 0.59, respectively, using log-log transformations. This correlation also depended on DSR frequency. The correlation to high-temperature PG was 0.63 without transformation. All of the r²'s are given in table 9.

The correlations to G*/sind at 50°C and 70°C using a DSR frequency of 0.125 rad/s are shown in figures 6 and 7, respectively. The correlation at 50°C is very good. The largest deviation was provided by Elvaloy, followed by SBS Radial Grafted. Figure 7 shows that the G*/sind 's for EVA and SBS Radial Grafted are low at 70°C . G*/sind underpredicted their resistances to rutting. Because the data for EVA significantly affected the position of the trend line, the trend line in figure 7 was drawn without the data for EVA. The data point for SBS Radial Grafted did not significantly affect the position of the trend line.

Figures 8 and 9 show the data at 70°C using DSR frequencies of 2.0 and 10.0 rad/s, respectively. Both figures show that the G*/sind for PG 70-22 is high, while the G*/sind 's for EVA and Elvaloy are low. The G*/sind for SBS Radial Grafted is low at 2.0 rad/s, but not at 10.0 rad/s based on 95-percent confidence bands.

Figure 10 shows that the correlation using high-temperature PG is poor, although the r² increased from 0.63 to 0.79 after excluding the data for SBS Radial Grafted. The 95-percent confidence band for cumulative permanent shear strain at the mean PG of 74 is 22 000 to 36 000 mm/m with SBS Radial Grafted and 25 000 to 36 000 mm/m without SBS Radial Grafted.

The higher r² using G*/sind at 0.125 rad/s compared to G*/sind at 10.0 rad/s suggests that, according to cumulative permanent shear strain, a low DSR frequency might provide a better grading system. If the frequency is changed, then the criterion, which is currently 2200 Pa after RTFO, must also be changed. Figure 11 shows the relationship between cumulative permanent shear strain and temperature if the frequency is changed to 0.125 rad/s, but the criterion is not changed. The temperatures are very low and the correlation is poor. The lack of a known correlation between cumulative permanent shear strain and pavement rutting makes it difficult to choose a criterion. A preliminary recommendation based on the ranking in table 6 is to use a maximum allowable shear strain of around 30 000 mm/m to 40 000 mm/m. Figure 7 shows that a shear strain of 30 000 mm/m (log 30 000 = 4.477) provides a criterion of around 60 Pa (log 60 1.8), although the scatter in figure 7 shows that this will not provide a perfect grading system. Furthermore, the relationship based on 0.125 rad/s in figure 7 is not better than the relationship based on high-temperature PG in figure 10.

Table 10 and figure 12 provide the cumulative permanent shear strains for the asphalt mixtures at 70°C and 5,000 cycles vs. the cumulative permanent shear strains for the asphalt binders from repeated creep at 70°C and 100 cycles. The correlation is poor, having an r² of 0.58. If the data for the PG 64-28 materials are removed, the r² drops to 0.21. Based on the mixture test results, the cumulative permanent shear strains for the Elvaloy and EVA Grafted asphalt binders are high, while they are low for the PG 70-22 asphalt binder. The relationship should start at the zero-zero origin, but it does not. Therefore, the relationship must be curvilinear. Figure 13 provides a log-log relationship. This did not improve the correlation. The r² of 0.38 is poor. The repeated creep is a new asphalt binder test and it is not known if the protocols are the optimal protocols.

F. French PRT

The rut depths from the French PRT at 70°C are given in tables 11 and 12.⁽¹⁾ Table 12 shows that the mixture with SBS Radial Grafted had a high coefficient of variation. Tests on the mixtures with EVA and SBS Radial Grafted were repeated. The range in the replicate rut depths for these mixtures is relatively large compared to the range in average rut depth for all modified asphalt binders. The statistical ranking in table 11 shows that the rut depths for all mixtures, except for the mixture with PG 64-28, had rut depths that were not different at a 5-percent level of significance.

Figures 14 and 15 show the relationships between rut depth and G*/sind using DSR frequencies of 0.9 and 10.0 rad/s, respectively. The G*/sind 's for EVA may be low. If so, G*/sind underpredicted the relative rutting resistance provided by EVA. Without EVA, frequencies of 0.9 and 10.0 rad/s provided r²'s of 0.83 and 0.91, respectively, compared to 0.54 and 0.56 with EVA. Without both EVA and PG 64-28, frequencies of 0.9 and 10.0 rad/s provided r²'s of 0.69 and 0.67, respectively. The data point for PG 64-28 increases the upward curvature of the relationship, while the data point for EVA tends to flatten the relationship. If the 9.9-percent rut depth for SBS Radial Grafted in table 12 were to be eliminated, G*/sind would also underpredict the relative rutting resistance of this asphalt binder.

Figure 16 provides the correlation with high-temperature PG. The PG of EVA agrees with mixture performance. This means that the G*/sind for EVA is not increasing as rapidly as it should at temperatures immediately below its high-temperature PG. EVA was found to have the lowest slope (G*/sind divided by temperature) around the grading temperatures. The r² of 0.89 drops to 0.57 without the data for PG 64-28. Even so, no data point is more than 1.5°C away from the regression line.

Unlike cumulative permanent shear strain, low and high DSR frequencies provided approximately the same degree of correlation with rut depth, even though a cursory review of the G*/sind 's in table 11 showed that the two frequencies did not provide identical rankings for the asphalt binders. To examine this in more detail, the G*/sind 's were linearly regressed. The r² of 0.81 in figure 17 shows that the relationship was good. Without Elvaloy, the r² is 0.97. Based on the rut depths in table 11, the G*/sind of 4110 Pa for Elvaloy at 10 rad/s is low relative to the other asphalt binders. It should have the highest G*/sind .

Table 9. Coefficients of determination between RSCH and DSR properties.

Figure 6. Log RSCH cumulative permanent shear strain at 50°C
vs. log (G*/sind) of the asphalt binder at 50°C and 0.125 rad/s.

Figure 7. Log RSCH cumulative permanent shear strain at 70°C
vs. log (G*/sind) of the asphalt binder at 70°C and 0.125 rad/s.

Figure 8. Log RSCH cumulative permanent shear strain at 70°C
vs. log (G*/sind) of the asphalt binder at 70°C and 2.0 rad/s.

Figure 9.Log RSCH cumulative permanent shear strain at 70°C
vs. log (G*/sind) of the asphalt binder at 70°C and 10.0 rad/s.

Figure 10. RSCH cumulative permanent shear strain
at 70°C vs. high-temperature PG of the asphalt binder.

Figure 11. RSCH cumulative permanent shear strain
at 70°C vs. PG temperature based on 0.125 rad/s.

Table 10. Cumulative permanent shear strain at 70°C .

Figure 12. Asphalt mixture cumulative permanent shear strain
vs. asphalt binder cumulative permanent shear strain.

Figure 13. Comparison of cumulative permanent
shear strain using a log-log transformation.

Table 11. DSR data and French PRT rut depths with the materials
listed from lowest to highest rut depth at 6,000 wheel passes.

Table 12. Replicate data for the French PRT at 6,000 wheel passes.

Figures 14. French PRT rut depth at 70°C vs.
G*/sind of the asphalt binder at 70°C and 0.9 rad/s.

Figure 15. French PRT rut depth at 70°C vs.
G*/sind of the asphalt binder at 70°C and 10.0 rad/s.

Figures 16. Figure 16. French PRT rut depth
at 70°C vs. high-temperature PG.

Figure 17. G*/sind at 0.9 rad/s vs. G*/sind at 10.0 rad/s.

Figure 18 provides the correlation with the cumulative permanent shear strains from the asphalt binder repeated creep test. The r² of 0.83 drops to 0.19 without the data for PG 64-28. The narrow range in rut depth provided by the French PRT makes it difficult to make a firm conclusion.

Figure 19 provides the relationship between the cumulative permanent shear strain from RSCH at 70°C and the French PRT rut depth at 70°C . If the data for the mixture with the PG 64-28 asphalt binder are excluded, the remaining data indicate that the French PRT provided a narrower range in performance compared to RSCH. The statistical rankings in tables 6 and 11 support this finding. Table 6 shows that shear strain provided five statistical groups (A through E), while table 11 shows that only the mixture with the PG 64-28 asphalt binder had a significantly different resistance to rutting according to the French PRT. In a previous FHWA study, both tests agreed with full-scale pavement rutting tests, although only five asphalt binders were evaluated and only two of these binders were polymer-modified asphalt binders.⁽⁵⁾

The mixtures with EVA, EVA Grafted, and SBS Linear at an asphalt binder content of 4.85 percent were tested using RSCH at 70°C to determine if the reduction in asphalt binder content contributed to the differences between RSCH and the French PRT. Table 13 shows that the cumulative permanent shear strains for EVA Grafted and SBS Linear at an asphalt binder content of 4.85 percent were not repeatable, so a conclusion could not be made.

G. Conclusions

The cumulative permanent shear strains from RSCH at 70°C were correlated to the G*/sind 's of the asphalt binders at 70°C and three DSR frequencies: 10.0, 2.0, and 0.125 rad/s. The best correlation was provided by a frequency of 0.125 rad/s. At 0.125 rad/s, G*/sind underpredicted the relative rutting resistance provided by EVA and SBS Radial Grafted. G*/sind at the standard frequency of 10.0 rad/s underpredicted the rutting resistances provided by EVA and Elvaloy, and overpredicted the relative rutting resistance provided by the unmodified PG 70-22 asphalt binder. High-temperature PG underpredicted the relative rutting resistance provided by SBS Radial Grafted.

Based on the French PRT at 70°C , G*/sind underpredicted the relative rutting resistance provided by EVA at both high and low DSR frequencies. However, the high-temperature PG of EVA agreed with mixture performance. This means that the G*/sind for this asphalt binder did not increase as rapidly as it should have at temperatures immediately below its high-temperature PG of 75°C . The correlation between high-temperature PG and the French PRT provided no obvious outliers.

Grafting did not improve the rutting resistance of EVA. Grafting and geometry had no effect on the rutting resistances of the SBS-modified asphalt binders at 50°C . The effect at 70°C was marginal.

H. Recommendations

• The French PRT indicated that the current Superpave binder specification is valid. The G*/sind for one asphalt binder, EVA, was low at 70°C , but its high-temperature PG agreed with mixture rutting performance. No changes to the specification are recommended based on the French PRT results.
  
  
• The cumulative permanent shear strains from RSCH suggested that a low DSR frequency, such as 0.125 rad/s, might provided a better grading system than 10.0 rad/s. However, it is not known whether this finding applies to pavements or if it is related to the accelerated nature of the RSCH test. This requires full-scale validation. Furthermore, G*/sind at 0.125 rad/s and 70°C did not clearly provide a better correlation to RSCH than high-temperature PG.
  
  
• Based on the coefficients of variation in tables 2, 7, and 13, a minimum of five replicate specimens should be tested by RSCH per mixture.
  
  
• Additional research is needed to evaluate various protocols for the asphalt binder repeated creep test.
  
  
• For similar studies involving fewer asphalt binders, it is imperative that the high-temperature PG's of the asphalt binders be as close to each other as possible. If the performances of the asphalt binders at one particular temperature are of interest, then the most important property of the asphalt binders, such as their G*/sind 's, must be as close to each other as possible at this temperature. In this study, the high-temperature PG's of the polymer-modified asphalt binders varied from 71°C to 77°C . The G*/sind 's of the asphalt binders at the test temperature of 70°C also varied significantly. Regression and ranking statistical analyses were used to find which asphalt binders had properties that did not agree with the properties of the other asphalt binders based on mixture rutting resistance. For studies involving fewer asphalt binders, these analyses may not provide a valid conclusion because of insufficient data points. Therefore, the deviation in the high-temperature properties of the asphalt binders must be so small that they should not affect mixture rutting resistance. An alternative approach is to adjust the mixture test temperature according to the asphalt binder property so that all of the asphalt binders should provide the same resistance to rutting.

Figures 18. French PRT rut depth vs. asphalt
binder cumulative permanent shear strain.

Figure 19. Cumulative permanent shear strain of the asphalt
mixture vs. French PRT rut depth of the asphalt mixture.

Table 13. RSCH cumulative permanent strains at 70°C
and 5,000 cycles using two asphalt binder contents.

References

1. K. D. Stuart and W. S. Mogawer, Understanding the Performance of Modified Asphalt Binders in Mixtures: Permanent Deformation Using a Mixture With Diabase Aggregate (Publication No. FHWA-RD-02-042), Federal Highway Administration, McLean, VA, December 2001, 67 pp.
   
   
2. NCHRP Project 90-07, "Understanding the Performance of Modified Asphalt Binders in Mixtures," Work Plan, Study in Progress, National Cooperative Highway Research Program (NCHRP), Transportation Research Board, National Research Council, Washington, D.C., 2001.
   
   
3. AASHTO TP5, "Method for Determining the Rheological Properties of Asphalt Binder Using a Dynamic Shear Rheometer," AASHTO Provisional Standards, American Association of State Highway and Transportation Officials, Washington, D.C., April 2000 Edition.
   
   
4. AASHTO TP7, "Method for Determining the Permanent Deformation and Fatigue Cracking Characteristics of Hot Mix Asphalt (HMA) Using the Simple Shear Test (SST) Device," AASHTO Provisional Standards, American Association of State Highway and Transportation Officials, Washington, D.C., April 2000 Edition.
   
   
5. K. D. Stuart, W. S. Mogawer, and P. Romero, Validation of Asphalt Binder and Mixture Tests That Measure Rutting Susceptibility Using the Accelerated Loading Facility, Publication No. FHWA-RD-99-204, Federal Highway Administration, McLean, VA, December 1999, 348 pp.
   
   
6. H. U. Bahia, D. I. Hanson, M. Zeng, H. Zhai, M. A. Khatri, and R. M. Anderson, Characterization of Modified Asphalt Binder in Superpave Mix Design, Appendix IV, NCHRP Report 459, National Cooperative Highway Research Program, Transportation Research Board, Washington, D.C., 2001.

back to top

Previous | Table of Contents