Characterization of Cold Recycled Asphalt Mixtures including Reinforcing Fibers

: In recent years, cold recycling techniques have been widely used all over the world thanks to their huge environmental advantages. However, their performances are lower than the traditional hot-mix asphalt, both for the longer time to develop the ﬁnal mechanical properties, which leads to delays in the reopening to road trafﬁc, and for the lower fatigue resistance. The present paper deals with the characterization of cold recycled asphalt mixtures (CRAM), made with 100% reclaimed asphalt pavement, where synthetic ﬁbers were included to improve the fatigue performance. The investigation involved the analysis of the curing time, volumetric properties, stiffness, strength, rheological behavior and resistance to cyclic loading. The results showed that the use of synthetic ﬁber, with the optimum dosage, determined a higher CRAM performance, especially in terms of fatigue resistance.


Introduction
Nowadays, cold recycling techniques are becoming more and more important thanks to their environmental and economic benefits.In particular, they combine the advantage of mixing and laying the material at room temperature together with the reuse of high percentages of reclaimed asphalt pavement (RAP), up to 100% [1].These advantages produce several benefits such as a reduction in the emissions, reduced contribution to landfills, virgin aggregate preservation and reductions in construction time and costs [2][3][4].
Cold mixtures for structural layers (not considering the cold mixtures for surface courses and pothole repair) can be classified into four different categories based on their composition [5]: Bitumen stabilized materials (BSM), where the bitumen/cement ratio (B/C) is greater than 1 and the bitumen content is lower than 2%; • Cement treated materials (CTM), in which only cement is used; • Cement-bitumen treated materials (CBTM), with a cement dosage higher than 1% and a bitumen/cement ratio B/C ≤ 1; • Cold asphalt mixtures (CAM), with a bitumen content higher than 2% and a cement content lower than 2%.
Moreover, based on the production method, cold mixtures can be classified into cold in-place recycling (CIR) and cold central plant recycling (CCPR) [6].
Cold recycled asphalt mixtures (CRAM) are CAM including high RAP contents, even 100%.Moreover, CRAM mixtures contain bitumen emulsion or foamed bitumen, water (to facilitate the blending process), active filler (Portland cement or hydrated lime) and, if necessary, natural aggregates for the correction of the gradation [7][8][9].Even if profitable from economic and environmental points of view, cold mixtures do not guarantee the same performance as hot-mix asphalt (HMA), since they need 3-7 days of curing to develop a good early resistance, necessary for the reopening to road traffic [10,11].Moreover, cold mixtures have a higher air-void content than HMA, which reflects in lower mechanical properties and, in particular, fatigue resistance [4,12].All these reasons have led to the use of CRAM materials only in base or sub-base layers, typically overlaid with not less than 10 cm of asphalt concrete [13].They could be used in more superficial layers, but only by solving these problems [12].
Numerous research works tried to solve the low early-stage strength issue.For example, an increase in cement dosage seemed to be an effective method to accelerate the curing rate, but entailed a higher brittleness and a significant shrinkage cracking [14].Du et al. proposed the use of fast hardening cement (e.g., sulfoaluminate cement) to reduce the curing time and increase the ITS of CRAM [15].Lin et al. found that the dynamic modulus of CRAMs in the first days of curing was significantly higher when surfactant and early-stage additives were used [14].Wang et al. improved early-stage strength using the microwave irradiation method, which not only accelerates cement hydration but also promotes asphalt emulsion demulsification [16].
The other big problem associated with CRAMs, as previously mentioned, is the low fatigue resistance.This problem is associated with the high air-void content in these materials, usually ranging between 10% and 15% [17].Several studies investigated the factors that influence the fatigue resistance of CRAM.Leandri et al. found that an increase in the cement content leads to an improvement in the fatigue resistance at low strain levels, but a worsening at high strain levels [18].Zhao et al. observed that the resistance to fatigue decreases as the RAP content and the stress level increase [19].Cheng et al. compared the fatigue properties of CRAM subjected to traffic loading and laboratory cyclic tests.They observed that the initial stiffness modulus can be used as an indicator to predict the in-service fatigue performance of the CRAM, while indirect tensile strength is not a reliable parameter [20].
One of the solutions to improve the fatigue resistance of HMA deals with the use of reinforcement fibers, which also showed a positive effect in terms of tensile strength and permanent deformation resistance [21,22].Moreover, some research showed that fiber can enhance the low-temperature crack resistance of the mixture [23,24].However, the effectiveness of the fiber reinforcement is strongly influenced by the length of the fiber.In particular, short fibers cannot properly reinforce the mixture, while long fibers tend to clump together [25].Few recent studies explored the possibility to use reinforcing fibers in CRAM.The first attempt to include fibers in cold mixtures was carried out by de S. Bueno et al., who added polypropylene fibers to a BSM.They obtained a decrease in the mix density, which resulted in a slight decrease in the strength properties and resilient modulus [26].Some years later, Ferrotti et al. carried out Marshall tests, indirect tensile tests and Abrasion tests on cold mix asphalt for pothole repair including three different types of fibers (cellulose, glass-cellulose, nylon-polyester-cellulose) with different dosages (0.15% and 0.30% by aggregate weight).The results showed that the mixture with 0.15% cellulose fibers provided similar or better performance than the reference mixture without fibers [27].The effects brought by the insertion of natural and synthetic fibers in cold asphalt mixtures for surface courses were investigated by Shanbara et al.; significant improvement was gained in terms of stiffness, creep resistance, resistance to permanent deformation and to the effect of water [28].More recently, research on fiber-reinforced CRAMs has undergone increased interest.In 2022, Kong et al. tested the mechanical properties of CRAMs modified with polyester fiber and basalt fiber.The results from Marshall, indirect tensile, semicircular bending and acoustic emission tests showed that both types of fibers improve the mix strength and delay the development of the crack state [29].Jiang et al. studied the effect of polyester fibers (0.4%) and cement (1.5%) in CRAM and found that their addition led to an improvement in the performance at both low and high temperatures [30].Wang et al. investigated the influence of the mixing process of CRAMs reinforced with basalt fibers on the resistance to cracking.In particular, various blends were produced by inserting the fibers in different steps.The first mix was made without adding fibers (reference), the second by mixing the fibers with the aggregates, the third by inserting the fibers after adding the water and before adding the emulsion, and finally the fourth inserting the fibers after adding the binder.The results showed how the insertion of the reinforcing fiber after the addition of the water and before the addition of the binder led to a better distribution within the mixture (as confirmed by the Scanning Electron Microscopy (SEM) analysis) and to higher fracture energy, splitting strength, and fracture work [31].The influence of the type of fiber and its content, within CRAMs, has been studied by Du.In particular, polyester fibers, polypropylene fibers, polyacrylonitrile fibers, lignin fibers and basalt fibers were investigated, with quantities ranging from 0 to 0.5% (in 0.1% increments).The results showed that polyester fibers overall guaranteed the best mechanical performance, while the optimum dosage of each fiber was close to 0.3% by aggregate weight [32].Zhu et al. also confirmed enhanced performances, especially in terms of fatigue resistance, for CRAMs reinforced with polyester and brucite fibers [24].
The research on fiber-reinforced cold mixture is still in the early stage.Due to the many variables, such as binder type and proportion, aggregate type (virgin and/or recycled), fiber type, geometry and dosage, the scientific community needs further investigation exploring the possibility of using fiber as a reinforcement in CRAM for structural pavement layers.

Objective
The objective of this study was the volumetric, mechanical, rheological and performance characterization of CRAM including synthetic fibers (a blend of aromatic polyamide, polypropylene monofilament and polyolefin).Two mixtures containing different fiber contents (0.05% and 0.1% by aggregate weight) were investigated and compared with a reference mix without fibers.Each mixture was manufactured according to the typical job mix formula adopted by the Italian national road authority, which provides a total water content of 5.0%, a cement content of 2.0% and a SBS-modified bitumen emulsion content of 4.0% [33].All the dosages are referred to as the aggregate weight.The comparison between the three CRAM mixtures was carried out in terms of:

Materials and Methods
The RAP was collected from an asphalt plant in Ancona (Italy).Figure 1 shows the RAP gradation.It can be observed that the RAP lacked fine particles, which is a rather frequent problem.For this reason, limestone filler with a dosage of 6% by aggregate weight was added (Figure 1).

Materials and Methods
The RAP was collected from an asphalt plant in Ancona (Italy).Figure 1 shows the RAP gradation.It can be observed that the RAP lacked fine particles, which is a rather frequent problem.For this reason, limestone filler with a dosage of 6% by aggregate weight was added (Figure 1).A cationic over-stabilized bitumen emulsion, containing 60% of SBS-modified bitumen (styrene-butadiene-styrene) and 40% of water and specifically designed for CRAM, was used.The emulsion characteristics are shown in Table 2.In order to accelerate the emulsion breaking and improve the short-term performance of the material, a pozzolanic cement type IV/A (P) with strength class 32.5R (EN 197-1) was added with a dosage of 2% by aggregate weight.
The synthetic fibers were a blend of aramid and polypropylene, complying with ASTM D8395-23 [34] (Figure 2).In particular, they are manufactured for HMA and cement concrete reinforcement.Since CRAM mixtures include both bituminous and cementitious binders, synthetic fibers are supposed to provide a good compatibility with the mix components.Moreover, they may contribute to the mix performance in the case of either hightemperature and room-temperature mix production.The dosage of 0.05% by aggregate weight replicated the dosage that is usually achieved with this kind of fibers in HMA mixtures [35].The fibers properties are shown in Table 3 while the CRAM mix composition is summarized in Table 4.A cationic over-stabilized bitumen emulsion, containing 60% of SBS-modified bitumen (styrene-butadiene-styrene) and 40% of water and specifically designed for CRAM, was used.The emulsion characteristics are shown in Table 2.In order to accelerate the emulsion breaking and improve the short-term performance of the material, a pozzolanic cement type IV/A (P) with strength class 32.5R (EN 197-1) was added with a dosage of 2% by aggregate weight.
The synthetic fibers were a blend of aramid and polypropylene, complying with ASTM D8395-23 [34] (Figure 2).In particular, they are manufactured for HMA and cement concrete reinforcement.Since CRAM mixtures include both bituminous and cementitious binders, synthetic fibers are supposed to provide a good compatibility with the mix components.Moreover, they may contribute to the mix performance in the case of either high-temperature and room-temperature mix production.The dosage of 0.05% by aggregate weight replicated the dosage that is usually achieved with this kind of fibers in HMA mixtures [35].The fibers properties are shown in Table 3 while the CRAM mix composition is summarized in Table 4.

Specimen Preparation
The mixtures were manufactured using an automatic mixer at 30 rpm.The laborato protocol provided the mixing of RAP, filler and fibers (when present) for 60 s; then, ment, half of the water content, bitumen emulsion, and the remaining water were add 60 s mixing was carried out after each addition.The visual analysis of the fiber-reinforc CRAM after mixing confirmed a good dispersion of the fibers without the adoption o specific addition procedure.
Specimens were prepared using a gyratory compactor equipment [40] with a gy tion speed of 30 rpm, an angle of 1.25°, a vertical pressure of 600 kPa and a mold diame of 150 mm.The specimen for ITSM and ITS tests were compacted with 100 gyrations a to a final height of approximately 70 mm (specimen mass of order 2840 g), while the sp imens for UCC and ITF tests were compacted with 120 gyrations and to a final height order 170 mm (specimen mass of order 6570 g).The specimens cured in the oven at 40 for a time ranging from 3 to 30 days, in order to assure an complete curing [41].Accord to Graziani et al., within 28 days the water in the specimens almost totally evaporates this curing conditions and this indicates the end of the curing time [42].To check this, ea specimen was periodically weighed over 30 days.At the end of the curing, before testi the tall specimens were cored to 100 mm diameter and the top and bottom bases were to obtain a specimen height of 130 mm. Figure 3 shows the pictures of the mix product and specimen preparation.

Specimen Preparation
The mixtures were manufactured using an automatic mixer at 30 rpm.The laboratory protocol provided the mixing of RAP, filler and fibers (when present) for 60 s; then, cement, half of the water content, bitumen emulsion, and the remaining water were added; 60 s mixing was carried out after each addition.The visual analysis of the fiber-reinforced CRAM after mixing confirmed a good dispersion of the fibers without the adoption of a specific addition procedure.
Specimens were prepared using a gyratory compactor equipment [40] with a gyration speed of 30 rpm, an angle of 1.25 • , a vertical pressure of 600 kPa and a mold diameter of 150 mm.The specimen for ITSM and ITS tests were compacted with 100 gyrations and to a final height of approximately 70 mm (specimen mass of order 2840 g), while the specimens for UCC and ITF tests were compacted with 120 gyrations and to a final height of order 170 mm (specimen mass of order 6570 g).The specimens cured in the oven at 40 • C for a time ranging from 3 to 30 days, in order to assure an complete curing [41].According to Graziani et al., within 28 days the water in the specimens almost totally evaporates in this curing conditions and this indicates the end of the curing time [42].To check this, each specimen was periodically weighed over 30 days.At the end of the curing, before testing, the tall specimens were cored to 100 mm diameter and the top and bottom bases were cut to obtain a specimen height of 130 mm. Figure 3 shows the pictures of the mix production and specimen preparation.

Test Procedures
The experimental program included the analysis of the compactability parameters, the evolution of the water content, the stiffness, and the strength during the curing time, finally, the complex modulus and the fatigue resistance at the end of curing.
The compactability was evaluated for each sample by analyzing the reduction in the air-void content during gyratory compaction.The air-void content (vm) was determined according to Equation (1): The dry density ρdry was calculated according to Equation (2): where mRAP, mfiller, mcement and mbitumen, respectively, indicate the mass of RAP, filler, cement and residual bitumen from the emulsion, while Vtot is the total volume of the specimen, measured by geometry.The maximum density ρmax was calculated according to Equation (3): where Pi and ρi are, respectively, the percentage and the density of the i-th component in the mix.According to Grilli et al., the dry density is an intrinsic material property that does not change with curing [43].With this approach, it is assumed that the volume occupied by water becomes air void as the curing goes on.
During the curing time, the sample mass was measured to monitor the water evaporation and assess the end of the curing process.In particular, each specimen was weighed after 3, 7, 14 and 30 days, in correspondence of the ITSM and ITS tests.
The ITSM of the specimens was measured using a Nottingham Asphalt Tester (NAT) according to EN 12697-26, Annex C [44].During the test, the cylindrical specimens were

Test Procedures
The experimental program included the analysis of the compactability parameters, the evolution of the water content, the stiffness, and the strength during the curing time, finally, the complex modulus and the fatigue resistance at the end of curing.
The compactability was evaluated for each sample by analyzing the reduction in the air-void content during gyratory compaction.The air-void content (v m ) was determined according to Equation (1): The dry density ρ dry was calculated according to Equation (2): where m RAP , m filler , m cement and m bitumen , respectively, indicate the mass of RAP, filler, cement and residual bitumen from the emulsion, while V tot is the total volume of the specimen, measured by geometry.The maximum density ρ max was calculated according to Equation (3): where P i and ρ i are, respectively, the percentage and the density of the i-th component in the mix.According to Grilli et al., the dry density is an intrinsic material property that does not change with curing [43].With this approach, it is assumed that the volume occupied by water becomes air void as the curing goes on.
During the curing time, the sample mass was measured to monitor the water evaporation and assess the end of the curing process.In particular, each specimen was weighed after 3, 7, 14 and 30 days, in correspondence of the ITSM and ITS tests.
The ITSM of the specimens was measured using a Nottingham Asphalt Tester (NAT) according to EN 12697-26, Annex C [44].During the test, the cylindrical specimens were subject to 15 load pulses (10 conditioning pulses and 5 test pulses) along the vertical diameter, with rise time of 124 ms.Loading was applied in a half-sine wave form to achieve a target horizontal deformation of 2 µm.The measurements were repeated along two diameters and an average ITSM was calculated.The test was performed at 20 • C and 3 repetitions under the same conditions were carried out.
The ITS tests were carried out through a servo-hydraulic device by applying a constant rate of deformation of 50 mm/min, according to EN 12697-23 [45].The test was performed at a temperature of 25 • C and on 3 replicates for each material and curing time.
The complex modulus of the CRAM mixtures was determined by means of a Universal Testing Machine UTM-30 device, in accordance with the American standard AASHTO R82 [46], at different temperatures (5, 20, 35 and 50 • C) and frequencies (20, 10, 5, 2, 1, 0.5, 0.2 and 0.1 Hz).Haversine compression load waves were applied to the specimen to achieve a vertical strain amplitude of 30 microstrain, measured as the average output of three LVDTs placed 120 • apart on the specimen lateral surface.
The ITF test was carried out in the control-stress mode using the UTM-30 device, according to EN 12697-24, Annex E [47].The specimens from ITF tests were obtained by extracting 100 mm diameter cores from 150 mm samples.Then, the cores were cut to obtain two specimens (top and bottom part).The ITSM was measured on one diameter of each specimen prior to the ITF test, in order to calibrate the stress level and achieve a number of cycles between 1000 and 1,000,000.For each load cycle, the device acquires the cumulative deformation of the specimen.The test protocol provided a temperature of 20 • C, a loading pulse duration of 0.1 s with a rest period of 0.4 s.Failure was established at the complete fracture of the specimen.

Compactability and Air Voids
Figure 4a illustrates the compaction curves, which depict the air-void content (average of 12 specimens for each mix) as a function of the number of gyrations, for the different CRAM mixtures.The curves showed the typical trend that can be approximated with a line in the semi-log scale (v m vs. logN) as for Equation (4): where v 1 is the air-void content at the first gyration and k represents the mix compactability.
The equation of the compaction curves for mixes REF, F0.05% and F0.1% are shown in the boxes in Figure 4a.It can be observed that the increase in fiber content led to a lower compactability, with the k values decreasing from 3.95 (REF) to 3.75 (F0.05%) to 3.57 (F0.1%).Differently, the v 1 values were comparable (order 32%) between the mixtures with and without fibers.The result was that the compaction curves were rather close in the left side of the graph, while they slightly departed when increasing the compaction energy.This indicates that the fiber filaments increased the internal friction between the CRAM solid particles, hindering their movement towards a denser configuration.
Figure 4b shows the average air-void content for each material at the end of the compaction.It can be noted that, a higher fiber content in the mixture led to higher air voids, because of the lower compactability.The reference material (REF) had the lowest air-void content of order 13.4%.When including 0.05% of fibers the average v m increased to 14.0% and for CRAM mix F0.1% it reached 15.3%.Despite there is a certain data scattering, the increase in the air-void content with the fiber dosage was statistically confirmed (t-test probabilities lower than 0.5%).In particular, the difference in the air-void content between the mixes F0.1% and REF was significant (order 2%) while between F0.05% and REF the different (approximately 0.6%) was less relevant.

Evolution of the Water Content
The amount of water lost from the specimens during the curing time was measured and reported in a graph as a function of the curing days (Figure 5).It can be noted that the water loss values after 30 days of curing were higher than 4%.Considering that the CRAM mixtures were manufactured with a total water content of 5% and that part of this water was captured by the cement hydration products, almost all the water in the specimens evaporated during the 30 days of curing at 40 °C.The data were simulated using the Michaelis-Menten (MM) model [48], defined by Equation ( 5): where f(t) is the parameter evolving over time, in this case the water loss, yA is the asymptotic value for t → ∞ (i.e., in the long-term) and Kc is the curing time when the parameter reaches a value equal to yA/2.A previous study demonstrated that the MM model can well simulate the evolving behavior of CRAM mixtures, with particular reference to the water loss, ITS and ITSM [49].From the results, it was observed that the measured water loss values after 30 days of curing were comparable to the yA from the MM model for all the mixtures, allowing assuming the complete curing.Indeed, the water loss values at 14 and 30 days were similar, indicating that most of the curing already ended after 14 days.Water

Evolution of the Water Content
The amount of water lost from the specimens during the curing time was measured and reported in a graph as a function of the curing days (Figure 5).It can be noted that the water loss values after 30 days of curing were higher than 4%.Considering that the CRAM mixtures were manufactured with a total water content of 5% and that part of this water was captured by the cement hydration products, almost all the water in the specimens evaporated during the 30 days of curing at 40 • C. The data were simulated using the Michaelis-Menten (MM) model [48], defined by Equation ( 5): where f (t) is the parameter evolving over time, in this case the water loss, y A is the asymptotic value for t → ∞ (i.e., in the long-term) and K c is the curing time when the parameter reaches a value equal to y A /2.A previous study demonstrated that the MM model can well simulate the evolving behavior of CRAM mixtures, with particular reference to the water loss, ITS and ITSM [49].From the results, it was observed that the measured water loss values after 30 days of curing were comparable to the y A from the MM model for all the mixtures, allowing assuming the complete curing.Indeed, the water loss values at 14 and 30 days were similar, indicating that most of the curing already ended after 14 days.

Evolution of the Water Content
The amount of water lost from the specimens during the curing time was measured and reported in a graph as a function of the curing days (Figure 5).It can be noted that the water loss values after 30 days of curing were higher than 4%.Considering that the CRAM mixtures were manufactured with a total water content of 5% and that part of this water was captured by the cement hydration products, almost all the water in the specimens evaporated during the 30 days of curing at 40 °C.The data were simulated using the Michaelis-Menten (MM) model [48], defined by Equation ( 5): where f(t) is the parameter evolving over time, in this case the water loss, yA is the asymptotic value for t → ∞ (i.e., in the long-term) and Kc is the curing time when the parameter reaches a value equal to yA/2.A previous study demonstrated that the MM model can well simulate the evolving behavior of CRAM mixtures, with particular reference to the water loss, ITS and ITSM [49].From the results, it was observed that the measured water loss values after 30 days of curing were comparable to the yA from the MM model for all the mixtures, allowing assuming the complete curing.Indeed, the water loss values at 14 and 30 days were similar, indicating that most of the curing already ended after 14 days.Water

The Indirect Tensile Stiffness Modulus
The average ITSM results for all the mixtures are shown in Figure 6.It is possible to notice that the ITSM increased during the curing until reaching a stable value after 14 days, as observed for the water loss.The data showed that there was no substantial difference between the mixtures REF and F0.05% in terms of stiffness at the various curing times.In particular, the ITSM ranged between approximately 3500 MPa (3-day cured) and 4900 MPa (30-day cured) for the CRAM without fibers and with 0.05% of fibers.Instead, the mix F0.1% had a lower stiffness at all the curing time.Specifically, the ITSM reduction was approximately 20%, with values ranging between 2800 MPa (3-day cured) and 3800 MPa (30-day cured).Also the y A parameter from MM model confirmed the lower stiffness for the F0.1% specimens.

The Indirect Tensile Stiffness Modulus
The average ITSM results for all the mixtures are shown in Figure 6.It is possible to notice that the ITSM increased during the curing until reaching a stable value after 14 days, as observed for the water loss.The data showed that there was no substantial difference between the mixtures REF and F0.05% in terms of stiffness at the various curing times.In particular, the ITSM ranged between approximately 3500 MPa (3-day cured) and 4900 MPa (30-day cured) for the CRAM without fibers and with 0.05% of fibers.Instead, the mix F0.1% had a lower stiffness at all the curing time.Specifically, the ITSM reduction was approximately 20%, with values ranging between 2800 MPa (3-day cured) and 3800 MPa (30-day cured).Also the yA parameter from MM model confirmed the lower stiffness for the F0.1% specimens.

The Indirect Tensile Strength
Figure 7 shows the ITS values for the different CRAM mixes as a function of the curing time.The average ITS values for each series of specimens increased over time until a horizontal asymptote after 14-30 days.After 3 days of curing, all the mixtures showed the same strength values (approximately 0.37 MPa).Instead, from 7 days to 30 days, the ITS values for F0.1% mixture (from 0.39 MPa to 0.43 MPa) were lower (approximately 15%) than those measured for REF and F0.05% mixtures (from approximately 0.47 MPa to 0.51 MPa).This difference in the strength properties was also noticed in the yA values from MM model.However, all the CRAM mixtures respected the minimum ITS value typically provided by Italian technical standards, that is 0.35 MPa after 7 days of curing at 40 °C.The comparison between the MM model parameters optimized during the simulation of water loss, ITSM and ITS data evolution allowed noticing that the Kc values were very similar between the mixtures with and without fibers.This denoted that the presence of the fibers did not influence either the water evaporation or, in general, the curing of the CRAM.So, the dissimilar behavior between the mixtures can be likely explained with the different air-void content, particularly for F0.1% mix.

The Indirect Tensile Strength
Figure 7 shows the ITS values for the different CRAM mixes as a function of the curing time.The average ITS values for each series of specimens increased over time until a horizontal asymptote after 14-30 days.After 3 days of curing, all the mixtures showed the same strength values (approximately 0.37 MPa).Instead, from 7 days to 30 days, the ITS values for F0.1% mixture (from 0.39 MPa to 0.43 MPa) were lower (approximately 15%) than those measured for REF and F0.05% mixtures (from approximately 0.47 MPa to 0.51 MPa).This difference in the strength properties was also noticed in the y A values from MM model.However, all the CRAM mixtures respected the minimum ITS value typically provided by Italian technical standards, that is 0.35 MPa after 7 days of curing at 40 • C. The comparison between the MM model parameters optimized during the simulation of water loss, ITSM and ITS data evolution allowed noticing that the K c values were very similar between the mixtures with and without fibers.This denoted that the presence of the fibers did not influence either the water evaporation or, in general, the curing of the CRAM.So, the dissimilar behavior between the mixtures can be likely explained with the different air-void content, particularly for F0.1% mix.

The Complex
The measured rheological data were plotted in the Black and Cole-Cole diagrams to evaluate the reliability of the experimental results and assess the thermo-rheological simplicity of the mixtures (Figure 8).The graphs report the results from all the tested specimens, highlighting the same CRAM type with a specific color and symbol.A first observation of the results allowed estimating the complex modulus norm |E*| and phase angle ϕ variations.In particular, the complex modulus approximately ranged between 800 MPa and 9000 MPa, while the phase angle ranged between 4° and 15°.These results are consistent with other studies [50,51], where CRAM mixtures showed a lower time and temperature sensitivity and a less pronounced viscous behavior compared to HMA.The graphs of Figure 8 show that the measured data were slightly dispersed and the alignment between the points corresponding to different temperatures were not perfect, especially for the mixtures containing fibers.However, these imprecisions were considered tolerable, as the contemporary presence of aged bitumen, virgin polymer-modified bitumen, cement hydration products and synthetic fibers made the system behavior very complex.For this reason, the materials were considered thermo-rheologically simple and the timetemperature superposition principle (TTSP) was applied to determine the complex modulus master curves and shift factor relationships.

The Complex Modulus
The measured rheological data were plotted in the Black and Cole-Cole diagrams to evaluate the reliability of the experimental results and assess the thermo-rheological simplicity of the mixtures (Figure 8).The graphs report the results from all the tested specimens, highlighting the same CRAM type with a specific color and symbol.A first observation of the results allowed estimating the complex modulus norm |E*| and phase angle φ variations.In particular, the complex modulus approximately ranged between 800 MPa and 9000 MPa, while the phase angle ranged between 4 • and 15 • .These results are consistent with other studies [50,51], where CRAM mixtures showed a lower time and temperature sensitivity and a less pronounced viscous behavior compared to HMA.The graphs of Figure 8 show that the measured data were slightly dispersed and the alignment between the points corresponding to different temperatures were not perfect, especially for the mixtures containing fibers.However, these imprecisions were considered tolerable, as the contemporary presence of aged bitumen, virgin polymer-modified bitumen, cement hydration products and synthetic fibers made the system behavior very complex.For this reason, the materials were considered thermo-rheologically simple and the timetemperature superposition principle (TTSP) was applied to determine the complex modulus master curves and shift factor relationships.

The Complex Modulus
The measured rheological data were plotted in the Black and Cole-Cole diagrams to evaluate the reliability of the experimental results and assess the thermo-rheological simplicity of the mixtures (Figure 8).The graphs report the results from all the tested specimens, highlighting the same CRAM type with a specific color and symbol.A first observation of the results allowed estimating the complex modulus norm |E*| and phase angle ϕ variations.In particular, the complex modulus approximately ranged between 800 MPa and 9000 MPa, while the phase angle ranged between 4° and 15°.These results are consistent with other studies [50,51], where CRAM mixtures showed a lower time and temperature sensitivity and a less pronounced viscous behavior compared to HMA.The graphs of Figure 8 show that the measured data were slightly dispersed and the alignment between the points corresponding to different temperatures were not perfect, especially for the mixtures containing fibers.However, these imprecisions were considered tolerable, as the contemporary presence of aged bitumen, virgin polymer-modified bitumen, cement hydration products and synthetic fibers made the system behavior very complex.For this reason, the materials were considered thermo-rheologically simple and the timetemperature superposition principle (TTSP) was applied to determine the complex modulus master curves and shift factor relationships.The rheological data were shifted with respect to time until the isothermal curves merged into a single smooth function at the reference temperature of 20 • C. The estimation of the temperature shift factors carried out according to the closed form shifting algorithms, based on the minimization of the area between two successive isothermal curves [52].
The complex modulus results were simulated using Huet-Sayegh (HS) model, which includes two spring and two parabolic elements [53].An upgraded version of the model was proposed by Graziani et al. for cold asphalt mixtures [51], including a dimensionless frequency-independent parameter that applies a correction to the phase angle.However, as the estimation of this parameter requires a high data accuracy, as the phase correction usually assumes values lower than 2 • , the base HS model was used because of the dispersion of the measured data.
The HS analytical expression of the Complex Young Modulus, at a specific temperature, is expressed by Equation ( 6): where E 0 is the static shear modulus when ω→0; E ∞ is the glassy shear modulus when ω→∞; ω = 2πf is the angular frequency; i is the imaginary unit defined by i 2 = −1; τ is the characteristic time; k, h and δ are dimensionless parameters such that 0 < k < h < 1.Based on the TTSP, τ can be determined as in Equation ( 7): where a T is the shift factor at the temperature T and τ 0 = τ(T 0 ) is determined at the reference temperature T 0 .Figures 9 and 10 show the master curves of |E*| and φ at the reference temperature of 20 • C. In the graphs, the points represents the shifted data measured for each specimen while the smooth curves depict the HS model of the single specimens.From Figure 9, it can be noted that mix F0.1% had lower stiffness than the other mixtures at all frequencies/temperatures.In particular, the difference between F0.1% and REF mix was more pronounced at high frequencies/low temperatures while the |E*| master curves of these mixes got closer at high frequencies/low temperatures.The mixture F0.05% showed |E*| values comparable to those of the REF mix at high frequencies/low temperatures, while at high frequencies/low temperatures, the stiffness of F0.05% was higher than that of the REF.Comparing F0.05% and F0.1% mixes, it was observed that the |E*| master curves were approximately equidistant (in the log-log scale).These results indicated that the presence of the fibers had an influence on the complex modulus of the mix at high temperatures by assisting the cement hydration products in hindering the particle mobility when the stiffness of the bituminous component (residual bitumen from emulsion and RAP binder) reduced.However, the excess of fibers, as in the case of F0.1%, determined an increase in air voids in the mix, which entailed decrease in the stiffness and mitigated the fiber effect.
From Figure 10, it can be observed that the phase angles of F0.1% and REF were the same at high frequencies/low temperatures.Instead, at low frequencies, F0.1% showed a lower phase angle.The mix F0.05% had the lowest phase angle values at all frequencies/temperatures denoting a more marked elasticity than the other materials.A possible explanation of these results can be associated to the contribution of the fibers in emphasizing the effect of the cement in the binder matrix, i.e., the increase in the mix stiffness and the mitigation of the viscous features.

Fatigue Resistance
The ITF test results (number of cycles to failure) were plotted in log-log-scale graphs, respectively as function of the horizontal stress (Figure 11) and as function of the horizontal strain measured on the undamaged specimen at the beginning of the test (Figure 12).For each material, a power function (with a linear trend in the log-log plane) was superimposed to the experimental data in order to describe the fatigue behavior.The results in Figure 11 showed that for high values of stress the fatigue resistance of the REF mix was slightly higher, vice versa for low stress values.When the applied stress was high and failure was quicker, the contribution of the fibers was probably attenuated and the effect of the different air-void content prevailed.When the stress was lower, the efficacy of the fibers in opposing to the specimen crack opening was higher, resulting in a lower slope of the fatigue curve.The mix F0.1% showed the same slope of the trend (exponents of the power function) of mix F0.05%, but the curve was positioned below, indicating a lower number of cycles to failure at all stresses.This was probably related to the lower stiffness, due to the higher air voids.
When plotting the data as a function of the initial horizontal strain (Figure 12), the CRAM including fibers showed a noticeably higher resistance to fatigue.The specimens of the mix F0.05% withstood a number of cycles approximately one-order higher than those of REF mix, for each strain value.The fatigue behavior of mix F0.1% was even improved at high strain values, despite the higher voids content, but at low strains it became

Fatigue Resistance
The ITF test results (number of cycles to failure) were plotted in two log-log-scale graphs, respectively as function of the horizontal stress (Figure 11) and as function of the horizontal strain measured on the undamaged specimen at the beginning of the test (Figure 12).For each material, a power function (with a linear trend in the log-log plane) was superimposed to the experimental data in order to describe the fatigue behavior.The results in Figure 11 showed that for high values of stress the fatigue resistance of the REF mix was slightly higher, vice versa for low stress values.When the applied stress was high and failure was quicker, the contribution of the fibers was probably attenuated and the effect of the different air-void content prevailed.When the stress was lower, the efficacy of the fibers in opposing to the specimen crack opening was higher, resulting in a lower slope of the fatigue curve.The mix F0.1% showed the same slope of the trend (exponents of the power function) of mix F0.05%, but the curve was positioned below, indicating a lower number of cycles to failure at all stresses.This was probably related to the lower stiffness, due to the higher air voids.
When plotting the data as a function of the initial horizontal strain (Figure 12), the CRAM including fibers showed a noticeably higher resistance to fatigue.The specimens of the mix F0.05% withstood a number of cycles approximately one-order higher than those of REF mix, for each strain value.The fatigue behavior of mix F0.1% was even improved at high strain values, despite the higher voids content, but at low strains it became

Fatigue Resistance
The ITF test results (number of cycles to failure) were plotted in two log-log-scale graphs, respectively as function of the horizontal stress (Figure 11) and as function of the horizontal strain measured on the undamaged specimen at the beginning of the test (Figure 12).For each material, a power function (with a linear trend in the log-log plane) was superimposed to the experimental data in order to describe the fatigue behavior.The results in Figure 11 showed that for high values of stress the fatigue resistance of the REF mix was slightly higher, vice versa for low stress values.When the applied stress was high and failure was quicker, the contribution of the fibers was probably attenuated and the effect of the different air-void content prevailed.When the stress was lower, the efficacy of the fibers in opposing to the specimen crack opening was higher, resulting in a lower slope of the fatigue curve.The mix F0.1% showed the same slope of the trend (exponents of the power function) of mix F0.05%, but the curve was positioned below, indicating a lower number of cycles to failure at all stresses.This was probably related to the lower stiffness, due to the higher air voids.
When plotting the data as a function of the initial horizontal strain (Figure 12), the CRAM including fibers showed a noticeably higher resistance to fatigue.The specimens of the mix F0.05% withstood a number of cycles approximately one-order higher than those of REF mix, for each strain value.The fatigue behavior of mix F0.1% was even improved at high strain values, despite the higher voids content, but at low strains it became similar to that of mix F0.05%.This result demonstrated the clear ability of the fibers to improve the fatigue resistance of the CRAM mixtures.Probably, the fibers allowed delaying the evolution of the microcracks into macrocracks by sewing the fracture edges, as was observed in HMA mixtures by many authors [54].Before reaching the specimen failure, the fibers in needed to be broken by or unthreaded from the binder matrix, entailing a significant increase in the necessary number of cycles.
Sustainability 2023, 15, x FOR PEER REVIEW 13 of 16 similar to that of mix F0.05%.This result demonstrated the clear ability of the fibers to improve the fatigue resistance of the CRAM mixtures.Probably, the fibers allowed delaying the evolution of the microcracks into macrocracks by sewing the fracture edges, as was observed in HMA mixtures by many authors [54].Before reaching the specimen failure, the fibers in needed to be broken by tension or unthreaded from the binder matrix, entailing a significant increase in the necessary number of cycles.

Conclusions
The present study aimed at characterizing CRAM mixtures including synthetic fibers with two different contents (0.05% and 0.1% by aggregate weight) in comparison with the reference material without fiber.The investigation included the analysis of mix compactability and air-void content, evolution of water loss, stiffness and strength during the curing time (up to 30 days) and rheological and fatigue behavior at the end of curing.
Based on the results shown in this paper, the following conclusions can be drawn:  The insertion of the fibers led to a reduction in mixture compactability and an increase in air voids.These differences were not significant with 0.05% fibers.


The insertion of 0.05% of fiber did not lead to a variation in ITSM and ITS values.
Instead, these values decreased (20% and 15%, respectively) when doubling the fiber content (F0.1%).similar to that of mix F0.05%.This result demonstrated the clear ability of the fibers to improve the fatigue resistance of the CRAM mixtures.Probably, the fibers allowed delaying the evolution of the microcracks into macrocracks by sewing the fracture edges, as was observed in HMA mixtures by many authors [54].Before reaching the specimen failure, the fibers in needed to be broken by tension or unthreaded from the binder matrix, entailing a significant increase in the necessary number of cycles.

Conclusions
The present study aimed at characterizing CRAM mixtures including synthetic fibers with two different contents (0.05% and 0.1% by aggregate weight) in comparison with the reference material without fiber.The investigation included the analysis of mix compactability and air-void content, evolution of water loss, stiffness and strength during the curing time (up to 30 days) and rheological and fatigue behavior at the end of curing.
Based on the results shown in this paper, the following conclusions can be drawn:  The insertion of the fibers led to a reduction in mixture compactability and an increase in air voids.These differences were not significant with 0.05% fibers.


The insertion of 0.05% of fiber did not lead to a variation in ITSM and ITS values.

Conclusions
The present study aimed at characterizing CRAM mixtures including synthetic fibers with two different contents (0.05% and 0.1% by aggregate weight) in comparison with the reference material without fiber.The investigation included the analysis of mix compactability and air-void content, evolution of water loss, stiffness and strength during the curing time (up to 30 days) and rheological and fatigue behavior at the end of curing.
Based on the results shown in this paper, the following conclusions can be drawn:

•
The insertion of the fibers led to a reduction in mixture compactability and an increase in air voids.These differences were not significant with 0.05% fibers.

•
The insertion of 0.05% of fiber did not lead to a variation in ITSM and ITS values.

•
The final stiffness and strength of the CRAM mixtures were reached after 14 days of curing at 40 • C, as confirmed by the analysis of the water loss, independently from the presence of fibers.

•
The LVE characterization showed that F0.05% and REF mixtures had a similar stiffness at high frequencies/low temperatures, while F0.05% showed a higher stiffness at low frequencies/high temperatures.Mix F0.1% had the lowest stiffness in the whole reduced frequency spectrum.• F0.05% had a slightly lower phase angle than the other materials at all frequencies/temperatures, denoting a higher elasticity.

•
The ITF test results showed that the fibers allowed a noticeable improvement in the fatigue resistance of the CRAM, probably related to the ability of the filaments in hindering crack propagation.
In conclusion, the use of fibers in CRAM mixtures can lead to a significantly higher performance against fatigue, without reducing (or even increasing) the other mix properties.Indeed, the correct fiber dosage must be accurately identified to achieve the desired mix behavior.The main limitation of this research was related to the use of just one fiber type, specifically high-quality fibers with aramid and polypropylene.Moreover, only one laboratory fatigue test protocol (ITF in control stress mode, at 20 • C and 0.1 s of load pulse duration) was applied for the characterization of the fatigue behavior.Due to the very promising results of this study, future works will involve CRAM mixtures including fibers of different materials and dimensions, in order to validate these findings with different types of reinforcement and identify the most effective.Then, the investigation will be extended to the full-scale application, with the construction of fiber-reinforced CRAM base layers in a trial section, the performance of which will be monitored over years.

Figure 4 .
Figure 4. Compaction curves (a) and air-void content (b) for the different CRAM mixes.

Figure 5 .
Figure 5. Water loss as a function of the curing time for the different CRAM mixtures.

Figure 4 .
Figure 4. Compaction curves (a) and air-void content (b) for the different CRAM mixes.

Figure 4 .
Figure 4. Compaction curves (a) and air-void content (b) for the different CRAM mixes.

Figure 5 .
Figure 5. Water loss as a function of the curing time for the different CRAM mixtures.

Figure 6 .
Figure 6.ITSM results for the different CRAM mixtures: (a) average values; (b) evolution with curing time.

Figure 6 .
Figure 6.ITSM results for the different CRAM mixtures: (a) average values; (b) evolution with curing time.

Figure 7 .
Figure 7. ITS results for the different CRAM mixtures: (a) average values; (b) evolution with curing time.

Sustainability 2023 , 16 Figure 7 .
Figure 7. ITS results for the different CRAM mixtures: (a) average values; (b) evolution with curing time.

Figure 11 .
Figure 11.Stress fatigue lines of each mixture.

Figure 12 .
Figure 12.Initial strain fatigue lines of each mixture.

Figure 11 .
Figure 11.Stress fatigue lines of each mixture.

Figure 12 .
Figure 12.Initial strain fatigue lines of each mixture.

Figure 12 .
Figure 12.Initial strain fatigue lines of each mixture.

Table 1 .
Nomenclature of the tested materials.

Table 3 .
Characteristics of the synthetic fibers.

Table 3 .
Characteristics of the synthetic fibers.
Stress fatigue lines of each mixture.