By: E. Ray Brown, Ph.D.
The first two-year cycle of tests at the National Center for Asphalt Technology (NCAT) Test Track, Auburn, Ala., was completed recently. The testing of 46 different sections was conducted by subjecting the pavements to the equivalent of 10 to 12 years of traffic loading in two years. Heavier-than-normal trucks punished the pavements with 10 million equivalent single axle loads (an ESAL equals one 18,000-lb single axle load).
The results of this first series of tests were presented at the National Transportation Symposium at NCAT held on Nov. 13-14, 2002. Most of the 250 attendees were representatives of the state departments of transportation that sponsored test sections at the track.
Here are some of the findings:
* A number of mixes with varying aggregate types, asphalt grades and aggregate gradations provided good performance;
* The total average rut depth was very small--.12 in. after 10 million ESALs. The worst-performing sections had .25 in. of rutting;
* Mixes that were intentionally designed to be more likely to rut did result in the greatest rut depths, but the magnitude of these ruts was still small;
* Fine-graded and coarse-graded mixes provided approximately equal performance;
* The higher PG grades clearly resulted in lower rut depths, and the results indicate that potentially more asphalt binder may be added to mixes using higher PG grades to improve durability without increasing rutting; and
* Several of the tests that were evaluated have been found to be useful in helping to identify mixes with rutting potential.
Carrying a lot of weight
Empirical laboratory tests have been used for years to test hot-mix asphalt (HMA) to determine the potential for various mixtures to perform satisfactorily. As the amount of traffic has increased (higher volumes, higher loads and increased tire pressures) the ability of these laboratory tests to evaluate potential performance has become more important.
The best way to quickly evaluate and verify the potential for a test to be used is with accelerated loading facilities. The test track at NCAT is one method that can be used to rapidly test a large number of mixes simultaneously using full-scale vehicles.
The sponsors of the first cycle at the track included the Alabama Department of Transportation, which financed the overall construction of the track, as well as the Federal Highway Administration (FHWA) and other state DOTs, which sponsored test sections. The participating DOTs were Florida, Georgia, Indiana, Mississippi, North Carolina, Oklahoma, South Carolina and Tennessee.
The 46 test sections constructed used various aggregates, grades of asphalt and various mixture types. Some mixtures were designed with marginal aggregates and some mixtures were designed with .5% additional asphalt. Several mixture types were used including fine- and coarse-graded Superpave, stone matrix asphalt, open-graded friction courses, as well as some variations of these mixtures.
As the NCAT Test Track was constructed, 184 temperature probes--four for each of the 46 test sections--were installed at various depths within the pavement layer. Moisture gauges also were installed in each of the test sections at the surface of the improved subgrade.
Four tractors pulled triple-trailer assemblies around the track at 45 mph for 17 hours a day, six days a week, in order to apply 10 million ESALs to the pavements within two years. The seven axles of the trailers were loaded to a total weight of 20,000 lb each (including 12,000 lb on the steer axle), resulting in a total gross vehicle weight of each rig being approximately 152,000 lb. Although the main focus of the research was the accelerated performance of the test sections, data collected and observations made in support of the trucking operations have provided valuable information upon which future trucking operations can be refined.
Measurements and observations were made weekly to determine the overall performance of all of the sections.
Effects of distress
The primary purpose of the track was to evaluate various mixture types and to evaluate the ability of laboratory tests to predict performance. The only distress expected at the test track was some type of surface-related problem such as rutting. The pavement was designed to be strong enough to prevent fatigue cracking. Due to the relatively short time of evaluation, durability problems such as stripping were not expected.
All of the sections performed very well for the first 10 million ESALs. In fact, no maintenance was required on any section other than a section in one of the curves that had a friction problem. The mixture in this section used a limestone aggregate that was known to polish. It was selected for use in stone matrix asphalt (SMA) to determine if the coarse surface texture would continue to provide good skid resistance as it was subjected to traffic. The friction was monitored on a regular basis and when the friction fell below an acceptable point, the test section was immediately overlaid with a maintenance course to improve friction.
The overall level of rutting was very small. The worst section had only about .25 in. of rutting. Most state DOTs don't begin to consider rutting to be significant until it exceeds about .5 in. Clearly, none of the 46 sections approached this amount of rutting. For that reason, some of the sponsors plan to continue traffic on their sections for another two years.
Generally, the rutting was higher in the right wheel path than in the left path. There were probably at least two reasons for this. First of all, there was a 2% transverse slope on each of the pavement sections in the tangents. This slope likely resulted in a slightly heavier load on the right side of the traffic lane than on the left side. Secondly, the material adjacent to the right side of the lane likely did not provide as much confinement as the well-compacted mix on the left side of the traffic lane. Hence, more rutting would be expected in the right wheel path.
It also is interesting to note that the variability of the rutting within a section was very low. Even though the overall rutting was low, it appeared to be consistent between tests within each section. Hence, the measured rutting appeared to be actual rutting and not some random variability.
Some of the sections appeared to have statistically significantly higher rutting values than others. Some were designed so they would be more likely to rut.
For example, the sections with the most rutting were designed with .5% additional asphalt binder so they might experience rutting when exposed to this high level of traffic. These sections also included asphalt binders that were not bumped to the next higher PG (performance graded) level. Also, some of the sections used aggregates that were marginal, and it was expected this could cause some rutting problems.
Another item of interest was the observed rutting rates. The rutting rate was relatively higher during the first 800,000 ESALs even though much of the traffic was applied during the winter months. It appears this first significant rate of rutting was caused more by initial seating of the aggregate and initial compaction. For example, if the average density in the top 4 in. increased by 1% this alone should result in an average rut depth of approximately .04 in.
After this initial densification and seating of the aggregate, the rutting rate was reduced to near zero until the average seven-day maximum daily temperature reached approximately 28°C (82°F), at which time the rutting rate again began to climb. However, the rate appeared to be a little less than the initial rate. After the seven-day maximum daily temperature dropped below approximately 28°C, the rate of rutting almost went to zero again. The rate of rutting stayed near zero until the temperature exceeded approximately 28°C, at which time the rate began to increase a little.
The rate was much lower during the second summer than it was during the first summer, even though the temperature was higher during the second.
Based on this observation, it appears that a mix that is properly designed will stabilize within a couple of years due to aging and compaction. The data also shows that the initial seating and densification resulted in an overall average rut depth of approximately .03 in. The first summer resulted in an additional .07 in. of rutting and the second summer resulted in an additional .02 in. of rutting on average. As stated earlier, the total average rut depth was very small--.12 in.
A mini experiment involving 10 sections was set up to look at the effect of PG grade, asphalt content and fine-graded vs. coarse-graded mixes. One observation from this evaluation was that the mixes with modified asphalts (PG-76) had significantly lower rutting (66% lower) than the mixes without modified asphalts. This indicates the importance of bumping the PG grade on high-volume roads, as specified by Superpave.
Another observation was that binders modified with SBS and SBR gave very similar results. The fine-graded and coarse-graded mixes provided approximately equal performance.
Increasing the asphalt content by .5% resulted in an increase of 54% in the rutting of the unmodified mixes. When the mixes were modified, the increase in rutting as a result of the increased asphalt content was very small (less than .04 in.). This indicates that one may be able to use slightly higher asphalt content with modified asphalts to improve durability without causing a loss in performance due to rutting.
There were three mini experiments to look at fine-graded vs. coarse-graded mixes. The data from these three mini experiments clearly shows that there was very little difference in the amount of rutting of fine-graded and coarse-graded mixes. Hence, from a rutting standpoint, good performance can be obtained with fine-graded as well as with coarse-graded mixes.
Of course, one of the keys to ensuring good performance is to have a test accurately related to performance. Tests have been used over the years to predict performance, and new tests are being evaluated. Several laboratory tests were evaluated for the purpose of predicting performance, including wheel tracking tests, Superpave simple shear, dynamic modulus and confined repeated load test.
Keep in mind that the rutting observed at the track was very small, so it is difficult for these tests to accurately predict the rutting. If the rutting numbers were higher, then a better evaluation could be made concerning the potential for each of these tests to predict performance.
Still good
The rutting results are very preliminary; hence, one must be careful in making too many conclusions. However, it is worth noting that there appeared to be no relation between rutting and dynamic modulus. There was a reasonable relationship between the confined repeated load test and the Superpave shear test. The rut testers also indicated a correlation with performance.
In summary, the first series of tests showed that a number of mixes with varying aggregate types, asphalt grades and aggregate gradations provided good performance when subjected to 10 million ESALs of traffic. Mixes intentionally designed to be more likely to rut did result in the greatest rut depths, but the magnitude of these ruts was still small. Fine-graded and coarse-graded mixes provided approximately equal performance. The higher PG grades clearly resulted in lower rut depths, and the results indicate that potentially more asphalt binder may be added to mixes using higher PG grades to improve durability without increasing rutting.
