High-Temperature Rutting Resistance of Inverted Asphalt Pavement Structure

To improve the high-temperature rutting resistance of asphalt pavements, an inverted asphalt pavement structure (IAPS), 4 cm AC-13 mixture + 8 cm AC-25 mixture + 6 cm AC-20 mixture + 54 cm cement-stabilized macadam, was proposed herein by considering engineering practice, theoretical calculation, and analysis. A rutting prediction equation of asphalt pavements was then proposed via rut-development trends found by laboratory 18 cm thick rutting test. Subsequently, the rutting resistance of the IAPS was evaluated. *e results show that, compared with the traditional asphalt pavement structure (TAPS), 4 cm AC-13 mixture + 6 cm AC-20mixture + 8 cm AC-25mixture + 54 cm cement-stabilizedmacadam, the maximum shear stress of the IAPS can be reduced by ∼1.7% along with improvements in rutting resistance by ∼16% and ∼12% under wheel loads of 0.7 and 1.2MPa, respectively. Wheel-load increase affects the rutting resistance of both structures in a similar manner: when the wheel load increases from 0.7MPa to 1.2MPa, the rut depths of both pavement structures increase by at least 63%.*e IAPS clearly has better rutting resistance than the TAPS and is thus the better choice for asphalt pavement structure design.


Introduction
Asphalt pavements offer advantages such as a flat surface, comfortable driving, low noise, fast construction, and simple maintenance and repairs. erefore, they are widely used for paving roads all over the world [1,2]. However, as global temperatures continue to rise and as heavy-duty and channelized traffic become increasingly commonplace, the usage conditions of asphalt pavements have become increasingly demanding, resulting in the reduced service life of typical asphalt pavements. Nowadays, insufficient durability has become the primary problem associated with the usage of asphalt pavements [3], often caused by pavement rutting [4,5].
Road engineers throughout the world have investigated various strategies for improving the asphalt pavement's resistance to rutting at high temperatures . Additives such as styrene-butalene-styrene (SBS) [6], ethylene copolymer [7], epoxy resin [8], rock asphalt [9], nanoparticles [10], highmodulus modifier [11], crumb rubber [12], and anti-rutting additives [13] have been used to enhance the rheological properties of asphalt binders and to improve the rutting resistance of the asphalt mixtures. Kim et al. [14] and Bernier et al. [15] studied the effect of the amount of recycled asphalt pavement materials on the anti-rutting performance of polymer-modified asphalt. ey observed that the incorporation of recycled asphalt pavement materials can improve the rutting resistance of the asphalt mixtures. Hydrated lime [16,17], graphite [18], and steel slag [19,20] have been mixed into the asphalt mixtures to partially replace the mineral filler or the asphalt mixture. Subsequently, the high-temperature performances of the corresponding modified mixtures were evaluated. e results indicated that such mixtures exhibit better rutting resistance than the control asphalt mixture. Coleri et al. observed that dense gradation and large aggregate size improve aggregate interlocking, which can effectively disperse the shear stress distribution within the asphalt pavement, improving the shear resistance and the high-temperature performance [21]. Shanbara et al. found that, compared with the conventional cold and hot mixtures, the cold mix asphalt reinforced with the natural and synthetic fibers shows better resisting permanent deformation damage [22,23]. Reflective cool pavements can prevent the occurrence of urban heat islands as well as the occurrence of pavement rutting to a certain extent [24][25][26][27]. e aforementioned studies have undoubtedly helped improve the high-temperature rutting resistance of asphalt pavements by improving the high-temperature performance of the asphalt binder and asphalt mixtures or reducing the pavement temperature. However, they have rarely considered performing improvements by optimizing the asphalt pavement structure combination.
In China, the upper, middle, and lower layers of asphalt pavements have conventionally used aggregates in the order of fine, medium, and coarse, or in the order of fine, medium, and medium granular in asphalt mixtures. Furthermore, the asphalt mixture of each structural layer is designed to exhibit the complete range of functional requirements; i.e., each asphalt structural layer is designed to achieve good rutting resistance, cracking resistance, and moisture sensitivity. Researchers have observed that pavement rutting mainly appears in the middle layer of the asphalt pavement structure; therefore, the middle-layer mixtures must be designed to be the most resistant to rutting [28][29][30]. However, low-temperature crack resistance and moisture sensitivity of the middle-layer mixtures are also considered during the asphalt mixture design; thus, maximizing the rutting resistance of the middle layer is impossible. Moreover, a previous study has proved that the lower-layer mixtures mainly provide resistance to fatigue cracking in asphalt pavements; therefore, the lower-layer mixtures should exhibit good crack resistance. However, these mixtures are typically designed to provide temperature stability and moisture sensitivity; therefore, the crack resistance of these mixtures is not maximized [31]. erefore, it is significantly important to design the asphalt pavement structure according to the individual functional needs of each asphalt structural layer.
is study thereby intends to improve the typical semirigid base asphalt pavement structure with an inverted asphalt pavement structure (IAPS), wherein the middle and lower layers of the standard asphalt pavement are switched. Subsequently, the BISAR 3.0 software and a custom-built rut-testing rig were used to test the resistance of 18 cm thick slab specimen. e test results offer significant improvements in the asphalt pavement structure design.

Inverted Asphalt Pavement Structure and
Theoretical Verification 2.1. Inverted Asphalt Pavement Structure. Structural engineers have observed that, in three-layered asphalt pavements, ruts typically originate in the middle layer [28][29][30]. eoretical studies have also denoted that the maximum shear stress of the semi-rigid base asphalt pavement gradually shifts downward with an increase in load pressure; however, this stress is always observed in the middle layer [32,33]. erefore, the middle layer of asphalt pavements should have sufficient high-temperature rutting resistance [32,33]. e nominal maximum particle size of the aggregate significantly affects the high-temperature stability of an asphalt mixture. Large nominal particle sizes have been observed to improve the high-temperature rutting resistance of the asphalt mixture [34]. is observation inspired the IAPS, in which the bottom two layers of the pavement structure typically applied in China were switched. e properties of all the layers in both pavement structure types are presented in Table 1, where TAPS represents the traditional asphalt pavement structure.  [35], and the test temperature was set to 20°C. e Poisson's ratios of the monolithic materials, including the asphalt mixtures and the cement-stabilized macadam base, and the bulk material (soil base) were observed to be 0.25 and 0.4, respectively.

Calculation
Results. e stress distributions in the two asphalt pavement structures were calculated using the BISAR 3.0 software. e calculation results of the maximum shear stress in the two asphalt pavement structures are presented in Table 3, where σ TAPS and σ IAPS denote the maximum shear stress of the TAPS and IAPS, respectively. It is noteworthy that an asphalt pavement structure is considered to be a linear elastic layer system in these stress calculations, and the interlayer contact is modeled to be completely continuous. Table 3 shows that the maximum shear stress of the IAPS is lower than that of the TAPS by 1.7%. is indicates that the IAPS can improve the high-temperature rutting resistance of the asphalt pavement.

Raw Materials and Mixtures
3.1.1. Asphalt Binder. e SBS (I-C)-modified asphalt, produced by Zhejiang Yinji Petrochemical Co. Ltd. (Zhejiang Province, China), was used as the binder for the AC-13 asphalt mixture and as the gluing material between two asphalt layers. Class A-70# matrix asphalt, produced by the Zhenhai Refining & Chemical Company (Zhejiang Province, China), was used as the binder for the AC-20 and AC-25 asphalt mixtures. e technical properties of the two asphalt binders are presented in Table 4.

Aggregate.
Crushed basalt was used as the coarse aggregate for the AC-13 asphalt mixture, whereas crushed limestone was used as the coarse aggregate for the AC-20 and AC-25 asphalt mixtures. e fine aggregate and mineral powder that were used for the AC-13 and those for AC-20 and AC-25 asphalt mixtures were crushed limestone and limestone ore, respectively. e coarse aggregate, fine aggregate, and mineral powder were produced by the Jinhua Pan'an Stone Material Factory (Zhejiang Province, China), and each material satisfied the requirements mandated by the Technical Specification for Construction of Highway Asphalt Pavements (JTG F40-2004) [36].

Gradation.
e aggregate gradations of the asphalt mixtures that are used in this study are presented in Table 5.

Mix Design.
In this study, the AC-13, AC-20, and AC-25 asphalt mixtures were designed using the Marshall design procedure. e mixture design and road performance test results of the AC-13, AC-20, and AC-25 asphalt mixtures are presented in Tables 6 and 7, respectively, where VV, VMA, VFA, DS, ε B , MS 0 , and TSR denote the air void volume, voids in mineral aggregate, volume of voids filled with asphalt, dynamic stability, flexural tensile strain, retained Marshall stability, and freeze-thaw splitting strength ratio, respectively. VV, VMA, VFA, DS, ε B , MS 0 , and TSR were verified according to the Standard Test Methods of Bitumen and Bituminous Mixtures for Highway Engineering (JTG E20-2011) [35]. e results in Table 7 denote that the road performance indicators, such as DS, ε B , MS 0 , and TSR, of the asphalt mixtures used in this study satisfy the technical requirements of the Technical Specification for Construction of Highway Asphalt Pavements (JTG F40-2004) [36]. e dynamic stability of the AC-25 asphalt mixture is notably higher than that of the AC-20 asphalt mixture.

Preparation of the 18 cm
ick Slab Specimen. Considering that the traditional slab specimen forming instrument can produce slab specimens with a thickness of 3-10 cm, the 18 cm thick slab specimen was prepared using a custom-built slab specimen forming instrument. e main technical indexes and parameters of the custom-built instrument are identical to those of the traditional instrument; however, the custom-built instrument can produce slab specimens with a thickness of 3-20 cm.
To simulate the actual field conditions of the upper, middle, and lower asphalt layers, the slab specimen was prepared in the laboratory as follows: (1) e lower-layer slab specimen: first, a 300 mm × 300 mm layer of board (see Figure 1) was produced using the custom-built slab specimen forming instrument, satisfying the layer-wise thickness requirement denoted in Table 1. When the lower layer was AC-20 asphalt mixture, the thickness of the produced slab specimen was 6 cm; when AC-25 asphalt mixture was used as the lower layer, the    Advances in Civil Engineering 3 board thickness was 8 cm. en, the slab specimens were placed at room temperature for at least 24 h. Subsequently, the SBS-modified asphalt weighing 0.4 L/m 2 was evenly sprayed on the slab specimen surface and was cured for at least 2 h. (2) e middle-layer slab specimen: the middle-layer slab specimen was produced and applied on the lower-layer slab specimen sprayed with the sticky layer oil, with thicknesses as specified in Table 1.
When the AC-20 asphalt mixture was used as the middle layer, the board thickness was 6 cm; when the AC-25 asphalt mixture was used as the middle layer, its thickness was 8 cm. e SBS-modified asphalt weighing 0.4 L/m 2 was evenly sprayed on the surface of the prepared middle-layer slab specimen, which was subsequently cooled at room temperature for 24 h.
(3) e upper-layer slab specimen: finally, the 4 cm thick top-layer slab specimen was produced and applied on the middle-layer slab specimen sprayed with the sticky layer oil. Subsequently, it was cooled at room temperature for 48 h. e production process of the upper-layer slab specimen is depicted in Figure 2.
Each asphalt layer was produced with the optimal asphalt content and the mixture density specified in the Standard Test Methods of Bitumen and Bituminous Mixtures for Highway Engineering (JTG E20-2011) [35]. To ensure the reliability of the rutting tests, the porosity difference in all slab specimens was maintained within 1% from that of the respective standard cylindrical specimens [37]. e demolded 18 cm thick slab specimen is denoted in Figure 3.

Rutting Test.
e rutting test was conducted on a custom-made rutting tester for a thick slab specimen (18 cm thick). e main difference between the traditional rutting tester and the custom-made rutting tester is that the former can be used only to test the rutting depth of the slab specimen with a height of 5-10 cm, whereas the custommade rutting tester can be used to measure the rutting depth of the slab specimen with a height of 5-20 cm. Other test procedures of the custom-made and traditional rutting tester are identical. For the rutting test, the test temperature was 60°C, the loading rate was 42 times/min, the total testing time was 10 h, and the wheel loads were 0.7 and 1.2 MPa.
e Standard Test Methods of Bitumen and Bituminous Mixtures for Highway Engineering (JTG E20-2011) [35] specifies that the prepared slab specimen should be placed into a constant-temperature (60°C) oven before testing to ensure that the internal temperature of the slab specimen is equal to the temperature at which rutting occurs. Generally, when the slab specimen is 5-10 cm thick, the holding time of the slab specimen in a 60°C oven should be 5-12 h. However, the slab specimen investigated in this study is thicker than most other samples (at 18 cm), and we could find no reference data to indicate the duration for which thick slab specimens should be kept in the oven. If the holding time is too short, the internal temperature of the slab specimen will not reach the mandated test temperature; however, if the    holding time is too long, the binder and mixtures will age, affecting the reliability of the test results. erefore, the relationship between the holding time and the internal temperature of the 18 cm thick slab specimen was recorded in preliminary tests and is depicted in Figure 4. During the test, the temperature of the oven was maintained to be constant at 60°C. e internal temperatures between the upper and middle asphalt layers and those between the middle and lower asphalt layers were tested using temperature sensors that were embedded within the pavement layers during the slab specimen preparation. e temperature test procedure is shown in Figure 5.
As shown in Figure 4, when the holding time of a slab specimen is lower than 4 h, the internal temperatures of the 18 cm thick slab specimen increase sharply with increase in the holding time. When the holding time is 4 h, the internal temperatures of the slab specimen reach 55°C; furthermore, when the holding time is 8 h, the internal temperatures of the slab specimen are stable at ∼60°C, which is the temperature at which the loading tests are conducted. erefore, to ensure that the internal temperatures of the slab specimen are equal to the mandated rutting test temperature, we suggest that the holding time of an 18 cm thick slab specimen should be kept in an oven at 60°C for 8 h. Figure 6 shows the laboratory rutting test process of the 18 cm thick slab specimen.

Advances in Civil Engineering
Moreover, the rutting test results of slab specimens in different pavement structures are presented in Table 8.
As denoted by the results in Table 8, the cumulative rut depths of the two asphalt pavement structures under different wheel loads increase as the number of load cycles increases. Furthermore, the cumulative rut depths of the IAPS samples tend to be smaller than those of the TAPS when the load cycles and wheel loads are the same. Considering that the number of the testing samples is small, an analysis of variance test was conducted to examine the differences between the rut depths of two pavement structures [38,39]. Tables 9 and 10 present the significant test results of the two pavement structures under wheel loads of 0.7 MPa and 1.2 MPa, respectively.
As presented in Tables 9 and 10, the differences between the rut depths of the TAPS and IAPS samples that are subjected to the same number of load cycles are significant. ese findings show that the IAPS sample is more resistant against rutting than the TAPS sample.

Development of Rut.
According to the results presented in Table 8, the development trend of the cumulative rut depth of different pavement structures is plotted against the number of the loading cycles in Figure 7.
It can be seen from Figure 7 that the shapes of the rutdevelopment curves of both pavement structures are very similar when plotted against the increase of the number of loading cycles. When the number of loading cycles is less than 2550, the cumulative rut depths of both pavement structures increase sharply as the number of load cycles increases. However, after 2550 load cycles, the rate of increase in the cumulative rut depths of both structures decreases gradually until the rut depth stabilizes. e permanent deformation of asphalt mixture is mainly classified into three stages. During the first stage, the cumulative rut depths increase sharply as the number of load cycles increases. During the second stage, the rate of increase of the cumulative rut depths decreases gradually and eventually stabilizes. In the third stage, the strain rate grows rapidly with the increases of loading cycles. e material structure loses its stability under shear and finally results in the failure of the material [40]. us, Figure 7 only shows the first and second stage of the permanent deformation of asphalt mixture during the rutting experiment.

Rutting Prediction Equation.
e rut-development stages of asphalt pavement structures, as shown in Figure 7, indicate that there exists a horizontal asymptotic line during the second stage of permanent deformation of slab specimen in the rut-development curve. Furthermore, the corresponding value of the asymptote during the second stage of permanent deformation can be interpreted as the maximum rut depth (RD max ). We assume that the development of ruts during the first and second stages in asphalt pavement structures follows a rutting prediction equation, which should satisfy the following two boundary conditions: Where N denotes the number of loading cycles, RD(0) denotes the rut depth (mm) of the slab specimen before any loading has been applied, and RD(∞) denotes the rut depth (mm) of the slab specimen at the number of final loading cycles of the second stage.
Furthermore, we fitted the resulting curves to establish the following equation according to the aforementioned boundary conditions: where RD(N) denotes the rut depth (mm) of slab specimen after N loading cycles and RD max and ξ are the regression coefficients. Equation (1) was used to fit the data presented in Table 8, and the resulting values of the coefficients RD max , ξ, and R 2 are presented in Table 11. Table 11 shows that the correlation coefficients R 2 of the rutting prediction equation are greater than 0.98, indicating that the rutting prediction equation proposed in this study can accurately predict the development trend of rutting in the asphalt pavement structures.

Effect of Wheel Load on Pavement Rutting.
According to the values presented in Table 11, the effect of the wheel load on the maximum rut depth of the asphalt pavement structure is denoted in Table 12, where (RD 1.2 /RD 0.7 ) denotes the ratio of the maximum rut depth of the slab specimen under the wheel load of 1.2 MPa to that under the wheel load of 0.7 MPa. As shown in Table 12, when the wheel load is increased from 0.7 MPa to 1.2 MPa, the rut depths of the different pavement structures increase by 63% or 72% depending on the pavement layer structure.

Effect of the Pavement Structure on Pavement Rutting.
Based on the values presented in Table 11, the effect of the type of pavement structure with the same wheel loads on the maximum rut depth is shown in Table 13, where RD IAPS and RD TAPS denote the maximum rut depths of the IAPS and TAPS, respectively. Furthermore, (RD IAPS /RD TAPS ) denotes the ratio of the maximum rut depth of the IAPS to that of the TAPS under the same wheel load.
As shown in Table 13, the IAPS sample contained ruts that were 16% and 12% shallower than those in the TAPS sample under the same wheel loads of 0.7 MPa and 1.2 MPa, respectively.

Conclusions
An inverted asphalt pavement structure was proposed in this study, and the rutting resistance of the inverted asphalt pavement structure was evaluated. e following conclusions can be obtained from the abovementioned results.
(1) When compared with the TAPS, the maximum shear stress of the IAPS can be reduced by approximately 1.7%. ese results show that the IAPS can reduce the maximum shear stress of asphalt pavements. (2) Two rutting prediction equations of asphalt pavement structures were established, and the correlation coefficients R 2 exceeded 0.98. is high R 2 value indicates that the proposed rutting prediction equation can accurately predict the rut development in asphalt pavements.
(3) When compared with the TAPS, the rutting resistance of the IAPS under the wheel load of 0.7 MPa and 1.2 MPa can be increased by 16% and 12%, respectively. Furthermore, the increase in wheel load affects the rutting performances of both pavement structures in a similar manner. When the wheel load is increased from 0.7 MPa to 1.2 MPa, the rut depths of the different pavement structures increase by 63% or 72% depending on the pavement layer structure. e aforementioned tests compared the rutting resistance of the TAPS and IAPS with different numbers of load cycles (≤25500). e effect of more load cycles will be observed in our future studies to further verify the reliability of the proposed rutting prediction equations and the superiority of the IAPS. Moreover, the paper primarily focused on the rutting resistance of the IAPS; the fatigue property of the IAPS also will be studied in our further work.

Data Availability
e data used to support the findings of this study are included within the article.

Conflicts of Interest
e authors declare that there are no conflicts of interest regarding the publication of this paper.