Evaluation of the Contribution Rates of Different Phases on Pavement Performance of Cement-Casting Asphalt Mixture

Cement-casting asphalt mixture (CCAM) exhibits excellent antirutting properties and has been widely used to address the severe rutting distress across the world. However, CCAM’s failure mechanism has not been comprehensively understood. To this end, the contribution rates of porous asphalt mixture (PAM), cementitious network, and asphalt-cement interface phases to CCAM’s pavement performance are evaluated. A novel method was developed to fabricate the cementitious network with the aid of plastic balls. Abrasion loss, strength, peak force, and fracture energy of PAM, cement samples, and CCAM samples are obtained through dierent tests. Analysis results show that the contribution rate of cementitious network to Cantabro loss is around 60% and that of the interface phase is around 30%. Interface phase and PAM phase approximately contribute 90% to the overall strength and fracture energy of CCAM. With the increase of polymer contents in asphalt binder, the contribution rate of PAM phase increases and the contribution rate of interface decreases. Contribution rates results together with samples’ fracture surface observation verify that the interaction between asphalt skeleton and cement network plays a vital role during CCAM’s failure process. Attention should be given to the physical-chemical interactions between asphalt and cement in the future study.


Introduction
Cement-casting asphalt mixture has drawn great attention during the past several years in both research and application elds in China [1][2][3][4][5]. It has been widely utilized to address the severe rutting distress of asphalt pavement in heavy-duty sections, such as intersections, BRT lanes, and bus stops [6].
is composite material is fabricated by pouring cementitious slurry into the asphalt mixture skeleton, which takes good advantage of both asphalt mixture and cement concrete [7][8][9][10][11][12][13][14][15][16]. Generally, cement-casting asphalt mixture exhibits much better antirutting performance than asphalt mixture, while no joints-cutting is needed. However, eld applications have shown that cement-casting asphalt pavement exhibits relatively poorer antirutting cracking properties when compared with traditional asphalt mixture, which limits its further application [17]. Since then, a great number of researchers have been conducting research focusing on the failure mechanism of CCAM.
Hou evaluated the mechanical properties and durability characteristics of CCAM [9]. It was found that the brittleness of hardened cement paste would decrease the low-temperature cracking resistance of CCAM. e properties of CCAM are determined by the internal friction of asphalt mixture skeleton, the network structure of hardened cement paste, and the adhesion between porous asphalt mixture and hardened cement paste. Ding evaluated the in uence of volume parameter of asphalt mixture skeleton on the performance of CCAM [18]. As for the asphalt mixture skeleton with the same air voids but di erent pore structures, it is shown that the performance of CCAM with homogenous gradation is prior to that with consecutive gradation. Ding further investigated the mechanical behavior and failure mechanism of recycled CCAM with splitting tests [19]. e results revealed that the cracks initially appeared in the recovered binder and the hardened cement paste at the pore connected position. Modified asphalt binder would reduce the residual strain under cyclic loading and therefore delay the development of cracks. Cai evaluated the mechanical features of CCAM with different grouting materials and porous asphalt mixtures with different air voids [20]. e contribution rate and contribution efficiency of each part in the two-phase CCAM are calculated. Specifically, the contribution efficiency refers to the influence of each phase material per volume on the strength of the composite grouting material. According to Cai's research, it was found that the uniaxial compressive strength of CCAM is close to that of grout material, while CCAM's compressive modulus is slightly larger than that of control asphalt mixture. Although CCAM has much better antirutting performance than asphalt mixture, and the contribution rate and contribution efficiency of matrix asphalt mixture to CCAM are larger than those of grouting materials from the perspective of strength [21]. Considering that CCAM consists of open graded asphalt skeleton and the cement skeleton formed through the connected pores in porous asphalt, Cai employed micromechanics to distinguish the different reinforcing mechanism including cement filling skeleton interlock and physiochemical stiffening. Results illustrated that the interlock effect plays a vital role in the stiffness of SFP at low frequencies and porous asphalt mixtures contribute most to the stiffness at high frequencies.
e interlock effect would become more stable at higher frequencies as the porosity increases. Afterwards, the damage modes of CCAM under axial compression test were analyzed based on acoustic emission (AE) technique to investigate the failure mechanism of CCAM [22]. Based on these results, the damage processes of CCAM are divided into three phases: (1) compression phase, where the composite material is compressed with little AE signal or damage; (2) microcrack development phase, where the microcrack initiates and develops mostly along the interface between different phases in shearing cracking mode; and (3) macrocrack development phase, where nucleation of microcracks and development of tensile cracks take place to form macrocracks.
Based on the literature review above, it was clearly shown that the mechanical properties of CCAM are determined by asphalt mixture skeleton, grouting material, and the asphaltcement interface. e contribution rate of asphalt mixture skeleton and grouting material has been quantitatively analyzed [20]. However, the contribution rate of the interface has not been taken into consideration. To this end, this paper aims to develop a method to evaluate the contribution rate of three different phases in CCAM. e findings would hopefully provide an insight into the failure mechanism of CCAM.

Objectives
e main objective of this paper is to investigate the contribution rate of different phases on the properties of cement-asphalt composite and provide a guidance for the design of CCAM. Porous asphalt mixture (PAM), cement/ cement network, and CCAM cylindrical samples were fabricated in the first step. Afterwards, the pavement performances of different samples were obtained through Cantabro loss test, indirect tension (IDT) test, and semicircular bending (SCB) test. Finally, the contribution rates of different phases were calculated and compared. e research flowchart is illustrated in Figure 1.

Asphalt Binder.
ree different asphalt binders were used. eir basic properties are listed in Table 1. e contents of SBS polymer in SBS modified asphalt and high viscosity modified asphalt are 4% and 14%, respectively.

Aggregates.
Basalt aggregate is used for asphalt mixture fabrication. e aggregates with size from 10 to 16 mm are seized by passing the 9.5 mm mesh screen which is shown in Figure 2. e apparent density of aggregate is 2.94 g/cm 3 .

Asphalt Mixture.
e seized single-size aggregates were collected and mixed with asphalt binder at the binder-aggregate ratio of 3%. No fine aggregates or fillers were used. e aggregate was heated to 190°C and asphalt binder was heated to 140°C for base asphalt, 155°C for SBS-modified asphalt, and 165°C for high-viscosity-modified asphalt, respectively. Cylinder samples with two geometries (Φ101.6 mm * 63.5 mm and Φ152.4 mm * 100 mm) were prepared by Marshall compactor. e samples with the size of Φ101.6 mm * 63.5 mm are used for Cantabro loss test and IDT test, while the samples with the size of Φ152.4 mm * 100 mm are prepared for SCB test.
e air voids of all samples with different binders were measured and listed in Table 2. It can be seen that all samples exhibit similar air voids around 32%, which makes sure that the differences in the pavement performance of CCAM come from PAM and the interactions between cement and different binders. e influences of grouting material type and grouting material content in CCAM which is directly related to the air voids of PAM are not considered in this research.

Grouting Material.
Grouting material is developed by mixing the cement with filler and admixtures. e water/ binder ratio is set as 0.36. e grouting slurry is prepared by using a high shear mixer. Properties of grouting slurry are displayed in Table 3.

Cementitious Sample.
Cylinder cementitious samples were prepared by pouring the grouting slurry into the polyvinyl chloride (PVC) tubes with the same size as the Marshall asphalt mixture samples (shown in Figure 3). e bottom surfaces of PVC samples were sealed with tape.
To simulate the actual state of cement in CCAM, plastic balls with the diameter of 1 cm were used to fill in the PVC tubes and work as the skeleton (shown in Figure 4) [23].
Grouting materials were then poured into the PVA tubes. For PAV tubes with plastic balls, a metal mesh was used to cover the top of the tubes during grouting to prevent the plastic balls from oating out of the tubes. After curing for 7 days, the samples were got out of the tubes. e cementitious samples with and without balls are demonstrated in Figure 5. By counting the number of plastic balls, the volume of plastic ball skeleton can be calculated and the residue volume of

CCAM Samples.
e PAM samples were firstly sealed with tape. Grouting materials were then poured into the sealed PAM samples. In the end, these samples were cured in the ambient environment with a temperature of 22°C and humidity of 50% for 7 days.

Pavement Performance Characterization
Cantabro loss test and indirect tensile test were conducted according to Standard Test Methods of Bitumen and Bituminous Mixtures for Highway Engineering [24]. Semicircular bending test was conducted according to AASHTO Standard Method of Test for Determining the Fracture Potential of Asphalt Mixtures Using Semicircular Bend Geometry (SCB) at Intermediate Temperature [25].
IDT tests were conducted at 15°C and −10°C. e loading rates were set as 50 mm/min and 1 mm/min, respectively. Indirect tension strengths were calculated as follows: where ITS is indirect tension strength, MPa; P is peak force, N; and h is the height of Marshall sample, mm. For SCB tests, 25°C and −10°C were chosen as the test temperatures. e corresponding loading rates were 50 mm/ min and 1 mm/min, respectively. e height, thickness, and diameter of SCB specimen were 50 mm, 50 mm, and 150 mm, respectively. e notch depth and width are set as 15 mm and 1.5 mm, respectively.
Fracture energy (Gf ) can be expressed as G f is fracture energy, J/m 2 ; W f is the work of fracture, J. Pavement performances of different samples, including Cantabro loss, IDT strength, IDT fracture energy, SCB peak force, and SCB fracture energy, were then compared and analyzed.

Contribution Rate Calculation
e contribution rates of different phases are calculated as follows: where k A , k G , and k I are the contribution rates of PAM phase, grouting cementitious network phase, and interface phase, respectively. E A , E G , and E C represent the pavement performance (e. g., Cantabro loss, strength, peak force, and fracture energy) of PAM phase, grouting cementitious network phase, and CCAM, respectively. e contribution of interface phase on the overall strength of CCAM comes from the bonding between cement and asphalt as well as the friction between asphalt mixture skeleton and grouting network.

Cantabro Loss Test.
e Cantabro losses of base asphalt mixture (BAM), SBS-modified asphalt mixture (SAM), and high-viscosity-modified asphalt mixture (HAM) are demonstrated in Figure 6. As single-size aggregates were used, all PAM's Cantabro losses are higher than 80%. e antiraveling ability of PAM can be enhanced by using high-    Advances in Materials Science and Engineering viscosity-modified asphalt. It is clearly shown that cementgrouting would significantly decrease the Cantabro loss of PAM. As the viscosities of asphalt binders increase, the Cantabro losses of CCAM go down slightly. Table 4 illustrates the Cantabro losses of cement samples with and without plastic balls. It is found that cement sample has much better antiraveling ability than PAM but worse antiraveling ability than CCAM. e Cantabro loss of the cement sample with balls is about twice as much as that of the cement sample with no balls. Samples of CCAM and cement samples after test are shown in Figure 7.

IDT Test.
Force-displacement curves of PAM, CCAM, and cement samples are shown in Figure 8. For IDT test at 15°C, it is found that CCAM and cement sample without balls show similar peak forces. Cement sample with balls has the lowest peak force among all samples. Both the cement samples with and without balls display brittleness characteristics as the force drops quickly after peak force. When the test is conducted at −10°C, the peak force of cement sample without balls is close to that of CCAM. However, the displacement at the peak force of cement sample with balls is higher than that of cement sample without balls. is can be attributed to the existence of plastic balls. e balls hardly provide any contribution to the strength of CCAM at 15°C but may become brittle at low temperatures which impacts the failure process of cement sample with balls. Nevertheless, the peak force of cement sample without balls is still much higher than that of the cement sample with balls which proves that the stability of fabricated cementitious network is much lower than that of the integral cementitious sample. e pictures of the fractured samples are shown in Figure 9. It is shown that the cracks generally develop along with two routines for the cement sample without balls. e samples were fractured into three pieces. For the cement samples with balls, the crack mainly develops along with the plastic ball-cement interface and through the cement matrix. e plastic balls would distract the stress concentration during loading. Figures 10 and 11 demonstrate the IDT strength and fracture energy results of PAM and CCAM samples, respectively. Typically, using polymer-modified asphalt would slightly increase the strength of PAM but lower the strength of CCAM. However, both PAM and CCAM's fracture energies at 15°C would increase as more polymers were used in asphalt. At −10°C, SAM has the highest fracture energy among the three kinds of mixtures. HAM-PAM exhibits higher fracture energy than BAM-PAM, while HAM-CCAM's fracture energy is lower than that of BAM-PAM. is can be attributed to the failure mode of PAM. It has been pointed out in the literature that cohesive failure is the main cause at high temperatures, while adhesive failure is predominant at low temperatures of PAM. Although high viscosity asphalt binder was used, the failure mainly takes place in the asphalt-aggregate interface at low temperature [26,27]. us, the high cohesion energy of high viscosity asphalt binder cannot be fully taken advantage of. IDT strength and fracture energy of cement samples are      Table 5.
e strength of cement samples without balls is around six times higher than that of cement samples with balls at 15°C and about four times higher than that of cement samples with balls at −10°C. e 15°C fracture energy of cement sample without balls is much higher than that of cement samples with balls. However, the difference between the fracture energies of these two samples at −10°C was not as distinct as at 15°C. Advances in Materials Science and Engineering 6.3. SCB Test. SCB force-displacement curves of the samples are shown in Figure 12. At 25°C, CCAM sample has much higher strength than other samples. However, CCAM and cement samples without balls exhibit similar peak force values at −10°C. Regardless of the test temperature, the peak force value of the cement sample with balls is the lowest among all the samples. Unlike IDT test results, Figure 12 shows that CCAM behaves more like a cement sample rather than PAM. e differences can be attributed to the loading mode. e sample loses its strength due to tensile failure in IDT test, while the sample fails due to flexural bending in SCB test. e cementitious materials in CCAM make the composite materials more sensitive to flexural failure.
Pictures of failed samples are demonstrated in Figure 13. ere are visible cracks in CCAM, cement sample with balls, and cement samples without balls except PAM. It is shown that there are some similarities in the crack development patterns between CCAM and cement sample with balls. e cracks mainly develop across the cement phase and along the cement-aggregate/ball interface. SCB peak forces and fracture energies of PAM and CCAM samples are further displayed in Figures 14 and 15, respectively. It is found that using polymer would slightly increase the peak force of PAM at 25°C. e increase would become relatively distinct at −10°C. BAM-CCAM has the highest peak force among CCAM samples at 25°C. But the differences among the peak forces of BAM-CCAM, SAM-CCAM, and HAM-CCAM at −10°C are not significant. For fracture energy data, there is a clear trend that more polymer in asphalt binder would correspond to higher fracture energies in PAM and CCAM. It is shown that PAMs have higher fracture energies than CCAMs for all three kinds of mixtures at 25°C. is phenomenon shows that cementitious grouting would decrease the anticracking ability of PAM. Contrarily, all CCAMs show higher fracture energies than their PAMs. is verifies that different phases have different mechanical responses to loading at different temperatures. PAM becomes stiffer at low temperature and cementitious network can reinforce its anticracking ability. It is notable that all the PAM's fracture energy data show high variance. is can be explained as that the strength of PAM is formed by point-point stone contact [26,27]. e crack develops more randomly and scatteringly in PAM which can also be reflected in the PAM forcedisplacement curve in Figure 12. Table 6 illustrates the SCB peak force and fracture energy of cement samples at 25°C and −10°C. It is shown in the IDT test that cement samples have similar strengths at 15°C and −10°C. However, e SCB peak force of cement sample at −10°C is about twice higher than that at 25°C. Moreover, the peak force of cement samples without balls is around three to four times higher than that of cement samples with balls.
is again indicates that the failure mechanisms of samples in SCB and IDT tests are different. e SCB fracture energies of cement samples are much lower than those of PAM and CCAM. It seems that the cementitious network alone does not have the potential to make PAM as tough as CCAM. e interface phase should also have vital contributions to the anticracking ability of CCAM. However, the contribution rate of cement-asphalt interface has not been evaluated in the literature.

Contribution Rate Analysis.
e contribution rates of PAM phase, cementitious network phase, and interface phase are calculated according to equation (3) to (5) and illustrated in Table7 in sequence. As the cementitious network was fabricated for the first time, the cementitious network phase's contribution rate is acquired by using test data from both cement samples with balls and without balls to see if there are significant differences between them.
As can be seen in Table 7, the contribution rate of cement phase is much higher than that of cementitious network especially for IDT and SCB test results. By using the data from the cement sample without balls, the contribution from cement phase to the overall performance of CCAM is overestimated and the contribution from interface phase is underestimated. For Cantabro loss test result, PAM almost contributed nothing to CCAM's antiraveling ability. e contribution rate of cementitious network is around 60% and that of interface phase is around 30%. e contribution rates of different phases for BAM and SAM are close with each other. Unlike Cantabro loss test result, the interface phase contributes most to the IDT strength at 15°C and −10°C, while cementitious network phase contributes least. As more polymer modifiers are used, the contribution rate of interface phase to IDT strength decreases, while the contribution rates of PAM and cementitious network increase. When referring to SCB test results, similar trends can also be found. Higher viscosity of asphalt binder would lower the contribution rate of interface phase while increasing the other two phases' contribution rates. At −10°C, the contribution rate of cementitious network to SCB peak force is relatively stable. For SAM and HAM samples, PAM phases contribute more to the SCB peak force than the interface phases.
Due to space limitation, the contribution rates calculated using the data of cement sample without balls are not shown further. e contribution rates of the three phases to IDT and SCB fracture energy are demonstrated in Figures 16 and  17. e contribution rates of PAM in SAM and HAM are Besides PAM, the interaction between asphalt and cement also plays a vital role in the contribution to the IDT fracture energy especially at lower temperature. e bonding together with friction between asphalt mixture skeleton and grouting network becomes dominated. When looking at Figure 17, it is interesting to nd that the contribution rate of the interface phase to SCB fracture energy at 25°C is negative. is phenomenon can be explained as that the interface between PAM and cementitious network may be the weak zone of CCAM. At higher temperatures PAM would become soft and deform more under same loading. However, the existence of cementitious network would limit the deformation of PAM. Due to di erent modulus of cement and asphalt materials, the discontinuous displacement of cement network and PAM at the interface would induce stress concentration. e temperature would also in uence the bonding strength and friction between PAM and cementitious network. All these together contribute to the di erent contribution rates of interface phase at di erent temperatures. e grouted cementitious materials can be Advances in Materials Science and Engineering regarded as a reinforcing phase from the perspective of SCB fracture energy at 25°C. As discussed in Figure 13, the crack in PAM develops more randomly and scatteringly across the asphalt binder phase. Moreover, there exists a compression force from the loading strip in the top part of PAM during loading at 25°C. Since there are large-size air voids between aggregates in PAM, the space between aggregates would be squeezed. As reflected in the force-displacement curve, it will show that the PAM deforms more under loading. is crack development pattern and compression induced deformation     Advances in Materials Science and Engineering would together make the fracture energy higher. However, as pointed out by Cai, the compression in CCAM will be transformed into tension and interfacial shearing force [21]. As can be seen in Figure 18, the fracture surface of CCAM mainly contains cement fracture, asphalt binder cohesive failure, and asphalt-cement adhesive failure (marked as red). Both cement fracture and asphalt-cement adhesive failure would contribute much lower fracture energy than asphalt binder cohesive failure. At −10°C, all the contribution rates for SCB test results are positive. Similar to IDT results, PAM contributes most to the fracture energy at −10°C and the contribution rate keeps increasing as more polymers are used in the mixture.

Conclusions
e main objective of this paper is to evaluate the contribution rates of asphalt skeleton, cement, and interface phases on the pavement performance of CCAM. e cementitious network is simulated by using plastic balls with a diameter of 1 cm as single-seized aggregates. e Cantabro loss test, IDT test, and SCB test were employed to characterize the performance of PAM, CCAM, and cement samples with and without balls. e main conclusions are drawn as follows: (1) Cement sample with balls generally has higher Cantabro loss, lower IDT strength/fracture energy, and lower SCB peak force/fracture energy than cement sample without balls. (2) By using the data from cement sample without balls, it is found that the contribution from cement phase to the overall performance of CCAM is overestimated and the contribution from interface phase is underestimated. (3) e contribution rate of cementitious network to Cantabro loss is around 60% and that of interface phase is around 30%. Interface phase and PAM phase approximately contribute 90% to the overall strength and fracture energy of CCAM. With the polymer contents in the asphalt binder increasing, the contribution rate of PAM phase increases and the contribution rate of interface decreases. (4) It is inferred that the formation of an interface between PAM and cementitious network would change the developing routines of cracks in PAM. e crack in PAM is formed due to asphalt binder cohesion failure between aggregates under tensile force, while the crack in CCAM consists of asphalt binder  cohesion failure, cement failure, and asphalt-cement interface adhesive failure. e result in this paper verifies that the asphalt-cement interface adhesive failure is nonnegligible.

Data Availability
Data are available upon request from the authors.

Additional Points
Limitations.
(1) As only one type of grouting material is used in this paper, the contribution rate of cementitious network is relatively stable. Combinations of different types of asphalt binder and grouting materials can be used in future research to comprehensively evaluate the contribution rates of different phases in CCAM. (2) Although it is assumed that light-weight plastic balls would have no interactions with cement and they can be regarded as air bubbles in the samples, some clues from the force-displacement curve in −10°C IDT test were found that the elasticity of plastic balls may slightly impact the test results. (3) e aggregate is not perfectly round and the diameter is not exactly 1 cm. erefore, the air void of the fabricated PAM is 32%, while the volume within the plastic ball skeleton is 17%. ere is still a gap between PAM and the fabricated plastic ball skeleton.

Conflicts of Interest
e authors declare that they have no conflicts of interest.