Asphalt Binder Cracking Device to Reduce Low-Temperature Asphalt Pavement Cracking
NO-TRIM ABCD EXPERIMENT
As an attempt to improve precision of ABCD testing, a no-trim ABCD test procedure was developed. The test procedure used in the ABCD ILS consists of seven simple steps:
- Lubricating the ABCD rings and silicone rubber molds.
- Pouring the sample into the ABCD ring-mold assemblies.
- Trimming excess asphalt on the ABCD ring-mold assemblies.
- Twisting the ABCD rings to break the bond at the binder/ABCD ring interface.
- Connecting the ABCD rings to a data acquisition system.
- Starting the ABCD test software and completing the test.
- Cleaning the ABCD rings and the silicone molds.
The only step that requires some degree of skill to obtain high-quality data is the trimming process. Trimming disturbs the specimen and if done carelessly may pull asphalt binder off of the ABCD ring surface; pull the asphalt binder upward, causing excessive trimming; and damage the silicone rubber mold.
No-Trim ABCD tests were performed by pouring the exact volume of asphalt binder to form the test specimen. The exact volume of an ABCD specimen is 14.38 cm3 (0.88 in3), the target volume of asphalt binder at 25 °C (77 °F). When the 200 με/°C CTE is assumed, the volume of asphalt binder at 170 °C (338 °F) pouring temperature becomes 15.63 cm3 (0.95 in3). For the No-Trim ABCD experiments, the depth of annulus space between the silicone rubber mold and the ABCD ring was increased from 12.7 mm (0.50 in.) to 13.71 mm (0.54 in.) to accommodate the expanded volume of asphalt binder at the pouring temperature. Several designs of pouring device were tried with varying levels of success. Due to the high viscosity of asphalt binder, any pouring device relying on gravitational force could not control volume accurately and efficiently. A precision-made, syringe-type, pouring device is needed and under development.
The No-Trim ABCD experiment proceeded without the syringe-type pouring device. Instead, it was decided to control the mass of asphalt binder. The specific gravity of asphalt binder was assumed as a value between 1.01 and 1.03. Then, the mass of asphalt binder occupying 14.38 cm3 (0.88 in3) volume would range from 14.5 g to 14.8 g (from 0.51 oz to 0.52 oz). The mass of trimmed asphalt specimens for ABCD testing was about 14.0 g, significantly less than the theoretical 14.5-14.8 g (0.51-0.52 oz) due to the demolding agent (glycerin-talc mixture) and possible over-trimming. Different operators would use different amounts of demolding agent, resulting in different amounts of asphalt binder being used to form the ABCD test specimens. These differences could also contribute to the variability of ABCD test results. In the No-Trim ABCD experiment, the ABCD ring and mold assembly was placed on an electrical balance and the heated asphalt binder was carefully poured while the mass of the sample was closely monitored. The target mass for the No-Trim experiment was between 14 g and 16 g (0.49 oz and 0.56 oz). A few samples were intentionally prepared with 7-8 g (0.25-0.28 oz) or 16-17 g (0.56-0.60 oz) of asphalt binder. When the binder sample was poured while it was sufficiently warm, the top surface of the resulting ABCD specimen was always clean and uniform. Post-test visual inspection of the ABCD specimen also showed that the test specimens did not contain entrapped air bubbles or cold joints if they were prepared with sufficiently warm asphalt binder.
Three asphalt binders (unaged AAA-1, unaged AAB-1, and unaged AAF-1) used in this No-Trim experiment had been previously tested following the current trimming procedure for 20, 5, and 8 times, respectively. Their masses were not determined in the trimmed tests. In addition to the Trim versus No-Trim comparison, the effects of the ring type (open versus covered), rotation (rotation versus no rotation), and mold lubrication (lubed mold versus not-lubed mold) on ABCD test results were also determined.
The current ABCD ring has a bottom and a top plastic cover protecting the inside sensors and allowing easy handling. However, the covers influence the strain readings during the ABCD tests due to greatly different CTEs between the plastic covers and the ABCD rings. This cover-ring interaction seemed to cause the relatively large variability of the strain-jump magnitude.
"Rotation" refers to twisting the ABCD ring before the start of the test to break bonds be-tween the ABCD ring and the asphalt binder specimen. To avoid causing unwanted specimen deformation, the rotation is usually done after about an hour, when the test specimen is sufficiently cooled. If the rotation is not needed in the No-Trim ABCD test, the binder specimens can be tested right after pouring without waiting an hour, thus saving time.
In the current test procedure, the ABCD mold is lubricated with a glycerin-talc mixture before the binder sample is poured. However, as pointed out earlier, the mold lubrication might be a variable affecting ABCD test results by displacing different amounts of the asphalt binder forming the ABCD test specimen. The trimming also leaves smeared asphalt binder on the top surface of the mold, and cleaning with soap water is always necessary. If the lubrication is not needed, cleaning the silicone rubber molds after the test may not be needed. The elimination of lubrication saves testing time and potentially improves the precision of ABCD test by improving control of the volume of asphalt binder specimen.
A total of 111 specimens were tested following the No-Trim ABCD test procedure; the results are presented in the appendix and plotted in figures 7-9. For unaged AAA-1 binder with a specimen mass between 13.5 g and 16 g (0.48-0.56 oz), the range of No-Trim ABCD cracking temperature was almost identical to the range of 20 trimmed ABCD data previously obtained following the current test procedure. No-Trim data were obtained from tests performed with various procedures that included variations in rotation, ring type, and lubrication of mold. A similar trend was also observed for unaged AAB-1 and unaged AAF-1 binders. The ranges of both trimmed and No-Trim ABCD cracking temperatures were about 4 °C (7 °F), except for the trimmed cracking temperature for unaged AAB-1, probably due to the small number of data. For statistical analysis, 10 data obtained from specimens with less than 13 g (0.46 oz) or more than 16 g (0.56) mass were excluded. The similarity of the ABCD cracking temperatures of trimmed and No-Trim samples of unaged AAA-1 is also shown in table 12. Their means and standard deviations are very similar.
|Trim||N||Mean||Standard Deviation||Standard Error Mean|
|No-Trim||37||-38.12 °C||1.43 °C||0.23|
|Trimmed||20||-38.25 °C||1.37 °C||0.31|
Regression analyses were performed to identify factors significantly affecting the ABCD cracking temperature (table 13) and strain jump (table 14). For the 5 percent significance level (p value < 0.05), the effects of rotation, ring type, and mold lubrication on the cracking tem-perature were not significant. The only significant factor was the mass of sample. If the mass of sample increases by 1.0 g (0.04 oz), the cracking temperature would decrease by 0.62 °C (1.12 °F). For the strain jump, the ring type, mold lubrication, and sample mass did not have a significant effect. The only significant factor affecting the strain jump was rotation of the ABCD ring. The ABCD test with ring rotation would reduce the strain jump by 7.6 µe, probably due to reduced adhesion between the ABCD ring and the binder sample. When the No-Trim data were separately analyzed, the standard deviation of ABCD strain jump was 10.84 με for tests without rotation and 4.40 με for tests with rotation. Rotation of the ABCD ring significantly reduced the variability of the strain jump and was an essential step to be kept in the ABCD test procedure.
|Analysis of Variance|
|Sum of Squares||df||Mean Square||F||p-value|
|R Square = 0.932; Standard Error of the Estimate = 1.18 °C|
|Analysis of Variance|
|Sum of Squares||df||Mean Square||F||p-value|
|R Square = 0.203; Standard Error of the Estimate = 8.46 με|
Based on this experiment, a revised ABCD test procedure is recommended, which has two major changes from the current version:
- The specimen trimming step is eliminated. Instead, the exact volume (14.38 cm3 [0.88 in3] at 25 °C [77 °F] or 15.63 cm3 [0.95 in3] at the pouring temperature) of binder sample is poured.
- The silicone rubber mold is not lubricated with glycerin-talc mixture. The surface of the ABCD ring is still lubricated to facilitate the rotation of the ring prior to the test.
The elimination of the mold lubrication and trimming steps also eliminated the need for clean-ing the silicone mold with soap and water. As shown in figure 10, the silicone molds remained very clean after the test and were immediately ready for the next test without the cleaning step.
Among the No-Trim data, there were 14 results from tests following the revised test proce-dure with covered ABCD rings (excluding data from no rotation, lubrication of mold, and open ring). For these data, the standard deviations (pooled among three binder types) for the cracking temperature and the strain jump were 0.96 °C (1.73 °F) and 3.52 με (or 0.55 MPa [79.8 psi] for fracture strength), respectively. To compare the standard deviations obtained from the ABCD ILS in table 6 (the current procedure), the single-operator precision for the cracking temperature of the revised procedure would be about the same (0.96 °C versus 0.95 °C [1.73 °F versus 1.71 °F]), and the precision for the strain jump would be improved (3.52 με versus 5.48 με). For the multilaboratory testing of the revised procedure, precision of both the cracking temperature and the strain jump would be improved greatly from 1.36 °C (2.45 °F) and 7.21 με (shown in table 6) to close to 0.96 °C (1.73 °F) and 3.52 με because of the simplified test procedure and the elimination of a couple of steps that required careful execution by operators. The average and the standard deviation of No-Trim sample mass were 14.56 g (0.51 oz) and 0.35 g (0.01 oz), respectively. With a proper pouring device, the standard deviation of the sample mass can be reduced significantly and the precision of the ABCD test may be further improved for both single-operator and multilaboratory testing.