May 10, 2005

The old with the New

University of New Hampshire research shows effects of RAP in Superpave

Many states currently allow the use of some proportion of reclaimed asphalt pavement (RAP) in hot-mix asphalt (HMA). The amount of RAP allowed varies depending upon individual agency policies and practices that have been developed through application of national guidelines. An agency’s previous experience with RAP materials also is a consideration. NCHRP project 9-12 addressed the use of RAP in the Superpave mix design system, providing guidance on how to adjust virgin binder grades to balance the stiffer RAP binder that is introduced into the mixture. This research was conducted using the original Superpave performance tests. The new simple performance test for Superpave mix design is a dynamic modulus test. In addition, the new mechanistic-empirical (M-E) design guide requires dynamic modulus as an input. As the M-E design guide is implemented and as states start using the simple performance test as part of the Superpave mix design procedure, it will become increasingly important to evaluate the effect of RAP on the dynamic modulus of mixtures. This article summarizes work that has been conducted at the University of New Hampshire on RAP mixtures.
The addition of RAP to an asphalt mixture changes the mechanistic properties (i.e., strength, durability) of the mixture and affects its performance (i.e., resistance to cracking and deformation) in the field. The objective of the research was to determine the effect of the addition of RAP on the dynamic modulus of the mixture. As the research progressed, it became apparent that the change in volumetric properties with the addition of RAP also is important. The mechanistic properties change as a result of the aged binder introduced to the mixture as part of the RAP. The RAP binder will have a different chemical composition and different properties from the virgin binder added during the mixing process. These two binders mix to some extent, changing the properties of the mixture containing RAP from one that contains only virgin material. The extent of blending of the binders also will affect the volumetrics of the mix; the RAP particles that do not break down in the mixing process will act like black rock and change the effective gradation of the mixture. To date, two RAP sources have been used to study the change in volumetric properties of the mixtures, and one RAP source has been used for dynamic modulus testing. A control mixture containing only virgin materials (0% RAP) was tested along with mixtures containing 15%, 25% and 40% RAP.

Effect of RAP on volumetrics

A 19-mm Superpave gradation designed for low-volume roads was used with an unmodified PG 58-28 binder. The two types of RAP used in this study are a processed RAP (with some amount of portland cement concrete) and an unprocessed RAP (straight millings), hereinafter referred to as grindings. The processed RAP has an asphalt content of 3.6% with an extracted binder grade of PG 94-14. The grindings have an asphalt content of 4.9% with an extracted binder grade of PG 82-22.
The mixes were designed based on a New Hampshire DOT-approved 19-mm Superpave mix containing 15% RAP. The control, 25% RAP and 40% RAP were designed to achieve an overall mixture gradation similar to the original 15% RAP gradation. The relative proportions of blast rock and sand stockpiles were held constant for the different mixtures to maintain the same relative structure (particle angularity, type of material) for the virgin material in the mixture. For the processed RAP mixtures, the increasing percentages of RAP caused the gradations to become finer in the No. 8 to No. 50 size range. The grindings mixtures had slightly finer gradations at the No. 30 sieve size, but were essentially the same as the control mix. The same mixing and compaction temperatures, based on the virgin binder, were used for all of the mixtures.
The volumetric properties of the processed and grindings RAP mixtures are shown in Table 1. For the processed RAP mixtures, the VMA and VFA values for the 25% and 40% RAP mixtures were higher than those for the control and 15% mixtures. For the grindings RAP mixtures, the VMA values increased with RAP percentage and the VFA values for all of the RAP mixtures were higher than the control mix. It is hypothesized that this difference is due to the extent of blending of the RAP material with the virgin materials.
As part of the mixing procedure, the RAP was preheated in the oven for a period of two hours prior to mixing with the virgin asphalt and binder. This is the procedure that the New Hampshire DOT uses to simulate plant operations. If the RAP material is not heated sufficiently, the RAP binder does not blend with the virgin binder to the extent possible and the RAP then tends to act more like a black rock material. The RAP particles have a coarser gradation than the RAP aggregate. Therefore, if the RAP particles do not completely break down and blend with the virgin materials the overall mixture gradation will be coarser and, with the same compaction effort, an increase in VMA is expected.
To test whether heating time had an effect on the mixture volumetrics, several specimens with the 40% proc-essed RAP were fabricated by heating the RAP for two hours, 3.5 hours, and eight hours at the mixing temperature. The two-hour time is the standard procedure, the 3.5-hour time was the time required for the RAP to reach mixing temperature, and eight hours is equivalent to the aggregate heating time (usually overnight). The same compaction effort was used in fabricating all of the specimens.
The %VMA decreased by 0.5 when the heating time increased from 2 to 3.5 hours, and then increased by almost three with the longer heating time. At the shorter heating time, the RAP was not heated enough to allow the RAP particles to break up into smaller pieces and blend with the virgin materials. At the longer heating time, the RAP has aged further and fewer of the RAP particles break down and blend with the virgin material. The shorter and longer heating times create a situation where more of the RAP acts like a black rock. This indicates that there is an optimum heating time for the RAP material to allow for the greatest extent of blending between the virgin and RAP materials. The different heating times may affect the mixture design and design asphalt content. A RAP mixture may not meet the Superpave VMA requirements when the RAP is heated for a particular amount of time, but may meet the requirements if the RAP is heated for a different amount of time. Therefore, it is very important that the laboratory procedures for producing RAP mixture simulate the plant operations as closely as possible. More research is needed in this area, as the way RAP is handled in the lab can significantly affect the mix design and therefore the performance of the mix in the field.

Effect of RAP on dynamic modulus

Specimens 100 mm in diameter and 150 mm tall were fabricated for the compression tests, and specimens 75 mm in diameter and 150 mm tall were fabricated for the tension tests. These are the specimen geometries currently recommended for the simple performance test and used in constitutive modeling of asphalt concrete in tension and compression. The dynamic modulus test measures the response of the material to cyclic loading at different frequencies (usually ranging from 0.1-30 Hz) in the undamaged state. Asphalt concrete is a viscoelastic material, meaning that its response to a particular load depends on the magnitude of the load, the rate of application and the duration of the load. Therefore, it is important to evaluate how the material responds to different frequencies or rates of loading, which correspond to the different traffic speeds a pavement could experience in the field. The dynamic modulus at each frequency is calculated by dividing the steady state stress amplitude by the strain amplitude. The dynamic modulus values measured over a range of frequencies and temperatures (-10°C, 0°C, 10°C, 20°C and 30°C) are used to create a master curve, which can be shifted to different temperatures.

The addition of 15% RAP increases the stiffness over the control mix in both cases; however, that increase was greater in compression than in tension. At lower frequencies, the mixtures had similar stiffness in both tension and compression. The difference in stiffness between the two mixtures increased with frequency in compression and appears to reach a constant difference after a point in tension. The dynamic modulus curves for the 25% RAP and 40% RAP mixtures were similar to that for the control mixture, when it was expected that they would have higher dynamic modulus values than the 15% RAP mixture due to the greater amount of stiff RAP binder in the mix.
There are several possible explanations for the unexpected results. First, the 25% RAP mixture had a higher asphalt content than the 15% RAP mix. Mixtures with higher asphalt content have been shown to have lower dynamic modulus values. Secondly, the gradations for the 25% and 40% processed RAP mixtures were finer, especially in the 0.3-mm to 2.36-mm particle size range, and finer gradations typically have lower stiffness. The gradations among the different mixtures in this project were allowed to vary to maintain the same relative percentages of blast rock and sand for the virgin aggregate. This was done to maintain consistent aggregate properties (aggregate angularity and specific gravity) for the virgin material in each mixture. A study of the fine aggregate angularity of the different RAP mixtures showed that all of the combined gradation values were above the Superpave minimum value of 40%. The effect of the aggregate angularity is expected to be more pronounced in compression testing, where the aggregate structure has a greater contribution to the overall mixture stiffness. The dynamic modulus master curves did not show much difference between tensile and compressive testing, indicating that the differences in fine aggregate angularity between the different mixtures with varying percentages of RAP were not significant enough to affect the measured values. A Florida study found that there was no correlation between fine aggregate shear strength and fine aggregate angularity and that fine aggregate angularity alone did not reflect the rutting performance measured in the lab.
It is hypothesized that the lower-than-expected dynamic modulus curve for the 25% RAP mixture was due to a combination of higher asphalt content, finer gradation and higher VMA and VFA. The 40% RAP mixture likely exhibited lower dynamic modulus due to the finer gradation and higher VMA and VFA. This follows the findings of the Florida study where mixes with higher VMA exhibited lower stiffness.

Soft numbers

The purpose of this study was to examine the effect of RAP on the volumetric properties and stiffness of HMA. The addition of RAP increased the VMA and VFA of the mixtures. Also, the study indicated that there is an optimal preheating time for RAP to allow the particles to soften, break down and blend with the virgin materials. Further research in this area is needed to determine how best to simulate the plant operations in the lab, especially for mix design.
The addition of 15% RAP increased the dynamic modulus of the mixture, as would be expected. However, the mixtures containing 25% and 40% RAP did not follow the expected trend of increasing stiffness. Instead, the dynamic modulus values were similar to those for the control mixture. It is likely that a combination of variables is influencing the material behavior for these mixtures: the 25% RAP mixture had higher asphalt content and a finer gradation, and the 40% RAP mixture had a gradation finer than that for the 25% RAP mixture. The 25% and 40% RAP mixtures also had higher VMA and VFA values than those for the control and 15% mixtures. All of these effects will tend to soften the mixture, decreasing the dynamic modulus. More research is needed to determine if similar trends are seen with various RAP sources and virgin binders.

About the author

Daniel is an assistant professor, Department of Civil Engineering, at the University of New Hampshire, Durham, N.H.