Practice makes perfect

Preparing the AMPT tests for routine use


The current practice of design and construction of asphalt mixtures is mostly based upon volumetric and material properties. Over the past two decades, technologists have been trying to shift the design and QC/QA procedures of asphalt pavements from component measurements towards more performance-based measures.

The philosophy behind a performance-based design is to establish criteria based upon expected output of the entire pavement construction process; in contrast to a volumetric-based design, which is mostly based on properties of materials that are put into the mix. For example, performance-based tests can show the potential rutting of a mixture as compared to just achieving four percent air voids in a volumetric design.

To estimate the overall performance of a mixture in the field at an agency level, a practical yet simple tool was needed for mass implementation. In the late 1990s, research studies investigated a test method as a simple tool to capture the performance of the asphalt mixtures with regard to rutting and fatigue cracking. As the result of these projects, the dynamic modulus and flow number tests were recommended for incorporation in the Superpave mix design.

These simple performance tests were further evaluated and refined through NCHRP Project 9-29, and a testing tool was developed which is today known as the Asphalt Mixture Performance Tester, or AMPT. This test device was designed to be operable at a relatively low cost, and without extensive technician training.

The AMPT dynamic modulus test has been utilized in numerous research studies and has proven to be sensitive to the mix properties. It can also be well correlated to the mixture’s field performance; however, this high sensitivity comes at the cost of reproducibility. If the AMPT test specimens are not fabricated exactly alike, the dynamic modulus test detects the difference and produces different results. An inter-laboratory study was conducted as a part of the NCHRP 9-29 project concluded that factors relative to preparation and storage of the AMPT specimens are a major source of variability in test results.

This may not seem to be an important issue at the first glance but from a practitioners’ perspective, a test is only useful if different laboratories can reproduce it especially when pay factors are considered. For instance, when a contractor and an agency perform the same test on the same material in two different labs, they should be able to obtain similar values.

To improve the repeatability and reproducibility of the AMPT tests, the Asphalt Institute and the Federal Highway Administration embarked upon several cooperative research projects over the past few years. The primary objective of these projects was not to refine the test methods. Instead, the research focused on the effects that sample preparation and storage could have on the test data.

One of these projects examined the effect of various mixing temperature, conditioning temperature, and conditioning time on dynamic modulus test data for modified and neat binders. This study concluded that neat binders are more influenced by changing these variables compared to the modified binders. Additionally, this research revealed that conditioning time and temperature have a highly significant impact on stiffening the specimen and consequently, dynamic modulus data.

In another study, the Asphalt Institute laboratory evaluated the effect of specimen storage at ambient laboratory temperature. This project’s goal was to figure out if leaving the specimens on a shelf for a longer duration would influence the dynamic modulus and flow number. The impetus for this study was the understanding that labs could make test specimens relatively quickly, but could take a longer time to actually conduct performance-related tests in the AMPT.

For this study, the fabricated specimens were stored at room temperature for various durations up to 12 weeks. The results were interesting. Asphalt Institute found that the specimens became stiffer as they were stored for a longer time, and the dynamic modulus increased. This increase in dynamic modulus was more rapid during the first 3 weeks, and became negligible after 4 weeks. This could be due to oxidative aging or steric hardening of asphalt. Nonetheless, such a stiffening effect exists, and its magnitude is different from one asphalt binder to another. The results indicated that neat binders are more prone to stiffening during storage as compared to polymer binders.

Some other factors relative to specimen storage were evaluated in this research. The time of trimming the AMPT specimen from the gyratory compacted sample was one of these factors. It seemed that when the specimen was trimmed soon after compaction, the initial stiffening phase passed faster. In other words, if the test is to be run after passing the initial stiffening phase, it is better to trim the samples as soon as possible after compaction, maybe within two days. Furthermore, storing the specimens in zip-top bags did not seem to have an effect on the dynamic modulus and flow number test data.

Most of the technicians and researchers who have been using the AMPT follow the same practice for preparing cylindrical samples that is standardized under AASHTO PP 60 and AASHTO T 312. However, as mentioned earlier, the repeatability and reproducibility of the test is not where we need it yet.

For example, to heat the asphalt sample to the mixing temperature, the current standard says: “place the aggregate and binder container in the oven, and heat them to the required mixing temperature.” According to this procedure, a binder container can be left in the oven for several days, which might excessively age small asphalt samples and affect the modulus test results. In practice, most know to use asphalt binder when it reaches temperature and to not reheat used samples.

Issues like this sparked the idea of another cooperative research project between the Asphalt Institute, FHWA and Advanced Asphalt Technologies. In this study, 12 different variables of sample preparation are being evaluated.

One of the factors that was expected to affect the AMPT test results was the conditioning temperature of the loose asphalt mix before compaction. This conditioning temperature has a high impact on asphalt aging, and the resulting mixture modulus. The Asphalt Institute laboratory ran a detailed experimental study on different forced-draft ovens to see if different ovens have different ways of aging the mix. Various conditioning scenarios with multiple sample production and mix stirring were included in this evaluation. The data showed a surprisingly large difference between various ovens. It was clear that consistent sample conditioning could only be achieved with a high-quality, forced-draft oven. High-quality ovens recover faster after opening the door and move air across the sample.

The lower-quality ovens that are mainly designed for heating and drying purposes in a laboratory do not condition the samples at a uniform temperature and can struggle to maintain temperature after the door is opened for stirring. In such ovens, there is a significant difference between the temperatures throughout the oven and within the loose mixture sample in a single pan. In fact, conditioning of loose mix samples should be thought of as a standard practice of changing the material properties to a more aged state; which requires not only heating the mix for a period of time at certain oven temperature, but using a competent oven as well.

As this study continues, the research team is looking into more variables such as using a planetary mixer or bucket mixer, gyratory sample’s compaction height, mixing and conditioning temperatures, air voids tolerance of the specimen, stirring the loose mix during conditioning and conditioning duration to name a few. Hopefully this study will recommend refinements to the preparation procedures for AMPT samples and improve repeatability and reproducibility of performance-based, mechanical tests.

Acknowledgements: This research is based upon work supported by the U.S. Department of Transportation under Cooperative Agreements DTFH61-08-H-00030 and DTFH61-11-H-00033. Thanks to John Bukowski (FHWA) and Michael Arasteh (FHWA) for their support.

Alireza Zeinali is a PhD graduate of the University of Kentucky who collaborated with Asphalt Institute on research studies.