Characterization of hot mix asphalt containing post-consumer recycled asphalt shingles and fractionated recycled asphalt pavement

 

Transportation agencies are increasingly investigating new technologies that will reduce the cost of asphalt pavement materials while maximizing long-term performance.  According to the American Society of Civil Engineer’s (ASCE) 2009 Infrastructure Report Card, $186 billion is needed annually for rehabilitation and maintenance of United State roadways, but only $70.6 billion is spent annually.  Since 96%t of hard surfaced roadways in the United States are paved with asphalt, there is a strong need for lowering the cost of asphalt pavements that also meet superior performance standards of being safe, smooth and structurally capable of supporting heavy traffic loads.  The need for well performing asphalt pavements together with the rising prices of liquid asphalt and the scarcity of quality aggregates have placed additional pressure on agencies and owners to create effective economic solutions.

The cost of asphalt materials can be reduced by replacing the new (virgin) asphalt cement and mineral aggregates with recycled products derived from construction waste or byproducts that contain asphalt mix components.  Using recycled products saves not only on the cost of asphalt materials, but also on the amount of construction waste since it is not being placed in landfills.  Recycling products into asphalt pavement also means less energy is needed to produce the pavement, making it a more sustainable product that minimizes its impact on the environment. Furthermore, properly designed asphalt mixes that contain recycled products can exhibit no performance differences or even improved performance for certain applications compared to typical mixes (Al-Qadi et al. 2007).

The most common source for secondary materials comes from reclaimed asphalt pavement (RAP).  RAP is old pavement that has been milled from the roadway, crushed into smaller aggregates sizes, and stockpiled.  At the end of an asphalt pavement’s service life, the pavement is still valuable since it contains mineral aggregates and asphalt cement that can be reheated and reincorporated with new hot mix asphalt (HMA).  The Federal Highway Administration (FHWA) and the Environmental Protection Agency (EPA) report that of the 100.1 million tons of asphalt pavement removed each year, 80.3 million tons is reused as part of new roads, roadbeds, shoulders and embankments, making asphalt America’s number one recycled material (FHWA 1993).

Most transportation agencies have a construction specification in place that allows asphalt producers to add RAP to HMA, but only up to a certain percentage, typically 25%.  By increasing the amount of RAP usage, the cost of asphalt pavement can be even further reduced. Adding higher amounts of RAP to HMA reduces the amount of control engineers have when combining different crushed aggregate sizes to formulate a well-performing mixture.  Since a typical RAP stockpile contains aggregate particles of varying sizes and binder contents, increasing the RAP content in pavements can increase the variability of the HMA end product (NCHRP 2001).  To maintain quality and consistency of mixes when increasing the RAP percentage, RAP can be fractionated into stockpiles of different sizes similar to the processing of virgin aggregates. Using fractionated reclaimed asphalt pavement (FRAP) allows for an increase in the level of quality control during the construction process, which in turn allows higher RAP mixes to be produced more consistently.  Nationally, the concept of fractionating RAP is becoming recognized as an efficient way to lower the cost of a new mix and reduce the inconsistencies of the high RAP mix properties without sacrificing quality (Vavrik et al. 2008).  With this advancement in technology, research efforts of transportation agencies have focused on increasing RAP usage up to 50%.

Another source for secondary materials is recycled asphalt shingles (RAS).  Asphalt shingles, like RAP, also contain mineral aggregates and asphalt cement, making RAS a candidate for product replacement in HMA.  RAS comes from two different sources, post-manufactured shingles and post-consumer shingles.  Post-manufactured shingles are the waste products of the shingle manufacturing process, which include factory rejects and tab cut-outs, while post-consumer shingles are shingles that come directly from roofs of commercial and residential buildings after their service life including damage from severe weather.  Historically, the vast majority of research on RAS has focused on post-manufactured shingles since government engineers and regulators have traditionally accepted post-manufactured shingles over post-consumer shingles in the development of construction materials specifications and environmental regulations.  With more recent technological advances in processing asphalt shingles, research efforts are trending toward the utilization of post-consumer shingles.  A major factor driving this interest is that 10 million tons of post-consumer shingles are placed in landfills in the United States each year, while only 1 million tons of post-manufactured shingles are placed in landfills each year (FHWA and EPA 1993).  With this large pool of post-consumer shingle resource, there is significant potential for cost savings in mix constituents and landfill space.

Recycling manufactured shingle scrap has been occurring for the last 25 years due to the many applications of RAS as a construction material (Krivit and Associates 2007).  RAS has been used mostly as a secondary material for HMA in commercial and private pavements.  Recently, it has become more widely used in highway pavements by transportation agencies.

One transportation agency evaluating options that minimize construction costs and optimize the selection of materials used in its asphalt pavements is the Illinois Tollway.  The Tollway is a user-supported system of public roadways.  The Tollway receives no state or federal funding for maintenance and operation of its system, relying primarily on tolls paid by travelers on the roadway (Bentsen 2010).

Over the last few years, the Tollway has begun implementing an unprecedented rehabilitation/expansion program for its highway network.  Due to financial constraints, economic demands, and the need to improve as much of the network as possible, it faces many challenges as it continues to update its system of roadways.  It is important for the Tollway to look to new technologies for solutions that answer the economic and performance challenges it faces.

With more transportation agencies studying the options of adding RAS or using higher amounts of RAP through fractionation, the Tollway became interested in adopting these techniques in its construction specifications.  The Tollway selected Iowa State University to conduct research investigating the performance of asphalt pavements with RAS that contain higher amounts of fractionated RAP. The results of this research project are presented herein for this study.

Asphalt Shingles
Understanding the composition and properties of asphalt shingles is necessary for fully characterizing asphalt mixtures that incorporate their use. The American Society for Testing and Materials (ASTM) has specifications for their production. There are two different types of specifications, ASTM D225 which specifies asphalt shingles made with organic (cellulose or wood fiber) backing and ASTM D3462 which specifies asphalt shingles made with fiberglass backing. These specifications are fairly broad so the exact composition of shingles will vary among different manufacturers.  

Shingles are manufactured by saturating and coating both sides of organic or fiberglass backing felt with liquid asphalt. The asphalt used to coat the felt material is different than asphalt used in paving materials.  The asphalt used in roofing shingles is much harder and stiffer because the manufacturers use an “air-blown” process to increase the viscosity of the asphalt. The process infuses oxygen into the asphalt which changes the chemical make-up of the asphalt making it stiffer. The shingles are then covered with sand and crushed-stone granules to increase their durability and resistance to weathering.  

The percentages of the individual component materials in asphalt shingles are different in shingles manufactured with organic felt compared to shingles manufactured with fiberglass felt. Brock (2007) summarized the composition of each type of shingles and his data is shown in Table 1. The shingles manufactured with organic felt have substantially more liquid asphalt then shingles manufactured with fiberglass felt due to the different absorption of the materials. Since asphalt binder is the most valuable product in RAS for paving materials, RAS made from organic felt will have a higher economic value. RAS made from post-consumer shingles (indicated as “Old” in Table 1) also has higher asphalt contents than RAS made from post-manufactured shingles due to the loss of a portion of the surface granules from weathering. McGraw et al. (2007) found similar asphalt contents as Brock in post-manufactured shingles and post-consumer shingles after conducting extractions on multiple samples.

Table 1: Asphalt Shingle Composition (Brock 2007)

The other components used in the manufacturing of shingles are also a valuable commodity in HMA. The crushed-stone granules for example can reduce the amount of manufactured sand needed for an asphalt mixture. Next to asphalt binder though, the component of particular interest to researchers and HMA producers is the fibers that come from the felt backing. Fibers are used as an additive in asphalt with a gap-graded or open-graded aggregate structure to prevent drain-down of the asphalt binder.  Several studies conducted in the United States found that significant benefits can be gained from asphalt paving mixtures that incorporate fibers by increasing the tensile strength and toughness of the mixes (Newcomb et al. 1993).

Processing Roofing Waste
For shingles to be successfully used in asphalt paving mixtures they need to be shredded or ground down to relatively small particle sizes. Different types of crushers, including rotary shredders and hammer mills, are used to process the shingles. The 2006 AASHTO provisional standard on the use of RAS as an additive in HMA requires 100% of the RAS passes a 12.5 mm (0.5 inch) sieve. Some state agencies that have a construction specification in place for RAS require an even smaller maximum particle size by specifying 100% of the RAS passes either the 9.5 mm (0.375 inch) sieve or the 4.75 mm (No. 4) size.  

In order to maximize the benefits of RAS, past research has helped identify how the RAS gradation affects its performance in HMA. Research completed by Button et al. (1996) and Abdulshafi et al. (1997) found that a finer grind produced a more consistent and better performing mix. Button et al. (1996) also found that the mixes containing a finer ground post-consumer RAS increased the tensile strength more than a coarser grind.  

The size of the recycled asphalt shingle can also be expected to affect the fraction of shingle asphalt binder that contributes to the final blended binder (Krivit and Associates 2007). A smaller RAS particle will have a larger surface area and more exposed binder.  With more binder exposed on the surface of the RAS particle, more binder will be activated and fully blended with the virgin asphalt. Mix designs developed by the Iowa Department of Transportation (DOT) have revealed that not all of the RAS binder becomes activated in the asphalt mixture.  Their mix designs revealed that approximately two-thirds of RAS binder behaves as liquid when heated and contributes to the final binder blend. The other third behaves as an aggregate coated with asphalt.

As more recyclers gain experience processing shingles, better quality facilities and processes are being developed to solve the challenges faced during production. The continuing challenges in using RAS are found to be in the quality control and quality assurance of the final product along with identifying mix designs that meet the requirements of specifying agencies (Scholz 2010). When processing post-consumer shingles, construction debris must be removed from the shingle. This includes wood, nails and other contaminates. Usually manual labor is used to separate the shingle from the wood. Removal of nails and other material removal is accomplished by using magnets at different locations on plant conveyer belts before and after the crushing process.  

Asphalt Shingle Binder Properties
While using recycled products in HMA helps achieve a more economical asphalt pavement and lowers its impact on the environment, adding recycled products to HMA can also impact its performance due to the rheological behavior change of the final binder blend (NCHRP 2001). It has been well established that the rheological properties of the asphalt binder affect pavement performance (Roberts et al. 1996).  It is important to understand the behavior of asphalt binder to design and characterize HMA with RAS or higher amounts of RAP. Before understanding the implications of adding asphalt via recycled shingles to HMA, it is helpful to understand how RAP binder affects HMA properties since RAP has been researched extensively over the last 30 years and is commonly added to HMA.

When RAP is added to HMA, it contributes aggregate and asphalt binder to the final mixture. The final blend of recycled materials and virgin materials needs to meet certain physical properties for design and construction specifications. Asphalt binder from RAP is stiffer than virgin asphalt because during the construction and service life of the pavement from which the RAP came, the binder aged and hardened. If the asphalt binder from the RAP is very stiff or if higher amounts are added (more than 20%), the influence of the RAP binder can have a large effect stiffening the final binder blend (NCHRP 2001). Asphalt mixtures containing stiffer asphalt binder can exhibit higher resistance to rutting but decreased resistance to low temperature cracking and fatigue cracking (SHRP-A-367 1994). Because the asphalt in post-consumer roofing shingles not only undergoes a stiffening process during production, but also has typically undergone years of oxidative aging, agencies are concerned RAS could have stiffening effects on the final binder blend (Scholz 2010) impacting pavement performance.  

To counter the effect of adding a stiffer binder, a softer virgin asphalt is often used.  Historically, blending charts have been used in designing HMA with two different grades of binder (Asphalt Institute 2007). With the advent of Performance Graded (PG) binders, “grade bumping” is practiced by agencies as an easy method to account for the introduction of stiffer binder in the mixture matrix.  When the percentage of reclaimed asphalt pavement exceeds a certain amount, the specified virgin binder performance is reduced one or two grades on the low temperature and/or high temperature side.

Because RAP only contains 4% to 5% asphalt content, adding 5%, 10% or 15%, RAP will not make a large reduction in the percentage of virgin binder added to the mix as compared to RAS. As a result, agencies have been accustomed to writing specifications for RAP by allowing a certain percentage of RAP. In contrast, when RAS is added to HMA, a much larger percentage of virgin asphalt is reduced because RAS can contain up to 30% asphalt. To help regulate the amount of recycled asphalt being added to HMA when RAS is used so the final blend is not too stiff, the concept of “percent binder replacement” has emerged (Bentsen 2010).  Typical percent binder replacement specifications require a maximum of 20% to 40% in combination with grade bumping requirements. The percent binder replacements of the mixes conducted in this study are between 25% and 65%.

Past Experience Using Asphalt Shingles in HMA
Literature associated with performance testing of asphalt pavements containing post-consumer RAS has increased over the last few years. A challenge for most states is to determine and integrate RAS properties into HMA mix design properties that must be taken into consideration when using post-consumer RAS. Monitoring the end product through well defined specifications helps ensure an owner/agency is receiving a quality final product that will lead to realizing the benefits of RAS.

Johnson et al. (2010) of the Minnesota Department of Transportation recently investigated the incorporation of RAS in HMA through a laboratory study and field investigation. Mixtures containing no RAS, post-consumer RAS, or post-manufactured waste RAS at 3% or 5% with either 0%, 15% or 30% RAP were developed and tested in the laboratory for binder and mixture properties.  The conclusions from the study included the following:

• Dynamic modulus laboratory tests revealed the stiffness of mixtures containing RAS/RAP was significantly higher than mixtures containing no recycled materials at high temperatures/low frequencies suggesting an increased resistance to rutting;
• The low temperature binder grade was increased with the addition of RAP and/or RAS suggesting an increase in thermal cracking potential;
• The evaluation of a failed highway section revealed a high binder replacement ratio, which was shown to be related to the amount of recycled material;
• The use of a softer grade (from PG 58-28 to a PG 51-34) reduced the stiffness of the RAP/RAS asphalt mixtures;
• Post-consumer RAS or post-manufactured waste RAS can be used for MNDOT projects; and
• The current specification of 30% binder replacement can be maintained.

 

Scholz at Oregon State University (2010) conducted a study on asphalt mixtures containing 5% post-consumer RAS with 0%, 10%, 20%, 30%, 40% and 50% RAP to determine how the addition of these materials would affect the final binder blend.  The control mixture used contained no recycled materials. The virgin binder grade for all mixtures was a PG 70-28. The results of the study revealed the following:

• Inclusion of 5% RAS and no RAP resulted in an increase in both the high and low temperature performance grades; and
• At RAP contents of 30% or more, in combination with 5% RAS, the low temperature grade exceeded that of the mixture containing only 5% RAS while the high temperature grade equaled that of the mixture containing 5% RAS.

Summary
The composition of RAS provides both an economical and engineering benefit that can enhance the performance of asphalt pavements. One of the primary engineering aspects of adding these recycled materials to HMA mixes is the rheological behavior change of the final binder blend. Because the asphalt in post-consumer roofing shingles not only undergoes a stiffening process during production, but also years of oxidative aging, it tends to have a higher performance grade in both the low and high temperatures than virgin asphalts typically used in highways. Likewise, increasing the amount of RAP in asphalt mixes has a similar stiffening effect. Blending binder from recycled materials with virgin binder can increase the performance grade of the final blend at both the high and low temperatures. An increase in the high temperature performance grade can decrease the risk of permanent deformation, but an increase in the low temperature performance grade can increase the risk of low temperature cracking. However, when RAS is used in asphalt mixtures, fibers from the roofing shingles have the potential to increase the tensile strength and ductility of the mixture and counteract the adverse effects of using recycled products at low temperatures.

Project Description
In the summer of 2009, the Tollway conducted a field demonstration project on the applicability and feasibility of using RAS in asphalt mixes that contain an increased amount of fractionated RAP. The project took place on the Jane Addams Memorial Tollway (I-90) in the Rockford, Ill., area.  From research previously conducted by the Tollway in 2007 (Vavrik et al. 2008), a new construction specification was developed that allowed asphalt mixes to have up to 40% FRAP on shoulder binder course mixes and up to 50% FRAP on shoulder base course mixes. The objective of this new research was to determine how replacing 5% of the FRAP in these new mixes containing higher amounts of FRAP with 5% post-consumer RAS would affect the performance of asphalt pavements. This was accomplished by evaluating the performance characteristics of field and laboratory produced samples of the asphalt mixtures containing either RAS or no RAS. Laboratory performance tests measured and analyzed the response of the mixtures to different loading and environmental conditions. The testing results were analyzed using statistical methods to determine differences among the sample means.  

Eight different mixes were developed that contained various percentages of RAS and FRAP. Of the eight designs developed, there were three different types of mixes: a base course, a binder course and a surface course, as shown in Table 2. These mixes were placed as test strips in the shoulder of the highway in different sections of the pavement structure.  

Table 2: Material Study Matrix

The HMA shoulder pavement structure was 6 in. deep and comprised of a 4-in. base layer and a 2-in. surface layer. (The base and binder mixes both functioned as the 4-in. base layer of the pavement structure.) Each of the base and binder course mixes were placed in alternating mile long test strips, 4 in. deep, and overlaid with mile long test strips of surface course mixes, two inches deep. Figure 2 details the shoulder pavement structure layout.

Figure 2: Shoulder pavement structure.

The HMA was produced by Rock Road Co. at the Janesville/Beloit plant in Wisconsin. For each experimental section, field and laboratory produced samples were obtained to determine if the performance characteristics of the field produced mix significantly deviated from performance characteristics of the laboratory produced mix.  For each control section, either a field or laboratory sample was obtained. Samples were obtained by Iowa State University with the assistance of Rock Road. All together, 13 mixes were collected from the project; seven field and six laboratory mixes.

The laboratory testing plan included a combination of empirical based tests and mechanistic based tests, which measured fundamental engineering properties on the asphalt mixture and the extracted asphalt binder from field produced samples and laboratory produced samples. The performance tests selected for this study were based on the type of distresses the pavement would be subjected to during its service life. These distresses include rutting, fatigue cracking, thermal cracking, and freeze-thaw distresses and were evaluated by performing the tests outlined in Table 3. In order to fully characterize the pavement performance, each sample was tested for these distresses as they are considered to be the principle types of distresses for flexible (asphalt) pavement design in mechanistic-empirical methods (Huang 2004).

Table 3: Testing Plan Matrix

All tests were conducted at Iowa State University in the asphalt materials research laboratory except for the DC(T) fracture test, which was conducted by the University of Illinois Urbana-Champaign.

Materials
The mix design summary is shown in Table 4, and provides the characteristics of each mix design. Each mix used the same PG 58-22 virgin binder. The surface mix design had a nominal maximum aggregate size of 9.5 mm (3/8 inch) and the base and binder mixes had a nominal maximum aggregate size of 19 mm (3/4 inch). The base and binder course mixes are designed as asphalt-rich, fatigue resistant mixes for layers of a perpetual pavement structure. The base mix was designed at 2% air voids while the binder mix was designed at 3% air voids.

Table 4: Mix Design Summary

The FRAP used in the mixes is a Category 2 as defined in Tollway specifications, meaning it contained natural sand.  The FRAP for each mix was comprised of two different RAP stockpiles. One stockpile contained the coarse portion RAP (primarily consisting of RAP above a number 4.75 mm screen) and the other stockpile contained the fine portion RAP (consisting of the RAP below the 4.75 mm screen). The fine portion RAP asphalt content of 6.0% is greater than the coarse portion RAP asphalt content of 3.3%. The difference between these two materials indicates that fractionating RAP into separate stockpiles will increase the control of the RAP during mix design and construction. With respect to the RAS gradation, approximately 100% of the RAS material passed the 9.5 mm sieve, and approximately 92% passed the 2.36 mm sieve for a “fine grind.” The RAS contained an asphalt content of 28.1%.

Performance Grades of the Extracted Binders
The performance grades of the extracted asphalt binders were determined from DSR and BBR test results. The high temperature performance grades are presented in Figure 3 and Figure 4. The results indicate the addition of recycled materials had a positive impact on the final binder high temperature performance grade as it increased several grades above the virgin asphalt’s high temperature performance grade of 58. The increase of grade levels should improve these mixtures rutting resistance whether 5% FRAP was replaced with 5% RAS or not.

Figure 3: High temperature grade of binder extracted from field produced samples.

Figure 4: High temperature grade of binder extracted from lab produced samples.

A comparison of the binders from field and laboratory samples indicate the binders from laboratory samples are stiffer than the binders from the field samples. Although AASHTO laboratory mixing protocols were followed, the stiffness of the binder in the laboratory mixes is high enough to conclude the samples were possibly aged too long and do not represent the true grade of the blended binder based on the properties/grades of the recovered field produced mixes. The results of the field mixes appear to better represent more reasonable changes in performance grade of the virgin binder when combined with binder from FRAP and/or RAS. The binders follow the general trend that as more recycled materials are added to the asphalt mixtures, the resulting high temperature performance grade of the binder blend will increase.

Figure 5 and Figure 6 present the low temperature performance grades of binders extracted from field and laboratory produced samples. In these results, binders from laboratory samples are also stiffer than the binders from field samples. Binders from the field samples appear to give more reasonable binder grades as the virgin low temperature grade was a PG-22.  

Figure 5: Low temperature grade of binder extracted from field produced samples.

Figure 6: Low temperature grade of binder extracted from lab produced samples.

In the binders from the field samples the low temperature grade did not change in mixes with 25% FRAP/0% RAS and 25%FRAP/5% RAS and only slightly increased in mixes with 20% FRAP/5% RAS and 35%FRAP/5% RAS. For mixes with 50% recycled materials, the increase was more substantial as the grades changed from -22 to -10. However, replacing 5% FRAP with RAS for the mix with 50% recycled materials had essentially no difference in the binder grade. While some increase in performance grade is expected with these percentages of recycled materials, the field mixes indicate adding up to five percent RAS and 35% FRAP (40% total recycled materials) only increases the low temperature grade by one half a grade (3˚C).

Dynamic Modulus Test Results
The dynamic modulus test procedure used to test the asphalt pavement samples in this study followed AASHTO TP62 “Standard Test Method for Dynamic Modulus of Asphalt Concrete Mixtures.” During the dynamic modulus test, a cyclical load is applied vertically to a cylindrical sample. It is classified as an unconfined triaxial compression test with cyclical one-dimensional loading.  

A batch of five samples was fabricated for each of the 13 HMA samples obtained in the study. Each mixture was tested at three temperatures (4˚C, 21˚C and 37˚C) and nine frequencies of cyclical loading (0.1, 0.3, 0.5, 1, 3, 5, 10, 15 and 25 Hz). With the combination of three temperatures and nine frequencies, 27 E* values were measured for each sample. Rather than comparing all 27 values, three E* values were selected from each mixture to analyze its stiffness properties at high, intermediate and low temperatures for the performance indicators of rutting, fatigue cracking resistance and thermal cracking resistance. The dynamic modulus values used were from the results of testing each mixture at 4˚C at 25 Hz (high freq), 21˚C at 10 Hz (med freq) and 37˚C at 0.1 Hz (low freq).  

Results of the one-way ANOVA tests conducted to compare the dynamic modulus values of the RAS mixes to the no-RAS mixes are presented in Figure 7. One-way ANOVA essentially reduces the analysis to a two sample t-test since a comparison is being made between two samples to test the null hypothesis if they came from the same population. The t-tests results indicate no significant differences between RAS and no-RAS mixtures at high temperature/low frequencies for all three mix types although the average E* values increased each time 5% FRAP was replaced with 5% RAS. In contrast, at intermediate temperatures and frequencies there are significant differences between RAS and no-RAS mixtures for all three mix types. The reason why the t-test detected differences at intermediate temperatures and not at high temperatures appears to be from the larger variation in test results at high temperatures. The larger variation in E* values at 37˚C may be due to softening of the epoxy used to attach the LVDT metal holders on to each sample.  

Figure 7: T-test results comparing E* values of mixes with RAS and no RAS.

The decrease in stiffness at low temperatures is an interesting observation since a larger percentage of the virgin binder is being replaced when RAS replaces FRAP.  The decrease in stiffness could be the result of fibers influencing the ductility of the mix.

While the dynamic modulus at low temperatures can give an indication of an asphalt pavements tendency to fracture or develop fatigue related distress, it does not directly measure it. For a more accurate correlation between laboratory results with field performance for cracking and fatigue, the results from the disc shaped compact tension test and the flexural beam test will be used to evaluate the performance of the RAS mixtures under these distress criteria and further investigate the influence of the RAS fibers.

Flow Number
The flow number of an asphalt mixture corresponds to the number of cycles needed to accumulate 0.5% strain in the sample tested.  A higher flow number indicates a higher resistance to rutting. The test ends at 10,000 load cycles even if the sample has not accumulated 0.5% strain. Since all of the samples tested reached 10,000 load cycles, the response measured in this test was the accumulated strain and is presented in Figure 8. The values are the average strain levels of three samples. A lower strain level was interpreted as a higher resistance to rutting.

Figure 8: Percent strain after 10,000 cycles in flow number test.

Based on the flow number test results, very little rutting is likely to occur in the Tollway mixes since all samples accumulated strains less than 5% after 10,000 load cycles. The surface course mix with 25% FRAP and no RAS produced the greatest strain accumulation of 1.92% indicating this mix has the least resistance to rutting. When 5% FRAP was replaced with RAS in the surface course and binder course the strain accumulation significantly decreased. As the percentage of FRAP increased from 25% to 35% in the base course, the strain accumulation also significantly decreased. From 35% to 45% FRAP, there was no significant change in the strain accumulation.  

Tensile Strength Ratio Test Results
The tensile strength ratio test (TSR) results of the field produced mixtures are presented in Figure 9. Many transportation agencies require that an asphalt mixture have a tensile strength ratio (TSR) value of 0.80 or greater for good performance in a freeze-thaw environment. The results indicate good freeze-thaw performance can be expected for five of the seven mixtures as their TSR values are greater than 0.80. Although the other two mixtures did not meet the criteria, they came very close. Based on these results, mixtures that contain RAS and higher percentages of FRAP can have low moisture sensitivity and perform adequately in freeze-thaw environments.

Figure 9: Tensile strength ratios of field produced mixes.

Beam Fatigue Test Results
The fatigue data from the Tollway mixes was modeled using the following relationship to characterize their fatigue behavior.



The K1 coefficient characterizes the flexural modulus, and the K2 coefficient indicates the rate of damage accumulation in a sample. When using this relationship as failure criterion for a pavement design, a lower K2 value is more conservative as it assumes faster accumulation of fatigue damage. Suggested values for K2 are 4.477 by The Asphalt Institute, 4.0 by Shell and 3.571 by the University of Nottingham (Huang 2004). Carpenter (2006) recommended the Illinois DOT use a K2 value in the range of 3.5 to 4.5.

K2 values for all the mixes are presented in Figure 10. The data does not appear to show any clear trends among the different mix types. However, only two mixtures do not have K2 values greater than 3.5, indicating sufficient fatigue performance for most of the Tollway mixes. The one mix that may incur early fatigue cracking is the base course with 45% FRAP and 5% RAS since both the field and lab samples have relatively low K2 values of 3.5 and 2.9.

Figure 10: K2 values of Tollway mixes.

Disk Compact Tension Results
The DC(T) test results from the University of Illinois Urbana-Champaign for each of the 13 mixtures (four replicates in each case) are presented in Table 5 and Figure 11. In addition, a plot of average CMOD fracture energy grouped according to the total percentage of recycled material (addition of RAS and FRAP) is displayed in Figure 12.

Figure 11: Disc compact tension DC(T) test results.

Table 5: DC(T) mix IDs

Figure 12: Average fracture energy sorted by % recycled materials.

In general, a moderate threshold for a sufficiently resistant HMA mixture to thermal and reflective cracking lies between 350-400 J/m2 (Buttlar et al. 2010). Transverse cracking frequency is found to be minimal if the pavement core fracture energy average is greater than 400 J/m2 (Buttlar et al. 2010). Figure 11 shows a decreasing trend in fracture energy as the total percentage of recycled materials is increased. None of the Tollway mixtures produced fracture energies greater than 400 J/m2. Mixtures containing 40%-50% recycled materials did not meet the lower recommended fracture energy limit of 350 J/m2 limit. When analyzing these mixtures in terms of total percentage of recycled materials, mixtures containing 40%-50% recycled materials may warrant the use of a slightly softer virgin binder grade in order to improve crack resistance.

Figure 13 presents the fracture energy of the mixtures based on percent binder replacement. Industry standards are trending toward specifying recycled materials in HMA based on percent binder replacement rather than total percentage of recycled materials. Figure 13 can be used to assist in developing specifications for HMA containing RAS and FRAP. Since the threshold for HMA to sufficiently resist thermal and reflective cracking lies between 350-400 J/m2, a more conservative specification would limit the fracture energy to a minimum of 400 J/m2 while a less conservative specification would limit the fracture energy to a minimum of 350 J/m2. Based on the trend of the RAS and FRAP mixtures in Figure 13, the maximum percent binder replacement of the Tollway mixtures could range from approximately 25% to 50% depending on the specification. An average threshold of 375 J/m2 would limit the maximum percent binder replacement to approximately 35%.

Figure 13: Fracture energy vs. percent binder replacement

Recommendations
The primary finding of this study identified low temperature cracking as the most critical distress to be addressed when implementing recycled post-consumer asphalt shingles and fractionated reclaimed asphalt pavement in Tollway mixtures. The virgin binder utilized for the mixes in this study was a PG58-22. The laboratory test results indicate that mixtures containing 5% RAS and up to 35% FRAP with the virgin PG58-22 may exhibit sufficient crack resistance, while mixtures containing more than 40% total recycled materials will be more prone to cracking. The cracking performance of the mixes with more than 40% total recycled materials may improve with a softer grade of virgin asphalt such as a PG58-28. In addition, the percent binder replacement of the mixtures that adequately performed in low temperature cracking laboratory testing ranged from 25% to 50% binder replacement with an average of approximately 35%. Monitoring of the field test sections should continue in order to identify when thermal cracking related distresses develop to see if the field performance correlates with the laboratory test results. Obtaining field cores for DC(T) testing and subsequent analysis could then be further used to determine the actual criteria for prescreening mixtures containing a high amount of recycled materials.

Andrew Cascione, P.E., and R. Christopher Williams, Ph.D., are with Iowa State University. Debra S. Haugen is with Debra S. Haugen, LLC. William G. Buttlar, Ph.D., is with the University of Illinois Urbana-Champaign. Steven L. Gillen is with the Illinois Tollway. Jay Behnke, P.E. is with STATE Testing. Ross A. Bentsen, P.E., is with Quigg Engineering, Inc.

---