Investigation of the Effect of Filler on Cohesive Bond Strength of Asphalt Mastic Using Binder Bond Strength (BBS) Test

Abstract

Water damage leads to spalling and loosening of asphalt pavement, which is ultimately due to a reduction in the bond strength of the asphalt mastic. The filler, as an important component of the asphalt mastic, has a great impact on the various properties of the asphalt mastic and even the asphalt mix itself. Therefore, it is important to study the influence of fillers on the performance of asphalt mastic. In this paper, the various properties of twelve types of fillers, including mineral powder, mineral powder partially replaced with cement, slaked lime, and recycled powder, and their effects on the bond strength of asphalt mastic are investigated. Binder bond strength (BBS) was used to evaluate the performance of the different asphalt mastics. The moisture resistance of the asphalt mastic was evaluated by measuring the bond strength of the asphalt mastic after treatment in a water bath. A grey correlation analysis was then carried out to derive the relationship between the filler index and the bond strength of the asphalt mastic. The results show that the addition of slaked lime and cement helped to improve the moisture resistance of the asphalt mastic. The fineness modulus and other indicators have a great influence on the bond strength of asphalt mastic. The interaction between hydrothermal coupling and filler type has a non-negligible effect on the moisture resistance of the asphalt mastic.

1. Introduction

Asphalt mixtures are granular materials, and their mechanical strength is generally considered to depend on the friction and embedding forces between the aggregate particles, the cohesion of the asphalt binder, and the bond between the asphalt and the aggregate [1]. Erosion due to water can damage the interface between asphalt and aggregate, reducing the adhesion of the asphalt bond [2,3,4,5], resulting in spalling and loosening and reducing the life of the asphalt pavement [6,7].

Many pioneering studies have been carried out on the adhesion between asphalt and aggregate. In general, commonly used tests fall into one of three main categories: energy-based methods (contact angle tests) [8,9,10], observation-based methods (immersion tests) [11] and mechanical methods (pull-off tests) [12,13]. The binder bond strength (BBS) test, a pull-off test, was developed to obtain mechanical parameters and is widely used to evaluate the adhesion and cohesion of asphalt [14,15,16]. Canestrari et al. [17] used a modified pull-off head to improve the accuracy of the BBS test. The pull-off tensile strength (POTS) and damage type were obtained from the BBS test and used to evaluate the bond/cohesion properties of the asphalt [18]. Moraes et al. noted that the BBS test results were consistent with surface energy measurements [19]. Wang et al. [1] used the skeleton penetration test, the BBS test and the MBBS test to investigate several aspects of the asphalt–aggregate bond strength, the mastic–aggregate bond strength, the mortar–aggregate bond strength and the cohesive strength of the asphalt, the mastic and the mortar itself. The results show that spalling damage in medium- and low-temperature conditions is dominated by cohesive damage to the mortar; spalling damage due to short-term water damage is dominated by cohesive damage to the mortar, and spalling damage due to long-term water damage is dominated by bonding damage between the asphalt and aggregate.

Studies have shown that asphalt can only form a slurry with mineral fillers in order to adsorb and bond coarse and fine aggregates, forming an asphalt mixture with a certain strength and deformability [20,21,22,23]. Therefore, it is necessary to study the influence of the fillers on asphalt mastic and asphalt mixes. Zhu et al. [24] tested six mineral powders for some basic physicochemical parameters (density, hydrophilicity coefficient, methylene blue value) and analyzed the particle size of each mineral powder in detail using a laser particle size analyzer. A slurry was prepared using No. 70 matrix asphalt at different powder-to-mastic ratios and the filler properties were evaluated based on the slurry properties. Li et al. [25] characterized the particle size distribution, chemical composition and microstructure of nine fillers and used the bond strength (BBS) to evaluate the performance of different asphalt mastics.

As can be seen, the influence of fillers on the bond strength of asphalt and asphalt mastic has been of interest to numerous researchers, and many attempts have been made to explore it. However, in regard to the selection of fillers, existing studies have mainly chosen single fillers such as limestone, fly ash, etc., without adding additives to replace the existing fillers in equal amounts and measuring the performance of the newly composed mixed fillers. Therefore, in this study, three grades of limestone mineral powder were prepared by regulating the original grade of limestone mineral powder and adding three additives, cement, slaked lime and recycled powder, to the original grade to replace the mineral powder, and the properties of these 12 fillers were characterized. The asphalt mastic was then prepared for BBS tests, followed by analysis of variance (ANOVA) of the test data to consider the effect of filler type and hydrothermal coupling on the bond strength of the asphalt mastic. Finally, the correlation between the different filler indicators and the bond strength of the asphalt mastic was analyzed using grey correlation analysis.

2. Materials and Methods

2.1. Materials

2.1.1. Fillers

Using limestone mineral powder (powder obtained by crushing mined ore), the filler grade distribution was first analyzed using a laser particle size analyzer (Microtrac S3500, Microtrac Inc., Largo, FL, USA), and the original grade was used as the median grade, on the basis of which coarse and fine grades were adjusted (the adjustment method considered a constant grade, increasing the weight of the coarse or fine particle size). The density and hydrophilicity coefficients of the three particle size gradations were tested, as shown in

Table 1. Filler density and hydrophilic coefficient.

Filler	Density (g/cm3)	Hydrophilic Coefficient
Mineral powder (fine)	2.708	0.509
Mineral powder (medium)	2.669	0.520
Mineral powder (coarse)	2.659	0.565
Mineral powder + 10% recycled powder	2.720	0.582
Mineral powder + 20% recycled powder	2.742	0.632
Mineral powder + 30% recycled powder	2.791	0.658
Mineral powder + 20% cement	2.822	0.714
Mineral powder + 40% cement	2.854	0.738
Mineral powder + 60% cement	2.913	0.792
Mineral powder + 30% slaked lime	2.639	0.623
Mineral powder + 50% slaked lime	2.528	0.651
Mineral powder + 70% slaked lime	2.482	0.684

.

2.1.2. Additives

Three additives were considered: slaked lime, recycled powder (dust generated during the production by asphalt mixing plants) and cement. The substitution ratios for slaked lime were 30%, 50% and 70%, and for recycled powder 10%, 20% and 30%, while the cement substitution ratios were 20%, 40% and 60%. The densities and hydrophilic coefficients of the nine fillers containing additives were tested, as shown in Table 1.

2.1.3. Asphalt

As the asphalt, 70# base asphalt and finished SBS modified asphalt were selected. Their basic specifications are shown in

Table 2. Properties of 70# base asphalt.

Items		Measured Value	Test Method [26]
Penetration, 0.1 mm		65.4	T0604
Ductility (5 cm/min, 15 °C), cm		120.4	T0605
Softening Point, °C		48.1	T0606
Thin film oven test (163 °C, 5 h)	Mass variation, %	0.45	T0609
	Residual penetration ratio, %	61.6	T0604
	Ductility (5 °C), cm	79.2	T0605
Dynamic viscosity (60 °C, Pa·s)		290	T0620

and

Table 3. Properties of SBS modified asphalt.

Items		Measured Value	Test Method [26]
Penetration, 0.1 mm		55.5	T0604
Ductility (5 cm/min, 5 °C), cm		37.9	T0605
Softening Point, °C		86	T0606
Thin film oven test (163 °C, 5 h)	Mass variation, %	−0.15	T0609
	Residual penetration ratio, %	73.2	T0604
	Ductility (5 °C), cm	25	T0605
Dynamic viscosity (60 °C, Pa·s)		4841	T0620

.

2.1.4. Preparation of Asphalt Mastics

The mass ratio of filler to asphalt was 1, prepared as follows:

A sufficient amount of mineral powder was put in a tin box and placed in an oven at 105 °C for 4 h to eliminate moisture in the mineral powder, kept warm and set aside; a sufficient amount of asphalt was placed in a constant-temperature oven and heated to a flowing state. The heating temperature for SBS modified asphalt was 160 °C, and the heating temperature for 70# base asphalt was 150 °C. Heating time was about 2 h. The asphalt was dispensed into beakers (the mass of bitumen per cup dispensed was recorded). Manual pre-mixing was performed. The divided asphalt was placed on an electric stove to keep the SBS modified asphalt at 160 °C and the 70# base asphalt at 150 °C. The mineral powder was gradually added (powder to mastic ratio of 1) while stirring. The asphalt mastic was put in an insulation jacket and placed under high-speed shear with the rotor immersed in the asphalt mastic. The rotor was turned on and speed was gradually increased to 1500 rpm. The asphalt mastic was stirred for 30 min at a constant stirring temperature. After completion, the mastic was obtained according to the mixture.

2.2. Test and Evaluation Methods

2.2.1. Characterization of Filler Properties

The particle size distribution of fillers reflects their gradation distribution. In this paper, the particle size distribution of 12 fillers was analyzed using a laser particle size analyzer, which applies the Mie scattering theory and tests particle size distribution based on the physical phenomena of diffraction and scattering of light by particles. When the laser beam, which is highly monochromatic and highly directional, is blocked by a particle, a portion of the light is scattered, the angle of the scattered light being inversely proportional to the diameter of the particle. It has been shown that the intensity of the scattered light represents the number of particles of that size and that the scattered light is processed optically. When scattered light at different angles is directed through a Fourier lens onto a multivariate photodetector, the light signals are converted into electrical signals and transmitted to a computer, where they are processed by special software to obtain an accurate particle size distribution [24].

Three parallel analyses were carried out for each filler sample to ensure that the values obtained were reliable. In addition to the particle size distribution curves, indicators such as particle size d (0.1) for 10% passage, d (0.5) for 50% passage, d (0.9) for 90% passage and volume mean particle size D [4,3] were obtained, as shown in Equation (1).

$$\mathrm{D}[4,3]=\frac{\sum m_i\times d_i^4}{\sum m_i\times d_i^3}$$

(1)

where m_i is the frequency of particles in size class i, having a mean d_i diameter.

In addition, the fineness modulus (FM) is used to reflect the particle size distribution of the filler, as shown in Equation (2) [27].

$$\mathrm{FM}=\frac{\sum (100-p_i)}{100}$$

(2)

where: i ∈ [75,50,30,20,10,5,3,1] μm, and pi is the cumulative passage rate of the i^th particle size, %.

2.2.2. Binder Bond Strength Test

The BBS test system consists of an automatic adhesion tester (Positest AT-A, DeFelsko, NY, USA.), an aggregate substrate, a pulling head and an asphalt sample [28]. The AT-A, with its self-aligning system, produces a constant pulling rate, separating the asphalt from the aggregate substrate and allowing the measurement of POTS. There are two types of damage in the BBS test, adhesion damage and cohesion damage, and the corresponding POTS can be divided into adhesion strength (between asphalt and aggregate) and cohesion strength (within the asphalt) [29].

The original pulling head proved to be somewhat problematic in terms of controlling the thickness of the asphalt mastic and the bonding of the asphalt mastic to the pulling head, so in order to improve the accuracy of the BBS test data, some modifications were made to the pulling head. Compared to the original pulling head, the improved pulling head was 20 mm in diameter and 0.2 mm thick around the edge, which provided a lateral restraint for the asphalt mastic during the sample preparation process. The surface of the pulling head was then roughened to improve the adhesion between the pulling head and the asphalt paste. In addition, four cuts were made along the edges to allow excess bitumen to escape when the pulling head was pressed against the aggregate substrate. These modifications allowed the thickness of the asphalt to be controlled and ensured a complete bond between the asphalt and the pulling head [29].

In order to determine the thickness and loading rate of the asphalt tested in the BBS test, Huang et al. [30] recommended a thickness of 0.2 mm and a loading rate of 0.7 MPa/s for the impact test of asphalt. The BBS test followed the following steps:

(1)

Slab preparation

Firstly, large pieces of basalt obtained from the quarry were cut into rectangular blocks of 12 cm × 12 cm × 2.5 cm. As the saw teeth used during the cutting process can make the surface of the slabs uneven and of varying roughness, if used directly for the preparation of BBS specimens, this can lead to large errors in the test results, so the surface of the slabs needed to be polished [31]. The stones were sanded with 60 and 180 grit sandpaper for 3 min each and then placed in an ultrasonic cleaner at 60 °C for 30 min to remove the dust from the surface. The stones were then dried in an oven at 160 °C for 2 h. The stones were then removed and set aside.

(2)

Specimen preparation

To match the actual asphalt mix, the asphalt mastic was heated at the mixing temperature specified in the specification, while the stone and the pulling head were kept at the same temperature. After the asphalt mastic had softened, a glass rod was dipped into the asphalt mastic and dripped onto the pulling head. The pulling head was then quickly pressed vertically against the surface of the stone and pressure was applied so that the excess asphalt flowed out of the opening, and was held for 10 s. This process must be carried out in an oven to reduce the cooling of the slab, and the drawing head may not be moved again during this period. After the specimen was formed, the slab was removed and the excess asphalt mastic extruded from the bottom was removed using a scraper to prevent it from affecting the drawing results. After completion, the aggregate slab was placed on the drawing head and left at room temperature for 24 h.

(3)

Working conditions

The prepared BBS specimens were placed in water at 60 °C, and the effect of three water bath regeneration times of 1 d, 3 d and 5 d on the bond strength of different asphalt mastics was considered.

(4)

Testing

The specimens were placed in an oven at 25 °C for 1 h before testing. The bond strength of the asphalt mastic to the aggregate was tested using a Positest AT-A bond tester (Positest AT-A, DeFelsko, NY, USA). The process must be carried out in the oven so as to minimize the effect of temperature variations on the pull-out strength and increase the reliability of the test, as shown in Figure 1. If the bitumen breaks from the interface between the pulling head and the bitumen, this indicates that the bond strength between the bitumen and aggregate is greater than the bond strength between the bitumen and the pulling head. If more than 50% of the final fracture surface slab is adhered to the asphalt, the damage is cohesive; otherwise the damage is adhesive. The data on the display was read and recorded. Each set of tests was repeated three times, and the average of the valid data was taken as the test result.

2.2.3. Grey Correlation Analysis

The aim of grey correlation analysis is to find important relationships between the factors of a system, which are based on the grey correlation degree. The grey correlation analysis method is a quantitative description and comparison of the developmental trends of a system. The grey correlation is calculated for each comparison series and a reference series, where the greater the correlation with the reference series, the closer the relationship with the reference series. The grey correlation analysis method does not result in a discrepancy between quantitative results and qualitative analysis results. The basic idea is to dimensionlessly process the original observation data of the evaluation indexes, calculate the correlation coefficient and the correlation degree, and rank the indicators to be evaluated according to the magnitude of the correlation degree [1].

3. Results and Discussion

3.1. Characterization of Filler Properties

The data obtained by the laser particle size analyzer was plotted as a particle size distribution graph [32,33], as shown in Figure 2.

As can be clearly seen from the particle size graph of the filler shown in Figure 2, the particle size of the mineral powder (coarse) is the largest, and the filler without additives is somewhat larger than the filler with additives, which indicates that the particle size of additives is generally smaller than that of the mineral powder. The particle sizes of the nine fillers containing additives are relatively similar, requiring the calculation of d (0.1), d (0.5), d (0.9), d [4,3] and FM to quantify the particle size distribution. As shown in Figure 2, the particle size is obviously smaller when replacing mineral powder with additives. In terms of the fineness modulus, 70% slaked lime replacing mineral powder has the smallest particle size, while 50% slaked lime replacing mineral powder is obviously larger, followed by 60% cement replacing mineral powder and 40% cement replacing mineral powder, which shows that the filler particle size not only depends on the type of additives but also on the replacement ratio of additives. The volume average particle size D [4,3] shows that 30% recycled powder replacing mineral powder has the smallest particle size, followed by 60% cement replacing mineral powder and then 40% cement replacing mineral powder, indicating that the cement replacing mineral powder has a significant effect on reducing the particle size of the filler, as shown in

Table 4. Calculated evaluation indicators of different fillers.

Filler	d (0.1)	d (0.5)	d (0.9)	D [4,3]	FM
Mineral powder (fine)	4.54	18.41	61.01	173.56	4.71
Mineral powder (medium)	4.60	19.98	82.79	204.56	4.90
Mineral powder (coarse)	4.93	23.86	137.70	244.53	5.23
10% recycled powder	4.90	16.70	47.87	49.49	4.49
20% recycled powder	5.28	16.84	44.97	49.49	4.43
30% recycled powder	5.27	15.00	34.55	36.82	4.30
20% cement	4.74	16.86	47.59	55.89	4.47
40% cement	4.44	15.19	40.11	44.94	4.28
60% cement	4.55	15.91	39.8	43.67	4.27
30% slaked lime	4.89	15.37	46.85	52.99	4.44
50% slaked lime	5.73	15.01	44.76	58.25	4.41
70% slaked lime	4.96	13.23	42.32	58.39	4.18

.

3.2. Analysis of Asphalt Mastic Bond Strength and Damage Forms

Data measured with the Posi Test At-A bond tester was recorded and photographed for analysis of the form of damage to the stones and pulling heads, and the average of three parallel specimens under the same working conditions for each sample was taken for analysis to obtain the bond strength of different asphalt mastics under different hydrothermal action times, as shown in Figure 3.

The X-axis represents how the specimen was maintained after preparation, and the time duration.

As can be seen from

Table 5. Forms of damage in asphalt mastic bond strength tests.

Filler	Forms of Damage
	Room Temperature 1 Day		60 °C Water Bath for 1 Day		60 °C Water Bath for 3 Day		60 °C Water Bath for 5 Day
	SBS	70#	SBS	70#	SBS	70#	SBS	70#
Mineral powder (fine)	C	C	C	C	C	C	C	C
Mineral powder (medium)	C	C	C	C	C	C	A	C
Mineral powder (coarse)	C	C	C	C	C	C	A	C
10% recycled powder	C	C	*	C	*	C	*	C
20% recycled powder	C	C	*	C	*	C	*	C
30% recycled powder	C	C	*	C	*	C	*	C
20% cement	C	C	A	A	*	A	*	A
40% cement	C	C	C	A	C	A	C	A
60% cement	C	C	C	C	C	C	C	C
30% slaked lime	C	C	C	C	A	A	*	A
50% slaked lime	C	C	C	C	C	C	A	C
70% slaked lime	C	C	C	C	C	C	C	C

* represents damage between the asphalt and the pulling head; A represents asphalt–aggregate adhesion damage; C represents cohesive damage to the asphalt itself.

, under the action of hydrothermal coupling, the damage form of the 24 asphalt mastic bond strength tests changed gradually from cohesive damage to the asphalt mastic itself to adhesion damage between the asphalt mastic and the aggregate, as shown in Figure 4. When the adhesion damage between the asphalt mastic and the aggregate is changed, it indicates that the bond strength between the asphalt mastic and the aggregate is less than the bond strength of the asphalt mastic itself.

3.2.1. Influence of Filler and Hydrothermal Coupling on Bond Strength of Mastic

As can be seen from Figure 3, the bond strength after 24 h in the water bath is significantly lower than that after 24 h at room temperature under hydrothermal coupling, although the bond strength of the asphalt mastic increases again after 72 h in the water bath. The results show that the hydrothermal coupling does not necessarily result in a continuous reduction of the bond strength of the asphalt mastic in the short term, but that the bond strength fluctuates due to the asphalt, the filler and the hydrothermal coupling.

In regard to the mineral powder, the highest bond strength was found in the asphalt mastic made from mineral powder (fine) and SBS modified asphalt, while higher bond strength was found in the mastic made from 70# base asphalt and mineral powder (medium). The results show that the particle size of the mineral powder does not completely determine the bond strength of the asphalt mastic, and that the type of asphalt is also an influencing factor. In terms of the form of damage, the asphalt mastic made using mineral powder as filler changes from cohesive damage to the asphalt mastic itself to adhesion damage between the asphalt mastic and the aggregate under the effect of hydrothermal coupling.

In regard to the recycled powder, the bond strength of the asphalt mastic made from 20% recycled powder replacing mineral powder and 70# base asphalt was the highest, while the bond strength of the mastic made from SBS modified asphalt and filler replaced with recycled powder was higher for the 10% recycled powder replacing mineral powder. The results show that the proportion of recycled powder to replace mineral powder should not be too large, but the mineral powder can be mixed with a small amount of recycled powder. This achieves environmental protection and resource recycling, but does not produce a large change in the performance of the asphalt mastic. In terms of the form of damage, the asphalt mastic made from recycled powder replacing mineral powder as filler with SBS modified bitumen changed from cohesive damage of the asphalt mastic itself to adhesion damage between the asphalt mastic and the aggregate under hydrothermal coupling, and in some specimens to damage between the asphalt mastic and the pulling head, indicating that moisture penetrated between the asphalt mastic and the aggregate plate and between the asphalt mastic and the pulling head for damage, thus indicating that the recycled powder was not as effective as mineral powder as a filler.

In regard to the cement, the bond strength of the asphalt mastic made from 20% cement-replaced mineral powder and 70# base asphalt was the highest, while the bond strength of the slurry made from SBS modified asphalt and filler replaced with cement was higher for the 40% cement-replaced mineral powder. The results indicate that the proportion of cement replacing mineral powder should not be too large, ideally 20% to 40%. The difference is that the bond strength of the 70# base asphalt and the cement-replaced filler is higher when kept in a water bath for 3 days than at room temperature for 1 day, probably due to the reaction of the cement itself with water. When cement meets water, the calcium silicate in it reacts with the water to produce hydrated calcium silicate and calcium hydroxide. The hydrated calcium silicate and calcium hydroxide will rely on hydrogen bonds and intermolecular forces to bind together to form a very high-strength curing material, which has an impact on the bond strength of the asphalt mastic. It is also clear that when cement replaces mineral powder, water does not necessarily have a negative effect on the asphalt mastic, but may also have a beneficial effect on the performance of the asphalt mastic.

In regard to the slaked lime, the bond strength of the asphalt mastic made from 50% slaked lime replaced with mineral powder and 70# base asphalt was highest, while the bond strength of the slurry made from SBS modified asphalt and filler replaced with slaked lime was higher for the mastic made from 30% slaked lime replaced with mineral powder. As with the cement, the bond strength of the 70# base asphalt with filler replaced with slaked lime was higher when kept in a water bath for 3 days than at room temperature for 1 day, which may be related to the effect of slaked lime on the stone. The slaked lime interacts with the acidic stone surface and takes on the nature of an alkaline stone, and the acidic stone surface becomes hydrophobic and lipophilic. When water penetrates the stone to the asphalt interface, the stone becomes more easily bonded to the asphalt and distant from the water and adhesion is improved.

3.2.2. Influence of Type of Asphalt on Bond Strength of Mastic

As can be seen in Figure 3, the bond strength of the mastic varies for the same filler and the same hydrothermal coupling conditions, due to the asphalt. Overall, asphalt mastics made from base asphalt have a greater bond strength than those made from SBS modified asphalt.

From the “mechanical theory” of asphalt adhesion theory, we can learn that there are usually many rough, porous structures on the surface of the aggregate. The asphalt can enter these pores after high temperature flow, and when the asphalt cools and hardens, it forms a “mechanical anchoring force” in the pores. However, the incorporation of SBS causes different degrees of agglomeration within the asphalt, so that the asphalt is not easily adsorbed by the small pores on the surface of the slab, leading to a decrease in the adhesion capacity of the asphalt.

Conversely, SBS modified asphalt has good elastic properties. When evaluating its bond strength, it is necessary to consider not only pulling strength but also the type of pulling process it can withstand. Testing clearly reveals that the changes in the values of the two types of asphalt mastic are different. The pull-out strength of the matrix asphalt drops to zero immediately after reaching its peak, while the pull-out strength of the SBS modified asphalt drops gradually to zero after reaching its peak. This indicates that the mastic specimen made from SBS modified bitumen was still able to withstand a certain tensile load after reaching the pull-out damage value, which indicates that it required more external tensile work than the matrix asphalt mastic specimen to break it completely. In this regard, the bond strength data for 70# base asphalt is more informative.

3.3. Analysis of Variance for BBS Test Data

BBS tests were carried out to investigate the effect of filler type and hydrothermal coupling and their interaction with the bond strength of the outgoing asphalt mastic. When testing 70# base asphalt mastic, 12 filler types were used to evaluate the hydrothermal coupling effect. The material was kept at room temperature for 1 day, then placed in a 60 °C water bath for 1 day, 3 days, and 5 days under 4 kinds of working conditions, for a total of 48 tests. Each test was performed 3 times, and the obtained asphalt mastic bond strengths are shown in

Table 6. Asphalt mastics bond strength.

Filler Types	Asphalt Mastics Bond Strength/MPa
	B1 (Room Temperature 1 Day)	B2 (60 °C Water Bath for 1 Day)	B3 (60 °C Water Bath for 3 Day)	B4 (60 °C Water Bath for 5 Day)
A1 (Fine mineral powder)	1.98	1.12	1.40	1.22
	2.47	1.14	1.19	1.36
	2.42	1.31	1.22	1.12
A2 (Medium mineral powder)	2.61	1.28	1.55	1.31
	2.42	1.29	1.29	1.28
	2.35	1.16	1.31	1.36
A3 (Coarse mineral powder)	2.25	1.18	1.27	1.21
	2.59	1.24	1.29	1.25
	2.41	1.06	1.26	1.28
A4 (20% cement)	1.64	1.52	2.42	1.96
	1.75	1.73	2.03	1.72
	1.63	1.60	2.12	2.12
A5 (40% cement)	1.87	1.65	1.82	1.71
	1.77	1.43	2.02	1.64
	1.71	1.49	2.17	2.13
A6 (60% cement)	1.78	1.57	2.03	1.88
	1.57	1.69	2.09	1.86
	1.90	1.32	2.04	1.83
A7 (30% slaked lime)	1.70	1.29	2.10	1.62
	1.64	1.15	2.22	1.77
	1.57	1.17	1.98	1.51
A8 (50% slaked lime)	1.64	1.26	2.46	1.67
	1.74	1.34	2.26	1.72
	1.83	1.24	2.30	1.83
A9 (70% slaked lime)	1.88	1.19	2.31	1.65
	1.52	1.14	2.30	1.58
	1.72	1.22	2.32	1.55
A10 (10% recycled powder)	1.86	1.65	1.68	1.62
	2.02	1.71	1.72	1.51
	1.91	1.83	1.67	1.64
A11 (20% recycled powder)	2.41	1.98	1.84	1.72
	2.42	1.79	1.83	1.59
	2.34	1.80	2.06	1.57
A12 (30% recycled powder)	2.23	1.82	1.83	1.20
	2.34	1.90	2.01	1.46
	2.38	1.80	1.81	1.46

.

A two-factor repeated experiment ANOVA was used to analyse the significance of the effect of filler type and hydrothermal action and their interaction on the bond strength of the outgoing asphalt mastic at a significance level of α = 0.01 [34].

As can be seen from

Table 7. Two-factor repeated trial ANOVA table.

Sources of Variance	Sum of Squared Deviations	Degrees of Freedom	Sum of Mean Deviations Squared	F	Threshold Fα (n1, n2)	Significance
Factor A (Filler types)	2.5021	11	0.2275	15.9091	F0.01 (11, 96) = 2.484	Significant
Factor B (Water-heat coupling)	7.1495	3	2.3832	166.6573	F0.01 (3, 96) = 4.022	Significant
Interaction A × B	10.2901	33	0.3118	21.8042	F0.01 (33, 96) = 1.526	Significant
Errors E	1.3769	96	0.0143
Total	21.3186	143

, Factor A (filler type), Factor B (hydrothermal coupling effect), and the interaction of Factor A and Factor B all have a significant effect on the bond strength of the asphalt mastic.

In general, the larger the difference between the critical value and the F-value, the greater the influence of the factor corresponding to the F-value on the experimental results. It also indirectly indicates that this factor is more important and requires special attention. The magnitude of the F-value can be used to rank the factors in order of priority: factor B (hydrothermal coupling effect) is the most important, followed by factor A and factor B interaction, and finally factor A (filler type). The interaction is often also a factor influencing the results of experiments performed by the subjects.

3.4. Correlation Analysis of Different Filler Indicators with Bond Strength of Asphalt Mastic

In this paper, the bond strength of each 70# base asphalt mastic after 1 day at room temperature was selected with reference to the actual measured data for physical and chemical indices of different fillers as a comparative series (see

Table 8. Grey correlation analysis of raw data.

Mastic Number	A1	A2	A3	A4	A5	A6	A7	A8	A9	A10	A11	A12
X0 (Bond strength)	2.29	2.46	2.42	1.67	1.78	1.75	1.64	1.74	1.71	1.93	2.39	2.32
X1 (Density)	2.71	2.67	2.66	2.72	2.74	2.79	2.82	2.85	2.91	2.64	2.53	2.48
X2 (hydrophilic coefficient)	0.51	0.52	0.57	0.58	0.63	0.66	0.71	0.74	0.79	0.62	0.65	0.68
X3 (Fineness modulus)	4.71	4.90	5.23	4.49	4.43	4.30	4.47	4.28	4.27	4.44	4.41	4.18
X4 (Average particle size)	174	205	245	49.5	49.5	36.8	55.9	44.9	43.7	53.0	58.3	58.4
X5 [d (0.1)]	4.54	4.60	4.93	4.90	5.28	5.27	4.74	4.44	4.55	4.89	5.73	4.96
X6 [d (0.5)]	18.4	20.0	23.9	16.7	16.8	15.0	16.9	15.2	15.9	15.4	15.0	13.2
X7 [d (0.9)]	61.0	82.8	138	47.9	45.0	34.6	47.6	40.1	39.8	46.9	44.8	42.3

).

Each series is averaged; that is, the individual data in the series is divided by the average of each series, and the averaging process produces a new series. The new series after the averaging process is shown in

Table 9. Series generated from correlation analysis.

Mastic Number	A1	A2	A3	A4	A5	A6	A7	A8	A9	A10	A11	A12
X0 (Bond strength)	1.14	1.23	1.20	0.83	0.89	0.87	0.82	0.87	0.85	0.96	1.19	1.15
X1 (Density)	1.00	0.99	0.98	1.00	1.01	1.03	1.04	1.05	1.07	0.97	0.93	0.92
X2 (hydrophilic coefficient)	0.80	0.81	0.89	0.91	0.99	1.03	1.11	1.16	1.24	0.97	1.02	1.07
X3 (Fineness modulus)	1.04	1.09	1.16	1.00	0.98	0.95	0.99	0.95	0.95	0.98	0.98	0.93
X4 (Average particle size)	1.94	2.29	2.74	0.55	0.55	0.41	0.62	0.50	0.49	0.59	0.65	0.65
X5 [d (0.1)]	0.93	0.94	1.01	1.00	1.08	1.07	0.97	0.91	0.93	1.00	1.17	1.01
X6 [d (0.5)]	1.09	1.19	1.42	0.99	1.00	0.89	1.00	0.90	0.94	0.91	0.89	0.78
X7 [d (0.9)]	1.09	1.48	2.47	0.86	0.81	0.62	0.85	0.72	0.71	0.84	0.80	0.76

.

Results of the correlation series:

ε₁ = (0.856, 0.771, 0.785, 0.829, 0.873, 0.840, 0.782, 0.814, 0.784, 0.996, 0.758, 0.772)

ε₂ = (0.700, 0.659, 0.720, 0.922, 0.897, 0.836, 0.730, 0.732, 0.672, 1.000, 0.827, 0.908)

ε₃ = (0.900, 0.857, 0.958, 0.836, 0.902, 0.915, 0.823, 0.913, 0.899, 0.983, 0.793, 0.781)

ε₄ = (0.494, 0.423, 0.337, 0.741, 0.704, 0.632, 0.811, 0.687, 0.688, 0.683, 0.594, 0.612)

ε₅ = (0.791, 0.737, 0.804, 0.832, 0.813, 0.800, 0.845, 0.962, 0.919, 0.967, 0.985, 0.854)

ε₆ = (0.951, 0.963, 0.792, 0.841, 0.888, 0.990, 0.814, 0.967, 0.904, 0.953, 0.727, 0.682)

ε₇ = (0.951, 0.759, 0.382, 0.983, 0.913, 0.761, 0.967, 0.849, 0.858, 0.873, 0.671, 0.667)

Correlation of filler indicators with the bond strength of asphalt mastic:

β₁ = 0.822, β₂ = 0.800, β₃ = 0.880, β₄ = 0.617, β₅ = 0.859, β₆ = 0.873, β₇ = 0.803

As can be seen in Figure 5, the correlation between filler index and asphalt mastic bond strength, in descending order, is:

β₃ > β₆ > β₅ > β₇ > β₁ > β₂ > β₄

The correlation between the three indicators of fineness modulus, d (0.5), d (0.1) and the bond strength of asphalt mastic, is greater than 0.85, while the correlation between density, d (0.9) and the hydrophilic coefficient is greater than 0.8. The correlation between the average particle size and the bond strength of asphalt mastic is slightly lower than the others. Therefore, in regard to asphalt mastic bond strength, we should focus on the fineness modulus, d (0.5), d (0.1), to conduct targeted analysis and research.

4. Conclusions

This paper investigates performance indicators for limestone fines and a total of 12 fillers after replacing the fines with cement, slaked lime and recycled powder. BBS tests were used to evaluate the adhesion and cohesion of the asphalt, while ANOVA data were used to consider the effect of filler type and hydrothermal coupling on the bond strength of the asphalt mastic. Finally, the correlation between different filler indicators and the bond strength of the asphalt mastic was analyzed using grey correlation analysis. The following conclusions can be drawn from this study:

• Hydrothermal coupling does not necessarily result in a continuous reduction in the bond strength of the asphalt mastic in the short term; rather, the bond strength will fluctuate due to the asphalt, filler and hydrothermal coupling.
• Filler type, hydrothermal coupling, and the interaction between the two have a significant effect on the bond strength of the asphalt mastic. The order of influence is: hydrothermal coupling > interaction of hydrothermal coupling and filler type > filler type.
• In regard to asphalt mastic bond strength, the three indicators of fineness modulus, d (0.5) and d (0.1) should be focused on for targeted analysis and research.
• In regard to mineral fillers, the particle size of the mineral powder should not be too large. The addition of slaked lime and cement helps to improve the moisture resistance of asphalt mastic, but more is not necessarily better.

It should be noted that the asphalt mastic selected for this test was only made with a powder-to-mastic ratio of 1 and a test temperature of 25 °C. The effect of different powder-to-mastic ratios and different test temperatures on the bond strength of the asphalt mastic will be considered in further studies.