ENVIRONMENTAL IMPACT OF CIVIL ENGINEERING INFRASTRUCTURES: FLEXIBLE PAVEMENTS USING END-OF-LIFE TIRES AND MATERIAL INTENSITY FOR WIND TURBINES By Angela Farina A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Civil Engineering—Doctor of Philosophy 2021 ABSTRACT ENVIRONMENTAL IMPACT OF CIVIL ENGINEERING INFRASTRUCTURES: FLEXIBLE PAVEMENTS USING END-OF-LIFE TIRES AND MATERIAL INTENSITY FOR WIND TURBINES By Angela Farina Construction projects that involve flexible pavements and wind turbines typically utilize and handle large volumes of materials. Polymer modifications can improve the durability of flexible pavements and reduce the amount of virgin materials used over the service life. Crumb rubber (CR) from end-of-life tires might substitute the synthetic polymer in the asphalt mixtures, which would reduce the use of virgin materials and the number of tires deposited in landfills. Wind power contribution to global electricity generation is expected to increase from 5 to 30% by 2050. This growing capacity will lead to increasing construction of wind farms globally. The scope of the study presented in this dissertation includes evaluating the mechanical, environmental, and toxic impact of different asphalt mixtures containing CR and the material demand for wind turbines in the USA and globally until 2050. Polymer coated rubber (PCR) and devulcanized rubber (DVR) are enhanced CR products used as recycled modifiers in asphalt mixtures. However, a comprehensive evaluation of the mechanical, environmental, and toxic impact of asphalt mixtures containing enhanced CR is still missing and necessary to evaluate whether the replacement of synthetic polymers with recycled modifiers is viable. In this study, life cycle assessment (LCA) and mechanistic-empirical pavement design (MEPDG) were used to compare the environmental impact of enhanced CR (PCR, dry and wet technology, and DVR) used in pavement materials with reference mixes (Control, unmodified, and SBS, modified with the synthetic polymer styrene-butadiene-styrene). Moreover, the potential leaching of metals from the same mixtures was assessed through the mass transfer rate of constituents, the 1315 EPA method. PCR dry mixture performed mechanically and environmentally as well as the SBS mix. Over the service life of road pavement, it was observed that the use of PCR dry and SBS mixtures leads to material savings up to 2.4 times compared to the control mixture. Leaching tests revealed that zinc was present in all asphalt mixtures, with a concentration greater than the drinking water standard limit. The number of wind turbines that are expected to be built by 2050 to meet the renewable energy targets will require a large amount of material mining and transformation. The material demand for wind turbines installed in the USA and the rest of the world were estimated and compared to the expected production until 2050. In addition, the carbon footprint and the cumulative energy demand associated with the material production were quantified based on the annual addition capacity using three different outlooks for the USA and one for the rest of the world. The material demand in the USA and globally was lower than the expected material production. The carbon footprint for material manufacturing for all wind turbines built in 2050 will be eight times lower than the CO2 equivalent emitted by coal power plants in the USA in that year. Lowering the environmental impact of wind turbine manufacturing will increase the competitiveness of wind energy compared to non-renewable sources. Copyright by ANGELA FARINA 2021 Ad meliora et maiora semper v ACKNOWLEDGMENTS I would sincerely thank my advisors Dr. Annick Anctil and Dr. M. Emin Kutay, for trusting me and this project. Without their guidance and support, I would never have become a Spartan. I learned a lot from them in these years at Michigan State University. I would also like to thank the committee members Dr. Rafael Auras and Dr. Bora Cetin for their availability and for providing good inputs and suggestions for my thesis. I want to extend my appreciation to Dr. Barbara Ruffino from the Polytechnic of Turin, Italy, for her friendship and collaboration. The Ph.D. journey would not have been incredible without my ‘A119-mates’: Dr. Eunsang Lee, Dr. Dipti Kamath, Dr. Seyed Mohammadreza Heidari, Siddharth Shukla (soon to be Dr.), and Ben Cecil. Thanks for our coffees and talks! I wish the best Ph.D. adventure to Cassandra Valcourt, Francis Hanna, and Luyao Yuan that just joined our group. I want to thank all my colleagues in the pavement group, especially Mahdi Ghazavi, for being consistently supportive. I appreciate my parents for their endless encouragement. I would never have arrived here without them. I want to dedicate a particular wish to my niece Martina to pursue whatever she desires for her future. Thanks to my family and friends in Italy for their encouragement, making easier the challenging periods. Last but not least, I would like to thank Michele for always having been with me, not making me perceive the 7,000 miles distance that has separated us in this long period. This experience that I strongly desired to do was probably somehow in my destiny. My educational journey started in my hometown Taranto (Italy), founded by Spartans, and it concluded here, at the motto of 'Spartan Will'! vi TABLE OF CONTENTS LIST OF TABLES ......................................................................................................................... ix LIST OF FIGURES ..................................................................................................................... xiii KEY TO ABBREVIATIONS ..................................................................................................... xvii 1. INTRODUCTION .................................................................................................................. 1 1.1 End-of-life tires in flexible pavements ............................................................................. 2 1.2 Potential toxicity of asphalt mixtures ............................................................................... 6 1.3 Material usage for wind turbines ...................................................................................... 8 1.4 Literature review ............................................................................................................ 10 1.4.1 Mechanical performance of CR mixes ................................................................... 10 1.4.2 Life cycle assessment .............................................................................................. 11 1.4.3 Toxicity assessment ................................................................................................ 13 1.4.4 Material usage for wind turbines ............................................................................ 15 1.5 Motivation and knowledge gaps .................................................................................... 17 1.6 Dissertation outline ........................................................................................................ 19 2. ENHANCED CRUMB RUBBER IN ASPHALT MIXTURES .......................................... 22 2.1 Preliminary study on pre-swollen crumb rubber in asphalt mixtures ............................ 22 2.2 Laboratory investigation of pre-swollen crumb rubber.................................................. 23 2.2.1 Dynamic modulus (|E*|) test ................................................................................... 24 2.2.2 Repeated Load Permanent Deformation (RLPD) test ............................................ 26 2.2.3 Disc Shaped Compact Tension (DCT) test ............................................................. 27 2.3 Field implementation of pre-swollen crumb rubber ....................................................... 29 2.4 Polymer coated rubber and devulcanized rubber in asphalt mixtures ............................ 30 2.4.1 Asphalt mixture design ........................................................................................... 31 2.4.2 Mechanical performance testing ............................................................................. 34 2.4.3 Results ..................................................................................................................... 37 2.5 Summary of chapter findings ......................................................................................... 42 3. MECHANISTIC-EMPIRICAL PAVEMENT ANALYSIS AND LIFE CYCLE ASSESSMENT ............................................................................................................................. 43 3.1 Analyses based on the Mechanistic-Empirical Pavement Design Guide (MEPDG) ..... 44 3.1.1 Methodology ........................................................................................................... 44 3.1.2 Results ..................................................................................................................... 53 3.2 Life cycle assessment ..................................................................................................... 61 3.2.1 Goal and scope ........................................................................................................ 61 3.2.2 System boundaries and functional unit ................................................................... 62 3.2.3 Life cycle inventory ................................................................................................ 64 3.2.4 Results ..................................................................................................................... 68 3.2.5 Material usage over the service life of the pavements ............................................ 82 3.3 Summary of chapter findings ......................................................................................... 83 vii 4. LEACHING ASSESSMENT: METALS CHARACTERIZATION .................................... 86 4.1 Introduction .................................................................................................................... 86 4.2 Methodology .................................................................................................................. 88 4.2.1 Sample preparation ................................................................................................. 88 4.2.2 Leaching testing procedure ..................................................................................... 88 4.2.3 Chemical analyses ................................................................................................... 90 4.2.4 Digestion and metals characterization .................................................................... 90 4.3 Results ............................................................................................................................ 92 4.3.1 Leaching test ........................................................................................................... 92 4.3.2 Assessment ratio ..................................................................................................... 97 4.3.3 Digestion and metals characterization .................................................................. 104 4.4 Summary of chapter findings ....................................................................................... 112 5. MATERIAL USAGE FOR THE USA AND GLOBALLY WIND TURBINES .............. 114 5.1 Methodology ................................................................................................................ 114 5.1.1 Historical material intensity .................................................................................. 115 5.1.2 Future material requirement .................................................................................. 116 5.2 Results .......................................................................................................................... 118 5.2.1 Historical material intensity .................................................................................. 118 5.2.2 Future material requirement .................................................................................. 119 5.2.3 Future environmental impact of material production ........................................... 127 5.3 Summary of chapter findings ....................................................................................... 129 6. CONCLUSIONS ................................................................................................................ 132 6.1 Environmental impacts of roads pavements................................................................. 132 6.2 Material usage for wind energy .................................................................................... 133 APPENDICES ............................................................................................................................ 136 APPENDIX A: Supplementary information Chapter 2 .......................................................... 137 APPENDIX B: Supplementary information Chapter 3 .......................................................... 141 APPENDIX C: Supplementary information Chapter 4 .......................................................... 158 APPENDIX D: Supplementary information Chapter 5 .......................................................... 173 REFERENCES ........................................................................................................................... 183 viii LIST OF TABLES Table 2.1 Mixture volumetric properties and gradations for asphalt mixtures ............................. 33 Table 2.2 Indirect tensile strength parameters .............................................................................. 41 Table 3.1 A summary of key inputs for the pavement structures analyzed in this study ............. 44 Table 3.2 Material-level damage models ...................................................................................... 48 Table 3.3 Damage accumulation model and transfer functions .................................................... 49 Table 3.4 Calibration factors for fatigue cracking model ............................................................. 50 Table 3.5 Calibration factors for rutting ....................................................................................... 51 Table 3.6 Limit values for pavement distresses at the end of the design life ............................... 53 Table 3.7 Number of years before distress line passed the threshold ........................................... 59 Table 3.8 Reference flow (ton/1-mile single lane) ....................................................................... 63 Table 3.9 Material inventory to produce 1 metric ton of hot mix asphalt for each alternative .... 68 Table 3.10 Price and economic allocation coefficient for scrap tires ........................................... 70 Table 3.11 Global warming potential (GWP) and fossil depletion (FD) of 1 ton of Control and rubberized asphalt mixtures compared to the SBS blend ............................................................. 73 Table 3.12 Environmental impact of electricity mix per kWh ..................................................... 78 Table 3.13 Construction equipment specifications ....................................................................... 78 Table 3.14 Fuel consumption to build the low and high traffic structures ................................... 78 Table 3.15 Difference in tons of asphalt mixtures used in Michigan to build the surface layers of low and high traffic structures over 50 years compared to the Control and SBS mixtures .......... 82 Table 4.1 Total metal concentration after 63 days ........................................................................ 92 Table 4.2 Net infiltration, P1 and P2, and number of ≤1 day and >1 days events, N1 and N2 .... 98 Table 4.3 Assessment ratio (AR) in Lansing, Grand Rapids, and Muskegon areas ................... 100 Table 4.4 Parameters used to calculate the assessment ratio ...................................................... 101 ix Table 4.5 Average leaching concentration (Ciav) and effective concentration C1 and C2, Lansing ..................................................................................................................................................... 102 Table 4.6 Average leaching concentration (Ciav) and effective concentration C1 and C2, Grand Rapids ......................................................................................................................................... 103 Table 4.7 Average leaching concentration (Ciav) and effective concentration C1 and C2, Muskegon ..................................................................................................................................................... 104 Table 4.8 Hypothetical P1 and P2............................................................................................... 104 Table 4.9 Material content in each mixture, in percentage ......................................................... 107 Table 4.10 Weight of the cylindrical specimens ......................................................................... 107 Table 4.11 Metals content in each material per mixture (mg/specimen) .................................... 109 Table A1 VECD results: Number of cycles to failure (Nf) for all mixtures calculated at 5 Hz . 140 Table B1 Asphalt binder content ................................................................................................ 142 Table B2 Measured |G*| of the asphalt binders used in all different mixtures ........................... 142 Table B3 Measured |E*| of the asphalt mixtures (psi) ................................................................ 143 Table B4 Creep compliance for all mixtures (1/psi) ................................................................... 144 Table B5 Traffic input data ......................................................................................................... 145 Table B6 Number of axles per truck ........................................................................................... 145 Table B7 Vehicle class distribution ............................................................................................ 146 Table B8 Volume monthly adjustment factors ........................................................................... 147 Table B9 Climate stations input.................................................................................................. 147 Table B10 Electricity generated in the USA in 2020 [146, 147] ................................................ 148 Table B11 Energy consumption to produce 1 t of gravel and sand. [120] ................................. 149 Table B12 Energy and Transport used to produce 1 ton of Reclaimed Asphalt Pavement ........ 149 Table B13 Imported crude oil in the USA in 2016 (Unit: Thousand barrels) ............................ 150 x Table B14 Unit processes used for 1 ton of imported crude oil, based on the databases available in SimaPro (Unit:tons) .................................................................................................................... 151 Table B15 Distances for imported crude oil in the USA in 2016 (Unit: km) ............................. 151 Table B16 Transportation of 1 ton of imported crude oil. .......................................................... 151 Table B17 Quantities of domestic crude oil transported between PADDs and related distances covered by pipeline, barge tanker, and rail in 2016 .................................................................... 152 Table B18 Materials and energy used to refine1 ton of crude oil ............................................... 152 Table B19 Input material for 1 kg of SBS .................................................................................. 153 Table B20 Input material for 1 ton of SBR................................................................................. 153 Table B21 Input material for 1 ton of Sasobit ............................................................................ 153 Table B22 Materials and energy used to produce 1 ton of crumb rubber................................... 153 Table B23 Material input and operation to obtain 1 ton of steel for........................................... 154 Table B24 Materials and energy used to produce 1 ton of DVR ................................................ 154 Table B25 Electricity and energy mix to produce 1 ton of hot mix asphalt. .............................. 154 Table B26 Cumulative energy demand (CED) of 1 ton of Control and rubberized asphalt mixtures compared to the SBS blend ......................................................................................................... 155 Table B27 Statistical data from Monte Carlo simulation ........................................................... 156 Table C1 Eluate concentrations, on average of the three replicates ........................................... 159 Table C2 Interval flux mass release, on average of three replicates ........................................... 160 Table C3 Cumulative mass release, on average of three replicates ............................................ 162 Table C4 Steps for the microwave digestion of the aggregates .................................................. 164 Table C5 Steps for the microwave digestion of the RAP ........................................................... 164 Table C6 Steps for the microwave digestion of PCR, DVR, SBS, Sasobit, and neat bitumen .. 164 Table C7 Metals content in each material used in asphalt mixtures (mg/kg) ............................. 165 Table C8 Metals content in Control mixtures (three replicates, mg/sample) ............................. 166 xi Table C9 Metals content in SBS mixtures (three replicates, mg/sample) .................................. 167 Table C10 Metals content in PCR dry mixtures (three replicates, mg/sample).......................... 168 Table C11 Metals content in PCR wet mixtures (three replicates, mg/sample) ......................... 169 Table C12 Metals content in DVR mixtures (three replicates, mg/sample) ............................... 171 Table D1 Processes used in Simapro .......................................................................................... 174 Table D2 Transportation of wind turbines components ............................................................. 175 Table D3 Normalized values for each material for global warming .......................................... 176 Table D4 Installed capacity and corresponding percentage with respect to the total generated electricity for each outlook ......................................................................................................... 179 Table D5 Capacity addition projected for onshore and offshore wind energy ........................... 179 Table D6 Future steel, cement, and rare earth elements production ........................................... 180 xii LIST OF FIGURES Figure 2.1.Dynamic modulus (|E|*) master curves of (a) 5E1 mixtures and (b) 4E1 mixtures .... 25 Figure 2.2. Repeated Load Permanent Deformation test results for (a) 5E1 and (b) 4E1 mixtures ....................................................................................................................................................... 26 Figure 2.3. Disc Shaped Compact Tension test’s specimen shape and specimen placed in the Material Testing System device during the test. ........................................................................... 27 Figure 2.4. Results of the Disk-shaped Compact Tension tests for (a) 5E1 and (b) 4E1 mixtures ....................................................................................................................................................... 28 Figure 2.5. Transverse crack on Control section not propagated on the PSCR lane .................... 30 Figure 2.6. Status of the PSCR section after four years from construction .................................. 30 Figure 2.7. Comparison of unconfined dynamic modulus at 10 Hz and 4, 20, 40°C for all mixtures ....................................................................................................................................................... 38 Figure 2.8. a) log-log, and b) linear-log unconfined dynamic modulus master curves of all mixtures ....................................................................................................................................................... 39 Figure 2.9. Repeated load permanent deformation test results for all mixtures. .......................... 40 Figure 2.10. Number of cycles to failure calculated at 5 Hz, 20°C, and 150, 200, 400 microstrains for all mixtures. ............................................................................................................................. 40 Figure 2.11. Indirect tensile strength comparison for all mixtures ............................................... 41 Figure 3.1. Framework of the Mechanistic-Empirical Pavement Design Guide. ......................... 47 Figure 3.2. Distresses of flexible pavements: a) fatigue cracking, b) thermal cracking, and c) permanent deformation ................................................................................................................. 52 Figure 3.3. Pavement failures overtime for low and high traffic structures in Michigan ............. 55 Figure 3.4. Pavement distresses overtime for low and high traffic structures in Idaho ................ 56 Figure 3.5. Pavement distresses overtime for low and high traffic structures in Florida ............. 57 Figure 3.6. Pavement distresses overtime for low and high traffic structures in California......... 58 Figure 3.7. Number of reconstructions over 50 years for the low and high traffic structures in all four climate zones ......................................................................................................................... 60 xiii Figure 3.8. Boundaries of the cradle-to-grave system for 1-mile single lane road pavement ...... 63 Figure 3.9. System boundaries considered for the crumb rubber production using a) cut-off method, b) economic allocation, and c) system expansion. .......................................................... 70 Figure 3.10. Contribution to Global Warming Potential of each process associated with the production of 1 ton of CR, using cut-off, economic allocation, system expansion, and system expansion considering the natural gas substitution. ...................................................................... 71 Figure 3.11. a) Global warming potential and b) fossil depletion comparison of PCR dry, PCR wet, and DVR mixtures to Control and SBS mixes based on the allocation methods ......................... 74 Figure 3.12. Uncertainty from Monte Carlo analysis for 1 ton of asphalt mixture using different allocation methods for a) global warming potential and b) fossil depletion................................. 76 Figure 3.13. a) Global warming potential (kg CO2eq) and b) fossil depletion (MJ Surplus) associated with the virgin bitumen content in asphalt mixtures ................................................... 77 Figure 3.14. Global warming potential (GWP), fossil depletion (FD), and cumulative energy demand (CED) from cradle-to-grave LCA of a low traffic pavement using reference and rubberized mixtures in a) Michigan, b) Idaho, c) Florida, and d) California ............................... 79 Figure 3.15. Global warming potential (GWP), fossil depletion (FD), and cumulative energy demand (CED) from cradle-to-grave LCA of a high traffic pavement using reference and rubberized mixtures in a) Michigan, b) Idaho, c) Florida, and d) California ............................... 80 Figure 3.16. Global warming potential in kg of CO2eq per 1-mile single lane over 50 years: contribution of each LCA phase for a) low traffic and b) high traffic pavement in Michigan ..... 81 Figure 4.1. a) Asphalt mix specimens, b) plastic support, and c) specimens placed in deionized water.............................................................................................................................................. 89 Figure 4.2. a) Eluate sample took from the bucket was filtered through a membrane by using a pump to speed up the process, b) 45 μm filter membrane, c) sample preparation for chemical analysis: 2% nitric acid (HNO3) added to the filtered eluate (10 mL).......................................... 90 Figure 4.3. a) Microwave digestor, b) Filtration of PCR and DVR digestated, and c) ICP-OES analysis.......................................................................................................................................... 91 Figure 4.4. Eluate concentrations for As, Cd, Cr, Cu, Pb, and Zn compared with lower limit of quantitation (LLOQ) and method detection limit (MDL)............................................................. 94 Figure 4.5. Interval flux rates of mass released over specific intervals of time, compared with the ideal diffusion trend ...................................................................................................................... 95 xiv Figure 4.6. Cumulative release of mass per exposed surface area of test samples compared with the trend line ................................................................................................................................. 96 Figure 4.7. pH and conductivity measured at each interval of the leaching test .......................... 97 Figure 4.8. Metals content in mg per kg of each material used in asphalt mixtures................... 106 Figure 4.9. a) Total mg of metals presents in each mixture, average of the three replicates; b-f) average of metals (mg) per specimen of each mixture, contribution of all materials. ................ 108 Figure 4.10 Metals content in mg per kg of Michigan, North Carolina, Wisconsin, and Maryland aggregates ................................................................................................................................... 111 Figure 4.11 Difference in mg of metals contained in asphalt mixtures samples ........................ 112 Figure 5.1. System boundaries including raw materials used for onshore and offshore turbines (foundation, tower, nacelle, and rotor blades) and transportation between assembly and construction site. Raw materials: m1=concrete; m2=gravel; m3=steel and iron; m4=copper; m5=aluminum; m6=fiberglass; m7=polyethylene and polymers; m8=epoxy; m9=rare earth materials. ..................................................................................................................................... 116 Figure 5.2. Material intensity for a) onshore and b) offshore wind turbines in terms of ton per MW from 1991 to 2017, based on data available in the literature. ..................................................... 119 Figure 5.3. Percentage of increase in material demand for concrete, steel and REEs compared to 2018 level for the USA (a, b, c, d) and the rest of the world (e, f, g, h) wind turbines. ............. 124 Figure 5.4.Material demand of concrete, steel and iron, REE based on DNVGL, Wind Vision, and EIA for the USA (a, b, c, d) and the rest of the world (e, f, g, h) compared to the expected material production. .................................................................................................................................. 125 Figure 5.5. Global warming potential (GWP) of onshore and offshore wind turbines in the USA based on a) DNVGL, b) Wind Vision, c) EIA outlook, and (d) in the rest of the world based on the DNVGL outlook, in terms of thousand tons of CO2 eq. ............................................................. 128 Figure 5.6. Cumulative Energy Demand (CED) of onshore and offshore wind turbines in the USA based on a) DNVGL, b) Wind Vision, c) EIA outlook, and in the rest of the world based on the DNVGL outlook (d), in terms of terajoule (TJ). ......................................................................... 129 Figure A1. C vs S curves for Control mixture, results of the VECD model .............................. 138 Figure A2. C vs S curves for SBS mixture, results of the VECD model.................................... 138 Figure A3. C vs S curves for PCR dry mixture, results of the VECD model ............................. 139 Figure A4. C vs S curves for PCR wet mixture, results of the VECD model ............................ 139 xv Figure A5. C vs S curves for DVR mixture, results of the VECD model .................................. 139 Figure B1. Blend gradation of aggregates based on the Superpave Aggregate Gradation Specification. .............................................................................................................................. 149 Figure B2. Cumulative energy demand comparison of PCR dry, PCR wet, and DVR mixtures to Control and SBS mixes based on the allocation methods........................................................... 155 Figure D1. Global warming potential (GWP) of onshore and offshore wind turbines in terms of ton CO2 eq./MW: a,b) critical and non-critical materials, c,d) non-critical materials contribution broken down into each material, and e,f) components and transportation contribution. ............ 177 Figure D2. Cumulative Energy Demand (CED) of onshore and offshore wind turbines in terms of GJ/MW: a,b) critical and non-critical materials, c,d) non-critical materials contribution broken down into each material, and e,f) components and transportation contribution. ........................ 178 xvi KEY TO ABBREVIATIONS AMPT: Asphalt Mixture Performance Testing AR: Assessment Ratio CED: Cumulative Energy Demand CR: Crumb Rubber DAF: Dilution Attenuation Factor DCT: Disc Shaped Compact Tension DFIG: Double Fed Induction Generator DDSG: Direct Driven Synchronous Generator DI: Deionized Water DOT: Department of Transportation DVR: Devulcanized Rubber Dy: Dysprosium ESAL: Equivalent Single Axle Load FD: Fossil Depletion FHWA: Federal Highway Administration GWP: Global Warming Potential HMA: Hot Mix Asphalt ICP-MS: Inductively Coupled Plasma Mass Spectometry ICP-OES: Inductively Coupled Plasma Optical Emission Spectrometry IDT: Indirect Tensile Strength LEAF: Leaching Environmental Assessment Framework LCA: Life Cycle Assessment xvii LLOQ: Lower Limit of Quantitation MDL: Method Detection Limit MEPDG: Mechanisti-Empirical Pavement Design Guide MTS: Material Testing System Nd: Neodymium NdFeB: Neodymium-Iron-Boron PAH: Polycyclic Aromatic Hydrocarbons PCR: Polymer Coated Rubber PM: Permanent Magnet PMSG: Permanent Magnet Synchronous Generator PSCR: Pre Swollen Crumb Rubber RAP: Reclaimed Asphalt Pavement REE: Rare Earth Elements RLPD: Repeated Load Permanent Deformation TCLP: Toxicity Characteristic Leaching Procedure SBR: Styrene-Butadiene-Rubber SBS: Styrene-Butadiene-Styrene SFIG: Single Fed Induction Generator SPLP: Synthetic Precipitation Leaching Procedure VECD: Visco-Elastic Continuum Damage VOC: Volatile Organic Compound WMA: Warm Mix Asphalt xviii 1. INTRODUCTION1 The construction of civil infrastructures implies the use of large volumes of materials. In the USA, most of the major civil infrastructures were built in the 1960s and many of them will reach the end of their service life soon [1]. Those billions of tons of materials currently in use will be potentially available for recycling in new infrastructures. Recycling contributes to the circularity of materials by reducing the environmental impact of civil infrastructures and minimizing the use of virgin materials and waste. However, some factors can limit the use of recycled materials like technical and economic feasibility of the recycling process (e.g., rare earth elements in wind turbines), materials stocked in infrastructures for extended periods (e.g., cement concrete in buildings, asphalt mixtures in roads pavements, steel, aluminum, copper in wind turbines), and material demand of infrastructures expected to grow fast (e.g., wind turbines). The research presented in this dissertation focused on the material intensity and the environmental assessment of flexible pavements and wind turbines. If properly designed and built, road pavements can last longer and need only periodically surface layers replacements, without major reconstructions of the entire structure. Moreover, recycled materials can improve the durability of asphalt mixtures, reduce the use of virgin materials and disposal in landfills. For wind turbines, the most challenging components are the blades since they are difficult to recycle and need to be replaced more often than the rest of the structure. The road pavement industry is one sector where end-of-life composite material from the blades can be reused to replace the natural aggregates or the asphalt binder modifier [2, 3]. 1 Part of this chapter have been published as Angela Farina, Annick Anctil, “Material consumption and environmental impact of wind turbines in the USA and globally”, Resources, Conservation & Recycling, DOI: https://doi.org/10.1016/j.resconrec.2021.105938 Copyright Elsevier 2021 1 Flexible pavements cover 95% of the paved roads in the USA. They require approximately 400 million tons of asphalt mixtures(natural aggregates and asphalt binder) for their construction and maintenance every year [4]. Polymer modifications can improve the durability of flexible pavements and reduce the quantity of virgin materials used over the service life. Crumb rubber (CR) from end-of-life tires might substitute the synthetic polymer in the asphalt mixtures. This can further reduce virgin materials and the amount of tires deposited in landfills. Moreover, the energy transition will require large-scale deployment of resources to guarantee the expected capacity from renewable sources in the next 30 years. For instance, wind power is one of the fastest-growing global energy. Its contribution to global electricity generation is expected to increase from 5 to 30% by 2050. This growing contribution will lead to increasing construction of wind farms globally. Therefore, the quantification of the materials is essential to ensure that the material demand will be lower than the future material production. Evaluating the environmental impact due to the production of the materials necessary to meet the expected wind capacity growth is important to avoid creating new issues related to material scarcity, which could increase the carbon footprint of future electricity production. 1.1 End-of-life tires in flexible pavements In the United States, more than 270 million tires from cars and trucks are disposed every year. Scrap tires can be used as fuel for energy recovery, aggregates for construction, and crumb rubber (CR) [5]. These applications have contributed to reducing the number of scrap tires ending in landfills from 75% in 1990 to 17% today [5, 6], but additional applications could further reduce this amount. The CR is currently used in playgrounds, sports turfs surfaces, and asphalt mixtures. In 2019, in the USA, asphalt pavements accounted for only 17% of the total CR market. This percentage could be increased by demonstrating the benefit of using scrap tires in asphalt mixes. 2 The material composition of scrap tires includes synthetic and natural rubber, carbon black, sulfur, zinc oxide as a vulcanizing agent, antioxidants, steel, and textile fibers. Tires from trucks have a higher natural rubber percentage and lower sulfur and zinc oxide content than car tires. Different technologies are used to reduce rubber from scrap tires into CR particles. The most common process is the mechanical ambient grinding, in which scrap tires are shredded using machinery that cuts tires into granules of different sizes. Shear and abrasion operations of blades and knives occur at room temperature. During this process, steel and textile fibers are removed and recycled. Another method is cryogenic grinding, where liquid nitrogen is used to freeze the rubber that is subsequently shredded by using a hammer mill. Finally, the last technique involves the use of pressurized water jets (55000 psi) to produce CR particles [7–9]. Mechanical ambient grinding is the most used technique to produce CR for asphalt applications. The CR has been used as an additive in asphalt pavements since the 1960s in the USA [10]. One significant advantage of using CR is that it can improve the mechanical performance of the traditional asphalt mixes if designed and blended properly [11, 12]. The traditional asphalt mixes are composed of aggregates and bitumen. Aggregates constitute almost 95% of the mix by weight and define the structure of asphalt mixtures, through which load is transferred. The bitumen, or asphalt binder, derived from the portioned distillation of crude oil, is used as ‘glue’ between the aggregates and useful for its waterproof, thermoplastic, and viscoelastic properties. The use of CR can be an alternative technique to the common polymer modification methods (e.g., Styrene- Butadiene-Styrene, SBS), or CR can be used in conjunction with polymers to improve mechanical performance. The CR is incorporated in the asphalt mixture through two broad methods: wet and dry techniques [13]. In the wet process, CR is blended with asphalt binder before mixing with aggregates. During the mixing phase at high temperatures (150-190°C), CR tends to absorb the 3 oily fraction of the bitumen [14, 15]. As a result, each rubber particle swells up to 2-3 times its original size, and it is thought that a layer of gel-like material is formed around the core of the rubber particles [16]. On the other hand, in dry technology, CR particles are added as partial replacement of the aggregates during the mixing phase of the asphalt mixture. In this case, when CR particles are in contact with bitumen, the phenomenon of interaction is minimal as compared to the wet process [17]. The wet process is the most adopted technique around the world. The wet asphalt mixes enhanced the resistance to permanent deformation and fatigue cracking of asphalt mixtures significantly [18–20]. However, the presence of such swollen particles is both an advantage and a significant drawback of wet technology. On one side, the final mixture takes advantage of the elastic properties of swollen rubber particles in the bitumen. On the other side, the swelling of rubber reduces the distance between the particles and hardens the unabsorbed bitumen, which leads to an overall increase in viscosity and brittleness [21]. The high viscosity can affect pumping, transportation, mixing, and lay-down stages. By comparison, the mixtures made with dry technology may exhibit an increase in fracture toughness and ductility, with a corresponding improvement in resistance to crack propagation and the overall elasticity of the mix [22]. Moreover, less reflective and thermal cracking result because CR particles absorb the stress, and less fatigue cracking occurs because the rubber absorbs and release the energy under repeated load condition [23]. However, problems related to homogeneity and compaction may happen with consequent deterioration of the pavement with loss of asphalt and aggregates (called raveling phenomenon) caused by moisture-related damage [24, 25]. Drawbacks of wet and dry technologies have been addressed by developing enhanced CR technologies. Some examples of these enhanced materials are the pre-swollen crumb rubber 4 (PSCR), devulcanized rubber (DVR), and polymer coated rubber (PCR). The PSCR is a pelletized rubber produced by the reaction of scrap tire rubber particles with bitumen-compatible oil at elevated temperatures. It does not require a prior blending with bitumen, and it can be added directly at an asphalt plant [26]. The absorption of the oil avoids the rubber absorbing the oily fraction of the bitumen (typical phenomenon issued in the wet technology using regular crumb rubber), which usually leads to increased viscosity of the asphalt binder [26]. The devulcanization process is used to improve the processability and plasticity of the rubber from tires, which then facilitates its use in pavement materials. The sulfur-to-sulfur bonds of the rubber can be broken by chemical, thermal, mechanical, and thermomechanical processes [27]. The DVR is used to modify the asphalt binder, and the particles are able to dissolve almost completely when mixed with the asphalt binder at high temperatures. The PCR is a chemically enhanced/polymer-coated rubber where a polymer film partially covers the surface area of the rubber particles [28]. The polymer promotes and controls the reaction with asphalt and allows the development of strong bonds between materials during the production process of the mixture. The advantage of the PCR is that it can be used as dry technology, where it can directly be added to aggregates at the asphalt plant. The use of CR from scrap tires is supported by the Departments of Transportation (DOTs) and the Federal Highway Administration (FHWA) when mechanical performance and costs of CR mixes are comparable with those of the asphalt mixtures modified with a synthetic polymer such as the SBS. Decision-making for pavement preservation, rehabilitation, or reconstruction has traditionally been driven by mechanical performance and cost. However, more recently, agencies started considering environmental impacts as well. Environmental assessment is essential to evaluate the actual benefits of recycled materials such as CR used in asphalt pavements. 5 1.2 Potential toxicity of asphalt mixtures The use of asphalt and recycled materials like reclaimed asphalt pavement (RAP) and CR leads to questions regarding their toxicity and potential for leaching and contaminating the environment. RAP contains aged asphalt binder and chemicals released on the road surface during the use phase, like lubricating oils, gasoline, metals from brake pads, and vehicle exhaust [29]. Asphalt binders include heavy metals and polycyclic aromatic hydrocarbons (PAHs), pollutants classified as carcinogenic, mutagenic, and teratogenic [29, 30]. In 2015, in a research study conducted in Italy on 16 different CR samples, CR was assessed to contain metals (e.g., iron, zinc, chrome), PAHs, and volatile organic compounds (VOCs) [8]. The PAHs are hydrocarbons, organic compounds containing only carbon and hydrogen, such as anthracene, tetracene, and benzopyrene. The VOCs are organic chemicals both humans made (e.g., from paints, adhesives, fossil fuel, like acetone, formaldehyde, and benzene) and naturally occurring by plants, animals, and microbes. In this mentioned research study, the CR, produced by mechanical and cryogenic processes, was provided by 11 different plants in Italy and Portugal. Two different methods performed the chemical analyses to assess the presence of metals, PAHs and VOCs. Metals were determined by using the inductively coupled plasma optical emissions spectrometer, PAHs and VOCs through solvent extraction and gas chromatographic analysis. Elemental analyses for carbon, hydrogen, nitrogen, and sulfur were also performed by employing a thermal conductivity detector. Carbon and hydrogen percentages resulted in being similar in all CR samples; instead, nitrogen and sulfur contents were variable, depending on the origin of the rubber was from car or truck. The level of cleanliness and purity in terms of metals were also different. However, all CR samples were found to have zinc content higher than 1% due to the oxide used for the vulcanization. Other studies also confirmed that zinc is the only metal 6 present in percentage by mass higher than 1% with respect to the other heavy metals [8, 31, 32]. In California, currently, more than 40 waterways have been found in exceedance of zinc with respect to the standard limit established by the Clean Water Act. Since the large use of rubberized asphalt pavements in California, the California Department of Transportation (CalTrans) attributed the zinc exceedance to the stormwater runoff from rubberized pavements [33, 34]. The calcium carbonate has been assessed to absorb the dissolved zinc from a solution [35]. Therefore, CalTrans suggested adding the calcium carbonate up to 3% of CR mass by weight to the asphalt mixtures to try to fix this issue [31, 33]. Leaching can be defined as the transfer of metals or pollutants into a liquid from a material. The leaching is typically quantified using batch and column experiments. The batch tests consist of 3 procedures: Toxicity Characteristic Leaching Procedure (TCLP) [36], Synthetic Precipitation Leaching Procedure (SPLP) [37], and Deionized Water Leaching Procedure. The TCLP determines the mobility of organic and inorganic substances in liquid, solid, and multiphasic wastes, and it is employed to simulate leaching through landfills and verify whether a waste is hazardous. The SPLP experiment is an alternative test used to simulate acid rainfalls in non-landfill conditions. The Deionized Water Leaching Procedure follows the same procedure of TCLP and SPLP, but the deionized water is used as a leaching solution. The column leaching experiments simulate what happens in the field and can be performed in saturated and unsaturated conditions. The Leaching Environmental Assessment Framework (LEAF) is an integrated framework that includes four laboratory methods for characterizing the leaching behavior of solid materials [38]. The four methods assess (i) liquid-solid partitioning as a function of extract pH using a parallel batch extraction procedure (method 1313), (ii) liquid-solid partitioning as a function of liquid-solid ratio for constituents in solid materials using an up-flow percolation column procedure (method 7 1314), (iii) mass transfer rates of constituents in monolithic or compacted granular materials using a semi-dynamic tank leaching procedure (method 1315), and (iv) liquid-solid partitioning as a function of liquid-solid ratio using a parallel batch extraction procedure (Method 1316). The methods can be applied individually or in combination. The LEAF approach has been used to evaluate the reuse of coal fly ash as road base material, and construction of embankments; municipal solid waste incineration bottom ash as road base; and secondary materials (e.g., coal fly ash, recycled concrete aggregate, furnace slags) used as partial substitutes for Portland cement or admixtures in cement and concrete construction products [39]. In Europe, leaching procedures analogous to the LEAF methods have been used to evaluate materials such as coal fly ash, recycled concrete aggregate, or municipal solid waste incinerator bottom ash for reuse in road bases and embankments [40, 41], and use of a byproduct from the aluminum industry as soil amendment [42]. In China, LEAF methods have been used to evaluate the environmental safety of the use of sewage sludge compost as an agricultural amendment [43, 44]. The advantage of the LEAF methods compared to the TCLP and SPLP is that LEAF is a flexible framework that can be tailored for use over a wide range of material types and release scenarios over time, while the TCLP and SPLP are single-point tests. 1.3 Material usage for wind turbines Onshore and offshore wind turbines are made of three main parts: the tower, rotor, and nacelle. Towers are made from steel and sometimes concrete, and their average height is 100 m [45]. The rotor includes two or three blades generally made of fiberglass and the hub. The blades lift and rotate, causing the rotor to spin. The nacelle contains the gearbox, generator, controller, and brakes [46]. There are two types of generators: fed induction generator (single SFIG or double DFIG) or direct driven synchronous generator (DDSG) [47, 48]. In the induction turbines (SFIG 8 and DFIG), the rotor is electromagnetic and made from copper coils. Electromagnetic rotors need a higher speed than direct-driven generators to induce a current and generate electricity. Therefore, a gearbox is required to increase the rotor’s speed inside the generator [48]. Some DDSGs use a permanent magnet (Permanent Magnet Synchronous Generator - PMSG) and both the DDSG and the PMSG are gearless. The generator rotates at low speed in the gearless direct drive because it is directly connected to the turbine rotor hub. To deliver a specific power, the lower speed requires higher torque. Hence, the direct-driven generators' size and weight are higher than for fed induction generators [47]. Globally, the most common wind turbines installed are those with the double induction generator (DFIG) (73.4%). Instead, the turbines using PM are 20.8%, and the direct driven are 5.8% [49, 50]. There are different types of permanent magnets (PMs) used in wind turbines, such as the samarium-cobalt and the neodymium-iron-boron (NdFeB) [48]. The most common PM in wind turbines is NdFeB because it is 2.5 stronger than the samarium-cobalt magnet [48]. The NdFeB was developed in the 80s by General Motors and its production has increased rapidly in China after 2000 [48, 51]. The NdFeB magnets are made from 65-70% iron (Fe), 1% boron (B), and 29- 34% rare earth elements (REEs) that are used for their magnetic properties [48]. Neodymium (Nd) (27-31% by weight of PM) is the most common REE in PMs. Dysprosium (Dy) (2-5% by weight of PM) is added to the magnet to improve the magnetic properties at high temperatures and reduce corrosion [48, 52]. REEs comprise 17 different elements with similar properties. They are abundant in nature but rare due to the number of deposits and the difficulty in the extraction. REEs are classified as light, medium, and heavy. Nd is light, while the Dy is a heavy element [53]. China is the primary producer of REEs globally, with the most significant deposit in Bayan Obo. In 2018, China extracted 75% of the total world production, followed by Australia (12%), the United States 9 (9%), Russia (2%), and India, and Brazil (1%) [54]. There is a growing concern about REEs consumption which are deemed critical due to their high economic importance and supply risk [55, 56]. Critical materials have unique properties and are difficult to substitute [57]. Materials criticality assessment considers the economic importance and the supply risk [55, 58]. 1.4 Literature review 1.4.1 Mechanical performance of CR mixes Performance of asphalt mixtures was improved in the last decades with the use of synthetic rubber such as the triblock copolymer Styrene-Butadiene-Styrene (SBS) as a modifier for asphalt binder in hot mix asphalt (HMA) or in warm mix asphalt (WMA) [59]. In recent years, the use of CR from scrap tires started to be widely adopted in lieu of, or conjunction with SBS. Asphalt mixes modified with CR with the wet process can be used in all climate conditions [60]. The CR, like most synthetic polymers, increases the elasticity of mixes and stiffens the bitumen. Asphalt mixtures containing CR typically show improved resistance to permanent deformation (rutting), fatigue (load-related distress), and reflective cracking (propagation of old cracks in rehabilitated surface layers) [7]. In the 1990s, most states in the USA had already adopted the use of CR in road pavements. The overall finding was that the wet mixes had better performance than dry mixes. In particular, the experience in California reported reduced reflective cracking and good performance with a reduced thickness of the surface layer. Moreover, the life cycle of pavements made with wet CR mixes was twice longer than that of traditional unmodified mixtures [20]. The main benefits experienced in the past years included improved durability, lower maintenance costs, and lower use of resources. A more recent study reported that the use of CR improves rutting and fatigue 10 resistance with respect to a traditional mix [61]. According to Feiteira Dias et al., dry CR mixes perform better than traditional mixtures and at the same level as wet mixes. Results of the dynamic modulus testing (measuring the stiffness of the material) suggest a more resistance to high temperature (rutting) for the dry mix, and the higher binder content improved the fatigue resistance as well as that of the wet mixes [62]. Typically, the CR content in the wet process is 15 to 20% by weight of the total binder, corresponding roughly to 0.5 to 1% by weight of the total mixture, and the typical size of the particles is 0.15 to 0.6 mm. By comparison, in the dry mixes, the CR percentage varies from 1 to 3% by weight of the total mixture, and the particle size is 0.85 to 6.4 mm [63]. However, lower CR content in the wet process may increase rutting and fatigue resistance [64]. According to Shirini and Imaninasab, the comparison among wet mixes modified with 10, 15, and 20% of CR, a mix containing 5% of SBS, and a traditional unmodified mix, showed that the stiffness of the mix with 10% CR content was almost equivalent to the SBS mix and higher than mixes with the other CR contents. Moreover, the traditional mix had higher stiffness than the mix modified with 20% CR content [65]. Results of the laboratory investigation of asphalt binder modified with SBS compared with DVR modified binder at different concentrations showed that 3% of DVR reached the same effect of 1% SBS in performance grade testing. In terms of fatigue, results indicated that the SBS performed slightly better than DVR [66]. 1.4.2 Life cycle assessment Life Cycle Assessment (LCA) is a comprehensive methodology used to assess and quantify the environmental impacts (e.g., Global Warming Potential, Cumulative Energy Demand, Human Health, etc.) of products and services [67]. It is a standardized methodology following the ISO 14040-44 [68, 69]. LCA considers all phases of the product’s life cycle, such as i) materials 11 production, ii) construction, iii) use, iv) maintenance, and v) end-of-life. When applied to road pavements, this methodology can help support paving solutions decisions by adding environmental factors to economic and mechanical performance. In literature, LCA was used by many practitioners to compare asphalt mixtures containing different types of recycled materials. LCA is particularly useful to evaluate the benefits of asphalt mixtures containing recycled materials such as reclaimed asphalt pavement (RAP, recycled material obtained from milling operations of old pavements used in partial substitution of the natural aggregates), CR, plastic, fly ash, and glass by estimating the resource depletion. Multiple studies using LCA have found that the global warming potential [70], cumulative energy demand [71], and all impacts from Recipe [72] and Eco-indicator 99 [73] methods were lower for mixes containing RAP compared to a traditional mix [74–76]. The Recipe method includes eighteen impacts like fossil depletion, acidification, ozone depletion, and human and ecotoxicity. Instead, the method Eco-indicator 99 uses dimensionless eco-indicators points to evaluate the environmental impacts in terms of human health, ecosystem quality, and resources. The carbon footprint and embodied energy are lowered by using 30-40% by weight of RAP in new mixes and lower mixing temperature (150°C instead of 160-170°C) [77]. The environmental benefit of using CR in current LCA literature is not very clear. The outcomes of one study reported a 16% increase in eco-points from the Eco-indicator 99 method by using CR compared to the unmodified mix [74]. The higher results were associated with a higher asphalt binder content and heat usage to produce the rubberized mix. However, the same thickness of the surface layer and equal service life were used for all mixes considered. Other studies showed that these two factors, thickness and service life, play an important role since CR improves the mechanical performance of road pavements. The global warming and the cumulative energy demand reduction varied from 36 to 12 45% when using CR instead of the traditional mix, due to higher durability and reduced thickness of the surface layer [75, 76]. 1.4.3 Toxicity assessment In the existing literature, the potential leaching of the RAP used in new mixes has been assessed. For instance, in 1999, Brantley et al. [29] performed leaching tests to determine pollutants from six samples of RAP collected in different locations in Florida. The authors performed the leaching experiments in batch (TCLP and SPLP) and in columns to assess PAHs and VOCs, heavy metals (Ba, Ca, Cr, Cu, Ni, and Zn). PAHs and VOCs were determined by using a Gas Chromatograph Mass Spectrometer, while metals through the Atomic Absorption Spectrometer. All samples passed the tests and therefore were not characterized as hazardous waste. Legret et al. [78] analyzed the percolating water from different core samples containing 10% and 20% of RAP. The leaching tests were performed by following the standard French procedure (AFNOR XP X31-210 [79]) on several RAP samples collected from the RN 76 highway in France. The authors evaluated pH, hydraulic conductivity, the number of hydrocarbons, and heavy metals. They obtained leachate concentrations below the detection limits. However, results derived from batch leaching tests were lower than standard values for drinking water, while from the column tests, the authors found pollutant concentrations higher than the drinking water limit in the initial phase of leaching. However, these values decreased rapidly below the detection limits after the initial peak. Results showed a significant presence of heavy metals, except for lead. Nevertheless, concentrations remained below the limit values for drinking water. In 2012, Shedivy and Meier [30] performed TCLP, DI, and column tests on samples of RAP collected in five locations in the United States (Ohio, Wisconsin, California, New Jersey, and Colorado). They compared the number of heavy metals and PAHs in RAP samples with laboratory asphalt mixes. 13 Results reported concentrations of manganese and arsenic significantly higher than the minimum concentration level allowed in drinking water. A recent study carried out a literature review on the leaching and environemtnal assessment risk for RAP [80]. The authors reviewed 17 studies for a total of 41 RAP sources. The following constituents exceeded the limits established by the US EPA RSLs (Regional Screening Levels) for tap water: for metals, arsenic (7 RAP sources), lead (13 RAP sources), antimony (8 RAP sources), and manganese (5 RAP sources); for PAHs, naphthalene (6 RAP sources), dibenz(a,h)anthracene (6 RAP sources), benzo(a)anthracene 5 RAP sources), benzo(a)pyrene (4 RAP sources) [80]. The high zinc content in CR samples concerned many research studies in the literature. Rhodes et al. [32] stated that an assessment of zinc leaching data from 14 studies published in the literature indicates that increasing zinc leaching is associated with lower pH and longer leaching times. Moreover, in the same study, the correlation between the size of the CR particles and zinc leaching was investigated. Results reported that zinc leaching increases with smaller CR and longer exposure time. Another study assessed the zinc concentration in eluate solutions from leaching test in batch on CR particles from passenger car and truck tires and different particle sizes [31]. Results of this study reported that CR from passenger cars had a higher concentration of zinc than CR from truck tires, and smaller CR particles contained more zinc than bigger CR particles [31]. In another study, four types of CR from car, truck, and CR produced with the cryogenic procedure were tested with the SPLP and the column leaching test to assess the release of 23 metals (e.g., zinc, chromium, copper, lead, mercury) and 15 semi-VOCs like aniline, phenol, and benzothiazole. The zinc was the only metal leached from the CR for every sample tested, with an average concentration close to the groundwater standard. The aniline, detected in all samples, had the highest concentration of the selected compounds. The phenol content was 13 times the 14 groundwater standard. The benzothiazole was higher in CR made from truck tires than in samples from car tires [81]. In 2003, the asphalt mixes modified with CR were tested by performing leaching tests in batch to assess the presence of organic and metallic contaminants. The benzothiazole, mercury, and aluminum were found in the collected leachate in potentially harmful concentrations. Moreover, the leachate was tested for algae (Selenastrum Capriconutum) and animal species (Daphnia Magna). Results reported a gradual increase in the toxicity of the leachates with leaching time from both tests, associated with the gradual release of aluminum, mercury, and benzothiazole [82]. 1.4.4 Material usage for wind turbines There are different estimations of material demand for wind turbines in the literature. The future global use of non-critical materials for the energy transition has been assessed and, except for copper, there was no concern for the other materials, such as steel and concrete [83, 84]. For all applications including wind, the steel reserves were estimated to be enough for 100 to 200 years maintaining the same production rate as in 2009 [85]. Materials like sand, gravel, and limestone for concrete production were deemed not scarce and easy to recycle and reuse [83]. The total demand for copper in 2050 was estimated to be 200-350% higher than in 2010 [84]. The cumulative global copper production is expected to exceed the reserves by 2040 [84]. For wind turbines, the consumption of non-critical materials was not considered an issue either in the USA [86] or in Germany [87]. For the USA, there should not be a shortage of any of the abundant materials for the construction of wind turbines by 2030 [86]. For German wind turbines installation until 2050, the demand for iron, steel, copper, and aluminum was calculated to be less than 6% of the current domestic consumption [87]. There will be no aluminum shortage soon, and the peak production might happen in 20 to 40 years [88]. The bauxite reserves are still increasing. The 15 scarcity happens gradually through different stages, the first of which is when the peak is reached, the demand goes up, but the production is flat with an increasing price [88]. It appears that we are far from this stage today. The annual demand for wind turbines (6.8 Mt of concrete, 1.5 Mt of steel, 310,000 t of cast iron, and 40,000 t of copper) was estimated to meet 20% of the electricity-generating capacity by wind energy in the USA in the period 2011 to 2030 [86]. These results suggested that there should not be a shortage of any of the principal materials. However, there may be shortages from a manufacturing perspective if the total demand for these materials from all markets is greater than the available supply of raw materials or component manufacturing capacity [86]. The only possible expected shortage was estimated for REE with an annual demand of 380 t by 2030 [86]. An expected shortage of REE was also identified in other studies. In 2019, a cumulative Nd demand of 15 kt was estimated to meet the 80 GW capacity for the USA offshore wind turbines until 2050 based on the Wind Vision forecast, assuming that all turbines will be installed with a permanent magnet [89]. With the current growth REE demand from the electric vehicles and global wind energy, the USA wind sector will deal with a strong competition for Nd [89]. In another study, the cumulative global REE requirement was estimated to be between 460 and 902 kt in 2021-2050 period, with a consequent needed to expand the REE supply 11-26 times the current level to meet the global demand [90]. The results were estimated based on the Global Wind Energy Council outlook (2,870 GW in 2050) and comparing different scenarios based on the market share of turbines using REE, installed capacity, material efficiency of the wind turbine, and lifetime. The global need of Nd for permanent magnets was estimated to become four times higher in 30 years, and the related Chinese production rate was considered insufficient to satisfy the demand [83]. By comparison, the REEs requirement for the USA wind installation was estimated 16 to be 4-12% of the total production in the USA (from Mountain Pass, Bear Lodge, and phosphate rock mines), assuming an installation rate in the USA of 10 GW/year, with 20% of the total turbines using direct PM driven generator, [91]. The gap between global supply and global demand of REEs was expected to increase in the last decade [92]. Moreover, whether or not the REEs’ demand will exceed the supply will also depend on their demand for other applications (e.g., other technologies) [93]. Considering the entire electricity sector (oil, coal, natural gas, nuclear, biomass, wind, and solar), the total material requirement associated with the mineral production was expected to increase by 200 to 900% globally by 2050 due to an increase demand for copper, silver, cobalt, and steel [94]. Another study estimated that the raw materials demand for green technologies (i.e., renewable energy and mobility sector) will increase by 30% in 2050 [95]. The materials with the highest demand (six times higher than current values) were cobalt, lithium, magnesium, titanium, and zinc, which are not used in wind turbines. The total raw material exergy demand (expressed in Mtoe) associated with wind energy was negligible compared to the other sources, such as bioenergy, battery electric vehicles, and plug hybrid electric vehicles [95]. 1.5 Motivation and knowledge gaps In the past, the use of CR in road applications has been widely investigated from a technical and economic perspective, to assess the interaction between CR and bitumen and the influence of the rubber content on the mechanical performance. However, a comprehensive evaluation including environmental impact of asphalt mixtures containing CR is still necessary to evaluate whether the trade-off between recycled modifiers and synthetic polymers is viable. The existing LCA literature on rubberized asphalts focuses on the use of wet and dry technologies referred as regular CR in asphalt mixes. Asphalt mixes modified with enhanced CR like PCR and DVR have 17 not been evaluated from the mechanical, environmental, and toxic perspectives. Today, more accurate software and approaches in the determination of mechanical performance over time have been developed, allowing more realistic simulations. The use of the mechanistic-empirical pavement analysis method (e.g., the AASHTOWare Pavement ME software) is considered the best method to predict the long-term performance of road pavements. This approach can reduce the number of unjustified assumptions about the performance and lifetime, which are commonly found in LCA studies. For this reason, in this dissertation, the mechanistic-empirical pavement analysis methods results were used in LCA. In the literature, the health risks associated with the potential leaching from the asphalt mix using CR were assessed specially to simulate landfill conditions or to evaluate the potential leaching from materials tested individually and not englobed in the asphalt mixture. Although scrap tires are classified as non-hazardous waste, results of leachate tests from literature suggested a potential for environmental problems. The knowledge gaps include the assessment of innovative and emerging CR technologies in terms of mechanical performance over time, environmental impacts, and toxicological effects in field conditions. A comprehensive comparison of the enhanced CR technologies with synthetic polymer modifying agents and unmodified mixes is also missing and essential to understanding whether the trade-off is viable. In literature, there are studies assessing the future use of materials in specific regions, studies only focusing on the consumption of critical materials, and studies evaluating the consumption of selected materials for all energy technologies. There is a knowledge gap about the future use of critical and non-critical materials for onshore and offshore wind turbines in the USA and the rest of the world compared to the expected material availability. In addition to material 18 availability, we need to ensure that manufacturing wind turbines will have a lower carbon impact than our current electricity production. The overall goal of this dissertation was to provide a comprehensive evaluation (including engineering performance and sustainability aspects) of enhanced CR technologies and evaluate the material demand to construct wind turbines in the USA and globally compared to the expected material production until 2050. 1.6 Dissertation outline The first goal of this study was to evaluate whether asphalt mixtures containing recycled materials such as CR from end-of-life tires might have as good as or better environmental impact than mixes modified with synthetic polymers and unmodified mixtures. The environmental impact was quantified by considering the entire life cycle of pavements, including the mechanical performance of the asphalt mixtures. Better mechanical performance increases the design life and consequently decrease the materials used for their construction over the service life. In addition, the potential leaching of the metals from asphalt mixtures containing CR were evaluated in this study and compared to the reference mixtures. The reference mixtures included unmodified materials and materials modified with a synthetic polymer. The second goal of this study was to estimate the material needed to build wind turbines that will guarantee the growing demand for wind energy in the next 30 years in the USA and globally. The material estimation is important to ensure that there will be enough materials in the future and that the production of these materials will not increase the environmental impact of future electricity production. The specific objectives of this study were: 1. Evaluate mix design and mechanical performance of Control, SBS, PCR dry, PCR wet, and DVR asphalt mixtures through laboratory testing. 19 2. Use mechanistic-empirical pavement analysis methodology to forecast the progression of various distresses with time to determine the reconstruction schedule over the service life of the pavements built with various materials (rubber-based and control). 3. Conduct a life cycle assessment to quantify and compare environmental impacts associated with all five asphalt mixtures. 4. Evaluate leaching potential of various metals out of asphalt mixtures through leaching test. 5. Estimate the material demand for wind turbined in the USA and globally compared to the expected material production until 2050, and quantify the environmental impact associated to the material production. Chapter 2 presents preliminary results on a previous investigation on a type of enhanced CR in Michigan: the pre-swollen crumb rubber. The experience with this technology used to modify the asphalt mixtures led to satisfying results in the field. The investigation then assessed other CR products, such as the DVR and the PCR, on which this study focused. This chapter also includes an evaluation of the mechanical performance of the five mixtures (Control, SBS, PCR dry, PCR wet, DVR) in terms of dynamic modulus, permanent deformation, fatigue, and thermal cracking. Chapter 3 presents the results of the mechanistic-empirical pavement analyses of two structures using the AASHTOWare Pavement ME software. These structures were designed to withstand low and high traffic with 2 and 3 asphalt layers, respectively. Each of the structures was analyzed using five different asphalt mixtures in four climate zones in the USA. Results of the laboratory investigation in Chapter 2 were used to calibrate the models in the Pavement ME software for an accurate prediction of the mechanical behavior of the structures considered. Moreover, Chapter 3 includes the results of an LCA of the five asphalt mixtures considering the 20 entire life cycle of the road pavement. The LCA included quantification of global warming potential, fossil depletion, and cumulative energy demand of the low and high traffic structures. A sensitivity analysis is also presented to assess three different allocation methods (i.e., cut-off, economic allocation, and system expansion) for crumb rubber. Moreover, a cradle-to-gate analysis of the asphalt mixture production is compared to the cradle-to-grave assessment to highlight the importance of including the mechanical performance of the asphalt mixtures when calculating their environmental impact. Results from Chapter 2 and the first part of Chapter 3 were used as input in the Simapro 9.1 software for the LCA. In order to present a comprehensive assessment of the asphalt mixtures containing PCR and DVR compared to the reference mixtures (unmodified Control and SBS), Chapter 4 includes the results of leaching tests to evaluate whether or not there is a potential of leaching of metals from the asphalt mixtures to the groundwater. The 1315 Method from the EPA LEAF framework was used to conduct the experiments. Moreover, Chapter 4 presents the results of the microwave digestion and the consequent metal trace analysis conducted on every single material used in the asphalt mixtures to better understand and interpret the leaching results. Chapter 5 reports the results of an analysis on the material demand for wind turbines in the USA and the rest of the world and comparisons of those values with the expected production until 2050. The demand for critical (rare earth elements, REEs) and non-critical materials (e.g., concrete, steel, copper, fiberglass) was forecasted by considering the size-scaling effect of onshore and offshore wind turbines and the future capacity. Chapter 5 quantifies the carbon footprint and the cumulative energy demand associated with the material production based on the annual capacity installation. 21 2. ENHANCED CRUMB RUBBER IN ASPHALT MIXTURES2 In this chapter, first, the use of a CR product named pre-swollen crumb rubber (PSCR) is presented as a preliminary investigation conducted in Michigan on the use of enhanced CR materials in asphalt mixtures. Secondly, the use of the polymer coated rubber (PCR) in dry and wet mixtures and the devulcanized rubber (DVR) were evaluated. The specific objective of this part of the work was to assess if the mechanical performance of rubberized mixtures were as good or better than the reference mixtures. 2.1 Preliminary study on pre-swollen crumb rubber in asphalt mixtures To produce the PSCR, rubber particles are blended with a small amount of bitumen- compatible oil at elevated temperatures. During this procedure, rubber particles absorb the oil and swell, which mimics the wet technology, where the crumb rubber particles are directly in contact with the bitumen and absorb its oily fraction. When the regular crumb rubber absorbs the oily fraction of the bitumen in the wet technology, the CR particles tend to swell, simultaneously reducing the inter-particle distance and the cause hardening of the base bitumen, which leads to an overall increase in viscosity. On the other hand, the PSCR is added to the aggregates and the bitumen at an asphalt plant and pre-swollen crumb rubber particles do not absorb the oil in neat bitumen. This is useful since the oily fractions stay in the neat bitumen, improving the durability. The only interaction between the pre-swollen CR particles and neat bitumen is thought to be simple 2 Parts of this chapter have been published as Angela Farina, M. Emin Kutay, Michele Lanotte, “Laboratory and Field Performance Investigation of Pre-Swollen Crumb Rubber Modified Asphalt Mixtures”, International Journal of Pavement Research and Technology, ISSN: 1997-1400 DOI: https://doi.org/10.1007/s42947-020-0191-0. Copyright Springer Nature 2021 22 adherence to each other, creating a bonded flexible medium between the aggregates. Therefore, the use of the PSCR addresses the drawbacks of using the regular crumb rubber in wet technology. The PSCR can be defined as an improved dry process, a product that is instantly ready to use and can eventually be pre-mixed with the reclaimed asphalt pavement (RAP) and then added to the aggregates. None of the asphalt plants in Michigan have the equipment needed to produce wet technology, making the PSCR a valuable option to scrap tire rubber in roadways. 2.2 Laboratory investigation of pre-swollen crumb rubber A laboratory investigation was carried out to evaluate the performance of two plant produced PSCR-modified asphalt mixtures (surface and intermediate course). For comparison purposes, the investigation included two mixtures with neat bitumen (Control) and one with SBS polymer-modified bitumen (SBS). The Control and the PSCR asphalt mixtures were used for the construction of a field test sections in August 2015 in Haslett, Michigan, while the SBS-modified mix was used for laboratory comparison only. The results of a visual inspection of the status of the pavement structure are here reported in support of the laboratory findings. The mixes evaluated in this study were designed in accordance with the Michigan Department of Transportation (MDOT) specifications [96]. In the MDOT specifications, asphalt mixtures are designated by an alpha- numerical code in which the first character indicates the design layer, and the following denotes the design traffic. For example, a 5E1 mixture is a top layer designed to withstand 1 million equivalent single axle loads (ESALs). Mixture code starting with ‘5’ are for those mixtures designed for surface layer only, while ‘4’ can be used for both surface and intermediate courses, depending on the project. PSCR and Control mixtures were produced using the same PG58-28 bitumen, while the SBS-modified bitumen was a PG70-28 produced with 3% SBS starting from a neat PG58-28 23 bitumen. The PSCR, provided by Liberty Tire Recycling, was treated as fine aggregates and pre- mixed with the RAP before introducing it into the heating drum of the asphalt plant. A single PSCR content of 0.6% by weight of the total mixture was investigated. The SBS mixture has not been used in the field as this type of material is not commonly employed for the construction of low-volume traffic roads in Michigan. However, the laboratory comparison of this mixture with the PSCR mix was needed to assess the potential of PSCR technology in high-traffic roads. 2.2.1 Dynamic modulus (|E*|) test The dynamic modulus (|E*|) is a fundamental viscoelastic material property for asphalt mixtures. Typically, 4” diameter and 6” tall samples are prepared for the |E*| test by cutting and coring 6” diameter 7” tall specimens compacted using a gyratory compactor. The samples were tested using the Asphalt Mixture Performance Tester (AMPT) in uniaxial cyclic compression mode at different temperatures (25, 10, 1 and 0.1 Hz) and loading frequencies (4ºC, 20ºC, 40ºC, and 54ºC). Three replicates were tested for each temperature/frequency combination. The applied load level is small enough such that there is no damage induced to the specimen. This test provides the linear viscoelastic (LVE) |E*| master curve of the asphalt mixture, which is a key input of the mechanistic-empirical pavement design software to predict pavement distresses such as fatigue cracking and rutting. The |E*| master curves at a reference temperature (Tref=21°C) were obtained in accordance with the AASHTO T378 and R84 protocols [97, 98], using the time-temperature superposition (TTS) principle. A second-order polynomial function (Eq. 2.1) and a sigmoidal model (Eq. 2.2) were used to fit shift factors a(T) and the |E*| master curve, respectively. log ( aT ) = a1 (T 2 - Tref2 ) + a2 (T - Tref ) Eq. 2.1 24 b2 Eq. 2.2 ( ) log E * = b1 + 1 + exp ( -b3 - b4 log ( f R ) ) where Tref is the reference temperature, a1 and a2 the shift factor polynomial coefficients, fR the reduced frequency (log(fR)=f·a(T)), and c1, c2, c3, and c4 the sigmoidal coefficients. The dynamic modulus master curves of PSCR, Control, and SBS mixtures are shown in Figure 2.1. Results revealed that the presence of the pre-swollen rubber particles have a softening effect on the mixtures when compared to the Control and SBS mixes. In particular, the PSCR curve seems to be simply shifted down with respect to the Control mixture, indicating the possibility to have better performance for low temperature cracking (high reduced frequencies, low- temperatures), but more susceptible to rutting (low reduced frequencies, high-temperatures). The master curve of the SBS mixture, compared to Control 5E1 and 4E1 mixes, shows the peculiar behavior of the polymer modified asphalt. The |E*| is higher at high temperature (low frequency) and lower at low temperature (high frequency) which indicates potential better performance for both rutting and thermal cracking. 5 5 10 10 (a) (b) 5E1 PSCR 4E1 PSCR 4 5E1 Control 4 4E1 Control Dynamic Modulus (MPa) Dynamic Modulus (MPa) 10 10 4E1 SBS 4E1 SBS 1000 1000 100 100 10 -5 10 0.001 0.1 10 1000 10 5 10 -5 5 10 0.001 0.1 10 1000 10 Reduced Frequency (Hz) Reduced Frequency (Hz) Figure 2.1.Dynamic modulus (|E|*) master curves of (a) 5E1 mixtures and (b) 4E1 mixtures 25 2.2.2 Repeated Load Permanent Deformation (RLPD) test The repeated load permanent deformation (RLPD) test evaluates the susceptibility of asphalt mixtures to rutting in the field. A haversine load with a duration of 0.1s is applied, followed by a 0.9s rest period. The test was run at a relatively high temperature (54°C) and permanent (accumulated) strain at the end of each loading-resting cycle is measured. Flow Number (FN) is defined as the cycle at which the mixture starts to flow, or the rate of permanent deformation accelerates. The FN is typically correlated to the field rutting. Mixtures with high FN are less prone to rutting than the mixtures with low FN. The RLPD test results reported in Figure 2.2 are in agreement with the results of the dynamic modulus tests discussed above. The 5E1 and 4E1 PSCR mixtures are more prone to the accumulation of permanent deformation with respect to either the Control or the SBS mixtures. During the test, the 4E1 PSCR is the only type of material to reach the tertiary flow (Figure 2.2 (b)), a clear indication of the high susceptibility to rutting. As expected, the SBS mixture performed better than the control and PSCR mixtures in both 5E1 and 4E1 mix types. 4 4 2 10 6 10 (a) (b) 4 5 10 Plastic Microstrain 4E1 PSCR Plastic Microstrain 4 1.5 10 5E1 PSCR 5E1 Control 4 4 10 4E1 Control 4E1 SBS 4E1 SBS 4 4 1 10 3 10 4 2 10 5000 4 1 10 0 0 4 4 0 2000 4000 6000 8000 1 10 0 2000 4000 6000 8000 1 10 Number of Cycles Number of cycles Figure 2.2. Repeated Load Permanent Deformation test results for (a) 5E1 and (b) 4E1 mixtures 26 2.2.3 Disc Shaped Compact Tension (DCT) test The Disc Shaped Compact Tension (DCT) test was performed in accordance with the ASTM D 7313 specification [99]. Figure 2.3 shows the typical shape of the specimens for the DCT test. The DCT specimens were placed in a controlled chamber and conditioned for 2 hours at - 10°C. After conditioning, the specimens were inserted in the loading fixture and subjected to a pre-load not higher than 0.2kN (Figure 2.3). Samples were tested in a Material Testing System (MTS) device under tensile loading with a constant crack mouth opening displacement (CMOD) rate of 1mm/min. Figure 2.3. Disc Shaped Compact Tension test’s specimen shape and specimen placed in the Material Testing System device during the test. The results of DCT tests are illustrated in Figure 2.4. Figure 2.4(a) shows the comparison between the response of those mixtures used in the wearing course of pavement structures. This comparison is of extreme interest as thermal cracking usually appears in the top layer and then propagates throughout the structure. Even though the peak of the load is comparable between the 5E1 PSCR and Control mixtures, the shape of the load-displacement curves clearly indicates a more ductile behavior of the 5E1 PSCR. In fact, the total area under the curve, which represents the energy needed to create and propagate the crack into the test specimen, is higher for the PSCR 27 mix. Thus, in the field, cracks will be more likely to happen in the sections with Control mixtures. In the case of PSCR, time to crack initiation may be similar to the Control but the rate of crack growth is expected to be slower, which is a desirable behavior. On the other hand, the HMA with SBS-modified bitumen showed a peculiar behavior, significantly different from both the PSCR and Control mixtures. The pre-loading peak part of the curve reveals that the energy needed for initiating the crack into the material is very high and not comparable with those of the PSCR and Control mixtures. However, the speed at which the crack propagates is higher than what is observed for the PSCR mixture. Similar trends were observed among the materials used in the leveling layers in the field (Figure 2.4(b)). It can be noticed that the behavior of the 5E1 and 4E1 Control mixtures is comparable. The energy needed for the formation of the crack in the case of the 4E1 mixture is slightly higher, but the rate of crack propagation is analogous between the two mixes. The two PSCR-modified mixes, however, showed different behavior. In fact, even though the loading peak is the same, the 4E1 mix reaches it much faster than the 5E1 mix indicating a lower energy level needed for creating the crack. 3.5 3.5 (a) (b) 3 3 2.5 2.5 Load (kN) Load (kN) 2 2 1.5 1.5 1 1 0.5 0.5 0 0 0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2 Displacement (mm) Displacement (mm) 5E1 PSCR 1 5E1 Control 1 4E1 SBS 1 4E1 PSCR 4E1 SBS 1 4E1 SBS 3 5E1 PSCR 2 5E1 Control 2 4E1 SBS 2 4E1 Control 4E1 SBS 2 5E1 PSCR 3 5E1 Control 3 4E1 SBS 3 Figure 2.4. Results of the Disk-shaped Compact Tension tests for (a) 5E1 and (b) 4E1 mixtures 28 2.3 Field implementation of pre-swollen crumb rubber The results of a visual inspection of the status of the pavement structure are here reported in support of the laboratory findings. The field construction was conducted in the summer of 2015 in Haslett, Michigan. The asphalt mixture layers of the existing pavement structure were milled and replaced for a total length of around 2 km in both traffic directions. All test sections were placed on the top of the natural aggregate base. The performance of the test sections after five years of service life are acceptable. Only a few transverse cracks are visible on the surface of the sections with 5E1 Control mixture in the top layer. Figure 2.5 shows an example of a transverse crack propagating towards the entire 5E1 Control section. However, it was interesting to see that the cracks did not propagate into the lane with the PSCR mix. The field observations are in agreement with the DCT test results. It is also worth noting that the PSCR sections stayed a lot darker than the control mixtures. This is a significant issue for the visibility of the road marking for many State agencies. Even though lab results led to concerns related to the poor rutting performance, no evidence of rutting was observed in the field. This was despite the fact that the rubber-modified section was at a left-turn lane at the intersection, where braking/acceleration and slow movement of vehicles are common (Figure 2.6). Since rutting is more likely to happen in the early service life of the pavement structure, the visual inspection of the field section indicates that, for low-traffic volume roads, the PSCR mixtures can be safely used. 29 Figure 2.5. Transverse crack on Control section not propagated on the PSCR lane Control section PSCR section Figure 2.6. Status of the PSCR section after four years from construction 2.4 Polymer coated rubber and devulcanized rubber in asphalt mixtures After the successful application of the pre-swollen CR in the field, the EGLE (Environment, Great Lakes & Energy – State of Michigan) funded another study to investigate the feasibility of using two additional new technologies: polymer coated rubber (PCR) and the devulcanized rubber (DVR) in asphalt mixtures. This portion of the work includes the development of the asphalt mixture designs based on the Superpave guidelines and the hot mix asphalt production manual for Michigan [100, 101]. The main objective of this task was to obtain the 30 recipes of the mixes involved in the investigation. All mixtures needed to have a similar mix design in terms of aggregates, target design air voids content, and percentage of binder used. Once the asphalt mixture designs were completed, each mixture was tested for: • Dynamic modulus (|E*|) - confined and unconfined testing, • Repeated load permanent deformation (RLPD) test, • Fatigue cracking test, • Indirect Tensile (IDT) strength test. 2.4.1 Asphalt mixture design Five different mixtures were considered in this work. The design of the mixes consists of three steps: (i) selection of the gradation of the aggregates, (ii) selection of bitumen type, and (iii) determination of the optimum asphalt binder content. During the last step, multiple trial mixes with different binder contents are prepared until Superpave volumetric requirements are met. This process simulates the actual manufacturing of the asphalt mixes, and it results in the recommended recipe of the specific asphalt mix. As part of this stage, three different CR mixtures were tested and compared with the reference blends: the SBS modified mixes and the Control unmodified mixture. All mixtures are hot mix asphalts, laboratory prepared at temperatures between 150-175°C. The mixtures are: • Control mix: unmodified mix, • SBS mix: bitumen modified with 3.5 % of SBS and 0.4% of cross-linker by the weight of the total asphalt binder, • PCR dry mix: mixture modified with 0.5% PCR by weight of the asphalt mixture (dry technology), 31 • PCR wet mix: bitumen modified with 7% of PCR, 0.5% SBS, and 0.4% of sulfur catalyst solution (cross-linker), 3% Sasobit additive by weight of total bitumen, • DVR mix: bitumen modified with 7% of DVR and 2% of SBS, 0.4% of sulfur catalyst solution (cross-linker) by weight of total bitumen, in wet technology. 32 Table 2.1 Mixture volumetric properties and gradations for asphalt mixtures Mixture ID Control(1) SBS(2) PCR dry(3) PCR wet(4) DVR(5) E30 ≤ 30 E30 ≤ 30 E30 ≤30 E30 ≤30 E3 ≤3 Design ESALs millions millions millions millions millions NMAS (mm) 12.5 12.5 12.5 12.5 12.5 Performance grade PG58-28 PG70-28 PG58-28 PG82-28 PG70-28 Asphalt binder (%) 5.48 5.83 5.67 5.83 5.7 Design gyration (Ndesign) 109 109 109 109 86 Design air voids (%) 3.5 2.9 3.3 2.9 3.0 VMA (%) 15.2 14.7 17.0 14.8 18.9 VFA (%) 77.1 80.2 80.6 80.5 63.2 RAP (%) 15 15 15 15 20 Sieve size Percent passing, % inch mm 1" 25.4 100.0 100.0 100.0 100.0 - 3/4" 19 100.0 100.0 100.0 100.0 100.0 1/2" 12.5 91.1 91.1 91.1 91.1 99.3 3/8" 9.5 86.1 86.1 86.1 86.1 89.4 #4 4.75 72.8 72.8 72.8 72.8 65.9 #8 2.36 51.6 51.6 51.6 51.6 53.3 #16 1.18 36.2 36.2 36.2 36.2 40.9 #30 0.6 25.6 25.6 25.6 25.6 28.4 #50 0.3 15.7 15.7 15.7 15.7 12.9 #100 0.15 6.4 6.4 6.4 6.4 6.2 #200 0.075 3.7 3.7 3.7 3.7 4.5 Note: (1) Control = unmodified mixture. (2) SBS = Styrene-Butadiene-Styrene modified asphalt binder. (3) PCR dry = Polymer-Coated Rubber modified asphalt mixture dry technology. (4) PCR wet = Polymer-Coated Rubber modified asphalt mixture wet technology. (5) DVR = Devulcanized Rubber modified asphalt binder. 33 The mixtures compared in this study were prepared in the Michigan State University asphalt materials laboratory. Mixtures had similar aggregate gradation with a nominal maximum size of 12.5 mm, and they were designed with target air voids of 3±0.5% for surface layers. All mixtures had ESALs equal to 30 million, except the DVR mixture that was designed for a lower- level traffic (3 million) because this mix was part of a previous project. The RAP content in DVR mix (20% by weight of the total mix) also differed from the other mixtures (15% by weight of the total mix). A summary of the mix design characteristics and volumetric properties are provided in Table 2.1. 2.4.2 Mechanical performance testing The loose mixes produced in the asphalt laboratory at Michigan State University based on the mix design were compacted by using the Superpave Gyratory Compactor to prepare the performance testing samples. The samples had a cylindrical shape with a height of 180 mm and a diameter of 150 mm. The cylindrical samples were then cut and cored based on the shape required in the standard for each test. All specimens must have 7±0.5% air void content to reproduce the field conditions. The dynamic modulus and the repeated load permanent deformation testing were described, respectively, in sections 2.2.1 and 2.2.2. A description of the fatigue cracking test and the indirect tensile strength is provided below. 2.4.2.1 Tension-Compression Fatigue Cracking (FC) test In this test, 3” diameter and 6” tall asphalt samples were prepared by cutting and coring specimens compacted using a gyratory compactor. Then, displacement-controlled (0.12-0.15 mm) cyclic fatigue tests were conducted. The testing frequency was 5 Hz, the strain level was between 250-350 microstrain and the test was conducted at 10°C and 20°C. Data were then analyzed 34 following the Visco-Elastic Continuum Damage (VECD) model in order to compute the number of cycles to failure which represents the most common indicator of a mixture’s susceptibility to fatigue cracking. The VECD is the most common model used to analyze results of fatigue tests correlates the fatigue life (Number of cycles to failure - Nf) to tensile strain (! ! ) and modulus (E) of the asphalt mixture. The VECD theory uses the elastic–viscoelastic correspondence principle and the Schapery’s work potential theory to model the mechanical behavior of asphalt mixtures [102]. Based on this principle, equations for a viscoelastic material are equivalent to equations of an elastic material through the use of the ‘pseudo-strain’ instead of the actual strain [102]. The pseudo-strain is calculated as in Eq. 2.3: ! 1 +! Eq. 2.3 ! " (#) = " ( '(# − *) ,* ' +* # where, ! " is the pseudo-strain, ER is a reference modulus, E(t) is the linear viscoelastic relaxation modulus, t is time and τ is the (time) variable of integration. As damage accumulates in a continuum material, the deviation of the stress from the pseudo-strain is quantified in terms of a parameter called ‘pseudo-stiffness’ (C) and the damage parameter (S). C is the ratio of modulus at a certain time of loading to the initial modulus, calculated as follows (Eq. 2.4): |' ∗ |$ -$ = ∗ |' |&'( Eq. 2.4 where, CN is the pseudo-stiffness at the end of each cycle, |E*|N is the dynamic modulus for each cycle and |E*|LVE is the LVE (initial/undamaged) dynamic modulus. The S represents the amount of damage. The failure criterion is set to a pseudo-stiffness reduction of 50% (C=0.5) [102]. From the C versus S curves, the number of cycles to failure were calculated using the following relationship (Eq. 2.5) 35 ,! -. !#* |' ∗ |*&'( ,- /) = 0 1− 4 5 6∆3/ Eq. 2.5 2 ,3 +!, ,01 where, Nf is the number of cycles to failure, ε0 is the strain amplitude, ‫׀‬E*‫׀‬LVE is the LVE dynamic modulus, α is the damage exponent (1/m, where m is the maximum slope of the relaxation modulus vs. time in log-log scale), and f is the frequency. 2.4.2.2 Indirect Tensile (IDT) strength test The IDT test is used for the prediction of thermal cracking [103]. In this test, conducted at -10°C, a controlled load is applied on a cylindrical specimen diametrically across the circular cross-section. The loading causes a tensile deformation perpendicular to the loading direction, which yields a tensile failure. The IDT strength of the tested materials is computed by using the relationship between the maximum load before sample failure and the geometric characteristics of the specimen. However, due to an insufficient amount of material of the mixtures used in this work, the IDT tests could not be completed. For all mixtures, the IDT strength model in Eq. 2.6 was used: !" = −9.901 + 20.737,- + 2.674,0#$%& − 6.407,0'() + 0.669120 + 356.593415 + Eq. 2.6 1.027,1/2′′ + 2.517,3/8′′ − 3.768,#4 + 5.151,#8 + 3.452,#100 − 62.733,#200 − 0.0171: where, PM is a polymer modification factor (either 1 for polymer modified or 0 for unmodified binder), PGHigh is the high PG temperature, PGLow is the absolute value of low PG temperature, ANG is the angularity percentage, FAR is the fines/asphalt ratio, P1/2” is the percent passing the 1/2” sieve, P3/8” is the percent passing the 3/8” sieve, P#4 is the percent passing the #4 sieve, P#8 is the percent passing the #8 sieve, P#100 is the percent passing the #100 sieve, P#200 is the percent passing the #200 sieve, and AV is the percent of air voids. 36 It should be noted that the equation above was developed using 200+ mixtures tested in MSU, using the local materials in Michigan [103]. Therefore, its accuracy would be quite reasonable for the mixtures included in this testing program. 2.4.3 Results Figure 2.7 shows a comparison between rubberized and reference mixtures in terms of measured unconfined dynamic modulus, at 10 Hz frequency and at 4, 10, and 40°C temperatures. Results indicate that at low temperatures, the PCR dry mix may perform better than other mixes because mixes with lower stiffness are typically more flexible and exhibit better cracking resistance. On the other hand, at high temperatures, the PCR wet and the DVR mixtures are less prone to rutting than other mixes, because their moduli were higher. Figure 2.8 shows the master curve of the mixtures. The part of the curves shown in Figure 2.8a represents the behavior of the materials at low frequencies and high temperatures. Better rutting resistance is typically expected from the mixtures with a higher dynamic modulus at high temperatures/low frequencies. In Figure 2.8b, the curves represent the behavior of the materials at high frequencies and low temperatures, where better cracking resistance is typically expected from the mixtures with a lower dynamic modulus at low temperatures/high frequencies. 37 1.4 10 4 4 1.2 10 Control Dynamic Modulus (MPa) 1 10 4 SBS PCR dry 8000 PCR wet DVR 6000 4000 2000 0 4 20 40 Temperature ( 0C) Figure 2.7. Comparison of unconfined dynamic modulus at 10 Hz and 4, 20, 40°C for all mixtures Figure 2.9 shows results on the repeated load permanent deformation test (RLPD). The curves represent the accumulated deformation as a function of the number of cycles. As shown, the PCR dry and wet mixtures accumulated more plastic deformation than DVR, SBS, and Control mixes. This indicates that PCR mixtures are potentially more prone to rutting in the field. Except for the PCR wet mixture, these results are inconsistent with the dynamic modulus results. Figure 2.10 shows the results of the fatigue tests run on the mixes investigated in this study. The number of cycles to failure values depicted in the figure were calculated at a frequency of 5 Hz, temperature of 20°C, and strain levels of 150, 200, and 400 microstrains. These results are generally in agreement with the dynamic modulus results, where softer mixtures (at intermediate temperatures) performed better in fatigue. The SBS and the PCR dry mixtures had better cracking resistance than the other investigated blends. Moreover, the PCR dry mix had a number of cycles to failure higher than the SBS mixture with increasing microstrains. Table 2.2 shows the parameters used in the mathematical model to predict the IDT, and the IDT values are shown in Figure 2.11. The mixtures containing CR (PCR dry and wet, DVR) had 38 better resistance to thermal cracking than the reference blends (Control and SBS mixes). Except for the DVR mix, these values confirmed the trend observed in the dynamic modulus master curves. 5 10 a) 4 10 Dynamic Modulus (MPa) 1000 Control Control SBS 100 PCR dry PCR wet DVR 10 1 -8 -6 4 6 8 10 10 0.0001 0.01 1 100 10 10 10 Reduced frequency (Hz) 4 3 10 b) 2.5 10 4 Control Dynamic Modulus (MPa) Control 2 10 4 SBS PCR dry PCR wet 4 1.5 10 DVR DVR 1 104 5000 0 -8 -6 4 6 8 10 10 0.0001 0.01 1 100 10 10 10 Reduced frequency (Hz) Figure 2.8. a) log-log, and b) linear-log unconfined dynamic modulus master curves of all mixtures 39 5 104 Control Control 4 10 4 SBS Plastic microstrain PCr PCRdry dry PCR wet 3 104 DVR DVR 2 104 1 104 0 100 1000 104 Number of cycles Figure 2.9. Repeated load permanent deformation test results for all mixtures. 8 10 150 microstrain 107 Number of cycles to feilure (Nf) 200 microstrain 106 400 microstrain 5 10 4 10 1000 100 10 1 Control SBS PCR dry PCR wet DVR Figure 2.10. Number of cycles to failure calculated at 5 Hz, 20°C, and 150, 200, 400 microstrains for all mixtures. 40 Table 2.2 Indirect tensile strength parameters Control SBS PCR dry PCR wet DVR PM 0 1 0 1 1 PG High 58 70 58 82 70 PG Low 28 28 28 28 28 ANG % 41 41.2 41 41.2 42.5 FAR 0.72 0.72 0.95 0.72 0.89 P 1/2" 91.1 91.1 91.1 91.1 99.3 P 3/8" 86.1 86.1 86.1 86.1 89.4 P#4 72.8 72.8 72.8 72.8 65.9 P#8 51.6 51.6 51.6 51.6 53.3 P#100 6.4 6.4 6.4 6.4 6.2 P#200 3.7 3.7 3.7 3.7 4.5 Av % 7 7 7 7 7 500 Indirect tensile strength (psi) 400 300 200 100 0 Control SBS PCR dry PCR wet DVR Figure 2.11. Indirect tensile strength comparison for all mixtures 41 2.5 Summary of chapter findings The objective of this chapter was to investigate the mechanical performance of the asphalt mixtures modified with CR (PCR dry, PCR wet, and DVR) compared to the reference mixtures (Control and SBS). The major findings have been summarized as follows: • Based on the dynamic modulus at low temperatures, the PCR dry mix had better cracking resistance than other mixes because it was less stiff (lower modulus). Instead, at high temperatures, the PCR wet and the DVR mixtures were less prone to rutting than other mixes, because their moduli were higher. • The repeated load permanent deformation tests indicated that the plastic deformations of the PCR dry and PCR wet mixtures were higher than that measured for the DVR, SBS, and Control mixes, meaning that the mix modified with PCR was more prone to rutting. Except for the PCR wet, these results confirmed the dynamic modulus results. • The SBS and the PCR dry mixtures had better cracking resistance than the other investigated blends, based on the push-pull test. These results confirmed the dynamic modulus results. Moreover, the PCR dry mix had a number of cycles to failure higher than the SBS mixture with increasing microstrains. • PCR dry, PCR wet, and DVR mixes had better resistance to thermal cracking than the reference mixtures (Control and SBS mixes). Except for the DVR mix, these values confirmed the trend observed in the dynamic modulus master curves. 42 3. MECHANISTIC-EMPIRICAL PAVEMENT ANALYSIS AND LIFE CYCLE ASSESSMENT Although laboratory tests can provide valuable insight into the expected performance of asphalt mixtures, it is important to predict the long-term response of the pavement structures using the Mechanistic-Empirical Pavement Design Guide (MEPDG) approach [104]. This is because MEPDG approach can consider many factors such as the fluctuations on the pavement temperatures (and their effect on the modulus), oxidative aging, change in traffic with time etc. The main objective of this part of the project was to use data from all the laboratory tests as input to calibrate models that can predict failures such as fatigue cracking, thermal cracking, and rutting which can occur over the service life of pavements. Results were used to demonstrate whether the CR modified mixes performed similarly to SBS mixture and better than the Control unmodified blend. In addition to performance results, the modeling approach was used to calculate the reconstruction schedule over the lifetime of the pavement. Life cycle assessment (LCA) was used to assess the environmental evaluation of the two reference and the three rubberized asphalt mixtures. I used data from the mechanical performance investigation and from the MEPDG results as input for LCA. The inventory for the asphalt mixture composition (type and quantity of materials) was based on the mix design reported in Chapter 2. The reconstruction schedule, as well as the thicknesses of the surface layer, from the MEPDG. The goal of the LCA was to assess if asphalt mixtures containing CR form scrap tires had better or similar environmental performance as the reference mixtures, and what rubberized mixture among those considered in this work was environmentally better for Michigan and the other climate zones (Idaho, Florida, and California). 43 3.1 Analyses based on the Mechanistic-Empirical Pavement Design Guide (MEPDG) 3.1.1 Methodology The analyses with the MEPDG software were performed with the highest level of accuracy, i.e., Level 1. Level 1 requires site-specific input data determined from actual measurements. Two structures were selected to run the analyses, among the road pavements existing in Michigan. Table 3.1 reports the main characteristics of the two structures. The low traffic structure, designed for 3 million ESALs, had two asphalt concrete layers (top and base courses). Instead, the high traffic structure, designed for 30 million ESALs, had a thicker asphalt concrete package, composed of top, leveling, and base courses. The ESALs is a standard factor expressing the number and weight of all axle-loads from the different vehicles expected during the pavement design life of a road pavement, expressed in 80 kN equivalent single axle load. Table 3.1 A summary of key inputs for the pavement structures analyzed in this study Low traffic High traffic structure structure ESALs (million) 3 30 Asphalt concrete layers Thickness (in) - Top 1.5 1.5 Thickness (in) - Leveling - 2 Thickness (in) - Base 2 3 Modulus (E* - psi) from master curves Unbound base layer Thickness (in) 8 6 Resilient modulus MR (psi) 33,000 33,000 Unbound subbase layer Thickness (in) 18 18 Resilient modulus MR (psi) 20,000 20,000 Subgrade Thickness Semi-infinite Semi-infinite Resilient modulus MR (psi) 5,200 7,000 44 The five asphalt mixtures considered in this work (Control, SBS, PCR dry, PCR wet, DVR) were used as asphalt concrete layers of the two structures. Although a multi-layer structure is usually composed of different types of asphalt mixtures to deal with the different pavement distresses, in this work, the same asphalt mixture was used for all layers to study the overall behavior of the material with respect to all pavement failures. The design life of the pavement structures was set to 20 years, and the results obtained were used to project the number of reconstructions over a service life of 50 years. The analyses were performed for the wet freeze climate conditions of Michigan and results were compared to the other climate zones (Idaho - dry freeze, Florida - wet no-freeze, and California - dry no freeze). The main input data implemented in the AASHTOWare Pavement ME software, reported in Appendix B (Table B1 - Table B9), were: • Resilient modulus and thickness for unbound layers, as reported in Table 3.1. • Volumetric parameters (air voids, binder content), dynamic shear modulus and phase angle of asphalt binders, dynamic modulus of mixes at various frequencies and temperatures, creep compliance, indirect tensile strength, for asphalt layers. • Traffic data (e.g., vehicles distribution, axles per truck). • Climate data for Michigan (wet freeze zone), Idaho (dry freeze zone), Florida (wet no- freeze zone), and California (dry no-freeze zone). Figure 3.1 illustrates the framework of the MEPDG methodology implemented in the Pavement ME software. The methodology has been developed to use the input data to predict field distresses. The main steps are: 1) Initially, a finite-difference climatic model is run to predict the hourly pavement temperatures and moisture in base and subbase layers during the entire analysis period. The 45 pavement temperatures are used in conjunction with the frequency estimated from the vehicle speed to compute the moduli of the asphalt pavement sublayers. 2) Layered elastic analyses were performed (through the algorithm JULEA) to calculate strains due to the traffic loading in critical locations inside the structure, depending on the distress investigated (e.g., at the bottom of the asphalt mix layer for fatigue cracking). 3) Strains are then used as input in the materials damage models, shown in Table 3.2, to calculate the material level damage parameters. The parameters are Nf = number of cycles to failure for fatigue cracking (bottom-up, bu, and top-down, td); εp = plastic strain for rutting, for asphalt and unbound layers (named asphalt concrete, AC, and unbound respectively); ΔC = Change in thermal crack length, used in thermal cracking. 4) The parameters are finally used in the damage accumulation models and transfer functions to calculate the field distresses (Table 3.3). 46 Figure 3.1. Framework of the Mechanistic-Empirical Pavement Design Guide. 47 Table 3.2 Material-level damage models Distress Material model in MEPDG Bottom-up fatigue 1 (!" )!" 1 (!# )!# !!"#$ = $!% %!% &&&"#$ ' * ' * (asphalt layers) )'"#$ + (!" +!" Top-down fatigue 1 1 (!# +!# !!"'* = $!% %!% &&&"'* ' * ' * (asphalt layers) )'"'* + Rutting ),"-. = $/% ,-($" # !($# 0 (asphalt layers) )/ Rutting )5 7 % ),"$1#2$1* = )3 $4% ' * . "689 (unbound layers) )/ Low-temperature ∆& = 0∆1 1 cracking (asphalt layers) 0 = 10(& (;.=>?"@.A@ BCD(EF' 1) Description of the parameters: Nf-bu and Nf-td = number of cycles to failure for bottom-up and top-down cracks, respectively, βf1,2,3, βr1,2,3, kf1,2,3, C, CH, = calibration factors, εt-bu and εt-bu = tensile strain at the bottom and at the surface for bottom-up and top-down cracks, respectively, E = dynamic modulus of the asphalt material, εp-AC and εp-unbound = plastic strain for asphalt mix and unbound layer, respectively, εr = resilient strain, εν= vertical strain, a, b, c = empirical constant, T = temperature, N =number of cycles, ΔC and ΔK = change in crack length and in stress intensity, respectively, σm = indirect tensile strength of the mixture, n and m = coefficients. 48 Table 3.3 Damage accumulation model and transfer functions Distress Damage accumulation model Transfer function IJ Bottom -up fatigue 5H 1 &;"#$ 8; 3#$ = 4 6&#$ = ' * ' * (asphalt layers) !!"#$"H 60 1 + . . ()*+ .( L.")*+ ." BCD M*+ HK% IJ Top-down fatigue 5H &;"'* 3'* = 4 6&'* = 10.56 ' * (asphalt layers) !!"'*"H 1 + . .()&, .( L.")&, BCD M&, HK% 8N Rutting (asphalt layers <3 = 4 ),H ℎH None and unbound) HK% Δ&H = 0(Δ1H )1 IJ Low temperature & log J 2Kℎ L cracking (asphalt &2 = 4 Δ&H P0 -& = $'0% ! 〈 〉 layers) HK% M 1 = A'H, B0.45 + 1.99&55.AO E Description of the parameters: Dbu and Dtd = bottom-up and top-down crack damage, respectively, Nf-bu-i and Nf-td-i = number of cycles to failure for bottom-up and top-down cracks for the period i, respectevely, TP= number of periods, ni = traffic cycles in a period, εpi = εp-AC and εp-unbound = plastic strain in sublayer I for asphalt mix and unbound layer, respectively, C0 = crack depth, K = stress intensity factor, σtip = stress at crack tip, C1,2,4-bu and C1,2,4-td and βtc1= field calibration factors, λ= standard deviation, hac = height of the asphalt mix layer, N〈 〉 = standard normal distribution. 49 3.1.1.1 Calibration of the models There are two different types of calibration in the MEPDG methodology. The first calibration is at the level of the material damage models (Table 3.2), and the second one is at the level of the transfer functions (Table 3.3). Results from laboratory testing were used to calibrate the material damage models. The transfer functions were calibrated using the default field calibration factors valid for the USA. A laboratory test for fatigue cracking (push-pull test) was performed at different temperatures and strain levels. Test results consist of different numbers of cycles to failure Nf (measured), one for each temperature and strain. The equations for bottom-up and top-down fatigue cracking in Table 3.2 calculate the Nf(predicted). The calibration of the model means to vary the β factors included in the equations to have the minim error between the measured and predicted Nf. Table 3.4 shows β factors calculated at different temperatures and microstrain levels at 10 Hz, for all mixtures, by using the VECD model. The rutting model was calibrated using the curves of plastic strain versus cycle obtained with the flow number test. The β factors were calculated to minimize the error between the measured and calculated plastic strain (Table 3.5). The k coefficients are global calibration factors, and the default values were used in both cases. Table 3.4 Calibration factors for fatigue cracking model k factors k1 k2 k3 3.75 2.87 1.46 β factors β1 β2 β3 Control 5.106 1.173 0.987 SBS 2.226 1.831 1.617 PCR dry 2.462 1.324 1.035 PCR wet 2.206 1.350 1.109 DVR 3.147 1.768 1.562 50 Table 3.5 Calibration factors for rutting k factors k1 k2 k3 -2.45 3.01 0.22 β factors β1 β2 β3 Control 266.0 0.086 1.164 SBS 31.80 0.197 1.330 PCR dry 689.1 0.013 1.141 PCR wet 0.095 0.600 1.673 DVR 0.115 0.567 1.579 3.1.1.2 Distresses of flexible pavements Asphalt pavements fail due to distresses that develop over their service life. The main factors influencing the occurrence of the pavement distresses are traffic, environment (e.g., temperature, moisture), subgrade soil support, materials used for construction (mix design). Fatigue cracking occurs in areas subjected to repeated traffic loadings (Figure 3.2a). Cracks usually initiate at the bottom of the asphalt layers where the tensile stress is the highest, then propagate to the surface as one or more longitudinal cracks. This is commonly referred to as bottom-up fatigue cracking. However, cracks can also initiate from the top of the surface due to tire-pavement interaction creating areas of high localized tensile stresses (top-down cracking) [105, 106]. Thermal cracking is a phenomenon where transverse cracks propagate perpendicularly to pavement centerline are mostly attributed to shrinkage of the asphalt mixture placed on the surface due to low temperatures or to the asphalt binder hardening (Figure 3.2b) [105, 106]. Permanent deformation, or rutting, is a longitudinal surface depression in the wheel path, as shown in Figure 3.2c. Rutting is a temperature-dependent distress that accelerates at high temperatures, mostly due to structural deficiency, asphalt mixture design, or compaction issues. 51 Figure 3.2. Distresses of flexible pavements: a) fatigue cracking, b) thermal cracking, and c) permanent deformation Roughness can be defined as the variation in the longitudinal pavement elevation in the wheel paths, referred to the pavement profile. Pavement profile consists of a wide range of wavelengths having different amplitudes affecting the excitation of the vehicles, depending on their traveling speed and dynamic characteristics (e.g., suspension configuration, wheel and frame inertial properties) [107]. The irregularities in the pavement profile affect the ride quality of a vehicle and the costs in terms of fuel consumption and maintenance [108, 109]. The International Roughness Index (IRI) is a standardized roughness measurement, expressed in m/km or in/mi, representing the vehicle suspension motion over the traveled distance. The IRI can be generally described ad indicative of ride comfort [106]. Results obtained by the MEPDG software were the prediction of pavement distresses over the design life compared to thresholds values, limits representing the maximum values allowed to ensure that a pavement will perform satisfactorily over its design life [104]. Each distress requires limit and reliability value. The limits used in this work, reported in Table 3.6, were established by the Michigan DOT (MDOT) User Guide for Mechanistic-Empirical Pavement Design [110]. Reliability is the probability that the distress will be less than the limit over the design life. For example, a reliability of 95 would indicate that there is a 95% chance that the distress will not exceed the limit value entered during the design life. By comparison, this also means that there 52 would be a 5% chance that the distress will exceed the limit value. The reliability used in this work was 95%, as also adopted by MDOT. Table 3.6 Limit values for pavement distresses at the end of the design life Terminal IRI (in/mile) 172 Permanent deformation - AC layers (in) 0.50 AC bottom-up fatigue cracking (% lane area) 20 AC top-down fatigue cracking (% lane area) 25 AC thermal cracking (ft/mile) 1,000 3.1.2 Results Figure 3.3 shows the simulation results of the long-term response of the pavement structures in Michigan, performed using the MEPDG software. For both low and high traffic levels, the Control mixtures failed for bottom-up fatigue cracking occurring, respectively, after 13.67 and 9.83 years from the construction (Figure 3.3 e and f). Except for IRI (International Roughness Index), the distresses of other mixture types did not exceed the limit. The IRI, calculated as the summation of the fatigue cracking, permanent deformation, and thermal cracking, increased linearly over time. However, the trend shows a sudden increase due to the thermal cracking prediction trend (Figure 3.3-Figure 3.6 a and b). Figure 3.3 c and d show that the accumulated permanent deformations (rutting) were higher for the PCR dry and the Control mixtures compared to the other mixes, although all were below the limit. The behavior of the DVR mix was the best against the rutting in the low and high traffic structures. These results were consistent with those obtained from the RLPD tests. The transverse cracks due to the thermal cracking propagated in the pavement built with the DVR mixture in the low traffic pavement, and the failure reached the limit after 13.83 years from the construction (Figure 3.3i). 53 Similarly, in the high traffic structure, DVR and Control mixtures reached the thermal cracking limit after 11.83 and 12.75 years, respectively. However, the thermal cracking results were not considered in this work for the calculations of the number of reconstructions. The reason is that the thermal cracking model implemented in the MEPDG software needs to have an accurate local calibration to minimize the error between the predicted thermal cracking and the field observations [111]. 54 Figure 3.3. Pavement failures overtime for low and high traffic structures in Michigan 55 Figure 3.4. Pavement distresses overtime for low and high traffic structures in Idaho 56 Figure 3.5. Pavement distresses overtime for low and high traffic structures in Florida 57 Figure 3.6. Pavement distresses overtime for low and high traffic structures in California 58 The thermal cracking prediction was worse in Idaho, having a dry freeze climate, (Figure 3.4i,j), but also in Florida and California that are no-freeze zones (Figure 3.5 and Figure 3.6, respectively), where all mixtures reached the threshold limit of the transverse cracking. This is clearly evidence of the fact that the thermal cracking predictions of the MEPDG software are not accurate. The Control mixture reached the limit for the bottom-up cracking for all climates and the two structures, except for the low traffic level in Idaho, where the IRI was reached before the fatigue cracking (Figure 3.4 a and i). The PCR wet in California in the high traffic structures failed for permanent deformation before the other distresses (Figure 3.6 d). It should be noticed that the results calculated considered a reliability factor of 95%. In practice, this means that there is only a 5% chance that the pavement structure will fail before the predicted time. The number of reconstructions over the service life was computed by dividing the number of years between the construction (year 0) and the cross of the limit reported in Table 3.7 by the total service life of the pavement (50 years). Figure 3.7 summarizes the number of reconstructions for the two structures in all climate zones. The IRI was the distress that drove the calculation of the reconstructions in the majority of the cases because most of the mixtures reached the limit of the IRI before that of the other distresses. Table 3.7 Number of years before distress line passed the threshold Low traffic High traffic Michigan Idaho Florida California Michigan Idaho Florida California years Control 13.67 13.08 14.67 18.25 9.83 11.00 7.25 9.08 SBS 19.75 15.75 17.17 25.17 23.42 17.75 20.67 29.00 PCR dry 19.25 13.00 17.25 24.58 22.92 16.67 21.67 29.17 PCR wet 19.00 12.90 14.80 23.80 22.58 15.42 16.92 13.25 DVR 14.83 13.58 15.75 24.67 16.08 16.5 18.5 24.75 59 Figure 3.7. Number of reconstructions over 50 years for the low and high traffic structures in all four climate zones In Figure 3.7, mixtures with a lower number of reconstructions over the pavement service life of 50 years are expected to perform better than mixtures with a higher number. In Michigan, the PCR dry, PCR wet, and DVR performed better than the Control mix for both structures, and both mixtures with PCR had similar behavior as the SBS. In Idaho, with a dry freeze climate, the rubberized mixtures (PCR dry, PCR wet, and DVR) had a number of reconstructions greater than those in Michigan. The PCR dry and PCR wet performed similarly to the Control mix for the low traffic structure. PCR dry and DVR mixes had a mechanical behavior similar to the SBS mix for the high traffic structure. In Florida, with a wet no freeze climate, the number of reconstructions for SBS, PCR dry and wet were higher than in Michigan for the low traffic structure and for the high traffic level, also for the Control mix the reconstructions were higher than in Michigan. The PCR dry had similar mechanical behavior of the SBS mix in low and high traffic conditions. In California, with a dry no-freeze climate, the mixtures modified with PCR and DVR performed better than in other climate conditions, especially for the low traffic structure. PCR dry, PCR wet, and DVR had a number of reconstructions close to the SBS mix in low traffic conditions. However, these results are affected by the assumptions made and, therefore, present some limitations. Materials used in the mixtures such as aggregates and bitumen in other States would 60 have different characteristics from those used in Michigan. This means that the mix design (the aggregates gradation and the percentage and type of bitumen) would be different elsewhere, and, consequently, the material would respond differently to the mechanical distress. Moreover, in this study, the same asphalt mixture was used in all layers on the top of the unbound materials. In the field, the DOTs sometimes use different materials for different layers to better deal with the distress that most affects the pavement based on the location, climate, and traffic. For example, The PCR wet mix in California failed for rutting for a high traffic structure before the other mixtures did, but this does not mean that the PCR wet mixture is not valuable in such a condition. This means that in that particular case, an accurate design of the pavement is needed, choosing the right combination of asphalt mixture per each layer. 3.2 Life cycle assessment Life cycle assessment (LCA) is a comprehensive yardstick of the environmental performance of goods and services [112]. In this study, an LCA was performed to quantify and compare the environmental impacts of rubberized and reference asphalt mixtures designed for surface layers of two structures, in accordance with the ISO 14040-44 [68, 69]. 3.2.1 Goal and scope The objective of the LCA was to compare global warming potential (GWP), fossil depletion (FD), and cumulative energy demand (CED) of pavement structures built with the asphalt mixes modified with CR from scrap tires (PCR dry, PCR wet, and DVR) compared to the two reference mixtures (unmodified Control and SBS mixtures). The impact categories were selected to assess the carbon footprint, the direct and indirect energy, and the impact of the fossil fuels used throughout the entire life cycle of road pavements. An attributional approach is used to 61 carry out the LCA. The attributional method estimates the total emissions associated with processes and material flows directly used in the life cycle of a product within a specific temporal window. It does not consider possible changes in the system production and indirect effects arising from these changes. In the attributional approach, the inventory is based on average data that represents the actual physical flows [113, 114]. The context of this study can be classified as situation A based on the ILCD Handbook classification [67]. Situation A is defined as micro-level decision support, used in attributional LCA for an individual product not having a significant influence on the background system and depicting the existing supply chain. The analysis was conducted for the United States, and all data regarding electricity production, chemicals, and transportation were adjusted to represent the average conditions. Material transportation was included based on the commodity flow survey for the United States [115]. 3.2.2 System boundaries and functional unit In this work, the boundaries of the cradle-to-grave system included material acquisition (e.g., aggregates, bitumen, CR, SBS, etc.), material production (asphalt mixtures), construction, use phase, and the end of life of the surface layers (Figure 3.8). The use phase included the reconstructions of the surface layers based on the response of the pavement structure to the distresses under two different levels of traffic and in 4 climate conditions. The use phase did not include the emissions coming from vehicles traveling on the roads over the service life. This type of emissions varies based on surface roughness, which is different for various asphalt mixtures and changes over time. These factors certainly affect the carbon footprint of the use phase of a road pavement, but were not considered in this research. Other factors such as noise or albedo phenomenon were also excluded from the system boundaries. 62 The functional unit was a 1-mile single lane pavement structure with a fixed thickness over the service life of the road pavement (50 years). The low traffic structure, designed for 3 million ESALs, had two asphalt layers with a total thickness of 3.5 inches (1.5 and 2 inches). The high traffic structure, designed for 30 million ESALs, had three asphalt layers with a total thickness of 6.5 inches (1.5, 2 and 3 inches). The reference flow, which is the measure of the output required to fulfill the functional unit, was the amount, in metric tons, of the different asphalt mixtures used to build 1 mile of pavement. The mixtures had different weights due to small fluctuations in their specific gravity. Table 3.8 reports the quantity of each asphalt mixture in the two structures. Table 3.8 Reference flow (ton/1-mile single lane) Structure Control SBS PCR dry PCR wet DVR Low traffic (3M ESALs) 1211.7 1214.9 1185.8 1228.6 1214.2 High traffic (30M ESALs) 2246.4 2252.3 2198.3 2277.7 2251.0 Figure 3.8. Boundaries of the cradle-to-grave system for 1-mile single lane road pavement In this study, the cut-off method was applied to evaluate the impact of CR from scrap tires and a sensitivity analysis was performed by using economic allocation and system expansion. Cut- off method and economic allocation were applied to the reclaimed asphalt pavement (RAP) and 63 the asphalt binder production, respectively. The TRACI 2.1 method [116] was used to evaluate GWP and FD impacts and the Cumulative Energy Demand impact category for the energy demand [71]. SimaPro 9.1 [117] software was used for the assessment. 3.2.3 Life cycle inventory Development of the Life Cycle Inventory is the stage where data are collected from various sources. The material inventory for the mixtures was determined from the mix design described in Chapter 2. Data on the energy spent in quarry activities to produce aggregates, fuels and energy used for crude oil production and its distillation for asphalt binder and scrap tires grinding for CR production were collected from the available literature. The PCR and the DVR manufacturers provided a description of the production processes of the two enhanced CR technologies. Existing databases such as Ecoinvent 3.6 [118], USLCI [119], and US-EI 2.2 [120] were used. 3.2.3.1 Aggregates and RAP Aggregates from natural rocks define the structure of asphalt mixtures and are classified based on their size as gravel, sand and fines (silt+clay). The quarry activities needed to produce natural aggregates were modeled using data from the Portland Concrete Association [121] as recommended in the Product Category Rules for asphalt mixtures [122]. Since RAP is recycled countless times, the burden associated with the virgin materials becomes smaller over time and is excluded from the system boundaries. The system boundaries for the RAP considered the milling operations of old pavements and the avoided burdens for waste disposal in landfills as well as environmental credits for avoiding the use of primary resources in new asphalt mixes. The fuel consumption of a milling machine producing 350 tons/hour in the recycling phase was assumed to be 56,397 Btu per ton of RAP produced [123]. A distance of 50 km was assumed for transporting 64 aggregates and RAP to the asphalt plant by truck. Data on electricity and fuel used in quarries to produce aggregates and on the milling operations for RAP sourcing are reported in Appendix B. 3.2.3.2 Asphalt binder The asphalt binder is used in asphalt mixtures for its adhesive and viscoelastic properties. It is a petroleum product derived from the portioned distillation of crude oil. In this study, the system boundaries for the asphalt binder production included crude oil extraction, flaring, domestic and foreign crude oil transportation, refinery, and storage. Extraction operations in foreign countries were not included. Besides asphalt binder, crude oil has many valuable co- products such as kerosene, fuel oil, and petroleum coke. Therefore, the economic allocation was used to assess the impact of crude oil refining for the asphalt, according to the example provided in the ISO/TR 14049 [124] and also reported in the Eurobitume report [125]. Detailed data on the origin of the foreign crude oil, processes in SimaPro 9.1 for imported crude oil, distance traveled, transoceanic tankers and pipelines, domestic crude oil transportation by pipeline, barge tanker, and train and materials and energy used to refine 1 ton of crude oil are reported in Appendix B. The allocation coefficient, mass residue yield, and allocation factor were calculated based on Eq. 3.1, Eq. 3.2, and Eq. 3.3. The allocation coefficient was calculated using the ratio between the price of the asphalt binder multiplied by its mass over the summation of all co-product’s prices multiplied by their mass, as in Eq. 3.1. The mass residue yield was obtained by dividing the mass of the asphalt binder produced by the total mass of all co-products, using Eq. 3.2. Finally, the allocation factor in Eq. 3.3 corresponds to the ratio between the allocation coefficient and the mass residue yield [126]. 65 $23!4567 ∙ E?FF23!4567 Eq. 3.1 ;<<=>?#@=A >=B66@>@BA# = ∑ $+99 :;<=;>4:!/ ∙ E?FF+99 :;<=;>4:!/ E?FF23!4567 Eq. 3.2 E?FF HBF@,IB J@B<, = ∑ E?FF+99 :;<=;>4:!/ ;<<=>?#@=A >=B66@BA# Eq. 3.3 ;<<=>?#@=A 6?>#=H = E?FF HBF@,IB J@B<, Based on the mass [127] and the price [128] of all co-products produced from domestic and imported crude oil in 2016, the allocation coefficient was 0.0132, while the mass residue yield was 0.0215. Therefore, 1.32% of the total economic output of the refinery and 2.15% of total mass output was asphalt binder. The allocation factor was calculated to be 0.615 and multiplied by materials input and energy used to refinery the crude oil. Additional information is available in Appendix B. Neat asphalt binder was used in Control and PCR dry mixtures. The asphalt binder in the SBS, PCR wet and DVR mixtures was modified by adding the associated agent modifier based on the mix. The technique used to modify the asphalt binder was the same for all different mixes in terms of equipment used, time and temperatures (Appendix B). The electricity consumption for asphalt binder modification was set to 72 MJ per ton of total asphalt binder produced [125]. 3.2.3.3 Crumb rubber In 2019, in the United States, the primary use of scrap tires was for fuel in cement kilns (41%), followed by CR production (27%), landfill disposal (17%), and other applications (16%), which includes exports and tire-derived aggregates. At the end of their life, tires are collected and transported to the CR facility where components such as steel, textile and rubber are separated. An average distance of 300 km was assumed for the transportation of scrap tires to the CR facility, 66 based on the commodity flow survey for the United States [115]. Steel is recycled and reduces the production of primary steel. Textile fibers are used as fuel in kilns, preventing the production and usage of petroleum coke [9]. Cut-off, economic allocation, and system expansion are used to model CR production. The distance from the CR to the PCR facility was assumed to be 600 km. Table B22 and Table B23 provide additional details on electricity and material input used in the CR production. 3.2.3.4 Polymer coated and devulcanized rubber The PCR is made by mixing crumb rubber from scrap tires with a styrene-butadiene-rubber (SBR) polymer emulsion. The emulsion creates a film that partially covers the surface area of the rubber particles [28]. In this study, the PCR composition was 5% of SBR and 95% of CR. The mixing process of these two materials is similar to the mixing for the asphalt binder modification and therefore the same amount of electricity (72 MJ/ton PCR) was assumed to be used. The PCR was assumed to be transported for 1,000 km by truck from the facility to the asphalt mixture plant. The DVR is a pellet obtained after applying a chemical and mechanical process to a regular 0.595 mm CR. The chemical process uses compatibilizers, plasticizers, and reagents, which are needed to break the sulfur-to-sulfur bonds of the rubber. The mechanical process is used to create pellets. According to the manufacturer (Full Circle Technologies, LLC, Cleveland Ohio) and based on the chemicals available in the databases, Table B24 shows the unit processes used to model the DVR. The transportation distance was assumed to be 500 km. 3.2.3.5 Synthetic polymers The information available to model synthetic polymers such as SBS, SBR, and Sasobit are scarce in literature or in the databases for LCA software. The SBS was modeled as reported in 67 Table B19 [129]. The process used for the SBR is reported in Table B20 [130], and Sasobit as in Table B21 [131]. 3.2.3.6 Hot mix asphalts Table 3.9 reports the quantity of each material per mixture, based on the mix design described in section 2.4.1. The energy spent to produce 1 ton of asphalt mixture was 317 MJ based on a previous LCA study [132]. Details on the electricity and fuels used in asphalt plants such as diesel, gasoline and natural gas, are reported in Appendix B. Table 3.9 Material inventory to produce 1 metric ton of hot mix asphalt for each alternative Control SBS PCR dry PCR wet DVR (kg) Virgin aggregates 804.0 800.4 797.6 800.5 754.5 Aggregates from RAP 142.1 141.2 140.7 141.2 188.5 Asphalt binder 47.00 48.50 48.90 44.99 42.18 Asphalt binder from RAP 7.87 7.83 7.80 7.82 10.45 SBS - 1.770 - 0.250 0.93 Sasobit - - - 1.510 - PCR - - 5.00 3.53 - DVR - - - - 3.26 Sulfur cross-linker - 0.20 - 0.20 0.19 3.2.4 Results 3.2.4.1 Life cycle assessment of crumb rubber Figure 3.9 illustrates the system boundaries considered to produce crumb rubber from scrap tires using three different allocation methods. Figure 3.9a shows the processes included in the study for the cut-off method. The impacts for recycling are attributed to the current system, and 68 the burdens begin with the transportation of the scrap tires between the facility where tires are collected and that where tires are recycled. Recycling steel reduces primary production, and burning the textile replaces another type of fuels such as petroleum coke. Economic allocation (Figure 3.9b) was preferred to mass allocation because scrap tires are a valuable waste. Table 3.10 summarizes the average price for each scrap tires’ application [133] and the associated allocation factors. A coefficient of 0.88 means that 88% of the environmental benefits and impact for transporting and recycling tires and steel and textile recycling were assigned to the CR production. It was assumed that the same recycling process is used for producing CR, for recovery energy, and disposing of tires in landfills since most of the States ban whole tires disposal [134]. Figure 3.9c shows the boundaries of the system expansion. The system producing the CR was extended to the other applications using scrap tires and to the rubber’s co-products. Scrap tires were considered as recovered energy in cement kilns and landfills disposal. The other applications, such as tire-derived aggregates and exportation were not included in the expansion. There is an avoided burden due to the elimination of incineration and landfill disposal. The equivalent amount of petroleum coke that substitutes the scrap tires for recovery energy was added to the system. 69 Figure 3.9. System boundaries considered for the crumb rubber production using a) cut-off method, b) economic allocation, and c) system expansion. Table 3.10 Price and economic allocation coefficient for scrap tires Price[1] Allocation coefficient $/ton Tire-derived fuel 40 0.09 Crumb rubber 400 0.88 Other 15 0.03 [1] BCA Industries, 2018 The lowest carbon footprint was calculated to be -54.4 kg CO2eq/ton for the production of 1 ton of CR using the cut-off method (Figure 3.10). The result is negative because the benefits of recycling steel and textile are higher than the impact due the scrap tires recycling. Similar results were observed using the economic allocation (-35.1 kg CO2eq/ton). 70 Figure 3.10. Contribution to Global Warming Potential of each process associated with the production of 1 ton of CR, using cut-off, economic allocation, system expansion, and system expansion considering the natural gas substitution. For system expansion, the environmental benefits due to the avoided incineration of scrap tires (green bar in Figure 3.10, is -2,236 kg CO2eq/ton CR produced). It represents 80% of the environmental impacts of burning petroleum coke in cement kilns (orange bar in Figure 3.10, equal to 2,795 kg CO2eq/ton CR produced). Figure 3.10 also includes an alternative scenario (System exp – NG) where the scrap tires substitute the natural gas instead of the petroleum coke. The benefit of recycling scrap tires rather than burning them is clearer if natural gas is used instead of petroleum coke in cement kilns. The CO2 emissions from natural gas burning (orange line pattern bar in Figure 3.10, equal to 1,479 kg CO2eq/ton CR produced) are 1.5 lower than scrap tire incineration. The reduction in carbon footprint from using natural gas instead of coke illustrates 71 the potential from decarbonization of cement manufacturing. At present, the emissions due to cement production can only be reduced by increasing the use of waste derived fuels and biomass, but in the future alternative fuels will be used such as synthetic fuels or the electrification of heat generation [135]. The use of natural gas requires specific adjustments to the plant in order to avoid dissipation of heat, as happened to the plant already switched to natural gas in North America [136]. The future decarbonization of cement manufacturing using alternative fuels will affect the scrap tires market. More scrap tires will be available for recycling materials with the consequent decrease of the CR price. Lower CR market prices will result in higher environmental credits associated with the use of scrap tires in second life. Consequently, the economic allocation coefficient will be lower than the actual factor (0.88), with a consequent decrease of the environmental impact associated with the CR production but also a consequent decrease of the benefits that can be considered, as shown in Figure 3.10 for the economic allocation. 3.2.4.2 Impact of allocation methods on the asphalt mixtures The different allocation methods applied to the CR production influenced the environmental impact of the asphalt mixtures. Table 3.11 reports the results in terms of GWP and FD of 1 ton of each mixture. In parenthesis, I indicated the difference in percentage between the Control and rubberized mixtures compared to the SBS mix. Figure 3.11 depicts in percentage the data reported in Table 3.11 to clearly illustrate the comparison between the mixtures containing CR and the reference mixtures based on the allocation methods. Table 3.11 summarizes the GWP and FD for all materials computed using all four methods. As shown in Table 3.11, the cut-off and economic allocation methods revealed that three rubberized mixtures had lower GWP and FD than the SBS blend (65.7 kg CO2eq/ton, 289.6 MJ surplus/ton). In the case of the system expansion, the PCR dry mix had a FD 3.5% higher than the 72 SBS mix. The GWP and FD obtained by considering the natural gas instead of the petroleum coke in cement kilns (System exp-NG) were lower than those from the system expansion considering the petroleum coke only (Figure 3.11). The GWP of the system exp-NG was also lower than the results obtained using the cut-off method. Instead, fossil depletion results were higher than the cut- off method due to the use of fossil fuel (petroleum coke and natural gas) in both allocations (system expansion and system exp-NG). If in the future the cement manufacturing were to be decarbonized, also the impact of the fossil depletion of asphalt mixtures containing CR would be reduced, the results may be lower than those obtained by using the cut-off method. Results of the CED (Cumulative Energy Demand) were similar to the FD and are reported in Appendix B (Table B26 and Figure B2). Table 3.11 Global warming potential (GWP) and fossil depletion (FD) of 1 ton of Control and rubberized asphalt mixtures compared to the SBS blend Cut-off Economic allocation System expansion System exp-NG GWP kg CO2 eq/ton Control 57.3 (-12.9%) SBS 65.7 PCR dry 60.1 (-8.5%) 60.2 (-8.4%) 62.8 (-4.5%) 56.5 (-14.0%) PCR wet 63.5 (-3.3%) 63.6 (-3.2%) 65.4 (-0.5%) 61.0 (-7.2%) DVR 56.1 (-14.6%) 56.2 (-14.5%) 57.9 (-12.0%) 53.8 (-18.1%) FD MJ surplus/ton Control 257.9 (-10.9%) SBS 289.6 PCR dry 270.2 (-6.7%) 270.7 (-6.5%) 299.8 (+3.5%) 287.9 (-0.6%) PCR wet 256.2 (-11.5%) 256.6 (-11.4%) 277.2 (-4.3%) 268.8 (-7.2%) DVR 218.2 (-24.7%) 218.5 (-24.5%) 237.4 (-18%) 2297 (-20.7%) 73 Figure 3.11. a) Global warming potential and b) fossil depletion comparison of PCR dry, PCR wet, and DVR mixtures to Control and SBS mixes based on the allocation methods In LCA studies of products containing recycled materials, the cut-off method is often used because it tends to favor the system using recycled material, by excluding impacts associated with virgin materials and co-functions. Moreover, it is easy to apply and to explain to a broad audience, and it requires only input data related to the product under investigation. On the other hand, it oversimplifies the environmental impact of the cradle and the grave stages [137]. Results obtained by using the cut-off method remain constant over time unless there is a change in the scrap tire recycling process. The cut-off method highlights the environmental benefits of using scrap tires in asphalt mixtures, and it is not influenced by the changes in the scrap tire market. Not everybody agrees with using market data in LCA studies (Pelletier and Tyedmers, 2011), because prices for recycled materials are often not stable and influenced by governmental policies [139]. The price fluctuation can lead to significant uncertainties. However, companies use economic estimations for the decision to recycle, and these economic values should be used for the environmental analyses as well. The market characteristics of a material are part of an adequate life cycle model by adding realism to the analysis [140, 141]. Even if, at present, results of the economic allocation were close to those of the cut-off method, results can be different in the future, influenced by the market price and the other scrap tires applications (e.g., tire-derived fuel). 74 From a life cycle thinking perspective, an environmental assessment of recycled materials should give importance to other aspects as well, such as benefits and impacts due to recycling, avoiding new resources, and waste management [137, 142–144]. The system expansion is helpful to accomplish this need. The system producing the recycled material receives credit for avoided production, with the rationale that the first system supplies material as a feedstock for future systems [145]. The system expansion is the most comprehensive method that analyzes the impact of a broad system. However, results were influenced by the assumptions that are made to extend the boundaries of the system. The system expansion did not always highlight the advantages of using PCR in asphalt mixtures as alternative material to the synthetic polymer. The cut-off method is recommended herein to compare the asphalt mixtures modified with CR to the reference mixes because this method is constant over time and not influenced by external factors, such as the price of the CR or the fuels used for the cement production. The same rationale was found in literature, in a work analyzing alternative cementitious materials to replace Portland cement using different allocation methods versus a no allocation approach [146]. 3.2.4.3 Uncertainty The uncertainty of the life cycle studies can be due to model parameters (e.g., measurements, temporal and spatial variability, and data representativeness) and modeling procedures (e.g., selection of functional unit and allocation methods). In this research, the Monte Carlo analysis was applied to evaluate the effect of data uncertainty for different allocation methods on the production phase of 1 ton of each asphalt mixture. Figure 3.12 shows the 2.5, 50 (median), and 97.5 percentile for the global warming potential and fossil depletion of the asphalt mixtures considered in this research. Results indicated that the cut-off method is characterized by low uncertainties, while the system expansion is the approach with the highest variability, 75 confirming the decision of selecting the cut-off method for comparative purposes. All the statistical data are reported in Appendix B, Table B27. Figure 3.12. Uncertainty from Monte Carlo analysis for 1 ton of asphalt mixture using different allocation methods for a) global warming potential and b) fossil depletion 3.2.4.4 Material and energy contribution For all asphalt mixtures, the asphalt binder had the largest contribution in terms of GWP and FD. Figure 3.13 shows the contribution of the virgin asphalt binder (production only, excluding modification and transportation) added in each mixture based on the quantity used in the blends to GWP and FD. The asphalt binder was, on average of the five mixtures, 41.2% of the total GWP and 75.5% of the total FD. Considering the asphalt modifications and the transportation processes, those percentages were higher (58.2% of the total GWP and 96.8% of the total FD). 76 Figure 3.13. a) Global warming potential (kg CO2eq) and b) fossil depletion (MJ Surplus) associated with the virgin bitumen content in asphalt mixtures The second large contribution to the GWP and FD was due to the energy consumption in the asphalt plants to produce the asphalt mixtures. Impacts associated with the asphalt plant operations were 32.1% of the total CO2 emissions, and 13.4% of the total fossil depletion, on average of the five mixtures. The asphalt plant operations require the use of electricity and fuels. The contribution due to the fuels was 92% of the CO2 emissions and 94% of the fossil depletion of the total asphalt operation. The contribution of the electricity was very low in comparison, and alternative fuels can be used to lower the environmental impact of the asphalt mixtures. The overall contribution of the electricity in all processes involved in the asphalt mixture production (e.g., aggregates, CR, synthetic polymer, asphalt binder modification, asphalt mix production) was 7.67% of the total GWP and 2.27% of the total FD. The electricity mix used in this work was calculated based on the EIA energy outlook [147, 148], as average for the USA (Table B10). The environmental impact of the road pavements in four different states (Michigan, Idaho, Florida, and California) were evaluated using the average electricity mix for the USA. The environmental impact attributed to the electricity by using the electricity mix specific of each State would increase or decrease based on values reported in Table 3.12. The GWP, FD, and CED per kWh were computed based on the electricity mixes of each State and the difference in percentage between each State and the USA average [149]. 77 Table 3.12 Environmental impact of electricity mix per kWh Global warming Cumulative energy Fossil depletion potential demand kg CO2eq/kWh MJ surplus/kWh MJ/kWh Electricity mix, USA 0.532 0.703 11.38 Electricity mix, Michigan 0.639 (+20%) 0.572 (-19%) 13.23 (+16%) Electricity mix, Idaho 0.166 (-69%) 0.371 (-47%) 5.635 (-50%) Electricity mix, Florida 0.672 (+26%) 1.259 (+79%) 12.50 (+10%) Electricity mix, California 0.296 (-44%) 0.658 (-6%) 8.293 (-27%) 3.2.4.5 Construction phase The fuel consumption needed for building the two pavement structures was calculated based on the hourly fuel consumption and working speed of the milling machine, paver, and compactor, as reported in Table 3.13 [150–152]. Table 3.14 reports the calculations of the total liters per machine per structure. Table 3.13 Construction equipment specifications Liter/hour Meter/hour Milling machine [150] 76 3,000 Paver [151] 13 3,000 Compactor [152] 10 5,000 Table 3.14 Fuel consumption to build the low and high traffic structures hour/1-mile Liter/1-mile Low traffic High traffic [1] single lane single lane structure structure [2] Liters Milling machine 0.536 40.76 40.8 40.8 Paver 0.536 6.97 13.9 20.9 Compactor 0.322 3.22 38.6 57.9 [1] 2 asphalt layers [2] 3 asphalt layers 78 3.2.4.6 Life cycle assessment of road pavements After having discussed the results of the asphalt mixtures per ton of material produced (cradle-to-gate analysis), this section provides an explanation of the importance of performing a cradle-to-grave LCA of road pavement by showing how results change including the mechanical performance of the asphalt mixtures in the analysis. Figure 3.14 and Figure 3.15 show the total GWP, FD, and CED for all five mixtures in four climate zones (Michigan, Idaho, Florida, and California), respectively, for the low and high traffic structures. The results presented in this section were calculated by using the cut-off method for the CR production and the USA average electricity mix for all processes involved. Figure 3.14. Global warming potential (GWP), fossil depletion (FD), and cumulative energy demand (CED) from cradle-to-grave LCA of a low traffic pavement using reference and rubberized mixtures in a) Michigan, b) Idaho, c) Florida, and d) California 79 Figure 3.15. Global warming potential (GWP), fossil depletion (FD), and cumulative energy demand (CED) from cradle-to-grave LCA of a high traffic pavement using reference and rubberized mixtures in a) Michigan, b) Idaho, c) Florida, and d) California The results of the cradle-to-gate analysis presented in the previous section revealed that the Control and SBS were the mixtures with the lowest and highest impacts, respectively. However, the results of the complete LCA were different than the cradle-to-gate analysis. The Control mix had the highest GWP, FD, and CED in Michigan and California for the low traffic road pavement (Figure 3.14a and d) and in all climate zones for the high traffic structure (Figure 3.15). In Idaho, having a dry freeze climate, and in Florida, having a wet no-freeze climate, the PCR wet mix was the mixture with the highest carbon footprint in the low traffic pavement (Figure 3.14b and c). For the high traffic level, the PCR wet mix had the highest impacts among the modified mixtures (SBS, PCR dry, DVR) in Idaho, Florida, and California (Figure 3.15 b, c, d). Since the greater contribution to the total impacts was attributed to the use phase, as shown in Figure 3.16, the results of the cradle-to-grave LCA are driven by the distress response predicted using the mechanistic- empirical pavement analysis. 80 Overall, in Michigan, the mixtures containing CR (PCR dry, PCR wet, DVR) performed similar to the SBS mix, especially the PCR dry mixture, having the lowest environmental impact compared to the PCR wet and DVR. Regarding the asphalt mixtures with high environmental impact (such as the PCR wet in California in the high traffic level), it does not necessarily mean that the mixtures modified with the PCR wet technology is not good in California or in places with similar climate. It just means that the same mix design performed for Michigan, using the materials available in Michigan (aggregates and asphalt binder), may not be appropriate for other places. For instance, in California, aggregates would have a different gradation, and consequently, the asphalt mixture would have a different percentage of asphalt binder to satisfy the volumetric requirements of the Superpave mix design (section 2.4.1). The response of the asphalt mixture to the pavement failure would be different and which would affect the environmental impact. This strengthens the importance of having site-specific data regarding the asphalt material to better calibrate the distress prediction model and to predict the design life more accurately. Figure 3.16 illustrates that the major contribution of the total GWP and FD was due to the use phase. Figure 3.16 also shows the difference in terms of environmental impact between low traffic and high traffic pavement structures, having two different thicknesses. Figure 3.16. Global warming potential in kg of CO2eq per 1-mile single lane over 50 years: contribution of each LCA phase for a) low traffic and b) high traffic pavement in Michigan 81 3.2.5 Material usage over the service life of the pavements Based on the tons of asphalt mixtures used for each lane of the two different structures in Michigan and the number of reconstructions calculated in section 0, the total material usage for each mixture over 50 years was calculated. Then the percentage difference of materials saved/gained with respect to the Control and the SBS mixtures were calculated and summarized in Table 3.15. Table 3.15 Difference in tons of asphalt mixtures used in Michigan to build the surface layers of low and high traffic structures over 50 years compared to the Control and SBS mixtures Low traffic High traffic Compared to Control SBS -30.7% -58.0% PCR dry -30.5% -58.1% PCR wet -27.1% -55.9% DVR -7.69% -38.8% Compared to SBS PCR dry +0.30% -0.11% PCR wet +5.19% +5.13% DVR +33.2% +45.9% Using asphalt mixtures modified with SBS or PCR (dry technology) in a low and high traffic level structure saved up to approximately 31% and 58% of materials, respectively, compared to the unmodified Control mixtures. Similar results were reported for the PCR wet mixtures compared to the Control mix. Among the modified mixtures, the PCR dry was the blend saved the same quantity of materials over 50 years as the SBS mixture in Michigan. PCR dry mixture can reduce the use of virgin materials up to 2.3 times compared to the Control mix for a high traffic structure in Michigan over 50 years. PCR dry uses 271.7 ton/1-mile single lane of asphalt binder and approximately 3,830 ton/1-mile single lane of natural aggregates over the service life. By comparison, Control mix uses 626.6 ton/1-mile single lane of asphalt binder and approximately 9,140 ton/1-mile single lane of natural aggregates in 50 years. 82 However, using the PCR wet mixture, we would need around 5.1% more material than placing the SBS mixture for both low and high traffic road pavements. Results for the DVR mixture were higher than those reported for PCR (both technologies). The mixture modified with DVR was a mix designed slightly differently from the other mixtures. It was designed for 3 million ESALs instead of 30 million, with different aggregates (per type and gradation), and a percentage of RAP higher than that used in the others (20% instead of 15% by weight of the total mix). For this reason, the DVR mix performed differently in terms of long-term pavement design. 3.3 Summary of chapter findings The objective of the part of this study presented in this chapter was to investigate the long- term response of the pavement structures (low and high traffic level) built in all climate zones (Michigan – wet freeze, Idaho – dry freeze, Florida – wet no freeze, California - dry no freeze) with five different asphalt mixtures (Control, SBS, PCR dry, PCR wet, DVR) by using the mechanist-empirical pavement design approach. Results of this part of the work (thicknesses of the asphalt layers and number of reconstructions over 50 years, based on the response of the structures to the failures) were used as input in the life cycle assessment. The major findings have been summarized as follows: • In Michigan, with a wet freeze climate, the PCR dry, PCR wet, and DVR had a number of reconstructions lower than the Control mix for low and high traffic structures. Mixtures containing PCR had a similar long-term response to the SBS mix for both structures. • In Idaho, with a dry freeze climate, the rubberized mixtures (PCR dry, PCR wet, and DVR) had a number of reconstructions greater than those in Michigan. The PCR dry and PCR wet performed similarly to the Control mix for the low traffic structure. PCR dry and DVR mixes had a mechanical behavior similar to the SBS mix for the high traffic structure. 83 • In Florida, with a wet no freeze climate, the number of reconstructions for SBS, PCR dry and wet were higher than in Michigan for the low traffic structure and for the high traffic level, also for the Control mix the reconstructions were higher than in Michigan. The PCR dry had similar mechanical behavior to the SBS mix in low and high traffic conditions. • In California, with a dry no-freeze climate, the mixtures modified with PCR and DVR performed better than in other climate conditions, especially for the low traffic structure. PCR dry, PCR wet, and DVR had a number of reconstructions close to the SBS mix in low traffic conditions. • Results of the mechanist-empirical pavement design were calculated with 95% of reliability, as also adopted by the Michigan Department of Transportation, meaning that there is a 95% of chance that distresses will not exceed the limit earlier than predicted. • Some assumptions affected the results: aggregates and bitumen in other States different from Michigan would have different characteristics. Therefore, the mix design would be different elsewhere, and, consequently, the material would respond differently to the mechanical distress. Moreover, in this study, the same asphalt mixture was used in all layers on the top of the unbound materials. In the field, the DOTs assume different materials per layer to better deal with the distress that most affects the pavement based on the location, climate and traffic. The objective of the life cycle assessment was to investigate the environmental impact of the road pavements comparing the rubberized mixtures to the reference mixtures. The major findings have been summarized as follows: • From the cradle-to-gate analysis, Control and SBS mixtures were the mixtures with lower and higher environmental impacts, respectively, compared to the PCR dry and PCR wet 84 mixtures. The DVR mix had GWP, FD, and CED even lower than Control mixture per ton of material produced. • Results of the cradle-to-grave analysis were different. The Control mix had the highest GWP, FD, and CED in Michigan and California for the low traffic road pavement and in all climate zones for the high traffic structure. In Idaho and Florida, the PCR wet mix was the mixture with the highest carbon footprint in the low traffic pavement. For the high traffic level, the PCR wet mix had the highest impacts among the modified mixtures (SBS, PCR dry, DVR) in Idaho, Florida, and California. • Since the greater contribution to the total impacts can be attributed to the use phase, the results of the cradle-to grave LCA are driven by the distress response predicted using the mechanist-empirical pavement design. • Overall, in Michigan, the mixtures containing CR (PCR dry, PCR wet, DVR) performed as good as and better than the SBS mix, especially the PCR dry mixture, having the lowest environmental impact compared to the PCR wet and DVR. • Using asphalt mixtures modified with SBS or PCR (dry technology) in a low and high traffic level structure allows saving, respectively, up to approximately 31% and 58% of materials compared to the unmodified Control mixtures. Among the modified mixtures, the PCR dry was the blend allowing to save the same quantity of materials over 50 years as the SBS mixture in Michigan. 85 4. LEACHING ASSESSMENT: METALS CHARACTERIZATION The part of the research presented in this chapter aims to investigate the potential risk of toxicity due to the metals leaching from the materials in the asphalt mixtures used in road pavements, such as crumb rubber, asphalt binder, and reclaimed asphalt pavement, deemed to be the main responsible of the metal leaching. 4.1 Introduction According to Epps [13], the toxicological aspect represents a complex issue due to the characteristics of the materials in use. The chemical composition of the asphalt binder is not precisely known because the bitumen is obtained by refining a wide variety of crude oils, meaning that not all asphalt binders have the same composition. The CR as well is produced by a variety of scrap tires and chemicals not well-defined. Moreover, the CR used in this work is further processed to produce the PCR and the DVR. Asphalt binders include heavy metals and polycyclic aromatic hydrocarbons (PAHs), pollutants classified as carcinogenic, mutagenic, and teratogenic [29, 30]. RAP contains aged asphalt binder and chemicals released on the road surface during the use phase, like lubricating oils, gasoline, metals from brake pads, and vehicle exhaust [29]. CR contains metals (e.g., iron, zinc, chrome), PAHs, and volatile organic compounds (VOCs) [8]. In particular, the high zinc content in CR samples concerned many research studies in literature, as reported in section 1.4.3. There is a potential risk that the stormwater runoff may transfer the substances mentioned above contained in the materials used in road pavements to the groundwater. Leaching is the transfer of chemicals from a solid material into contacting water. The contacting water may result from rainwater infiltration through overlying soils or direct material contact with groundwater or 86 surface water. Constituents that leach into the water can contaminate adjacent soils or disperse into groundwater or surface water bodies. Leaching occurs when constituents contained in materials are released into the environment through the contacting water. The leaching process is driven by principles of mass transport, which defines the movement of constituents from a solid phase to contacting water [38]. This part of the research aimed to assess the potential metal leaching of rubberized asphalt mixtures compared to an unmodified mixture and another blend modified with the synthetic polymer styrene-butadiene-styrene (SBS) as two reference mixes. The central question that this work wanted to address was if metals englobed in the asphalt mixes placed as surface layers in road pavements could be released into the groundwater due to the rainwater after construction and in what concentration. Method 1315, part of the LEAF framework, was used to conduct the leaching testing of the five asphalt mixtures [38]. Method 1315 measures the mass transfer rates of constituents in monolithic and compacted granular materials using a semi-dynamic tank leaching procedure. Results from this method indicate the rate of mass transport from the interior of the material to its external surface (i.e., the interface of the material with the surrounding environment). Method 1315 is applicable for cases where water flows primarily around the material and then percolates through it. However, actual liquid to surface area ratios is often much less than the test conditions in the field. Therefore, the rate of leaching in the field can be less than measured in the laboratory, and Method 1315 data often are used to estimate the parameters (e.g., observed diffusivity) that control mass transfer for each constituent. The data from Method 1315 may represent mass transport rates over short time durations with the mass transport parameters used to estimate leaching rates in the field over longer time durations. Today, all the four LEAF test methods have undergone interlaboratory validation for use with inorganic constituents of 87 potential concern (COPCs) only, such as metals [38]. For this reason, since Method 1315 is the method that is better suited to test asphalt materials, only the leaching of the metals was investigated. The EPA will adapt and extend LEAF methods to organic substances as well [38]. In this work, the trace of the metals in each material used in the asphalt mixtures was also investigated through the microwave digestion, in collaboration with the Polytechnic of Turin (DIATI - Department of Environment, Land and Infrastructure Engineering), to better understand what metals are included in the materials used in the mixtures considered in this dissertation and better interpret the leaching results. 4.2 Methodology 4.2.1 Sample preparation The same five asphalt mixtures included in Chapters 2 and 3 were considered in this part of the study: Control, SBS, PCR dry, PCR wet, DRV mixtures. After having prepared the loose mixes, based on the mix design reported in section 2.4.1, 15 cylindrical samples were compacted, three specimens for each mix, using a gyratory compactor. The cylinders had a diameter of 150 mm and a height of 60 mm. The air voids were 7% ±0.5% to replicate the field conditions and were checked by following the standard procedure for compacted specimens [153]. 4.2.2 Leaching testing procedure Method 1315, part of the LEAF framework, was followed while carrying out the experiments [38]. Two gallons buckets were used to guarantee the geometric requirements requested by the standard (at least 2 cm between the sample and all walls of the bucket, and a minimum of 5 cm of distance between the solid-liquid interface and the top of the bucket). Before starting the test, the buckets (never used for previous experiments) were pre-washed with a 88 detergent and rinsed with the reagent water. After measuring the weight of the 15 samples, plastic supports with knots were attached to easily remove the specimens from the buckets to replace the water (Figure 4.1). The plastic supports ensured that 98% of the surface was exposed to the water, as the standard requires. As indicated in the guide of method 1315, the deionized water (DI) was selected as a reagent solution. Each sample was immersed in 5.8 liters of DI (9 mL/cm2) for the first 2 hours (Figure 4.1). After this step, the samples were removed from the water, weighted, then immersed immediately in the fresh water in another set of 15 buckets for the next 23 hours. Within 15 minutes, the pH and conductivity of the DI water were measured by following the methods 9040 and 9050, respectively [154, 155]. Then a sample of eluate from each bucket was collected and filtered by using 45 μm membranes. Then, the solution for the chemical analysis was prepared by measuring 9.8 mL of eluate and adding 2% of nitric acid (Figure 4.2). All these steps were repeated for each interval of time required in the standard (after other 23 hours, 5, 7, 14, 14, 7, and 14 days) for a total of 63 days. Figure 4.1. a) Asphalt mix specimens, b) plastic support, and c) specimens placed in deionized water 89 Figure 4.2. a) Eluate sample took from the bucket was filtered through a membrane by using a pump to speed up the process, b) 45 μm filter membrane, c) sample preparation for chemical analysis: 2% nitric acid (HNO3) added to the filtered eluate (10 mL) 4.2.3 Chemical analyses The concentrations of the metals were assessed in the ICP-MS (Inductively coupled plasma mass spectrometry) laboratory in the Department of Earth and Environmental Sciences at Michigan State University. The Thermo iCAP-Q ICP-MS was used to assess the presence of the following metals: aluminum (Al), antimony (Sb), arsenic (As), barium (Ba), cadmium (Cd), calcium (Ca), chromium (Cr), cobalt (Co), copper (Cu), iron (Fe), lead (Pb), magnesium (Mg), manganese (Mn), mercury (Hg), nickel (Ni), potassium (K), sodium (Na), tin (Sn), titanium (Ti), vanadium (V), and zinc (Zn). To assess the purity of the reagent and equipment, the concentrations of metals in blank solutions (deionized water and 2% of nitric acid) were also tested. 4.2.4 Digestion and metals characterization Quantitative analytical techniques, such as inductively coupled plasma atomic emission spectrometry (ICP-AES), inductively coupled plasma mass spectrometry (ICP-MS), and Inductively coupled plasma optical emission spectrometry (ICP-OES), are used to determine trace metals contained in materials with sufficient sensitivity [156]. For these kinds of analyses, the 90 samples must be converted into a liquid form. For solid samples, this can be achieved by undertaking the digestion procedure. The digestion must efficiently decompose the sample matrix so that the analytes of interest are completely released and solubilized [156]. The technique of digestion involves placing the sample in vials with reagents and placing the vials in the microwave oven for irradiation by microwave energy [156]. The microwave digestion and the consequent quantification of the metals in the materials used in the asphalt mixtures were carried out in collaboration with the Polytechnic of Turin (DIATI - Department of Environment, Land and Infrastructure Engineering), at the Environmental Chemistry laboratory. The investigation on each material was done individually: aggregates, RAP, neat bitumen, PCR, DVR, SBS, and Sasobit. The procedures for the digestion, performed by using a Milestone mls 1200 mega oven, are described in Appendix C (Figure 4.3a). After this step, 20 mL of deionized water was added to the digested material, that was then filtered by using 542 Whatman filters (2.7 μm pore size) and finally diluted with 50 mL with deionized water (Figure 4.3b). These samples were analyzed with the ICP-OES to identify the metals trace (Figure 4.3c). Figure 4.3. a) Microwave digestor, b) Filtration of PCR and DVR digestated, and c) ICP-OES analysis 91 4.3 Results 4.3.1 Leaching test Results from Method 1315 were the eluate concentrations in μg/L (Figure 4.4), the interval flux rate in μg/m2-sec (Figure 4.5), and the cumulative mass release in μg/m2 (Figure 4.6). Figure 4.7 shows the pH and the conductivity (μS/cm) of the eluates measured at each specific time interval. For each interval, pH and conductivity values were similar for all mixtures tested. Among all metals considered, certain metals were the focus of this study. These include toxic (Pb) or harmful for the human body under prolonged exposure (As, Cd, Cr, Cu, and Zn). Mercury was not considered because its quantification was unreliable. Figure 4.4f shows that zinc had higher concentrations than the other metals and that it was present in the eluates of all mixtures, rubberized (PCR dry, PCR wet, and DVR) and reference blends (Control and SBS), over the duration of the test (63 days). Table 4.1 reports the total metal concentrations after 63 days, on average of the three replicates for each mixture. Table 4.1 Total metal concentration after 63 days As Cd Cr Cu Pb Zn μg/L Control 0.528 1.704 10.50 82.75 6.940 501.9 SBS 0.358 312.1 10.20 40.57 7.262 559.2 PCR dry 0.421 1.211 7.903 20.77 4.306 668.6 PCR wet 0.354 31.15 7.493 23.80 18.87 553.2 DVR 0.610 8.235 10.84 31.04 7.167 850.4 After seven days, SBS and DVR mixtures showed a peak for cadmium. Instead, for the PCR wet the peak for Cd was after 14 days (Figure 4.4b). In the case of lead (Figure 4.4e), the data at day 7 was zero due to a possible error in the quantification of the metals during the chemical analysis because the lead in the eluate solution was lower than that in the blank solution. 92 Eluate concentrations were compared with the method detection limit (MDL) and the lower limit of quantitation (LLOQ) to indicate quantitation of measured concentrations. The MDL is the minimum concentration of an analyte for a given analytical technique and sample matrix. The LLOQ is a minimum concentration of an analyte that can be measured within specified limits of precision and accuracy [38]. The comparison with these two limits is due to the leaching test's quality assurance/quality control step, in case a quality assurance project plan is required [38]. Eluate concentrations should be greater than the MDLs values but may be higher or lower than the LLOQs. If concentrations are lower than the MDLs, they should be reported at one-half the MDLs values for further calculations or decision-making. Figure 4.5 shows the interval flux rate of mass released over an interval (μg/m2-sec) for each mixture, calculated by multiplying the eluate concentration (μg/L) by the ratio of the volume of leachate to the surface area of the sample (L/m2) and dividing by interval specific time in second [38]. Mass transport describes the diffusion of the constituents from the solid material to the water. The specific rate of diffusion is the speed at which a constituent travels through the water, proportional to the magnitude of the concentration gradient (diffusion is faster when the incremental difference in concentrations is greater) [38]. The rate of diffusion depends on the distance the constituent has to travel, the effective porosity of the material, the tortuosity of the porous network. The interval flux typically decays from left to right. When the leaching interval flux decays constantly, it ideally follows the Fickian diffusion’ law [157]. The ideal diffusion trend -1/2 is shown as the negative square root of time (t ) trend (grey dot-dash line) in Figure 4.5. All metals decay faster than the ideal trend. The diffusion of copper and zinc in all mixtures is greater than that of other metals because the trend of the curves is steeper than the diffusion of other metals. 93 In the field, the rate that constituents can leach may be different from that obtained from the laboratory testing because it can be affected by chemical and physical factors that remain constant under laboratory conditions but may vary in the field (e.g., pH, change in liquid-solid ratio due to precipitations, biological activities that alter the oxidation of the state of a constituent, physical degradation due to freeze/thaw cycles or erosion) [38]. Figure 4.4. Eluate concentrations for As, Cd, Cr, Cu, Pb, and Zn compared with lower limit of quantitation (LLOQ) and method detection limit (MDL) 94 Figure 4.5. Interval flux rates of mass released over specific intervals of time, compared with the ideal diffusion trend Figure 4.6 shows the cumulative mass release (μg/m2) of the metals from all mixtures, calculated by multiplying the interval mass flux by the interval specific time and summing across all previous leaching intervals. The cumulative release was compared to a trend line proportional to the square root of time (t1/2). The cumulative zinc release follows the ideal trend. The other metals showed less leaching than predicted by the Fickian diffusion trend. 95 Figure 4.6. Cumulative release of mass per exposed surface area of test samples compared with the trend line 96 Figure 4.7. pH and conductivity measured at each interval of the leaching test 4.3.2 Assessment ratio Data from Method 1315 represents the rate of mass transport from the interior of the material to its external surface. This method is applicable for cases where water flows primarily around the material and then percolates through it, like in the case of the pavements. In the field, the actual liquid to surface area ratios is often much less than the test conditions and leaching into limited contacting liquid reduces the concentration gradients. The rate of leaching in the field can be less than measured in laboratory testing [38]. Eluate concentrations from Method 1315 are used for calculating fluxes and are not reflective of the expected maximum leaching concentrations. Therefore, results should not be compared with the threshold concentrations [38]. However, there is a way to correlate the mass transport to the field conditions and compare the test results to the threshold of the drinking water concentrations through the assessment ratio (AR). The assessment ratio (Eq. 4.1) correlates the average leaching concentration Ciav, calculated as in Eq. 4.2, to the threshold concentration Cthres (drinking water limits, in this work) and the dilution attenuation factor, DAF. 97 Ciav is calculated as a function of the effective concentration C1 and C2 and the infiltration events N1 and N2. C1 and C2 are the effective concentration for ≤1 day and > 1-day infiltration events, respectively. They are calculated based on Eq. 4.3 and Eq. 4.4. C1 and C2 are estimated using the cumulative mass release from Method 1315 results: the first three values in Figure 4.6 (R1, R2, R3 in Eq. 4.3 and Eq. 4.4). These values (the mass release from over 2-day cumulative leaching) have been assessed to be a good approximation for release over extended infiltration event >1 day in the field [158]. -3+? Eq. 4.1 ;K = -!@=6/ ∙ L;M /1 ∙ -1 + /2 ∙ -2 Eq. 4.2 -3+? = N P /1 + /2 (∑ K* − K1 ) ∙ ;6A< SB Eq. 4.3 -1 = R V Q1 ∙ ;37) 1000U (∑ KB − K1 ) ∙ ;6A< SB Eq. 4.4 -2 = R V Q2 ∙ ;37) 1000U Q = LQ − '<+7 − 'W# Eq. 4.5 P1 and P2 represent the net infiltration of the rain events less or equal to 1 day and greater than 1-day, respectively. The net infiltration is the infiltration rate of water in the environment. It is calculated as in Eq. 4.5 through the EPA dataset that includes data collected over 30-year (from 1961 to 1990) [159]. The dataset for Lansing, Michigan was used to calculate P1 and P2 as a function of the daily precipitation (DP), the pan evaporation (Epan) depending on climate elements 98 (e.g., temperature, humidity, rainfall), and the evapotranspiration (ET0) depending on the evaporation from the land surface and transpiration from plants. The results of the Lansing area were compared to those obtained by using data of other areas in Michigan, like Grand Rapids and Muskegon (Table 4.2). Table 4.2 Net infiltration, P1 and P2, and number of ≤1 day and >1 days events, N1 and N2 Lansing, MI Grand Rapids, MI Muskegon, MI P1 P2 P1 P2 P1 P2 (cm) Average 0.85 0.97 0.98 1.15 0.85 1.00 Max 10.27 5.96 7.87 8.06 6.82 9.44 Min 0.01 0.01 0.01 0.01 0.01 0.01 N1 N2 N1 N2 N1 N2 Number 403 64 383* 72* 438 66 event/30 y Number 13.43 2.13 14.19 2.67 14.60 2.20 event/y *Average over 27 years Table 4.3 reports the AR for chromium, lead, and zinc calculated for each mixture, based on Eq. 4.1, for the three areas considered in Michigan. Only these three metals were assessed because their concentration was higher than the detection limit in the first three intervals of time (Figure 4.4). The Cthres was the drinking water limit reported in Table 4.4, established by U.S. EPA [38]. The dilution attenuation factor, DAF, was set to one [158]. Table 4.4 - Table 4.7 reports the coefficients calculated for Lansing, Grand Rapids, and Muskegon, based on the mass release and net filtration. The AR of each metal was similar in all three areas considered. The AR related to the zinc was higher than 1, instead in all other cases the assessment ratio was less than 1. Assessment ratio less than or equal to 1 (AR≤1) indicates that the constituent is not likely to leach at concentrations higher than the threshold limit. By comparison, an assessment ratio higher than 1 (AR>1) does not 99 necessarily mean that constituents will leach at a level greater than the threshold limit in the field, but further evaluation is required. For instance, additional leaching tests using different methods can be performed, or field conditions may be changed in order to reduce the potential release [38]. In the case of zinc, the use of calcium carbonate in asphalt mixtures may absorb the dissolved zinc from the solution [31, 33, 35]. The hypothetical P1 and P2 were back calculated to have an AR for zinc equal to 1 based on each asphalt mixture (Table 4.8). The hypothetical P1 and P2 represent the minimum values for avoiding further evaluations or the need to take actions in the field to limit the potential leaching of the substance (the zinc in this case). Table 4.3 Assessment ratio (AR) in Lansing, Grand Rapids, and Muskegon areas Cr Pb Zn AR – Lansing, MI Control 0.17 0.31 16.36 SBS 0.19 0.12 33.76 PCR dry 0.23 0.52 97.98 PCR wet 0.20 0.49 72.48 DVR 0.22 0.11 32.10 AR – Grand Rapids, MI Control 0.15 0.26 14.33 SBS 0.17 0.11 29.20 PCR dry 0.20 0.45 84.81 PCR wet 0.18 0.43 63.08 DVR 0.19 0.10 28.43 AR – Muskegon, MI Control 0.17 0.30 16.17 SBS 0.19 0.11 33.60 PCR dry 0.23 0.52 97.47 PCR wet 0.20 0.49 71.89 DVR 0.22 0.11 31.52 100 Table 4.4 Parameters used to calculate the assessment ratio Cr Pb Zn Drinking water limit (μg/L) 100 15 10 Control mixture R1 (μg/m2) 142.35 136.59 9,490.16 2 R2 (μg/m ) 272.87 176.21 10,724.66 2 R3 (μg/m ) 413.85 176.21 12,201.32 SBS mixture 2 157.83 46.69 4,849.88 R1 (μg/m ) 2 R2 (μg/m ) 307.54 56.68 7,681.75 R3 (μg/m2) 425.05 100.37 8,395.97 PCR dry mixture 2 R1 (μg/m ) 197.93 49.07 2,583.80 2 R2 (μg/m ) 380.34 113.11 10,760.44 2 R3 (μg/m ) 523.29 142.57 13,174.94 PCR wet mixture 2 181.59 77.83 4,878.04 R1 (μg/m ) 2 R2 (μg/m ) 337.14 138.36 10,662.62 2 R3 (μg/m ) 496.12 165.42 14,609.58 DVR mixture R1 (μg/m2) 206.36 88.32 4,985.69 2 R2 (μg/m ) 371.23 99.09 7,164.01 2 R3 (μg/m ) 588.72 130.09 12,049.93 101 Table 4.5 Average leaching concentration (Ciav) and effective concentration C1 and C2, Lansing Cr Pb Zn Control mixture C1 (μg/L) 15.36 4.66 145.24 C2 (μg/L) 27.99 4.08 279.50 av 17.09 4.58 163.64 Ci (μg/L) SBS mixture C1 (μg/L) 17.61 1.18 333.16 C2 (μg/L) 27.55 5.53 365.58 Ciav (μg/L) 18.97 1.77 337.60 PCR dry mixture C1 (μg/L) 21.46 7.53 961.96 C2 (μg/L) 33.54 9.64 1,091.87 Ciav (μg/L) 23.12 7.82 979.76 PCR wet mixture C1 (μg/L) 18.30 7.12 680.54 C2 (μg/L) 32.43 9.03 1,003.25 av Ci (μg/L) 20.24 7.38 724.77 DVR mixture C1 (μg/L) 19.40 1.27 256.27 C2 (μg/L) 39.42 4.31 728.27 Ciav (μg/L) 22.14 1.68 320.96 102 Table 4.6 Average leaching concentration (Ciav) and effective concentration C1 and C2, Grand Rapids Cr Pb Zn Control mixture C1 (μg/L) 13.32 4.04 125.97 C2 (μg/L) 23.61 3.45 235.75 av Ci (μg/L) 14.95 3.95 143.34 SBS mixture C1 (μg/L) 15.28 1.02 288.97 C2 (μg/L) 23.24 4.67 308.36 av 16.54 1.60 292.03 Ci (μg/L) PCR dry mixture C1 (μg/L) 18.61 6.53 834.35 C2 (μg/L) 28.29 8.13 920.97 Ciav (μg/L) 20.15 6.79 848.06 PCR wet mixture C1 (μg/L) 15.87 6.18 590.26 C2 (μg/L) 27.35 7.62 846.22 av 17.69 6.40 630.77 Ci (μg/L) DVR mixture C1 (μg/L) 16.82 1.10 222.28 C2 (μg/L) 33.25 3.63 614.28 av 19.42 1.50 284.31 Ci (μg/L) 103 Table 4.7 Average leaching concentration (Ciav) and effective concentration C1 and C2, Muskegon Cr Pb Zn Control mixture C1 (μg/L) 15.36 4.66 145.24 C2 (μg/L) 27.15 3.96 271.12 av 16.90 4.57 161.72 Ci (μg/L) SBS mixture C1 (μg/L) 17.61 1.18 333.16 C2 (μg/L) 26.72 5.37 354.61 Ciav (μg/L) 18.81 1.72 335.97 PCR dry mixture C1 (μg/L) 21.46 7.53 961.96 C2 (μg/L) 32.54 9.35 1,059.11 Ciav (μg/L) 22.91 7.77 974.68 PCR wet mixture C1 (μg/L) 18.30 7.12 680.54 C2 (μg/L) 31.45 8.76 973.15 av Ci (μg/L) 20.02 7.34 718.86 DVR mixture C1 (μg/L) 19.40 1.27 256.27 C2 (μg/L) 38.24 4.18 706.42 Ciav (μg/L) 21.86 1.65 315.22 Table 4.8 Hypothetical P1 and P2 P1 P2 (cm) Control 12.35 27.11 SBS 28.32 35.46 PCR dry 81.77 105.91 PCR wet 57.85 97.32 DVR 21.78 70.64 4.3.3 Digestion and metals characterization To understand what metals were included in the materials used in each asphalt mixture and that are expected to leach out from the samples, the microwave digestion of the materials was performed. Figure 4.8 shows the results of the digestion in terms of mg of metals per kg of material (aggregates, RAP, bitumen, SBS, Sasobit, PCR, and DVR). The complete list of the results 104 including all metals are reported also in Table C7. Almost all metals concentration was higher in the aggregates and RAP (approximately constituted by 95% of aggregates by mass) than in the organic materials, such as bitumen and crumb rubber. Nickel was present more in the bitumen than in the other materials, and copper and zinc content was higher in the PCR and DVR. The high value of zinc in the CR products confirms data published in the literature saying that zinc is present in CR more than 1% by weight [8, 31, 32]. In PCR particles, zinc was 1.09% by weight, and in DVR particles, it was 0.83% by weight. This result may also confirm the hypothesis that zinc is more present in smaller particles than in CR with bigger size since the PCR and DVR particles have, respectively, around 0.6 and 5 mm of diameter. 105 Figure 4.8. Metals content in mg per kg of each material used in asphalt mixtures 106 Table 4.9 Material content in each mixture, in percentage Mix ID Control SBS PCR dry PCR wet DVR Materials (%) Aggregates 80.30 80.00 79.70 80.00 80.10 RAP 14.90 14.90 14.90 14.90 14.90 Bitumen 4.70 4.86 4.89 4.49 4.45 SBS - 0.17 - 0.15 0.09 PCR - - 0.50 0.35 - DVR - - - - 0.34 Sasobit - - - 0.15 - Table 4.10 Weight of the cylindrical specimens Mix ID Control SBS PCR dry PCR wet DVR Replicate (kg) 1 2.47 2.49 2.46 2.48 2.50 2 2.46 2.48 2.44 2.48 2.51 3 2.47 2.48 2.45 2.48 2.50 Based on the metal contents present in each material, the percentage of materials in each mixture (Table 4.9), and the weight of each specimen used for the leaching test (Table 4.10), the content of metals in each sample was calculated (Table C8, Table C9, Table C10, Table C11, Table C12). Figure 4.9 shows the average metal contents of the three specimens for each mixture. Figure 4.9a depicts that, except for the zinc, all the other metals' content is constant in each mixture. Zinc content is higher in PCR dry, PCR wet, and DVR mixtures because PCR and DVR give their contribution besides the zinc contained in the aggregates and RAP. Figure 4.9b-f shows that the major contribution of the metals is due to the aggregates. Table 4.11 reports the average metal content in all materials for each mixture. 107 Figure 4.9. a) Total mg of metals presents in each mixture, average of the three replicates; b-f) average of metals (mg) per specimen of each mixture, contribution of all materials. 108 Table 4.11 Metals content in each material per mixture (mg/specimen) Control SBS PCR dry Aggregates RAP Bitumen Aggregates RAP Bitumen SBS Aggregates RAP Bitumen PCR mg/specimen for each mixture As 51.16 47.79 0.00 51.27 47.88 0.00 0.00 50.38 47.10 0.00 0.00 Cd 6.52 2.10 0.00 6.53 2.11 0.00 0.00 6.42 2.07 0.00 0.00 Cr 99.73 13.52 0.17 99.95 13.55 0.18 0.01 98.22 13.32 0.18 0.08 Cu 94.65 8.18 0.00 94.86 8.20 0.00 0.00 93.22 8.06 0.00 6.99 Pb 174.15 40.13 0.00 174.53 40.20 0.00 0.00 171.51 39.55 0.00 0.03 Zn 68.42 29.30 0.00 68.57 29.35 0.00 0.00 67.38 28.88 0.00 133.94 PCR wet DVR Aggregates RAP Bitumen SBS PCR Sasobit Aggregates RAP Bitumen SBS DVR mg/specimen for each mixture As 51.20 47.85 0.00 0.00 0.00 0.00 51.79 48.39 0.00 0.00 0.00 Cd 6.53 2.11 0.00 0.00 0.00 0.00 6.60 2.13 0.00 0.00 0.00 Cr 99.82 13.54 0.16 0.01 0.06 0.01 100.96 13.69 0.17 0.00 0.04 Cu 94.74 8.19 0.00 0.00 4.95 0.00 95.82 8.28 0.00 0.00 0.03 Pb 174.31 40.19 0.00 0.00 0.02 0.00 176.29 40.64 0.00 0.00 0.00 Zn 68.48 29.34 0.00 0.00 94.94 0.00 69.26 29.67 0.00 0.00 70.59 109 Since a high contribution to the total metals content in asphalt mixtures seems to derive from aggregates, we tested the aggregates from other States such as Maryland, North Carolina, and Wisconsin to evaluate and compare the difference in metals content to those used in Michigan. In North Carolina and Maryland, the zinc was even higher than in Michigan, and Figure 4.10 gives an overall idea of the difference in metal content per kg of aggregates between Michigan and North Carolina, Wisconsin, and Maryland. Figure 4.11 shows the difference in mg of metals included in the asphalt mixtures with respect to the content present in the Michigan samples (zero line). The different metal content in the aggregates may lead to having different results from the leaching test. 110 Figure 4.10 Metals content in mg per kg of Michigan, North Carolina, Wisconsin, and Maryland aggregates 111 Figure 4.11 Difference in mg of metals contained in asphalt mixtures samples 4.4 Summary of chapter findings The objective of this part of the work was to investigate through the leaching test if metals englobed in the asphalt mixes placed as surface layers in road pavements could be released into the groundwater due to the rainwater after construction and in what concentration. Moreover, the metals trace in each material was assessed by performing microwave digestion. The main findings have been summarized as follows: • In terms of metal concentrations in the eluate from the leaching test (μg/L), zinc concentrations were higher than the other metals, and it was present in the eluates of all mixtures, rubberized (PCR dry, PCR wet, and DVR) and reference blends (Control and SBS), in similar concentrations, over the duration of the test (63 days). • The cumulative mass released (μg/m2) was used to calculate the assessment ratio that correlates the leaching results laboratory conducted to the field conditions. The assessment ratio for zinc was higher than one, meaning that further evaluation is required to understand if zinc will leach at a level greater than the drinking water limit. • Based on the results of the digestion, zinc was present in the aggregates, RAP, PCR and DVR. Per kg of each material, PCR and DVR contained more zinc than aggregates. However, since the asphalt mixtures are composed of approximately 80% of aggregates 112 and 0.34-0.5% of PCR and DVR by weight of total mix, the major contribution of zinc derives from aggregates. • Based on the digestion results, the total content of As, Cd, Cr, Cu, and Pb was constant in each asphalt mixture considered in this work. Zinc content was higher in PCR dry, PCR wet and DVR mixtures with respect to Control and SBS mixes due to the contribution of zinc in PCR and DVR besides that coming from aggregates and RAP. • Aggregates from other States such as North Carolina, Wisconsin, and Maryland have been investigated as well. Zinc content in Wisconsin was lower than that in Michigan. Maryland and North Carolina aggregates had a zinc content higher than that in Michigan. Different metal content in the aggregates may influence leaching results. 113 5. MATERIAL USAGE FOR THE USA AND GLOBALLY WIND TURBINES3 In this part of the work, as part of a National Science Foundation project (Grant number 1801785), I assessed the material usage in wind turbines manufacturing and the associated environmental impacts in the USA and the rest of the world until 2050. The goal was to ensure that there will be enough materials to meet the expected global capacity growth of wind energy, whose global electricity generation is expected to increase from 5 to 30% in 2050. Moreover, it is important to evaluate the potential impact of large-scale deployment of wind energy to avoid as much as possible creating new issues related to material scarcity, which could increase the carbon footprint of future electricity production. 5.1 Methodology The overall goal of this work was to evaluate the material requirement associated with the large-scale deployment of wind energy in the USA and the rest of the world in the past (from 1991 and 2017) and in the future (until 2050). The annual historical material inventory for manufacturing and transporting onshore and offshore wind turbines components was compiled on a capacity basis (tons/MW). The amount of critical and non-critical material used was compiled and the associated environmental impact in terms of carbon footprint and cumulative energy demand were calculated. The material demand until 2050 was forecasted based on the size scaling effect of wind turbines and the expected capacity additions for the USA and the rest of the world. 3 This chapter have been published as Angela Farina, Annick Anctil, “Material consumption and environmental impact of wind turbines in the USA and globally”, Resources, Conservation & Recycling, DOI: https://doi.org/10.1016/j.resconrec.2021.105938 Copyright Elsevier 2021 114 I compared three different projections for the installed capacity by 2050 in the USA and one for the rest of the world and considered two future scenarios for the use of the REEs in wind turbines. The aim was to evaluate the total material need for wind turbines compared to future global production. 5.1.1 Historical material intensity The historical material intensity for onshore and offshore wind turbines constructed from 1991 to 2017 was compiled from journal articles, life cycle assessment reports, and the Ecoinvent database [47, 48, 165–174, 52, 83, 114, 160–164]. The functional unit was 1 MW wind turbine since the wind turbines capacity varies. Materials were divided into non-critical and critical, and their quantity was reported in ton/MW. I calculated the raw elements mined from the ground, carbon footprint and cumulative energy demand per 1 MW wind turbine. The system boundaries included the materials used for each component and the transportation of the components from the wind turbine assembly to the construction site (Figure 5.1). The LCA was conducted using SimaPro 9.1 [117] and Ecoinvent 3.6 [118] and US-EI 2.2 [120] databases for the primary production of all non-critical materials and transportation. For the critical materials (Nd and Dy), I modeled the ore extraction based on data available in the literature [175, 176]. The transportation of all wind turbines’ components was included. Wind turbines were assumed to be constructed in the USA and the average distances and mode of transportation for onshore [177] and offshore [171] turbines was used. The impact categories were the IPCC 2013 Global Warming Potential (GWP) [178] and the Cumulative Energy Demand [71]. 115 Figure 5.1. System boundaries including raw materials used for onshore and offshore turbines (foundation, tower, nacelle, and rotor blades) and transportation between assembly and construction site. Raw materials: m1=concrete; m2=gravel; m3=steel and iron; m4=copper; m5=aluminum; m6=fiberglass; m7=polyethylene and polymers; m8=epoxy; m9=rare earth materials. 5.1.2 Future material requirement To forecast the material intensity for wind turbines between 2018 and 2050, we considered three different types of onshore (DFIG, DDSG, and PMSG) and offshore turbines (conventional land-based, with and without PM, and floating). I used an established method that takes into account the size scaling effect over time [52] to calculate the size of the turbine as a function of the capacity. It also calculates the weight and size of the tower, nacelle, and blades but not the foundation. The ratio of foundation per weight of wind turbine was calculated using historical weight of wind turbines (tower, nacelles, and blades) and weight of the foundation. The ratio was used to calculate future foundation weight based on the weight of wind turbines. 116 Three different wind energy forecasts were considered: the energy transition outlook from government (EIA) and a private company (DNVGL) as well as the Department of Energy Wind Vision study [179–181]. The DNVGL outlook is related to North America and all other regions in the world. In order to have data for the USA only, the contribution associated with Canada has been subtracted by using data of the National Energy Board (NEB) reported in the Government of Canada website [182]. Also, this database includes data until 2040. To have a complete projection, I extrapolated data for the period 2041-2050. For the extrapolation were considered a linear trend that better-fitted data from 2030 to 2040. The percentage for each generator for both onshore and offshore was assumed to change from the initial values in 2018 (73.4% DFIG, 20.8% PMSG, and 5.8% DDSG) [50]. I considered two scenarios, in the first one I assumed a linear increase in PMSG up to 75% in 2050 for both onshore and offshore turbines (high market share scenario, High) [183–186]. By comparison, in the second scenario I assumed a linear decrease in PMSG down to 14.6% (30% lower of the initial value in 2018) in 2050 (low market share scenario, Low). The Low scenario considers the possible partial substitution of the turbines with PM with other technologies, such as the DFIG [187]. In both scenarios, the amount of REE used in the PM was assumed constant between 2018 to 2050. I calculated the percentage of DFIG and DDSG turbines in 2050 proportionally. I adjusted the percentages of the other turbines until 2050 using the same average increment, dividing the difference between the percentage of the turbines in 2050 and 2018 by the number of years. The percentage of floating turbines for offshore wind turbines is expected to grow from 1% in 2018 to 15% in 2050 [50]. The future total production of steel, cement, and REE was compared to the material demand for wind turbines (Table A5). The steel production was assumed to increase 1.5 times compared to current production by 2050 to meet the needs of the growing population [188]. 117 Similarly, global cement production is expected to increase between 12-23% by 2050 [189]. I considered the value of 12% in our model to verify if the material demand for wind turbines was satisfied using the lowest range value. For the REE production forecast, I used a previous report for China that was calculated using the Generalized Weng model [190]. 5.2 Results 5.2.1 Historical material intensity The material intensity per MW of global onshore and offshore turbines constructed from 1991 to 2017 (Figure 5.2 a and b, respectively) shows that concrete and steel are the most important material by mass. Although REEs receive a lot of attention, they represented on average only 0.03% and 0.01% by weight of all materials used for onshore and offshore wind turbines, respectively. By comparison, concrete and steel were on average 75.1% and 22.8%, respectively, by weight of the total materials used in onshore turbines. On the other hand, for offshore turbines, concrete and gravel represent on average respectively 38.83% and 33.29% of the total mass, and steel is 26.24% by weight. The share of all materials is almost constant over the considered period. The different symbols (circle, triangle, and square) indicate multiple values for the same material in the same year from different sources. Concrete and steel were used to construct the foundation of onshore and conventional offshore turbines. In the case of floating offshore turbines, gravel and steel were used as ballast and anchorage. Fiberglass and epoxy resins were used for blades, and steel for towers. For nacelle were used copper, aluminum, steel, cast iron, fiberglass, and Nd and Dy in the PMSG generator. Steel and iron in the following figures include reinforced steel, unalloyed, low-alloy, chromium, and cast iron. 118 Figure 5.2. Material intensity for a) onshore and b) offshore wind turbines in terms of ton per MW from 1991 to 2017, based on data available in the literature. 5.2.2 Future material requirement The main question I wanted to address in this manuscript is whether there will be enough materials to build wind turbines in the future in the USA and globally. To answer this question, I calculated the material demand for wind turbines in the USA and the rest of the world between 2018 and 2050. Figure 5.3 shows that concrete, steel, and REE demand will change overtime based on the three different scenarios (DNVGL, Wind Vision, and EIA) in the USA (Figure 5.3 a, b, c, d) and the rest of the world (Figure 5.3 e, f, g, h). For the REE, the demand will vary also based on the future market share of the wind turbines using permanent magnets (Low and High scenarios, 119 Figure 5.3 c, d, g, h). In both the Low and High scenarios, the overall demand for non-critical materials will remain the same. In DNVGL scenario, material demand for the USA turbines shows a peak in the period 2032-2036 (Figure 5.3 a, b, c, d), while the material consumption in the rest of the world increases overtime (Figure 5.3 e, f, g, h). By comparison, in the Wind Vision case, the material demand remains constant because the capacity addition increases linearly in the periods 2020-2030 and 2030-2050. The EIA scenario shows a completely different trend with respect to the previous two. The material demand shows a peak in 2023 and then it decreases by 85% in 2050, with respect to the peak level. The reason is that government projection as EIA is often conservative about renewable growth because driven by climate policy necessary to encourage renewable energy investments [191]. Overall, the material demand for wind turbines is lower than the expected material production by 2050 (Figure 5.4 a-h). However, a shortage of materials might happen depending on the future wind capacity installed and the global demand for the same materials in other applications. 5.2.2.1 Concrete In the case of concrete, cement might be consumed more than the average amount used in the energy sector. For instance, 1.95 Mt of cement per year will be used to meet the concrete annual demand of 13 Mt in the USA in the peak period (2032-2036) in the DNVGL scenario (Figure 5.4 a). This amount of cement, calculated by assuming 0.15 ton of cement per ton of concrete, corresponds to more than 2% of the total USA cement production in the same period. This value is more than double of the average cement tonnage used for both wind energy and pipeline construction (1%) [192]. In the EIA scenario the amount of cement that will be used overcomes the 1% level in the peak period. In the Wind Vision scenario, on average, I estimated that the 120 cement will be 0.7% of the expected cement production over the 2018-2050 period. A demand of cement higher than the average supply will lead to an increase cement production to meet the concrete demand for the wind turbines segment and the needs of the growing population, with a consequent increase of the greenhouse gas emissions (890 kg CO2 eq. per ton of cement produced). The use of blast furnace slag as partial substitution of the cement can reduce the environmental impact associated with the concrete in wind turbines, without affecting mechanical properties and durability [193]. 5.2.2.2 Steel Steel is a fundamental material in the clean production technologies, such as the wind sector, where it is used in all turbines components. Even if the percentage of steel demand for wind turbines compared to the total expected production seems to be low, the USA will probably experience difficulties in procuring steel in the next future [194]. Recently, the USA faced up to a steel shortage by importing 18% of the steel to meet the demand for all needs in 2020 and the ongoing coronavirus pandemic raised the steel price with the highest increment in the last 13 years [194]. Recycling more steel is the solution to deal with the expected shortage and ensure a future supply. I estimated 511% increase in steel demand in 2033 with respect to the current level in the USA, based on the DNVGL scenario (Figure 5.3 b). In 2018, the steel consumption was 1% of the total USA production in the same year in all three scenarios investigated. I estimated that the use of steel will be growth up to 2.3% (EIA) and 4.7% (DNVGL) of the total production in the respective periods of peak capacity. The increase corresponds to a maximum consumption of 2.2 Mt (EIA) and 5.2 Mt (DNVGL) in the two mentioned periods (Figure 5.3 b). In Wind Vision scenario, on average, 1.2% of steel will be consumed with respect to the total USA production until 2050, corresponding to 1.37 Mt of steel annually used (Figure 5.3 b). In the rest of the world, 121 the steel demand for wind turbines will increase linearly, consuming from 0.3% to 1.1% of the total global production overtime. On average, the rest of the world will annually consume 15.8 Mt of steel in wind energy sector, and the steel use will increase 5 times in 2050 with respect to the 2018 level (Figure 5.3 f). 5.2.2.3 Rare earth elements In the case of the DNVGL outlook, which is the most ambitious scenario (Figure 5.3 c), on average, the REE demand in the USA will consume 0.7-1.8% of the Chinese production until 2050. The increase demand with respect to the 2018 level is different based on the market share of installed wind turbines using permanent magnets. The REE demand in the USA in 2033 will increase by 254% in the Low scenario and by 815% in the High scenario (DNVGL) (Figure 5.3 d). In the rest of the world, the REE demand will increase by 160% to 1240% (Low and High, respectively) in 2050 with respect to the 2018 level, based on the DNVGL outlook (Figure 5.3 g, h). Moreover, in the High scenario, the REE demand in the rest of the world is 38% of the expected Chinese production. Cleary the global demand of REE cannot be satisfied by producing REE from the ore extraction in China only. The demand for REE depends on the future market share of wind turbines using PM and the amount of REE that will be used in the magnet. On the other hand, the market share of the wind turbines will depend on the price of the REE and the techno-economic advantages of using PMSG turbines instead of other technology [187]. In the future, the amount of REE used in the PM will likely be reduced by increasing the material efficiency with the optimization of the PM design or lowering the work temperature of the turbines [187]. The amount of Dy, which currently is 2-5% by weight of the PM, might be lowered to 1% or completely eliminated [187]. In this study, I considered only Chinese production to meet the future demand for wind turbines. Other 122 deposits should also be considered, such as those in Australia and the USA. Moreover, REEs can be extracted by alternative materials. Generally, there are many methods to extract REEs from wastewater and solid waste, but there are pros and cons. For example, the precipitation used to remove REEs from industrial water is an easy method, but it needs many chemicals, and it generates a considerable quantity of sludges. Moreover, the REEs concentration extracted is low [195]. The pyro-metallurgy used for solid waste and mineral concentrations is energy-intensive because it requires high temperatures and involves many chemicals [195]. Instead, adsorption seems to be the most simple, efficient, and economical method to extract REEs from wastewater [195]. However, methods may change based on the specific REE that can be extracted. The alternative sources to recover Nd and Dy used in wind turbines are the coal ash through the supercritical method or leaching and solvent extraction or, alternatively, they can be recycled from old permanent magnets through extraction and purification [196–198]. The option to produce REE from recycling instead of the ore extraction should be also considered to estimate the material availability of critical materials. 123 Figure 5.3. Percentage of increase in material demand for concrete, steel and REEs compared to 2018 level for the USA (a, b, c, d) and the rest of the world (e, f, g, h) wind turbines. 124 Figure 5.4.Material demand of concrete, steel and iron, REE based on DNVGL, Wind Vision, and EIA for the USA (a, b, c, d) and the rest of the world (e, f, g, h) compared to the expected material production. 125 5.2.2.4 Copper Besides concrete, steel and REE, I calculated the global demand for the other materials used in wind turbines until 2050. The current global copper reserves are estimated at 830 million tons and the global annual copper demand is 28 million tons [199]. Copper is used in small quantities in wind turbines. Globally, I calculated 54 thousand tons of copper used for wind turbines in 2018. Copper demand for global onshore and offshore turbines is 14 times (754 thousand tons) higher in 2050 with respect to the current level, based on the DNVGL outlook. Even if results indicate that there is enough copper to meet current demand, the cumulative global copper production is expected to exceed the reserves by 2040 [84]. Even if we have enough copper, we need to use it efficiently. There is low risk of copper shortage since it is available in more than 20 countries, such as Chile, Peru, China, and the USA and is highly recyclable. About 35% of the world’s copper consumption comes from recycled copper, which significantly reduces the need for mining [199]. However, recycling alone will not be sufficient to meet the demand and ensure a stable supply of copper and we will continue to need to mine copper. 5.2.2.5 Aluminum Aluminum is one of the most important metals used in many sectors, such as transportation, energy, packaging, and electrical. The global production was 64,000 thousand tons in 2019 [200] and the global aluminum demand is expected to reach 298 million tons per year by 2050 [201]. The global resource of bauxite, the primary ore for aluminum, is estimated to be between 55 to 75 billion tons, which is sufficient to meet the demand for metal in the future (USGS, 2020b). I calculated a total consumption of 83 thousand tons of aluminum for onshore and offshore turbines in 2050, globally and our estimate did not assume an increase of recycling material over time, which is likely to happen [201, 202]. 126 5.2.2.6 Fiberglass Even if fiberglass and plastics represent only 1.3% by weight of wind turbines, they are made by using chemicals and high temperatures processes. In particular, fiberglass requires natural minerals (i.e., silica sand, limestone, and soda ash) and chemicals that are melted into a furnace using electricity and fossil fuel [203]. In the USA, the fiberglass manufacturing process use 73% of natural gas, 24% of electricity, and 3% of other fuels [204]. I estimated 3,601 thousand tons of fiberglass used for wind turbines worldwide in 2050. Assuming the process will not change in the future (0.109 m3 of natural gas per 1 kg of fiberglass produced), I calculated slightly less than 324 million cubic meters of natural gas for fiberglass production for wind energy, which is 0.00016% of the current global natural gas reserves (198.8 trillion cubic meters) located in Russia, Iran, Qatar, Turkmenistan, and the USA [205]. The USA has enough dry natural gas to last about 92 years [206]. The use of fiberglass in wind turbines to produce blades is not a concern for the consumption of natural resources. However, the fiberglass is a composite material difficult to separate and recycle, and the landfill is currently the most convenient solution for the end-of-life of the blades. Wind turbines manufacturers, private companies and NREL will improve the recyclability of future blades by using a thermoplastic resin, which easier to melt and recycle than fiberglass [207]. 5.2.3 Future environmental impact of material production In addition to the material demand estimate for future wind turbines in the USA and the rest of the world, I quantified the environmental impact associated with the material production. Fig. 9 show the annual GWP based on yearly installation for the USA (Figure 5.5 a-c) based on the DNVGL, EIA and Wind Vision forecasts the rest of the world based on the DNVGL projection (Figure 5.5 d). The CED was also calculated and reported in Figure 5.6. The total carbon footprint associated with the global production of all materials to construct wind turbines in 2050, based on 127 the DNVGL outlook, was 108,970 thousand ton of CO2 eq. which is 20 times lower than the CO2 eq. due to the total current electricity generation in the USA. By comparison, the total CED to produce materials to manufacture wind turbines globally in 2050 was 1,566,787 TJ, which is 33 times more than the CED of the total current electricity generation in the USA. The two processes that contribute the most to the energy consumption are the steel and fiberglass production. The current estimate assumes only primary manufacturing. I can anticipate that the CED of the manufacturing phase will decrease over time as we used more recycled steel and the fiberglass is replaced by materials that are easier to recycle such as the thermoplastic resin. Figure 5.5. Global warming potential (GWP) of onshore and offshore wind turbines in the USA based on a) DNVGL, b) Wind Vision, c) EIA outlook, and (d) in the rest of the world based on the DNVGL outlook, in terms of thousand tons of CO2 eq. 128 Figure 5.6. Cumulative Energy Demand (CED) of onshore and offshore wind turbines in the USA based on a) DNVGL, b) Wind Vision, c) EIA outlook, and in the rest of the world based on the DNVGL outlook (d), in terms of terajoule (TJ). 5.3 Summary of chapter findings The goal of this work was to estimate the material demand for wind turbines in the USA and in the rest of the world until 2050 and the associated carbon footprint and cumulative energy demand. The material demand was compared with the expected material production and availability. The main findings have been summarized as follows: • The material intensity per MW of global onshore and offshore wind turbines constructed from 1991 to 2017 shows that concrete and steel are the most important material by mass. REEs represented instead on average only 0.03% and 0.01% by weight of all materials used for onshore and offshore wind turbines, respectively. 129 • The material intensity did not change depending on the type of turbines overtime, but mostly between onshore and offshore. • The material demand for wind turbines was overall lower than the expected material production by 2050. • The predicted USA cement usage for onshore and offshore turbines during the 2023-2036 period was expected to be the double of the average cement tonnage used for both wind energy and pipeline construction in the USA. The use of blast furnace slag as partial substitution of the cement can reduce the environmental impact associated with the concrete in wind turbines, without affecting mechanical properties and durability. • The steel demand was estimated to increase by 511% in 2033 with respect to the current level in the USA, arriving to consume up to 4.7% of the total production. Recycling more steel is the solution to deal with expected increasing demand and ensure a future global supply. • The rare earth demand will increase up to 815% in the USA and 1,240% in the rest of the world in 2050 with respect to the today level, considering that 75% of the global installed turbines will use permanent magnet and that the rare earth materials will be produced by ore extraction in China only. However, there might be a higher or much lower proportion of wind turbines using permanent magnets in the future and rare earth composition of those magnets might change. In the future, besides the ore extraction, rare earth elements might be more likely to be recovered or recycled from old devices or extracted from alternative sources. The availability rare earth materials will also depend on the future market of the other technologies, and therefore shortage remains possible. 130 • The total CO2 emissions associated with the production of the materials to construct wind turbines in the USA and in the rest of the world in 2050 was 20 times lower than the CO2 due to the total current electricity generated in the USA. The total CED was 33 times more than the CED of the total current electricity generation in the USA. The two processes that contribute the most to the energy consumption are the steel and fiberglass production. The current estimate assumes only primary manufacturing. The CED of the manufacturing phase will decrease over time as we used more recycled steel and the fiberglass is replaced by materials that are easier to recycle such as the thermoplastic resin. • The main limitation of this work relates to the forecasted installed capacity and the potential for drastic change in wind turbines design. Consequently, the percentage of the materials used for wind turbines with respect to the expected production might differ. • This work only considered material use for wind turbines and did not consider the simultaneous increasing demand for other technologies, such as solar energy, electric vehicles, and energy storage. 131 6. CONCLUSIONS This dissertation focused on a comprehensive evaluation (engineering performance and sustainability aspects) of road pavements using asphalt mixtures modified with crumb rubber from end-of-life tires and the material usage estimation for wind turbines in the USA and globally. Potential for replacement of synthetic polymer with scrap tire rubber was investigated from the technical, environmental, and toxicity perspectives. In addition, this study investigated whether the material demand for wind turbines is lower than the expected material production on the USA and in the rest of the world until 2050. 6.1 Environmental impacts of roads pavements This part of the dissertation aimed to evaluate whether asphalt mixtures containing recycled modifying agent such as polymer coated rubber (PCR) and devulcanized rubber (DVR) might perform as good as or better than mixes modified with synthetic polymers and unmodified mixtures, from mechanical, environmental, and toxic perspectives. Environmental impacts were different based on the allocation methods used to model the crumb rubber. The cut-off method was evaluated as the better method to compare different types of asphalt mixtures. This research highlighted the importance of including mechanical performance results in life cycle assessment to better quantify the environmental impact of the asphalt mixtures over the service life. A cradle- to-gate analysis does not provide an exhaustive idea of the environmental performance of the asphalt mixtures because the overall impacts changed when considering the mechanical performance results in the cradle-to-grave LCA. Results may be affected by the assumptions made in this work. For instance, considering the same mixture designed for a specific location to assess the mechanical performance in other climate zones in the USA may provide not accurate response 132 of that mixture to the mechanical distress prediction over time. In other locations, asphalt mixtures would have used materials with different characteristics (e.g., aggregates different per type and gradation, different content of asphalt binder). Consequently, the asphalt mixture would respond differently to the distress, changing the overall environmental performance as well. Therefore, this leads to the importance of having specific mix design for each location to better calibrate the models for the distress prediction. Moreover, in this study, the same asphalt mixture was used in all layers on the top of the unbound materials. In the field, the Departments of Transportation typically use different materials per layer to better deal with the distress that most affects the pavement based on the location, climate and traffic. For instance, the PCR wet mix in California failed for rutting for a high traffic structure before the other mixtures did, but this does not mean that the PCR wet mixture is not valuable in such a condition. This means that in that particular case, an accurate design of the pavement is needed, choosing the right combination of asphalt mixture per each layer. Finally, results from leaching test and metals trace analysis through the microwave digestion indicate that not only the crumb rubber, reclaimed asphalt pavement and bitumen are responsible for metal leaching, but the aggregates as well, and that the location or origin of the aggregates (e.g., river, quarries) may influence the leaching results. 6.2 Material usage for wind energy This part of the dissertation aimed to estimate the material demand for wind turbines in the USA and in the rest of the world until 2050 and the associated carbon footprint and cumulative energy demand. The material demand was compared with the expected material production and availability. The material demand for wind turbines was overall lower than the expected material production by 2050 considering the most ambitious energy outlook with the highest future installed capacity. However, this work only considered material use for wind turbines and did not consider 133 the simultaneous increasing demand for other technologies, such as solar energy, electric vehicles, and energy storage. Therefore, a shortage of materials may remain possible. The total CO2 emissions associated with the production of the materials to construct wind turbines in the USA and in the rest of the world in 2050 was 20 times lower than the CO2 eq. due to the current total electricity generated in the USA. This work considered only primary material, and the use of recycled materials might lower the environmental impact of the wind turbines, contributing to decrease the impact of the future electricity production. Lowering the environmental impact of wind turbine manufacturing will make wind energy more competitive compared to non-renewable sources. The main limitation of this work relates to the forecasted installed capacity and the potential for drastic change in wind turbines design. As shown from the three forecasts used in this study, there is a lot of uncertainty about wind deployment. Overall, forecasts have generally underestimated the annual capacity addition [208, 209], therefore the demand for critical and non- critical materials might be higher than estimated in this work. Consequently, the percentage of the materials used for wind turbines with respect to the expected production might be higher. On the other hand, if the forecasts overestimated the capacity addition for wind energy, we overestimated the material demand, meaning that we will need less materials in the future compared to the availability. For turbines using permanent magnets I considered a scenario with growing market share and another with decreasing market share, both using a constant mass intensity for rare earth materials. However, there might be a higher or much lower proportion of wind turbines using permanent magnets in the future and rare earth composition of those magnets might change. The availability of metals and rare earth materials will also depend on the future market of the other technologies, and therefore shortage remains possible. Engineers and policymakers can use results 134 of this work to evaluate and select new designs with lower environmental impacts. Reducing the material intensity and associated environmental impact of wind turbines is critical considering the increasing number of wind farms expected to be built in the next 20-30 years. 135 APPENDICES 136 APPENDIX A Supplementary information Chapter 2 137 A1. Viscoelastic continuum damage model results Figure A1. C vs S curves for Control mixture, results of the VECD model Figure A2. C vs S curves for SBS mixture, results of the VECD model 138 Figure A3. C vs S curves for PCR dry mixture, results of the VECD model Figure A4. C vs S curves for PCR wet mixture, results of the VECD model Figure A5. C vs S curves for DVR mixture, results of the VECD model 139 Table A1 VECD results: Number of cycles to failure (Nf) for all mixtures calculated at 5 Hz Control SBS PCR dry PCR wet DVR Temp ‫׀‬E*‫׀‬ ‫׀‬E*‫׀‬ ‫׀‬E*‫׀‬ ‫׀‬E*‫׀‬ ‫׀‬E*‫׀‬ Microstrains Nf Nf Nf Nf Nf (Celsius) (Mpa) (Mpa) (Mpa) (Mpa) (Mpa) 10 100 8,238 113,450 7,285 1,214,100 5,736 1,121,500 8,249 397,140 9,350 478,410 10 150 8,238 19,785 7,285 149,770 5,736 174,210 8,249 49,593 9,350 42,850 10 200 8,238 5,730 7,285 33,931 5,736 46,482 8,249 11,333 9,350 7,736 10 250 8,238 2,192 7,285 10,726 5,736 16,681 8,249 3,607 9,350 2,050 10 300 8,238 999 7,285 4,186 5,736 7,221 8,249 1,415 9,350 693 10 350 8,238 514 7,285 1,889 5,736 3,557 8,249 642 9,350 277 10 400 8,238 289 7,285 948 5,736 1,927 8,249 323 9,350 125 15 100 6,112 86,231 5,340 1,101,500 4,266 938,350 6,960 220,980 7,202 450,630 15 150 6,112 15,037 5,340 135,880 4,266 145,760 6,960 27,594 7,202 40,362 15 200 6,112 4,355 5,340 30,783 4,266 38,892 6,960 6,306 7,202 7,286 15 250 6,112 1,666 5,340 9,731 4,266 13,957 6,960 2,007 7,202 1,931 15 300 6,112 759 5,340 3,797 4,266 6,041 6,960 787 7,202 653 15 350 6,112 391 5,340 1,714 4,266 2,976 6,960 357 7,202 261 15 400 6,112 220 5,340 860 4,266 1,612 6,960 180 7,202 118 20 100 4,220 98,561 3,680 1,549,000 2,970 1,192,400 5,538 182,420 5,304 598,620 20 150 4,220 17,187 3,680 191,080 2,970 185,230 5,538 22,780 5,304 53,617 20 200 4,220 4,978 3,680 43,289 2,970 49,423 5,538 5,206 5,304 9,679 20 250 4,220 1,904 3,680 13,684 2,970 17,736 5,538 1,657 5,304 2,566 20 300 4,220 868 3,680 5,340 2,970 7,677 5,538 650 5,304 867 20 350 4,220 447 3,680 2,410 2,970 3,782 5,538 295 5,304 346 20 400 4,220 251 3,680 1,210 2,970 2,048 5,538 149 5,304 157 25 100 2,737 162,830 2,420 3,132,800 1,964 2,153,400 4,154 223,640 3,758 1,083,900 25 150 2,737 28,394 2,420 386,450 1,964 334,510 4,154 27,928 3,758 97,083 25 200 2,737 8,224 2,420 87,550 1,964 89,252 4,154 6,382 3,758 17,526 25 250 2,737 3,145 2,420 27,675 1,964 32,029 4,154 2,031 3,758 4,645 25 300 2,737 1,434 2,420 10,800 1,964 13,864 4,154 797 3,758 1,570 25 350 2,737 738 2,420 4,874 1,964 6,830 4,154 361 3,758 627 25 400 2,737 415 2,420 2,447 1,964 3,699 4,154 182 3,758 283 140 APPENDIX B Supplementary information Chapter 3 141 B1. Mechanistic-empirical pavement design input data B1.1 Asphalt binders Table B1 Asphalt binder content % by weight % by volume (Pb) (Vbeff) Control 5.48 11.47 SBS 5.83 11.50 PCR dry 5.67 13.38 PCR wet 5.83 11.37 DVR 5.70 11.22 Table B2 Measured |G*| of the asphalt binders used in all different mixtures Temperature Complex Shear Phase Shift (°F) Modulus (Pa) Angle (°) Control and PCR dry (PG 58-28) 136 3,885 64.1 147 3,024 64.9 158 1,733 65.5 169 1,022 65.9 SBS (PG 70-28) 57 2,665,800 60.7 75 510,505 65.7 93 101,016 67.3 111 21,800 67.0 129 6,641 67.3 147 2,341 68.7 165 903 70.9 PCR wet (PG 82-28) 57 4,816,200 57.6 75 1,072,200 61.6 93 133,240 63.7 111 40,192 61.7 129 13,766 59.7 147 5,372 57.3 165 2,229 55.8 142 Table B2 (cont’d) DVR (PG 70-28) 75 1,124,600 59.3 93 212,355 63.1 111 50,128 64.1 129 15,278 64.5 147 5,228 66.3 165 1,953 69.8 B1.2 Asphalt mixtures Table B3 Measured |E*| of the asphalt mixtures (psi) |E*| (psi) Frequency (Hz) Temperature [°F] 0.1 1 10 25 Control 39 728,012 1,189,163 1,731,398 1,959,797 68 152,250 363,087 727,673 916,255 104 15,926 41,911 127,794 193,566 130 3,589 9,644 28,340 43,675 SBS 39 667,054 1,070,198 1,566,086 1,785,486 68 153,101 329,415 635,414 796,420 104 21,751 49,482 131,292 189,889 130 6,211 15,136 38,740 56,230 PCR dry 39 524,335 846,444 1,221,335 1,374,698 68 117,915 263,861 515,481 642,767 104 14,519 38,224 106,788 155,007 130 4,294 11,340 31,318 46,642 PCR wet 39 837,736 1,226,125 1,674,262 1,798,311 68 271,892 512,240 852,733 946,692 113 35,806 77,765 193,130 238,835 130 23,694 47,525 109,978 144,412 DVR 14 1,956,967 2,498,621 2,878,012 - 40 1,019,956 1,427,721 1,871,988 - 70 274,891 493,832 824,311 - 100 45,552 103,737 234,688 - 130 17,409 35,276 92,054 - 143 Table B4 Creep compliance for all mixtures (1/psi) Creep compliance D(t) (1/psi) Temperature (˚F) -4 14 32 Time (sec) Control 1 3.93E-07 5.35E-07 9.03E-07 2 4.07E-07 5.74E-07 1.04E-06 5 4.33E-07 6.45E-07 1.25E-06 10 4.54E-07 7.12E-07 1.47E-06 20 4.79E-07 7.89E-07 1.77E-06 50 5.23E-07 9.29E-07 2.26E-06 100 5.61E-07 1.07E-06 2.81E-06 SBS 1 4.34E-07 6.29E-07 1.10E-06 2 4.60E-07 6.95E-07 1.26E-06 5 5.00E-07 8.02E-07 1.54E-06 10 5.41E-07 8.96E-07 1.84E-06 20 5.87E-07 1.02E-06 2.20E-06 50 6.59E-07 1.23E-06 2.83E-06 100 7.33E-07 1.43E-06 3.52E-06 PCR dry 1 5.64E-07 7.77E-07 1.32E-06 2 5.87E-07 8.37E-07 1.51E-06 5 6.26E-07 9.46E-07 1.81E-06 10 6.58E-07 1.04E-06 2.15E-06 20 6.99E-07 1.16E-06 2.56E-06 50 7.67E-07 1.38E-06 3.27E-06 100 8.24E-07 1.58E-06 4.06E-06 PCR wet 1 4.42E-07 5.98E-07 9.33E-07 2 4.63E-07 6.50E-07 1.04E-06 5 4.98E-07 7.27E-07 1.23E-06 10 5.31E-07 7.97E-07 1.42E-06 20 5.65E-07 8.90E-07 1.63E-06 50 6.23E-07 1.03E-06 2.02E-06 100 6.79E-07 1.16E-06 2.41E-06 144 Table B4 (cont’d) Temperature (˚F) Time (sec) -4 14 32 DVR 1 3.54E-07 4.86E-07 7.82E-07 2 3.68E-07 5.20E-07 8.78E-07 5 3.89E-07 5.72E-07 1.02E-06 10 4.09E-07 6.25E-07 1.16E-06 20 4.31E-07 6.84E-07 1.34E-06 50 4.65E-07 7.76E-07 1.62E-06 100 4.98E-07 8.70E-07 1.90E-06 B1.3 Traffic Table B5 Traffic input data Structure type Low traffic High Traffic Initial two-way AADTT: 590 7,664 Number of lanes in design direction 1 2 Percent of trucks in design direction (%) 51.0 51.0 Percent of trucks in design lane (%) 90.0 90.0 Operational speed (mph) 55 60 Table B6 Number of axles per truck Single Tandem Tridem Quad Class 4 1.65 0.36 0.00 0.00 Class 5 2.00 0.05 0.00 0.00 Class 6 1.00 1.00 0.00 0.00 Class 7 1.06 0.06 0.58 0.37 Class 8 2.28 0.74 0.00 0.00 Class 9 1.29 1.85 0.00 0.00 Class 10 1.54 1.00 0.33 0.55 Class 11 4.99 0.00 0.00 0.00 Class 12 3.85 0.96 0.00 0.00 Class 13 2.03 1.40 0.36 0.62 145 Table B7 Vehicle class distribution Low traffic structure High traffic structure Class % Growth (%) Growth Type Class % Growth (%) Growth Type Class 4 1.77 1.5 Compound 2.08 1.5 Compound Class 5 27.37 1.5 Compound 49.78 1.5 Compound Class 6 5.01 1.5 Compound 6.62 1.5 Compound Class 7 0.77 1.5 Compound 1.09 1.5 Compound Class 8 4.42 1.5 Compound 4.27 1.5 Compound Class 9 45.43 1.5 Compound 22.08 1.5 Compound Class 10 7.07 1.5 Compound 6.43 1.5 Compound Class 11 1.12 1.5 Compound 0.41 1.5 Compound Class 12 0.22 1.5 Compound 0.04 1.5 Compound Class 13 6.82 1.5 Compound 7.2 1.5 Compound 146 Table B8 Volume monthly adjustment factors Class 4 Class 5 Class 6 Class 7 Class 8 Class 9 Class 10 Class 11 Class 12 Class 13 Jan 0.8 0.8 0.8 0.8 0.91 0.91 0.91 0.88 0.88 0.88 Feb 0.89 0.89 0.89 0.89 0.95 0.95 0.95 0.89 0.89 0.89 Mar 0.88 0.88 0.88 0.88 0.98 0.98 0.98 0.88 0.88 0.88 Apr 0.93 0.93 0.93 0.93 1.01 1.01 1.01 0.96 0.96 0.96 May 1.02 1.02 1.02 1.02 1.06 1.06 1.06 1.05 1.05 1.05 Jun 1.14 1.14 1.14 1.14 1.12 1.12 1.12 1.16 1.16 1.16 Jul 1.18 1.18 1.18 1.18 0.98 0.98 0.98 1.07 1.07 1.07 Aug 1.19 1.19 1.19 1.19 1.08 1.08 1.08 1.1 1.1 1.1 Sep 1.13 1.13 1.13 1.13 1.03 1.03 1.03 1.07 1.07 1.07 Oct 1.06 1.06 1.06 1.06 1.05 1.05 1.05 1.11 1.11 1.11 Nov 0.96 0.96 0.96 0.96 0.96 0.96 0.96 1 1 1 Dec 0.82 0.82 0.82 0.82 0.87 0.87 0.87 0.83 0.83 0.83 B1.4 Climate Table B9 Climate stations input Climate Station Cities US, MI US, ID US, FL US, CA Location, latit long 45.50000 43.50000 28.50000 36.00000 elevation (ft) -85.00000 856 -114.37500 5697 -81.25000 108 -119.37500 223 Mean annual air 43.94 41.15 71.88 66.50 temperature (ºF) Mean annual precipitation 41.03 17.88 45.24 15.27 (in) Freezing index 1054.65 1428.00 0.43 0.06 (ºF - days) Average annual number of 57.59 120.89 2.73 2.84 freeze/thaw cycles 147 B2. Life Cycle Assessment B2.1 Life cycle inventory B2.1.1 Electricity mix Table B10 Electricity generated in the USA in 2020 [147, 148] kWh 1 kWh Coal 8.88E+11 0.228 Petroleum 1.56E+10 0.004 Natural Gas 1.44E+12 0.368 Nuclear 7.93E+11 0.203 Hydropower 2.90E+11 0.075 Geothermal 1.63E+10 0.004 Biogenic Municipal Waste 1.83E+10 0.005 Wood and Other Biomass 1.32E+10 0.003 Solar Thermal 3.58E+10 0.001 Solar Photovoltaic 8.97E+10 0.023 Wind 3.36E+11 0.086 B2.1.2 Aggregates and RAP Figure B1 shows the results of the sieve analysis performed in the MSU asphalt laboratory on aggregates used for the mixtures. The curve is close to the control points and avoids the restricted zone [210]. Table B11 summarizes data used for sands and gravels with size between 0.075 and 6.30 mm, and crushed stones with size between 6.30 and 12.50 mm. Based on the mix design, 70% of the virgin aggregates was composed of sands and 30% by crushed stones. Data regarded energy consumption needed for the quarries’ activities, including blasting, wet drilling in unfragmented rock, product loading in open truck, unloading primary and crushing, screening, conveyor, transfer, and storage piles. Transportation between quarry and asphalt plant, where aggregates are usually stored, was also included, based on the commodity flow survey for the United States [115]. The RAP is usually stored in asphalt mixing plants with other aggregates after 148 milling operations of old pavements. Data shown in Table B12 regards diesel consumption of a milling machine having an hourly production of 350 tons/hour. 100.0 90.0 80.0 70.0 60.0 % Passing 50.0 40.0 30.0 20.0 Blend Cont. Pts. 10.0 Restricted Max Den. 0.0 0.075 0.6 1.18 2.36 4.75 6.3 9.5 12.5 19 25.4 Sieve Size mm Figure B1. Blend gradation of aggregates based on the Superpave Aggregate Gradation Specification. Table B11 Energy consumption to produce 1 t of gravel and sand. [121] Gravel, sand Crushed stone Unit process (unit) 0.075-6.30 mm 6.30-12.50 mm Gasoline, combusted in equipment/US S (gal) 5.43E-3 9.4E-3 Diesel, burned in building machine/GLO US (Btu) 9,681 15,087 Natural gas, combusted in industrial boiler NREL/US (ft3) 1.33 3.45 Electricity, at grid, 2020/US (kWh) 2.41 2.96 Electricity, hard coal, at power plant/US US-EI U (kWh) - 4E-6 Transport, lorry 16-32t, EURO4/US- US-EI (tkm) 50 50 Table B12 Energy and Transport used to produce 1 ton of Reclaimed Asphalt Pavement [123] Unit process (unit) Diesel, burned in building machine {GLO} (Btu) (1) 56,397 Transport, lorry 16-32t, EURO4/US- US-EI (tkm) 50 Disposal, inert waste, 5% water, to inert material landfill/US (ton) -1 B2.1.3 Crude oil and bitumen production Based on data reported in the Energy Information Administration (EIA) website, in 2016, the production of crude oil in the United States was 3,235,353 thousand barrels, of which 7% was exported [211]. The total imported crude oil was 2,873,086 thousand barrels [212]. Table B13 149 reports the origin of the foreign crude oil (98% of the total imported crude oil considered) and the cities chosen as a port entry in this study. Based on the processes available in SimaPro 9.1, the crude oil imported from different countries was considered as shown in Table B14. Table B15 shows the approximate distances between the origin country and corresponding Petroleum Administration for Defense Districts (PADDs). Table B16 summarizes the transport covered by transoceanic tankers and pipelines. Table B17 reports data on the domestic crude oil transported between PADDs (only 36% of the total) in 2016 by pipeline, barge tanker, and train [213–215]. Table B18 summarizes the materials and energy used to refine 1 ton of crude oil [126]. Table B13 Imported crude oil in the USA in 2016 (Unit: Thousand barrels). [212] PADD4 PADD1 PADD2 PADD3 (Rocky PADD5 (East Coast) (Midwest) (Gulf Coast) Mountain) (West Coast) Philadelphia Peoria Huston Rocky Los Angeles PA IL TX Mountain WY CA Canada 77,372 788,347 133,324 96,485 85,420 Saudi Arabia 21,512 14,251 257,654 - 108,701 Venezuela 13,753 - 256,729 - 884 Mexico 4,941 510 206,928 - 700 Colombia 29,976 - 85,936 - 45,689 Iraq 19,229 - 120,248 - 14,054 Ecuador - - 12,762 - 74,097 Kuwait - - 48,517 - 27,942 Nigeria 65,998 - 9,119 - 690 Angola 28,232 - 19,195 - 10,706 Brazil 5,274 - 38,438 - 9,224 Chad 10,509 - 13,878 - - Algeria 10,646 - 7,869 - - Russia - - 2,545 - 11,427 Norway 11,764 - 1,224 - - Oman - - 15 - 10,935 150 Table B14 Unit processes used for 1 ton of imported crude oil, based on the databases available in SimaPro (Unit:tons) Unit process Quantity Note Crude oil, at production NREL/RNA 0.699 Canada, Mexico, Ecuador, Venezuela, Colombia, Brazil Crude oil, at production onshore/RME US-EI U 0.228 Saudi Arabia, Oman, Iraq, Kuwait Crude oil, at production onshore/RAF US-EI U 0.036 Algeria, Angola, Ciad Crude oil, at production/NG US-EI U 0.027 Nigeria Crude oil, at production offshore/NO US-EI 0.005 Norway Crude oil, at production onshore/RU US-EI U 0.005 Russia Table B15 Distances for imported crude oil in the USA in 2016 (Unit: km) PADD4 PADD1 PADD2 PADD3 (Rocky PADD5 (East Coast) (Midwest) (Gulf Coast) Mountain) (West Coast) Philadelphia Peoria Huston Rocky Los Angeles PA IL TX Mountain WY CA Canada 2,090 1,930 2,000 1,450 1,930 Saudi 10,600 11,450 12,700 - 13,350 Arabia Venezuela 3,830 - 3,980 - 6,120 Mexico 3,270 2,000 1,220 - 2,450 Colombia 3,940 - 3,570 - 5,590 Iraq 9,750 - 11,800 - 12,300 Ecuador - - 3,960 - 5,787 Kuwait - - 12,370 - 12,800 Nigeria 8,850 - 10,850 - 12,670 Angola 11,074 - 12,870 - 14,920 Brazil 6,500 - 6,760 - 8,830 Chad 9,054 - 11,135 - - Algeria 7,000 - 9,126 - - Russia - - 9,700 - 8,700 Norway 5,900 - 7,870 - - Oman - - 13,560 - 13,800 Table B16 Transportation of 1 ton of imported crude oil. Unit process (unit) Transport, crude oil pipeline, onshore/RER U (tkm) 3,948 Operation, transoceanic tanker/OCE US-EI U (tkm) 198,36 151 Table B17 Quantities of domestic crude oil transported between PADDs and related distances covered by pipeline, barge tanker, and rail in 2016. [213–215] Unit process (unit) Transport, crude oil pipeline, onshore/US- US-EI (tkm) 4427.01 Operation, barge tanker/US- US-EI (tkm) 107.61 Transport, train, diesel powered NREL/US (tkm) 615.21 Table B18 Materials and energy used to refine1 ton of crude oil. [126] Unit Process (units) Natural gas, sweet, burned in production flare (m3) 0.013 Natural gas, sour, burned in production flare (m3) 0.002 Electricity, at grid, 2020/US (kWh) 0.105 Steam, in chemical industry, production (kg) 0.124 Natural gas, burned in boiler modulating >100kW/US (MJ) 2.200 Diesel, burned in diesel-electric generating set (MJ) 0.028 Bituminous coal, combusted in industrial boiler NREL/US (kg) 2.4E-5 Residual fuel oil, combusted in industrial boiler NREL/US (l) 0.001 Refinery gas, burned in furnace (MJ) 0.360 Liquefied petroleum gas, combusted in industrial boiler NREL/US (l) 0.009 B2.1.4 Asphalt modification The styrene-butadiene-styrene (SBS) used to modify the bitumen was 3.5% by weight of the total asphalt binder. The modification was carried out in the asphalt laboratory by using high and low shear mixers in two consecutive steps. SBS pellets were first milled into the hot base binder (Performance Grade 58-28) at 163°C by using a mixer at 5,000 revolutions per minute (RPM) for approximately 30 minutes. During the low shear mixing phase, the SBS-binder compound was mix at 1000 rpm for 120 minutes at a temperature ranging between 170°C and 180°C. A sulfur catalyst solution was also used as a cross-linker with a ratio of 0.4% by weight of the total binder after 90 minutes. The modification of the bitumen with the DVR and SBS flowed the same steps. 7% DVR and 2% SBS (by weight of the total asphalt binder) were added to the hot neat bitumen before starting the low shear mixer, and the cross-linker after 90 minutes from the starting of the high shear mix. The modification for the PCR wet mixture followed the same steps and range of temperatures as before, but the modifying agents were added as following: 7% PCR 152 and 0.5% SBS (by weight of the total asphalt binder) were added to the hot neat bitumen before starting the low shear mixer, 3% of Sasobit after 80 minutes from the start of the high shear mixing and the 0.4% of cross-linker after 90 minutes, as for the other modifications. According to the Eurobitumen report [125], the energy consumption spent for modification is 72 MJ per ton of modified bitumen. B2.1.5 Synthetic polymer Table B19 Input material for 1 kg of SBS [129] Unit process (unit) Transport, lorry 16-32t, EURO4/US- US-EI (kgkm) 950 Polybutadiene E (kg) 0.34 Polystyrene, general purpose| production (kg) 0.66 Electricity, at grid, 2020/US (MJ) 0.072 Table B20 Input material for 1 ton of SBR [130] Unit process (unit) Acrylonitrile-butadiene-styrene copolymer resin, at plant NREL/RNA (ton) 1 Transport, lorry 16-32t, EURO4/US- US-EI (tkm) 950 Table B21 Input material for 1 ton of Sasobit [131] Unit process (unit) Fatty acid {USA}| market for | Cut-off, U 1 Transport, lorry 16-32t, EURO4/US- US-EI (tkm) 3,000 B2.1.6 Crumb rubber from scrap tires Table B22 Materials and energy used to produce 1 ton of crumb rubber. [9] Unit Process Input Scrap tires for ground rubber (ton) 1.45 Steel blades (kg) 0.9 Tap water, tap water production, conventional treatment 67 (kg) Electricity, at grid, 2020/US (kWh) 513 Avoided products Cast iron| production (ton) 0.23 Petroleum coke| petroleum refinery operation (ton) 0.16 Emissions to air Particulates, unspecified (kg) 0.09 153 Table B23 Material input and operation to obtain 1 ton of steel for blades used to shred tires (Unit: ton). Unit process Steel, unalloyed| steel production, converter, unalloyed 1 Hot rolling, steel| processing 1 Sheet rolling, steel| processing 1 Table B24 Materials and energy used to produce 1 ton of DVR Unit Process Crumb rubber (kg) 1000 Ethylene vinyl acetate copolymer (USA) market for ethylene vinyl acetate 20 copolymer | APOS, U Sulfuric acid (USA)| production | APOS, U 20 Acrylonitrile-butadiene-styrene copolymer resin, at plant NREL/RNA U 20 System Electricity, at grid, 2020/US (kWh) 429 B2.1.7 Hot mix asphalt The energy spent in the asphalt plant to produce 1 ton of hot asphalt mixture was 317 MJ, divided into electricity and energy mix [132]. Based on lower heating values available in the reference, Table B25 shows the quantities of fuel supposed in this study. Table B25 Electricity and energy mix to produce 1 ton of hot mix asphalt. Unit process (unit) Diesel produced and combusted, at industrial boiler/US (l) 0.45 Gasoline, combusted in equipment NREL/US (l) 0.46 Liquefied petroleum gas, combusted in industrial boiler NREL/US (l) 0.48 Natural gas, combusted in industrial boiler NREL/US (l) 0.6 Bituminous coal, combusted in industrial boiler NREL/US (kg) 0.69 Residual fuel oil, combusted in industrial boiler NREL/US (l) 0.43 Crude oil, at production NREL/RNA (kg) 0.45 Hydrogen (reformer) E (kg) 1.5 Kerosene, at refinery/US- US-EI (kg) 0.46 Electricity, at grid, 2020/US (kWh) 3.32 154 B2.2 Cradle-to-gate results Table B26 Cumulative energy demand (CED) of 1 ton of Control and rubberized asphalt mixtures compared to the SBS blend Cut-off Economic allocation System expansion System exp-NG CED MJ/ton Control 1,956 (-11.1%) SBS 2,2201 PCR dry 2,052 (-6.8%) 2,056 (-6.6%) 2,262 (-2.8%) 2,170 (-1.4%) PCR wet 2,003 (-9.0%) 2,005 (-8.9%) 2,151 (-2.3%) 2,086 (-5.2%) DVR 1,697 (-22.9%) 1,699 (-22.8%) 1,833 (-16.7%) 1,774 (-19.4%) Figure B2. Cumulative energy demand comparison of PCR dry, PCR wet, and DVR mixtures to Control and SBS mixes based on the allocation methods B2.3 Uncertainty 155 Table B27 Statistical data from Monte Carlo simulation Global warming potential kg CO2 eq/ton PCR dry PCR dry PCR dry PCR dry Control SBS Cut off Econ all Sys Exp Sys Exp NG Mean 57.25 65.73 60.14 60.22 62.79 56.53 Median 57.22 65.71 60.10 60.17 62.73 56.45 SD 0.212 0.252 0.279 0.360 0.862 0.776 CV 0.370 0.383 0.465 0.598 1.373 1.373 2.5% 56.92 65.36 59.71 59.79 61.25 55.30 97.5% 57.76 66.34 60.82 60.87 64.45 58.27 SEM 0.007 0.008 0.009 0.011 0.027 0.025 Fossil fuel depletion MJ surplus/ton Mean 257.85 289.58 270.21 270.72 299.72 287.83 Median 257.79 289.52 270.12 270.62 298.99 287.68 SD 0.6344 0.7460 0.8624 0.9934 6.5938 2.4953 CV 0.2460 0.2576 0.3192 0.3670 2.2000 0.8670 2.5% 256.79 288.38 268.80 269.32 288.56 283.14 97.5% 259.22 291.14 272.13 272.62 314.52 292.89 SEM 0.020 0.024 0.027 0.031 0.209 0.079 Values with 42.20% 40.80% 37.40% 37.40% 49.90% 49.90% uncertainty data 156 Table B27 (cont’d) Global warming potential kg CO2 eq/ton PCR wet PCR wet PCR wet PCR wet DVR DVR DVR DVR Cut off Econ all Sys Exp Sys Exp NG Cut off Econ all Sys Exp Sys Exp NG Mean 63.56 63.61 65.38 61.00 56.14 56.21 57.90 53.80 Median 63.51 63.58 65.34 60.91 56.13 56.19 57.84 53.75 SD 0.695 0.651 0.886 0.823 0.144 0.158 0.645 0.458 CV 1.094 1.023 1.356 1.349 0.257 0.282 1.114 0.851 2.5% 62.35 62.40 63.71 59.57 55.92 56.00 56.88 53.04 97.5% 64.98 64.97 67.23 62.72 56.50 56.54 59.02 54.85 SEM 0.022 0.021 0.028 0.026 0.005 0.005 0.020 0.014 Fossil fuel depletion MJ surplus/ton Mean 256.19 256.51 277.36 268.74 218.15 218.52 237.54 229.71 Median 256.11 256.43 276.90 268.81 218.12 218.47 236.98 229.67 SD 0.890 0.847 4.775 1.891 0.411 0.446 4.530 1.459 CV 0.348 0.330 1.722 0.704 0.189 0.204 1.907 0.635 2.5% 254.73 255.17 269.21 265.13 217.42 217.82 230.29 227.00 97.5% 257.93 258.19 288.13 272.65 219.05 219.56 247.59 232.80 SEM 0.028 0.027 0.151 0.060 0.013 0.014 0.143 0.046 Values with uncertainty 58.40% 58.40% 59.9% 59.9% 69.20% 69.20% 69.50% 69.50% data 157 APPENDIX C Supplementary information Chapter 4 158 C1 Leaching results Table C1 Eluate concentrations, on average of the three replicates As Cd Cr Cu Pb Zn μg/L day Control 0.083 0.13 0.19 1.58 74.49 1.51 105.08 1 0.04 0.09 1.45 3.73 0.44 13.67 2 0.02 0.00 1.56 0.79 0.00 16.35 7 0.02 0.34 2.82 0.93 0.18 35.35 14 0.04 0.50 2.02 2.37 2.62 40.67 28 0.03 0.22 0.00 0.26 1.88 30.67 42 0.11 0.03 0.42 0.00 0.12 103.23 49 0.05 0.31 0.53 0.01 0.06 63.79 63 0.09 0.03 0.13 0.17 0.13 93.04 SBS 0.083 0.08 0.10 1.75 17.47 0.52 53.70 1 0.05 0.08 1.66 6.43 0.11 31.36 2 0.01 0.04 1.30 0.86 0.48 7.91 7 0.01 311.02 3.09 13.88 0.00 45.72 14 0.03 0.43 1.57 0.87 4.48 60.62 28 0.04 0.23 0.06 0.67 1.15 94.45 42 0.07 0.05 0.35 0.69 0.14 64.24 49 0.02 0.10 0.29 -0.27 0.05 56.53 63 0.06 0.04 0.14 -0.03 0.32 144.68 PCR dry 0.083 0.11 0.83 2.19 14.38 0.54 28.61 1 0.09 0.08 2.02 4.20 0.71 90.54 2 0.03 0.00 1.58 0.66 0.33 26.73 7 0.02 0.00 0.73 0.00 0.00 30.62 14 0.03 0.04 0.61 0.49 0.69 39.45 28 0.05 0.20 0.00 0.25 1.71 108.2 42 0.03 0.02 0.28 0.36 0.10 107.7 49 0.03 0.01 0.37 0.09 0.00 137.9 63 0.04 0.02 0.12 0.34 0.23 98.90 159 Table C1(cont’d) As Cd Cr Cu Pb Zn μg/L day PCR wet 0.083 0.13 0.17 2.01 11.34 0.86 54.01 1 0.05 0.07 1.72 1.71 0.67 64.05 2 0.04 0.00 1.76 9.50 0.30 43.70 7 0.00 0.18 0.34 0.00 0.00 8.94 14 0.01 30.55 0.58 0.24 15.27 68.74 28 0.03 0.16 0.00 0.54 1.65 59.87 42 0.03 0.01 0.32 0.46 0.06 53.18 49 0.03 0.01 0.29 0.03 0.01 88.62 63 0.04 0.01 0.47 0.00 0.04 112.04 DVR 0.083 0.12 0.57 2.29 15.40 0.98 55.20 1 0.08 0.04 1.83 1.46 0.12 24.12 2 0.12 0.06 2.41 12.30 0.34 54.10 7 0.04 7.03 1.82 0.97 0.00 113.8 14 0.06 0.31 0.95 0.26 4.15 57.71 28 0.05 0.13 0.53 0.60 1.25 55.49 42 0.05 0.01 0.46 0.04 0.09 83.23 49 0.05 0.07 0.14 0.00 0.21 327.5 63 0.04 0.01 0.43 0.00 0.03 79.18 Table C2 Interval flux mass release, on average of three replicates As Cd Cr Cu Pb Zn 2 μg/m s day Control 0.083 0.002 0.002 0.020 0.934 0.019 1.318 1 0.000 0.000 0.002 0.004 0.000 0.015 2 0.000 0.000 0.002 0.001 0.000 0.018 7 0.000 0.000 0.001 0.000 0.000 0.007 14 0.000 0.000 0.000 0.000 0.000 0.006 28 0.000 0.000 0.000 0.000 0.000 0.002 42 0.000 0.000 0.000 0.000 0.000 0.008 49 0.000 0.000 0.000 0.000 0.000 0.010 63 0.000 0.000 0.000 0.000 0.000 0.007 160 Table C2 (cont’d) As Cd Cr Cu Pb Zn 2 μg/m s day SBS 0.083 0.001 0.001 0.022 0.219 0.006 0.674 1 0.000 0.000 0.002 0.007 0.000 0.034 2 0.000 0.000 0.001 0.001 0.001 0.009 7 0.000 0.065 0.001 0.003 0.000 0.010 14 0.000 0.000 0.000 0.000 0.001 0.009 28 0.000 0.000 0.000 0.000 0.000 0.007 42 0.000 0.000 0.000 0.000 0.000 0.005 49 0.000 0.000 0.000 0.000 0.000 0.008 63 0.000 0.000 0.000 0.000 0.000 0.011 PCR dry 0.083 0.001 0.010 0.027 0.180 0.007 0.359 1 0.000 0.000 0.002 0.005 0.001 0.099 2 0.000 0.000 0.002 0.001 0.000 0.029 7 0.000 0.000 0.000 0.000 0.000 0.006 14 0.000 0.000 0.000 0.000 0.000 0.006 28 0.000 0.000 0.000 0.000 0.000 0.008 42 0.000 0.000 0.000 0.000 0.000 0.008 49 0.000 0.000 0.000 0.000 0.000 0.021 63 0.000 0.000 0.000 0.000 0.000 0.007 PCR wet 0.083 0.002 0.002 0.025 0.142 0.011 0.678 1 0.000 0.000 0.002 0.002 0.001 0.070 2 0.000 0.000 0.002 0.010 0.000 0.048 7 0.000 0.000 0.000 0.000 0.000 0.002 14 0.000 0.005 0.000 0.000 0.002 0.010 28 0.000 0.000 0.000 0.000 0.000 0.004 42 0.000 0.000 0.000 0.000 0.000 0.004 49 0.000 0.000 0.000 0.000 0.000 0.013 63 0.000 0.000 0.000 0.000 0.000 0.008 DVR 0.083 0.001 0.007 0.029 0.193 0.012 0.692 1 0.000 0.000 0.002 0.002 0.000 0.026 2 0.000 0.000 0.003 0.013 0.000 0.059 7 0.000 0.001 0.000 0.000 0.000 0.024 14 0.000 0.000 0.000 0.000 0.001 0.009 28 0.000 0.000 0.000 0.000 0.000 0.004 42 0.000 0.000 0.000 0.000 0.000 0.006 49 0.000 0.000 0.000 0.000 0.000 0.049 63 0.000 0.000 0.000 0.000 0.000 0.006 161 Table C3 Cumulative mass release, on average of three replicates As Cd Cr Cu Pb Zn 2 μg/m day Control 0.083 11.42 17.20 142.35 6,727.82 136.59 9,490.16 1 15.09 25.08 272.87 7,064.57 176.21 10,724.66 2 16.78 25.08 413.85 7,136.28 176.21 12,201.32 7 19.03 56.02 668.13 7,220.36 192.11 15,394.09 14 22.56 100.82 850.65 7,434.31 428.94 19,067.22 28 25.25 120.45 850.65 7,457.46 599.05 21,837.01 42 35.24 123.41 888.46 7,457.46 609.55 31,160.41 49 39.45 151.56 936.27 7,458.17 615.20 36,921.24 63 47.73 153.87 948.31 7,473.69 626.81 45,323.73 SBS 0.083 6.80 8.64 157.83 1,578.02 46.69 4,849.88 1 11.11 16.21 307.54 2,159.17 56.68 7,681.75 2 12.15 19.97 425.05 2,236.80 100.37 8,395.97 7 13.03 28,109.36 704.02 3,490.43 100.37 12,524.82 14 15.34 28,147.94 846.20 3,569.20 505.16 17,999.57 28 18.69 28,169.07 851.64 3,629.91 609.44 26,530.03 42 25.25 28,173.64 883.14 3,691.84 622.30 32,331.68 49 27.39 28,182.51 909.10 3,667.09 626.96 37,436.97 63 32.37 28,185.86 921.50 3,664.21 655.86 50,503.70 PCR dry 0.083 10.03 75.32 197.93 1,298.87 49.07 2,583.80 1 17.78 82.32 380.34 1,678.26 113.11 10,760.44 2 20.33 82.32 523.29 1,737.63 142.57 13,174.94 7 22.06 82.32 588.78 1,737.63 142.57 15,939.89 14 24.87 86.08 643.95 1,781.73 204.72 19,502.90 28 29.01 103.88 643.95 1,803.98 359.24 29,271.32 42 31.86 105.94 669.01 1,836.63 368.57 38,996.89 49 34.71 107.23 702.61 1,844.89 368.49 51,446.89 63 38.04 109.39 713.70 1,875.48 388.90 60,378.52 PCR wet 0.083 11.63 15.07 181.59 1,023.84 77.83 4,878.04 1 16.49 21.04 337.14 1,178.19 138.36 10,662.62 2 19.87 21.28 496.12 2,036.16 165.42 14,609.58 7 19.87 37.23 526.47 2,036.16 165.42 15,416.92 14 20.92 2,796.23 578.93 2,057.49 1,544.69 21,624.87 28 23.20 2,810.58 578.93 2,105.87 1,694.00 27,031.92 42 25.56 2,811.73 608.08 2,147.10 1,699.03 31,834.37 49 28.13 2,812.96 634.32 2,149.47 1,699.88 39,838.00 63 31.93 2,813.64 676.73 2,149.47 1,703.87 49,956.79 162 Table C3 (cont’d) As Cd Cr Cu Pb Zn 2 μg/m day DVR 0.083 10.41 51.56 206.36 1,391.03 88.32 4,985.69 1 17.34 55.22 371.23 1,522.84 99.09 7,164.01 2 28.20 60.77 588.72 2,634.10 130.09 12,049.93 7 32.14 695.80 752.88 2,721.82 130.09 22,331.76 14 37.48 723.79 838.35 2,745.15 504.62 27,543.70 28 42.06 735.55 885.83 2,799.69 617.26 32,554.93 42 46.78 736.64 927.38 2,803.54 625.76 40,071.72 49 51.05 743.08 940.12 2,803.54 644.38 69,650.07 63 55.11 743.76 978.80 2,803.54 647.25 76,800.89 C2 Digestion procedure C2.1 Aggregates and RAP Aggregates and RAP were pulverized before proceeding with the digestion (Figure C1). The digestion was performed on 0.15 g of aggregates, in a single phase, by adding 2 mL of HNO3 65%, 3 mL HCl 32%, and 1 mL HF 40%. Figure C1. Aggregates preparation for the microwave digestion: a) batch of aggregates used in Michigan, b) crusher, and c) pulverized aggragates sample ready for the digestion 163 Table C4 Steps for the microwave digestion of the aggregates Step Time (min) Power (W) 1 5 250 2 5 400 3 10 500 4 5 ventilation The RAP was digested by using 2 mL HNO3 65%, 2 mL HF 40%, 1 mL H2O2 30%, in a single phase. Table C5 Steps for the microwave digestion of the RAP Step Time (min) Power (W) 1 6 250 2 6 400 3 6 650 4 6 250 5 5 ventilation The PCR, DVR, SBS, Sasobit, and the neat bitumen were digested in two phases, on 0.15 g of each material, by using 3 mL H2SO4 96%. Table C6 Steps for the microwave digestion of PCR, DVR, SBS, Sasobit, and neat bitumen Step Time (min) Power (W) 1 1 250 2 1 0 3 4 250 4 4 400 5 4 600 6 5 ventilation 164 Table C7 Metals content in each material used in asphalt mixtures (mg/kg) Aggregates RAP[1] Bitumen SBS[2] Sasobit PCR[3] DVR[4] Michigan Maryland North Carolina Wisconsin mg/kg Al 7,273 15,450 24,147 14,098 8,875 < DL 6.54 < DL 895.32 232.35 As 25.79 89.38 46.84 73.70 129.39 < 17 < 17 < 17 < DL < 17 Ba 1,675 1,782 799 2,469 3,670 1.01 < DL 1.25 11.27 14.07 Ca 16,684 6,872 2,557 4,819 4,563 < 60 < 60 < 60 839.50 638.87 Cd 3.29 3.40 3.56 4.52 5.69 < DL < DL < DL < DL 0.29 Co - 0.00 0.00 0.00 - < DL < DL < DL 199.22 2.86 Cr 50.28 68.62 51.35 32.50 36.61 1.49 1.52 1.39 6.37 4.27 Cu 47.72 31.36 44.74 23.36 22.15 < DL < DL < DL 570.46 3.93 Fe 34,659 59,001 14,431 11,579 8,952 20.59 19.20 14.59 3,315 136.35 K 8,428 11,172 23,036 26,699 10,064 < DL < DL < DL 1,332 298.16 Mg 7,439 2,054 148 2,061 3,484 < DL 28.72 < DL 251.49 190.85 Mn 696.45 963.17 295.40 209.83 408.04 0.49 0.23 0.19 21.45 3.18 Na 7,422 8,926 13,262 7,733 7,746 < DL < DL < DL 151.57 83.25 Ni 43.31 81.73 28.32 28.92 32.81 75.72 0.74 1.27 3.13 2.20 Pb 87.80 87.00 116.83 105.72 108.65 < 1,5 < 1,5 < 1,5 2.18 < 1,5 Sb 100.42 92.46 106.41 116.20 121.54 < DL < DL < DL 39.14 < DL Sn < DL 0.00 11.90 8.02 11.18 1.17 < DL < DL < DL 3.42 Ti 2,155 11,863 1,344 807 1,282 3.14 1.34 0.50 61.92 408.19 Zn 34.49 111.01 43.28 32.33 79.33 < DL < DL < DL 10,936 8,287 Note: [1] Reclaimed Asphalt Pavement; [2] Styrene-Butadiene-Styrene; [3] Polymer Coated Rubber; [4] Devulcanized Rubber 165 Table C8 Metals content in Control mixtures (three replicates, mg/sample) Control Aggregates RAP Bitumen Replicate 1 2 3 1 2 3 1 2 3 [mg] Ca 33,130 32,999 33,156 1,687 1,680 1,688 - - - Ba 3,326 3,313 3,328 1,357 1,352 1,358 0.117 0.116 0.117 Zn 68.49 68.22 68.55 29.33 29.22 29.36 - - - Fe 68,821 68,549 68,874 3,310 3,297 3,312 2.392 2.382 2.394 Mn 1,383 1,377 1,384 150.9 150.3 151.0 0.057 0.057 0.057 Cr 99.84 99.44 99.91 13.54 13.48 13.55 0.173 0.172 0.173 Mg 14,772 14,713 14,783 1,288 1,283 1,289 - - - Cu 94.75 94.38 94.83 8.190 8.157 8.196 - - - Na 14,739 14,680 14,750 2,864 2,853 2,866 - - - K 16,735 16,669 16,748 3,721 3,706 3,724 - - - As 51.21 51.01 51.25 47.84 47.65 47.88 - - - Pb 174.3 173.6 174.5 40.18 40.02 40.21 - - - Cd 6.526 6.500 6.531 2.105 2.096 2.106 - - - Sb 199.4 198.6 199.5 44.94 44.76 44.98 - - - Ni 86.00 85.66 86.07 12.13 12.08 12.14 8.796 8.761 8.803 Co - - - - - - - - - Sn - - - 4.133 4.117 4.136 0.136 0.135 0.136 Ti 4,280 4,263 4,283 474.0 472.1 474.4 0.364 0.363 0.365 Al 14,442 14,384 14,453 3,282 3,269 3,284 - - - 166 Table C9 Metals content in SBS mixtures (three replicates, mg/sample) SBS Aggregates RAP Bitumen SBS Replicate 1 2 3 1 2 3 1 2 3 1 2 3 [mg] Ca 33,258 33,152 33,093 1,693 1,688 1,685 - - - - - - Ba 3,339 3,328 3,322 1,362 1,357 1,355 0.122 0.121 0.121 - - - Zn 68.76 68.54 68.42 29.43 29.34 29.29 - - - - - - Fe 69,086 68,867 68,745 3,321 3,311 3,305 2.492 2.484 2.480 0.081 0.081 0.081 Mn 1,388 1,384 1,381 151.4 150.9 150.6 0.060 0.059 0.059 Cr 100.2 99.90 99.73 13.58 13.54 13.51 0.180 0.179 0.179 0.006 0.006 0.006 Mg 14,829 14,781 14,755 1,293 1,289 1,286 - - - 0.122 0.121 0.121 Cu 95.12 94.82 94.65 8.218 8.192 8.178 - - - - - - Na 14,795 14,748 14,722 2,874 2,865 2,860 - - - - - - K 16,800 16,746 16,717 3,734 3,722 3,715 - - - - - - As 51.41 51.24 51.15 48.01 47.85 47.77 - - - - - - Pb 175.0 174.4 174.1 40.31 40.19 40.11 - - - - - - Cd 6.551 6.530 6.519 2.112 2.105 2.101 - - - - - - Sb 200.2 199.5 199.2 45.10 44.95 44.87 - - - - - - Ni 86.33 86.06 85.91 12.17 12.13 12.11 9.163 9.134 9.118 - - - Co - - - - - - - - - - - - Sn - - - 4.148 4.134 4.127 0.141 0.141 0.141 - - - Ti 4,296 4,282 4,275 475.6 474.1 473.3 0.380 0.379 0.378 0.006 0.006 0.006 Al 14,497 14,451 14,425 3,293 3,283 3,277 - - - 0.028 0.028 0.028 167 Table C10 Metals content in PCR dry mixtures (three replicates, mg/sample) PCR dry Aggregates RAP Bitumen PCR Replicate 1 2 3 1 2 3 1 2 3 1 2 3 [mg] Ca 32,732 32,476 32,571 1,668 1,655 1,660 - - - 10.33 10.24 10.28 Ba 3,286 3,260 3,270 1,342 1,331 1,335 0.121 0.120 0.120 0.139 0.137 0.138 Zn 67.67 67.14 67.34 29.00 28.77 28.86 - - - 134.5 133.5 133.8 Fe 67,995 67,462 67,661 3,272 3,247 3,256 2.477 2.457 2.465 40.78 40.46 40.58 Mn 1,366 1,356 1,360 149.2 148.0 148.4 0.059 0.059 0.059 0.264 0.262 0.263 Cr 98.64 97.87 98.15 13.38 13.28 13.32 0.179 0.177 0.178 0.078 0.078 0.078 Mg 14,594 14,480 14,523 1,274 1,264 1,267 - - - 3.093 3.069 3.078 Cu 93.62 92.88 93.16 8.10 8.03 8.06 - - - 7.017 6.962 6.982 Na 14,562 14,447 14,490 2,832 2,809 2,818 - - - 1.864 1.850 1.855 K 16,534 16,405 16,453 3,679 3,650 3,661 - - - 16.39 16.26 16.31 As 50.60 50.20 50.35 47.30 46.93 47.06 - - - - - - Pb 172.24 170.89 171.39 39.72 39.41 39.52 - - - 0.027 0.027 0.027 Cd 6.45 6.40 6.42 2.08 2.06 2.07 - - - - - - Sb 197.00 195.46 196.03 44.43 44.08 44.21 - - - 0.481 0.478 0.479 Ni 84.97 84.31 84.55 11.99 11.90 11.93 9.109 9.037 9.064 0.038 0.038 0.038 Co - - - - - - - - - 2.450 2.431 2.438 Sn - - - 4.086 4.054 4.066 0.141 0.139 0.140 - - - Ti 4,228 4,195 4,207 468.6 464.9 466.3 0.377 0.374 0.376 0.762 0.756 0.758 Al 14,268 14,156 14,198 3,244 3,219 3,229 - - - 11.01 10.93 10.96 168 Table C11 Metals content in PCR wet mixtures (three replicates, mg/sample) PCR wet Aggregates RAP Bitumen Replicate 1 2 3 1 2 3 1 2 3 [mg] Ca 33,124 33,188 33,064 1,687 1,691 1,684 - - - Ba 3,325 3,332 3,319 1,357 1,360 1,355 0.109 0.109 0.109 Zn 68.48 68.61 68.36 29.34 29.40 29.29 - - - Fe 68,809 68,942 68,684 3,311 3,317 3,305 2.237 2.241 2.233 Mn 1,383 1,385 1,380 150.9 151.2 150.6 0.054 0.054 0.053 Cr 99.8 100.0 99.6 13.54 13.56 13.51 0.161 0.162 0.161 Mg 14,769 14,798 14,742 1,289 1,291 1,286 - - - Cu 94.7 94.9 94.6 8.192 8.207 8.177 - - - Na 14,736 14,764 14,709 2,865 2,870 2,859 - - - K 16,732 16,765 16,702 3,722 3,729 3,715 - - - As 51.20 51.30 51.11 47.85 47.94 47.76 - - - Pb 174.3 174.6 174.0 40.18 40.26 40.11 - - - Cd 6.525 6.538 6.513 2.105 2.109 2.101 - - - Sb 199.4 199.7 199.0 44.95 45.04 44.87 - - - Ni 85.99 86.15 85.83 12.13 12.16 12.11 8.226 8.242 8.211 Co - - - - - - 0.000 0.000 0.000 Sn - - - 4.134 4.142 4.127 0.127 0.127 0.127 Ti 4,279 4,287 4,271 474.1 475.0 473.2 0.341 0.342 0.340 Al 14,439 14,467 14,413 3,282 3,289 3,276 - - - 169 Table C11 (cont’d) PCR wet SBS PCR Sasobit Replicate 1 2 3 1 2 3 1 2 3 [mg] Ca - - - 7.288 7.302 7.275 - - - Ba - - - 0.098 0.098 0.098 0.005 0.005 0.005 Zn - - - 94.94 95.12 94.77 - - - Fe 0.071 0.072 0.071 28.78 28.84 28.73 0.054 0.054 0.054 Mn 0.001 0.001 0.001 0.186 0.187 0.186 0.001 0.001 0.001 Cr 0.006 0.006 0.006 0.055 0.055 0.055 0.005 0.005 0.005 Mg 0.107 0.107 0.107 2.183 2.187 2.179 - - - Cu - - - 4.952 4.962 4.943 - - - Na - - - 1.316 1.318 1.313 - - - K - - - 11.57 11.59 11.55 - - - As - - - - - - - - - Pb - - - 0.019 0.019 0.019 - - - Cd - - - - - - - - - Sb - - - 0.340 0.340 0.339 - - - Ni 0.003 0.003 0.003 0.027 0.027 0.027 0.005 0.005 0.005 Co - - - 1.729 1.733 1.726 - - - Sn - - - - - - - - - Ti 0.005 0.005 0.005 0.538 0.539 0.537 0.002 0.002 0.002 Al 0.024 0.024 0.024 7.773 7.788 7.759 - - - 170 Table C12 Metals content in DVR mixtures (three replicates, mg/sample) DVR Aggregates RAP Bitumen Replicate 1 2 3 1 2 3 1 2 3 [mg] Ca 33,379 33,597 33,528 1,700 1,712 1,708 - - - Ba 3,351 3,373 3,366 1,368 1,377 1,374 0.112 0.113 0.112 Zn 69.0 69.5 69.3 29.56 29.76 29.70 - - - Fe 69,339 69,792 69,648 3,336 3,358 3,351 2.292 2.307 2.302 Mn 1,393 1,402 1,400 152.1 153.1 152.7 0.055 0.055 0.055 Cr 100.6 101.2 101.0 13.64 13.73 13.70 0.165 0.167 0.166 Mg 14,883 14,980 14,949 1,298 1,307 1,304 - - - Cu 95.47 96.09 95.89 8.255 8.308 8.291 - - - Na 14,849 14,946 14,915 2,887 2,906 2,900 - - - K 16,861 16,971 16,936 3,750 3,775 3,767 - - - As 51.60 51.93 51.83 48.22 48.53 48.43 - - - Pb 175.6 176.8 176.4 40.5 40.8 40.7 - - - Cd 6.575 6.618 6.604 2.121 2.135 2.131 - - - Sb 200.9 202.2 201.8 45.30 45.59 45.50 - - - Ni 86.65 87.22 87.04 12.23 12.31 12.28 8.429 8.485 8.467 Co - - - - - - - - - Sn - - - 4.166 4.193 4.184 0.130 0.131 0.131 Ti 4,312 4,340 4,331 477.7 480.9 479.9 0.349 0.352 0.351 Al 14,550 14,645 14,615 3,308 3,329 3,322 - - - 171 Table C12 (cont’d) DVR SBS DVR Replicate 1 2 3 1 2 3 [mg] Ca - - - 5.422 5.457 5.446 Ba - - - 0.119 0.120 0.120 Zn - - - 70.33 70.79 70.64 Fe 0.047 0.047 0.047 1.157 1.165 1.162 Mn 0.001 0.001 0.001 0.027 0.027 0.027 Cr 0.004 0.004 0.004 0.036 0.036 0.036 Mg 0.070 0.071 0.071 1.620 1.630 1.627 Cu - - - 0.033 0.034 0.034 Na - - - 0.707 0.711 0.710 K - - - 2.530 2.547 2.542 As - - - - - - Pb - - - - - - Cd - - - 0.002 0.002 0.002 Sb - - - - - - Ni 0.002 0.002 0.002 0.019 0.019 0.019 Co - - - 0.024 0.024 0.024 Sn - - - 0.029 0.029 0.029 Ti 0.003 0.003 0.003 3.464 3.487 3.480 Al 0.016 0.016 0.016 1.972 1.985 1.981 172 APPENDIX D Supplementary information Chapter 5 173 D1 Process used in LCA Table D1 Processes used in Simapro Process Description Concrete, normal, at plant/US* Included processes: includes the whole manufacturing US-EI U processes to produce ready-mixed concrete, internal processes (transport, etc.) and infrastructure. No administration is included. Special outputs: wastewater, average data of 11 German concrete plants Remark: Part of total Swiss concrete production: 55%. Density: 2'380 kg/m3. Ingredients: Cement 300 kg, Water 190 kg, Gravel 1'890 kg. Wastewater is an average from data on 11 German concrete mixing plant. Gravel, unspecified, at Included processes: includes the whole manufacturing mine/US* US-EI U process, internal processes (transport, etc.) and infrastructure. No administration is included. Recultivation is taken into account. Remark: Mix of round and crushed gravel: 79% round gravel and 21% crushed gravel. Steel, converter, low-alloyed, at Included processes: Transports of hot metal and other input plant/US- US-EI U materials to converter, steel making process and casting. Remark: This process produces primary steel. Scrap is only used for cooling the liquid steel. Steel, converter, unalloyed, at Included processes: Transports of hot metal and other input plant/US- US-EI U materials to converter, steel making process and casting. Remark: This process produces primary steel. Scrap is only used for cooling the liquid steel. Steel, converter, chromium steel Included processes: Transports of hot metal and other input 18/8, at plant/US- US-EI U materials to converter, steel making process and casting. Remark: This process produces primary steel. Scrap is only used for cooling the liquid steel. Reinforcing steel, at plant/US- Included processes: Mix of differently produced steels and US-EI U hot rolling. Remark: represents Average of World and European production mix. Cast iron, at plant/US- US-EI U Included processes: Transports of metal and other input materials to electric arc furnace, melting and refining process and casting. Remark: 35% scrap and 65% pig iron assumed as iron input; Technology: Electric arc furnace for melting. 174 Table D1 (cont’d) Process Description Copper {RNA}| production, This activity represents the production of 1 kg of refined primary | U copper. It is used as pure metal or as alloying element in various technical applications. The module includes the pre- treatment of the ore, the reduction and the refining copper to cathode copper, including the country-specific share in hydrometalurgically won copper (SX-EW). The inventory is modelled for Northern America. Aluminum ingot, production Included processes: Mixing process - calls US LCI existing mix, at plant NREL/US U processes Epoxy resin, liquid, at Included processes: Aggregated data for all processes from plant/US- US-EI U raw material extraction until delivery at plant. Technology: Production from epichlorohydrin and bisphenol-A. Polyethylene, HDPE, granulate, Included processes: Aggregated data for all processes from at plant/US- US-EI U raw material extraction until delivery at plant. Glass fibre reinforced plastic, Included processes: Gate to gate inventory for the injection polyamide, injection moulding, moulding of glass fibre with polyamide resin including at plant/US- US-EI U material inputs, process and infrastructure. Transport, barge/US- US-EI U Included processes: The module calls the modules addressing: operation of vessel; production of vessel; construction and land use of port; operation, maintenance and disposal of port. Remark: Inventory refers to the entire transport life cycle. Transport, lorry 16-32t, Included processes: operation of vehicle; production, EURO3/US- US-EI U maintenance and disposal of vehicles; construction and maintenance and disposal of road. Remark: Inventory refers to the entire transport life cycle Table D2 Transportation of wind turbines components Onshore [1] Offshore [2] Distance Mode of Distance Mode of [km] transportation [km] transportation Tower 1,100 truck 500 truck 1,500 boat Nacelle 1,025 truck 200 truck 1,500 boat Blades 600 truck 700 boat Foundation 50 truck 500 boat Note: [1] Garrett and Rønde, 2012); [2] Weinzettel et al., 2009. 175 D2 Environmental impact Table D3 Normalized values for each material for global warming potential and cumulative energy demand Global warming Cumulative energy potential demand ton CO2 eq/ton GJ/ton Concrete 0.120 0.647 Gravel 0.010 0.084 Steel reinforcing 1.590 23.49 Steel unalloyed 1.694 31.25 Steel - low alloy 2.185 31.25 Steel - chromium 5.647 85.78 Cast iron 1.505 24.97 Copper 5.311 72.82 Aluminum 8.013 96.76 Fiberglass 8.961 147.9 PE/Polymer 2.008 77.31 Epoxy 6.940 135.9 Neodymium 100.0 1,894 Dysprosium 132.0 613.4 176 Figure D1. Global warming potential (GWP) of onshore and offshore wind turbines in terms of ton CO2 eq./MW: a,b) critical and non-critical materials, c,d) non-critical materials contribution broken down into each material, and e,f) components and transportation contribution. 177 Figure D2. Cumulative Energy Demand (CED) of onshore and offshore wind turbines in terms of GJ/MW: a,b) critical and non-critical materials, c,d) non-critical materials contribution broken down into each material, and e,f) components and transportation contribution. 178 D3 Future wind capacity Table D4 Installed capacity and corresponding percentage with respect to the total generated electricity for each outlook 2020 2030 2050 Scenarios Reference Region GW [%] GW [%] GW [%] DNVGL, 2020 North America 169 8 387 25 830 48 Canada energy Canada 15 - 21 - 27 - regulator, 2020 DNVGL Calculated, USA 154 366 803 DNVGL Wind U.S. Department USA 113 10 224 20 409 35 Vision of Energy, 2020c EIA EIA 2020 USA 121 8 173 13 205 14 DNVGL DNVGL, 2020 Rest of the 591 4 133 10 4144 26 world Table D5 Capacity addition projected for onshore and offshore wind energy ONSHORE OFFSHORE USA World USA World Wind Wind DNVGL EIA DNVGL DNVGL EIA DNVGL Vision Vision GW/yr GW/yr 2018 7.06 7.6 7.04 37.5 - - - 4.47 2019 19.2 12.1 7.04 37.1 4.46 - - 0.89 2020 18.2 14.6 7.04 48.1 4.47 - 3.05 1.11 2021 16.0 3.66 9.19 51.2 3.37 - 1.87 1.52 2022 13.2 15.2 9.19 50.7 2.48 0.42 1.87 2.04 2023 14.0 17.4 9.19 53.9 2.42 0.40 1.87 2.83 2024 13.7 1.87 9.19 57.2 3.14 - 1.87 3.66 2025 12.9 0.46 9.19 60.8 4.30 - 1.87 4.46 2026 14.3 0.44 9.19 69.4 5.63 - 1.87 5.45 2027 15.5 0.26 9.19 78.9 7.09 - 1.87 6.48 2028 16.0 0.87 9.19 83.2 8.72 2.00 1.87 7.51 2029 18.0 1.54 9.19 89.7 10.5 - 1.87 9.11 2030 17.9 0.03 9.19 92.3 12.2 7.10 1.87 10.8 2031 18.7 0.15 5.80 94.9 13.8 - 3.21 12.8 2032 18.2 0.12 5.80 95.7 15.1 - 3.21 15.1 2033 17.4 0.43 5.80 96.1 16.1 - 3.21 17.8 2034 15.8 0.61 5.80 97.4 16.6 - 3.21 21.0 2035 14.3 0.32 5.80 99.1 16.8 8.20 3.21 24.9 2036 12.5 1.06 5.80 99.4 16.7 - 3.21 29.6 2037 10.8 0.42 5.80 97.4 16.3 - 3.21 35.1 2038 8.81 1.35 5.80 93.6 15.7 - 3.21 41.4 2039 7.46 - 5.80 88.5 15.2 - 3.21 48.5 2040 6.14 1.11 5.80 83.4 14.6 - 3.21 55.7 179 Table D5 (cont’d) ONSHORE OFFSHORE USA World USA World Wind Wind DNVGL EIA DNVGL DNVGL EIA DNVGL Vision Vision GW/yr GW/yr 2041 4.75 1.26 5.80 79.6 14.2 - 3.21 61.9 2042 4.12 1.50 5.80 77.1 13.8 - 3.21 67.0 2043 3.39 1.72 5.80 74.7 13.4 - 3.21 71.4 2044 2.78 1.30 5.80 72.5 13.1 - 3.21 75.5 2045 2.25 1.79 5.80 71.5 12.9 - 3.21 79.6 2046 1.75 1.37 5.80 71.9 12.7 - 3.21 83.8 2047 1.25 1.86 5.80 73.1 12.5 - 3.21 87.8 2048 0.78 2.14 5.80 74.4 12.4 - 3.21 92.0 2049 0.39 2.47 5.80 75.5 12.3 - 3.21 96.6 2050 0.12 2.84 5.80 76.6 12.1 - 3.21 102 D4 Future material production Table D6 Future steel, cement, and rare earth elements production STEEL [1] CEMENT [2] REE [3] United United World World China States States Thousand Thousand tons Thousand tons tons 2018 86,607 1,730,004 87,000 3,963,000 138.5 2019 87,961 1,757,035 87,326 3,977,861 138.4 2020 89,314 1,784,066 87,653 3,992,723 138.0 2021 90,667 1,811,098 87,979 4,007,584 137.3 2022 92,020 1,838,129 88,305 4,022,445 136.4 2023 93,374 1,865,160 88,631 4,037,306 135.2 2024 94,727 1,892,191 88,958 4,052,168 133.7 2025 96,080 1,919,223 89,284 4,067,029 132.1 2026 97,433 1,946,254 89,610 4,081,890 130.2 2027 98,787 1,973,285 89,936 4,096,751 128.2 2028 100,140 2,000,317 90,263 4,111,613 126.0 2029 101,493 2,027,348 90,589 4,126,474 123.7 2030 102,846 2,054,379 90,915 4,141,335 121.3 2031 104,200 2,081,411 91,241 4,156,196 118.8 2032 105,553 2,108,442 91,568 4,171,058 116.2 2033 106,906 2,135,473 91,894 4,185,919 113.5 2034 108,259 2,162,504 92,220 4,200,780 110.7 2035 109,612 2,189,536 92,546 4,215,641 108.0 2036 110,966 2,216,567 92,873 4,230,503 105.2 2037 112,319 2,243,598 93,199 4,245,364 102.4 180 Table D7 (cont’d) STEEL [1] CEMENT [2] REE [3] United United World World China States States Thousand Thousand tons Thousand tons tons 2038 113,672 2,270,630 93,525 4,260,225 99.53 2039 115,025 2,297,661 93,851 4,275,086 96.70 2040 116,379 2,324,692 94,178 4,289,948 93.87 2041 117,732 2,351,724 94,504 4,304,809 91.07 2042 119,085 2,378,755 94,830 4,319,670 88.28 2043 120,438 2,405,786 95,156 4,334,531 85.52 2044 121,792 2,432,818 95,483 4,349,393 82.80 2045 123,145 2,459,849 95,809 4,364,254 80.11 2046 124,498 2,486,880 96,135 4,379,115 77.47 2047 125,851 2,513,911 96,461 4,393,976 74.87 2048 127,205 2,540,943 96,788 4,408,838 72.32 2049 128,558 2,567,974 97,114 4,423,699 69.82 2050 129,911 2,595,005 97,440 4,438,560 67.38 D5 Future material demand For each type of turbine, the future intensity for each material used in all components was calculated by (Eq. D1). W$%&'# ∙ W!,$ &'(()#$ (Eq. D1) I!,#,$ = W$%&'$,&'(()#$ Where, Im,n = intensity of the material m in ton/MW (concrete m1, gravel m2, steel and iron m3, copper m4, aluminum m5, fiberglass m6, PE/Polymer m7, Epoxy m8, REE m9) n = year (n = 2018-2050) t = type of turbine (t= DFIG, DDSG, PMSG, floating offshore) Wcompn = mass of the component (foundation, tower, nacelle, rotor blades) Wm,t current = current intensity of the material m for the type of turbine t (ton/MW). Wcompt,current = current mass of the component for the type of turbine t (ton/MW). 181 After calculating the future mass intensity, the future material requirement was calculated by multiplying the mass intensity of each material for each type of turbine by the annual capacity addition for the USA and the rest of the world until 2050 (Eq. D2). MR !,# = +,-.!,#,$ ∙ t %,# 01 ∙ C# (Eq. D2) Where, MRm,n = total requirement for the material m (concrete m1, gravel m2, steel and iron m3, copper m4, aluminum m5, fiberglass m6, PE/Polymer m7, Epoxy m8, REE m9) for the year n (n = 2018-2050) (ton). t%, n = percentage of the type of turbine constructed in in year n. Cn = capacity addition installed during year n (MW). 182 REFERENCES 183 REFERENCES [1] Council on Foreign Relations The State of U.S. Infrastructure. https://www.cfr.org/backgrounder/state-us-infrastructure [2] Fox TR (2016) Recycling wind turbine blade composite material as aggregate in concrete. Master thesis. Iowa State University [3] Adams J (2021) Digestion of Unconventional Materials to Make Asphalt Recycling Additives. 58th Petersen Asphalt Research Conference. Western Research Institute [4] National Asphalt Pavement Association (NAPA) NAPA FAST FACTS. https://www.asphaltpavement.org/uploads/documents/NAPA_Fast_Facts_Sept_2020_Fina l_Version.pdf [5] U.S. Tire Manufacturers Association (2020) U.S. Scrap Tire Management Summary. https://www.ustires.org/sites/default/files/2019 USTMA Scrap Tire Management Summary Report.pdf [6] Kearney AT (1990) Scrap Tire Management Council - Scrap Tire Use/Disposal Study. Final Report [7] Lo Presti D (2013) Recycled Tyre Rubber Modified Bitumens for road asphalt mixtures: A literature review. Constr Build Mater 49:863–881. https://doi.org/10.1016/j.conbuildmat.2013.09.007 [8] Zanetti MC, Fiore S, Ruffino B, et al (2015) Characterization of crumb rubber from end-of- life tyres for paving applications. Waste Manag 45:161–170. https://doi.org/10.1016/j.wasman.2015.05.003 [9] Feraldi R, Cashman S, Huff M, Raahauge L (2013) Comparative LCA of treatment options for US scrap tires: Material recycling and tire-derived fuel combustion. Int J Life Cycle Assess 18:613–625. https://doi.org/10.1007/s11367-012-0514-8 [10] United States Environmental Protection Agency (1991) Markets for Scrap Tires -EPA/530- SW-90-074A October 1991 [11] Kocak S, Kutay ME (2016) Use of crumb rubber in lieu of binder grade bumping for mixtures with high percentage of reclaimed asphalt pavement. Road Mater Pavement Des 0629:1–14. https://doi.org/10.1080/14680629.2016.1142466 [12] Santagata E, Baglieri O, Riviera PP, et al (2018) Influence of lateral confining pressure on flow number tests. Bear Capacit Roads, Railw Airfields - Proc 10th Int Conf Bear Capacit Roads, Railw Airfields, BCRRA 2017 237–242. https://doi.org/10.1201/9781315100333- 35 184 [13] Epps JA (1994) NCHRP 198. Uses of Recycled Rubber Tires in Highways [14] Artamendi I, Khalid HA (2006) Diffusion Kinetics of Bitumen into Waste Tyre Rubber (With Discussion). J Assoc Asph Paving Technol 75: [15] Abdelrahman M, Carpenter S. (1999) Mechanism of interaction of asphalt cement with crumb rubber modifier. Transp Res Rec J Transp Res Board 1661:106–113 [16] Jamrah A, Kutay ME, Varma S (2015) Backcalculation of swollen crumb rubber modulus in asphalt-rubber binder and its relation to performance. Transp Res Rec 2505:99–107 [17] Santagata E, Lanotte M, Baglieri O, Dalmazzo D (2016) Analysis of bitumen – crumb rubber affinity for the formulation of rubberized dry mixtures. Mater Struct 49:1947–1954. https://doi.org/10.1617/s11527-015-0625-3 [18] Hanson DI, Foo KY, Brown ER, Denson R (1994) Evaluation and characterization of rubber-modified hot mix asphalt pavement. Transp Res Rec 1436:98–107 [19] Heitzman MA (1992) FHWA-SA-92-022 Report - State of the practice - Design and construction of asphalt paving materials with crumb rubber modifier. Washington D.C., USA [20] Epps JA (1994) NCHRP Report - Synthesis of Highway Practice 198 - Uses of recycled rubber tyres in highways. Washington D.C., USA [21] Peralta J, Silva HMRD, Hilliou L, et al (2012) Mutual changes in bitumen and rubber related to the production of asphalt rubber binders.Constr Build Mater 36:557–565. https://doi.org/10.1016/j.conbuildmat.2012.06.030 [22] Santagata E, Lanotte M, Dalmazzo D, Zanetti MC (2013) Potential performance-related properties of rubberized bituminous mixtures produced with dry technology. Liverpool [23] Buncher MS (1995) Evaluating the effects of the wet and dry processes for including crumb rubber modifier in hot mix asphalt. Auburn, AL, USA [24] Amirkhanian SN (2001) Utilization of crumb rubber in asphaltic concrete mixtures – South Carolina’s Experience. Clemson University, Clemson, SC, U.S.A. [25] State of California Department of Transportation (Caltrans) (2005) Synthesis of Caltrans rubberized asphalt concrete projects. Sacramento, CA, U.S.A. [26] Farina A, Kutay ME, Lanotte M (2021) Laboratory and field performance investigation of pre-swollen crumb rubber modified asphalt mixtures. Int J Pavement Res Technol 14:513– 518. https://doi.org/10.1007/s42947-020-0191-0 [27] Garcia PS, de Sousa FDB, de Lima JA, et al (2015) Devulcanization of ground tire rubber: 185 Physical and chemical changes after different microwave exposure times. Express Polym Lett 9:1015–1026. https://doi.org/10.3144/expresspolymlett.2015.91 [28] Kurgan G. and Dongre R. (2015) International Trends in Low-Carbon/Low-Energy Pavement Construction. Workshop on the Adoption of Innovative Technologies and Materials for Road Construction in India [29] Brantley AS, Townsend TG (1999) Leaching of Pollutants from Reclaimed Asphalt Pavement. Environ Eng Sci. https://doi.org/10.1089/ees.1999.16.105 [30] Shedivy R, Meier A (2012) Leaching Characteristics of Recycled Asphalt Pavement Used as Unbound Road Base. Student Project Report [31] Duin PA (2020) A Comparison of Zinc Levels in Runoff From Rubberized Hot Mix Aspahlt and Conventional Pavement [32] Rhodes EP, Ren Z, Mays DC (2012) Zinc Leaching from Tire Crumb Rubber. https://doi.org/10.1021/es3024379 [33] California Department of Transportation -Caltrans- (2008) Water Quality and Toxicity Evaluation of Discharge Generated from Asphalt Pavement Surfacing Materials [34] California Stormwater Quality Association (CASQA) (2014) Zinc Sources in California Urban Runoff [35] Murphy LU, Cochrane TA, O’Sullivan A (2015) The Influence of Different Pavement Surfaces on Atmospheric Copper, Lead, Zinc, and Suspended Solids Attenuation and Wash- Off. Water Air Soil Pollut 226:. https://doi.org/10.1007/s11270-015-2487-2 [36] U.S. Environmental Protection Agency (EPA) (1992) SW-846 Toxicity Characteristic Leaching Procedure. Test Method 1311 [37] U.S. Environmental Protection Agency (EPA) (1994) SW-846 Synthetic Precipitation Leaching Procedure. Test method 1312 [38] U.S. Environmental Protection Agency (EPA) (2017) Leaching Environmental Assessment Framework (LEAF) Methods and Guidance. https://www.epa.gov/hw-sw846/leaching- environmental-assessment-framework-leaf-methods-and-guidance [39] Kosson D, van der Sloot HA, Garrabrants A, Seignette. P (2014) Leaching Test Relationships, Laboratory-to-Field Comparisons and Recommendations for Leaching Evaluation using the Leaching Environmental Assessment Framework (LEAF). Cincinnati [40] Engelsen CJ, Van Der Sloot HA, Wibetoe G, et al (2010) Leaching characterisation and geochemical modelling of minor and trace elements released from recycled concrete aggregates. Cem Concr Res 40:1639–1649. 186 https://doi.org/10.1016/j.cemconres.2010.08.001 [41] Engelsen CJ, Wibetoe G, van der Sloot HA, et al (2012) Field site leaching from recycled concrete aggregates applied as sub-base material in road construction. Sci Total Environ 427–428:86–97. https://doi.org/10.1016/j.scitotenv.2012.04.021 [42] Carter CM, van der Sloot HA, Cooling D (2009) pH dependent extraction of soils and soil amendments to understand the factors controlling element mobility - New approach to assess soil and soil amendments. Eur J Soil Sci 622–637 [43] Fang W, Qi G, Wei Y, et al (2018) Leaching characteristic of toxic trace elements in soils amended by sewage sludge compost: A comparison of field and laboratory investigations. Environ. Pollut. 237:244–252 [44] Fang W, Delapp RC, Kosson DS, et al (2017) Release of heavy metals during long-term land application of sewage sludge compost: Percolation leaching tests with repeated additions of compost. Chemosphere 169:271–280 [45] U.S. Department of Energy (2015) Wind Vision: A New Era for Wind Power in the United States [46] U.S. Department of Energy (2020) The Inside of a Wind Turbine. https://www.energy.gov/eere/wind/inside-wind-turbine [47] Schreiber A, Marx J, Zapp P (2019) Comparative life cycle assessment of electricity generation by different wind turbine types. J Clean Prod 233:561–572. https://doi.org/10.1016/j.jclepro.2019.06.058 [48] Venås C (2015) Life cycle assessment of electric power generation by wind turbines containing rare earth magnets [49] Vázquez Hernández C, Telsnig T, Pradas Villalba A (2017) JRC Wind Energy Status Report - 2016 Edition : Market, technology and regulatory aspects of wind energy. 62. https://doi.org/10.2760/332535 [50] IRENA International Renewable Energy Agency (2019) Technology , Grid Integration and Socio-Economic Aspects [51] Lucas J, Lucas P, Le Mercier T, et al (2015) Permanent Magnets Based on Rare Earths [52] Cao Z, O’Sullivan C, Tan J, et al (2019) Resourcing the Fairytale Country with Wind Power: A Dynamic Material Flow Analysis. Environ Sci Technol. https://doi.org/10.1021/acs.est.9b03765 [53] Koltun P, Tharumarajah A (2014) Life Cycle Impact of Rare Earth Elements. ISRN Metall 2014:1–10. https://doi.org/10.1155/2014/907536 187 [54] USGS (2020) Rare Earths Statistics and Information. https://www.usgs.gov/centers/nmic/rare-earths-statistics-and-information [55] Blengini GA, Latunussa C EL, Eynard U, Matos C (2020) Study on the EU’s list of Critical Raw Materials (2020) Final Report [56] Schrijvers D, Hool A, Blengini GA, et al (2020) A review of methods and data to determine raw material criticality. Resour Conserv Recycl 155:104617. https://doi.org/10.1016/j.resconrec.2019.104617 [57] Peck D, Kandachar P, Tempelman E (2015) Critical materials from a product design perspective. Mater Des 65:147–159. https://doi.org/10.1016/j.matdes.2014.08.042 [58] U.S. Department of Energy (2020) Critical Minerals and Materials. 50 [59] Lu X, Isacsson U (1997) Rheological characterization of styrene-butadiene-styrene copolymer modified bitumens. Constr Build Mater 11:23–32. https://doi.org/10.1016/S0950-0618(96)00033-5 [60] ASTM D6114 (2020) Standard Specification for Asphalt-Rubber Binder [61] Irfan M, Ali Y, Ahmed S, Hafeez I (2018) Performance Evaluation of Crumb Rubber- Modified Asphalt Mixtures Based on Laboratory and Field Investigations. 1795–1806. https://doi.org/10.1007/s13369-017-2729-2 [62] Dias JLF, Picado-santos LG, Capitão SD (2014) Mechanical performance of dry process fine crumb rubber asphalt mixtures placed on the Portuguese road network. 73:247–254 [63] Huang Y, Bird RN, Heidrich O (2007) A review of the use of recycled solid waste materials in asphalt pavements. Resour Conserv Recycl. https://doi.org/10.1016/j.resconrec.2007.02.002 [64] Khalid H, Artamendi I (2006) Post-consumer tyre rubber as a performance-enhancing additive in asphalt. Asphalt Professional; 2006 [65] Shirini B, Imaninasab R (2016) Performance evaluation of rubberized and SBS modified porous asphalt mixtures. 107:165–171 [66] Kocak S, Kutay ME (2016) Combined Effect of SBS and Devulcanized Rubber (DVR) Modification on Performance Grade and Fatigue Cracking Resistance of Asphalt Binders. 269–274. https://doi.org/10.1007/978-94-024-0867-6 [67] EC JRC (2010) ILCD Handbook: General Guide for Life Cycle Assessment – Detailed Guidance. European Commission, Joint Research Centre, Institute for Environment and Sustainability, Luxembourg. 188 [68] ISO 14040 (2006) Environmental management-Life cycle assessment-Principles and Framework. International Organization for Standardization, Geneva, Switzerland. [69] ISO 14044 (2006) Environmental management -- Life cycle assessment -- Requirements and guidelines. International Organization for Standardization, Geneva, Switzerland. [70] IPCC (2013) Guidelines for National Greenhouse Gas Inventories [71] Frischknecht R, Jungbluth N (2003) Implementation of Life Cycle Impact Assessment Methods. Final report ecoinvent 2000 [72] Goedkoop M, Heijungs R, Huijbregts M, et al (2009) ReCiPe 2008. Potentials 1–44. https://doi.org/10.029/2003JD004283 [73] Ministry of Housing Spatial Planning and Environment (2000) Eco-indicator 99 Manual for Designers [74] Chiu C-T, Hsu T-H, Yang W-F (2008) Life cycle assessment on using recycled materials for rehabilitating asphalt pavements. Resour Conserv Recycl. https://doi.org/10.1016/j.resconrec.2007.07.001 [75] Farina A, Zanetti MC, Santagata E, Blengini GA (2017) Life cycle assessment applied to bituminous mixtures containing recycled materials: Crumb rubber and reclaimed asphalt pavement. 117:204–212. https://doi.org/10.1016/j.resconrec.2016.10.015 [76] Bressi S, Santos J, Orešković M, Losa M (2019) A comparative environmental impact analysis of asphalt mixtures containing crumb rubber and reclaimed asphalt pavement using life cycle assessment. Int J Pavement Eng 0:1–15. https://doi.org/10.1080/10298436.2019.1623404 [77] Giunta M, Mistretta M, Pratico’ FG (2020) Environmental Sustainability and Energy Assessment of Bituminous Pavements Made with Unconventional Materials. pp 1–503 [78] Legret M, Odie L, Demare D, Jullien A (2005) Leaching of heavy metals and polycyclic aromatic hydrocarbons from reclaimed asphalt pavement. Water Res. https://doi.org/10.1016/j.watres.2005.06.017 [79] AFNOR XP X31-210 (1998). Waste leaching test. Afnor ed., Paris, France (in French) [80] Spreadbury CJ, Clavier KA, Lin AM, Townsend TG (2021) A critical analysis of leaching and environmental risk assessment for reclaimed asphalt pavement management. Sci Total Environ 775:145741. https://doi.org/10.1016/j.scitotenv.2021.145741 [81] Lim L, Walker R (2009) An Assessment of Chemical Leaching, Releases to Air and Temperature at Crumb-Rubber Infilled Synthetic Turf Fields. 189 [82] Azizian MF, Nelson PO, Thayumanavan P, Williamson KJ (2003) Environmental impact of highway construction and repair materials on surface and ground waters : Case study : crumb rubber asphalt concrete. Waste Manag 23:719–728. https://doi.org/10.1016/S0956- 053X(03)00024-2 [83] Garcia-Olivares A, Ballabera-Poy J, Garcia-Ladona E, Turiel A (2012) A global renewable mix with proven technologies and common materials. 41:561–574. https://doi.org/10.1016/j.enpol.2011.11.018 [84] Elshkaki A, Graedel TE, Ciacci L, Reck B (2016) Copper demand, supply, and associated energy use to 2050. Glob Environ Chang 39:305–315. https://doi.org/10.1016/j.gloenvcha.2016.06.006 [85] US Geological Survey (2009) Mineral Commodity Summaries. /http://%0Aminerals.usgs.gov/minerals/pubs/mcs/2009/mcs2009.pdf [86] USGS (2011) Wind Energy in the United States and Materials Required for the Land-Based Wind Turbine Industry From 2010 Through 2030 [87] Zimmermann T, Rehberger M, Gößling-Reisemann S (2013) Material flows resulting from large scale deployment of wind energy in Germany. Resources 2:303–334. https://doi.org/10.3390/resources2030303 [88] Sverdrup HU, Ragnarsdottir KV, Koca D (2015) Aluminium for the future: Modelling the global production, market supply, demand, price and long term development of the global reserves. Resour. Conserv. Recycl. 103:139–154 [89] Fishman T, Graedel TE (2019) Impact of the establishment of US offshore wind power on neodymium flows. Nat Sustain 2:332–338. https://doi.org/10.1038/s41893-019-0252-z [90] Li J, Peng K, Wang P, et al (2020) Critical Rare-Earth Elements Mismatch Global Wind- Power Ambitions. One Earth 3:116–125. https://doi.org/10.1016/j.oneear.2020.06.009 [91] Imholte DD, Nguyen RT, Vedantam A, et al (2018) An assessment of U.S. rare earth availability for supporting U.S. wind energy growth targets. Energy Policy 113:294–305. https://doi.org/10.1016/j.enpol.2017.11.001 [92] Kramer D (2010) Concern grows over China’s dominance of rare-earth metals. Phys Today 63:22–24. https://doi.org/10.1063/1.3431322 [93] Nassar NT, Wilburn DR, Goonan TG (2016) Byproduct metal requirements for U.S. wind and solar photovoltaic electricity generation up to the year 2040 under various Clean Power Plan scenarios. Appl Energy 183:1209–1226. https://doi.org/10.1016/j.apenergy.2016.08.062 [94] Watari T, McLellan BC, Giurco D, et al (2019) Total material requirement for the global 190 energy transition to 2050: A focus on transport and electricity. Resour Conserv Recycl 148:91–103. https://doi.org/10.1016/j.resconrec.2019.05.015 [95] Valero A, Valero A, Calvo G, et al (2018) An exergy flow analysis of the energy transition. ECOS 2018 - Proc 31st Int Conf Effic Cost, Optim Simul Environ Impact Energy Syst 159:1175–1184 [96] Michigan Department of Transportation (2012) Standard specifications for construction. [97] American Association of State Highway and TransportationOfficials (2018) AASTHO T 378-17. Standard Method of Test for Determining the Dynamic Modulus and Flow Number for Asphalt Mixtures Using the Asphalt Mixture Performance Tester ( AMPT ) Standard Method of Test for and Flow Number for Asphalt. [98] (2017) AASHTO R 84 - Practice For Developing Dynamic Modulus Master Curves For Asphalt Mixtures Using The Asphalt Mixture Performance Tester (Ampt). American Association of State Highway and Transportation Officials [99] ASTM D7313 - 13. Standard Test Method for Determining Fracture Energy of Asphalt- Aggregate Mixtures Using the Disk-Shaped Compact Tension Geometry [100] Asphalt Institute (2014) MS-2. Asphalt Mix Design Methods [101] Michigan Department of Transportation - online guideline (2019) HMA PRODUCTION MANUAL. https://www.michigan.gov/documents/mdot_HMA_ProductionManual_79005_7.pdf [102] Kutay ME, Lanotte M (2018) Viscoelastic continuum damage (VECD) models for cracking problems in asphalt mixtures. Int J Pavement Eng 19:231–242. https://doi.org/10.1080/10298436.2017.1279492 [103] Krcmarik M, Varma S, Emin Kutay M, Jamrah A (2016) Development of predictive models for low-temperature indirect tensile strength of asphalt mixtures. J Mater Civ Eng 28:1–10. https://doi.org/10.1061/(ASCE)MT.1943-5533.0001655 [104] NCHRP Project 1-37A (2004) Guide for Mechanistic-Empirical Design - online guide. http://onlinepubs.trb.org/onlinepubs/archive/mepdg/guide.htm [105] Miller JS, Bellinger WY (2003) Distress Identification Manual for the Long-Term Pavement Performance Program. Publ US Dep Transp Fed Highw Adm 129 [106] Federal Highway Administration U (2009) Pavement distress identification manual for the NPS Road Inventory. Cycle 4 1–25 [107] Papagiannakis AT, Masad EA (2008) Pavement Design and Materials. John Wiley & Sons, Inc., Hoboken, New Jersey 191 [108] Zaabar I, Chatti K Estimating Vehicle Operating Costs Caused by Pavement Surface Conditions. https://doi.org/10.3141/2455-08 [109] Zaabar I, Chatti K (2010) Calibration of HDM-4 models for estimating the effect of pavement roughness on fuel consumption for U. S. conditions. Transp Res Rec 105–116. https://doi.org/10.3141/2155-12 [110] Edition I (2021) Michigan DOT User Guide For Pavement Design [111] Lanotte M, Kutay ME, Haider SW, Musunuru GK (2020) New calibration approach to improve Pavement ME Design thermal cracking prediction: mixture-specific coefficients– the Michigan case study. Road Mater Pavement Des 21:1859–1871. https://doi.org/10.1080/14680629.2019.1570863 [112] American Center for Life Cycle Assessment (ACLCA) (2014) Environmental Life Cycle Assessment Measuring the Environmental Performance of Products. American Center for Life Cycle Assessment [113] Brander M, Tipper R, Hutchison C, Davis G (2008) Consequential and Attributional Approaches to LCA : a Guide to Policy Makers with Specific Reference to Greenhouse Gas LCA of Biofuels [114] Ekvall T, Azapagic A, Finnveden G, Rydberg T (2016) Attributional and consequential LCA in the ILCD handbook. 293–296. https://doi.org/10.1007/s11367-015-1026-0 [115] US Census Bureau (2012) 2007 Commodity Flow Survey [116] Environmental Protection Agency (2012) Tool for the Reduction and Assessment of Chemical and other Environmental Impacts (TRACI) - USER ’ S MANUAL [117] PRé Sustainability (2018) About SimaPro. In: https://simapro.com/about/ [118] Wernet G, Bauer C, Steubing B, et al (2016) The ecoinvent database version 3 (part I): overview and methodology. The International Journal of Life Cycle Assessment [119] Torcellini P, Deru M (2004) U.S. LCI Database Project Development Guidelines [120] LTS (2018) DATASMART LCI Package. http://ltsexperts.com/services/software/datasmart-life-cycle-inventory/ [121] Marceau ML, Nisbet MA, Vangeem MG (2007) Life Cycle Inventory of Portland Cement Concrete [122] National Asphalt Pavement Association (NAPA) (2016) Product Category Rules (PCR) for Asphalt Mixtures 192 [123] Mukherjee A, Dylla H (2017) Lessons learned in developing an environmental product declaration program for the asphalt industry in North America. In: Pavement Life-Cycle Assessment - Proceedings of the Pavement Life-cycle Assessment Symposium, 2017. pp 11–22 [124] ISO/TR 14049 (2012) Environmental management — Life cycle assessment — Illustrative examples on how to apply ISO 14044 to goal and scope definition and inventory analysis. ISO/TR 14049. Technical Report. [125] European Bitumen Association (2012) Life cycle inventory: Bitumen. Brussels, Belgium. [126] Yang RY (2014) Development of a pavement life cycle assessment tool utilizing regional data and introducing an asphalt binder model. University of Illinois at Urbana-Champaign [127] EIA (2019) Refinery&Blender Net Production. https://www.eia.gov/dnav/pet/pet_pnp_refp_dc_nus_mbbl_a.htm [128] EIA (2019) Open data. https://www.eia.gov/opendata/qb.php?category=714757 [129] Polymer Science Learning Center (2021) SBS rubber. https://pslc.ws/macrog/sbs.htm [130] Rasutis D (2014) Comparative Life Cycle Assessment of Conventional and Guayule Automobile Tires. Arizona State University [131] Santos J, Bressi S, Cerezo V, et al (2018) Life cycle assessment of low temperature asphalt mixtures for road pavement surfaces: A comparative analysis. Resour Conserv Recycl 138:283–297. https://doi.org/10.1016/j.resconrec.2018.07.012 [132] Mukherjee A (2016) Life Cycle Assessment of Asphalt Mixtures in Support of an Environmental Product Declaration [133] BCA Industries (2018) Scrap Tire Processing. http://www.profitablerecycling.com/tirerecycling.htm [134] U.S. Environmental Protection Agency (EPA) (2016) Wastes - Resource Conservation - Common Wastes & Materials - Scrap Tires. https://archive.epa.gov/epawaste/conserve/materials/tires/web/html/basic.html [135] World Cement Association WCA supports full decarbonisation of the cement industry. https://www.worldcementassociation.org/about-us/sustainability [136] Akhtar SS, Ervin E, Raza S, Abbas T (2016) From Coal to Natural Gas: Its Impact on Kiln Production, Clinker Quality, and Emissions. IEEE Trans Ind Appl 52:1913–1924. https://doi.org/10.1109/TIA.2015.2504554 [137] Shen L, Worrell E, Patel MK (2010) Open-loop recycling: A LCA case study of PET bottle- 193 to-fibre recycling. Resour Conserv Recycl 55:34–52. https://doi.org/10.1016/j.resconrec.2010.06.014 [138] Pelletier N, Tyedmers P (2011) An ecological economic critique of the use of market information in life cycle assessment research. J Ind Ecol 15:342 – 3 [139] Vogtländer J, Brezet H, Hendriks C (2001) Allocation in recycling systems an integrated model for the analyses of environmental im- pact and market value. Int J Life Cycle Assess 6:1 – 12 [140] Werner F, Althaus HJ, Richter K, Scholz R (2007) Post-consumer waste wood in attributive product LCA. Context specific evaluation of allocation procedures in a functionalistic conception of LCA. Int J Life Cycle Assess 12:160 – 1 [141] Guinée J (2002) Handbook on life cycle assessment: operational guide to the ISO standards. Springer Netherlands, Dordrecht [142] Guinée JB, Heijungs R, Huppes G (2004) Economic Allocation: Examples and Derived Decision Tree. Int J Life Cycle Assess 9:23–33. https://doi.org/10.1007/BF02978533 [143] Santero NJ, Masanet E, Horvath A (2011) Life-cycle assessment of pavements Part II: Filling the research gaps. Resour Conserv Recycl 55:810–818. https://doi.org/10.1016/j.resconrec.2011.03.009 [144] Johnson JX, Mcmillan CA, Keoleian GA (2013) Evaluation of life cycle assessment recycling allocation methods: The case study of aluminum Johnson et al. Evaluation of LCA allocation methods. J Ind Ecol 17:700–711. https://doi.org/10.1111/jiec.12050 [145] Curran MA (2012) Life Cycle Assessment Handbook - A Guide for Environmentally Sustainable Products [146] Chen C, Habert G, Bouzidi Y, et al (2010) LCA allocation procedure used as an incitative method for waste recycling : An application to mineral additions in concrete. Resour Conserv Recycl Elsevier 54:1231–1240. https://doi.org/10.1016/j.resconrec.2010.04.001 . hal-00595462 [147] EIA (2021) Annual Energy Outlook 2021 Table: Table 54. Electric Power Projections by Electricity Market Module Region. https://www.eia.gov/outlooks/aeo/data/browser/ [148] EIA (2021) Annual Energy Outlook 2021 Table: Table 16. Renewable Energy Generating Capacity and Generation. https://www.eia.gov/outlooks/aeo/data/browser/ [149] EIA (2021) Table C9. Electric Power Sector Consumption Estimates, 2019. https://www.eia.gov/state/seds/sep_sum/html/sum_btu_eu.html [150] Wirtgen Group Company (2021) Milling machine - Cold milling machine W220i. 194 https://www.wirtgen-group.com/binary/full/o4493v77_W_brochure_W220i_1216_US.pdf [151] Volvo (2021) Pavers - Volvo Construction Equipment. https://www.volvoce.com/- /media/volvoce/global/products/asphalt-pavers/abg-tracked- pavers/brochures/brochure_p4820d_t3_en_30_20050029_e.pdf?v=hHY9Pw [152] Volvo (2021) Compactors - Volvo Construction Equipment. https://www.volvoce.com/- /media/volvoce/global/products/compactors/soil- compactors/brochures/brochure_sd115b_t4i_en_a8_20046077_c.pdf?v=mvc_Pw [153] AASHTO (2018) T 166-16. Standard Method of Test for Bulk Specific Gravity of Compacted Asphalt Mixtures Using Saturated Surface-Dry Specimens. 3:1–5 [154] U.S. Environmental Protection Agency (EPA) (2004) SW-846 Test Method 9040C: pH Electrometric Measurement [155] U.S. Environmental Protection Agency (EPA) (1996) SW-846 Test Method 9050A: Specific Conductance [156] Lamble KJ, Hill SJ (1998) Microwave digestion procedures for environmental matrices. Analyst 123:. https://doi.org/10.1039/a800776d [157] Garrabrants AC, Kosson DS, Brown KG, et al (2021) Demonstration of the use of test results from the Leaching Environmental Assessment Framework (LEAF) to develop screening-level leaching assessments. Waste Manag 121:226–236. https://doi.org/10.1016/j.wasman.2020.12.016 [158] Garrabrants AC, Kosson DS, Brown KG, et al (2021) Methodology for scenario-based assessments and demonstration of treatment effectiveness using the Leaching Environmental Assessment Framework (LEAF). J Hazard Mater 406:. https://doi.org/10.1016/j.jhazmat.2020.124635 [159] U.S. Environmental Protection Agency (EPA) (2021) Metereological Data - Weather Stations, Center for Exposure Assessment and Modeling (CEAM). https://www.epa.gov/ceam/meteorological-data-weather-stations [160] Elshkaki A, Graedel TE (2014) Dysprosium , the balance problem , and wind power technology. Appl Energy 136:548–559. https://doi.org/10.1016/j.apenergy.2014.09.064 [161] Martinez E, Sanz F, Pellegrini S, et al (2009) Life cycle assessment of a multi-megawatt wind turbine. 34:667–673. https://doi.org/10.1016/j.renene.2008.05.020 [162] Savino MM, Manzini R, Della V, Accorsi R (2017) A new model for environmental and economic evaluation of renewable energy systems : The case of wind turbines. Appl Energy 189:739–752. https://doi.org/10.1016/j.apenergy.2016.11.124 [163] Oebels KB, Pacca S (2013) Life cycle assessment of an onshore wind farm located at the 195 northeastern coast of Brazil. Renew Energy 53:60–70. https://doi.org/10.1016/j.renene.2012.10.026 [164] Haapala K, Prempreeda P (2014) Comparative life cycle assessment of 2 . 0 MW wind turbines. 3:170–185 [165] Adibi N, Lafhaj Z, Yehya M, Payet J (2017) Global Resource Indicator for life cycle impact assessment : Applied in wind turbine case study. 165:1517–1528 [166] Ghenai C (2015) Life Cycle Analysis of Wind Turbine. https://doi.org/10.5772/29184 [167] Schleisner L (2000) Life cycle assessment of a wind farm and related externalities. 20:279– 288 [168] Ardente F, Beccali M, Cellura M, Lo Brano V (2008) Energy performances and life cycle assessment of an Italian wind farm. Renew Sustain Energy Rev 12:200–217. https://doi.org/10.1016/j.rser.2006.05.013 [169] Crawford RH (2009) Life cycle energy and greenhouse emissions analysis of wind turbines and the effect of size on energy yield. Renew Sustain Energy Rev 13:2653–2660. https://doi.org/10.1016/j.rser.2009.07.008 [170] Pacca S, Horvath A (2002) Greenhouse gas emissions from building and operating electric power plants in the upper Colorado river basin. Environ Sci Technol 36:3194–3200. https://doi.org/10.1021/es0155884 [171] Weinzettel J, Reenaas M, Solli C, Hertwich EG (2009) Life cycle assessment of a floating offshore wind turbine. Renew Energy 34:742–747. https://doi.org/10.1016/j.renene.2008.04.004 [172] Dolan SL, Heath GA (2012) Life Cycle Greenhouse Gas Emissions of Utility-Scale Wind Power Systematic Review and Harmonization. 16:. https://doi.org/10.1111/j.1530- 9290.2012.00464.x [173] Arvesen A, Birkeland C, Hertwich EG (2013) The importance of ships and spare parts in LCAs of offshore wind power. Environ Sci Technol 47:2948–2956. https://doi.org/10.1021/es304509r [174] Priyanka R, Garrett P (2015) Vestas - Life Cycle Assessment of Electricity Production from an onshore V110-2 . 0 MW Wind Plant [175] Marx J, Schreiber A, Zapp P, Walachowicz F (2018) Comparative Life Cycle Assessment of NdFeB Permanent Magnet Production from Different Rare Earth Deposits. ACS Sustain Chem Eng 6:5858–5867. https://doi.org/10.1021/acssuschemeng.7b04165 196 [176] Zapp P, Marx J, Schreiber A, et al (2018) Comparison of dysprosium production from different resources by life cycle assessment. Resour Conserv Recycl 130:248–259. https://doi.org/10.1016/j.resconrec.2017.12.006 [177] Garrett P, Rønde K (2012) Life Cycle Assessment of Electricity Production from an onshore V90-3 . 0MW Wind Plant September 2012 Authors : Peter Garrett & Klaus Rønde Vestas Wind Systems A / S. Vestas Wind Syst A/S Vestas 1–106 [178] Stocker T, Qin D (2013) Climate change 2013. The physical science basis. Intergovernmental panel on climate change. [179] DNVGL (2020) DNVGL. https://www.dnvgl.com/news/authors-of-energy-transition- outlook-make-its-dataset-freely-available--143475 [180] EIA (2020) Annual Energy Outlook 2020. https://www.eia.gov/outlooks/aeo/data/browser/ [181] U.S. Department of Energy (2020) Wind Vision. https://www.energy.gov/maps/map- projected-growth-wind-industry-now-until-2050 [182] Canada energy regulator (2020) Electricity generation. https://apps.cer- rec.gc.ca/ftrppndc/dflt.aspx?GoCTemplateCulture=en-CA+2019 [183] U.S. Department of Energy (2011) Critical Materials Strategy [184] Carrara S, Alves Dias P, Plazzotta B, Pavel C (2020) Raw materials demand for wind and solar PV technologies in the transition towards a decarbonised energy system [185] World Bank (2020) Minerals for Climate Action: The Mineral Intensity of the Clean Energy Transition. Clim Smart Min Initiat - World Bank Gr 110 pp. [186] Ballinger B, Schmeda-Lopez D, Kefford B, et al (2020) The vulnerability of electric-vehicle and wind-turbine supply chains to the supply of rare-earth elements in a 2-degree scenario. Sustain Prod Consum 22:68–76. https://doi.org/10.1016/j.spc.2020.02.005 [187] Pavel CC, Lacal-arántegui R, Marmier A, et al (2017) Substitution strategies for reducing the use of rare earths in wind turbines. Resour Policy 52:349–357. https://doi.org/10.1016/j.resourpol.2017.04.010 [188] World Steel Association (2020) Steel industry key facts. https://www.worldsteel.org/about- steel/steel-industry-facts.html%0A [189] IEA (2018) Cement technology roadmap plots path to cutting CO2 emissions 24% by 2050. https://www.iea.org/news/cement-technology-roadmap-plots-path-to-cutting-co2- emissions-24-by-2050%0A [190] Wang X, Lei Y, Ge J, Wu S (2015) Production forecast of China’s rare earths based on the 197 generalized Weng model and policy recommendations. Resour Policy 43:11–18. https://doi.org/10.1016/j.resourpol.2014.11.002 [191] Stark K (2019) EIA Outlook 2019: The ‘Extremely Conservative’ Case for Renewables Growth. https://www.greentechmedia.com/articles/read/eia-outlook-conservative- renewables [192] PCA (2017) Energy-Related Cement Consumption Outlook [193] Berndt ML (2004) Sustainable Concrete for Wind Turbine Foundations [194] Singh RK (2021) U.S. manufacturers grapple with steel shortages, soaring prices. https://www.reuters.com/article/us-usa-economy-steel-insight-idUSKBN2AN0YQ [195] Pereao O, Bode-Aluko C, Fatoba O, et al (2018) Rare earth elements removal techniques from water/wastewater: A review. Desalin Water Treat 130:71–86. https://doi.org/10.5004/dwt.2018.22844 [196] Peiravi M, Ackah L, Guru R, et al (2017) Chemical extraction of rare earth elements from coal ash. Miner Metall Process 34:170–177. https://doi.org/10.19150/mmp.7856 [197] Das S, Gaustad G, Sekar A, Williams E (2018) Techno-economic analysis of supercritical extraction of rare earth elements from coal ash. J Clean Prod 189:539–551. https://doi.org/10.1016/j.jclepro.2018.03.252 [198] Amato A, Becci A, Birloaga I, et al (2019) Sustainability analysis of innovative technologies for the rare earth elements recovery. Renew Sustain Energy Rev 106:41–53. https://doi.org/10.1016/j.rser.2019.02.029 [199] Copper Development Association (2019) How will the future copper demand be met? https://copperalliance.org.uk/coverage-future-copper-demand/ [200] USGS (2020) Aluminum Statistics and Information. https://www.usgs.gov/centers/nmic/aluminum-statistics-and-information/ [201] World Aluminium (2021) Global aluminium industry 2050 climate pathways. https://www.world-aluminium.org/news-opinions/ [202] European Aluminium Association (2019) Circular Aluminium Action Plan - a Strategy for Achieving Aluminium’s Full Potential for Circular Economy By 2030. 33 [203] EPA (1996) AP-42, CH 11.13: Glass Fiber Manufacturing. In: Mineral Products Industry, Fifth [204] EIA (2013) Glass manufacturing is an energy-intensive industry mainly fueled by natural gas. https://www.eia.gov/todayinenergy/detail.php?id=12631 198 [205] Looney B (2020) Statistical Review of World Energy, 2020 | 69th Edition [206] EIA (2020) How much natural gas does the United States have, and how long will it last? https://www.eia.gov/tools/faqs/faq.php?id=58&t=8 [207] NREL (2021) Advanced Thermoplastic Resins for Manufacturing Wind Turbine Blades. https://www.nrel.gov/manufacturing/comet-wind-blade-resin.html [208] AEEI (2015) Competitiveness of renewable energy and energy efficiency in the U.S. markets. Adv Energy Econ Inst 19 [209] EIA (2016) Wind and Solar Data and Projections from the U.S. Energy Information Administration: Past Performance and Ongoing Enhancements. US Energy Inf Adm 31 [210] Kandhal PS, Cooley LAJ (2001) NCHRP REPORT 464. The Restricted Zone in the Superpave Aggregate Gradation Specification [211] EIA (2019) Crude oil production in USA. https://www.eia.gov/dnav/pet/pet_crd_crpdn_adc_mbbl_a.htm [212] EIA (2019) Crude oil imported. https://www.eia.gov/dnav/pet/pet_move_impcus_a2_nus_epc0_im0_mbbl_a.htm [213] EIA (2019) Movements of Crude Oil by Pipeline. https://www.eia.gov/dnav/pet/pet_move_pipe_a_EPC0_LMV_mbbl_a.htm [214] EIA (2019) Movements of Crude Oil by Tanker and Barge. https://www.eia.gov/dnav/pet/pet_move_tb_a_EPC0_BMV_mbbl_a.htm [215] EIA (2019) Movements of Crude Oil by Rail. https://www.eia.gov/dnav/pet/pet_move_rail_a_EPC0_RAIL_mbbl_a.htm 199