As part of the Virginia Department of Transportation’s efforts to implement Balanced Mix Design, the indirect tensile (IDT) cracking test at intermediate temperature (formerly called the IDEAL-CT test) is required to assess the cracking susceptibility of mixtures during mix design and production. As a relatively new test method, there is no precision statement available to date, leaving test operators unable to assess their test results’ repeatability and reproducibility.
In light of this issue, Virginia Transportation Research Council (VTRC) and Virginia Asphalt Association (VAA) began collaborating in 2019 to develop a round robin testing program for the IDT cracking test to generate data to create precision statements. Phase I of the program was initiated in early 2020, and two mix designs having two significantly different CTindex results were selected for use in the study. These designs were used to fabricate sets of test specimens shipped to participants in early May 2020.
A total of 41 laboratories participated in the round robin. These labs consisted of VDOT, contractor, other DOT, and independent testing labs. Several labs received more than one set of test specimens per mix to perform testing using machines or load frames from different manufacturers. 46 sets of test specimens for each mixture were distributed, and seven devices from different manufacturers were evaluated.
Participants were asked to report detailed test results, including specimen measurements and bulk specific gravities, equipment manufacturer and model, loading mode (screw-drive or servo-
hydraulic), and CTindex values. Perhaps most importantly, participants were asked to submit the raw data files collected by their equipment in addition to reporting CTindex values for each specimen. These data files were used to perform quality checks on the data before further analysis. The quality checks revealed that only sixteen of the forty-six sets of test results fully met the test requirements; an additional fourteen sets of data met all requirements except for the loading rate.
ASTM D8225-19 Determination of Cracking Tolerance Index of Asphalt Mixture Using the Indirect Tensile Cracking Test at Intermediate Temperature sets forth the requirements for conducting the IDT cracking test. In summary, the specification requires specimens to be 150mm ± 2mm diameter cylinders having a height of 62 ± 1mm. VDOT specifies specimen air void contents of 7% ± 0.5% and test temperature of 25ºC. Test specimens are placed in the test fixture, and a constant load is applied to the specimen at a rate of 50 ± 2.0 mm/min, with no seating load applied before the test. The test is stopped when the load drops below 0.1 kN after reaching the peak load. Some key details are essential when evaluating data quality:
No seating load is applied,
The load is applied at the specified constant load rate of 50 ± 2 mm/min, and
The test ends only when the applied load drops to 0.1 kN or less after reaching the peak load.
An acceptable load-displacement curve will look like the example in Figure 1(a), and an acceptable displacement-time curve will look like that shown in Figure 1(b). These curves should be checked for each specimen as part of the test quality control.
There are many reasons that the load-displacement and displacement-time curves may not meet the test specification. Some load frames have a safety function limiting the ram travel that does not provide enough travel for the post-peak load to drop to 0.1kN or less during testing. Machine compliance may cause the rate of loading to change with increasing specimen resistance to loading. Improper LVDT setup, including misalignment, improper zeroing, or incorrect calibration, will affect displacement measurements. Performing data quality checks will permit the identification of issues and allow them to be addressed.
The time, load, and displacement measurements must be obtained from the testing software to identify non-compliant data. The operator manual or equipment manufacturer may have to be consulted to determine the best way to extract the raw data file. The time, load, and displacement measurements can then be plotted using a spreadsheet, and data quality can be assessed. If non-compliant data is found, the causes should be determined so they can be addressed and resolved. A single occurrence of non-compliance may indicate an isolated or random event; however, recurring events may be a sign of equipment issues, repetitive operator error, or other testing problems.
Some examples of non-compliant data are shown in Figure 2, along with possible causes. Figure 2(a) shows data from a test wherein the data acquisition has been incorrectly set up, as the measured load should reach approximately 10kN at its peak. The difference suggests that the load data shown are in US units (lbf) instead of kN as labeled, while displacements are shown correctly in SI units (mm); the mixed units will result in incorrect calculations of the CTindex. In addition, the test terminated before the load dropped to 0.1kN. A test with an error in displacement measurements is shown in Figure 2(b); the LVDT may have slipped out of position, or the range may need to be checked. Figure 2(c) displays a seating load applied at the beginning of the test and an LVDT error towards the test’s end. The test software should be configured to remove the seating load, and the LVDT installation and range should be checked. Another issue with the displacement measurement is shown in Figure 2(d). It appears that the LVDT may not have been installed or zeroed/initialized properly so that measurements were not collected as the test started. Figure 2(e) presents a load-displacement curve that complies with test requirements, although the test was performed using a non-linear loading rate. This demonstrates why evaluating both the load-displacement and displacement-time data are important. If the loading rate is non-compliant, the equipment does not meet the test requirements and may need troubleshooting or maintenance. While small deviations from 50±2 mm/min have not appeared to significantly affect calculated CTindex values for data collected to date, more work is necessary to validate the acceptable range of loading rates. Ongoing data quality checks are essential to implement with the use of performance testing. Non-compliance in test data can lead to incorrect CTindex values that do not describe the actual material performance—resulting in unnecessary redesigns or rejected materials. The next step in addressing overall data quality is checking that test results fall within acceptable precision estimates, but that is another discussion.