Novel Gradation Design of Porous Asphalt Concrete with Balanced Functional and Structural Performances

: To improve the permeability of porous asphalt concrete (PAC) with a small nominal maximum aggregate size (NMAS) of 10 mm (PAC10), a novel gradation design by excluding the 0.075–3 mm aggregate was developed. This study aims to evaluate the functional and structural performances of the novel PAC10 with various mineral ﬁller contents, using the conventional PAC10 and 13 mm NMAS PAC (PAC13) as reference, and develop the optimum gradation of the novel PAC10. The performance properties evaluated include moisture susceptibility, durability, high-temperature stability, low-temperature cracking resistance and permeability. The results indicated that for the two conventional PACs with the same ﬁne aggregate and mineral ﬁller content, PAC10 had worse permeability and rutting resistance, similar moisture susceptibility and durability, and better low-temperature cracking resistance, compared with the PAC13. The novel PAC10 showed better permeability than the conventional PAC10. With the increase of the mineral ﬁller content, the structural performance of the novel PAC10 is improved, but its permeability is decreased. With a mineral ﬁller content of 6%, the novel PAC10 can have balanced functional and structural performances, which are equivalent to those of the conventional PAC13.


Introduction
Porous asphalt concrete (PAC) has been widely used as a road surfacing material because of its various beneficial functions, such as low traffic noise, good visibility and better friction, obtained by the drainage capability of the mixture and by the macro-texture that is created due to the gradation in rainy days [1][2][3][4][5][6][7]. These attractive functions are mainly attributed to its high air void content (typically 15−20%) that allows water to penetrate into the material and absorbs noise. However, the high air void content also decreases durability of PAC, because water and oxygen have larger contact surface area with the mixture [8][9][10]. Thus, it is important to balance the functional performance (i.e., permeability) and structural performance of PAC in mixture design [11].
Many researches have been carried out to improve the overall performance of PAC. It has been reported that some measures were effective in improving the structural performance of PAC, such as using high-viscosity binder, fiber and so on [12,13], while aggregate gradation plays an important role in both the structural and functional performance of PAC [14,15]. Aggregate gradation is very important for PAC, because it controls the porosity of the aggregate structure, which in turn affects the permeability of PAC. The aggregate selected in this study was crushed basalt, which is widely used in the surface layers of highways in China. Table 2 lists the properties of the coarse and fine aggregate, which all meet the Chinese specification requirement. The mineral filler used in this study was from limestone; its properties are presented in Table 3.

. Additives
A common type of high viscosity additive (HVA) and polyester fiber were used to improve the durability of PAC. The HVA is a kind of polymer modifier, which can improve the kinetic viscosity of bitumen at 60 • C obviously. The HVA content was 8% weight of binder, and the dosage of the fiber was 0.25% by weight of aggregate.

Optimum Asphalt Content
According to the design method of PAC in China, the allowable range of asphalt content is determined based on the results of the Cantabro abrasion test and Schellenberg binder draindown test. The trial asphalt content was determined based on the assumed asphalt film thickness of 13 um and total aggregate surface area. Five trial mixtures of varying asphalt contents in 0.5% increment were prepared for the two tests, and each test was repeated three times at each trial asphalt content. As Figure 1 illustrates, the minimum asphalt content is determined as the content corresponding to the inflection point of the Cantabro loss curve, and the maximum asphalt content is determined as the content corresponding to the inflection point of the draindown ratio curve [12].
test. The trial asphalt content was determined based on the assumed asphalt film thickness of 13 um and total aggregate surface area. Five trial mixtures of varying asphalt contents in 0.5% increment were prepared for the two tests, and each test was repeated three times at each trial asphalt content. As Figure 1 illustrates, the minimum asphalt content is determined as the content corresponding to the inflection point of the Cantabro loss curve, and the maximum asphalt content is determined as the content corresponding to the inflection point of the draindown ratio curve [12]. Typically, the optimum asphalt content (OAC) of the PAC mixture is selected between the minimum and maximum asphalt contents. However, to provide better durability, the maximum asphalt content can be selected to produce larger asphalt film thickness. In addition, the PAC should also meet the mixture design requirements, such as the minimum Marshall stability of 3.5kN, the minimum tensile strength ratio (TSR) of 75%, and the target air void content, which was 21% in the study.

Schellenberg Binder Draindown Test
Since PAC has very few fine particles, too much asphalt will leads it to flow off the aggregates, which known as draindown. Therefore, the Schellenberg binder draindown tests were carried out on mixes at a mixing temperature of 185 °C to evaluate the binders' draindown potential according to T0732−2011(JTG E20-2011) [18].

Cantabro Abrasion Test
The Cantabro abrasion test is widely used to evaluate the durability of PAC [12]. The mass loss during this test is defined as the Cantabro mass loss, which is calculated as follows: where ∆S = Cantabro mass loss, %; m0 = sample weight before test, g; and m1 = sample weight after test. Typically, the optimum asphalt content (OAC) of the PAC mixture is selected between the minimum and maximum asphalt contents. However, to provide better durability, the maximum asphalt content can be selected to produce larger asphalt film thickness. In addition, the PAC should also meet the mixture design requirements, such as the minimum Marshall stability of 3.5 kN, the minimum tensile strength ratio (TSR) of 75%, and the target air void content, which was 21% in the study.

Schellenberg Binder Draindown Test
Since PAC has very few fine particles, too much asphalt will leads it to flow off the aggregates, which known as draindown. Therefore, the Schellenberg binder draindown tests were carried out on mixes at a mixing temperature of 185 • C to evaluate the binders' draindown potential according to T0732-2011 (JTG E20-2011) [18].

Cantabro Abrasion Test
The Cantabro abrasion test is widely used to evaluate the durability of PAC [12]. The mass loss during this test is defined as the Cantabro mass loss, which is calculated as follows: where ∆S = Cantabro mass loss, %; m 0 = sample weight before test, g; and m 1 = sample weight after test. The Cantabro test was conducted in accordance with the Chinese specifications T0733-2011 (JTG E20-2011) [18], and the specimens were conditioned in the following three ways in this study:

1.
Standard Cantabro test: specimens were tested after being immersed at 20 • C water for 20 h; 2.
Immersion Cantabro test: specimens were tested after being immersed at 60 • C water for one, two, and three days; 3.
Freeze-thaw Cantabro test: specimens were tested after one, two, and three freeze-thaw cycles.
For the freeze-thaw condition, the immersed specimens are placed in a programmable high and low temperature test chamber, as shown in Figure 2. One cycle temperature condition is that the temperature in the chamber is at 60 • C for 12 h and then at −18 • C for 12 h.

Immersion
Cantabro test: specimens were tested after being immersed at 60 °C water for one, two, and three days; 3. Freeze-thaw Cantabro test: specimens were tested after one, two, and three freeze-thaw cycles.
For the freeze-thaw condition, the immersed specimens are placed in a programmable high and low temperature test chamber, as shown in Figure 2. One cycle temperature condition is that the temperature in the chamber is at 60 °C for 12 h and then at −18 °C for 12 h. In this study, the standard Cantabro test was used to determine the minimum asphalt content of PAC, while the immersion and freeze-thaw Cantabro tests were used to evaluate the durability of PAC. The Cantabro abrasion test equipment and samples after Cantabro abrasion test are shown in Figure 3.

Moisture Susceptibility Test
The moisture resistances of PAC10 mixtures were evaluated by the TSR test according to the modified AASHTO T283 [23]. More specifically, the freeze-thaw condition in AASHTO T283 was modified to that for the aforementioned freeze-thaw Cantabro test, and the TSR tests were carried out after one, two, and three freeze-thaw cycles.

High-Temperature Permanent-Deformation Test
The high-temperature rutting resistances of the PCA10 mixtures were studied by the wheel tracking test in accordance with T0719-2011 (JTG E20-2011) [18]. The test slab has a dimension of 300 × 300 × 50 mm, and the wheel tracking tests were conducted under the standard condition, i.e., In this study, the standard Cantabro test was used to determine the minimum asphalt content of PAC, while the immersion and freeze-thaw Cantabro tests were used to evaluate the durability of PAC. The Cantabro abrasion test equipment and samples after Cantabro abrasion test are shown in Figure 3.

Immersion
Cantabro test: specimens were tested after being immersed at 60 °C water for one, two, and three days; 3. Freeze-thaw Cantabro test: specimens were tested after one, two, and three freeze-thaw cycles.
For the freeze-thaw condition, the immersed specimens are placed in a programmable high and low temperature test chamber, as shown in Figure 2. One cycle temperature condition is that the temperature in the chamber is at 60 °C for 12 h and then at −18 °C for 12 h. In this study, the standard Cantabro test was used to determine the minimum asphalt content of PAC, while the immersion and freeze-thaw Cantabro tests were used to evaluate the durability of PAC. The Cantabro abrasion test equipment and samples after Cantabro abrasion test are shown in Figure 3.

Moisture Susceptibility Test
The moisture resistances of PAC10 mixtures were evaluated by the TSR test according to the modified AASHTO T283 [23]. More specifically, the freeze-thaw condition in AASHTO T283 was modified to that for the aforementioned freeze-thaw Cantabro test, and the TSR tests were carried out after one, two, and three freeze-thaw cycles.

High-Temperature Permanent-Deformation Test
The high-temperature rutting resistances of the PCA10 mixtures were studied by the wheel tracking test in accordance with T0719-2011 (JTG E20-2011) [18]. The test slab has a dimension of 300 × 300 × 50 mm, and the wheel tracking tests were conducted under the standard condition, i.e.,

Moisture Susceptibility Test
The moisture resistances of PAC10 mixtures were evaluated by the TSR test according to the modified AASHTO T283 [23]. More specifically, the freeze-thaw condition in AASHTO T283 was modified to that for the aforementioned freeze-thaw Cantabro test, and the TSR tests were carried out after one, two, and three freeze-thaw cycles.

High-Temperature Permanent-Deformation Test
The high-temperature rutting resistances of the PCA10 mixtures were studied by the wheel tracking test in accordance with T0719-2011 (JTG E20-2011) [18]. The test slab has a dimension of 300 × 300 × 50 mm, and the wheel tracking tests were conducted under the standard condition, i.e., loading pressure of 0.7 MPa and testing temperature of 60 • C. The rutting resistance of each mixture was characterized by dynamic stability, which can be calculated as the following equation: where DS = dynamic stability of PAC, cycle/mm; t 1 = time 1 (min), 45min in this study; t 2 = time 1 (min), 45 min in this study; d 1 = deformation at time 1, mm; d 2 = deformation at time 2, mm; and N = speed of the test wheel, 42 cycle/min in this study.

Low-Temperature Cracking Resistance Test
A three-point flexural beam test was conducted to evaluate the cracking resistance of the PAC mixtures at low temperature in accordance with the Chinese test specification T0715-2011 (JTG E20-2011) [18]. The test beam has a dimension of 250 ± 2.0 mm in length, 30 ± 2.0 mm in width, and 35 ± 2.0 mm in thickness, and the three-point flexural beam tests were conducted under the standard condition, i.e., loading rate of 50 mm/min and testing temperature of −10 • C. The testing parameters were calculated as follows: where R B = the flexural strength at failure, MPa; ε B = the flexural strain at failure, µε ; S B = the flexural stiffness at failure, MPa; L = the span of the beam, mm; b = the width of the cross-section at mid-span, mm; h = the height of the cross-section at mid-span, mm; P B = the peak load at failure, N; and d = the disturbance of mid-span at failure, mm.

Permeability Test
Permeability was measured with a developed constant water-head permeability test equipment. As Figure 4 shows, the equipment can measure the vertical permeability coefficient and transverse permeability coefficient of the same specimen by changing components. The cubic specimen (length of each side is 15 cm) is compacted by static loading using a custom-designed device. To calculate the vertical permeability, the vertical baffles of the equipment were removed, and the water flowing from the cross section in the transverse direction. The same principle that the horizontal baffles were removed to calculated transverse permeability. According to the Darcy law [24], permeability coefficient was calculated as follows: where K = permeability coefficient of mixture, cm/s; Q = seepage quality of water, g; ρ w = density of water, g/cm 3 ; L = seepage length, cm; A = area of specimen, cm 2 ; ∆h = water-head, cm; and t = seepage time, s.  Based on a previous study [24], a hydraulic gradient of no more than 0.03 will meet the requirement of the Darcy law. Table 5 shows the final mix design properties of the five PAC mixtures, which all meet the design Based on a previous study [24], a hydraulic gradient of no more than 0.03 will meet the requirement of the Darcy law. Table 5 shows the final mix design properties of the five PAC mixtures, which all meet the design requirement for PAC in China [19]. Testing data presented in this paper are the average values of three replicate specimens, except the permeability test. It can be seen that among various mixtures, PAC13 has the lowest optimum asphalt content (OAC). For different novel PAC10s, with the increasing content of the mineral fillers, the OAC and the stability value increased, indicating that more mineral filler needs more asphalt binder to form the asphalt mastic and more asphalt mastic leads to higher Marshall stability. However, more asphalt mastic also decreased the air void content of PAC10.  Figure 5 shows the results from the moisture susceptibility tests of the PAC mixtures. It can be seen that after one freeze-thaw cycle, all the five mixtures showed good moisture resistance, with TSR values larger than 85%, and the differences in TSR among mixtures are small, implying that all PAC mixtures have excellent moisture resistance at the early stage of moisture effect. However, after two freeze-thaw cycles, the TSR value of PAC10-I4 became obviously smaller than the others, and after three freeze-thaw cycles, PAC10-I4 could not meet the minimum TSR requirement of 75%.

Moisture Susceptibility
Appl. Sci. 2020, 10, x FOR PEER REVIEW 8 of 13 two freeze-thaw cycles, the TSR value of PAC10-I4 became obviously smaller than the others, and after three freeze-thaw cycles, PAC10-I4 could not meet the minimum TSR requirement of 75%. Among the three novel PAC10s, the TSR value increased with the increasing content of the mineral fillers. After three freeze-thaw cycles, increasing the mineral filler content of PAC10 from 4% to 8% increased the TSR value by more than 10%. This can be attributed to more asphalt mastic, which improved the moisture resistance of PAC10, and such effect became more significant with longer moisture conditioning. After three freeze-thaw cycles, PAC13, PAC10-C, PAC10-I6 and PAC10-I8 showed similar TSR, which all met the minimum TSR requirement. Among the three novel PAC10s, the TSR value increased with the increasing content of the mineral fillers. After three freeze-thaw cycles, increasing the mineral filler content of PAC10 from Appl. Sci. 2020, 10, 7019 8 of 12 4% to 8% increased the TSR value by more than 10%. This can be attributed to more asphalt mastic, which improved the moisture resistance of PAC10, and such effect became more significant with longer moisture conditioning. After three freeze-thaw cycles, PAC13, PAC10-C, PAC10-I6 and PAC10-I8 showed similar TSR, which all met the minimum TSR requirement. Figure 6 presents the Cantabro test results of PAC mixtures under immersion and freeze-thaw conditions. Both conditions had similar effect on the durability of PAC mixtures: i.e., as the number of conditioning cycles increases, the Cantabro loss becomes larger. Among various mixtures, the Cantabro loss of PAC10-I4 is obviously larger, and those of the others are relatively close. For the three novel PAC10s, the Cantabro loss decreased with the increasing content of the mineral fillers, but the Cantabro loss of PAC10-I8 is slightly smaller than that of PAC10-I6, indicating that increasing the mineral filler content can improve the durability of novel PAC10 mixture, but when the mineral filler content reaches a certain level, further increase of mineral filler content might not be effective in improving the durability of PAC. In other words, for the three mineral filler contents considered in this study, 6% is the optimum value in terms of the durability of the novel PAC10.   Table 6 presents the average results of the wheel tracking rutting resistance tests. It can be seen that the DS values of all mixtures met the minimum acceptance criterion of 3,000 passes/mm commonly adopted in China (19). PAC13 had better high-temperature stability than PAC10-C. For the novel PAC10 mixtures, with the increasing content of the mineral fillers, the rutting depth was significantly reduced, and the DS value was obviously increased. The novel PAC10-I6 mixture showed better high-temperature stability than the conventional PAC10-C mixture. It is worth noting that rutting development of a mixture is dependent on various factors, such as aggregate gradation, asphalt content, and binder properties. Although PAC13 and PAC10-C have the same fine aggregate content, PAC13 showed better high-temperature stability than PAC10-C, because the former had a larger NMAS. The three novel PAC10s had the same aggregate structure and binder properties. So, a higher mineral filler content mixture led to more asphalt binder and  Table 6 presents the average results of the wheel tracking rutting resistance tests. It can be seen that the DS values of all mixtures met the minimum acceptance criterion of 3000 passes/mm commonly adopted in China [19]. PAC13 had better high-temperature stability than PAC10-C. For the novel PAC10 mixtures, with the increasing content of the mineral fillers, the rutting depth was significantly reduced, and the DS value was obviously increased. The novel PAC10-I6 mixture showed better high-temperature stability than the conventional PAC10-C mixture. It is worth noting that rutting development of a mixture is dependent on various factors, such as aggregate gradation, asphalt content, and binder properties. Although PAC13 and PAC10-C have the same fine aggregate content, PAC13 showed better high-temperature stability than PAC10-C, because the former had a larger NMAS. The three novel PAC10s had the same aggregate structure and binder properties. So, a higher mineral filler content mixture led to more asphalt binder and asphalt mastic, which may serve as an "extender" lubricating the aggregate structure. But the softening point of SBS modified binder is much more than 60 • C, and the HVA additive further increases the softening point of binder [12]. The Marshall stability results at 60 • C also indicated that more asphalt mastic improved the stiffness and strength of the novel PAC10s. So, the stiffening effect of more asphalt mastic was more significant than the increased lubrication effect, which resulted in stiffer mixtures. Therefore, with the increasing content of the mineral fillers, the novel PAC10 mixtures showed better high-temperature stability. Table 7 shows the low-temperature bending test results of PAC mixtures. As can be seen from this table, PAC13 had the lowest flexural strain at failure, while PAC10-C had the largest flexural strain at failure. The mixture has larger NMAS may be more brittle than that has smaller NMAS at −10 • C. Fine aggregate can improve the low-temperature cracking resistance of PAC10, so conventional PAC10-C mixture had larger flexural strain at failure than the novel PAC10 mixtures. For the three novel PAC10 mixtures, with the increasing content of the mineral fillers, the flexural strength and flexural strain at failure increased, but their flexural stiffness at failure are similar. This means increasing the mineral filler content can improve the low-temperature cracking resistance of PAC10 mixture, because more asphalt mastic improves the strength and toughness of the mixture.

Permeability
To evaluate the permeability of the PAC mixtures, permeability coefficients in both transverse and vertical directions of the same specimen were measured in this study. From the test results (average for two replicates) shown in Figure 7, it can be seen that there is a positive correlation between permeability coefficient and air void and the transverse permeability coefficient (T) is larger than vertical permeability coefficient (V). The points of four kinds of PAC10 mixture close to the regression line, but the point of PAC-13 above regression line obviously, means that use of smaller aggregate size may reduce the mixture permeability. The PAC10-I4 had the largest permeability coefficient in both vertical and transverse directions, while PAC10-C had the smallest permeability coefficient. Compared with the conventional PAC10-C mixture, the three novel PAC10 mixtures showed better permeability. However, the permeability of the novel PAC10 mixtures decreases with the increasing content of the mineral fillers. Under the condition of optimum asphalt content, the permeability of PAC10-I6 in both vertical and transverse directions are close to those of PAC13. aggregate size may reduce the mixture permeability. The PAC10-I4 had the largest permeability coefficient in both vertical and transverse directions, while PAC10-C had the smallest permeability coefficient. Compared with the conventional PAC10-C mixture, the three novel PAC10 mixtures showed better permeability. However, the permeability of the novel PAC10 mixtures decreases with the increasing content of the mineral fillers. Under the condition of optimum asphalt content, the permeability of PAC10-I6 in both vertical and transverse directions are close to those of PAC13.

Determination of Mineral Filler Content
Based on the above test results, it is obviously that with the increasing content of the mineral fillers, the durability increases but the permeability decreases. Therefore, careful selection of the mineral filler content would be required to keep the balance of durability and functionality for the novel gradation PAC10 mixture. In Figure 6, it was shown that Cantabro loss of freeze-thaw Cantabro test after three cycle condition is maximum for all the test condition of this study, and Figure 7 shows that the vertical permeability coefficient is smaller than transverse permeability coefficient. The Cantabro loss of freeze-thaw Cantabro test after three cycle condition is used to represent the durability and the vertical permeability coefficient is used to represent the functionality. Figure 8 presents the interaction between durability and functionality of PAC10 mixtures with different mineral filler content.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 11 of 13 Based on the above test results, it is obviously that with the increasing content of the mineral fillers, the durability increases but the permeability decreases. Therefore, careful selection of the mineral filler content would be required to keep the balance of durability and functionality for the novel gradation PAC10 mixture. In Figure 6, it was shown that Cantabro loss of freeze-thaw Cantabro test after three cycle condition is maximum for all the test condition of this study, and Figure  7 shows that the vertical permeability coefficient is smaller than transverse permeability coefficient. The Cantabro loss of freeze-thaw Cantabro test after three cycle condition is used to represent the durability and the vertical permeability coefficient is used to represent the functionality. Figure 8 presents the interaction between durability and functionality of PAC10 mixtures with different mineral filler content.
As shown in the Figure 8, increasing the mineral filler content from 4% to 6%, the Cantabro loss will decrease significantly, but further increase of mineral filler content to 8% has little effect on durability. At the same time, with the increasing content of the mineral fillers, the permeability coefficient decreases, and when increasing the content of the mineral filler from 6% to 8%, the permeability coefficient decreases more significantly than when increasing the content of the mineral filler from 4% to 6%. Therefore, the novel PAC10 mixture containing 6% mineral filler can better balance durability and functionality, and 6% mineral filler is recommended for the novel gradation design.

Conclusions
This study evaluated the effects of the gradation, especially mineral filler content, on the structural and functional performances of PAC through comprehensive laboratory tests. Based on the outcome of this study, the following findings can be obtained: 1.
When the mineral filler content is increased, more asphalt binder is needed to form the asphalt mastic, which can improve the strength of PAC10 mixtures.
PAC13 showed better rutting resistance than the conventional PAC10, but the high-temperature stability As shown in the Figure 8, increasing the mineral filler content from 4% to 6%, the Cantabro loss will decrease significantly, but further increase of mineral filler content to 8% has little effect on durability. At the same time, with the increasing content of the mineral fillers, the permeability coefficient decreases, and when increasing the content of the mineral filler from 6% to 8%, the permeability coefficient decreases more significantly than when increasing the content of the mineral filler from 4% to 6%. Therefore, the novel PAC10 mixture containing 6% mineral filler can better balance durability and functionality, and 6% mineral filler is recommended for the novel gradation design.

Conclusions
This study evaluated the effects of the gradation, especially mineral filler content, on the structural and functional performances of PAC through comprehensive laboratory tests. Based on the outcome of this study, the following findings can be obtained:

1.
When the mineral filler content is increased, more asphalt binder is needed to form the asphalt mastic, which can improve the strength of PAC10 mixtures.
PAC13 showed better rutting resistance than the conventional PAC10, but the high-temperature stability of the novel PAC10 can be equivalent to or even better than PAC13 by increasing the mineral filler content.

4.
Conventional PAC10 showed better low-temperature cracking resistance than PAC13. The low-temperature performance of PAC10 is compromised when 0.075-3 mm aggregate is excluded, but it can be improved by increasing the mineral filler content.

5.
The permeability of the conventional PAC10 is obviously worse than PAC13, but excluding the 0.075-3 mm aggregate from PAC10 is effective in improving the permeability of PAC10. 6.
Six percent mineral filler is recommended for the novel gradation design to keep the balance of durability and functionality.
For PAC mixtures, it is important to balance their structural performance and permeability. To achieve such balance, this study developed a novel gradation for PAC10 by considering the following two facts: on one hand, the permeability of PAC is improved by excluding the aggregate in the size range of 0.075-3 mm; and on the other hand, the durability of PAC10 is maintained by increasing the mineral filler content. The results of this study indicated that the novel PAC10 mixture containing 6% mineral filler could provide equivalent durability and permeability compared with the conventional PAC13 mixture, making PAC10-I6 a suitable porous mixture especially suitable for thin overlays. With increasing the mineral filler content, the OAC of PAC10 increases, and the price of asphalt mixture is closely related to the OAC. Therefore, compared with the conventional PAC10-C, the recommended PAC10-I6 will increase the cost slightly. In the study, several kinds of PAC mixtures with different gradation were compared. However, the type of bitumen has significant influence on the performance properties of PAC mixture, this is the further research will be conducted in future work.

Conflicts of Interest:
The authors declare no conflict of interest.