LABORATORY EVALUATION OF AN INNOVATIVE AND COST EFFECTIVE HIGH - FRICTION SURFACE TREATMENT By Xiaoyu Wendi Chen A THESIS Submitted to Michigan State University i n partial fulfilment of the requirements for the degree of Civil Engineering Master of Science 2018 A BSTRACT LABORATORY EVALUATION OF AN INNOVATIVE AND COST EFFECTIVE HIGH - FRICTION SURFACE TREATMENT By Xiaoyu Wendi Chen High Friction Surface Treatment (HFST) is a surface treatment that can effectively improve the frictional characteristics of pavements and enhance the users safety on the road. Among all existing application s of HFSTs, bauxite and oil - based epoxy resin are the most commonly us ed materials in the United States. However, bauxite and conventional epoxy resin are not cost - effective and provide limited preservation benefits to the pavement structure with existing distresses ( e.g. , top - down fatigue). In this study, the performance of a new and cost - effective HFST that uses waterborne epoxy, emulsified asphalt and corundum sand was investigated through a battery of laboratory tests. The performance of the innovative HFST was evaluated and compared with common HFSTs (copper slags and ba uxites) used in USA with respect to three aspects : (i) skid resistance improvement before and after the treatment, (ii) durability to environmental effects (moisture induced damage and freeze - thaw cycles), and (iii) the effect of application on surfaces wi th existing top - down cracking. The results showed that the new HFST with waterborne epoxy and corundum was able to improve the skid res istance similar to or better than the conventional HFSTs. The deterioration rate on skid resistance of new HFST after fre eze - thaw cycles and damage were found to be faster than bauxites but slower than copper slag. In addition, since corundum sand used in the low - cost HFST is much finer than either copper slags or bauxites and viscosity of waterborne epoxy is less than oil - b ased epoxy 1.5 ± 0.1 mm and relief the further propagation of the existing top - down cracks. Copyright by XIAOYU WENDI CHEN 20 18 iv ACKNOWLEDGMENTS First and foremost, I would like to sincerely express heartfelt gratitude and deep regards to my academic advisor and mentor Dr. M. Emin Kutay for his constant patience, motivation, immense knowledge , and guidance throughout all the time of research and wr iting of this thesis . Besides my advisor, I would like to thank the rest of my thesis committee: Dr. Chatti and Dr. Haider, for their insightful comments and encouragement , but also for the questions which incented me to widen my research from various pe rspectives. My sincere thanks also go to my fellow research group mates Dr. Antonio (Mike), Yogesh Kumbargeri, Akse l Seitllari, Angela Farina, and Mahdi Ghazavi for the time we were working together in the laboratory, for the assistance I received from th em technically and mentally. Last but not least, I would like to my family: my parents, my sister, and my boyfriend for the love and the spiritual support throughout writing this thesis and my life in general . v TABLE OF CONTENTS LIST OF TABLES ................................ ................................ ................................ ........................ vii LIST OF FIGURE S ................................ ................................ ................................ ..................... viii 1. INTRODUCTION ................................ ................................ ................................ ................... 1 2. LITERATU RE RVIEW ................................ ................................ ................................ .......... 2 2.1 High Friction Surface Treatments Procedures ................................ ................................ . 2 2.2 Testing Devices for High Friction Surface Treatments ................................ ................... 3 2.2.1 Field Testing Devices for Friction ................................ ................................ ............ 3 2.2.2 Laboratory Friction Measuring Devices ................................ ................................ ... 7 2.2.3 Laboratory Testing on Binders used in HFSTs ................................ ......................... 8 2.2.4 Laboratory Testing on Aggregates ................................ ................................ ............ 9 2.3 Effects of Aggregates ................................ ................................ ................................ ..... 10 2.3.1 Effect of Aggregates on Friction Loss ................................ ................................ .... 11 2.3.2 Effect of Aggregates on Texture Loss ................................ ................................ .... 12 2.3.3 Polishing Resistance of Aggregates ................................ ................................ ........ 13 2.4 Effects of Binders ................................ ................................ ................................ ........... 13 2.4.1 Effect of Binder Thickness on Aggregate Loss ................................ ...................... 14 2.4.2 Effect of Binder Type and Temperature on Aggregate Loss ................................ .. 14 2.4.3 Effect of Binder Type and Temperature on Texture Loss ................................ ...... 14 2.4.4 Thermal Compatibility of B inders ................................ ................................ .......... 15 2.4.5 Other Properties of Binders ................................ ................................ .................... 15 2.5 Limitations of High Friction Surface Treatments ................................ .......................... 16 2.5.1 Effect on Crack Relief ................................ ................................ ............................ 16 2.5.2 Cost ................................ ................................ ................................ ......................... 16 2.6 Synthesis of the Previous Work Motivation of the Current Study ................................ . 18 3. OBJECTIVE AND RESEARCH PLAN ................................ ................................ ............... 19 3.1 Skid Resistance ................................ ................................ ................................ .............. 19 3.2 Durability under Freeze - Thaw Cycles ................................ ................................ ........... 19 3.3 Crack Retardation ................................ ................................ ................................ ........... 20 4. MATERIALS ................................ ................................ ................................ ........................ 21 4.1 Aggregates ................................ ................................ ................................ ...................... 21 4.1.1 Gradation ................................ ................................ ................................ ................. 22 4.1.2 Origin s of sources and Components of Aggregates ................................ ................ 23 4.2 Binders ................................ ................................ ................................ ........................... 23 4.2.1 Components and Mixing ................................ ................................ ......................... 23 4.2.2 Physical and Chemical Properties ................................ ................................ ........... 24 4.3 Application Combination ................................ ................................ ............................... 24 4.4 Sample Preparation and Application Rates ................................ ................................ .... 26 vi 5. EXPERIMENTAL PROGRAM ................................ ................................ ............................ 28 5.1 Skid Resistance ................................ ................................ ................................ .............. 28 5.2 Durability Subjected to Moisture Induced Damage ................................ ....................... 29 5.2.1 Freeze - Thaw Cycles ................................ ................................ ................................ 30 5.2.2 Hamburg Wheel Tracking Device ................................ ................................ .......... 31 5.2.3 Sweep Test ................................ ................................ ................................ .............. 32 5.3 Effect on Crack Retardation ................................ ................................ ........................... 34 6. RESULT AND DISCUSSION ................................ ................................ .............................. 36 6. 1 Initial Skid Resistance ................................ ................................ ................................ .... 36 6.2 Skid Resistance after Moisture Induced Damage ................................ .......................... 38 6.2.1 Skid Resistance after Hamburg Wheel Tracking Device ................................ ....... 38 6.2.2 Skid Resistance after Freeze - Thaw Cycles and Sweep Test ................................ .. 40 6.3 Crack Initiation and Propagation ................................ ................................ .................... 44 6.3.1 Semi - Circular Bending Test ................................ ................................ ................... 44 7. CONCLUSIONS ................................ ................................ ................................ ................... 47 REFERENCES ................................ ................................ ................................ ............................. 49 vii LIST OF TABLES Table 1 Summary of Friction Measuring Devices ................................ ................................ .......... 7 Table 2 Examples of High Friction Surface Treatment Unit Cost ................................ ................ 17 Table 3 Gradation of the Aggregates Used in the Study ................................ .............................. 22 Table 4 Binder and Aggregate Application Rates (g/m 2 ) ................................ ............................. 26 Table 5 T - Test Results for before and after the Treatment ................................ ........................... 38 Table 6 Statistical Significance of Measurements (within cycles) after Conditioning Cycles ..... 41 Table 7 Comments on T - Test (within Cycles): H 0 = Unequal Mean ................................ ............ 42 Table 8 Result of T - Test on the BPN Measurements (for Certain HFST) ................................ ... 43 Table 9 Comments on T - Test on the BPN Measurements (for Certain HFST), H 0 = Unequal Means ................................ ................................ ................................ ................................ ............ 43 Table 10 SCB Test Results Summary ................................ ................................ ........................... 45 viii LIST OF FIGURE S Figure 1 British Pendulum Tester ................................ ................................ ................................ ... 8 Figure 2 Philosophy behind SCB Tests ................................ ................................ ........................ 20 Figure 3 Aggregates Tested in the study (a) Corundum (b) Copper Slag (c) Bauxite 1 and (d) Bauxite 2 ................................ ................................ ................................ ................................ ....... 21 Figure 4 Aggregate Gradation ................................ ................................ ................................ ...... 22 Figure 5 Diagram of Two HFST Systems Studied (a) Two - Layer System (b) One - Layer System (NOT TO SCALE) ................................ ................................ ................................ ........................ 25 Figure 6 Samples with Application (Unconditioned) (a) HFST 1, (b) HFST 2, (c) HFST 3 (d) HFST 4 (left to right) ................................ ................................ ................................ .................... 26 Figure 7 SCB Specimens with HFST s: (a) HFST 1 Top View, (b) HFST 1 Side View, (c) HFST 4 Top View, (d) HFST 4 Side View ................................ ................................ ............................. 27 Figure 8 Laboratory Setup of BPT ................................ ................................ ................................ 29 Figure 9 Processes of a Freeze Thaw Cycle ................................ ................................ ................. 30 Figure 10 Hamburg Wheel Tracking Device (a) Original, (b) Modified ................................ ..... 31 Figure 11 Specimen in HWT Molds ................................ ................................ ............................. 32 Figure 12 Original Sweep Tester (a) Device (b) Regular Plastic Brush ................................ ....... 33 Figure 13 Substituting Tools for Plastic Brush (a) V - S haped Knife, (b) Wire Brush .................. 33 Figure 14 Zoom - in on the Notch for HFST 1 ................................ ................................ ............... 35 Figure 15 BPN on Treated and Untreated Specimens without Conditioning ............................... 36 Figure 16 Result of HWT for HFST 1 and HFST 2 ................................ ................................ ..... 39 Figure 17 Flat Tire of HWT after 325 Cycles on HFST ................................ ............................... 39 Figure 18 BPN Measurements on Treated Samples after Each FT Cycle ................................ .... 41 Figure 19 SCB Test Result of HFSTs (a) HFST 1, (b) HFST 2, (c) HFST 3, (d) HFST 4 .......... 46 1 1. INTRODUCTION C rashes, injuries, fatalities and cost associated with them have always been significant concerns for state Departments of Transportation and local road agencies. Even though responsible driving practices are crucial for traffic safety, road safety is not only limited by human factors. The condition of pavement surface also plays a critical role in enhancing the pavement - tire interaction. Nowadays, High Friction Surface Treatment (HFST) is one of the surface treatment practice s that have been used in many states to restore the surface friction and reduce the crash rate. Successful HFST applications are available in thirty four (34) states in the US , most notable projects being in Pennsylvania, Iowa, South Carolina, Kentucky, and Tennessee. However, the cost of the HFST is typically higher than other surface treatments since high - qu ality materials need to be used (typically epoxy and bauxites). Therefore, the challenge becomes to lower the cost of HFST, yet the performance of HFST is maintained . Recently, an innovative HFST was developed in China, where the material and labor cost is approximately one third (1/3) of HFSTs available in the U SA. The special ized truck developed for this application has the capacity to apply 6700 square yards with one load , a nd finished HFST has two (2) lifts of material. However, its performance needed to be evaluated and compared with the traditional practices. The answer to the problem that if the new and low - cost HFST can perform similar to or better than the conventional application is expatiated in the thesis. The performance s of HFSTs are evaluate d in terms of skid resistance, durability of materials, and the ir effect on crack relief. 2 2. LITERATURE RVIEW 2.1 High Friction Surface Treatments Procedures High friction surface treatments (HFSTs) have been implemented nationally and internationally to restore skid resistance at the locations where high friction between the tire and pavement are needed [1]. The HFST ingredients include highly polishing - resistant aggregates and polymeric resin or other types of binder [2]. Depending on the temperature of the participating resin binder during the construction, HFSTs can be categorized into two groups: hot - applied surface treatments and cold - applied surface treatments [3]. For the hot - applied HFSTs, the pre - combined material with aggregates and binder is he ated and spread across the surface of the pavement slab. The mixture of granular material and binder are premixed at a high temperature and packaged in bags due to the fact that the binder used for hot applied HFSTs are thermoplastic binders. On the other hand, cold - applied surface treatments use chemically curing binders such as epoxy and polyurethane. Aggregates and binders are provided in separate containers, the components of the resin for spraying on the substrates were mixed and installed on the surfa ce before the aggregates are distributed across the slab. In the field, cold - applied surface treatment is more practical and convenient since this process is easier to handle and it required less equipment for construct ion . Typical application procedures f or cold - applied treatments contain five steps [1], as listed below : Preparation of the surface : The surface of the pavement must be dry and clean. grease, ice, and salt m ust be removed from the surface before applying the treatment on it. In general, tools like brooms, vacuum, and/or shot blasting are used to clean the 3 surface. It is suggested that the temperature of the surface for the application should be monitored and measured to ensure the minimum installation temperature . Blending of the epoxy components : The two components (epoxy resin and curing agency) of the epoxy binder are mixed together according to a specified proportion either by weight or by volume. The amou nt of blended binder is determined by the binder application rate. Epoxy resin and curing agency should be mixed thoroughly to obtain a uniform binder material property. This can be accomplished manually or automatically. Spread of epoxy binder : The mixed binder can be spread on the surface manually or automatically. The binder application rate should be controlled and ensured no matter how it is applied. Especially in the case of manual spread of epoxy resin, grids should be laid out on the surface with t ape before the application to guarantee the application rate is within the acceptable range. Application of aggregates : Granular material should be applied after the epoxy is spread with no delay manually or automatically according to the application rate as well. Usually, shovels and brooms are used to spread the aggregates uniformly on the binder if it is done manually. Curing of the system : The excess materials, binder and/or aggregates, should be removed from the surface before the pavement with treat ment is left for curing. A typical curing period for HFST ranges from two (2) to four (4) hours. 2.2 Testing Devices for High Friction Surface Treatments 2.2.1 Field Testing Devices for Friction Since high friction surface treatment is designed to improve the tire/pavement skid resistance, the most significant performance parameter is related to frictional characteristics . As part of a 4 National Cooperative Highway Research Program (NCHRP) project, various pavement friction testers were evaluated [4]. The co mparison among different devices to measure pavement friction was also summarized by Rajaei et al. [5]. In general, field pavement friction measuring devices can be categorized into four (4) types, based on the mode of operation: locked wheel, side force, f ixed slip, and variable slip. Examples with descriptions for each type are documented in this section. Table 1 summarizes the different testing devices used around the world to measure tire/pavement interface friction [4]. Locked Wheel Mode Field friction testers with locked wheel mode measure the steady - state friction force on a locked test wheel as it is dragged under a constant load at a constant speed over a wet pavement surface [ 6 ]. Locked w heel m ode simulates emergency braking without anti - lock brake s [4]. Different devices are available around the world. Following are some examples of field friction testers with locked wheel mode. - ASTM E274 Trailer : The trailer with one or more test wheels is incorporated into an automotive vehicle. The apparatus contains a transducer, instrumentation, a water supply, and proper dispensing system, and actuation controls for the braking of the test wheel [ 7 ]. - Diagonal Braked Vehicle : National Aeronautics and Space Administration (NASA) used the diagonal braked vehi cle to evaluate the vehicles stopping capability and tire friction performance in the 1970s [ 8 ]. The measurements were recorded at high speed with two diagonal wheels locked and two remaining wheels free rolling. - Japanese Skid Tester (Dynamic Friction Test er) : This device uses a disk that spins with its plane parallel to the test surface. There are three rubber sliders attached to the lower 5 surface of the disk [8]. Friction between the rubber sliders and the wet testing surface when the torque is generated as the sliders are compressed by the weight of the device. - French ADHERA Equipment : French ADHERA measures the longitudinal braking force coefficient with a locked wheel and PIARC smooth tire. The pavement friction is then determined by the braking force c oefficient. - Polish SRT - 3 : The polish SRT - 3 device is similar to ASTM E274 Trailer which is towed behind the measuring vehicle . Measurements on resistive drag force by braking to the system to a full stop are recorded at three test speeds: 30, 60 and 90 km per hour . - Others: Other pavement friction - measuring devices with locked wheel mode are also available in other countries. For instance, Skiddometer BV - 8 is used in Sweden, and Stuttgarter Reibungsmesser (SRM) is available in Germa ny. Side Force Mode Testers with side force mode do not lock the test wheel. Instead, they measure the side force friction of paved surfaces by pulling two or more freely rotating test wheels [ 6 ]. Side force mode measures the ability to maintain control in a curve. Following are the examples for this category. - MuMeter : The MuMeter consists of a small three - wheeled trailer with electronic measuring systems which is located in the towing vehicle . - Odoliograph : Belgian Road Research Center uses a device named O doliograph to measurement the pavement skid resistance. The measurement is recorded with a standardized smooth tire yawed to an angle of 20 degrees with respect to the surface, which will potentially skid on a wet condition. - SCRIM : The machine operates by applying a freely rotating fifth wheel on to the road surface under a known load at an angle of 20 degrees to the direction of travel. There is a 6 water supply system by a controlled jet that wets the road surface directly in front of the test wheel. The Si deway Force Coefficient value is calculated based on the output of the test which is the frictional resistance to skidding of the test wheel by transducers. - Stradograph : Stradograph is a Norwegian friction testing device that measures the side force of the system . Then the friction is calculated as the ratio between the applied vertical load onto the measuring wheel and the horizontal force acting on the measuring wheel. Fixed Slip Mode Testers with fixed slip mode consist of test wheels rotating with a con stant slip. For instance, the wheel is lightly braked to provide a difference in velocity between the test wheel and the speed of the tester. An example of this type of tester is shown below. - GripTester : The measuring wheel is mounted on a stub axle. Strai n gauges on this axle measure load and drag force. Then the level of friction is calculated by dividing drag force by the load. Variable Slip Mode Testers with variable slip mode are related to the braking with anti - lock brakes. - ROAR : The ROAR unit is spec ially designed to measure the pavement surface characteristics on the road. The variable slip technique is employed in the ROAR unit. The linear braking of the test wheel from free rolling to be almost entirely locked allows the recording the entire curve describing the friction process between the tire and the road surface. 7 Table 1 Summary of Friction Measuring Devices Device Operational Mode Country of Manufactory ASTM E274 Trailer Locked Wheel United States Diagonal Braked Vehicle Locked Wheel United State (NASA) Dynamic Friction Tester Locked Wheel Japan LCPC ADHERA Locked Wheel France Polish SRT - 3 Locked Wheel Poland Skiddometer BV - 8 Locked Wheel Sweden Stuttgarter Reibungsmesser Locked Wheel Germany MuMeter Side Force United Kingdom Odoliograph Side Force Belgium SCRIM Side Force United Kingdom Stradograph Side Force Denmark GripTester Fixed Slip Scotland DWW Trailer Fixed Slip The Netherlands Runway Friction Tester Fixed Slip United States Saab Friction tester Fixed Slip Sweden Skiddometer BV - 11 Fixed Slip Sweden ROAR Variable Slip Norway Oscar Variable Slip Norway SALTAR Variable Slip Norway IMAG Variable Slip France Komatsu Skid Tester Variable Slip Japan 2.2.2 Laboratory Friction Measuring Devices There are two testing devices used in the laboratory: Dynamic Friction Tester and British Pendulum Tester. Both devices can be used for field testing also ; however, they are categorized into laboratory testing devices because of their uniqueness th at they are portable and convenient to be utilized in a laboratory setting. Dynamic Friction Tester is included in the previous sub - section, and British Pendulum Tester (BPT) is described here. British Pendulum Tester British Pendulum Tester is a friction - measuring device developed by the British Transport Research Laboratory as shown in Figure 1 . It is a dynamic pendulum i mpact - type tester used to measure the energy loss when a rubber slider edge is propelled over a test surface [ 6 ]. 8 Figure 1 British Pendulum Tester 2.2.3 Laboratory Testing on Binders used in HFSTs Binders used in HFST should have sufficient viscosity to hold aggregate together in place. They should also provide enough strength to withstand the deterioration caused by traffic and weathering. The material and physical properties of binders should make them convenient and feasible fo r construction. Many Departments of Transportation have a similar set of requirements on binders to make sure that binders can be appropriately used for HFST [ 10 ] , [ 1 1 ] , [ 1 2 ]. - Viscosity (ASTM D2556): In the test, a rational viscometer is used for the measurement of the apparent viscosity of Newtonian and/or non - Newtonian liquid. Usually , these liquid are shear - rate dependent adhesives. - Gel Time (AASHTO M235, ASTM C881) : The test covers the method to d etermine the time it needs for an epoxy resin system to gel after epoxy and hardener are mixed. - Ultimate Tensile Strength (AASHTO M235, ASTM D638): The tensile properties of an epoxy binder with a standard dumbbell - shaped test specimen . 9 - Elongation at Break point (AASHTO M235, ASTM D638): Elongation of tested material is reported as part of the tensile test. The percent increase in length compared to the original length of test specimens is record ed when the sample break. - Durometer Hardness (ASTM D2240): In t he test, the indentation hardness of epoxy resin binder is determined using a durometer. - Compressive Strength (AASHTO M235, ASTM C579): The procedures of how to determine the compressive strength of epoxy resin with aggregates are covered. - Cure Rate (ASTM D1640): The time intervals for the substances to be at various stages (i.e., dry to touch, dry to hard) are recorded. - Water Absorption (AASHTO M235, ASTM D570): T he amount of water absorbed by the material by weight is determined by calculating the percent increase in weight and percentage of soluble matter loss. - Adhesive Strength at 24 Hours (ASTM C1583): The adhesive strength is tested similar ly to tensile strength. 2.2.4 Laboratory Testing on Aggregates Aggregates for HFST are supposed to provide adequat e surface skid resistance. Besides skid resistance, they should also have the ability to endure the polishing effect from the traffic and be resistant to the disintegration caused by weathering [1]. There are some tests commonly performed in the lab on HFS T aggregates to determine whether or not the applied aggregates are able to retain its texture to provide surface skid resistance to traffic for as long as possible. - Standard Practice for the Accelerated Polishing of Aggregates Using the British Wheel (AA SHTO T279, ASTM D3319) : In the test, the initial and conditioned frictional characteristics are measured by British Pendulum Tester. 10 - Resistance to Degradation and Abrasion (AASHTO T96, ASTM C131) : The resistance to degradation of coarse aggregates with a m aximum size of 37.5 mm is determined. The Los Angeles testing machined is used to apply abrasion. - Soundness of Aggregates by Freeze - thawing (AASHTO T103 - 91, ASTM C88) : the soundness of aggregate subjected to wea thering is estimated. Aggregate samples are r epeatedly immersed in saturated sodium or magnesium sulfate solution and dried by an oven. The mass loss of aggregates before and after the test is reported. 2.3 Effects of Aggregates National Center for Asphalt Technology (NCAT) performed a comprehensive study on HFST alternative aggregates on flexible pavements to evaluate different aspects of aggregates [1 3 ] . Both field and laboratory data were collected and analyzed. Eight types of aggregates were studied in this project: bauxite, granite, flint, basalt, silica sand, steel slag, emery, and taconite. The NCAT study concluded the effect of aggregate types on f riction reduction and texture loss (see the following portion). L ater on, the Florida Department (FDOT) of Transportation and the Federal Highway Administration (FHWA) have published a report about the evaluation of different aggregates for HFST [1 4 ] . Three types of aggregates were tested in the study. Three aggregates are calcined bauxite from China, calcined bauxite from India, and an unknown aggregate type from the United Kingdom. In New Zealand, the Transport Agency also reported the effect of aggregates on surface friction. The detailed types of aggregate used in the study were not identified even though the sources of aggregates were provided. A research group from China also compared the performance of four types of aggregates: bauxite, granite, limestone, and basalt [1 5 ] . Findings and conclusions from all the studies mentioned above are detailed below . 11 2.3.1 Effect of Aggregates on Friction Loss Three - study. TWPD is a polishing device with three wheels which tracked a circular path that matched the circular path used by the DF T. Water spray system was accomplished by attaching pipes to the frame that connects three wheels. In the NCAT study [1 3 ], the friction during TWPD conditioning was measured three times by the dynamic friction tester, and the change in friction was compute d twice. In the study, water was continuously introduced in the conditioning process on samples with different aggregates. Then the dynamic friction tester was run at 40 kilometers per hour to measure the surface friction. After 70,000 TWPD conditioning c ycles, the loss of friction (in increasing order) was ranked as follows: bauxite (11%), taconite (17%), slag (18%), basalt (19%), silica ( 21%), flint (23%), granite (29%), and emery (31%) However, the ranking changed after 140,000 condition cycles to: gran ite (4%), flint (5%), silica (5%), slag (5%), emery (5%), taconite (6%), bauxite (7%), and basalt (9%). Overall, the reduction in friction for bauxite was found to be the lowest among all aggregates. In other words, the result of this particular study show ed that bauxite had the highest skid res istance for a long - term [1 3 ]. Besides , The treatments, calcined bauxite is the only type of aggregates allowed in the practices [ 10 ]. The research project on evaluation of high friction surface treatments conducted by FDOT also revealed that two types of calcined bauxites with similar aluminum oxide content s (~ 85% by weight) performed very well with respect to wet friction after 2,000, 30,000, and 100,000 polishing cycles by TWPD. The friction loss of calcined bauxites was almost 5 times less than the aggregates with lower aluminum oxide content but higher s ilica content [ 1 4 ]. 12 Another study from the New Zealand Transport Agency showed that aggregates with higher resistance to polishing also had a higher initial value of skid resistance antecedent to any polishing condition [1 6 ]. The conditioning process was conducted on the samples and water that was used to polish the aggregates. The conditioning temperatures ranged from 4 °C to 55 °C, and the polishing durations were 15 - min intervals up to 7 hours. It was discovered in the study that the aggregates with hi gher initial skid resistance had a greater total skid resistance loss before the sample reached its equilibrium skid resistance level where the skid resistance did not decrease much [1 6 ]. 2.3.2 Effect of Aggregates on Texture Loss In the NCAT study [ 1 3 ] , t he surface texture was also measured three times and the texture loss was calculated twice by the Circular Texture Meter (CTM). During the first 70,000 cycles of conditioning using TWPD , there was a similar surface texture loss (20% to 30%) for all eight k inds of aggregates. Since the texture loss after second conditioning period was too low (0 to 6%) to be identified as a major change, there was no significant change in surface texture after the second 70,000 cycles as concluded in the report [1 3 ]. The stu dy concluded that the friction measurements by DFT did not correlate with the CTM measurements, which means that the texture loss was not an indicator of skid resistance of high friction surface treatments. In the study from China [1 5 ], the relationship b etween skid resistance measured by BPT and macro texture as well as the micro texture was established. For high friction surface treatments, the early skid resistance was significantly impacted by the macro - texture of aggregates, and long - term skid resista nce of HFST was greatly influenced by the micro - texture of aggregates. In that study, the macro - texture characteristics of aggregate were captured using image analysis, while the micro texture characterization was accomplished by using a 3D color 13 laser mic roscope system [1 5 ]. The polishing process consisted of two phases: one phase where corn emery was used to apply abrasion for 40,000 cycles to eliminate the manufacturing effect on the aggregates and a phrase of polishing with emery flour to damage the sur face. It was found in the study that bauxite had the lowest loss in macro texture and micro texture among four types of aggregates [1 5 ]. 2.3.3 Polishing Resistance of Aggregates The resistance of specimens with different aggregates to degradation by abrasi on in the Micro - Deval Apparatus was performed by the FDOT and FHWA study. The angularity of aggregates was determined using the Aggregate Imaging System (AIMS) based on 2D aggregate images. Among three types of aggregates, bauxites from China and India hav e a relatively higher resistance to degradation by abrasion than the aggregate from the UK, even though all three types of aggregate has similar unpolished and polished average angularity [1 4 ]. 2.4 Effects of Binders There is a comprehensive study performe d in cooperation with the Florida Department of Transportation and the Federal Highway Administration to evaluate the performance of alternative aggregates and materials for high friction surface treatments [ 1 4 ]. The study compares the performance of four different kinds of binders regarding the mass and texture loss of HFST after traffic as well as the thermal compatibility of HFST on both asphalt and concrete substrates [1 4 ]. These four types of binders were epoxy resin, epoxy urethane resin, cresol modif ied epoxy resin, and acrylic polyester resin. In the study, the binder was spread over different types of substrates in different ways. The aggregate loss and texture loss were evaluated at different test temperatures on specimens with various binder appli cation rates. Both single layer and double layer applications were evaluated. 14 2.4.1 Effect of Binder Thickness on Aggregate Loss The study by FDOT and FHWA [ 1 4 ] showed the evaluation of the performance of single layer application compared to double layer a pplication with epoxy resin and one kind of aggregate. Specimens were conditioned by 20,000 cycles of TWPD turning. It was concluded in the study that the thickness of binder is also related to aggregate loss . For epoxy resin, the aggregate loss significantly increased for binder thickness less than 50 mils. However , the aggregate loss could be significantly reduced by implementing double layer applications because there was no mass loss recorded when the HFST application is a double layer system . 2.4.2 Effect of Binder Type and Temperature on Aggregate Loss The FDOT study has shown that at an intermediate testing temperature (75 ° F ) the mass loss was negligible for all binder types except for acrylic polyester resin after 50,000 cycles of TWPD. When the testing temperature is high (140 ° F ), the sample mass loss is negligible for acrylic polyester resin, and the loss increases for all other types of binders. The mass loss is the most severe for cresol modified e poxy resin [1 4 ]. It should be noted that the test was conducted on one sample of each type because of time limitation and material shortage. Since the results were based on a limited amount of materials, the results may not be reliable; therefore, addition al testing is needed to validate the findings. 2.4.3 Effect of Binder Type and Temperature on Texture Loss In the FDOT study [1 4 ], texture loss was also measured when the aggregate loss was measured for each specimen with four different types of binders. I t is shown in the study that all four binder types have a lower texture loss (mean profile depth loss) at an intermediate temperature (75 ° F), and the loss in the texture of HFST increased for all binders at a high temperature (140 ° F) [1 4 ]. 15 2.4.4 Thermal Compatibility of Binders The scope of the FDOT study also included two sets of tests to determine the linear coefficient of thermal expansion for four types of binders according to American Association of State Highway and Transportation Officials (AASHTO) test standards (AASHTO T336) and American Society for Testing and Materials (ASTM) test standards (ASTM C531). The results showed different Coefficients of Thermal Expansion (CTE) ranking for four binders with different test sets, and the reason was not s tated in the report [1 4 ]. However, it was clear that the HFST had significantly higher CTE than the regular asphalt or concrete surface. In addition, the values of CTE for epoxy binders were found to be higher than the acrylic polyester. This is part of th e reason why the epoxy resin is the most common binder used in different states for HFST. 2.4.5 Other Properties of Binders Commonly used binders for HFST are epoxy resin and polymer resin in the United States [ 1 7 ] , [ 1 8 ]. The differences between the physi cal and chemical properties of epoxy and polyester resins can be summarized as follows [1 9 ]: - Cure Time : Polyester resin cures faster than epoxy resin in general. This is because the components of epoxy are less compliant with cross - linkage effect chemicall y [ 20 ]. Cure time should not be too long or too short for the fact that it may affect the spraying of aggregates. - Odor : The polyester resin has an unpleasant odor and it is sometimes toxic to work with [ 20], and epoxy resin typically has low odor. - Bonding Strength : Epoxy resin is able to obtain higher bonding capabilities than polyester resin both initially and over time [ 1 9 ]. 16 - Wearing Resistance : Epoxy resin is more resistant to wear because its bonding strength is higher. Epoxy resin is also less susceptib le to moisture than polyester resin because the cross - linkage of epoxy is stable to hydrolysis, while polyester resin loses its hydrolytic products when it is continuously in contact with water [ 1 9 ]. The physical and chemical properties of epoxy resin lead to a higher cost than polymer resin, but epoxy resin is still the most commonly used binder in practice. 2.5 Limitations of High Friction Surface Treatments 2.5.1 Effect on Crack Relief Even though HFST is well known for increasing the surface frictio n fo r pavements and reducing the cra sh rate, it is suggested in a n FHWA study [2 9 ] that it is typically applied for spot safety purposes. Unlike other surface treatments (i.e., chip seal) that are used for pavement preservation and/or rehabilitation, a sou nd pavement under HFST is more desirable because the HFST does not provide any mitigation to the evolution of the existing distress [ 30 ]. 2.5.2 Cost Since high friction surface treatments have been installed across 34 states in the United States for safety purposes in 2017, the benefit - cost analysis of HFST has been conducted in many Federal Highway Administration projects [1], [ 14 ]. High friction surface treatment is able to reduce the total crash numbers from 57 percent to 100 percent as reported from dif ferent states. The analysis looks into the cost of installation and maintenance of HFST and the benefit of the reduction in vehicles incidents and accidents to determine the benefit - cost ratio. Overall the benefit - cost ratios range between 2 and 8, which m eans that the benefit HFST brings in dollars is twice to eight times higher than the cost of implementing HFST. In other words, the HFST is 17 economically justified due to the fact that the impact to a safer driving surface and reduced fatalities is certainly worth the cost . Besides the benefit that HFSTs bring to the society, HFSTs have a cost associated with material s , labor, and equipment mobilization. Based on the construction area and the variety of materials, the unit cost of conventional HFSTs in the United States ranges from $25 to $50 per square yard including labor, material, and equipment mobilization [22], [23], [24], [25], [26], [ 27] , [28] . Some example s of unit prices for high friction surface treatments in different states across the United States are shown in Table 2 . Table 2 Exampl es of High Friction Surface Treatment Unit Cost State Unit Cost $/ sq. yd Year Iowa 39 2012 Washington 36.5 2012 Kentucky 20 2014 Florida 26 2014 Pennsylvania 35 2016 Tennessee 21.3 2017 Michigan 22 2017 Texas 18.7 2018 Even though the unit cost of HFSTs decreased over the p ast few years, the current cost, compared to other surface treatments [ 31 ] , is still relatively high, which prevents its widespread use. Therefore, there is a need for lower cost alternatives, with similar or better performance . The cost of new HFST techniques can be reduced in many ways. For instance, changing the types of aggregates and/or binders as well as improving the efficiency of the installation process may be two of the possible ways to reduce the cost. The unit cost o f new HFST developed in China range from $8 to $16 per square yard which is approximately a third of the cost of conventional HFST in the United States ; h owever, the performance of new HFST techniques must be thoroughly investigated. 18 2.6 Synthesis of the P revious Work Motivation of the Current Study It is clear from the literature that the HFSTs are very useful in improving the tire/pavement friction and reduce traffic accidents and resulting fatalities. However, their relatively high cost limits their appl ications in many roadway segments that need improvement. There is a dire need for lower - cost HFST alternatives. The scope of this study included a thorough laboratory evaluation of one of the potential low - cost HFSTs developed by a company in China. This n ew technology includes a waterborne epoxy emulsified asphalt and corundum sand. 19 3. OBJECTIVE AND RESEARCH PLAN In order to thoroughly investigate the performance of the newly - developed HFST technique, extensive laboratory experiments and associated data analys e s were carried out in th is study. The objective of this study was to eval uate the performance of the low - cost HFST technique through a laboratory program, where the evaluation of the new HFST and the comparison with common HSFTs in the United States were included. The laboratory program was designed to evaluate the performance of HFSTs in three (3) aspects: (i) skid resistance, (ii) durability of HFSTs subjected to Freeze - Thaw (FT) Cycles, and (iii) effect of HFST material on crack retardation. 3.1 S kid Resistance The initial skid resistance of the wet surfaces of HFSTs w as measured by a British Pendulum Tester. Skid resistance is a good indicator of surface friction; therefore, it is a required parameter for evaluating the performance of a n HFST , as friction and improve the safety . The reason for measurements of a wet surface was due to the fact that the friction on wet surfaces was usually lower than that on a dry surface. It was also due to the fact that procedures for using the BPT required wetting the surface. 3.2 Durability under Freeze - Thaw Cycles It is reasonable to state that the condition of HFSTs deteriorates over time and traffic due to environmental effects and traffic loading. The skid resistanc e of HFSTs after FT cycles and mechanistic loadings were measured to test the durability of HFST to moisture and temperature induced damage and observe the deterioration of skid resistance of treatments. Understanding the durability of HFST materials is no t only crucial for the new design of HFSTs but also helpful for maintaining the existing design. 20 3.3 Crack Retardation Even though there have been studies showing that current HFST materials do not contribute to preventing crack formation and propagation . The effect of HFSTs (new and conventional) on crack retardation was still evaluated in the study by testing Semi - Circular Bending (SCB) specimens with HFSTs. The philosophy behind SCB tests is that HFST materials flow into the notch when there is existin g top - down cracking , as shown in Figure 2 ; then how new cracking initiates and propagates in the pavement is tested . The significance of conducting SCB tests included two perspectives. On one hand, the low - cost HFST used different materials from conventional HFST in the United States, so its effect on crack retardation was not revealed. On the other hand, whether or not the new HFST had an effect on crack retardation ad ded to its additional benefits or limitations. Figure 2 Philosophy behind SCB Test s 21 4. MATERIALS F our (4) types of aggregates were tested in th is study: (i) Chinese corundum sand, (ii) copper slag, (iii) Alpha Star bauxite (Bauxite 1), and (iv) Round Kiln bauxite (Bauxite 2). Additionally, there were two types of binder used: (i) waterborne epoxy developed by a Chinese company, and (ii) oil - based e poxy used in USA practices. Details about some properties of materials used in HFSTs are presented in th is section . 4.1 Aggregates The aggregates used in this study had different gradations since they were made of different components as shown in Figure 3 . Sieve analysis was performed on each type of aggregate to determine the gradation, and the components of each type were provided by the corresponding source. (a) ( b ) ( c ) ( d ) Figure 3 Aggregates Tested in the study (a) Corundum (b) Copper Slag (c) Bauxite 1 and (d) Bauxite 2 22 4.1.1 Gradation Aggregate gradation is shown in Table 3 and Figure 4 , which both show that the gradation of the corundum sand is much finer than any other aggregates used for HFSTs in the United States. Bauxite 1 and Bauxite 2 are the coarsest ones among all aggregates. Bauxites have similar gradation to each other even if Bauxite 1 is slightly coarser than Bauxite 2. Table 3 Gradation of the Aggregates Used in the Study Sieve Size [mm] Corundum [%] Copper Slag [%] Bauxite 1 AS [%] Bauxite 2 RK [%] 4.76 100 100 100 100 3.36 100 94.8 99.7 99.8 2.36 100 78.0 70.5 77.9 2.00 100 65.6 51.1 63.1 1.70 100 55.6 36.6 44.9 1.40 100 48.0 19.3 23.1 1.18 100 41.5 4.1 3.9 0.85 99.8 30.8 2.3 2.2 0.60 81.4 23.7 0.3 0.3 0.425 49.3 17.8 - - 0.30 16.9 13.4 - - 0.25 12.6 11.4 - - 0.212 8.7 9.8 - - 0.15 0.5 - - - 0.075 0.1 - - - Figure 4 Aggregate Gradation 0 20 40 60 80 100 0.01 0.1 1 10 % Passing Sieve Size (mm) Corundum Sand Copper Slag Bauxite 1 Bauxite 2 23 4.1.2 Origins of sources and Components of Aggregates - Corundum Sand : This type of aggregate is provided from China. The chemical characterization of corundum sand provided by the company shows that it is primarily composed of corundum, magnetite, and hematite. - Copper Slag : Copper Slag is supplied by sources in the United States. Copper slag typically consists of iron (III) oxide and silicate. - Calcined Bauxite s : Calcined bauxite s are currently the most popular aggregate used in the United States. The compositions of calcined bauxite are aluminum oxide, iron ( III) oxide, and silicate. 4.2 Binders Both waterborne epoxy and oil - based epoxy resins were investigated in the study. Two kinds of epoxy binders are significantly different from each other in terms of their components, mixing method, physical and chemical properties. 4.2.1 Components and Mixing The waterborne epoxy is a combination of waterborne epoxy A, waterborne epoxy B, emulsified asphalt and salt. The proportion of each component cannot be made public since the formulation is proprietary. After the waterborne epox y is mixed and spread, the strength is gained through the chemical reaction among epoxy, asphalt and salt while water in the waterborne epoxy evaporates at room temperature . The oil - based epoxy binder is composed of a base resin and a hardener compound. Two components of oil - based epoxy were mixed at a ratio of 1:1 by volume. The bonding strength is obtained through the process of an exothermic chemical reaction between epoxy resin and hardener where heat or energy is released . 24 4.2.2 Physical and Chemical Properties The waterborne epoxy and oil - based epoxy are different in many aspects of physical and chemical properties. - Odor : In general, both types of epoxy binders have an o dor; however, the waterborne epoxy has the smell of asphalt since the proportion of asphalt is small, the odor is faint whereas oil - based - Color : The waterborne epoxy contains milky white epoxy, gray and black emulsified asphalt and brown salt. The color of the mixture is brown and black. The colors of components of oil - based epoxy are clear orange for the base epoxy resin and clear black for the hardener. When tw o parts are mixed uniformly, the resultant color is clear dark brown for the mixture. - Viscosity : The viscosity of waterborne epoxy is less than oil - based epoxy by visual inspection. The waterborne epoxy flows more easily than the oil - based epoxy. While thi s is advantageous for filling crack, this can create problems if this application is used in heavily rutted pavements or pavements on a slope. In such cases, the low viscosity may lead to flowing down, creating non - uniform application. 4.3 Application Com bination It should be mentioned that the new HFST is a two - layer system, while the conventional HFST is typically a one - layer system. A two - layer system means that the binder is spread twice and the aggregates are distributed twice as well. Figure 5 illustrates conceptual drawings of the difference between a two - layer system and a one - layer system. In the bottom layer of new HFST, the binder is composed by Epoxy A, emul sified asphalt and ammonium salt, while in the top 25 layer only Epoxy B and emulsified asphalt are used (exact weigh ratios are proprietary). In the study, the waterborne epoxy binder was combined with corundum sand since both of them came from China, and th e oil - base epoxy resin was matched with all three other aggregates. In summary, the following four (4) combinations of aggregates and binders were tested in this study (shown in Figure 6 ) : - HFST 1: Two layers of corundum sand and waterborne epoxy +emulsified asphalt - HFST 2: Copper slag and oil - based epoxy resin - HFST 3: Bauxite 1 and oil - based epoxy resin - HFST 4: Bauxite 2 and oil - based epoxy resin (a) (b) Figure 5 Diagram of Two HFST Systems Studied (a) Two - Layer System (b) One - Layer System (NOT TO SCALE) 26 Figure 6 Samples with Application (Unconditioned) (a) HFST 1, (b) HFST 2, (c) HFST 3 (d) HFST 4 (left to right) 4.4 Sample Preparation and Application Rates Superpave Gyratory Compactor (SGC) was used to compact cylindrical Hot Mix Asphalt (HMA) specimens with 15 0 mm diameter and 50 mm in height. These HMA specimens were used as substrates for all the HFST applications. In order to determine the amount of materials needed for each combination. The masses of binder and aggregates were computed based on the areas an d the application rates. The application rates for four (4) combinations are summarized in Table 4 . Then the binder with designed weight was spr ead on the surface of the substrate uniformly using a metal spreader, which was followed by the even distribution of aggregate on top of the binder. Since the new HFST was designed to be a double layer application, this step was repeated once more. Since t here was usually no compaction effort required for HFST construction, the specimens were left to be dry without any compaction. In total, forty (40) substrates were prepared ( 10 substrates for each HFST combination ) . Table 4 Binder and Aggregate Application Rates (g/m 2 ) HFST system Binders Aggregates HFST 1 bottom layer 400 300 HFST 1 top layer 400 350 HFST 2 750 2900 HFST 3 750 3350 HFST 4 750 2950 27 Semi - Circular Bending t est ( SCB ) specimens were sliced from Superpave Gyratory Compactor ( SGC ) samples that were compacted to a height of 180 mm and a diameter of 150 mm. From the center of the 180 - mm tall SGC samples, two cylindrical slices with a thickness of 50 mm were obtained. The two slices were later cut in the middle into four semi - circular specimens. A notch was saw cut along the symmetric axis for each semi - circular specimen. The dimensions of the notch were 15 ± 1 mm in height and 1.5 ±0.1 mm in width. The target air void content for SCB specimens was 7%±0.5 %. The procedures for applying HFST on SCB specimens were exactly the same as it was described above. Due to the fact that HFST 1 used a binder with relatively low viscosity and aggregates with fine gradation, the materials flowed into the notch as shown i n Figure 7 (a), and Figure 7 (b). For the other HFSTs, the epoxy binder covered (i.e., bridged over) the notch because of relatively high viscosity, and the size of aggregates did not allow it to fall into the gap as shown in Figure 7 (c) and Figure 7 (d). A total of sixteen (16) spe cimens (four (4) specimens for each HFS T) were prepared in this study. (a) ( c ) ( b ) ( d ) Figure 7 SCB Specimens with HFSTs: (a) HFST 1 Top View, ( b) HFST 1 Side View, (c) HFST 4 Top View, (d) HFST 4 Side View 28 5. EXPERIMENTAL PROGRAM As introduced in Chapter 3, the scope of this study includes laboratory tests to investigate three aspects related to the performance of HFSTs: initial skid resistance, durability after freeze - thaw cycles, and effect on crack retardat ion. The skid resistance was measured by British Pendulum Tester (BPT); the tests about durability incorporated freeze - thaw cycles and mechanistic loadings to maximize the surface damage; and the fracture - related properties (i.e., crack initiation) of HFST s were simulated and tested by Semi - Circular Bending Tests. 5.1 Skid Resistance In the study, BPT was used to determine the surface frictional properties of the samples treated with four HFSTs; however, slight modifications were made to th e BPT for labora tory testing. As shown in Figure 8 , a system of screw - suction cups replaces the three leveling screws of the original BPT to make the device stable on the testing bench and to adjust leveling. ASTM E303 specifies the procedures to use BPT. Measurements, called British Pendulum Numbers (BPN) were record ed at 20 ° C to avoid the temperature correction factor. Ten samples for each HFST were tested before and after the treatment. General steps of BPT testing are summarized as follows : Step - 1: Level the equipment by adjusting the screw - suction cup until the bubble is centered. Step - 2: Check and adjust the zero position of the device by releasing the pendulum wh ile no specimen is placed to measure. Step - 3: Choose the appropriate rubber slide for the test according to the contact area between pendulum and specimen surface. 29 Step - 4: Cover the test area with sufficient water thoroughly, then release the pendulum for one swing without recording the reading. Step - 5: Execute four more swing s and record the measurements be sure to rewet the surface after each release. Figure 8 Laboratory Setup of BPT 5.2 Durability Subjected to Moisture Induced Damage The durability of HFSTs to moisture induced damage was evaluated by exposing the samples to Freeze - Thaw ( FT ) conditioning cycles and simulation of surface damage caused by traffic . The mechanistic loading was initial ly applied by using Hamburg Wheel Device using a rubber wheel. H owever, it was later observed that the rubber tire was not stiff enough to create surface damage to the specimens. Therefore, modified sweep tester was finally used to apply mechanistic loading to the surface to damage the HF ST. The amount of deterioration after FT cycles and loading was measured in terms of skid resistance loss by BPT. 30 5.2.1 Freeze - Thaw Cycles The details on moisture induced damage by FT cycles are described in AASHTO T283 standard test method . The following steps of conditioning (one FT cycle) the specimen were followed in the study (pictures shown in Figure 9 ) : Step - 1: Saturate the specimens using a vacuum pump t o a degree of saturation between 50% and 80%. Step - 2: Vacuum - pack the specimens into leak - proof plastic bags. Step - 3: Condition the specimens in an environmental chamber at - 12°C for 40 hours. Step - 4: Remove samples from plastic bags, and then place them in to a water bath at 60 °C for 24 hours. Step - 5: Place the specimens into another water bath at 25 °C for 7 hours. (a) Before Frezzing ( b ) Freezing: - 12 ° C for 40 hours (c) Thawing: 60 ° C for 24 hours, 25 ° C for 7 hours Figure 9 Processes of a Freeze Thaw Cycle Even though the AASHTO T283 stand ard indicates that one FT cycle is enough to characterize the materials, the process of freezing and thawing and damaging the surface was repeated three (3) times to investigate the deterio ration rate after multiple cycle s . A total of ten (10) specimens were subjected to FT cycles (2 for each HFST as well as untreated samples). 31 5.2.2 Hamburg Wheel Tracking Device Hamburg Wheel Tracking (HWT) device is used to determine the failure susceptibi lity of HMA to permanent deformation due to inadequate binder stiffness, week aggregate structure or insufficient resistance to moisture damage. The device tests the specimen submerged under water using a reciprocating steel wheel as shown in Figure 10 (a). In this study, the steel wheel was substituted with a rubber wheel to better simulate the field condition as shown in Figure 10 (b). The rubber tire was locked in place to mimic the stopping and braking actions, so the wheel was sliding over the HFST in the test. The tire pressure of the attached r ubber wheel was 34 psi, and the wh eel rotated/slid at a rate of 52 pass es per minute. The load borne by the tire reduced to 125 lbs. from 158 lbs. with the modification. Procedures of using HWT device is detailed in AASHTO T324. The skid resistance was mea sured after every 25 passes, and the temperature of the water bath is 25 °C. (a) ( b ) Figure 10 Hamburg Wheel Tracking Device (a) Original, (b) Modified 32 Summary of steps in HWT test is listed as following: Step - 1: Cut the cylindrical specimens into the shape that fits the mold as shown in Figure 11 . Step - 2: Fasten the mounting trays into the empty water bath. Ste p - 3: Fill the device with water to a desir ed depth and monitor the temperature. Step - 4: Saturate the test specimens in the water for 30 minutes once the temperature is reached. Step - 5: Start the test for 25 passes then take specimens out and measure the sk id resistance. Step - 6: Repeat the previous step until the failure of the sample or constant skid resistance after deterioration. Figure 11 Specimen in HWT Molds 5.2.3 Sweep Test The purpose of sweep test after each FT cycle was t o maximize the surface damage on HFSTs. The traditional test device, described in the AASHTO D7000, is supplied with a regular plastic 33 brush as shown in Figure 12 . The plastic brush was found to be too soft to create any damage. Thus, the device was modified with a V - shaped knife instead of brush first as shown in Figure 13 (a). However, there were visible dents along the edge of the knife after a few loading cycles by the sweep tester. Finally, the device was modified with the wire brush shown in Figure 13 (b). The HFST surface was swept by the wire brush for 10 minutes after each FT cycle followed by the measurement of skid resistance. (a) ( b ) Figure 12 Original Sweep Tester (a) Device (b) Regular Plastic Brush (a) ( b ) Figure 13 Substituting Tools for Plastic Brush (a) V - Shaped Knife, (b) Wire Brush 34 5.3 Effect on Crack Retardation Since crack initiation and propagation belong to the area of fracture mechanics, SCB test was included in the experimental matrix to evaluate the effect of different HFSTs on the top - down cracking formation and propagation in the study. The existing standard for SCB test is AASHTO TP124, which describes how to determine the fracture energy of HMA at an intermediate test temperature based on the captured load and displacement curve. In addition, the Flexibility Index (FI) is calculated from the fracture energy and slope of the curve. Work of frac ture (W f ), Fracture Energy (G f ), and FI were calculated according to Equations ( 1 ) , ( 2 ) , and ( 3 ) . The procedures of the SCB test are summarized in the following: Step - 1: Condition the specimens in an environmental chamber at 25 ° C for 2 hours. Step - 2: Place the specimen into an axial loading device that measures loads and displacement (i.e., Material Test System (MT S )), and adjust the position of the specimen to the center of loading cell. Step - 3: Adjust the loading cell to a position where it barely touches the specimen. Step - 4: Start the test by releasing the load cell at a rate of 50 mm/min, the device records the displacement and the amount of loading at a frequency arou nd 200 Hz. Step - 5: The test will stop by itself once the duration is reach ed or the sample is damaged. ( 1 ) ( 2 ) ( 3 ) 35 where : W f1 is the work of fracture up to the load peak, W f 2 is the work of fracture from the load peak to the end of the test , u peak load is the displacement at the peak of the load (mm), u final is the cut - off displacement (mm), P 1 is the load recorded during the test up to the peak (kN), P 2 is the load from the peak to the end of the test (kN), A lig is the ligament area (mm 2 ), and | m | is the slope of the post - peak load - displacement curve ( kN/mm). Since the materials of HFST 1 allowed the notch to be partially filled with binder and aggregate during the sample preparation, it was expected that this phenomenon would lead to the propagation. Figure 14 is a zoomed - in image of the HFST 1 around the notch. It is clear ly shown in Figure 14 that waterborne epoxy and fine corundum sand flowed into the notch. Figure 14 Zoom - in on the Notch for HFST 1 36 6. RESULT AND DISCUSSION 6.1 Initial Skid Resistance The initial skid resistances (i.e., before FT cycles and abrasion testing) of both untreated and treated surfaces are shown in Figure 15 . All untrea ted specimens had a similar BPN of 60 ± 2, which was a good indication that the base skid resistance was almost the same for all surface treatments. As for treated specimens, samples with HFST 2 which was composed of the copper slag aggregate and oil - based e poxy resin, resulted in providing the least skid resistance to the untreated surface. The bauxites (HFST 3 and HFST 4) were able to provide the most significant increase in skid resistance, which is consistent with the findings from the literature. The BPN of the treated specimen with HFST 1 (the new technology) was comparable to the BPNs of HFST 3 and HFST 4 , which were made with bauxites . Figure 15 BPN on Treated and Untreated Specimens without Conditioning The t - Test is one of the inferential statistics test s used to determine the significance of the difference between the means of two groups of data. Based on the relationship between the two 37 groups considered in the test, the t - Test can be used to evaluate: i) independent group s with unequal variances, ii) indepen dent groups with equal variance and iii) paired groups that two groups are correlated to each other . Prior to that, the type of t - Tests ( t wo samples with equal variance or unequal variance) was determined by f - Test. In this investigation, all measurements within a cycle are independent with respec t to any other sample sets due to the fact that the measurement of one set does not impact the measurement on the other sets. In the case of paired data sets for a particular HF ST, f - Test was not needed because the most appropriate type t - Test was certain to be Type iii : t - Test with two paired groups . For a given HFST, the amount of ski d resistance at the end of a la ter cycle would be dependent on the skid resistance measured aft er the previous cycle(s). In order to identify the most appropriate type of t - Test (i or ii) to be considered for each pair of data within each cycle in this study, the f - Test has been used to verify whether the variance of two samples are equal or not. T he result of f - Test is the probability for two sets of data to have the same variance, while the result of t - Test is the probability for two sets of data to have the same mean. Therefore, the p - value of t - Test or f - Test is desired to be less than 0.05 to s how that the chance of two sample with the same mean or variance is less than or equal to 5%. For instance, p of T - Test for HFST 1 before and after the treatment is zero; it means that the probability for untreated HFST 1 samples to have the same average s kid resistance as the treated HFST 1 sample s is 0.0 %. Since the probability is less than 0.05, it is reasonable to reject the hypothesis that there is no statistical difference between the measurements of skid resistance before and after the HFST 1 applica tions. In other words, the means of skid resistances before and after HFST 1 applications have proved to be statistically different . Statistical significance of comparisons before and after the treatment for each combination is summarized in Table 5 where 38 t - Test results are listed. It has been shown in Table 5 that the treated and untreated BPN values (of each sample) are statistically significantly different. Table 5 T - Test Results for before and after the Treatment Untrea ted Treated (10 replicates each) Mean St. Dev. Mean St. Dev p ( T - test : Treated versus Untreated for each sample ) HFST 1 59.6 2.13 5 75.5 2.32 7 0 HFST 2 61. 2 2.62 8 64.1 3.37 8 0.01 HFST 3 58.4 3.55 9 78. 1 4.94 7 0.02 HFST 4 61. 3 2.74 2 76. 1 5.58 3 0 * P - 6.2 Skid Resistance after Moisture Induced Damage In this study, there were two approaches to mechanical loading after FT cycles. As described earlier, t he Hamburg Wheel Tracking device was tri ed with no success. Then the sweep tester with wire brush was deemed appropriate for testing HFSTs . The sections below include descriptions of the results. 6.2.1 Skid Resistance after Hamburg Wheel Tracking Device Figure 16 shows the change in BPN of HFST 1 and HFST 2 after testing using the HWT device. It is noted that the rubber wheel was locked in, i.e., it slid o ver the specimens, not rolled over them. The deterioration was captured after every 25 cycles of rubber wheel passing. It is clearly shown in Figure 16 that the BPN of HFST 1(the new material) continuously decreased after up to 200 cycles. However, for the HFST 1, the BPN only reduced for the first 25 cycles, and then the reading remained constant up to 125 cycles. The test in HFST 2 was terminated after 325 cycles because of the failure of the rubber wheel as shown in Figure 17 . The tire was flat while there was n o visible damage on the HFST 2 surfac e . It appeared that the HFST 2 was abrading the tire, 39 rather than tire abrading the HFST 2 . There fore, this test (in its current form) is not deemed appropriate for testing HFSTs. A stiffer tire in HWT or another material is needed to simulate abrasion caused traffic in the field. Figure 16 Result of HWT for HFST 1 and HFST 2 Figure 17 Flat Tire of HWT after 325 C ycles on HFST 0 10 20 30 40 50 60 70 80 90 0 50 100 150 200 250 BPN Number of HWT cycles HFST 1 HFST 2 40 6.2.2 Skid Resistance after Freeze - Thaw Cycles and Sweep Test Figure 18 illustrates the skid resistance test results of treated samples after each freeze - thaw cycle and the subsequent sweep test. The deterioration rat es of sp ecimens with waterborne epoxy and corundum sand (HFST 1) and the samples with copper slags (HFST 2) were similar. The specimens with oil - based epoxy and bauxites (HFST 3 and HFST 4) exhibited a lower deterioration rate , even though their initial unconditio ned BPNs were similar to the HFST 1 specimens. However, the concept of the minimum required skid resistance should be considered to identify whether or not the treatment can provide the minimum performance for safety . The scientific literature on this topi c is limited [3 1 ], [3 2 ]. N evertheless , the most restrict ive value for BPN for exterior pavement with a slope steeper than 1/14 was found to be 55 [3 2 ]. Based on this requirement and the results shown in Figure 18 , the HSFT 1 provides the minimum required skid resistance after two freeze - thaw cycles and is borderline after the third one. S amples treated with c opper s lag failed sooner than all the other combinations. It is worth noting that that the AASHTO T283 specification suggest conditioning the samples to only one freeze - thaw cycle. In that case, all the HFSTs were well above the threshold of 55. The result of statistical s ignificance test s , p valu es of 2 paired - sample T - Tests, are summarized in Table 6 and Table 7 . F - Tests were analyzed before T - Tests to determine either T - Test with equal variance or T - Test with unequal variances between two samples should be used. All the combinations in Table 7 showed their statistical significance except the combination of HFST 1 and HFST 4 in the unconditioned case. 41 Figure 18 BPN Measurements on Treated Samples after Each FT Cycle Table 6 Statistical Significance of Measurements (within cycles) after Conditioning Cycles HFST 1 HFST 2 HFST 3 HFST 4 HFST 1 Uncon ditioned - 0 0 0.36 1st Cycle - 0 0 0 2nd Cycle - 0 0 0 3rd Cycle - 0 0 0 HFST 2 Unconditioned - - 0 0 1st Cycle - - 0 0 2nd Cycle - - 0 0 3rd Cycle - - 0 0 HFST 3 Unco nditioned - - - 0 1st Cycle - - - 0 2nd Cycle - - - 0 3rd Cycle - - - 0 * P - between the measurements in a combination . 42 Table 7 Comments on T - Test (within Cycles): H 0 = Unequal Mean HFST 1 HFST 2 HFST 3 HFST 4 HFST 1 Uncon ditioned N/A Pass Pass Fail 1st Cycle N/A Pass Pass Pass 2nd Cycle N/A Pass Pass Pass 3rd Cycle N/A Pass Pass Pass HFST 2 Uncon ditioned N/A N/A Pass Pass 1st Cycle N/A N/A Pass Pass 2nd Cycle N/A N/A Pass Pass 3rd Cycle N/A N/A Pass Pass HFST 3 Uncon ditioned N/A N/A N/A Pass 1st Cycle N/A N/A N/A Pass 2nd Cycle N/A N/A N/A Pass 3rd Cycle N/A N/A N/A Pass The statistical differences were also evaluated for a particular HFST by the comparison of the combination of historical BPN measurements pairs on a single HFST. For instance, the uncondi tioned measurements on HFST 1 were compared with the measurements on HFST after one FT cycle . The probability for these two sets to have same mean was 2.81E - 3% which is less than 5%, so these two measurement s did not have the same average. In other words, there is a statistically significant deterioration in skid resistance for HFST 1 from unconditioned to the end measurem ents is shown in Table 8 and Table 9 . It is shown in Table 9 that for HFST 1 (with corundum sand), the deterioration was still significant after three cycles while HFST 3 and HFST 4 (with bauxites) did not have statistical deteriorations just after the first FT cycle. This is an indication that corundum sand was more susceptible to FT cycles and moisture induced damage because it continued d eteriorating after three FT cycles while the skid resistance of bauxites reduced after the first cycle but stayed the same for the following two cycles. HFST 2 (with Copper Slag) lay somewhere between 43 corundum sand and Bauxite since it stopped deterioratin g after two FT cycles; However, the initial skid resistance provided by HFST 2 was the lowest among four HFSTs (see Figure 15 ). Table 8 Result of T - Test on the BPN Measurements (for Certain HFST) Unconditioned . 1st Cycle 2nd Cycle HFST 1 1st Cycle 0 - - 2nd Cycle 0 0 - 3rd Cycle 0 0 0 HFST 2 1st Cycle 0 - - 2nd Cycle 0 0 - 3rd Cycle 0 0 0.07 HFST 3 1st Cycle 0 - - 2nd Cycle 0 0.45 - 3rd Cycle 0 0 0.86 HFST4 1st Cycle 0 - - 2nd Cycle 0 0.07 - 3rd Cycle 0 0 0.08 *P - different between the measurements in a combination. Table 9 Comments on T - Test on the BPN Measurements (for Certain HFST) , H 0 = Unequal Means Uncon. 1st Cycle 2nd Cycle HFST 1 1st Cycle Pass N/A N/A 2nd Cycle Pass Pass N/A 3rd Cycle Pass Pass Pass HFST 2 1st Cycle Pass N/A N/A 2nd Cycle Pass Pass N/A 3rd Cycle Pass Pass Fail HFST 3 1st Cycle Pass N/A N/A 2nd Cycle Pass Fail N/A 3rd Cycle Pass Pass Fail HFST4 1st Cycle Pass N/A N/A 2nd Cycle Pass Fail N/A 3rd Cycle Pass Pass Fail 44 It should be mentioned that in all cases, the reduction of skid resistance was due to a visible loss of aggregates . None of the samples failed at the interface between the base sample and the epoxy. In other words, there was no obvious separation of HFSTs from the base HMA afte r all three freeze - thaw cycles and sweep tests. Even though this was expected for the oil - based epoxy, which is commonly used in the United States without any issue, the bonding performance of waterborne epoxy was interestingly and surprisingly durable , wh ich turned out to perform as well as the oil - based epoxy. This shows that the HFST 1 is a viable alternative to traditional systems used in the U.S. However, this conclusion needs to be verified with field studies. 6.3 Crack Initiation and Propagation The fractural behavior of the specimens with or without HFST was tested using the Semi - Circular Beam ( SCB ) tests. 6.3.1 Semi - Circular Bending Test Table 10 and Figure 19 s ummarize the results of the SCB tests performed on untreated and treated samples and illustrate the effect of the HFST s on the crack formation and propagation during the SCB test . The behavi or of the samples before the loading peak is commonly associated with the initiation of the crack and it is the most a ffected by the HFSTs. A ll the applications were able to delay the crack initiation (in different ways) as compared to the untreated sample s. In fact, for the traditional HFST applications, the delay is mainly due to the action needed to crack /break the bridge between the two sides of the samples formed during the application of the treatment (see Figure 7 ). The initial irregularities or non - uniformity of the load - displacement curves, as well as the increase of the load peak values of the HFST 2 and HFST 3, clearly describe this phe nomenon ( Figure 19 (b) and Figure 19 (c)). The HFST 4 showed 45 a limited resistance to the crack formation wi th limited bumps noticed in the characteristic SCB curves. After the crack, as expected, the behavior of the material under load is similar to the one showed by the untreated samples ( Figure 19 (d)). The presence of the HFST 1 material in the notch delayed the formation of the crack as well. A visual inspection of the samples during and after the test highlighted that the crack start s and propagate at the interface between the filling material and the base sample. Thus, the characteristic load - displacement path shown in Figure 19 (a) seems to be represe ntative of the adhesive properties of the HFST 1 material and how it tends to redistribute the loading stress around the notch. Because of the delay in the crack initiation due to the HFSTs, the overall work of fracture as well as the fracture energy values are higher than those calculated for the untreated samples. Given Equation ( 3 ) , for the same slope of the post - peak load - displacement curve ( |m| ), the immediate consequence of the higher fracture energy is an increase in the FI values. However, this is not the only caus e of the increase in FI, which is also affected by the |m| . Table 10 shows that t he |m| values seem to be affected by the treatm ents as well, where th e lowest |m| was observed in the HFST 1 . Table 10 SCB Test Results Summary W f1 W f2 G f FI u peak - load |m| [J] [J] [J/m2] [ - ] [mm] [kN/mm] Untreated 3.3±0.2 3.2±1.8 1904.7±50.0 1.0 ±0.2 1. 1 ±0. 1 21. 7 ±5. 9 HFST 1 4.2±0.6 3. 4 ±0.5 2630. 1 ±233.7 4. 3 ± 1.0 1.8±0.2 6. 4 ±1. 2 HFST 2 7.0±1.5 3.2±0.3 3336. 5 ±60 1.0 4. 7 ±1.9 2. 6 ±0.2 13.9±5.0 HFST 3 4.9±0.7 3.0±0. 6 2717.6±326.8 3. 4 ±0.6 1.7±0.2 8.1±0.9 HFST 4 5.8±0.3 2.9±0. 2 3048. 8 ±118.9 2.5±0.9 1.9±0. 1 8.3±2. 9 46 Figure 19 SCB Test Result of HFSTs (a) HFST 1, (b) HFST 2, (c) HFST 3, (d) HFST 4 It should be recalled that the as |m| decreases, the flexibility index increases, which is an indication of slow crack propagation. However, the reasons behind this behavior are not entirely clear since |m| should be representative of the asphalt concrete in which the crack propagates. After the peak, it is expected that crack propagates within the asphalt sample, which is the same for all untreated and treated samples. 47 7. CONCL USION S High - friction surface treatments (HFSTs) are used by most state Departments of Transportation to restore skid resistance of existing pavements. An added benefit of the HFST includes limiting the moisture penetration into cracks. However, HFSTs are n ot typically expected to provide any mitigation to the evolution of the existing distresses and it is still an expensive surface treatment option. A low - cost HFST technique that uses waterborne epoxy and corundum sand was investigated in the study. The per formance of this low - cost HFST technique was evaluated and compared against materials commonly used in the United States. The evaluation focused on three performance aspects of the treatments: the initial skid resistance improvement , the deterioration of t he skid resistance after freeze - thaw cycles and water induced damage , and the effect on cracking formation and propagation. Laboratory test results demonstrated that the low - cost system (application with waterborne epoxy + emulsion + corundum sand) was abl e to provide an improvement of skid resistance to the original pavement surface comparable to the other traditional applications. However, the durability to moisture damage induced by freeze - thaw cycles of this technique is not as good as the current treat ments with bauxite aggregates, which remain by far the best performing material. Despite this, the new HFST still met the minimum required skid resistance of BPN 55. It should be noted that the reduction of skid resistance because of the sweep tests was du e to the loss of aggregates. In other words, the freeze - thaw conditioning cycles did not affect the interface between the base sample and the epoxy in any case. Results of the semi - circular bending (SCB) tests demonstrated that all HFST applications were a ble to delay the crack initiation as compared to untreated samples , but in different ways. 48 For the traditional HFST applications, the delay is mainly due to the action needed to crack /break the bridge between the two sides at the base of the sample while f or the HFST 1 seems to be due to its adhesive properties and the tendency to redistribute the loading stress around the notch with a width of 1.5±0.1 mm. Further evaluations is probably needed for better understanding the possible effect of this surface tr eatment on the crack propagation, especially on different crack sizes. Nonetheless, the viscosity of waterborne epoxy and emulsified asphalt also limits its application condition. The maximum degree of super - elevation of pavement needs to be determined in the future for the reason that the new HFST binder has the tendency to flow across the surface when the surface is not leveled. The relatively low viscosity of waterborne epoxy also limits its use in the present of rutting. When the pavement ruts, the bind er will flow toward the low - lying land due to the depression effect. In summary, it is presented in the thesis that an innovative and cost - effective high - friction surface treatment was evaluated and compared against common applications in the United States in the aspects of skid resistance, durability under freeze - thaw c ycles and its potential effect on crack retardation, and all testing was conducted in a laboratory setting. Therefore, the evaluation of other aspects of HFST and the correlation between lab oratory results to field performance require further investigation. 49 REFERENCES 50 REFERENCES [1] Izeppi, E. de L., G. W. Flintsch, and K. McGhee. Field Perfo r mance of High Friction Surfaces . Publication FHWA/VTRC 10 - CR6. FHWA, Washington D.C., CTRC, Virginia Transportation Research Council, Richmond, 2010. [2] Nicholls, J. C. Trials of High - Friction Surfaces for Highways . Publication TRL REPORT 125. TRL, Transporta t ion Research Laboratory, Cr owthorne, England, 1998. [3] Nicholls, J. C. Laboratory Tests on High - Fric t ion Surfaces for Highways TRL, Transporta t ion Research Laboratory , vol. 176, Crowthorne, England, 1997. [4] Henry, J. J. Evaluation of Pavement Friction Characteristics: A Synthesis of Highway Practice. NCHRP Synthesis 291, Washington D.C., 2000 . [5] Rajaei, S., Chatti, K. and Dargazany, R. A Re v i ew : P avement Surface Micro - texture and Its Contribution to Surface Friction. Transportation Research Board, Pu b lication 17 - 06773. Washington, D.C., 2017. [ 6 ] Lu, Q., and Steven, B. Friction Testing of Pavement Prese r vation Treatments: Literature Review. Technical Me m orandum UCPRC - TM - 2006 - 10. California Department of Transportation, Sacremento, California, 2006. [ 7 ] ASTM E274/E274M - 15, Standard Test Method for Skid Resistance of Paved Surfaces Using a Full - Scale Tire . ASTM International, West Conshohocken, PA, 2015, [ 8 ] Horne, W. B., NASA Diagonal - Braked Test Vehicle Evaluat ion of Traction Chara c teristics of Grooved and Ungrooved Runway Surfaces at Miami International Airport, Miami, Florida. Technical Me m orandum NASA TM X - 73912. National Aeronautics and Space Administration, Hampton, Virginia, 1973. [ 9 ] ASTM E1911 - 09a, Standard Test Method for Measuring Paved Surface Frictional Properties Using the Dynamic Friction Tester . ASTM International, West Conshohocken, PA, 2009. [ 10 ] Special Provision for High Friction Surface Treatment. Michigan Department of Transportation. 1 2SP - 800A - 03, 2016. [1 1 ] Qualified Product List of High Friction Surface Treatment. Illinois Department of Tran s portation, Springfield, IL, 2018. [1 2 ] Special Provision Regarding High Friction Surface Treatment . Tennessee Department of Transportation, 2017. 51 [1 3 ] Heitzman, M., T urner, P., and Greer, M. High Friction Sur fac e Treatment Alternative Aggregates Study . V ol. 15 0 4, National Center for Asphalt Technology at Auburn Univer si ty. Auburn, AL, 2015. [1 4 ] Wilson , B., and Mukhopadhyay, A., Alternati v e Aggregates and Materials for High Fric t ion Surface Treatments . Report BDR74 - 977 - 05. FHWA, Washington D.C., FDOT, 2016. [1 5 ] Guan, B., Wu, J., Xie, C., Fang, J., Zheng, H., and Chen, H., Influence of Macrotexture and Microtexture on the Skid Resistance of Aggregates . Advances in Materials Science and Engineering 1437069, 2018. https://doi.org/10.1155/2018/1437069 [1 6 ] Wilson , D., and Dunn, R., Polishing Aggregates to Equilibrium Skid Resistance. International Surface Frict i on Conference, Christchurch, New Zealand, 2005. [1 7 ] High Friction Surface Treatments . Transportation Rese a rch Synthesis TRS 1802. Min n esota Department of Transportation 2018. [1 8 ] Holzschuher, C., High Friction Surface Treatment Guidelines Project Selection, Materials, and Construction. Florida Department of Transportation, Gainesvill e , FL, 2018. [1 9 ] Plueddemann, E. P, Silane Coupling Agents . 2 nd ed. ISBN 978 - 1 - 4899 - 2072 - 0.Plenum Press, New York, 1991. [ 20 ] Luft, J. H., Improvements in Epoxy Resin Embedding Methods. The Journal of Biophysical and Biochemical Cytology, Vol. 9, 1961. http://doi.org/10.1083/jcb.9.2.409 [2 1 ] Bourne , L. B., and Milner, F. J. M., Polyester Resin Hazards. Brit. J. Ind ustr . Med , p. 100, 1963. [2 2 ] Cheung, J., and F. Julian. High Friction Surface Treatments: Not Just for Rural Curves . Publication FHWA - SA - 16 - 021. FHWA, U.S. Department of Transpor ta tion, 2016. [ 2 3 ] Anderson, K. W. , Russell, M. , Sheets, J., Burkey, D. , Uhlmeyer, J. S. , and Weston, J. \ Evaluation of Tyregrip ® High - Friction Surfacing. WA - RD 788.1, Washington State Department of Transportation. 2012. [ 24 ] Gray, G., High Friction Surface Treatment Benefit/Cost 2016. Pennsylvania D epartment of Transportation 2016. [ 2 5 ] Cheung, J., Julian, F., and Maravec, M., Frequently Asked Questions about High Friction Surface Treatments (HFST). Publication FHWA - SA - 1 4 - 019 . FHWA, U.S. Department of 52 Transpor ta tion, 2014. [ 2 6 ] Wilson, B. T., Brimley, B. K. , Mills, J., Zhang, J. , Mukhopadhyay, A. , and Holzschuher, C. Benefit Cost Analysis of Florida High - Friction Surface Treatments. Journal of the Transporta t ion Rese a rch Board , vol. 2550, no. 1, pp. 54 62, Jan. 2016. DOI: 10.3141/2550 - 08 [ 2 7 ] Tennessee Average Unit Price - 01Jan2017 thru 31Dec2017. Tennesse Department of Transportation, 2017. [ 2 8 ] Average Low BId Unit Price - Construction - Statewide. Texas Department of Transpotation [Online]. Available: https://www.dot.state.tx.us/insdtdot/orgchart/cmd/cserve/bidprice/s_3001.htm. [Accessed: 20 - Nov - 20 18]. [ 29 ] Merritt, D. K., C. A. Lyon, and B. N. Persaud. Evaluation of Pavement Safety Performance. Publication FHWA - HRT - 14 - 065. FHWA, U.S. Department of Transportion, 2015. [ 30 ] Cheung, J., and F. Julian. Every Day Counts: Frequently Asked Questions High Friction Surface Treatments (HFST) 2017 . Federal Highway Administration. https://www.fhwa.dot.gov/innovation/everydaycounts/edc - 2/pdfs/hfst_faqs.pdf. Accessed Jun. 26, 2018. [ 31 ] Fwa, T. F. Skid Resistance Determination for Pave ment Management a nd Wet - Weather Road Safety. International Journal of Transportation Science and Technology , Vol. 6, No. 3, 2017, pp. 217 227. https://doi.org/10.1016/j.ijtst.2017.08.001. [ 32 ] Harlow, A. Skid Resistance and Pavement Marking Materials. Surface Friction Conference, New Zealand, 2005.