Evaluation of Asphalt Mixtures Containing Metallic Fibers from Recycled Tires to Promote Crack-Healing

materials

Article Evaluation of Asphalt Mixtures Containing Metallic Fibers from Recycled Tires to Promote Crack-Healing

Alvaro González 1,*, José Norambuena-Contreras 2 , Lily Poulikakos 3, María José Varela 2, Jonathan Valderrama 1, Alexander Flisch 3 and Martín Arraigada 3

1 Department of Construction Engineering and Management, School of Engineering, Pontiﬁcia Universidad Católica de Chile, Avenida Vicuña Mackenna, Santiago 4860, Chile; [email protected] 2 LabMAT, Department of Civil and Environmental Engineering, University of Bío-Bío, Avenida Collao, Concepción 1202, Chile; [email protected] (J.N.-C.); [email protected] (M.J.V.) 3 Empa, Swiss Federal Laboratories for Materials Science and Technology, Überlandstrasse, 129 CH-Dübendorf, Switzerland; [email protected] (L.P.); alexander.ﬂ[email protected] (A.F.); [email protected] (M.A.) * Correspondence: [email protected]; Tel.: +56-2-2354-7031

Received: 19 August 2020; Accepted: 20 November 2020; Published: 16 December 2020

Abstract: This paper reports part of an international research project with the long-term aim of developing more sustainable asphalt mixture with crack-healing properties by the addition of recycled metallic waste from industrial sources. Specifically, this article presents an evaluation of the physical, thermophysical, and mechanical properties of asphalt mixtures with metallic fiber obtained from recycled tires for crack-healing purposes. Detailed results on the crack-healing of asphalt mixtures will be reported in a second article. Results showed a small reduction on the bulk density and increase in the air voids content was quantified with increasing fiber contents. The experimental results showed that mixing and compaction was more difficult for higher fiber contents due to less space for the bitumen to freely flow and fill the voids of the mixtures. Computed tomography (CT) results allowed to identify clustering and orientation of the fibers. The samples were electrically conductive, and the electrical resistivity decreased with the increase of the fiber content. Fiber content had a direct effect on the indirect tensile stiffness modulus (ITSM) and strength (ITS) that decreased with increasing temperature for mixtures and with increase in fiber content. However, the indirect tensile strength ratio (ITSR) was within acceptable limits. In short, results indicate that fibers from recycled tires have a potential for use within asphalt mixtures to promote crack-healing.

Keywords: asphalt mixture; metallic ﬁbers; waste tires; crack-healing; computed tomography.

1. Introduction Tire production for vehicles and trucks is globally increasing given the growing population, economy, and transportation. Worldwide, tire production exceeds 2.9 billion units per year [1]. It is estimated that for every tire placed in the market, another tire reaches its service life and becomes waste [2]. This massive amount of waste occupies a large area in landﬁlls and causes environmental hazards. Waste tires are nondegradable due to cross-linking of vulcanized rubber with sulfur bonds, and the addition of antioxidants and antiozonants during tire production [3]. Burning or using tire as fuel may produce toxic gases that are harmful for the environment and may cause destructive pollution of natural air [4,5]. Tire piles often provide breeding grounds for pests and insects such as mosquitoes, because their shape and impermeability allow them to hold water for extended periods, facilitating the spreading of contagious diseases [6]. Diﬀerent environmental policies, like extended

Materials 2020, 13, 5731; doi:10.3390/ma13245731 www.mdpi.com/journal/materials Materials 2020, 13, 5731 2 of 17 producer responsibility, control waste tire disposal in developed countries have aimed to address this problem. However, waste tires are still landfilled in many countries causing major public health risks and environmental issues [7]. Common components of tires are natural and synthetic rubbers, carbon black, metallic fiber, textile fabric, and additives [8]. Rubber from recycled tires has many applications in the industry. Tire pyrolysis, i.e. the thermal degradation of waste tire at elevated temperature, produces value-added products such as tire pyrolysis oil, pyro char, pyro gas, which are used as alternative fuel for engines [6,9]. Rubber from recycled tires are a source of sustainable new material that are also used to create environmentally friendly and low-cost composites [10]. In civil engineering materials, researchers have used rubber from recycled tires in mortars [5]; cement concrete [2,4,11]; geogrids in granular soils [12]; aggregates [13,14]; soil reinforcement [15]; composite membranes [16,17]; and asphalt [18–20]. The use of metal fiber waste from tires is less explored; however, researchers have investigated the effect of fiber waste from tires in the mechanical, fire resistance, and acoustic properties of concrete [21–26]. The use of metal fiber waste from tires in asphalt mixtures is less explored than concrete, with only few studies found in the literature [27]. Nevertheless, researchers have used different types of fiber mainly to increase the mechanical performance and durability of asphalt mixtures. Some of these fibers are products especially fabricated with the main purpose of material reinforcement [28–30], while others are waste byproducts from various industries [31]. In addition to the enhancement of the mechanical properties of asphalt mixtures, metallic fiber or particles from virgin materials or waste improve the electrical and thermophysical properties of asphalt mixtures [32]. This enhancement has led to the development of innovative materials, like pavements with crack-healing properties by external heating [33–36]. In these mixtures, metals, normally steel wool fibers, are added because they conduct and absorb more thermal energy than bitumen and aggregates, increasing the electrical conductivity of the asphalt mixtures [37]. To increase the temperature and heal this type of asphalt mixture, an external electromagnetic field, such as those applied by electromagnetic induction or microwaves, is artificially applied to heat the fiber. Later, the fiber heat transfers to the rest of the mixture, reducing the bitumen viscosity, making the bitumen flow and seal, repairing open cracks. Only a few studies have included waste materials for the preparation of asphalt mixtures with crack-healing purposes. Norambuena-Contreras et al. [38] evaluated the effect of metallic waste content on the thermophysical, electrical, and microwave crack-healing properties of asphalt mixtures, adding two types of metallic waste in different contents: steel wool fiber and steel shavings from the metal turnery industry. Gonzalez et al. [39] evaluated the effect of adding metal shavings and reclaimed asphalt pavement (RAP) on the properties of asphalt mixtures with crack-healing capabilities by microwave heating, concluding that asphalt mixtures with RAP and waste metal shavings have the potential of being crack-healed by microwave heating. The researchers have also compared different type of metal waste and other type of ceramic additives in the microwave crack-healing of asphalt mixtures [40]. Although a good number of articles on the crack-healing of asphalt mixtures with fibers or metallic waste are found in the literature, metallic fibers from recycled tires have been sparsely considered for this purpose [41]. This paper reports the first part of an international research project between Chile and Switzerland, with the long-term aim of developing a sustainable asphalt mixture with advanced mechanical, thermophysical, and crack-healing properties by the addition of recycled metallic waste from industrial sources. The objective of the paper is to evaluate the physical, thermophysical, and mechanical properties of asphalt mixtures with crack-healing potential containing different amounts of metallic fiber obtained from recycled tires. Detailed results on the crack-healing of asphalt mixtures will be reported in a second article. Asphalt mixture with one type of metallic fiber, bitumen, and aggregates, were studied by experimental tests. The results indicate that fibers from recycled tires have a positive effect on the properties of asphalt mixtures, as further discussed below. In addition, results indicate that fibers from recycled tires have a potential for use within asphalt mixtures to promote crack-healing. Materials 2020, 13, 5731 3 of 17

2. Materials and Methods

2.1. Raw Materials The aggregates, sourced from a local construction company located in Santiago, Chile, were provided in diﬀerent fractions and later combined to produce the target particle size distribution of a dense asphalt mixture. The aggregates supplied consisted of a 19 mm coarse gravel, a 12.5 mm gravel, and a crushed dust (Table1). These fractions were blended to achieve the particle size distribution of dense asphalt mixtures. The penetration grade of the CA24 bitumen used for mixture preparation was 80/100 mm at 25 ◦C and its absolute viscosity >2400 poise at 60 ◦C. The bitumen content targeted for all the asphalt mixtures was 5.2%, by mass. The virgin metallic ﬁbers with rubber scraps were obtained from a tire recycling plant from Tire Recycling Solutions, Préverenges, Switzerland.

Table 1. Particle size distribution of aggregates used.

Aggregate Aggregate Size (mm) Fractions Combination 19 mm 12.5 mm Crushed Dust 19 100 100 100 100 12.5 36 100 100 84 10 1 77 100 71 5 1 5 90 47 2.5 0 1 60 31 0.63 0 0 30 14 0.315 0 0 22 10 0.16 0 0 16 7 0.075 0 0 11 5

2.2. Preparation of Asphalt Mixture Specimens

The aggregates and bitumen were heated in the oven for a minimum of two hours at 150 ◦C before mixing. The mixing order for specimens with metallic ﬁbers followed these steps:

The bitumen was placed in a metallic container previously heated and used for the mixing to keep • the mixing temperature constant. Metallic fibers from recycled tires with rubber scraps bounded were initially cleaned in a solvent • solution for a few hours and then were heated to 500 ◦C to improve the ductility of the fibers during the mixing and compaction. Then, approximately 1% of the metallic fibers previously heated were blended with the mixture. • The remaining metallic fibers were gradually included after the addition of aggregate lots. Small lots of aggregates were added to the mixtures, separated by particle size. The lot with the • largest particles was the first to be added. Once the aggregates were entirely covered with bitumen, the next lot with smaller particles was • added to the mixture.

Marshall specimens were produced taking mixture from each batch (diameter = 100 mm; height = 60 mm, approximately). A sample of approximately 1150–1200 g of mixture was used for the preparation of each specimen. The sample was compacted using a Marshall hammer (Humboldt, IL, USA) giving 75 blows to each specimen face. The estimated bitumen content for each Marshall specimen was 58 g. The metallic ﬁber contents added to each specimen were 1.5%, 2.5%, and 3.5% of the bitumen volume, equivalent to 7 g, 11.6 g, and 16.3 g, respectively. In total, 117 Marshall test specimens were prepared. Materials 2020, 13, 5731 4 of 17 Materials 2020, 12, x FOR PEER REVIEW 4 of 18

2.3. Characterization of Metallic Fiber from Recycled Tires Size, surface aspect and cross-section of the metallic fibers fibers from recycled tires were characterized by optical and environmentalenvironmental scanningscanning electron microscopymicroscopy (ESEM),(ESEM), respectivelyrespectively following the methodology described by Norambuena-Contreras etet alal [[38].38]. To this end, one hundred metallic fibersfibers were randomlyrandomly selected.selected. Later, Later, the the diameter and length of each fiberfiber was determined by taking photographs under a stereoscopic microscope (Leica EZ4, Wetzlar, Germany)Germany) withwith 3535× magnificationmagnification ® × and measured with with the the image image processing processing software software ImageJ ImageJ® (version(version IJ 1.46r, IJ 1.46r, Bethesda, Bethesda, MD, MD, USA). USA). An Animage image of the of themetallic metallic fibers fibers from from recycled recycled tires tires used used and andthe frequency the frequency histogram histogram of their of their length length and anddiameter diameter are shown are shown in Error! in Figure Reference1a,b, respectively. source not found.a,b, respectively.

Figure 1. Characterization of metallic fiber fiber from recycled tires: ( a) group of fibers;fibers; (b) ESEM image of the cross-section ofof the cutting fiber;fiber; ((c)) frequency histogramhistogram ofof thethe length and diameter of fibersfibers data with WeibullWeibull and and normal normal fitting fitting respectively; respectively; (d )(d ESEM-EDX) ESEM-EDX elemental elemental analysis analysis of the of fibersthe fibers before before and; (and;e) after (e) manufacturingafter manufacturing indicates indicates more more mineral mineral residue residue and oxidation and oxidation after theafter mixing the mixing process. process. Moreover, the surface aspect ratio and cross-section of metallic fibers (see Figure1c) randomly Moreover, the surface aspect ratio and cross-section of metallic fibers (see Error! Reference selected was observed by using a ESEM Quanta FEG650 (ThermoFisher, Waltham, MA, USA) in the source not found.c) randomly selected was observed by using a ESEM Quanta FEG650 high vacuum mode. Additionally, metallic fibers were characterized through ESEM-EDX analysis (ThermoFisher, Waltham, MA, USA) in the high vacuum mode. Additionally, metallic fibers were before the mixing and compacting processes. To do it, asphalt mixture sample with 2.5% fibers was characterized through ESEM-EDX analysis before the mixing and compacting processes. To do it, placed in a toluene solution for a few hours and then the solution was filtered for that fibers were asphalt mixture sample with 2.5% fibers was placed in a toluene solution for a few hours and then extracted using a magnet. The ESEM images are a result of the interaction of the electron beam with the the solution was filtered for that fibers were extracted using a magnet. The ESEM images are a result atoms of the sample. The higher the atomic number the brighter the images. Energy dispersive X-rays of the interaction of the electron beam with the atoms of the sample. The higher the atomic number the brighter the images. Energy dispersive X-rays (EDX) are characteristic X-rays that are produced

Materials 2020, 13, 5731 5 of 17

(EDX) are characteristic X-rays that are produced due to displacement of electrons in the sample shells. In short, it was determined that metallic fibers from recycled tires had an average diameter of 0.256 mm (standard deviation (SD) = 0.079 mm), see Figure1b,c with an average aspect ratio of 81, and an initial length range of 11–40 mm, see histogram in Figure1b, which indicates that both short and long metallic fibers were randomly added to the asphalt mixture. Likewise, after the manufacturing process, the main difference between fibers is that there are particles of minerals attached to the fibers as a result of the mixing with aggregates. EDX analysis at two locations on the fibers is shown in Figure1d,e showing that after the mixing process there is more mineral residue and oxidation in the fibers.

2.4. Physical Characterization of Asphalt Test Samples Two important physical properties of asphalt mixtures are bulk density and air voids content. Bulk density is the ratio between the dry weight and the real volume of each specimen. The real volume of each specimen is determined using the weight of the water-submerged specimen. Since dry weight is known, bulk density (ρb) is calculated for each asphalt specimen. The material composition for and their density for each asphalt mixture type were known, hence the theoretical maximum density for each specimen was found (ρmax). Therefore, the air voids content (AV) of each test sample was calculated as [42]: ρmax ρb AV = − (1) ρmax Six test specimens were used for the calculation of the average bulk density and air voids content.

2.5. Fibers Distribution and Orientation by X-Ray Computed Tomography Depending on the amount of fiber used, clustering can increase the material’s inhomogeneity. This in turn can reduce its mechanical performance and affect the homogeneous distribution of the induced heat, as the fibers will be concentrated in some areas and will not be present in others. Furthermore, the direction in which the fibers are mostly aligned might influence the isotropy of the material. Therefore, X-ray computed tomography (CT) was used to analyze the distribution and orientation of the fibers of the three studied concentrations (1.5%, 2.5% and 3.5% of fibers in bitumen volume). Since X-rays were not able to penetrate the mass of the Marshall specimens (100 mm diameter), cylinders of approximately 40 mm diameter and 40 mm height were cored from the center of the 100mm specimens and used as samples for the scans. The CT system used is a RayScan500 (RayScan, Meersburg, Germany) equipped with a 300 kV micro focus X-ray source FineTec FORE 300.01Y RT (PerkinElmer, Waltham, MA, US) and a digital X-ray detector PerkinElmer XRD 1611 CP3 (PerkinElmer, Massachusetts, US). The detector is based on a 16” amorphous silicon sensor operating as a two-dimensional photodiode array. X-rays are converted into light using a CsI (Ti) scintillator with needles directly deposited on the photodiodes. The information is digitized in 16 bits to achieve a high dynamic range. The detector has a pixel size of 100 µm and an image size of 4096 4096 pixels. The operator of the CT system has three longitudinal × and one rotational axis and is able to handle samples with a weight up to 50 kg. The best achievable spatial resolution of the CT system is about 3 µm. The asphalt samples were scanned at X-ray source settings of 290 kV/490 µA. A total number of 1260 X-ray projections were acquired with an integration time of 999 ms and a detector pixel binning of 2. The source–detector distance of the scanner was set to 1534.9 mm and the source–object distance was 445.5 mm. In this study, the reconstructed three-dimensional (3D) images of each measurement achieved a resolution of (116 µm)3, i.e. this is the size of the smallest cubic units in the images called voxels. For visualization purposes, each voxel is assigned a gray tonality proportional to the X-ray attenuation coefficient, which in turn depends on the material’s density. Low-density materials are weak X-ray absorbers and exhibit low attenuation. When visualized on a typical gray-scale map these voxels appear as dark regions. On the other hand, high-density materials appear as brighter regions [43]. In order to perform the analysis of the fibers’ distribution and orientation it is first necessary to separate Materials 2020, 13, 5731 6 of 17 them from the aggregates, bitumen and void matrix through an image segmentation that splits the voxels according to the material. As described earlier, the 16 bit image digitalization corresponds to attenuation coefficients that range from 1 to 65,536 and can be represented as a histogram of gray-scale intensity distribution. Ideally, each material composing the asphalt concrete samples should present a clear peak in the histogram and segmentation could be easily performed by setting global thresholds between these peaks. However, due to several reasons like weak contrast between materials, noise or so-called partial volume effects, voxels often show attenuation coefficients different to those expected for a given material [44]. Thus, to avoid errors in the segmentation, the processing method should be carefully selected through a sensitivity analysis and validated against a known parameter. In this study, the volume of fibers calculated from the CT scan images was compared to the real volume of fibers in order to assess the quality of each methodology. To that end, the fibers contained in the scanned samples were recovered dissolving the asphalt binder with toluene. Fibers and aggregates were separated manually using sieves and magnets. Then, the weight of the fibers was measured to obtain their volume. The processing of the CT scans images was achieved with the software VGstudio Max 3.3 (VolumeGraphics, Heidelberg, Germany) [45]. This software offers the possibility of using a segmentation tool for the analysis of defects like pores or inclusions, considering two algorithms. The ‘Only threshold’ algorithm is based on a global gray threshold in which each voxel is examined and defined as pore or inclusion if it is below or above a certain value respectively. On the other hand, the VGDefX is a more sophisticated algorithm that checks if some voxels, considered as seeds, are defects. Then, it looks in the neighborhood for other voxels that are part of the defect following a certain probability criteria. This algorithm allows for gray value variations and includes noise reduction for seed point location. In this work, the fibers were assumed as inclusions in the asphalt matrix, as metal has higher density than bitumen or stones and could be differentiated through its gray value. It has to be mentioned that, due to the fact that the fiber’s diameter is close to the image resolution, blurred fibers edges can cause partial volume effects with a relatively big impact on the segmentation. Therefore, to perform the segmentation the most precise VGDefX’ algorithm was selected. As part of the sensitivity analysis, different threshold values were manually chosen based on the histogram of gray-scale intensity. The volume of the fibers was then calculated. However, it was found that the aggregates also have metallic inclusions with similar gray values as the fibers that cannot be separated only with the segmentation. Thanks to a tool from VGstudio Max 3.3 [45], selected inclusions can be filtered by their geometric features. Fibers have an elongated shape that differ completely from the typical round form of the inclusions in aggregates, which are also usually of small size. Therefore, three features were used to identify which of the inclusions correspond to fibers:

Size: amount of voxels that are part of the inclusion. Considering the average thickness and • length measured in last section, the fibers have an average diameter and length of ca. 0.256 mm and 20 mm respectively, which makes a volume of ca. 1 mm3. Values that differ too much of this volume are filtered out, as explained in Section 3.2. Sphericity: ratio between the surface of a sphere with the same volume as the inclusion and the • surface of the defect, with values between 0 and 1 being 1 a perfect sphere. Fibers have values towards 0. Compactness: ratio between the volume of the defect and the volume of the circumscribed sphere, • with values between 0 and 1, being 1 a perfect sphere. Compactness of fibers are close to 0.

In order to reduce white noise and enhance the quality of the images before segmentation, the software offers the possibility to apply different filters. According to the literature [46], median and Gaussian filters available in VGstudio Max 3.3 [45] exhibit good performance in reducing white noise and greyscale value outliers but they might weaken the contrast in the edges of the fibers and induce a bias in the gray scale values. A visual assessment of the use of these filters is presented in Figure2. Although the adaptive Gaussian filter seems to provide a better contrast between the fiber Materials 2020, 13, 5731 7 of 17

Materials 2020, 12, x FOR PEER REVIEW 7 of 18 and the surrounding bitumen-aggregate matrix, the fiber size appears to be reduced. On the other hand,blurred a blurredimage is image the consequence is the consequence of the application of the application of a median of a filter. median Conseque filter. Consequently,ntly, in this work in this no workimage no filter image was filter applied. was applied.

Materials 2020, 12, x FOR PEER REVIEW 7 of 18

blurred image is the consequence of the application of a median filter. Consequently, in this work no image filter was applied.

(a) (b) (c)

Figure Figure 2. 2.2D 2D slices slices showing showing ( a(a)) a a raw raw version version ( b(b)) the thee effectffectof ofthe the adaptiveadaptive GaussianGaussian andand ((cc)) medianmedian filtersfilters onon thethe imageimage ofof oneone fiber.fiber.

OnceOnce thethe sensitivitysensitivity analysis was performed and thethe best segmentation selected (see Figure 33),), thethe distributiondistribution Figure ofof thethe 2. fibers2Dfibers slices was wasshowing analyzedanalyzed (a) a raw version inin orderorder (b) the toto effect determinedetermine of the adaptive possiblepossible Gaussian accumulationaccumulation and (c) median in clusters. filters on the image of one fiber. TheThe specimen 3D image image was was divided divided into into eight eight cubic cubic subspaces subspaces with with a side a side length length of 150 of 150voxels voxels (ca. (ca.17.4 17.4 mm), mm), equivalent equivalentOnce the to sensitivity almost to almost 5.3analysis 5.3 cm cm3 was (see3 (see performed Figure Figure 3b).and3b). th Fore For best each each segmentation subspace, subspace, selected the percentage(see Figure 3), ofof fibersfibers waswas calculatedcalculatedthe bydistributionby addingadding of upup the allall fibers thethe was segmented analyzed voxelsinvoxe orderls assignedto determine asas possible fibers,fibers, a divideddividedccumulation byby in the clusters. defineddefined total The specimen 3D image was divided into eight cubic subspaces with a side length of 150 voxels (ca. volume of the subspace. The comparison of the percentage of fibers in each subspace was used as an volume of the17.4 subspace. mm), equivalent The to comparison almost 5.3 cm of3 (see the Figure percentage 3b). For each of fibers subspace, in eachthe percentage subspace of wasfibers used as anindication indication aboutwas about calculated possible possible by addingclusters. clusters. up all Based the Based segmented on on the the voxe same samels assigned segmentation segmentation as fibers, divided method, method, by the an defined an analysis analysis total of thethe orientationorientation of volume the the fibers fibersof the subspace.was was carried carried The comparison out out using using of another the another perc entagetool tool in of inthe fibers the software in software each subspace VGstudio VGstudio was usedMax Max as 3.3 an [45],3.3 [ 45the], thefiber fiber composite compositeindication material materialabout analysis possible analysis (FCMA)clusters. (FCMA) Based module. on module. the This same tool This segmentation can tool present can method, present the orientationan the analysis orientation of of the the offibers the orientation of the fibers was carried out using another tool in the software VGstudio Max 3.3 [45], the based on a spatial coding visualized on a color sphere, as displayed in Figure 3c. To quantify if there fibers basedfiber on a composite spatial coding material visualizedanalysis (FCMA) on amodule. color sphere,This tool ascan displayed present the orientation in Figure 3ofc. the To fibers quantify if thereis a direction is a directionbased or a onplane ora spatial a to plane which coding to whichthevisualized fibers the on are fibers a colormostly are sphere mostlyaligned,, as displayed aligned, each orientationin Figure each orientation3c. To is quantify projected is if projectedthere to a sphere to aas sphere a histogram as ais histograma directionwhere the or wherea frequency plane to the which frequency is thedisplayed fibers are is displayedmostlyas a color aligned, (see as each a Figure color orientation (see3d). Figure Theis projected polar3d). toplot The a sphere shown polar plotlater shownin the results later inas section,a the histogram results shows where section, the frequency surface shows the isof displayed surfacethe ha lf-sphereas of a color the half-sphere (see unrolled Figure 3d). into unrolled The 2D. polar Theinto plot centershown 2D. The later of center the plot of in the results section, shows the surface of the half-sphere unrolled into 2D. The center of the plot represent the fibers aligned vertically, coincident with direction of movement of the Marshall the plot representrepresent the the fibers fibers aligned aligned vertically,vertically, coincident coincident with with direction direction of movement of movement of the ofMarshall the Marshall hammerhammer usedusedhammer forfor compaction compaction used for compaction of of the the specimen ofspecimen the specimen (polar (polar (polar angle angle angleϕ φ= ==0 0°). ◦0°).). Instead, Instead, the the edgethe edge edge of the of ofpolar the the polar plot polar plot plot is perpendicularis perpendicularis perpendicular tothe to compactionthe compaction to the compaction (polar (polar angle (polar ϕangle= angle90 ◦φ). φ The= = 90°).90°). azimuth TheThe azimuth angleazimuth τanglerepresent angle τ represent τ therepresent orientationthe the oforientation the fiber inoforientation horizontal the fiber of the planein fiberhorizontal and in horizontal is less plane relevant plane and and is in isless the less context relevantrelevant ofin in thethe the compactioncontext context of the of method.compaction the compaction method. method.

Figure 3. (a)Figure Aspect 3. (a of) Aspect the specimen of the specimen with with 1.5% 1.5% of of fibers. fibers. (b)) View View of ofthe the subspaces subspaces used for used the for the distribution analysis.distribution (analysis.c) Space (c orientation) Space orientation of the of the fibers, fibers, according according to to the the color color mapping mapping shown shown in the in the sphere. (d) Schematic of the construction of a polar plot for fiber orientation analysis. sphere. (d) Schematic of the construction of a polar plot for fiber orientation analysis.

Materials 2020, 13, 5731 8 of 17

2.6. Electrical and Thermal Properties of Asphalt Test Samples Electrical resistivity and thermal conductivity of the asphalt mixture test samples without, and with, diﬀerent metallic ﬁber contents were measured following the experimental procedures described by Norambuena-Contreras et al. [38]. Firstly, to measure the electrical resistance of each asphalt test sample a megohmmeter device (IR4056-20, Hioki, Nagano, Japan) connected to two stainless steel electrode plates with dimensions 10 cm 15 cm was used. The electrodes were placed on the opposite × faces of the test samples and a pressure of 1 kPa was applied on them, ensuring that the samples were centered, and the plates horizontally aligned during the measurements. After that, electrical resistance measurement of test samples was recorded, and their electrical resistivity was calculated applying Ohm’s law [47]: R S ρ = · , (2) l where R is the electrical resistance of each test sample in Ω; S is the electrode plate area in m2; and l is the thickness of each asphalt sample in m. The electrical resistivity of each sample tested was determined as the average of three tests. Additionally, the thermal conductivity of each asphalt test sample was measured using the thermal needle probe method, based on the transient linear heat source theory. To this end, KD2-Pro thermal analyzer (Decagon Devices, Pullman, WA, USA) device with a handheld controller and a needle 1 1 thermal sensor (RK-1, Decagon Devices, Pullman, WA, USA) (in the range of 0.1–6.0 Wm− K− ) was used. To carry out the tests, the thermal sensor covered with a polysynthetic thermal compound was embedded in the test sample that was previously drilled. During the measurement, the tested sample was placed on two other test samples to ensure adiabatic conditions. The duration of each was 10 min. Lastly, for each test sample, the thermal conductivity (λ) was calculated using the following relationship [47]: q λ = , (3) 4 π m · · where q is the heat generated by the sensor in W/m; m is the slope of the linear relationship between the temperature range for the testing time; and λ is the thermal conductivity of each asphalt test 1 1 sample in Wm− K− . The thermal conductivity of each test sample was determined as the average of three measurements.

2.7. Stiffness Modulus and Indirect Tensile Strength The stiffness was measured using the indirect tensile stiffness modulus test (ITSM), following the European Standard UNE-EN 12697-26:2006 Annex C [48]. In the test, a vertical repetitive load is applied to the vertical diameter of the Marshall cylindrical specimen. The applied load deforms the specimen, reaching a 50 εµ horizontal deformation measured using two linear variable differential transducers (LVDTs) located parallel to the horizontal diameter. Ten repetitive loading pulsations were applied on two orthogonal diameters to measure the potential anisotropy of the material. For the calculation of the ITSM the following equation was used [42,49]:

P (0.27 + ν) ITSM = · , (4) d t · where P is the peak vertical load (in N) diametrically applied, ν is the Poisson’s ratio (assumed 0.35 based on [48] findings), d is the horizontal maximum deformation (in mm), t is the specimen height or thickness (in mm), and ITSM is the calculated stiffness modulus (in MPa). The stiffness modulus tests were performed in specimens with metal fiber contents of 0%, 1.5%, 2.5%, and 3.5%. The tests were carried out in a temperature-controlled chamber at temperatures of 5, 20, 30, and 40 ◦C. To ensure that the aim temperature was uniform in all the specimen volume, before testing the specimens were left in the temperature chamber for a minimum of four hours. Materials 2020, 13, 5731 9 of 17

MaterialsThe 2020 indirect, 12, x FOR tensile PEER REVIEW strength (ITS) test was conducted on Marshall specimens loaded vertically9 of 18 through its diameter at a constant speed of 5 0.2 mm/min. The compressive load induces transverse ± tensiontension stress stress on on the the central central portion portion through through plane plane lo loadad and and horizontal horizontal tension. tension. The The horizontal horizontal stress stress inducedinduced on on the the center center of of the the specimen specimen is is calculated calculated as as follows follows [42,50]: [42,50]:

2 = 2 Fmax (5) σmax = × (5) π D t × × where σh is the maximum tensile stress in the center of the specimen (N/mm2), Fmax is the maximum 2 F forcewhere (N),σh Dis is the the maximum specimen tensile diameter stress (mm), in the and center t is the of thespecimen specimen height (N/ mm(mm).), Marshallmax is the specimens maximum D t forforce the (N), ITS dryis thetest specimen were conditioned diameter and (mm), tested and atis 20 the °C. specimen Marshall height specimens (mm). for Marshall the ITS specimens wet test werefor the first ITS soaked dry test in werewater conditioned at 40 °C for and72 h tested and later at 20 tested◦C. Marshall at 20 °C. specimens for the ITS wet test were ﬁrst soaked in water at 40 ◦C for 72 h and later tested at 20 ◦C. 3. Results and Discussion 3. Results and Discussion

3.1.3.1. Effect Effect of of the the Metallic Metallic Fiber Fiber from from Recycled Recycled Ty Tyresres on on the the Physical Physical Propert Propertiesies of of Asphalt Asphalt Mixtures Mixtures 3 3 TheThe average average bulk bulk density density for for all all specimens specimens tested tested was was 2.331 2.331 g/cm g/cm3 (SD(SD == 0.0360.036 g/cm g/cm3).). The The general general effecteffect of of the the metal metal fiber fiber from from recycled recycled tires tires content content was wasa small a small reduction reduction on the on bulk the density, bulk density, with 3 averagewith average bulk densities bulk densities of 2.382, of 2.382, 2.312, 2.312, 2.329, 2.329, and and2.301 2.301 g/cm g/cm for3 formetal metal fiber fiber contents contents of of0%, 0%, 1.5%, 1.5%, 2.5%,2.5%, and and 3.5%, 3.5%, respectively (Figure(Figure4 a).4a). The The general general average average air air voids voids for for mixtures mixtures with with di ff erentdifferent fiber fibercontents contents was 6.6%.was 6.6%. Results Results show show an overall an overall air voids air contentvoids content increase increase with increasing with increasing fiber contents. fiber contents.The average The airaverage voids air measured voids measured were 4.7%, were 7.7%, 4.7% 6.1%,, 7.7%, and 6.1%, 7.7%, forand metal 7.7%, fibers for metal contents fibers of contents 0%, 1.5%, of2.5%, 0%, 1.5%, and 3.5% 2.5%, (Figure and 3.5%4b). (Figure 4b).

FigureFigure 4. 4. MetalMetal fiber ﬁber effect eﬀect from from recycled recycled tires tires on on the the (a (a) )bulk bulk density; density; and and (b (b) )air air voids voids content. content.

TheThe dissimilar dissimilar bulk bulk density density and and air air voids voids are are expl explainedained by by the the different different content content of of metal metal fibers, fibers, whilewhile the the aggregates aggregates and bitumenbitumen contentcontent remained remained constant constant in in the the mixtures. mixtures. The The higher higher air voidsair voids with withincreasing increasing metal metal fiber contentfiber content is explained is explained by the formationby the formation of metal of fibers metal clusters fibers in clusters the mixture, in the as mixture,will be shown as will laterbe shown in this later paper. in this This paper. result This is consistent result is withconsistent observations with observations made during made mixing during and mixingcompaction and compaction in the specimen in the preparation. specimen preparation. The mixing The and mixing compaction and compaction in mixtures in with mixtures higher with fiber highercontents fiber was contents more di wasfficult more because difficult there because is less spacethere is for less the spac bitumene for tothe easily bitumen flow. to Hence, easily forflow. the Hence,bitumen for is the more bitumen difficult is more to fill difficult the voids to of fill the the mixtures. voids of the mixtures. 3.2. Fibers Distribution and Orientation of the Metallic Fibers into the Asphalt Mixtures 3.2. Fibers Distribution and Orientation of the Metallic Fibers into the Asphalt Mixtures The sensitivity to the segmentation method was studied on volumes defined using the advance The sensitivity to the segmentation method was studied on volumes defined using the advance surface determination tool of VGstudio MAX 3.3, which stablishes a boundary between air and the surface determination tool of VGstudio MAX 3.3, which stablishes a boundary between air and the specimen through their gray values. Figure5a shows, as an example, the histograms of the distribution specimen through their gray values. Figure 5a shows, as an example, the histograms of the distribution of gray values for the specimen containing 1.5% of fibers, before and after surface determination. It is possible to observe peaks for the air, bitumen and aggregates, but not for the fibers as their gray values are blended with the aggregate’s values. Only by setting specific thresholds it is evident which region of the histogram represents the fibers. As explained in Section 2.5, different

Materials 2020, 13, 5731 10 of 17

of gray values for the specimen containing 1.5% of ﬁbers, before and after surface determination. It is possible to observe peaks for the air, bitumen and aggregates, but not for the ﬁbers as their

Materialsgray 2020 values, 12, x are FOR blended PEER REVIEW with the aggregate’s values. Only by setting specific thresholds it is10 evident of 18 which region of the histogram represents the fibers. As explained in Section 2.5, different thresholds thresholdsvalues (for values example (for example 17’000, 17’500,17’000, 18’000,17’500, 18’50018’000, and 18’500 19’000 and for 19’000 the 1.5%for the specimen) 1.5% specimen) were proposed were proposedfor the VGDefXfor the VGDefXmethod. method. These values These werevalues selected were selected based on based a visual on a assessment visual assessment of how theyof how affect theysegmentation. affect segmentation. Figure5 Figureb displays 5b displays the inclusions the inclusions obtained obtained using theusing 18’000 the 18’000 threshold threshold that contain that containnot only not only the fibers the fibers but alsobut also some some metallic metallic particles particles of theof the aggregates. aggregates. As As detailed detailed in in Section Section 2.6 , 2.6,fibers fibers and and aggregate aggregate inclusions inclusions diff differer up toup some to some degree degree in their in their size and size shape. and shape. This can This be can observed be observedin Figure in6 Figurea, where 6a, all where 2602 resultingall 2602 inclusionsresulting inclusions are plotted are regarding plotted theirregarding sphericity, their compactnesssphericity, compactnessand size. By and setting size. By thresholds setting thresholds in these parameters in these pa itrameters is possible it is topossible filter out to filter most out of the most aggregates’ of the aggregates’inclusions inclusions and only leaveand only the fibers leave forthe analysis. fibers for It analysis. was observed It was that observed all fibers that have all sizesfibers above have 0.1sizes mm 3 aboveand 0.1 sphericity mm3 and and sphericity compactness and compactness values below values 0.39 and belo 0.05w 0.39 respectively. and 0.05 respectively. Nevertheless, Nevertheless, any aggregate anyinclusion aggregate outliers inclusion still present outliers after still filtering present were after manually filtering erased were after manually visual assessment, erased after obtaining visual the assessment,final distribution obtaining of fibersthe final (see distribution Figure6b). Inof Figurefibers 7(see, the Figure fibers’ 6b). volumes In Figure obtained 7, the after fibers’ segmentation volumes obtainedand filtering after segmentation are plotted against and filtering the selected are plott threshold.ed against For thethe sampleselected containing threshold. 1.5% For ofthe fibers, sample 1.2 g containingof fibers 1.5% were of recovered, fibers, 1.2 g which of fibers represents were recovered, a volume which of 150 represents mm3. Using a volume Figure of 1507 it mm is possible3. Using to Figureinterpolate 7 it is possible and obtain to theinterpolate threshold and (18’322) obtain that the would threshold match (18’322) the real that volume would of match fibers, whichthe real was volumethen usedof fibers, for thewhich final was segmentation. then used for The the same final methodsegmentation. was used The tosame analyze method the was 2.5% used and to 3.5% analyzefiber samples.the 2.5% and 3.5% fiber samples.

FigureFigure 5. (a 5.) Histogram(a) Histogram of grayscale of grayscale distribution distribution of the ofim theages images of the 1.5% of the specimen 1.5% specimen before and before after and surfaceafter determination surface determination showing showing the materials the materials peaks peaksand the and threshold the threshold selected. selected. (b) View (b) View of the of the inclusions after segmentation (no ﬁlter), 1.5% ﬁber content. inclusions after segmentation (no filter), 1.5% fiber content.

Figure 6. (a) Overview the geometrical parameters of the inclusions, in red the values set to filter out the aggregate’s inclusions. (b) View of the inclusions after segmentation and filtering.

Materials 2020, 12, x FOR PEER REVIEW 10 of 18 thresholds values (for example 17’000, 17’500, 18’000, 18’500 and 19’000 for the 1.5% specimen) were proposed for the VGDefX method. These values were selected based on a visual assessment of how they affect segmentation. Figure 5b displays the inclusions obtained using the 18’000 threshold that contain not only the fibers but also some metallic particles of the aggregates. As detailed in Section 2.6, fibers and aggregate inclusions differ up to some degree in their size and shape. This can be observed in Figure 6a, where all 2602 resulting inclusions are plotted regarding their sphericity, compactness and size. By setting thresholds in these parameters it is possible to filter out most of the aggregates’ inclusions and only leave the fibers for analysis. It was observed that all fibers have sizes above 0.1 mm3 and sphericity and compactness values below 0.39 and 0.05 respectively. Nevertheless, any aggregate inclusion outliers still present after filtering were manually erased after visual assessment, obtaining the final distribution of fibers (see Figure 6b). In Figure 7, the fibers’ volumes obtained after segmentation and filtering are plotted against the selected threshold. For the sample containing 1.5% of fibers, 1.2 g of fibers were recovered, which represents a volume of 150 mm3. Using Figure 7 it is possible to interpolate and obtain the threshold (18’322) that would match the real volume of fibers, which was then used for the final segmentation. The same method was used to analyze the 2.5% and 3.5% fiber samples.

Figure 5. (a) Histogram of grayscale distribution of the images of the 1.5% specimen before and after

Materialssurface2020, 13determination, 5731 showing the materials peaks and the threshold selected. (b) View of the11 of 17 inclusions after segmentation (no filter), 1.5% fiber content.

Figure 6. (a) Overview the geometrical parameters of the inclusions, in red the values set to ﬁlter out MaterialsFigure 2020 6., 12(a,) x Overview FOR PEER theREVIEW geometrical parameters of the inclusions, in red the values set to filter out11 of 18 the aggregate’s inclusions. (b) View of the inclusions after segmentation and ﬁltering. the aggregate’s inclusions. (b) View of the inclusions after segmentation and filtering.

Figure 7. Sensitivity of the selected threshold on the volume of fibers and selection of the “right” Figure 7. Sensitivity of the selected threshold on the volume of fibers and selection of the “right” threshold based on the volume of fibers recovered from the sample. threshold based on the volume of fibers recovered from the sample. The columns in Figure8 present the analysis of distribution and orientation of the fibers for a The columns in Figure 8 present the analysis of distribution and orientation of the fibers for a fiber concentration of 1.5%, 2.5% and 3.5% by volume of bitumen. The subfigures presented in the first fiber concentration of 1.5%, 2.5% and 3.5% by volume of bitumen. The subfigures presented in the row (Figure8a) show a 3D vision of the fibers inside the cylindrical samples. Although they already first row (Figure 8a) show a 3D vision of the fibers inside the cylindrical samples. Although they give an indication about the formation of possible clusters, in the second row (Figure8b) the amount already give an indication about the formation of possible clusters, in the second row (Figure 8b) the of fibers per subspace is presented as charts. Similar heights in the columns of the chart indicate a amount of fibers per subspace is presented as charts. Similar heights in the columns of the chart homogenous distribution of the fibers. Conversely, a poor fiber distribution concentrates the fibers in indicate a homogenous distribution of the fibers. Conversely, a poor fiber distribution concentrates specific spots of the sample forming clusters. As a result, the columns of the chart should present large the fibers in specific spots of the sample forming clusters. As a result, the columns of the chart should differences, with the highest values representing clusters. This is the case of the sample with 3.5% of present large differences, with the highest values representing clusters. This is the case of the sample fibers, in which the subspace 4 has the smallest concentration and the subspace 8 the highest. It is then with 3.5% of fibers, in which the subspace 4 has the smallest concentration and the subspace 8 the evident that a cluster was formed on the top of the sample. This might indicate that using more than highest. It is then evident that a cluster was formed on the top of the sample. This might indicate that 2.5% of fibers might produce clusters. The orientation of the fibers is represented by a polar plot in using more than 2.5% of fibers might produce clusters. The orientation of the fibers is represented by Figure8c. The colors show the frequencies of the orientations projected on a sphere, with blue and red a polar plot in Figure 8c. The colors show the frequencies of the orientations projected on a sphere, as the least and most frequent orientations respectively. From the color distribution it can be observed with blue and red as the least and most frequent orientations respectively. From the color distribution that the fibers tend to orient towards a horizontal plane, i.e., perpendicular to the compacting force it can be observed that the fibers tend to orient towards a horizontal plane, i.e., perpendicular to the with the intensity increasing with fiber content. compacting force with the intensity increasing with fiber content.

Materials 2020, 13, 5731 12 of 17 Materials 2020, 12, x FOR PEER REVIEW 12 of 18

(a)

(b) 1.5% fiber 2.5% fiber 3.5% fiber

(c)

FigureFigure 8. (a )8. 3D (a) view3D view of theof the fibers fibers after after segmentation. segmentation. ( (bb)) Volume Volume of of fibers fibers for for each each subspace. subspace. (c) Polar (c) Polar plot showingplot showing the orientationthe orientation of theof the fibers. fibers. The The colors colors show the the frequencies frequencies of ofthe the fiber fiber orientations. orientations.

3.3. E3.3.ffect Effect of the of Metallic the Metallic Waste Waste on on the the Electrical Electrical and and ThermalThermal Properties Properties of of Asphalt Asphalt Mixtures Mixtures FigureFigure9a shows 9a shows all theall the electrical electrical resistivity resistivity results results measuredmeasured on on test test samples samples without, without, and andwith, with, differentdifferent metallic metallic fiber fiber contents. contents. The average average electr electricalical resistivity resistivity for the for reference the reference mixture mixturewas 2.05 was 8 2.05 × 108 ΩΩm,m, while while the the average average electrical electrical resistivity resistivity for test for samples test samples with 1.5%, with 2.5% 1.5%, and 2.5% 3.5% andof fiber 3.5% of × 7 6 4 fiber content was was 2.76 2.76 × 1010 Ω7 m,Ωm, 3.22 3.22 × 10 10Ω6mΩ andm and4.59 4.59× 10 Ω10m,4 Ωrespectively.m, respectively. Results Results show that show (1) that samples with metallic× fibers presented× an electrically conductive× behavior, and (2) the average (1) samples with metallic fibers presented an electrically conductive behavior, and (2) the average electrical resistivity decreased with the increase of the fiber content (Figure 9a), thus increasing the electricalelectrical resistivity conductivity decreased of samples, with theconsequent increase with of thethe fiberpercolation content theory (Figure on electrically9a), thus increasingconductive the electricalcomposites conductivity [36]. In ofsummary, samples, with consequent the range of with fiber the amounts percolation used in theory this study on electrically from 1.5% to conductive 3.5%, compositesthe percolation [36]. In point, summary, threshold with value the rangefrom where of fiber the amountselectrical resistivity used in this remains study constant from 1.5% with tothe 3.5%, the percolationcontent, was point, not observed. threshold In addition, value from Figure where 9b shows the electrical the average resistivity thermal conductivity remains constant results withand the content,error was bars not equivalent observed. to one In standard addition, deviation. Figure9 Thermalb shows conductivity the average was thermal measured conductivity on asphalt test results and errorsamples bars without, equivalent and towith, one different standard metallic deviation. fiber contents. Thermal Figure conductivity 9b shows was that, measured in general, on the asphalt test samplesthermal conductivity without, and decreases with, di withfferent the metallic addition fiberof metallic contents. fibers Figure with 9respectb shows to the that, reference in general, mixture; however, this reduction was not significant. The mixtures without fibers showed the highest the thermal conductivity decreases with the addition of metallic fibers with respect to the reference average thermal conductivity with a value of 1.293 Wm−1K−1 (SD = 0.106 Wm−1K−1). This result was mixture; however, this reduction was not significant. The mixtures without fibers showed the highest previously explained by Norambuena-Contreras et al., who related the air voids content increase 1 1 1 1 average thermal conductivity with a value of 1.293 Wm− K− (SD = 0.106 Wm− K− ). This result was previously explained by Norambuena-Contreras et al., who related the air voids content increase with the addition of metallic waste to the asphalt mixtures. Accordingly, the increase of air voids dissipates the heat in the mixtures, reducing the heat transmission of the mixtures [38]. Materials 2020, 12, x FOR PEER REVIEW 13 of 18

with the addition of metallic waste to the asphalt mixtures. Accordingly, the increase of air voids Materials 2020, 13, 5731 13 of 17 dissipates the heat in the mixtures, reducing the heat transmission of the mixtures [38]. Materials 2020, 12, x FOR PEER REVIEW 13 of 18

with11.E+10 x 10 the10 addition of metallic wasteAvg. to linethe asphalt mixtures. Accordingly, the increase of air voids 2.00 dissipates the heat in the mixtures, 0.0%reducing fibers the heat transmission of the mixtures [38]. 1.E+091 x 109 1.5% fibers 2.5% fibers 1.60 8 3.5% fibers 1.E+081 x 1011.E+10 x 1010 Avg. line 0.0% fibers 2.00 1.E+071 x 101.E+0917 x 109 1.5% fibers 1.20 2.5% fibers 1.60 1.E+061 x 101.E+0816 x 108 3.5% fibers 0.80 7 1.E+051 x 101.E+0715 x 10 1.20 1.E+061 x 106 0.40 1.E+041 x 104 0.80 Electricalresistivity (Ωm) 1.E+051 x 105 3 1.E+031 x 10 Thermalconductivity (W/mK) 0.00 0.40 1.E+041 0.0x 104 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0.01.52.53.5 Electricalresistivity (Ωm) 3Metallic fiber content (%) Metallic fiber content (%) 1.E+031 x 10 Thermalconductivity (W/mK) 0.00 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0.01.52.53.5 Metallic(a) fiber content (%) Metallic fiber(b) content (%)

Figure 9. Results of ( a) electrical (a) resistivity and (b) thermal conductivity ofof( b thethe) asphaltasphalt testtest samples.samples.

Figure 9. Results of (a) electrical resistivity and (b) thermal conductivity of the asphalt test samples. 3.4. Effect Effect of the Metallic Waste on the Mechanical Properties ofof AsphaltAsphalt MixturesMixtures Figure3.4. Effect 1010a aof showsthe Metallic that Waste the indirectindirecton the Mechanical tensiletensile stiPropertiesstiffffnessness modulusofmodulus Asphalt Mixtures (ITSM)(ITSM) decreasesdecreases with increasingincreasing temperatureFigure and 10a withwith shows increaseincrease that the inin indirect mixturemixture tensile fiberfiber sti contents.coffnessntents. modulus The physicalphysical (ITSM) decreases phenomenonphenomenon with increasing that explainsexplains these temperature resultsresults is is the theand temperature temperaturewith increase dependency dependencyin mixture of fiber bitumen of cobintents.tumen visco-elasticity, The visco-elasticity, physical i.e.,phenomenon a lossi.e., of a bitumen lossthat explainsof viscositybitumen andviscosity modulusthese and results whenmodulus is temperaturethe whentemperature temperature increases. dependency increases. Figure of 10bitumena Figure shows visco-elasticity, 10a that shows the addition that i.e., thea ofloss addition metal of bitumen fibers of metal from recycledfibersviscosity from tires recycled decreasesand modulus tires the decreaseswhen ITSM temperature of mixtures. the ITSM increases. The of averagemixtures. Figure ITSM 10aThe ofshowsaverage the fourthat ITSM the testing addition of temperaturesthe offour metal testing for thetemperatures mixturesfibers from without for recycled the fibers mixtures tires was decreases 4865 without MPa. the Conversely,ITSMfibers ofwas mixtures. the4865 average TheMPa. average ITSMConversely, consideringITSM of thethe four theaverage fourtesting testingITSM temperaturesconsideringtemperatures the and four allfor fibertesting the contentsmixtures temperatureswas without 2786 and MPa.fibers all Figurefiwasber contents4865 10b MPa. shows was Conversely, all 2786 the MPa. ITSM the Figure measured average 10b onITSMshows the test all considering the four testing temperatures and all fiber contents was 2786 MPa. Figure 10b shows all specimensthe ITSM measured in the longitudinal on the test (A-A’) specimens and orthogonal in the longitudinal (B-B’) directions, (A-A’) and respectively. orthogonal Figure (B-B’) directions,10b shows the ITSM measured on the test specimens in the longitudinal (A-A’) and orthogonal (B-B’) directions, that ITSM in A-A’ and B-B’ directions were similar almost for all the specimens tested. In other words, respectively.respectively. Figure Figure 10b 10b shows shows that that ITSM ITSM in in A-A’ A-A’ anandd B-B’ directionsdirections were were similar similar almost almost for allfor the all the the results indicate that including metal fibers recycled from tires does not induce anisotropy in the specimensspecimens tested. tested. In other In other words, words, the the results results indica indicatete that includingincluding metal metal fibers fibers recycled recycled from from tires tires axisdoes perpendicular doesnot inducenot induce anisotropy to A-A’anisotropy and in B-B’. inthe the Thisaxis axis behaviorperpendi perpendi hascularcular been to A-A’ reportedA-A’ and and inB-B’. B-B’. previous This This behavior researchbehavior has in has mixturesbeen been withreported crack-healingreported in previous in previous capabilities. research research in inmixtures mixtures with with crack-healing crack-healing capabilities. capabilities.

FigureFigure 10. E10.ffect Effect of of metal metal fiber fiber fromfrom recycled tires tires fiber fiber on on(a) (indirecta) indirect tensile tensile stiffness sti ffmodulusness modulus (b) anisotropy. (b) anisotropy. Figure 10. Effect of metal fiber from recycled tires fiber on (a) indirect tensile stiffness modulus (b) Theanisotropy. indirect tensile strength (ITS) results (Figure 11a) show consistency with ITSM. The highest dry ITS of 1.46 MPa was obtained in asphalt samples without steel fiber. The average dry ITS for the mixtures with fiber content of 1.5%, 2.5%, and 3.5% was 1.01 MPa, 0.89 MPa, and 0.84 MPa, respectively, indicating that ITS decreases with increasing fiber content. As expected, wet ITS was lower than dry Materials 2020, 12, x FOR PEER REVIEW 14 of 18

The indirect tensile strength (ITS) results (Figure 11a) show consistency with ITSM. The highest Materialsdry2020 ITS, 13of, 1.46 5731 MPa was obtained in asphalt samples without steel fiber. The average dry ITS for the14 of 17 mixtures with fiber content of 1.5%, 2.5%, and 3.5% was 1.01 MPa, 0.89 MPa, and 0.84 MPa, respectively, indicating that ITS decreases with increasing fiber content. As expected, wet ITS was ITS forlower asphalt than mixturesdry ITS for with asphalt 0.0%, mixtures 1.5%, andwith 3.5%0.0%, fiber 1.5%, content. and 3.5% However, fiber content. for However, 2.5% fiber for content 2.5% wet ITS werefiber highercontent than wet ITS dry were ITS. higher This result than dry is explained ITS. This result by the is lowerexplained air voids by the content lower air of voids asphalt content mixtures withof 2.5% asphalt fiber mixtures content with (Figure 2.5% 4fiber); which content was (Figure lower 4); thanwhich the was other lower mixtures than the other with mixtures fiber. The with lower air voidsfiber. content The lower reduces air voids the moisturecontent reduces susceptibility the moisture in asphalt susceptibility mixtures in asphalt making mixtures more di ffimakingcult water penetrationmore difficult during water the 72penetration h soaking. during A common the 72 h measuresoaking. A of common ITS water measure sensitivity of ITS iswater the indirectsensitivity tensile strengthis the ratio, indirect which tensile is thestrength ratio ratio, between which wet is th ITSe ratio and between dry ITS. wet Overall, ITS and alldry the ITS. measured Overall, all ITSR the was measured ITSR was within acceptable values (>80%) for asphalt mixtures, which has also been within acceptable values (>80%) for asphalt mixtures, which has also been observed in previous observed in previous studies [51]. studies [51].

Figure 11. (a)Effect of steel fiber on indirect tensile strength, and (b) corrosion of metal fibers after Figure 11. (a) Effect of steel fiber on indirect tensile strength, and (b) corrosion of metal fibers after soaking. The fibers were not coated with bitumen. soaking. The fibers were not coated with bitumen.

AfterAfter soaking, soaking, rust rust was was observed observed on on metal metal fibersfibers that that were were exposed exposed to towater water (Figure (Figure 11b). 11 Priorb). Priorto to waterwater soaking, soaking, there there were were no no signs signs of of rust rust on on thethe asphaltasphalt samples samples surface. surface. The The dimensions dimensions of the of fibers the fibers and airand voids air voids content content explain explain these these results. results. Recently, Li Li et et al al [52] [52 studied] studied the theeffect eff ofect moisture of moisture on on metallicmetallic fibers fibers used used in in asphalt asphalt concrete concrete with inductio inductionn properties. properties. The Theresults results showed showed that there that was there no was no corrosioncorrosion inside inside samplessamples mainly because because the the bitume bitumenn of ofthe the mixture mixture coated coated and andprotected protected the steel the steel fibersfibers from from water. water. The The average average dimensions dimensions of the fib fibersers in in Li Li et et al.’s al.’s (2020) (2020) research research were were 4.2 mm 4.2 in mm in lengthlength and and 0.07–0.13 0.07–0.13 mm. mm. In In thethe current research, research, the the length length of the of fibers the fibers ranged ranged from 11 fromto 40 mm, 11 to and 40 mm, the average diameter was 0.256 mm, making more difficult the coating with bitumen (Figure 11b). and the average diameter was 0.256 mm, making more difficult the coating with bitumen (Figure 11b). 4. Conclusions 4. Conclusions Waste tires pose a substantial environmental hazard worldwide. Therefore, finding avenues for Wastetheir reuse tires and pose recycling a substantial are sought. environmental Although there hazardhas been worldwide.substantial effort Therefore, to reuse crumb finding rubber avenues for theirin asphalt reuse pavements, and recycling little are effort sought. has been Although made to thererecycle has the been substantial substantial waste emetalffort tofibers reuse from crumb rubbertires. in asphaltThis work pavements, presented littlethe physical, effort has thermophysical, been made to recycleand mechanical the substantial properties waste of metalasphalt fibers frommixtures tires. This containing work presented different contents the physical, of metallic thermophysical, fiber obtained and from mechanical recycled tires. properties Based on of the asphalt mixturesresults, containing the following diff conclusionserent contents haveof been metallic obtained: fiber obtained from recycled tires. Based on the results, the following conclusions have been obtained: • In general, an increase in the fiber contents increases the air voids content and slightly reduces In general,the bulk an density increase of the in themixtures. fiber contents increases the air voids content and slightly reduces the • bulk• Mixing density and of thecompaction mixtures. was more difficult for higher fiber contents, which is attributed to less Mixingspace and for compaction the bitumen to was freely more flow di andfficult fill forthe highervoids of fiberthe mixtures. contents, which is attributed to less • space for the bitumen to freely flow and fill the voids of the mixtures.

Using the computed tomography technique, it was possible to identify clustering of the fibers • and that more fibers resulted in more clustering. Furthermore, it was possible to identify the orientation of the fibers that tend to orient towards a horizontal plane, i.e. perpendicular to the compacting force and with the intensity increasing with fibers content. This anisotropy would certainly have an effect in the mechanical performance of the material. Materials 2020, 13, 5731 15 of 17

Samples with metallic fibers presented an electrically conductive behavior; the average electrical • resistivity decreased with the increase of the fiber content, thus increasing the electrical conductivity of samples. The addition of metallic fibers did not affect the average thermal conductivity with respect to the • reference mixture significantly. Fiber content had a direct effect on the indirect tensile stiffness modulus (ITSM) that decreased • with increasing temperature for mixtures and with increase in fiber contents. ITS decreases with increasing fiber content; however, all the measured ITSR values were within • acceptable limits (>80%). Some fibers were not coated with bitumen; hence, fiber dimensions should be optimized to increase bitumen coating, reducing the risk of corrosion. Overall, the study shows that asphalt mixtures with metallic fiber from waste tires have the • potential to be used in asphalt mixtures with crack-healing purposes.

Author Contributions: J.N.-C., A.G. and L.P. conceptualized the work; A.G., J.N.-C., L.P. and M.A. developed the methodology; A.G., J.N.-C. and M.A. validated the data; J.N.-C., M.J.V., A.F. and J.V. developed the investigation; M.A., A.G. and J.N.-C. acquired the resources to conduct the research; M.J.V., M.A., A.F. and J.N.-C. curated the experimental data.; A.G. wrote the original draft; J.N.-C., A.G., M.A. and L.P. wrote, reviewed and edited the original version; J.N.-C., M.J.V., A.F. and M.A. developed the visualization; J.N.-C. and A.G. supervised the work; M.A. managed the administrative part of the project. All authors have read and agreed to the published version of the manuscript. Funding: Authors acknowledge the financial support given by the Leading House for the Latin American Region, Universität St. Gallen, Switzerland through the Seed Money Grant (SMG) project 1801. Acknowledgments: The authors acknowledge the experimental help provided by Patricio Pinilla, laboratory technician at Universidad Católica de Chile and researchers Victoria Valenzuela, Jose L. Concha and Irene Gonzalez-Torre from LabMAT of the Universidad del Bío-Bío. Conflicts of Interest: The authors declare no conflict of interest.

References

1. Raffoul, S.; Garcia, R.; Escolano-Margarit, D.; Guadagnini, M.; Hajirasouliha, I.; Pilakoutas, K. Behaviour of unconfined and FRP-confined rubberised concrete in axial compression. Constr. Build. Mater. 2017, 147, 388–397. [CrossRef] 2. Siddika, A.; Al Mamun, A.; Alyousef, R.; Amran, Y.M.; Aslani, F.; Alabduljabbar, H. Properties and utilizations of waste tire rubber in concrete: A review. Constr. Build. Mater. 2019, 224, 711–731. [CrossRef] 3. Sathiskumar, C.; Karthikeyan, S. Recycling of waste tires and its energy storage application of by-products—A review. Sustain. Mater. Technol. 2019, 22, e00125. [CrossRef] 4. Thomas, B.S.; Gupta, R.C. Long term behaviour of cement concrete containing discarded tire rubber. J. Clean. Prod. 2015, 102, 78–87. [CrossRef] 5. Muñoz-Sánchez, B.; Arévalo-Caballero, M.J.; Menor, M.C.P. Influence of acetic acid and calcium hydroxide treatments of rubber waste on the properties of rubberized mortars. Mater. Struct. 2016, 50, 75. [CrossRef] 6. Alsaleh, A.; Sattler, M.L. Waste Tire Pyrolysis: Influential Parameters and Product Properties. Curr. Sustain. Energy Rep. 2014, 1, 129–135. [CrossRef] 7. Winternitz, K.; Heggie, M.; Baird, J. Extended producer responsibility for waste tyres in the EU: Lessons learnt from three case studies—Belgium, Italy and the Netherlands. Waste Manag. 2019, 89, 386–396. [CrossRef] 8. Topçu, I.B.; Unverdi, A. Scrap tires/crumb rubber. In Waste and Supplementary Cementitious Materials in Concrete; Woodhead Publishing: Cambridge, UK, 2018. 9. Bobicki, E.; Liu, Q.; Xu, Z. Microwave heating of ultramafic nickel ores and mineralogical effects. Miner. Eng. 2014, 58, 22–25. [CrossRef] 10. Sienkiewicz, M.; Janik, H.; Borz˛edowska-Labuda, K.; Kucińska-Lipka, J. Environmentally friendly polymer-rubber composites obtained from waste tyres: A review. J. Clean. Prod. 2017, 147, 560–571. [CrossRef] 11. Pacheco-Torres, R.; Cerro-Prada, E.; Sánchez, F.E.; Varela, F. Fatigue performance of waste rubber concrete for rigid road pavements. Constr. Build. Mater. 2018, 176, 539–548. [CrossRef] Materials 2020, 13, 5731 16 of 17

12. Mittal, R.K.; Gill, G. Sustainable application of waste tire chips and geogrid for improving load carrying capacity of granular soils. J. Clean. Prod. 2018, 200, 542–551. [CrossRef] 13. Arulrajah, A.; Mohammadinia, A.; Maghool, F.; Horpibulsuk, S. Tire derived aggregates as a supplementary material with recycled demolition concrete for pavement applications. J. Clean. Prod. 2019, 230, 129–136. [CrossRef] 14. Arulrajah, A.; Mohammadinia, A.; Maghool, F.; Horpibulsuk, S. Tyre derived aggregates and waste rock blends: Resilient moduli characteristics. Constr. Build. Mater. 2019, 201, 207–217. [CrossRef] 15. Abbaspour, M.; Aflaki, E.; Nejad, F.M. Reuse of waste tire textile fibers as soil reinforcement. J. Clean. Prod. 2019, 207, 1059–1071. [CrossRef] 16. Norambuena-Contreras, J.; Aguilar, V.G.; Gonzalez-Torre, I. Physical and mechanical behaviour of a fibre-reinforced rubber membrane with self-healing purposes via microwave heating. Constr. Build. Mater. 2015, 94, 45–56. [CrossRef] 17. Norambuena-Contreras, J.; Silva-Robles, E.; Gonzalez-Torre, I.; Saravia-Montero, Y. Experimental evaluation of mechanical and thermal properties of recycled rubber membranes reinforced with crushed polyethylene particles. J. Clean. Prod. 2017, 145, 85–97. [CrossRef] 18. Liang, M.; Xin, X.; Fan, W.; Sun, H.; Yao, Y.; Xing, B. Viscous properties, storage stability and their relationships with microstructure of tire scrap rubber modified asphalt. Constr. Build. Mater. 2015, 74, 124–131. [CrossRef] 19. Cao, W. Study on properties of recycled tire rubber modified asphalt mixtures using dry process. Constr. Build. Mater. 2007, 21, 1011–1015. [CrossRef] 20. Yao, Z.; Zhang, J.; Gao, F.; Liu, S.; Yu, T. Integrated utilization of recycled crumb rubber and polyethylene for enhancing the performance of modified bitumen. Constr. Build. Mater. 2018, 170, 217–224. [CrossRef] 21. Medina, N.F.; Flores-Medina, D.; Hernández-Olivares, F. Influence of fibers partially coated with rubber from tire recycling as aggregate on the acoustical properties of rubberized concrete. Constr. Build. Mater. 2016, 129, 25–36. [CrossRef] 22. Samarakoon, S.S.M.; Ruben, P.; Pedersen, J.W.; Luís, E. Mechanical performance of concrete made of steel fibers from tire waste. Case Stud. Constr. Mater. 2019, 11, e00259. [CrossRef] 23. Sengul, O. Mechanical behavior of concretes containing waste steel fibers recovered from scrap tires. Constr. Build. Mater. 2016, 122, 649–658. [CrossRef] 24. Awolusi, T.; Oke, O.; Akinkurolere, O.; Sojobi, A.O. Application of response surface methodology: Predicting and optimizing the properties of concrete containing steel fibre extracted from waste tires with limestone powder as filler. Case Stud. Constr. Mater. 2019, 10, e00212. [CrossRef] 25. Caggiano, A.; Xargay, H.; Folino, P.; Martinelli, E. Experimental and numerical characterization of the bond behavior of steel fibers recovered from waste tires embedded in cementitious matrices. Cem. Concr. Compos. 2015, 62, 146–155. [CrossRef] 26. Gautam, P.K.; Kalla, P.; Jethoo, A.S.; Agrawal, R.; Singh, H. Sustainable use of waste in flexible pavement: A review. Constr. Build. Mater. 2018, 180, 239–253. [CrossRef] 27. Arabani, M.; Mirabdolazimi, S.; Sasani, A. The effect of waste tire thread mesh on the dynamic behaviour of asphalt mixtures. Constr. Build. Mater. 2010, 24, 1060–1068. [CrossRef] 28. Hassan, N.A.; Airey, G.D.; Hainin, M.R. Characterisation of micro-structural damage in asphalt mixtures using image analysis. Constr. Build. Mater. 2014, 54, 27–38. [CrossRef] 29. Wu, S.; Ye, Q.; Li, N. Investigation of rheological and fatigue properties of asphalt mixtures containing polyester fibers. Constr. Build. Mater. 2008, 22, 2111–2115. [CrossRef] 30. García, A.; Norambuena-Contreras, J.; Bueno, M.; Partl, M.N. Single and multiple healing of porous and dense asphalt concrete. J. Intell. Mater. Syst. Struct. 2015, 26, 425–433. [CrossRef] 31. Putman, B.J.; Amirkhanian, S.N. Utilization of waste fibers in stone matrix asphalt mixtures. Resour. Conserv. Recycl. 2004, 42, 265–274. [CrossRef] 32. Poulikakos, L.D.; Papadaskalopoulou, C.; Hofko, B.; Gschösser, F.; Falchetto, A.C.; Bueno, M.; Arraigada, M.; Sousa, J.; Ruiz, R.; Petit, C.; et al. Harvesting the unexplored potential of European waste materials for road construction. Resour. Conserv. Recycl. 2017, 116, 32–44. [CrossRef] 33. García, Á.; Bueno, M.; Norambuena-Contreras, J.; Partl, M.N. Induction healing of dense asphalt concrete. Constr. Build. Mater. 2013, 49, 1–7. [CrossRef] 34. Gallego, J.; Del Val, M.A.; Contreras, V.; Páez, A. Heating asphalt mixtures with microwaves to promote self-healing. Constr. Build. Mater. 2013, 42, 1–4. [CrossRef] Materials 2020, 13, 5731 17 of 17

35. Norambuena-Contreras, J.; Garcia, A. Self-healing of asphalt mixture by microwave and induction heating. Mater. Des. 2016, 106, 404–414. [CrossRef] 36. González, A.; Norambuena-Contreras, J.; Storey, L.; Schlangen, E. Effect of RAP and fibers addition on asphalt mixtures with self-healing properties gained by microwave radiation heating. Constr. Build. Mater. 2018, 159, 164–174. [CrossRef] 37. Menozzi, A.; García, Á.; Partl, M.N.; Tebaldi, G.; Schuetz, P. Induction healing of fatigue damage in asphalt test samples. Constr. Build. Mater. 2015, 74, 162–168. [CrossRef] 38. Norambuena-Contreras, J.; González, A.; Concha, J.; Gonzalez-Torre, I.; Schlangen, E. Effect of metallic waste addition on the electrical, thermophysical and microwave crack-healing properties of asphalt mixtures. Constr. Build. Mater. 2018, 187, 1039–1050. [CrossRef] 39. González, A.; Norambuena-Contreras, J.; Storey, L.; Schlangen, E. Self-healing properties of recycled asphalt mixtures containing metal waste: An approach through microwave radiation heating. J. Environ. Manag. 2018, 214, 242–251. [CrossRef] 40. González, A.; Valderrama, J.; Norambuena-Contreras, J. Microwave crack healing on conventional and modified asphalt mixtures with different additives: An experimental approach. Road Mater. Pavement Des. 2019, 20, S149–S162. [CrossRef] 41. Ajam, H.; Meijide, B.G.; Artamendi, I.; Garcia, A. Mechanical and healing properties of asphalt mixes reinforced with different types of waste and commercial metal particles. J. Clean. Prod. 2018, 192, 138–150. [CrossRef] 42. Hunter, R.N.; Self, A.; Read, J. The Shell Bitumen Handbook; ICE Publishing: London, UK, 2015. 43. Ketcham, R.A.; Carlson, W.D. Acquisition, optimization and interpretation of X-ray computed tomographic imagery: Applications to the geosciences. Comput. Geosci. 2001, 27, 381–400. [CrossRef] 44. Leu, L.; Berg, S.; Enzmann, F.; Armstrong, R.T.; Kersten, M. Fast X-ray Micro-Tomography of Multiphase Flow in Berea Sandstone: A Sensitivity Study on Image Processing. Transp. Porous Media 2014, 105, 451–469. [CrossRef] 45. Volume Graphics. VGStudio Max; Version 3.3; Volume Graphics: Heidelberg, Germany, 2020. 46. Garfi, G.; John, C.M.; Berg, S.; Krevor, S. The Sensitivity of Estimates of Multiphase Fluid and Solid Properties of Porous Rocks to Image Processing. Transp. Porous Media 2019, 131, 985–1005. [CrossRef] 47. Macchi, A. Problems in Classical Electromagnetism, 1st ed.; Springer: Cham, Switzerland, 2017. 48. British Standard BS EN 12697-26:2012. Bituminous Mixtures. Test Methods for Hot Mix Asphalt. Stiffness; BSI: London, UK, 2012. 49. González, A.; Cubrinovski, M.; Pidwerbesky, B.; Alabaster, D. Strength and Deformational Characteristics of Foamed Bitumen Mixes under Suboptimal Conditions. J. Transp. Eng. 2010, 137, 1–10. [CrossRef] 50. González, A.; Paniagua, F.; Thenoux, G.; Lopez, C. Evaluation of Additional Laboratory Tests for Design of Full-Depth Reclamation Mixtures Using Foamed Bitumen. Transp. Res. Rec. J. Transp. Res. Board 2016, 2573, 40–48. [CrossRef] 51. Norambuena-Contreras, J.; Serpell, R.; Vidal, G.V.; González, A.; Schlangen, E. Effect of fibres addition on the physical and mechanical properties of asphalt mixtures with crack-healing purposes by microwave radiation. Constr. Build. Mater. 2016, 127, 369–382. [CrossRef] 52. Li, H.; Yu, J.; Wu, S.; Liu, Q.; Wu, Y.; Xu, H.; Li, Y. Effect of moisture conditioning on mechanical and healing properties of inductive asphalt concrete. Constr. Build. Mater. 2020, 241, 118139. [CrossRef]

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional aﬃliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).