Performance-Based Mix Design for Porous Friction Courses (2018)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

5 Introduction The use of PFC mixes in the United States severely declined in the 1980s due to design and performance issues and was completely eliminated in some states. While conducting a literature review on PFC mixes, many of the reports and articles encountered were dated prior to this ces- sation. Some of these articles provided valuable background information, but they were based on prior mix design procedures and test methods. Hence, to better represent the currently used PFC materials, mix design procedures, and test methods, this literature review focuses most heavily on the post-cessation research. Benefits of Porous Mixtures There are many benefits to PFC mixes that have been recognized for years as reported in NCHRP Synthesis 49 (Halstead, 1978), NCHRP Synthesis 180 (Smith, 1992), and NCHRP Syn- thesis 284 (Huber, 2000). The majority of those benefits are safety-related. The use of PFC to remove water from the surface provides good contact between tires and the pavement surface, thus minimizing the potential for accidents and traffic fatalities during rainy weather. Some of the benefits of PFC mixes include: • Reduced risk of hydroplaning, • Increased friction resistance, especially during wet conditions, • Reduced backsplash and spray from vehicle tires, • Reduced noise resulting from tire-pavement interaction, and • Improved visibility of pavement markings. Depending on the required lift thickness, PFC pavements can also be economical because they can typically be placed in thinner lifts than dense-graded mixes (Kandhal, 2002). Reduced Hydroplaning and Improved Friction Nearly 6,000 people are killed and over 445,000 people are injured in weather-related crashes in the U.S. each year. The vast majority of most weather-related crashes, 73%, happen on wet pavement (Booz Allen Hamilton, 2016). A research study by the National Highway Traffic Safety Administration (NHTSA) reported that the lifetime economic cost for each fatality was found to be $1.4 million. Therefore, any reduction in accidents and especially a reduction in traffic fatalities may have a dramatic impact on our society as a whole (NHTSA, 2014). According to Huber (2000), many agencies have seen a decrease in wet pavement accidents on roadways with PFC mixes. A traffic study for Japan in 2010 showed that OGFC significantly C H A P T E R 2 Literature Review

6 Performance-Based Mix Design of Porous Friction Courses reduced the number of fatalities during rainy weather in that country when results were com- pared to standard dense-graded mix (Figure 3). The risk of hydroplaning during a rain event is increased in low-lying areas, at the transition of super-elevated curves, or when rutting of a dense-graded mix has occurred. This surface water may cause a water film to form between the tire and the pavement, affecting the tire-pavement interface friction and the driver’s ability to control the vehicle. A PFC surface allows water to drain through the surface and exit onto the shoulder (Figure 4). This limits the risk of hydro- planing and increases the friction resistance. By limiting the amount of water that is standing or flowing across the pavement surface, users are provided a safer traveling experience during rain events. Backsplash, Spray, and Glare Reduction Backsplash and spray from vehicles can diminish a driver’s view of the paint striping and surrounding vehicles. A PFC allows the water to drain through the pavement and consequently 0 2 4 6 8 Sunny Days Rainy Days Fa ta lit y Ra te p er 1 00 M ill io n M ile s T ra ve le d Safety Benefit of PFC in Japan Standard Mix PFC Figure 3. Fatality reduction on rainy days (Shimeno & Tanaka, 2010). porous asphalt standard asphalt Figure 4. Illustration of water transport on PFC and dense-graded mixes (N. B. West Contracting, n.d.).

Literature Review 7 reduces the effect of backsplash and spray significantly when compared to dense-graded mixes. A comparison of PFC to dense-graded mix in regards to backsplash and spray shows a dramatic difference (Figure 5). The United Kingdom reported a 90%–95% reduction in backsplash and spray for PFC mix- tures when compared to dense-graded mixes (Huber, 2000). As can be seen in Figure 5, the pavement markings on the PFC mixture have a higher degree of visibility during wet condi- tions when compared to the dense-graded mix. This is especially beneficial at night during wet weather. When a film of water is on the pavement surface, it can reflect a vehicle’s headlights, and the glare can prevent the driver from following the pavement markings. The film of water also mitigates the reflective beads in the pavement markings and keeps the driver from being able to distinguish lane stripes. Pavement Noise Reduction While some highway noise comes from the vehicles themselves, a large part of this noise comes from the pavement-tire interaction (Figure 6). This is especially true when the highway speed is above 45 miles per hour (mph). Metropolitan areas seem to have the most need for noise reduction due to the close proximity of businesses and homes to the highway. The most signifi- cant reason for the need of noise reduction is the quality of life of the population. The noise can become an annoyance to humans, which leads to negative impacts on the quality of life. It can also have an economic impact on real estate by keeping properties from being developed or sold (Donavan, 2007). There are a few different methods for mitigating highway noise. One of the methods approved by the Federal Highway Administration (FHWA) is the use of noise barrier walls. While being somewhat effective, depending on distance from the source, these walls are extremely costly and often unsightly. Noise reduction is dependent on the distance, both horizontally and vertically, from the highway to the point source in question. The walls offer little help for noise mitigation beyond about 500 feet from the roadway, and offer no help for residences that may be higher than the elevation of the top of the wall. Dense-Graded Mix PFC Mixture Figure 5. Backsplash comparison (NCAT photo).

8 Performance-Based Mix Design of Porous Friction Courses A typical reduction of 5 decibels dB(A) is expected from a noise wall. A gap-graded, thin lift PFC can provide on average a reduction of 3 dB(A) (Bernhard and Wayson, 2004). Joint research conducted by NCAT, FHWA, and several state agencies has shown that there can be as much as 1.5 dB(A) difference in noise between tires from different manufacturers. Mix Designs The main function of a PFC mix is to remove water from the surface of the pavement. This helps prevent vehicles from hydroplaning, eliminates backsplash and spray from the vehicles during rain events, and allows better visibility for all operators. A PFC pavement must be perme- able enough to drain the water away from the surface and off the roadway while still providing adequate friction. An added benefit is the noise reduction achieved from the open design of the pavement. Design of PFC mixtures requires four major components: 1. Suitable materials, 2. An adequate design blend gradation, 3. The optimum binder content, and 4. Evaluation of potential performance. These four components are critical in designing a PFC mix; however, the degree of variability between agency practices allows for differences in the performance of these mixes. There are two PFC mix design procedures currently available: ASTM D7064, Standard Practice for Open- Graded Friction Course (OGFC) Mix Design (ASTM, 2013); and AASHTO PP 77, Standard Prac- tice for Materials Selection and Mixture Design of Permeable Friction Courses (PFCs) (AASHTO, 2014). However, all of the agencies that responded to the NCAT survey in this study are using a state-specific method, or other method. The focus of this study is to take an analytical approach to these differences and to develop performance tests and related thresholds that will help pre- vent current distresses such as premature raveling and top-down cracking. The following sec- tions describe the four components in more detail. Suitable Materials PFC mixes consist primarily of coarse aggregate, fine aggregate, asphalt binder, and stabilizing additives. In order to attain the high air void content required for PFC mixes, an open-graded aggregate gradation is required. This consists primarily of coarse aggregate so that the mix can Figure 6. Contribution of highway noise (Donavan, 2007).

Literature Review 9 maintain stone-on-stone contact and create a stone skeleton of the coarse particles. Aggregate mineralogy is not specified in national specifications because of limitations and availability of local aggregates. The importance of the aggregate properties and type is most notable in the stone skeleton. The coarse aggregate provides the stone-on-stone contact, while the fine aggregate and stabilizing additives help to maintain the mix’s stability and cohesiveness. A summary of current aggregate property requirements can be found in Table 1. This table compares the requirements from ASTM D7064 and AASHTO PP 77. As stated previously, these requirements and recom- mendations may be altered by agencies if local materials cannot meet the minimum requirements. The use of modified binder has become common practice for most agencies due mostly to empirical results. The use of tire rubber, styrene-butadiene-styrene (SBS), and styrene-butadiene- rubber (SBR) as asphalt modifiers has proven to increase the durability of PFC mixtures by increasing the stiffness and ductility of the binder. The increased stiffness promotes increased film thicknesses while also preventing draindown of the asphalt binder during production, transport, and construction. Determining the optimum stiffness is important when choosing a modifier. Ruiz et al. (1990) suggested that an overly stiff binder will oxidize faster, which can lead to raveling issues prior to the expected design life. If modified appropriately, however, the binder may prevent short-term raveling that is caused by the shear forces at the tire-pavement interface (Molenaar and Molenaar, 2000). Stabilizing additives are used to improve the durability of the mixture by preventing drain- down and increasing the mixture’s tensile strength (Pasetto, 2000). When draindown occurs during production and transportation of the PFC mixture, a significant amount of the asphalt binder is lost from the mix. This loss of binder can cause decreased durability, which may lead to premature raveling or cracking. Stabilizing additives, such as mineral and cellulose fiber, can help prevent draindown along with reinforcing the film thickness of the asphalt binder. Design Gradation Selection After suitable aggregate sources have been chosen, the optimization of the mix can begin by creating three trial blends that fall on the coarse limit, fine limit, and in the middle of the Test Description Method ASTM D7064 AASHTO PP 77 min. max. min. max. Coarse Aggregate Los Angeles Abrasion, % Loss AASHTO T 96 - 30 - 30 Flat or Elongated, % (5 to 1) ASTM D 4791 - 10 - 10 Soundness (5 Cycles), % Sodium Sulfate AASHTO T 104 - - - 10 Magnesium Sulfate - - - 15 Uncompacted Voids AASHTO T 326, Method A - - 45 - Fine Aggregate Soundness (5 Cycles), % Sodium Sulfate AASHTO T 104 - - - 10 Magnesium Sulfate - - - 15 Uncompacted Voids AASHTO T 304 Method A 40 - 45 - Sand Equivalency AASHTO T 176 45 - 50 - Note: “-” = no data available or applicable. Table 1. Aggregate requirements for PFC mix designs.

10 Performance-Based Mix Design of Porous Friction Courses recommended gradation range as described in ASTM D7064. The agencies surveyed by NCAT that currently use PFC mixes provided their gradation specification ranges for PFC mix designs. Table 2 summarizes the response to the survey question regarding the gradation specification ranges. The asphalt content selected for the trial designs is based on the combined aggregate bulk specific gravity (Table 3) (Cooley et al., 2009) (AASHTO, 2014). Three specimens are prepared for each of the three trial designs; the voids in the coarse aggregate (VCA) and the air void con- tent are used to determine which trial will be selected for design. The VCA is used to determine if the mix has stone-on-stone contact. The VCA of the mix (VCAMIX) must be less than the VCA of the dry-rodded coarse aggregate (VCADRC) in order for the aggregate skeleton to have stone-on-stone contact. The survey showed that only one state agency (Louisiana) uses VCA as part of the design procedure while the rest rely on historical gradations and performance. According to ASTM D7064, the optimum design gradation should be the one that meets the VCA requirement and has the largest air void content, as long as it meets the minimum air void State Mix Design ¾ in. 3⁄8 in. 19 mm ½ in. 12.5 mm 9.5 mm No. 4 4.75 mm No. 8 or 10 2.36 mm No. 16 1.18 mm No. 30 or 40 0.6 mm No. 200 0.075 mm AL 100 85-100 55-65 10-25 5-10 2-4 AZ 1 100 30-45 4-8 0-2 AZ 2 100 31-46 5-9 0-3 CA 1 78-89 28-37 7-18 CA 2 99-100 29-36 7-18 FL 100 85-100 55-75 15-25 5-10 2-4 GA 1 100 85-100 20-40 5-10 2-4 GA 2 100 85-100 55-75 15-25 5-10 2-4 GA 3 100 80-100 35-60 10-25 5-10 1-4 LA 1 100 90-100 25-50 5-15 2-5 LA 2 100 85-100 55-75 10-25 5-10 2-4 MS 100 80-100 15-30 10-20 2-5 NC 1 100 75-100 25-45 5-15 1-3 NC 2 100 75-100 25-45 5-15 1-3 NC 3 100 85-100 55-75 15-25 5-15 2-4 NE 100 95-100 40-80 15-35 5-12 0-3 NJ 1 100 89-100 30-50 5-15 2-5 NJ 2 100 85-100 35-60 10-25 5-10 2-5 NJ 3 100 85-100 20-40 5-10 2-4 NM 100 90-100 25-55 0-12 0-8 0-4 NV 1 100 90-100 35-55 5-18 0-4 NV 2 100 95-100 40-65 12-22 0-5 OR 1 99-100 90-100 22-40 5-15 1-5 OR 2 99-100 90-98 18-32 3-15 1-5 SC 100 85-100 55-75 15-25 5-10 0-4 TN 100 85-100 55-75 10-25 5-10 2-4 TX 1 100 80-100 35-60 1-20 1-10 1-4 TX 2 100 95-100 50-80 0-8 0-4 0-4 UT 100 90-100 35-45 14-20 2-4 Table 2. Gradation specification ranges (% passing) for PFC designs currently used by agencies.

Literature Review 11 requirement. ASTM D7064 specifies a minimum accepted air void level of 18.0%. Table 4 lists the results of the survey conducted in this study in regard to state agency air void requirements for PFC mix designs. Determining the Optimum Asphalt Binder Content Once trial gradations have been completed and the design gradation has been selected, the optimum asphalt content needs to be determined. Additional specimens should be fabricated at three different asphalt contents. The asphalt contents should be in 0.5% increments above and below the trial asphalt content. As established in the previous section, the air void content must be at least 18.0% for most agencies. Combined Aggregate Bulk Specific Gravity Minimum Asphalt Content Based on Mass (%) 2.40 6.8 2.45 6.7 2.50 6.6 2.55 6.5 2.60 6.3 2.65 6.2 2.70 6.1 2.75 6.0 2.80 5.9 2.85 5.8 2.90 5.7 2.95 5.6 3.00 5.5 Table 3. Minimum asphalt content for PFC mix designs (Cooley et al., 2009). State Aid Void Requirement Alabama Min. 12% Georgia 18% – 20% for PFC 20% – 22% for PEM Louisiana 18% – 26% Maine 18% – 22% Maryland Min. 18% Mississippi Min. 15% Nebraska 17% – 19% New Jersey Min. 15%, 18%, or 20% depending on mix North Carolina Min. 18% Oklahoma Min. 18% Tennessee Min. 20% Texas 18% – 22% Virginia Min. 16% Table 4. Air void requirements of PFC mix designs (NCAT Survey).

12 Performance-Based Mix Design of Porous Friction Courses According to ASTM D7064, the mix must also have a draindown percentage less than 0.3%, a tensile strength ratio (TSR) of 0.80 or greater, and a VCAMIX ≤ VCADRC. Cooley et al. (2009) recommends selecting the optimum asphalt content based on the requirements in Table 5. Note that the TSR is decreased to 0.70 or greater and the VCAMIX must be less than VCADRC. NCHRP Report 640 (Cooley et al., 2009) states that “when VCAMIX is less than or equal to VCADRC stone-on-stone contact exists.” Previously, the optimum asphalt content for PFC mixes was selected based primarily on the pie plate test. The pie plate method is a visual and subjective evaluation of the draindown of asphalt binder and several other agencies have found alternative methods (McDaniel, 2015) for determining the optimum asphalt content. State of the Practice A PFC pavement provides many benefits over a dense-graded mix, but the obvious drawback of durability issues causes a shorter pavement performance life when compared to dense-graded mixtures. Timm et al. (2014) completed a study at the NCAT Test Track that concluded that the back-calculated structural number of OGFC was 0.15. This shows that while permeable surface courses are useful for water drainage, splash prevention, and noise mitigation, they provide only about a third of the structural capacity of a dense-graded mix. With little improvement to structural capacity, high asphalt contents, and a shorter service life, PFC mixes must become more resilient if they are going to be accepted by more state agencies. The subsequent sections will present the current state of the practice for PFC mixes. Selection of Materials Aggregate Characteristics As part of the recent survey conducted by NCAT, agencies were asked to provide the type of aggregate they specified along with what they deemed as the most important aggregate proper- ties when designing PFC mixes. The responses to the questionnaire can be seen graphically in Figure 7 and Figure 8. Granite and limestone are the predominant aggregate types, while abra- sion, polishing, and flat and elongated particles are the principal aggregate properties being evaluated for PFC designs. Aggregate characteristics that should be considered when designing PFC mixes are durability, polish resistance, angularity, cleanliness, abrasion resistance, and absorption. The most impor- tant of these characteristics are the polish resistance and durability (Cooley et al., 2009). The Mix Property Requirement Asphalt Binder (%) Table 3 18–22 Cantabro Loss (%) <15.0 Air Voids (%) VCAMIX (%) <VCADRC TSR 0.70 Draindown at Production Temperature (%) 0.30 Table 5. Requirements for selecting optimum asphalt content for PFC mixtures (Cooley et al., 2009).

Literature Review 13 most common way to define durability and polish resistance is with the sulfate soundness test and the polish stone value (PSV). The PSV relates to the aggregate’s ability to resist polishing. The minimum recommended PSV for porous mixes is 55 (German Asphalt Pavement Asso- ciation, 2006). Europe considers this to be the most important criteria when designing porous mixes (Lefebvre, 1993). Spain recommended a minimum value of 45, while Great Britain recom- mends a minimum value of 60 (Bolzan et al., 2001). Attaining a value of greater than 50 may be difficult depending on locally available aggregates. New Zealand typically attains a value of 55–61 for its porous mix designs (Fletcher and Theron, 2011), while South Africa recommends a value of greater than 50 (Masondo, 2001). The sulfate soundness test measures the durability of the aggregate in terms of weathering with regard to freeze–thaw cycles. The amount of accept- able loss for sulfate soundness depends on the agency. The state of Georgia allows up to 15% maximum loss (Watson et al., 1998), while Oregon only allows a 12% loss (Huber, 2000). Aggregate angularity and abrasion resistance are the next most important properties, both being specified in ASTM D7064. Aggregate angularity, more commonly known as fractured face count, gives requirements for the number of fractured faces a stone particle must contain. Granite, 14 Gravel, 5 Limestone, 11 Traprock, 4 Sandstone, 5 Slag, 2 Other, 4 Responses Figure 7. Aggregate type specified by agencies. Abrasion, 12 Polishing, 15 Absorption, 7 Flat/Elongated Particles, 13 Mineralogy, 6 Responses Figure 8. Required aggregate properties for use in PFC designs.

14 Performance-Based Mix Design of Porous Friction Courses If crushed gravel is used in the design, ASTM requires that 95% of the particles have two or more fractured faces. The criterion for two or more fractured faces ranges from 75% in Spain (Ruiz et al., 1990) to 100% in Florida (Huber, 2000). Based on survey responses, the percent of flat and elongated particles was one of the most emphasized properties for PFC aggregate. Agencies typically specify no more than 10% flat and elongated particles at a 5:1 ratio of minimum to maximum dimension. ASTM D7064 recommends the same criteria. The most common way to test for aggregate abrasion resistance is through the Los Angeles (L.A.) Abrasion Test. The L.A. Abrasion Test determines the aggregate’s resistance to crushing and degradation. The amount of allowed loss varies between agencies, but ranges from 12% to 50% loss allowed for the coarse aggregate (Alvarez et al., 2006) (Watson et al., 1998). According to ASTM and Kandhal (2002), the current standard in the United States is a maximum allowable loss of 30%. Asphalt Binder PFC mixes have been used successfully with both modified and unmodified binders. The use of modified binders became more prevalent after research showed that modifying the binder could increase the life of the pavement and prevent draindown of the mix. The binders are graded according to the Superpave Performance Grading (PG) system in the United States, but some European countries still implement a penetration grading system. The use of modifiers, such as SBS, SBR, and ground tire rubber (GTR), has significantly improved the mix performance of many different asphalt binders by increasing the modulus and elasticity of the asphalt. The most commonly used polymer is SBS (Kuennen, 2012). This elastic polymer absorbs the aromatics in the asphalt and increases the elastic recovery of the binder. SBS is a block copolymer of polybutadiene and polystyrene, whose combination increases strength and flexibility of the asphalt. Crumb rubber modifiers (CRMs) made from GTR have been used in asphalt since the 1960s (Carlson and Zhu, 1999). The most commonly used method for producing CRM is the cracker- mill process, which produces ground/torn particles ranging from 4.75 mm (No. 4 sieve) to 0.42 mm (No. 40 sieve) in size. CRM is typically added to the asphalt binder at a rate of 10%–20% by weight of binder, using a “wet” process. The “wet” process blends the CRM into the binder at a temperature range of 300°F–400°F for 45 minutes to an hour. The reported effect of the CRM on the performance of mixes varies, but it is suggested that the CRM can offer more resistance to asphalt oxidation while mitigating rutting and resisting thermal and reflec- tive cracking (Willis et al., 2013). Some results have shown a decrease in the permeability of PFC mixtures using CRM (Suresha et al., 2009). The purpose of using CRM is for binder modifica- tion, not as binder replacement; therefore, the 10%–20% rubber is in addition to the optimum asphalt content. This may be one of the contributing factors for observed decreased perme- ability for CRM mixtures. Both SBS and CRM modifiers provide a stiffer asphalt film that leads to increased cohesion of the aggregate stone skeleton. This provides a more durable PFC mix. In 2000, Huber reported that Britain used both modified and unmodified binders but required a 100-penetration value. Italy and Spain used only modified binders requiring an 80/100 pen (Huber, 2000). The Netherlands and Switzerland do not require modified binders, but Switzerland allows modified binders as an alternative to conventional binder (Alvarez et al., 2007). Europe primarily uses polymer modification, SBS specifically, while South Africa uses both polymer and rubber in their PFC mixes (Huber, 2000). While modified binders are beneficial to PFC performance, they are not always necessary. Consideration of the anticipated traffic volume and weather should be taken into account prior to designing the mix. The use of stabilizing agents, such as fibers, could replace the need for binder modifiers on low- to medium-traffic roads (Kandhal, 2002).

Literature Review 15 Stabilizing Agents Stabilizing agents come in several forms, the most common of which are fibers. Fibers pro- vide stability to the mix while also preventing draindown of the binder. There are many dif- ferent types of fibers, including cellulose, mineral, asbestos, polypropylene, acrylic, and glass fiber (Bennert and Cooley, 2014). Draindown is an issue in porous mixes because of the open- aggregate grading of the blend. The aggregate blend has little material passing the 4.75 mm (No. 4) sieve and a relatively low amount of P-200 material compared to dense-graded mixes. This results in a much lower aggregate surface area for PFC mixes, allowing for a thick coating of asphalt binder on the aggregate particles. A typical film thickness of a dense-graded mix is approximately 8 microns, while a porous mix is typically around 25–30 microns (Watson et al., 2004a). New Zealand has had success with their porous mixes that typically have a film thick- ness of 10 microns (Fletcher and Theron, 2011). This thick film can potentially draindown off the aggregate during production/construction and can cause a myriad of issues such as the loss of binder content in the mixture, flushing of asphalt in concentrated draindown spots on the roadway, and excessive adherence of the mix to the truck bed. In 1998, a survey conducted by Kandhal and Mallick showed that only 19% of state agencies were using fiber in their PFC designs (Kandhal and Mallick, 1998). By 2009, 85% of agencies reported using fiber in PFC mixtures, according to Cooley et al. (2009). This percentage has not changed significantly since 2009, according to the most current survey indicating that 82% of the responding agencies use some type of stabilizing agent. Figure 9 depicts the results of the most current survey. Several agencies allow cellulose and mineral fiber. The most common stabilizers used in the United States, Europe, and Australia are cellulose and mineral fiber (Cooley et al., 2009). They are typically added to the mix at a rate of 0.3% by total weight of the mixture, but can range from 0.2% to 0.5%. These values are recommended in the ASTM specification along with NCHRP Report 640 (Cooley et al., 2009) (ASTM, 2013). Cellulose fibers come in either pellet or loose form. The cellulose has high absorption so it can be an excellent method for maintaining high binder contents. The most common manu- factured mineral fibers are mineral wool or rock wool. The fibers are formed by melting the minerals down and spinning the minerals until they form fibers, which is similar to the process of making cotton candy (McDaniel, 2015). Figure 10 illustrates how effective fibers are at pre- venting draindown. Mineral Fiber, 11 Cellulose Fiber, 15 WMA Technology, 2 GTR, 4 Other (RAS), 1 None, 4 Responses Figure 9. Survey response to stabilizing additives used in PFC mixes.

16 Performance-Based Mix Design of Porous Friction Courses Filler/Anti-stripping Agents Hydrated lime is used as a filler material by many agencies. The Netherlands uses a limestone filler but requires that at least 25% of it must be hydrated lime. Australia not only uses hydrated lime as a filler, but also portland cement and limestone dust. Hydrated lime also doubles as an anti-strip agent to prevent moisture damage to the mixture. A respondent to the recent survey stated that one of the contributing factors for PFC mixtures performing for more than 12 years of service life was the use of 1.0% hydrated lime. Florida DOT and Georgia DOT have PFC pavements that often achieve in excess of 12 years of pavement life and attribute part of that success to the use of hydrated lime. Both agencies indicate that hydrated lime performs better than liquid anti-strip additives in regard to overall pavement life. Cooley et al. (2009) reports that filler contents vary depending on the maximum aggregate size of the design. Italy provides the lower limit of 0% passing the 0.075 mm (No. 200) sieve, while South Africa provides the upper limit by allowing as much as 8%. The recent survey shows that the responding agencies allow anywhere from 0% to 5% filler in PFC mix designs. Design Gradation Selection With suitable materials selected, trial gradations with initial asphalt contents should be estab- lished. There is no nationally accepted gradation band for PFC mixes; however, ASTM D7064 gives an “example” gradation. The example gradation in ASTM D7064 is the same recom- mended gradation (Table 6) in the National Asphalt Pavement Association (NAPA) Informa- tion Series 115 (Kandhal, 2002). Table 6 also shows the recommended gradation according to the FHWA Technical Advisory (FHWA, 1990). There are many different gradations used by agencies across the world. Some of these agencies define their mixtures according to nominal maximum aggregate size (NMAS), while others use a maximum aggregate size to define the mix type. The results of the Cooley et al. (2009) survey were converted to maximum aggregate size prior to summarizing them. According to Cooley’s survey, the only agencies to have a 25.0 mm (1 inch) design were Oregon and Great Britain, while Figure 10. Effect of fibers on draindown potential of PFCs (Watson et al., 2003).

Literature Review 17 the only 12.5 mm (1/2 inch) designs were from Louisiana and Great Britain. All other respond- ing agencies provided 19.0 mm (3/4 inch) designs. An illustration of these gradation bands for both the U.S. and international agencies can be seen in Figure 11 and Figure 12. The U.S. agencies primarily use a 19.0 mm maximum aggregate size design, and the majority of them are gapped around the 4.75 mm (#4) sieve. The same can be seen for the international agencies in Figure 12. The gradation band recommended by NCAT’s research in 2003 is also included in Figure 12 to show a comparison between the typical gradations used in the U.S. and other countries. This comparison is well illustrated on the 9.5 mm sieve. Spain has an upper limit of approximately 75% passing the 9.5 mm sieve, while Italy has a lower limit as low as 10% passing. The NCAT-recommended gradation band ranges from 35% to 60% passing the 9.5 mm sieve. The amount of filler allowed in the designs also varies significantly. According to this survey, the 19.0 mm U.S. designs range from 2% to 4% filler. The international agencies range from 0% for Italy to a high of 8% for South Africa. Percent Passing Sieve (mm) NAPA FHWA 19.0 100 100 12.5 85–100 100 9.5 35–60 95–100 4.75 10–25 30–50 2.36 5–10 5–15 0.075 2–4 2–5 Table 6. Recommended gradation for OGFC (Kandhal, 2002) (FHWA, 1990). Figure 11. 19.0 mm PFC Gradation requirements from U.S. agencies (Cooley et al., 2009).

18 Performance-Based Mix Design of Porous Friction Courses Optimum Binder Content Selection Trial asphalt contents, normally in 0.5% increments, are fabricated and then the properties of the mix are considered; however, there is no specific procedure that requires particular proper- ties of the mixture to be considered. While the ASTM standard suggests a minimum of 18.0% air voids and a draindown of less than 0.3%, it makes all other properties optional. The NAPA publication (Kandhal, 2002) has criteria for certain mixture properties in order to select the opti- mum binder content. A comparison of the properties can be found in Table 7. The minimum permeability requirement is based on research conducted by NCAT in 2000 (Mallick et al., 2000). According to ASTM D7064, design specimens are to be compacted using a Superpave Gyratory Compactor (SGC) to a design level of 50 gyrations. This criterion was developed on recommendation from a previous NCAT study (Kandhal and Mallick, 1999). The study Figure 12. 19.0 mm PFC gradation requirements from international agencies (Cooley et al., 2009). Mix Property NCHRP Report 640 (Cooley et al., 2009) ASTM D7064 NAPA Series 115 (Kandhal, 2002) Air Voids (%) 18 – 22 ≥18 ≥18 Unaged Cantabro Loss (%) ≤15.0 ≤20.0 (Optional) ≤20.0 VCAMIX (%) <VCADRC ≤VCADRC ≤VCADRC Tensile Strength Ratio ≥0.70 ≥0.80 ≥0.80 Draindown at Production Temperature (%) ≤0.30 ≤0.30 ≤0.30 Permeability (m/day) (Optional) 100 100 100 Table 7. Optimum asphalt content properties for PFC mixes.

Literature Review 19 indicated that 50 gyrations in the SGC provided approximately the same amount of compaction as a 50-blow Marshall design (European PFC design). A new design model (Figure 13) proposed by Bennert and Cooley (2014) considers a com- bination of draindown and Cantabro loss when selecting the optimum asphalt content. This allows the designer to select the optimum based on an acceptable range of binder contents. The model does not take into account air void content, so this should be considered when making the final selection. South Africa has a similar process for determining the optimum asphalt content that includes air void content. The optimum is selected by averaging the larger asphalt content of the “minimum” values (durability and abrasion resistance) with the smaller asphalt content of the “maximum” values (air void content and draindown) (Figure 14) (Masondo, 2001). The durability of the mix is determined according to the Cantabro Abrasion Test (AASHTO TP 108). This method of testing was developed in Spain in the 1980s (Lefebvre, 1993). It is the most common method for determining the durability of PFC mixes. The relationship shows Figure 13. Philosophy of designing PFC mixtures (Bennert and Cooley, 2014). Figure 14. South Africa typical graph for determining optimum binder content for porous mixtures (Masondo, 2001).

20 Performance-Based Mix Design of Porous Friction Courses that as the asphalt content increases, the durability of the mixture improves; but the risk of draindown is increased. Draindown in the U.S. is typically performed according to ASTM D6390 or AASHTO T 305; however, there are several different methods for assessing the draindown potential for PFC mixes. Some agencies use the Pyrex bowl method, better known as the “pie plate test,” for deter- mining optimum asphalt content. This method is based on draindown, is subjective, and is only a visual test. Approximately 1,000 grams of uncompacted PFC mix is placed in an 8–9 inch Pyrex glass plate. The plate is then placed in an oven at 121°C (250°F) for 1 hour, and a visual examination of the residual asphalt binder is conducted after the mix has cooled and the pie plate is inverted (Figure 15). This is done for all asphalt contents tested in the trials. The NCAT survey conducted during this study showed that 19% of the state agencies using PFC mixes are utilizing the pie plate test for determining the optimum asphalt content. The Schellenberg drainage test is also another method that has been used in the past. A 1,000 gram sample of uncompacted PFC mix is placed in a glass beaker, which is then placed in an oven at 170°C (338°F) for 1 hour. The loose mix is then removed from the beaker and the asphalt residue is quantified. This method has a more measureable result than the pie plate test, which makes it more valuable when determining draindown. The most common method for determining draindown was developed by NCAT. A wire basket with uncompacted PFC mix is placed in an oven at 15°C (27°F) greater than the antici- pated mixing temperature. The wire basket is placed on a container of known mass, and after 1 hour the container is removed from the oven. The amount of material that passed through the wire basket and into the container is then quantified as a percentage of the total mass. The wire baskets were originally a 4.75 mm (No. 4) mesh, but subsequent research by Watson et al. (2003) showed that some intermediate-sized stone could pass through the mesh and allow more than just asphalt binder to drain down onto the container. Based on that research, it was recom- mended that a 2.36 mm (No. 8) mesh be used for draindown testing instead. In addition to the mesh size change, amendments were made to the procedure in which any binder remaining on the basket after the 1-hour conditioning in oven should be considered part of the draindown percentage. Georgia DOT previously used both the pie plate test and the Schellenberg method for deter- mining draindown but has since moved to the draindown basket method. South Africa allows the designer to choose either the Schellenberg method or the draindown basket (Masondo, 2001). If significant draindown is occurring, fiber or the addition of a binder modifier will help to mitigate the draindown (Cooley et al., 2009). Figure 15. Pyrex bowl method for determining draindown of PFC mixtures.

Literature Review 21 Construction and Maintenance of PFC Mixes The main construction-related issues with PFC mixes are raveling and delamination. The fol- lowing factors are the main influences that lead to issues with PFC pavements during production and construction: • Homogenous mix gradation and temperature; • Asphalt content; • Tack bond strength, rate, and quality of application; • Layer thickness; and • Mixing temperature during placement. According to Bennert and Cooley (2014), production and construction issues may be more responsible for raveling than mix design properties. Inconsistent temperatures in the mix during construction can lead to both delamination and raveling. Delamination occurs when the bond between the underlying surface and the PFC is inadequate and causes a slip plane. A tack applica- tion is placed on the surface of the underlying layer so that the PFC can adhere. If the underlying layer is too cold or covered in dust, the tack material may not adhere, causing the pavement to delaminate. The amount and type of tack material is also important. Since PFC mixtures are coarse-graded, there is less contact area between aggregate particles in the PFC and the under- lying layer than for a dense-graded mix. It would therefore seem logical that the tack rate should be increased so that the contact area has the same tack bond strength as a dense-graded mix. Several studies have been conducted on the interface bond strength. An NCAT study in 2005 recommends a bond strength of 100 psi, when tested at 77°F, for newly constructed overlays (West et al., 2005). This study was primarily for dense-graded overlays but did include porous overlay data in the bond strength recommendation. By improving the bond of the two layers, the risk of delamination is diminished. The rate at which the tack is applied is also a critical component. Figure 16 shows tack rates provided by agencies that responded to the 2014 NCAT survey. Most tack material is an emul- sion. Emulsions consist of asphalt binder particles that are suspended in water. This allows the tack to be spread more evenly and allows it to be applied at lower temperatures for safety reasons. The percent of asphalt binder in the emulsion is known as the residual. Most applica- tion rates are based on the residual. There is a wide range of tack rates provided in the responses (0.02–0.15 gal/sy) depending on the type of tack material used. One example of the “Other” category shown in Figure 16 is from South Carolina, which provided a range of 0.05–0.15 gal/sy. 0 1 2 3 4 5 6 7 <0.04 0.04-0.08 0.08-0.12 > 0.12 Other N o. o f R es po ns es Application Rate, gal/sy Figure 16. Agency response for tack application rate (NCAT survey).

22 Performance-Based Mix Design of Porous Friction Courses While raveling can be linked to the interface bond, it is also a durability issue that begins at the top of the pavement. Mix temperature is one of the biggest concerns when constructing PFC mixtures. Consistent mix temperatures and short haul times are critical for adequate placement. Due to the open structure, a PFC will cool faster than a standard dense-graded mix. This can be mitigated somewhat with the use of insulated truck beds and tarpaulins during transport to resist crusting of the outer surface of the mix on the haul truck. However, initial production tempera- ture, haul time, and the ambient/pavement surface temperature are more critical. Great Britain specifies that from production until the mix is placed on the ground, no more than 3 hours can elapse (Alvarez et al., 2006). The FHWA Technical Advisory recommended a maximum haul distance of 40 miles and a travel time of less than 1 hour (FHWA, 1990). In order to mitigate the loss of heat in PFC mixtures during construction the following items have be considered: 1. Provide an adequate number of haul trucks so that there is no pause in construction. When the paver is required to wait on haul trucks due to a lack of mix, a cold transverse joint is created (Figure 17a). 2. Preheat the screed before the initial start-up at a transverse joint. A cold screed will pull some of the mix particles at the start-up transverse joint and will cause a lack of mix homogeneity. In dense-graded mix, the material can be raked to correct this issue; but raking a PFC, espe- cially with modified binder, is somewhat difficult. 3. Use a material transfer vehicle (MTV). An MTV is used to remix the asphalt mixture after it has been transported to the job site. This remixing should result in a homogenous mix tem- perature that will help eliminate cold spots in the asphalt mat. 4. Ensure adequate screed crown and temperature. Most pavers use multiple burners to heat the screed. These burners can go out during production and cause a cold spot in one section of the screed. It is important to provide proper adjustment of screed crown and screed exten- sions in order to obtain a smooth finish. Due to the relatively thin layer thickness and high proportion of coarse aggregate, failure to properly adjust the screed will cause the mix to pull and results in streaks in the mat (Figure 17b). Cold Transverse Joint (a) Center of Paver Streak (b) Figure 17. Raveling of PFC mixture due to construction practices.

Literature Review 23 Performance Testing Moisture Susceptibility The most recognized and widely used performance test for PFC mixes is a moisture suscep- tibility test. There are three different methods for determining moisture susceptibility of PFC mixes. The first is the modified Lottman method (AASHTO T 283). The test uses the indirect tensile strength of the mix to calculate a tensile strength ratio (TSR). Due to the open void structure of PFC mixtures, it is not possible to saturate the specimens to a certain degree of saturation. Therefore, the specimens are placed under water, and a partial vacuum of 26 in. Hg (660.4 mm Hg) below atmospheric pressure is applied to the sample for 10 minutes. The speci- men is then placed in a container and kept submerged (in order to maintain saturation) during the freeze cycle. The specimens are frozen for a minimum of 16 hours and are then thawed in a hot water bath at 60°C (140°F) for 24 hours. The specimens are then normalized in a water bath at the room temperature of 25°C (77°F) for 2 hours prior to testing. Some research has recom- mended using 5 freeze–thaw cycles prior to testing. In 2004, Watson et al. showed that there was no significant difference between 1, 3, and 5 freeze–thaw cycles (Watson et al., 2004a). The second moisture susceptibility test is the boil test (ASTM D3625). The boil test requires plac- ing a 250-g uncompacted PFC sample in a beaker of boiling water for 10 minutes. The sample is not to be conditioned and must be below 212°F and above 180°F prior to adding it to the boiling water. After the 10-minute boiling, the sample is removed from the beaker, and a visual inspection of the sample is performed to determine if any visual stripping of the aggregate has occurred. Texas and Georgia have used this method in the past, but its current use is primarily as a quality control tool. The third and final option for evaluating moisture susceptibility is the wheel-tracking test. Cooley et al. (2009) did some initial testing with PFC beams by submerging the specimens in 60°C water bath overnight and testing the specimens for rut depth in the Asphalt Pavement Analyzer (APA) for 8,000 cycles (GDT-115). This process is similar to the Hamburg Wheel- Tracking Test (HWTT), which also tests for moisture susceptibility of the mixture. The current standard for the HWTT (AASHTO T 324) can be performed on beams or Superpave gyratory specimens, but it is tested at 50°C under a constant load of 158 lbs. Alvarez et al. (2007) suggested not including HWTT as part of mix design criteria due to its large range of coefficient of variance (0.02–0.57). Louisiana actually requires HWTT be performed as part of the design requirements [Louisiana Department of Transportation and Development (LADOTD), 2016]. The require- ment for a mixture with a PG 76-22m binder states that the specimen must reach 5,000 passes prior to reaching 12.0 mm of rut depth. Cantabro Abrasion Testing According to the Cooley et al. (2009) survey, the most common way to determine the durabil- ity of a PFC is with the use of the Cantabro Abrasion Loss Test. The test is typically performed at 25°C (77°F). Herrington et al. (2005) proposed that testing the specimens below 0°C may provide a better differentiation between mixes with different binder types. While maintaining a tempera- ture lower than 25°C during testing would be difficult and expensive, they stated it would possibly be a better alternative when trying to compare different binder modifiers and types (Herrington et al., 2005). The British specification for the Cantabro Test requires a test temperature of 10°C (British Standards Institute, 2004). The Cantabro specimens can be classified as aged, unaged, or moisture conditioned. ASTM D7064 recommends a maximum stone loss of 20% for unaged specimens. Aged specimens are placed in a forced draft oven for 7 days at 140°F prior to testing. The criterion for the aged specimens is a maximum of 30% loss allowed. Some international agencies (Great Britain, South Africa, and Australia) require a moisture-conditioning period

24 Performance-Based Mix Design of Porous Friction Courses prior to testing. In the 2009 survey, Great Britain was the only agency that provided their aging protocol, which stated that the specimens were submerged for 24 hours at 140°F in a water bath prior to testing (Cooley et al., 2009). Field Performance Out of the 21 agencies that responded to the NCAT 2014 survey and are currently using PFC designs, the average service life for a PFC pavement is between 8 and 10 years. The distribution of the responses can be seen in Figure 18. Agencies that were achieving greater than 12 years of service life for PFC were asked what, if any, special consideration was given when designing and maintaining the pavement. Their responses are as follows: • Eliminated the use of gravel, lightweight aggregate, and slag due to past performance issues; • Required fiber stabilizer in all mixes; • Replaced liquid anti-strip with hydrated lime; and • Increased mix production temperature to 320°F, which resulted in improved smoothness and more uniform texture. Agencies were asked what primary distress was causing issues with their PFC pavements and the majority (76%) of responses claimed that raveling was the biggest issue. This is not only a local issue, but is also one of the main issues plaguing Europe’s porous pavements. Van der Zwan (2011) stated that over 90% of their maintenance practices for porous pavements are in regard to raveling. Figure 19 shows the distribution of the different distress types from the most recent survey. The agencies that are not currently using PFC designs were asked what improvements were needed in order for them to consider using a PFC. The majority of the responses requested improved durability (38%) along with safety and performance in colder climates (32%). Based on some of these responses, the lack of PFC use in the colder climates (which was shown in Fig- ure 1) seems to be primarily due to water freezing in the pavement and causing safety concerns. In the Netherlands, over 80% of the roadways have some form of porous pavement. The Netherlands has a very harsh climate and is subject to many freeze–thaw cycles throughout the year, yet still has adequately performing mixes. Arambula-Mercado et al. (2016) states that even though the Netherlands has good success with porous pavements, they also have a standard maintenance practice of rehabilitating 5%–7% of their PFC pavements annually. The design life of these pavements exhibits a wide range of 5 to 18 years. The Netherlands uses a two-layer 0 1 2 3 4 5 6 7 8 9 <6 6-8 8-10 10-12 >12 N o. o f R es po ns es Typical Service Life, Yrs Figure 18. Typical service life of PFC pavements.

Literature Review 25 system that has a coarse-graded PFC on the lower lift and a finer-graded PFC on the surface. A fixed binder content of 4.5 to 5.5 is also utilized. Even though there is a set range for the binder content, the mix designs must still meet certain criteria when tested using a Cantabro Abrasion Test, semicircular bending test, indirect tensile strength, and a rotating surface abrasion test (Ongel et al., 2007). Factors Affecting Performance of PFC Mixes In 1993, the long-term pavement performance (LTPP) program developed a list of the most common distresses for hot mix asphalt (HMA). A comprehensive list of these distresses can be found in Table 8. PFC pavements primarily fail by raveling (Huber, 2000) with longitudinal cracking second in rank. Delamination is also seen in PFC pavements. The following sections discuss factors that can potentially cause these distresses and other issues affecting the functional performance of PFC pavements. Raveling Raveling can be classified as short-term or long-term raveling. Short-term raveling occurs on new construction due to the shearing force between tires and the pavement surface. Poten- tial causes for this include, but are not limited to, placing the mix at an inadequate tempera- ture, or not properly compacting the mix that consequently prevents the creation of the stone 0 2 4 6 8 10 Ra ve lin g Cra cki ng Str ipp ing Clo gg ing De lam ina tio n 12 14 16 18 N o. o f R es po ns es Figure 19. Primary distress observed in PFC pavement (survey response). Cracking Patching and Potholes Surface Deformation Surface Defects Miscellaneous Fatigue Block Edge Longitudinal Reflection Transverse Patch Deterioration Potholes Rutting Shoving Bleeding Polished Agg Raveling Lane to Shoulder Drop-off Water damage Table 8. LTPP defined distresses for HMA pavements [Strategic Highway Research Program (SHRP), 1993].

26 Performance-Based Mix Design of Porous Friction Courses skeleton needed to maintain the structural integrity of the pavement. California’s design guide for OGFC also maintains that inadequate compaction can cause short-term raveling (Caltrans, 2006). Other construction issues such as waiting on trucks or long-haul distances can also play a part in short-term raveling. If construction is on hold while waiting on trucks, it can cause dif- ferences in the temperature profile of the pavement, resulting in cold areas. Likewise, if there is a long-haul distance, cold mix may be placed on joints or transition areas, which will not attain adequate compaction due to lack of heat. Long-term raveling is more difficult to define; however, it is the primary reason for the ter- mination of a PFC service life. If short-term raveling does not occur, Pucher et al. (2004) states that OGFC pavements will deteriorate slowly for the first 5 to 10 years, but deterioration will significantly increase at this point, and raveling is the most commonly observed distress that causes this increase in deterioration. Hamzah et al. (2011) suggested that binder creep (drain- down) is one explanation for raveling. Their research concluded that binder creep (draindown) results in (1) thinner film at the surface that increases aging at the surface and raveling poten- tial, and (2) drained binder that clogs air voids and restricts water flow, further accelerating the oxidative aging process. Once stone particles are dislodged from the surface, other stones are dislodged at an increas- ing rate because there is a lack of support in the stone structure. This, accompanied with the aging (oxidation and hardening) of the binder, results in diminished service life (Molenaar and Molenaar, 2000). The hardening of the binder potentially causes loss in cohesive and adhesive bonds between the aggregate and binder. According to Nicholls and Carswell (2001), the oxi- dation and hardening of the binder may cause the pavement to become brittle at lower tem- peratures, and therefore the strain from the traffic causes the binder-aggregate bond to fail. The indication is that micro-cracks form in the binder or mastic and over time form macro-cracks, which lead to the separation of binder and stone. It is well known that given time and higher temperatures, asphalt binder can be self-healing (Garcia et al., 2012). Due to the severe problem with raveling, the Netherlands built a self-healing test section on Dutch highway A58 in December of 2010. The idea was to use the process of induction to keep macro-cracks from forming in the bond between the binder and aggregate. Steel wool was added as a stabilizing additive and also as an inductive agent. At the end of each winter. The porous pavement was heated via induction energy as a form of preventative main- tenance. This is still an ongoing test, and no results could be found to determine what effect the induction process had on the potential of raveling. Delamination Delamination of PFC pavements is due largely to construction practices and tack rate. In the 2014 survey conducted by NCAT, tack coat rates as low as 0.02 gal/sy were reported. When placing a PFC mix over a dense-graded mix, it is imperative that a heavier tack coat be placed so that an adequate bond can be formed. Since there is less contact area for PFC mixes, a heavier tack coat is needed to compensate than for dense-graded mixes. In the survey conducted by Cooley et al. (2009), the tack coat rate ranged from 0.04 to 0.2 gal/sy. It was also noted that while most agencies specified an emulsion, some required a performance-grade binder instead. If the underlying layer is deemed permeable [>5% air voids (Alderson, 1996)], a slow-setting emulsion should be placed at a rate of 0.05 to 0.10 gal/sy residual asphalt in order to seal the layer. British Columbia requires a tack rate of 0.17 gal/sy for sealing underlying layers (Bishop and Oliver, 2001). Delamination can also be caused by moisture damage and may occur due to the mix being cold when placed on the receiving surface.

Literature Review 27 Top-Down Cracking Cracking of PFC mixes is not as common as raveling, but it does occur. There is much debate about the type of cracking that is predominant in PFC mixes. Top-down, fatigue, and reflective cracking are all possible, but the most common appears to be top-down cracking. Top-down cracking is attributed to the shear stress applied to the pavement by tires. A sche- matic of tire-pavement interaction can be found in Figure 20. Myers et al. (1999) stated that due to the relatively rigid tire wall and the structure of bias ply tires, the ribs of the tire cause an inward shear stress at the surface of the pavement by pulling the ribs into the center of the tire. Chen et al. (2012) concluded that the type of bond material at the interface between the PFC and underlying layer was critical in mitigating top-down cracking. Using a fracture mechanics analysis, Chen et al. (2012) suggested that the use of a thick polymer (SBS) modified tack increased the mixture’s fracture resistance over conventional anionic slow-setting emulsions. Loss of Permeability over Time PFC pavements are designed to reduce noise level, decrease the potential of hydroplaning, minimize splash-spray, and improve friction values; but this is only possible because of the high air void content in the mix. Because of this open void structure, PFC pavements become clogged with dust, silt, and other debris over time. This debris causes a reduction in permeability of the pavement. An impermeable PFC cannot perform according to design, and it loses some of the safety benefits accompanied with the use of PFC. According to the 2014 survey conducted by NCAT, agencies reported design air void contents for PFC mixes ranging from 12% to 26%. These same agencies also report that little to no effort is being put into maintenance practices to address the issue of loss of permeability over time. The use of a standard maintenance practice is imperative in order to maintain the serviceability and function of a PFC. It has been recom- mended that a vacuum sweeper with a high-pressure water system be used at least three times a year to prevent clogging of the pavement (Shirke and Shuler, 2008). The process of vacuum sweeping requires a large maintenance cost and can potentially damage the pavement. None of Figure 20. Load transfer diagram of tire-pavement interface (Baladi et al., 2003).

28 Performance-Based Mix Design of Porous Friction Courses the agencies surveyed indicated that field permeability testing was being conducted to track the performance of the PFC mixtures. Testing performed at the NCAT Pavement Test Track showed a decrease in permeability with traffic loading (Figure 21). However, a large portion of the loss in permeability appears to happen directly after construction. This could be due to densification of the mix under traffic loading, since the goal of PFC compaction is not to achieve a density requirement, but to “seat” the aggregate in place. Rolling and compaction methods for PFCs vary and are dependent on the agency. The graph shows that the trend begins to asymptote at a permeability level that is still relatively high compared to dense-graded mixes. The data imply that even when permeability stabilizes over time, the PFC mixture may still retain some of its functional capacity. In the 2014 NCAT survey, only Nebraska monitored field density by use of cores being removed from the pavement during construction. Mississippi responded that a specific roller pattern was used based on historical results. Alvarez et al. (2009) recommend a field density requirement for PFC mixes in order to prevent over-compaction and non-uniform densification, but additional research would be required to develop appropriate and efficient methods for determining the density of the mix. One method for achieving this is the use of a field permeameter (Figure 22). There is a direct correlation between air voids and permeability of a mix (Brown et al., 2004). A reasonable correlation between labora- tory design air voids and field permeability was produced by NCAT from PFC sections of the test track (Figure 23). According to this correlation, in order to achieve a permeability requirement of 100 m/day, the design air void content must be at least 15%. Argentina, Belgium, and Japan specify that field permeability be tested at the time of construction as a quality control procedure. Similarly, Spain conducts field permeability testing as a method for determining the degree of compaction that has been attained during construction (Cooley et al., 2009). The Danish Road Institute developed a method for determining field permeability of PFC, similar to the field permeameter in Figure 22, but with a much simpler process for determining if the pavement is clogged (Alvarez et al., 2006). A specially designed tube is used to direct 10 cm of water into the pavement. The time required to fully discharge the water is measured to assess the degree of clogging. Table 9 shows how the flow times correlate to the degree of clogging. 0 100 200 300 400 500 600 0 2 4 6 8 10 12 Pe rm ea bi lit y, (m /D ay ) Traffic, Million ESALs Figure 21. Reduction in permeability over time with traffic loading at NCAT Test Track (ESAL = equivalent single axle load).

Literature Review 29 Figure 22. NCAT truck-mounted field permeameter. Figure 23. Correlation between lab air voids and field permeability—NCAT Test Track. TIME TO DRAIN 10 CM OF WATER FROM SPECIAL TUBE PERFORMANCE EXPECTATION Less than 30 Seconds High permeability (expected of new PFC) 30 to 50 Seconds Medium permeability (partially clogged, but can be cleaned) More Than 75 Seconds Low permeability (clogged, cannot be cleaned) Table 9. Danish Road Institute field permeameter performance (Alvarez et al., 2006).

30 Performance-Based Mix Design of Porous Friction Courses Loss of Noise Reduction over Time One of the primary reasons for the loss of noise reduction is clogging of the pores with dust, sand, silt, and other types of debris. As noted in the previous section, there are methods for pre- venting and correcting clogging of the PFC, to a point. A study conducted by NCAT measured the noise levels of seven different PFC mixes across three states and compared that to the air voids of the pavement (Hanson et al., 2004). The air void range of the pavements was from 13% to 20%, which correlated to a 3 dB (A) range in noise results. The results of the study (Figure 24) show that as air void content decreases, the noise level increases. This seems to indicate that if the PFC becomes clogged and the air void content decreases, there will be an increase in the noise level. In another study conducted at the NCAT Test Track by Smit (2008), certain sections of the track were specifically designed with noise mitigation as the primary goal. The study involved testing three different types of asphalt pavement: dense-graded mix, stone matrix asphalt (SMA), and PFC. The results of the study clearly indicated decreased noise levels with PFC pavements, and that a thicker PFC section would provide a greater degree of noise reduction. The maximum aggregate size of the mixture also appears to play a role in noise reduction potential. Isenring et al. (2000) showed a reduced noise level with a smaller maximum aggregate size mixture, even after the pavement had been clogged. In order to avoid clogging, many European countries and South Africa now use a twin-layer porous pavement (Twinlay). Figure 25 shows an illustration of Twinlay pavements in South Africa and the Netherlands. Both layers are PFC designs, but the top mixture has aggregate ranging in size from 4 to 8 mm, while the bottom mixture has particles ranging in size from 11 to 16 mm. The Twinlay approach can also minimize the effect of traffic noise. A study conducted at the NCAT Test Track during the 2006 test cycle used a Twinlay PFC that was laid simultaneously using an imported European paver (Figure 26). This section of the test track was the quietest surface tested for the entire 2-year (10 million ESALs) research cycle. This section was also the most drainable section for this cycle (Willis et al., 2009). During the 2012 NCAT Test Track cycle, noise testing was conducted on a porous mix designed with SBS polymer and another one designed with GTR. The testing was conducted over a 2-year period using the on-board sound intensity (OBSI) procedure. The results for this Figure 24. Relationship of tire-pavement noise to air void content of PFC mixtures (Hanson et al., 2004).

Literature Review 31 South Africa Illustration Dutch Highway – The Netherlands 2.5 cm 4.5 cm Figure 25. Examples of porous Twinlay pavements (Masondo, 2001) (Vejdirektoratet, 2012). Figure 26. Imported European dual paver for Twinlay PFC at NCAT Test Track. testing can be found in Figure 27. It can be seen that as time increases the noise levels increase. This is mostly likely due to binder creep, densification of the pavement with increased traffic, and clogging of the pores with debris. Results show the GTR section, S1, did not dissipate the noise as well as the typical SBS polymer–modified section of W10. Issues Related to Cold Weather Porous pavements in cold regions pose two major concerns, the first of which is the use of ice-control materials such as fine-graded sand and salt (Bernhard and Wayson, 2004). These materials can cause the pavement to clog and subsequently lose permeability and increase the noise level between the tire and pavement. This problem can be avoided by using brine or wetted salt, which will prevent freezing of any surface water with little potential for clogging. The 2014 NCAT survey showed that many state agencies do not use PFC mixtures because of problems encountered with snow and ice removal.

32 Performance-Based Mix Design of Porous Friction Courses One of the northern states, Maine, reported that vacuum sweeping is conducted periodically in the winter months to keep the surface pores from clogging. In addition to the increased need for pavement maintenance in winter months, there are other financial costs associated with winter maintenance of PFC pavements. Due to the open nature of the pavement, a significant portion of the deicing materials settle into the PFC layer. This necessitates larger amounts of deicing material being used on the pavement, which leads to increased cost. France has ceased use of PFC mixes due to the increased winter maintenance cost of an additional 30 to 50% salt (Vejdirektoratet, 2012). Denmark uses special spreaders that apply a dry salt and brine simulta- neously. While this process is more efficient, it still costs 10% to 20% more than winter opera- tions for dense-graded mixes. It was originally thought that since PFC pavements cool faster than dense-graded mix, the PFC would form frost and ice before the dense-graded mixtures, and it would persist longer as well. But research conducted by Lebens and Troyer (2012) on Minnesota Road Research Facil- ity (MnRoad) test sections showed that snow and ice appeared to melt faster on the PFC pave- ment than on the standard dense-graded pavement (Lebens and Troyer, 2012). No evidentiary support could be given to explain the cause of this phenomenon, and it was stated that more research would be needed on the topic. Lift Thickness The ratio of NMAS to lift thickness (t) may be one of the factors leading to raveling. The rec- ommended lift thickness-to-NMAS ratio for coarse-graded mixes and SMA mixes is 4.0 (Brown et al., 2004). The most common use of PFC mixes is for the removal of surface water. This dic- tates that the thickness of the PFC mix should be based on the amount of expected rainfall and the amount of water storage needed (Cooley et al., 2009). While most agencies require a specific thickness for PFC mixes, it is primarily based on past experience rather than a calculation of stormwater run-off. A design chart for PFC lift thickness can be seen in Figure 28. As shown in the figure, the chart takes into account cross-slope, permeability (k), rainfall intensity (I), and length of flow path (L) but does not include NMAS. The 2014 NCAT survey showed ranges in W10 y = 0.331x + 94.336 R² = 0.5172 S1-GTR y = 0.5369x + 94.421 R² = 0.7156 93 94 95 96 97 98 99 So un d In te ns ity , d B( A) OBSI - Global Average W10 S1-GTR Date 9/12 3/13 5/13 8/13 12/13 5/14 8/14 10/14 Figure 27. Loss of noise abatement over time and traffic loading.

Literature Review 33 PFC thickness from 0.5 to 1.25 in. A typical PFC thickness in the U.S. is less than 1.0 inch, while in Europe, the PFC thicknesses range from 1.0 to 2.0 in. (Cooley et al., 2009). The 2-inch lift thickness is from the Netherlands and is dictated by the amount of expected rainfall. Clemson University conducted a study in 2012 with a conclusion that permeability and rainfall intensity had the greatest influence on PFC lift thickness selection, as long as the layer thickness was 2 times the maximum aggregate size. The study used the rainfall intensity at the 90th percentile and a minimum allowable permeability of 164 in./h and concluded that a PFC layer thickness should be between 1 and 1.25 in. thick for a single lane. Most highways that are paved with PFCs are two lanes or more in each direction, excluding the shoulders. The pave- ment thickness for PFC pavements has a linear relationship with pavement width (Figure 29) (Putnam, 2012). Figure 28. Design chart for PFC lift thickness (Cooley et al., 2009). Figure 29. Porous pavement lift thickness based on pavement width (I = 0.37 in./h, K = 164 in./h, ` = 2.0%) (Putnam, 2012).

34 Performance-Based Mix Design of Porous Friction Courses The NCAT Test Track demonstrated differences in performance of PFC mixes placed at different lift thicknesses (Watson, 2014). During the 2009 cycle, Sections S8 and N2 were surfaced with identical PFC mixtures with the only difference being the lift thickness. Sec- tion S8 was placed at 1.3 in. thick and performed well for the entire cycle, while section N2 was placed at 0.8 in. thick. N2 failed during the test cycle and had to be replaced. The thickness-to-NMAS ratio (t/NMAS) was 2.5 for S8 and 1.6 for N2. Additional research may show that increasing the lift thickness of a PFC may increase cohesion by providing more aggregate interlock. Another method for achieving better aggregate interlock may be to use a smaller NMAS mix design.