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.

103 PFC mixtures have been utilized to improve safety and reduce pavement noise and other environmental impacts. These applications are closely related to their high air void content and open-graded aggregate gradation. Despite these benefits, their use has diminished over the years due to durability and service life issues, prompting a need for improvement in mix design and additional laboratory performance testing of PFC mixtures. Several aspects of the current mix design and laboratory performance tests were evaluated in this study to promote the reliable use of these mixtures. In this chapter, the key findings presented in the previous chapters are first summarized and then incorporated in an improved performance-based mix design procedure that considers the balance between PFC mixture functionality and durability. The procedure is included in Appendix A, and further discussion is provided in this chapter. Key Mix Properties Affecting Field Performance Six mix designs were tested and compared in this study to determine if there was a discernable difference in the mix properties for the good and poor performance of those mixtures as experi- enced in the field. The original mix design components are summarized in Table 16. According to agency comments, the PFC mixtures in Virginia and South Carolina failed due to raveling and were replaced within 8 years. The two mixtures in Georgia and New Jersey showed good performance in the field and lasted up to 19 years before being replaced. The PFC mixture with a SBS-modified PG 76-22 in Florida was replaced within 8 years due to raveling, and a comparable PFC mixture with GTR modified binder is still in place after more than 9 years in service in a nearby pavement even though it carries 50% more traffic. With higher air void contents and open-graded aggregate gradations, PFC mixtures have been observed to have performance problems (primarily due to raveling) more often and sooner than dense-graded asphalt mixtures. To address these performance issues, several laboratory tests and performance parameters have been proposed over the years to evaluate the aggregate gradation for stone-on-stone contact and to determine the potential for draindown and durability prob- lems when selecting the optimum binder content of PFC mixtures. Several of these tests and mixtures properties were previously shown in Table 11, and they were evaluated in this study. A summary of the laboratory evaluation results follows. Design Air Voids The most common requirement for a PFC design is a minimum air void content. As shown in Table 83, the AASHTO standard recommends a design air void range of 18%–22% while ASTM requires a minimum design air void content of 18%. Those criteria are typically based on volumetric properties determined by dimensional measurements and calculations. The air voids C H A P T E R 7 Performance-Based Mix Design Procedure

104 Performance-Based Mix Design of Porous Friction Courses measured in this study were based on the CoreLok method, which generally results in about 2% less air voids. Three of the designs in this study, the Georgia design and the two Florida designs, had a design air void content below 18%; however, all of the designs were above 15%. Since the Georgia design and one of the Florida designs were considered good performers, a minimum of 15% air voids was targeted in this study. The South Carolina design with an air void content of 22.2% was outside the acceptable range according to the AASHTO standard. Based on these results, the design air voids can have a significant effect on the field performance, especially when they are above 20% (for Virginia and South Carolina mixes) when measured with the CoreLok device. Thus, a design air void range of 15%–20% when determined with the CoreLok device is proposed for PFC mix design. Film Thickness It was initially estimated that a minimum film thickness of 24 microns would be needed to provide a well performing PFC design although the AASHTO and ASTM standards do not specify a film thickness calculation for design purposes. All of the designs shown in Table 84, with the exception of New Jersey, have film thicknesses greater than the estimated 24 microns. The well performing New Jersey design had a film thickness of 18.6 microns, while the poorly performing South Carolina design had a film thickness of 34.7 microns. This range in film thick- ness values along with conflicting performance suggests that film thickness is not as vital to the performance of PFC pavements as originally believed. Voids in Mineral Aggregate The VMA for each design was conducted across three ACs and a VMA curve was plotted. The initial estimation for this study was that designing near the bottom of the VMA curve may be beneficial to mixture performance, as it is in dense-graded mixtures. There were distinct differ- ences in the designs according to the calculated VMA as shown in Table 85, but the differences Mix ID Criterion Results Recommended RequirementAASHTO ASTM Study Georgia - Good 18%–22% 18% Min. 15% Min 15.4 15%–20% Florida - Good 17.1 Florida - Poor 17.7 New Jersey - Good 19.5 Virginia - Poor 21.8 South Carolina - Poor 22.2 Table 83. Design air voids (percent) and field performance. Mix ID Criterion Results Recommended RequirementAASHTO ASTM Study New Jersey - Good Not Defined Not Defined 24 18.6 Not Applicable Virginia - Poor 25.4 Georgia - Good 27.1 Florida - Poor 32.5 South Carolina - Poor 34.7 Florida - Good 35.9 Table 84. Film thickness (microns) and field performance.

Performance-Based Mix Design Procedure 105 did not distinguish between well and poorly performing designs. The Florida designs and the Georgia design had VMA values of approximately 26%, while the New Jersey, South Carolina, and Virginia designs had VMA of approximately 32%. This large difference suggests that recom- mending a minimum VMA value may not be beneficial, as it had no apparent relationship to good field performance. In addition, the VMA curve for each design was practically flat, with the VMA values across a range of ACs for each design showing no relative change. The change in Gmb for some of the specimens was so minuscule that even some of the data points created an inverse parabolic “curve.” Designing near the bottom of the VMA curve for PFC mixtures, therefore, does not appear to be a reasonable approach for determining AC to optimize mixture performance. Voids in Coarse Aggregate Ratio Initially, all of the designs had a ratio of 1.00 or greater, as shown in Table 86, but after further investigation it was decided to perform the testing according to an alternative method. The ASTM standard states that the coarse aggregate in the mixture is defined as the material retained on the No. 4 sieve. This method was initially performed and resulted in failing VCA ratios for five of the six designs. The proportion of coarse aggregate, according to AASHTO, is defined by the breakpoint sieve. The breakpoint sieve is considered the finest sieve with at least 10% material retained. The AASHTO method was performed using the breakpoint sieve to define the coarse aggregate. Using this method all of the VCA ratios passed the requirement of 1.00 or less, as shown in Table 87. There was a distinct split between the designs, which was the same split as for the VMA calculations. The Florida and Georgia designs had approximate VCA ratios of 0.82 and the New Jersey, Virginia, and South Carolina designs were approximately 0.92. If VCA ratio is to be used in design, it should be stated that the coarse aggregate is defined by the breakpoint sieve which is the finest sieve to have at least 10% aggregate retained. Due to the small portion of fine Mix ID Criterion Results Recommended Requirement AASHTO ASTM Study Florida - Poor Not Defined Not Defined Minimum Point on Curve 26.2 Not Applicable Georgia - Good 26.3 Florida - Good 26.4 New Jersey - Good 31.1 South Carolina - Poor 32.3 Virginia - Poor 32.9 Table 85. Voids in mineral aggregate (percent) and field performance. Mix ID Criterion Results Recommended RequirementAASHTO ASTM Study Georgia - Good 1 1 1 1 Not Applicable Florida - Good 1.07 Florida - Poor 1.07 South Carolina - Poor 1.09 Virginia - Poor 1.18 New Jersey - Good 1.27 Table 86. VCA ratio using No. 4 sieve as breakpoint sieve.

106 Performance-Based Mix Design of Porous Friction Courses aggregate in the PFC designs, the data showed no significant difference in VCA ratio. VCA test- ing also seems impractical for PFC mixtures, since these mixtures are often placed at less than the thickness of two aggregate particles of maximum size. The VCA ratio for the designs varied but did not show a distinct difference between the good and poor designs. Therefore, it is concluded that VCA testing is not necessary for PFC design purposes. Permeability Results Based on the permeability results shown in Figure 45, the minimum amount of air voids needed to achieve a permeability rate of 100 m/day is approximately 17.0%. If the permeability rate of 100 m/day is essential, a minimum design air void content of 17.0% will be necessary. The permeability data for mixes in this study showed that two of the good mix designs had perme- ability rates less than 100 meters/day, as shown in Table 88. There is a direct correlation between air voids and permeability rates, so it is not surprising that those same two designs that had low air voids failed the 100 m/day recommended permeability rate. Since these two designs had good field performance, the proposed rate of 100 m/day may be too high. If a permeability rate of 50 m/day is used, then a corresponding minimum design air void con- tent of 15.0% would be required based on Figure 45. From the data analyzed for this study, the design range of 15%–20% air void content and minimum permeability rate of 50 m/day are pro- posed for PFC designs. However, if an agency experiences high rainfall amounts (such as Hawaii and Louisiana with more than 60 in. of rainfall annually), the minimum permeability of 100 m/day may be more acceptable. Likewise, if an agency has low annual rainfall amounts (such as Nevada with only 10 in. of rainfall annually), a minimum permeability may not be required at all. If a minimum permeability value is specified and the mix design does not meet the require- ment, a coarser gradation will typically increase the permeability. The coarser particles will increase the proportion of interconnected air voids and improve the flow of water. Mix ID Criterion Results Recommended RequirementAASHTO ASTM Study Florida - Poor 1 1 1 0.81 Not Applicable Georgia - Good 0.82 Florida - Good 0.82 New Jersey - Good 0.92 South Carolina - Poor 0.93 Virginia - Poor 0.93 Table 87. VCA ratio using No. 8 sieve as breakpoint sieve. Mix ID Criterion (m/day) Results Recommended Requirement (m/day)AASHTO ASTM Study Florida - Good 100 min. 100 min. 100 min. 77 50 min. Georgia - Good 80 Florida - Poor 107 New Jersey - Good 186 South Carolina - Poor 209 Virginia - Poor 237 Table 88. Permeability (m/day) related to field performance.

Performance-Based Mix Design Procedure 107 Draindown As shown in Table 89, none of the standard designs in this study exhibited draindown using a No. 8 mesh basket. This is most likely due to the fact that all the designs had cellulose fiber added to the mixture except for the binder type experiment in Part 2, Experiment 2. The laboratory observations suggested that fiber improved results with SBS binders but may not be necessary with GTR binders. No changes are recommended for the draindown requirement. Cantabro Stone Loss The Cantabro testing conducted for determining mix durability was fairly successful in distin- guishing the good from the poor performance mix designs. The Cantabro criterion for the ASTM standard is 20% loss for unaged specimens. The AASHTO standard requires a maximum of 15% loss. The initial threshold for this study was a maximum of 20% loss. As shown in Table 90, only the Georgia and New Jersey mix designs passed the 20% maximum loss criterion. The Georgia design (19.3% loss) failed the AASHTO criterion. The Florida good design was close to passing the ASTM standard with a percent loss of 21.9%. Based on these results, use of the maximum Cantabro loss of 20% would have eliminated the poorly performing mixes of South Carolina and Virginia during the mix design procedure. Thus, a maximum of 20% Cantabro loss for unaged specimens is proposed for PFC mix design. Moisture Susceptibility and Mixture Cohesiveness The durability of the PFC designs should also be evaluated in terms of cohesive strength. The ITS is a good indication of a mixture’s strength in terms of cohesiveness, so it was used to determine if cohesiveness could be differentiated between a good and poor mixture. The ITS of a mixture is calculated according to AASHTO T 283. AASHTO requires a minimum TSR of Mix ID Criterion Results Recommended Requirement AASHTO ASTM Study Georgia - Good 0.30% 0.30% 0.30% 0 0.30% max. New Jersey - Good 0 Florida - Good 0 Florida - Poor 0 South Carolina - Poor 0 Virginia - Poor 0 Table 89. Draindown and field performance. Mix ID Criterion Results Recommended Requirement AASHTO ASTM Study New Jersey - Good 15% max. 20% max. 20% max. 10.2 20% max. Georgia - Good 19.3 Florida - Good 21.9 Florida - Poor 23.8 Virginia - Poor 35.1 South Carolina - Poor 37.9 Table 90. Unconditioned Cantabro (percent loss) related to field performance.

108 Performance-Based Mix Design of Porous Friction Courses 0.70 and ASTM specifies a minimum of 0.80. This study shows that the TSR alone is not a good predictor of performance, since the Virginia poor design had a higher TSR value than the New Jersey and Georgia good designs. For this study, it was estimated that a minimum conditioned ITS of 50 psi should be required for design, while no initial unconditioned limit was set. As shown in Table 91, all of the designs except South Carolina had conditioned strengths greater than 50 psi. It is worth noting that PG 76-22 was the standard binder used in this study. If a softer grade binder is used, the tensile strength may be lower. For example, in another study conducted by the researchers, a compari- son of conditioned tensile strength with PG 76-22 and PG 70-28 binders resulted in the mix with PG 70-28 binder having about 30% reduction in conditioned tensile strength. The unconditioned strengths were also analyzed to determine if a minimum unconditioned strength criterion should also be required. The South Carolina poor design, Florida good, and Virginia poor designs had unconditioned ITS values of less than 60, while the Georgia good, New Jersey good, and Florida poor designs had unconditioned ITS values over 70. Based on the test results and PG grade used in this study, a minimum value of 50 psi for con- ditioned strength is proposed. It is further proposed that the 0.70 minimum criterion for TSR be kept as part of the procedure. This will prevent situations where the minimum tensile strength values are met, but the overall TSR may be relatively low. An optional minimum of 70 psi for unconditioned strength may be used; but if a minimum conditioned strength and a minimum TSR value are specified, a threshold value for unconditioned strength should not be needed. Shear Strength A third method used in this study to evaluate the cohesive strength and resistance to shear forces of PFC mixtures was to apply a shear force across a compacted sample. In this procedure, Mix ID Criterion Results Recommended Requirement AASHTO ASTM Study Conditioned ITS, psi South Carolina - Poor Not Defined Not Defined 50 min. 36.7 50 min. Florida - Poor 52.8 Virginia - Poor 53.2 Florida - Good 54 Georgia - Good 57.7 New Jersey - Good 64.5 Unconditioned ITS, psi South Carolina - Poor Not Defined Not Defined Not Defined 45.2 70 min. (Optional) Florida - Good 50.1 Virginia - Poor 59.5 Florida - Poor 72.5 Georgia - Good 74.3 New Jersey - Good 76.2 TSR Florida - Poor 0.70 min. 0.80 min. 0.70 min. 0.73 0.70 min. Georgia - Good 0.78 South Carolina - Poor 0.81 New Jersey - Good 0.85 Virginia - Poor 0.89 Florida - Good 1.08 Table 91. ITSs (psi), TSR, and field performance.

Performance-Based Mix Design Procedure 109 the peak load needed to shear a sample was determined and divided by the cross-sectional area of the specimen. This is a procedure that is easy to run and will not involve costly equipment since it is performed on a Marshall press commonly available in mix design laboratories. As shown in Table 92, a minimum shear strength of 125 psi would eliminate the Virginia and South Carolina poor performance mix designs, but it was not able to identify the Florida poorly performing mix design, which had a shear strength of 170 psi. Based on results of this study, an optional mini- mum shear strength of 125 psi is proposed for PFC mix design. Cracking Tests The OT and I-FIT were used to evaluate mixture susceptibility to cracking. The OT is included in the Texas design procedure but it only provides a criterion for fine-graded PFC designs. The criterion is a minimum of 300 cycles. The initial threshold considered for this study was a mini- mum of 200 cycles. As shown in Table 93, all of the designs met this criterion with the exception of the Florida good design (67 cycles). The same split that was seen for the VMA and VCA ratio was observed in the OT results as well. There was a distinct difference between the designs. The Florida designs and the Georgia design had less than 600 cycles to failure with the OT, while the New Jersey, Virginia, and South Carolina designs had 1,200 or more cycles to failure. The FI calculated from the I-FIT results showed a similar split in data (Table 94), with the lower group averaging 25 and the higher group averaging approximately 45. An initial mini- mum FI of 8 that has been proposed for dense-graded mixtures was selected for this study. All the mix designs passed this criterion. Based on these results, both the OT and I-FIT show a disconnect between the good and poor designs. These results indicate that it may be possible to design mixes that are resistant to cracking but may not be durable mixes. The results of this study do not imply that these tests are not good for dense-graded mixtures. The poor correla- tion between these tests and field performance may be due to the fact that these mixtures did not fail because of cracking, but the primary field distress was raveling. The SCB test for peak load with both notched and no-notch test specimens was also con- ducted for the PFC mix designs. As shown in Table 95, the discernable difference between the Mix ID Criterion Results Recommended RequirementAASHTO ASTM Study Virginia - Poor Not Defined Not Defined Not Defined 121 125 min. (Optional) South Carolina - Poor 122 New Jersey - Good 139 Florida - Good 149 Georgia - Good 162 Florida - Poor 170 Table 92. Shear strength (psi) and field performance. Mix ID Criterion Results Recommended Requirement AASHTO ASTM Study Florida - Good Not Defined Not Defined 200 Min. 67 Not Able to Validate in This Study Florida - Poor 370 Georgia - Good 583 Virginia - Poor 1,291 South Carolina - Poor 1,491 New Jersey - Good 1,866 Table 93. OT Results (cycles to failure) and field performance.

110 Performance-Based Mix Design of Porous Friction Courses mix designs in the modified no-notch I-FIT testing was similar to that of the unconditioned ITS. The peak load for the South Carolina poor, Florida good, and Virginia poor designs was less than 2.300 kN, while the Georgia good, New Jersey good, and Florida poor designs had peak loads greater than 2.900 kN. The peak load of notched specimens provided a better ranking related to field performance with the poor performing South Carolina and Virginia mixes having the lowest values. An excep- tion was that the Florida good performing mix ranked lower than the Florida poor performing mix. The peak load test, however, is similar to the tensile strength test in that it is a measure of the cohesiveness of the mix more so than cracking resistance. For that reason, it is reasonable that the Cantabro, conditioned tensile strength, and notched I-FIT peak load tests ranked the mixes similarly. If the I-FIT peak load of notched specimens is selected as an optional criterion for evaluating mix performance, a minimum value of 1.500 kN can be used. Hamburg Wheel-Tracking Test The HWTT results showed little variability in the results for each design, with the average COV being only 14.2%. The South Carolina design was the only design to fail the criterion of a maximum 12.5 mm rut depth before 20,000 passes, as shown in Table 96. All of the other designs had an average of approximately 7.5 mm of rut depth. The HWTT did not differentiate between all of the good and poor mixtures; it did, however, screen out the poorly performing South Carolina design. In order to achieve a balanced mix design approach, it is proposed that the HWTT, using the current TxDOT criteria for dense-graded mixes, be considered for final veri- fication of PFC mixtures if rutting is a concern. The TxDOT criteria are presented in Table 97. Proposed Revisions to AASHTO PP 77 This project included an evaluation of several mixture properties and performance tests based on the known field performance of six PFC mixtures. Two PFC mixtures failed due to Mix ID Criterion Results Recommended Requirement AASHTO ASTM Study Florida - Poor Not Defined Not Defined 8.0 min. 23.5 Not Able to Validate in This Study Florida - Good 25.2 Georgia - Good 28.7 New Jersey - Good 35.6 Virginia - Poor 57.5 South Carolina - Poor 57.7 Table 94. SCB I-FIT flexibility index and field performance. Mix ID Criterion No-Notch Results Notched Results Recommended Requirement AASHTO ASTM Study South Carolina - Poor Not Defined Not Defined Not Defined 1.645 1.263 1.500 min. (Optional- Notched) Florida - Good 2.057 1.347 Virginia - Poor 2.295 1.255 New Jersey - Good 2.951 1.663 Georgia - Good 3.111 1.796 Florida - Poor 3.286 1.556 Table 95. SCB peak load (kN) and field performance.

Performance-Based Mix Design Procedure 111 raveling and one failed due to cracking and were replaced within 8 years. The other three mix- tures showed good performance and lasted up to 19 years in the field. Based on the laboratory test results summarized in previous sections, the mix properties and test results as well as their respective thresholds that are important to the PFC mix designs are summarized in Table 98. The minimum air void content and permeability requirements are proposed for the drain- age functionality of PFC mixtures. The draindown and TSR requirements are in the current AASHTO standard for PFC mix design. The maximum Cantabro loss and minimum conditioned ITS requirements are proposed to evaluate the PFC mix resistance to raveling and moisture sus- ceptibility. These proposed requirements are incorporated in the draft AASHTO procedure for PFC mix design included in Appendix A. The results from other tests (Table 99) are promising but were not able to clearly separate the poor performance mixtures from the good performance ones in this study. These requirements are not incorporated in the draft standard. Also, as shown in Table 99, the overlay, SCB I-FIT, and HWTT have been proposed and/or used to evaluate the cracking and rutting resistance of dense-graded mixtures. Further validation of these test methods and their associated thresholds is needed for evaluating PFC mixtures. Mix ID Criterion Results, mm Recommended Requirement AASHTO ASTM Study New Jersey - Good Not Defined Not Defined 12.5 mm Max. 6.39 12.5 mm Max. after 20,000 passes for PG 76-22 Florida - Poor 6.81 Virginia - Poor 7.04 Florida - Good 8.47 Georgia - Good 8.99 South Carolina - Poor 15.26a Note: aRut depth recorded at maximum limit of LVDT, which occurred at 3200 passes. Table 96. HWTT rut depth (mm) and field performance. High Temp. Binder Grade Min. Number of Passes @ 0.5 in. Rut Depth, Tested at 122°F PG 64 or lower 10,000 PG 70 15,000 PG 76 or higher 20,000 Table 97. TxDOT rutting criteria for HWTT results. Proposed Property/Test Proposed Criteria Air Void Content, % (measured by CoreLok) 15–20 Permeability, meters/day 50 Min. (agency dependent) Draindown at Production Temperature, % 0.30 Max. Unconditioned Cantabro Loss, % 20 Max. Conditioned ITS, psi 50 Min. (agency dependent based on binder grade) TSR 0.70 Min. Hamburg Rut Depth, mm (PG 76-22 @ 20,000 passes) 12.5 mm Max. ** ** These criteria were proposed by other studies to evaluate dense-graded mixtures. Table 98. Criteria proposed for performance-based PFC mix design.

112 Performance-Based Mix Design of Porous Friction Courses The key mixture properties and performance tests evaluated in Part 1 of this research were utilized in other experiments later in this study to evaluate the effects of added P-200 material, binder modification, and layer thickness on the performance of PFC mixtures. The results of these experiments are summarized in the following sections. Effect of Added P-200 Material The evaluation of the increased P-200 specimens showed marked improvement in terms of durability and cohesiveness of the designs. It is recommended that agencies revise PFC specifica- tions to allow more P-200 in the mix where raveling is the primary form of distress, so long as drainage capability based on air voids and permeability can be attained. In this study the increase in P-200 material reduced the air void content and permeability rates; but for the South Carolina mix, test values were still within acceptable levels. Design Air Voids, Permeability, and Draindown The effects of P-200 on the air void content and permeability are shown in Table 100 and Table 101. For the Georgia mix design, adding more P-200 would cause the design to fail the proposed design requirements for minimum air void content and permeability. For the South Carolina design, adding P-200 would also reduce the air void content and permeability. How- ever, with 4% additional P-200, the South Carolina mix design still passed the proposed permea- bility requirement, and the design air void content fell within the proposed air void requirement for improving mix durability. In addition, adding P-200 should stiffen the mix and further improve resistance to draindown. All the designs passed the draindown requirement, as shown in Table 102. Optional Property/Test Requirement Unconditioned ITS, psi 70 min. Shear Strength, psi 125 Min. Notched SCB Peak Load, kN 1.500 kN min. OT, cycles to failure 200 min. ** SCB I-FIT 25 min. ** These criteria were proposed by other studies to evaluate dense-graded mixtures. Table 99. Criteria that may be agency specific or need further evaluation. Mix ID Proposed Design Requirement Results Improvement Over Control Georgia Control 15–20 15.4 Georgia +2BHF 12.8 No Georgia +4BHF 13.1 No South Carolina Control 22.2 South Carolina +2BHF 15–20 20.7 Yes South Carolina +4BHF 19.3 Yes Table 100. Effect of P-200 on air void content (percent).

Performance-Based Mix Design Procedure 113 Durability and Rutting Resistance The Cantabro, splitting tensile strength, and Hamburg test results are shown in Table 103, Table 104, and Table 105, respectively. The results showed improvement over the control mix design when the additional P-200 material was added as follows: • Adding more P-200 material to the mix designs reduced the Cantabro loss. The South Carolina Control mix design failed the proposed Cantabro loss requirement of 20% maximum, but adding an additional 2% P-200 material reduced Cantabro loss by 50% (from 37.9% to 18.9%), which passed the proposed requirement (Table 103). • It required 4% additional P-200 material for the South Carolina design to meet the proposed requirements for conditioned splitting tensile strength (Table 104). The additional P-200 increased tensile strength by 48%. Adding 2% more P-200 material improved the conditioned tensile strength for the Georgia mix by 50%; but 4% additional P-200 did not improve the strength. • Adding more P-200 material improved the mix rutting resistance (Table 105). The South Carolina mix design with 4% additional P-200 material marginally failed the proposed maxi- mum rutting requirement. Mix ID Proposed Design Requirement Results Improvement Over Control Georgia Control 50 min. 80 Georgia +2BHF 43 No Georgia +4BHF 38 No South Carolina Control 209 South Carolina +2BHF 50 min. 222 Yes South Carolina +4BHF 196 No Table 101. Effect of P-200 on permeability (m/day). Mix ID Proposed Design Requirement Results Improvement Over Control Georgia Control 0.3% max. 0 Georgia +2BHF 0 No Georgia +4BHF 0 No South Carolina Control 0 South Carolina +2BHF 0.3% max. 0 No South Carolina +4BHF 0 No Table 102. Effect of P-200 on draindown (percent). Mix ID Proposed Design Requirement Results Improvement Over Control Georgia Control 20 max. 19.3 Georgia +2BHF 13.0 Yes Georgia +4BHF 9.3 Yes South Carolina Control 37.9 South Carolina +2BHF 20 max. 18.9 Yes South Carolina +4BHF 16.9 Yes Table 103. Effect of P-200 on Cantabro loss (percent).

114 Performance-Based Mix Design of Porous Friction Courses Cracking Resistance The OT data in Table 106 show an increase in cycles to failure with additional P-200 for both the Georgia and South Carolina mix designs. Based on these results, it may be possible to improve resistance to reflective cracking by increasing the cohesiveness of the mixture. The I-FIT results in Table 107 show some improvement in FI for the Georgia design with +4BHF, but +2BHF designs showed a decrease in FI. In addition, Table 108 shows an increase in the notched SCB I-FIT peak load with increased P-200 for both the Georgia and South Carolina mix designs. For South Carolina mixes, only the mix with +4%BHF met the recommended minimum peak load of 1.500 kN. While add- ing additional P-200 material increased the peak load and fracture energy of the I-FIT, it also increased the slope at inflection point. Since FI is more sensitive to the slope, adding additional P-200 material would reduce FI in this case. Mix ID Proposed Design Requirement Results Improvement Over Control Conditioned ITS, psi Georgia Control 50 min. 57.6 Georgia +2BHF 86.4 Yes Georgia +4BHF 82.4 Yes South Carolina Control 50 min. 36.8 South Carolina +2BHF 38.8 Yes South Carolina +4BHF 54.4 Yes Unconditioned ITS, psi Georgia Control 70 min. (Optional) 74.3 Georgia +2BHF 100.3 Yes Georgia +4BHF 99.8 Yes South Carolina Control 70 min. (Optional) 45.2 South Carolina +2BHF 62.0 Yes South Carolina +4BHF 77.3 Yes TSR Georgia Control 0.70 min. 0.78 Georgia +2BHF 0.86 Yes Georgia +4BHF 0.83 Yes South Carolina Control 0.70 South Carolina +2BHF 0.70 min. 0.81 No South Carolina +4BHF 0.63 No Table 104. Effect of P-200 on ITS test results. Mix ID Proposed Design Requirement Results Improvement Over Control Georgia Control 12.5mm Max. for 20,000 passes for PG 76-22 8.99 Georgia +2BHF 5.54 Yes Georgia +4BHF 5.36 Yes South Carolina Control 12.5mm Max. for 20,000 passes for PG 76-22 15.26 South Carolina +2BHF 15.14 Yes South Carolina +4BHF 12.81 Yes Table 105. Effect of P-200 on HWTT rut depth (mm).

Performance-Based Mix Design Procedure 115 Mix ID Proposed Design Requirement Results Improvement Over Control Georgia Control 200 min. (Needs Further Validation) 583 Georgia +2BHF 682 Yes Georgia +4BHF 941 Yes South Carolina Control 200 min. (Needs Further Validation) 1,491 South Carolina +2BHF 2,335 Yes South Carolina +4BHF 1,662 Yes Table 106. Effect of P-200 on OT results (cycles to failure). Mix ID Proposed Design Requirement FI Improvement Over Control Georgia Control 8 min. (Needs Further Validation) 28.7 Georgia +2BHF 21.8 No Georgia +4BHF 39.1 Yes South Carolina Control 8 min. (Needs Further Validation) 57.7 South Carolina +2BHF 36.3 No South Carolina +4BHF 34.8 No Table 107. Effect of P-200 on SCB I-FIT (flexibility index). Since the OT and I-FIT results did not show the same trends, it is difficult to assess the effect of P-200 material on cracking resistance. However, all the mix designs had OT and I-FIT results that are greater than the proposed design requirements. Summary Adding more P-200 material into a mix design will have a positive effect on its resistance to raveling, but it will reduce its air voids and permeability. As was shown in Figure 63, the South Carolina and Georgia mix designs showed increased improvement in terms of percent Cantabro loss until a P-200 content of approximately 6.0% was reached (although some mixtures showed a trend of improvement up to around 8% P-200 material). At this point the percent loss either showed negligible improvement or a small increase. Agencies typically specify a maximum P-200 content maximum of 4% or 5% for PFC designs (Table 2). The AASHTO design procedure specifies a P-200 range of 0% to 4%, and ASTM specifies 2% to 4%. Based on this study, the gradation band for the No. 200 sieve should be revised to 2% to 6%. As shown in the literature review section, there are currently other international agencies that even allow up to 8%. This Mix ID Proposed Design Requirement Results Improvement Over Control Georgia Control 1.500 min. (Optional) 1.796 Georgia +2BHF 2.082 Yes Georgia +4BHF 1.892 Yes South Carolina Control 1.500 min. (Optional) 1.263 South Carolina +2BHF 1.441 Yes South Carolina +4BHF 1.592 Yes Table 108. Effect of P-200 on notched SCB I-FIT peak load (kN).

116 Performance-Based Mix Design of Porous Friction Courses research study indicates that specifying a gradation band of 2% to 6% will provide a potential for more durable mixes. Effect of Fiber The effect of fiber was evaluated based on the Georgia mix design. During the evaluation of binder effects, a control mix which used the standard PG 76-22 with SBS polymer modifier and 0.4% cellulose fiber stabilizer was compared to specimens made with the same binder without the fiber. The fiber appears to contribute to the stiffness of the mix as evidenced by increased tensile strength, reduced rutting in the HWTT results, and increased peak load with the I-FIT (Table 109). It was surprising that Cantabro loss was slightly higher with the fiber than with- out. A precision statement has not been developed for the Cantabro Test, so it may be that the observed difference is within the realm of testing variability. A main reason for adding fiber has been to reinforce the thick asphalt binder film to prevent draindown. Draindown results in this study showed a noticeable difference when the fibers were omitted in that draindown increased from 0.0% to 0.34%. A maximum draindown limit of 0.3% is recommended. There is an alternative approach to the draindown test that includes the amount of material retained on the mesh basket as part of the draindown percentage, as well as the binder that drained through the basket. This alternative approach is similar to the Schellenberg method mentioned in the literature review. This alternative approach was used along with the regular draindown testing for this section to see if there was a significant difference in the amount of draindown recorded. The results showed that the mesh basket retained a large amount of asphalt binder and there- fore increased the “percent draindown” of the designs. This seems to indicate that fiber provides a significant benefit for mixtures that are susceptible to draindown. Further studies should be conducted to determine if this alternative approach to measuring draindown would benefit the industry and if it should be included in the AASHTO procedure. If measures were implemented to keep mixture temperatures consistent through the production process, a lower mixing tem- perature for PFC designs could mitigate draindown. This would allow fiber to be omitted from the mix and therefore create a better performing PFC with limited draindown. One important thing to note is the increase in performance of the SBS design without fiber over the Control mixture with fiber in regards to cracking performance. The fiber in the design Test Georgia Control with 0.4% Fiber Georgia PG 76-22 with No Fiber Cantabro Loss, % 19.3 12.3 Draindown, % 0.00 0.34 Conditioned Tensile Strength, psi 57.6 48.8 Unconditioned Tensile Strength, psi 74.3 61.6 HWTT Rutting at 20,000 Passes, mm 8.99 10.56 OT, Cycles to Failure 583 2137 SCB I-FIT, Peak Load, kN 3.111 3.302 SCB I-FIT, FI 28.7 86.0 Table 109. Effect of fiber stabilizer.

Performance-Based Mix Design Procedure 117 may be increasing the mixture stiffness or reducing the elastic recovery of the mixture and there- fore decreasing the cracking resistance of the design. Since the fiber stabilizer appeared to stiffen and improve strength of the mixture, it was not surprising that the flexibility of mix with the fiber was reduced. Based on the OT and I-FIT FI, there was almost a four-fold increase in flex- ibility values when fiber was omitted. In summary, fiber stabilizer may reduce draindown and increase strength of the mix, but may decrease the flexibility of the mix. This should be a consideration for those areas where crack- ing is a primary form of distress. When draindown is an issue, a dosage rate of 0.3% to 0.4% by weight of total mix has typically been added to improve mix properties. Effect of Binder Modification The effect of binder modification was evaluated based on the Georgia mix design. Fiber was not used in the mixes prepared for this evaluation, except for the Georgia Control mix, to sepa- rate the possible effect of fiber on the performance of these mixtures with various binders. The effect of binder modification had mixed effects on mixture performance when compared to the Georgia Control. The test results are summarized in this section and the performance for each design is based on improvement over the control mix. Design Air Voids and Permeability The air voids of the designs decreased for the PG 82-22 SBS and GTR but increased slightly for the PG 76-22 SBS design as shown in Table 110. The permeability rates of the PG 76-22 SBS design did not change, but the GTR and PG 82-22 SBS had significant decreases in permeability (Table 111). The decrease in air voids and permeability for the GTR design was expected due to the higher AC of the mix. Also, the Control design with fiber had slightly lower air voids but a similar permeability rate compared to the PG 76-22 SBS design. Draindown Asphalt draindown was the most critical evaluation for this experiment. With the removal of the fiber (stabilizing agent) the mixture was solely dependent on the binder modification to prevent draindown. As shown in Table 112, the PG 82-22 SBS design failed the draindown test. This is likely due to the highly modified mix having a mixing and compacting temperature that was 10°F higher than the remaining mixes. For that reason the PG 82-22 mixes were tested at a high temperature of 367°F. The PG 76-22 SBS design showed evident draindown at the high temperature that was marginally outside the recommended criteria. The PG 76-22 SBS design was the same as the Control design with the exception of the fiber. The GTR mix and Control mix with fiber did not have draindown. Mix ID Proposed Design Requirement Results Improvement Over Control Georgia Control 15–20 15.4 Georgia PG 76-22 SBS 16.3 Yes Georgia PG 76-22 GTR 13.9 No Georgia PG 82-22 SBS 14.9 No Table 110. Effect of binder modification on air void content (percent).

118 Performance-Based Mix Design of Porous Friction Courses Durability and Rutting Resistance Table 113, Modified mixes without fiber had lower unconditioned tensile strength than the Control mix. This resulted in higher TSR values and is why TSR alone is not a sufficient criteria for evaluating resistance to moisture damage. The GTR mix is the only mix that had higher conditioned strength than the Control mix. The HWTT results for all mixes met the criteria of less than 12.5 mm rut depth after 20,000 passes. Only the HiMA modified design had less rutting than the Control mix. Table 114 and Table 115 show the effect of binder modification on the Cantabro loss, ITS, and HWTT rut depth. Compared to the Control mix design, the other mix designs showed improvement for the Cantabro loss, with the PG 82-22 SBS binder showing the best improve- ment. Using the GTR binder also improved the resistance to Cantabro loss. The Cantabro data showed improvement for all of the designs with modified asphalt binders. Modified mixes without fiber had lower unconditioned tensile strength than the Control mix. This resulted in higher TSR values and is why TSR alone is not a sufficient criteria for evaluat- ing resistance to moisture damage. The GTR mix is the only mix that had higher conditioned strength than the Control mix. The HWTT results for all mixes met the criteria of less than 12.5 mm rut depth after 20,000 passes. Only the HiMA modified design had less rutting than the Control mix. The OT performance was better for the HiMA and SBS design but not for the GTR. The GTR sample had worse performance than the Control in the OT (Table 116), which seems to trend with the Florida good design (GTR modified) performing so poorly compared to the Florida poor design (SBS modified). The FI data showed increased performance for all of the modified designs (Table 117). Only the GTR modification improved the I-FIT peak load compared to the Control (Table 118). The use of GTR as a binder modification will prove effective in terms of draindown and moisture susceptibility, but based on this data it may be more prone to cracking than polymer- modified designs. The HiMA design, while providing improvement in most of the performance tests, decreased the permeability of the mix while also failing the draindown criteria. With the Mix ID Proposed Design Requirement Results Improvement Over Control Georgia Control 50 Min. 80 Georgia PG 76-22 SBS 79 No Georgia PG 76-22 GTR 33 No Georgia PG 82-22 SBS 37 No Table 111. Effect of binder modification on permeability (m/day). Mix ID Proposed Design Requirement Results Improvement Over Control Georgia Control 0.3% max. 0 Georgia PG 76-22 SBS 0.34 No Georgia PG 76-22 GTR 0 No Georgia PG 82-22 SBS 0.99 No Table 112. Effect of binder modification on draindown (percent).

Performance-Based Mix Design Procedure 119 Mix ID Proposed Design Requirement Results Improvement Over Control Georgia Control 20 max. 19.3 Georgia PG 76-22 SBS 12.3 Yes Georgia PG 76-22 GTR 12.1 Yes Georgia PG 82-22 SBS 4.7 Yes Table 113. Effect of binder modification on unconditioned Cantabro loss (percent). Mix ID Proposed Design Requirement Results Improvement Over Control Conditioned ITS, psi Georgia Control 50 min. 57.6 Georgia PG 76-22 SBS 48.8 No Georgia PG 76-22 GTR 66.8 Yes Georgia PG 82-22 SBS 56.0 No Unconditioned ITS, psi Georgia Control NA 74.3 Georgia PG 76-22 SBS 61.6 No Georgia PG 76-22 GTR 70.2 No Georgia PG 82-22 SBS 63.3 No TSR Georgia Control 0.70 min. 0.78 Georgia PG 76-22 SBS 0.79 Yes Georgia PG 76-22 GTR 0.95 Yes Georgia PG 82-22 SBS 0.88 Yes Table 114. Effect of binder modification on ITS test results. Mix ID Proposed Design Requirement Results Improvement Over Control Georgia Control 12.5 mm max. for 20,000 passes for PG 76-22 8.99 Georgia PG 76-22 SBS 10.56 No Georgia PG 76-22 GTR 10.84 No Georgia PG 82-22 SBS 6.89 Yes Table 115. Effect of binder modification on HWTT rut depth (mm). Mix ID Proposed Design Requirement Results Improvement Over Control Georgia Control 200 min. (Needs Further Validation) 583 Georgia PG 76-22 SBS 2,137 Yes Georgia PG 76-22 GTR 388 No Georgia PG 82-22 SBS 2,877 Yes Table 116. Effect of binder modification on OT results (cycles to failure).

120 Performance-Based Mix Design of Porous Friction Courses HiMA design failing the draindown criteria, the addition of fiber or an alternative stabilizing agent will be needed. HiMA is significantly more expensive than standard SBS binders, and therefore it may not be economical to choose the HiMA over the SBS binder. Effect of Layer Thickness The lift thickness evaluation in Part 2 shows that the thin lifts in which most PFC pavements are placed do not have a significant effect on the durability of the mixture. The ITS of the 4.75 mm dense-graded mixture was significantly larger than the ITS of the PFC designs. However, the varying lift thicknesses were not statistically different. The 9.5 mm and 12.5 mm PFC designs showed no statistical difference in ITS when comparing the various lift thicknesses. However, there is concern that thin layers will result in greater tensile stress at the surface than thicker layers and may potentially lead to increased top-down cracking (Bennert and Cooley, 2014). Balanced Mix Design Framework for PFC Mixtures Figure 84 shows a balanced mix design framework for PFC mixtures based on the results reviewed and obtained in this project. The framework consists of two steps, including functional design and performance verification. The mix design should be porous to drain water while being durable to avoid early failures, especially the primary distress of raveling. A description of each step follows. • A PFC mixture can be designed with a 9.5-, 12.5-, or 19.0-mm NMAS gradation to meet the functional capacity requirements, including air voids and permeability, and to satisfy all per- formance criteria for unconditioned Cantabro loss and ITS. Materials needed for designing a PFC mixture can be selected based on the criteria provided in Section 6 of the proposed proce- dure. A stabilizing agent such as cellulose fiber, mineral fiber, and crumb rubber may be needed to prevent draindown of asphalt binder during transportation and placement. • Design gradation(s) can be selected for available aggregate materials based on the gradation bands shown in Table 1 of the revised procedure (Appendix A). These gradations can be selected depending on the primary objective of the mix. A PFC mixture can be designed with a coarse gradation for permeability and rutting resistance or a fine gradation for noise reduc- tion, and with high P-200 for improved durability. Table 117. Effect of binder modification on SCB I-FIT FI. Mix ID Proposed Design Requirement Results Improvement Over Control Georgia Control 8 min. (Needs Further Validation) 28.7 Georgia PG 76-22 GTR 31.1 Yes Georgia PG 76-22 SBS 38.7 Yes Georgia PG 82-22 SBS 86.0 Yes Mix ID Proposed Design Requirement Results Improvement Over Control Georgia Control 2.750 min. (Optional) 1.796 Georgia PG 76-22 SBS 1.603 No Georgia PG 76-22 GTR 2.015 Yes Georgia PG 82-22 SBS 1.437 No Table 118. Effect of binder modification on notched SCB I-FIT peak load (kN).

Performance-Based Mix Design Procedure 121 • Optimum binder content is determined for the selected gradation to meet the air void, Cantabro loss, and permeability criteria that are required to achieve the functional capability of a PFC mixture, as shown in Table 2 of the revised procedure (Appendix A). If test results show that the mix design meets the air void, permeability, and Cantabro requirements, perfor- mance testing can be conducted for the design. Otherwise, the mix is redesigned with a new trial aggregate gradation. • Performance testing is then conducted on the mix design that meets the functional require- ments. Draindown, rutting (optional), cracking (optional), and moisture susceptibility are conducted. If test results show that the mix design meets all the performance requirements shown in Table 2 of the revised procedure (Appendix A), the mix design is finalized for report- ing. Otherwise, the mix design needs to be adjusted. • In addition to changing the trial aggregate gradation, other material properties can be varied to achieve a successful PFC mix design. Both the binder content and percent passing the 0.075 mm (No. 200) can be varied to adjust the air voids and Cantabro loss of a PFC mix- ture. When needed, binder modification, including standard polymer modification, GTR modification, and high polymer content modification, can be used to improve performance properties of PFC mixes. Finally, stabilizing agents including cellulose fiber, mineral fiber, and crumb rubber can help prevent draindown of asphalt binder. Conduct Mixture Performance Testing Rutting (Optional) HWTT Cohesive Strength Indirect Tensile Strength Shear Strength (Optional) Performance Tests Passed? Final Mix Design NO YES Balanced Mixture Design Framework for PFC Mixtures Functional Capability Design with Performance Verification Select Trial Gradation for PFC Mixture Determine Optimum Binder Content based on Durability and Functional Application Redesign Select Materials (binder, aggregate, anti-strip, stabilizing agent) Air Voids Volumetric 15% Va Durability 20% Cantabro loss Functional Application Passed? Permeability 50 m/day Permeability YES NO Cracking (Optional) I-FIT FI Draindown Figure 84. Balanced mix design concept for PFC mixes.