Located on a 309-acre site, the 1.7-mile NCAT Pavement Test Track is a world-renowned accelerated pavement testing facility that combines full-scale pavement construction with live, heavy truck trafficking for rapid testing and analysis of asphalt pavements.
Since its original construction in 2000, six three-year research cycles have been conducted, and findings from the track have helped sponsors refine their materials specifications, construction practices and pavement design procedures for asphalt pavements.
In 2015, the sixth research cycle began a new chapter in full-scale pavement research through a partnership with the Minnesota DOT’s MnROAD facility. One of the major NCAT-MnROAD partnership efforts is to validate asphalt mixture cracking tests that are suitable for routine use in mix design and quality assurance testing. The experiment for the validation of cracking tests is called the Cracking Group (CG) Experiment and includes test sections built on both the NCAT Test Track and MnROAD’s mainline test road. In addition to the CG Experiment, another focus area of the sixth research cycle is placed on the responsible use of recycled materials in asphalt pavements. This effort includes the continued trafficking and performance monitoring of three test sections of the 2012 Cold-Central Plant Recycling (CCPR) and Stabilized Base Experiment. Key findings from the CG Experiment and the CCPR and Stabilized Base Experiment are summarized in the following sections.
The plant mix for the seven test sections was subject to laboratory testing; presently, plant mix samples are being lab-processed to simulate four years of in-field use and aging. Pictured are (top, middle) the production and loading process and (bottom) the lab interior where samples were excised and tested.
Cracking group experiment: validation of cracking tests
The ongoing CG Experiment includes seven new test sections at the NCAT Test Track for validating tests for top-down cracking and eight rebuilt test sections at MnROAD for validating tests for low-temperature cracking of asphalt mixtures. Preliminary results are from the NCAT Test Track experiment with results from the MnROAD experiment to be reported in the future.
The seven CG test sections on the NCAT Test Track are evaluated under the same traffic and environmental conditions and have similar pavement structures, except for the surface mixes. The surface mixtures were designed with a range of recycled materials contents, binder types and grades, and in-place densities to achieve various levels of cracking performance. They were constructed as 1.5-in. surface lifts over highly polymer-modified intermediate and base layers of asphalt. The target thickness for both the intermediate and base layers was 2.25 in. per layer. The asphalt pavement cross-section was relatively thin for the heavy loading on the track so that the surface layers would experience significant stress and strains but avoid bottom-up fatigue cracking by using the highly modified mix for intermediate and base layers. The construction of these sections was completed in the summer of 2015.
After two years of trafficking with 10 million accumulated ESALs, only one of the seven test sections (Section N8) had a substantial amount of top-down cracking in nearly 17% of the lane area. The surface mix of this section has 20% reclaimed asphalt pavment (RAP) and 5% recycled asphalt shingles (RAS) with a PG 67-22 binder. Limited coring showed that the cracking was confined to the surface layer with no evidence of de-bonding between layers. Analyses of back-calculated asphalt concrete moduli also indicated that the wheel path cracking in this section has resulted in damage to the pavement structure.
Three other test sections also showed evidence of very low-severity cracking on their surfaces. Based on the cores extracted at the cracked locations in these sections, there was no visible evidence that the observed hairline surface cracking in these sections had propagated into the surface layer. Analyses of back-calculated moduli indicated no damage to these pavement structures. The seven sections have remained in place for continued traffic and performance monitoring for another three-year research cycle to increase the amount and severity of cracking in several test sections in order to accomplish the experiment’s objectives.
During the construction of these sections, the plant mix for the seven surface layers was sampled for laboratory testing. Testing of plant mix samples that were reheated just enough to fabricate the specimens has been completed and analyzed in the test track report. Additional work is underway to test plant mix samples that have been laboratory-aged to simulate approximately four years of field aging in Auburn, Ala., as well as testing of laboratory-prepared mixtures that are aged to represent mix production aging and four years of in-service aging of surface mixtures. Based on the completed laboratory test results, the preliminary observations of the cracking tests evaluated are as follows:
The Energy Ratio (ER) method has several significant shortcomings. In its current procedure, it is not possible to properly characterize the variability of the ER parameter. The equipment cost and test complexity also render it impractical for routine use. The test results do not appear to properly separate the surface mixture with a substantial amount of cracking in Section N8 from the other surface mixtures that have had no signs of cracking. Although the field results are limited, and results of aged mixtures are yet to be reported, the ER method does not seem suitable for specification use in routine practice.
The overlay test (OT) results (both the Texas method and the NCAT-modified method) ranked the mixtures largely in accordance with their anticipated level of field cracking. Results of the two test methods were highly correlated. Both methods predicted that the surface mixture in Section N8 would be the most susceptible to cracking, as was confirmed in the field. However, one of the disadvantages of the OT methods is their relatively high variability. For the results of this study, the pooled coefficient of variation for the Texas method was approximately 45%, and for the NCAT-modified test, it was approximately 35%. These are similar to COVs reported in the literature for these methods. This diminishes the power of the test to distinguish mixtures with significant differences in composition. Furthermore, higher equipment costs and longer time to complete the tests are substantial disadvantages. On the other hand, both OT tests appear to appropriately rank the mixtures with different density levels. The mixture with a higher density level had higher cycles to failure than the control mixture with a lower density level.
The semi-circular bend (SCB) and Jc criteria (Louisiana method) were able to identify the surface mixture in Section N8 as susceptible to cracking, but also indicated very similar results for four of the other mixtures, two of which have had no signs of cracking to date. More field performance data are needed to judge the validity of the Jc criteria. Despite the relatively large number of specimens needed to obtain the Jc parameter, the test can be completed within a few days. Like the ER parameter, a disadvantage of the SCB method is the inability to assess variability of the Jc parameter with traditional statistical analyses.
The Illinois Flexibility Index Test (I-FIT) yielded a relatively large spread of Flexibility Index (FI) results for the seven mixtures. This kind of statistical spread in results for different mixtures would allow users to better assess how to improve mix designs and adjust field mixtures. The FI results indicated that the surface mixture in Section N8 was the most susceptible to cracking, as was confirmed on the track. Based on a similar calculation method, the indirect tensile asphalt cracking test (IDEAL-CT) data showed the same trends as the I-FIT data in most respects. More field performance data are needed to better judge the validity of the test and potentially set criteria for specification use. One concern with both the I-FIT and IDEAL-CT methods is the impact of specimen density. Counter to the expected outcome, higher density specimens have lower FI and CTIndex results than lower density specimens. The results of the I-FIT, IDEAL-CT and the two OT methods were strongly correlated. The I-FIT and IDEAL-CT have the lowest equipment cost and fastest testing time of the six cracking tests in the experiment, but the IDEAL-CT offers faster specimen fabrication than the I-FIT, since no specimen saw-cutting is required.
Technicians tested core samples for cracking depth and location, and to determine each mix’s propensity for further cracking.
CCPR and stabilized base experiment
CCPR—a method of combining RAP with foamed or emulsified asphalt and additives in a central recycling plant without the application of heat—has been used for rehabilitating low- and medium-volume roadways. To determine the viability of this technology for high-volume roadways, the Virginia DOT sponsored three test sections, complementing an existing project on I-81 in Virginia, to evaluate field performance of CCPR material and characterize its structural contribution. Sections N3 and N4 were designed to evaluate the difference between 6 in. and 4 in. of asphalt built on top of the same underlying layers, including 5 in. of CCPR material and 6 in. of aggregate base. Sections N4 and S12 were designed to evaluate the difference between underlying base material (6 in. of aggregate base vs. 8 in. of cement-stabilized base [CSB ]) in supporting the same upper layers, including 4 in. of HMA and 5 in. of CCPR material.
After 20 million ESALs, all three sections (N3, N4 and S12) have performed well with no cracking, minimal rutting, and no appreciable change in ride quality. Structural evaluations showed that CCPR material responds to temperature changes like conventional mix, which makes it appropriate to model CCPR material as a bituminous material in a mechanistic design. Compared to N3 and N4, the back-calculated AC/CCPR moduli in S12 had less temperature sensitivity and higher moduli, probably due to the back-calculation process attributing some of the CSB properties to the AC/CCPR layer. S12 also showed an increase in temperature-normalized modulus over time, possibly due to the CSB curing.
N3, with an additional 2 in. of AC, had lower strain levels than N4, and the CSB in S12 yielded much lower strain magnitudes and less temperature sensitivity. Strains normalized to 68°F showed that N3 and N4 had an increasing trend over time, while S12 was relatively constant. Thus, using a stabilized base may help control tensile strains and help eliminate bottom-up fatigue cracking.
Based on perpetual strain analysis, Section S12 with the CSB layer is expected to be perpetual as its strain distribution is less than the threshold distribution, while Sections N3 and N4 are expected to have bottom-up cracking in the future as its strain distribution exceeds the threshold distribution. Sections N4 and S12 remained in place for another research cycle to validate the assumption and criteria used in this analysis.
Findings from 11 other experiments from the 2015-2018 Test Track cycle are detailed in the Phase IV report available on the NCAT website: www.ncat.us.