*Asphalt Mix Components*

The bitumen used in the tests was 50/70 asphalt from ORLEN Asfalt Sp. z o.o. (Plock, Poland). The properties of asphalt binder are presented in Table 1.


**Table 1.** Properties of the base bitumen.

The aggregates used in laboratory tests were sand, dolomite, and limestone as a mineral filler. The dolomite used in the research comes from the Piskrzyn deposit (Kopalnie Dolomitu S. A.) while limestone from the Bukowa deposit (Lhoist Bukowa Sp. z o.o.). Both mines are located in the Swi ˛ ´ etokrzyskie Province (Poland). The sand used in the study comes from the Baranówka deposit. The microstructure of the initial carbonate raw materials (dolomites and limestone) used in the study is presented in Figure 1. The chemical composition of the minerals determined by the X-ray fluorescence spectrometry (XRF) method is presented in Table 2. For carbonate raw materials, the dominant chemical component is CaO, which is 80.94% for limestone filler and 39.62% for dolomite. In the case of the latter, MgO is also an important component, with a content of 14.76%. For quartz sand the main component is SiO2 (92.99%). Silica is also present in dolomite aggregate and its content is 10.05%. The remaining chemical components in the tested raw materials are present in minor quantities. Mineral composition of used aggregates and filler identified using the X-ray diffraction (XRD) method is presented in Figure 2. The content of individual mineral phases was determined by the Rietveld refinement. For dolomite aggregate, the content of dolomite equals 87.6%, 1.2% of calcite and 11.2% of quartz. In case of limestone filler the calcite content is 98.3%. The mineral composition of the filler is supplemented with quartz in the amount of 1.7%. In the sand used for the study, the main component is quartz, which is 93.6%, followed by marginal amounts of potassium feldspar identified as albite (6.4%). SEM images of the materials used are shown in Figure 3. The shape of dolomite grains is shown in Figure 3a and can be described as angular, while the character of their surface is presented in Figure 3b and can be clearly marked by a three-way cleavage which results in rough character. Limestone filler and crushed sand are presented in Figure 3d respectively. The limestone grains reach a size of about 1μm and represent a fairly regular shape. Quartz grains present in the sand most often reach irregular shapes emphasized by a clear fracture.

**Figure 1.** Microphotographs of dolomite (**a**) and limestone (**b**).



**Figure 2.** Mineral composition of used aggregates and filler.

**Figure 3.** Scanning electron microscopy (SEM) images of dolomite (**<sup>a</sup>**,**b**), limestone filler (**c**) and sand (**d**).

Aggregate grading was determined by dry sieving method [31]. In case of the filler air jet sieving test was applied [32]. The composition of the mineral mix was designed using the grading envelope method applicable in Poland [22].

The designed reference asphalt mix is intended for construction of asphalt concrete wearing layer (AC 11 S) on roads with traffic loads of KR category 1–2 in accordance with Polish technical standards [22]. The content of dosed asphalt was 5.7% in relation to the weight of the asphalt mix. Figure 4 shows the grading of the mineral mix. The composition of both mineral and asphalt is presented in Table 3.




### **3. Research Method**

### *3.1. Samples Preparation*

Samples made of reference asphalt mix were compacted in a Marshall device using 50 blows per side and named M3. The compaction temperature was 135 ◦C according to Polish requirements [19]. Samples were also compacted using different energies: 20, 35, and 75 blows per side, named M1, M2, M4 respectively. In order to reflect the compaction method in the real conditions prevailing during the placement of asphalt mix into the road surface and to assess the influence of the method on the properties samples were also made in a slab roller compactor that uses tilting pivoted circular sector with closed loop control system. To achieve various compaction indexes, samples were compacted to variable target heights: 63 mm, 66.2 mm, 69.3 mm, and 59.9 mm. The weight of the batch used for each slab was the same and was calculated taking into account the density of the asphalt mix, assuming that a 63 mm high slab sample should have a compaction index of 98% which is the minimum value required for compacting the asphalt mix under real conditions.

The attempt to produce a sample with a height of 56.7 mm (reduced by 10% compared to 63 mm) failed. The compaction device stopped at a slab height of 58.0 mm and the obtained sample was deformed, thus excluded from further testing.

Cylindrical cores were cut out of each slab for further testing and marked with symbols:

• Cores sampled from slab no. 1, 69.3 mm high (63 mm + 10%)—R1


*3.2. Test Methods*

> Volumetric parameters:

Compaction index was determined on core samples according to the following formula:

$$CI = \frac{\rho\_{core}}{\rho\_{Mars\text{half}}} \times 100 \text{ (\%)}$$

where:

CI—compaction index [%] core—bulk density tested on core sample Marshall—bulk density determined on Marshall samples

Bulk density was determined using the B method—saturated surface dry (SSD) [33] and calculated for each sample according to the formula:

$$
\rho\_{\text{bssd}} = \frac{m\_1}{m\_3 - m\_2} \times \rho\_{wr}
$$

where:

ρbssd—bulk density (SSD) (kg/m3) *m*1—mass of the dry specimen (g) *m*2—mass of the specimen in water (g) *m*3—mass of the saturated surface-dried specimen

ρw—density of water at test temperature (kg/m3)

Maximum density was determined experimentally according to the EN 12697:5:2018 standard [34] and calculated using the following formula:

 (g)

$$\rho\_{\text{\tiny{\tiny{\tiny{\tiny{\tiny{\tiny{\tiny{\tiny{\tiny{\tiny{\tiny{\alpha}}}}}}}}}}=\frac{m\_{2}-m\_{1}}}{1000\times\left(V\_{\text{pp}}-\frac{m\_{3}-m\_{2}}{\rho\_{\text{\tiny{\text{\tiny{\text{\tiny{\text{\tiny{\alpha}}}}}}}}}\right)}$$

where:

ρm—density of the asphalt mix (kg/m3)

*m*1—mass of the pycnometer (g)

*m*2—mass of the pycnometer with the specimen (g)

*m*3—mass of the pycnometer with the specimen and water (g)

ρw—density of water at test temperature (kg/m3)

*<sup>V</sup>*pp—Volume of the pycnometer filled to the measuring line (m3)

Air void content was calculated according to the EN 12697-8:2018 [35] using the following formula:

$$V\_m = \frac{\rho\_m - \rho\_b}{\rho\_m} \times 100\%$$

where:

*V*m—air void content in the asphalt mix specimen (%) ρm—density of the asphalt mix (kg/m3)

ρb—bulk density of the asphalt mix (kg/m3)

> ITSR:

Water and frost resistance tests of the asphalt concrete were performed based on the EN12697-12:2018 standard [36]. This test evaluates the effect of one freeze-thaw cycle of saturated mix asphalt samples on indirect resistance to stretching. Both the samples made in the Marshall compactor and the samples cut from the slabs were divided into two sets. The control series samples were conditioned in a laboratory on flat surface in room temperature of 20 ± 5 ◦C. The samples from the second set were conditioned in water at 40 ◦C for 72 h, then frozen at −18 ◦C for 16 h and re-conditioned in water at 25 ◦C for 24 h. After conditioning, the indirect tensile strength of all test pieces was tested according to EN 12697-23 at 25 ◦C. Based on the obtained results, an indicator of resistance ITSR was calculated:

$$ITSR = \frac{ITS\_w}{ITS\_d}$$

where:

ITSR—indirect tensile strength ratio (%)

ITSw—the average strength for the wet set of samples (kPa)

ITSd—the average strength for the dry set of samples (kPa)

Stiffness modulus:

The tests of the stiffness modulus were performed according to the EN 12697-26:2018 standard [37]. Controlled force was applied to each sample. The samples were subject to five dynamic loads applied to the sample vertically, along the diameter of the base. Force increase time, measured from zero to a maximum value, was 0.124 s. Maximum force generated a horizontal dislocation of sample equal to 5 μm. The sample, after performing the test, was rotated by 90◦ around the horizontal axis and tested again. A reliable stiffness modulus for each sample was a mean average out of two measurements. The final result was the arithmetic mean out of three tested samples. The tests were performed in three temperatures: 23 ◦C, 10 ◦C, −2 ◦C. The following Poisson's ratios were applied for the temperatures, respectively: 23 ◦C–0.4; 10 ◦C–0.3; −2 ◦C–0.25.

### **4. Results and Discussion**

### *4.1. Bulk Density and Compaction Index*

The results of bulk density of the tested samples are presented in Figure 5.

**Figure 5.** The results of bulk density.

Bulk density of the reference samples was 2467 kg/m3. The results for Marshall samples varied from 2364 kg/m<sup>3</sup> to 2475 kg/m3. The core samples cut from the slabs are characterized by bulk density ranging from 2321 kg/m<sup>3</sup> to 2459 kg/m3. In Figure 5, a linear decrease in bulk density can be observed as the slab height increases. A similar trend can be observed for Marshall samples under decreasing compaction energy. The average results of bulk density were the basis for determining the compaction indexes listed in Table 4.

**Table 4.** Specimen determination, bulk density results, air void content results, and compaction index values.


\* Height change of the slab samples (%); \*\* Compaction energy of Marshall samples (Blows)

According to the Polish technical regulations, the compaction index of asphalt concrete during the paving process is at least 98% [38]. On the other hand, slabs prepared in the laboratory for testing resistance to permanent deformations should have a compaction index ranging from 98% to 100% [22]. According to the assumptions, a change in the target height of slab compaction influenced the compaction index value. Samples cut from slabs with a height reduced by 5% (59.9 cm) had a compaction index of 99.7%. This value is within the requirements for both laboratory samples for further testing and for mixtures compacted with rollers during road surface preparation. Slabs of increased thickness were characterized by compaction index at the level of 97.3% and 94.3%, which indicates under-compaction of the mixture. Insufficiently compacted asphalt layers are more susceptible to deeper water penetration and more intensive asphalt oxidation and, as a consequence, to faster surface degradation. On the contrary, excessively compacted asphalt pavements are more susceptible to permanent deformations and low-temperature cracking [39].

The Marshall samples compacted using 50 blows per side are reference samples with a compaction rate of 100% in accordance with Polish requirements. As expected, the compaction index was dependent on the compaction energy. However, as in the case of the slab compactor, this parameter was found to be not very susceptible to compaction. Although similar correlations of density and compaction index were obtained. Proper compaction of the tested mixture is significantly influenced by the applied filler with regular grain shapes. As indicated by Zulkati et al. if the filler has a large enough diameter and a regular shape, it acts as a friction-lubricate agent. This facilitates a faster and smoother reorientation movement of larger aggregates, thereby resulting in high compaction susceptibility [40]. Conversely, Melloti et al. noticed that irregular shaped filler has a negative effect on the workability of asphalt mix [41].

As shown by previous studies, particle orientation and general aggregate structure is significantly different in samples compacted by different methods. Slab samples are characterized by an even particle size distribution across, while samples produced using other methods of laboratory compaction are susceptible to circumferential particle orientation. Contact and interlocking of aggregates, depending on the shape of the aggregate, also has a significant impact on the compaction. Use of angular aggregate particles means achieving more contact points and more uniform distribution of internal forces, with a better interconnection between elements and improvement to fatigue performance as well as permanent deformation resistance [42]. In the tests, crushed aggregates were used, which in combination with the asphalt binder guarantee the durability of the created bonds. Moreover, applied aggregates have an angular shape, which was confirmed by SEM analysis. Finally, the orientation and segregation of aggregates affects both the air void content and the mechanical properties of the asphalt mix [11].

#### *4.2. Density And Air Void Content of the Asphalt Mix*

The results of the air void content calculations are presented in Figure 6.

**Figure 6.** The results of air void content calculations.

Air void content was calculated according to the maximum density of the asphalt mix that was 2530 kg/m<sup>3</sup> and the bulk density of individual samples. The values of the parameter determined on Marshall samples varied from 2.2% to 6.5% and was adequate to the number of blows (Table 4). The air void content of AC 11 S reference samples made was 2.5%. According to technical regulations [19], asphalt concrete can be used for the wearing course if the samples' porosity is between 1 and 3%. The analysis of the obtained results shows that increasing the compaction energy above the standard requirements results in a slight change in the content of air voids; with an increase in compaction by 50% (from 50 × 2 to 75 × 2), a decrease in voids was from 2.5% to 2.2% and the compaction index was 100.3%. On the other hand, the reduction in the number of blows during the compaction of samples results in a considerable increase in free spaces. As indicated by Lucas Júnior et al. in samples with high air void content, the mutual blocking of aggregates is not fully possible because of their shape and texture. As a consequence, the functional properties of the produced mix deteriorate [43].

The content of air voids determined for samples cut out of slabs was from 2.8% to 8.3%. The value increased with the decrease of the compaction index, which is the expected relation. The core samples with a 99.7% compaction index were characterized by a free space content of 2.8%, which was closest to the results obtained for reference samples. It was also noted that in samples with the 98.8% compaction index, the free air void content was above the upper limit according to Polish requirements and was 3.7%, however the requirement of 1 to 3% porosity refers to samples made using the Marshall method. In case of surfaces made of AC 11 S mix compacted with rollers in real conditions, the technical recommendations [38] indicate that this layer should be compacted at a rate above 98% and contain between 1 and 4.5% air voids. It can be concluded that the samples made in a roller compactor with a compaction rate of at least 98% meet the requirement for the porosity of the asphalt layer. Nevertheless, the results of these tests are based on the assumed simulation of compaction by rolling in the laboratory, and the differences in the results are the effect of different slabs thicknesses. Under real conditions, the correct compaction of each layer is expected. In addition, one of the important elements influencing the proper effect of this process is the temperature of the asphalt mix—the lower the larger the voids in the layer, which is confirmed by both volumetric studies [16,44] and x-ray computed tomography analyses [45]. Wang L et al. noticed that not only the percentage content of air voids affects the performance of the surface, but also the spatial distribution and pore diameter [46]. Jiang W et al. proved that the characteristics of air voids are significantly influenced by aggregate

gradation [47]. Moreover, the structure of aggregate, number of contact points, and orientation of each aggregate are dependent on the compaction methods and conditions [48].

### *4.3. Water and Frost Resistance*

The results of the indirect tensile strength test are shown in Figure 7; Figure 8 as well as Table 5. The strength obtained for the samples compacted using Marshall method was in the range of 504 kPa to 648 kPa in the dry set and 421 kPa to 598 kPa in the wet set. It was observed that the results were dependent on the compaction energy, whereas in the case of 2 × 75 blows the strength increased only by 3% in the dry set and 4% after freezing cycle comparing to the reference samples.

**Figure 7.** The results of indirect tensile strength test for the wet set.

**Figure 8.** The results of indirect tensile strength test for the dry set.


**Table 5.** Average results of indirect tensile strength for the wet set and standard deviations.

Considering the core samples in general, it was noted that the strength increased with the growth of the compaction index, however the strength of the samples with the highest compaction index (99.7%) turned out to be lower than expected. This is probably due to the over-compaction of the asphalt mix. A further consequence of excessively "dense" mixtures is a decrease in resistance to permanent deformations and low-temperature cracking [39]. It was also observed that asphalt mix has reduced strength characteristics if the air void content is above the required limits. This is particularly the case for R1 samples where the porosity was the highest (8.3%) and the strength characteristics were lowest, 406 kPa for dry set and 334 kPa after the freezing cycle respectively. The acceptable limit of voids content was also exceeded in R4 samples, which is reflected in their strength.

Correlations between the compaction methods and the obtained strength results were observed. At a similar value of compaction index Marshall samples were more resistant than cores, in a similar range from 54 kPa to 91 kPa on average in dry set and from 33 kPa to 87 kPa after freezing cycle. The di fference may be caused by change in aggregate interlock pattern in the particular method [49]. Thus, the strength of the samples is directly related to the compaction process, which significantly affects all properties of the asphalt mix.

The results of water and frost resistance tests are presented in Table 4. The average ITSR parameter for reference Marshall specimens was 91.7%. In case of the other Marshall samples the value varied from 83.5% for the least compacted to 92.3% for the most compacted ones. R4 core samples, which on the basis of the test results were classified as over-compacted had high resistance to water and frost of 91.2%. The same was observed for most dense Marshall samples. This is probably the e ffect of very low air void content limiting water absorption in a natural way, which resulted in a slight decrease in strength of samples subjected to one freeze-thaw cycle. As predicted, the lowest ITSR value was obtained on R1 samples that were much under-compacted and had the highest air void content. Thus, moisture damage appeared, which is associated with both loss of cohesion and loss of adhesion [50]. In reference to the nominal value of Marshall compaction energy, R3 samples was less frost resistant than expected, even though they were properly compacted and had the highest strength among the samples cut out of slabs. However, in tests of asphalt mix properties there is not always a positive correlation between strength and frost resistance measured by ITSR [51,52].

As is well-known, moisture damage resistance of asphalt mix is strongly related to the aggregate-binder adhesion. Among the aggregates used, dolomites with CaO content of 39.62% prevailed. Alkaline aggregates are characterized by excellent adhesion to asphalt, which results not only in high water resistance but also in good strength parameters. Lucas J únior et al. indicates that the indirect tensile strength also depends on aggregate sphericity and texture [43].

The mixture also contains filler with a very high lime content. It is naturally hydrophilic material with a tendency to form strong bonds with hydrophobic organic compounds such as bitumen [53].

### *4.4. Sti*ff*ness Modulus*

The results of sti ffness modulus tests are presented in Figures 9–11 and Table 6. The values of the parameter determined on the Marshall samples varied from 1384 MPa to 2197 MPa in 23 ◦C, 4404 MPa to 6201 MPa in 10 ◦C, 11,075 MPa to 14,446 MPa in −2 ◦C. Compaction index was found to have a linear correlation to the increasing sti ffness modulus.

**Figure 9.** The results of stiffness modulus tests in 23 ◦C.

**Figure 10.** The results of stiffness modulus tests in 10 ◦C.

**Figure 11.** The results of stiffness modulus tests in −2 ◦C.



The stiffness of the samples cut out of the slabs increased as the density index increased, but only to a certain level. In case of 99.7% compaction index core samples, results were 1660 MPa in 23 ◦C, 4590 MPa in 10 ◦C, and 12162 MPa in −2 ◦C and were lower than the values obtained for samples with 98.8% CI value. These results confirm previous assumptions that R4 samples with the highest compaction index (99.7%) were over-compacted, whereas in case the Marshall samples compacted with increased energy a decrease in stiffness did not occur. Moreover, for similar compaction indexes, samples made using the Marshall method had higher stiffness than samples cored from slabs, which is consistent with the research of Hartman et al. [54]. This may be due to different compaction energies and the size and type of mold in the individual methods [11].

High stiffness modulus asphalt mixes are also characterized by higher resistance to rutting at positive temperatures [55]. Based on the analysis of the test results obtained, it can be concluded that over-compacted mixtures not only have lower strength but are also less resistant to high temperatures, including the development of permanent deformations. At negative temperatures, on the other hand, higher values of the asphalt mix stiffness modulus are expected.

Asphalt changes its properties as a result of oxidation processes and, as time passes, asphalt surfaces become even stiffer and consequently more susceptible to low temperature cracks and premature degradation [39,56]. Therefore, controlling the stiffness of the designed asphalt mix is a factor that can prevent early defects of the surface. It is particularly important in the case of uncertain compaction process and adaptation of additives softening the asphalt binder [57–59].

In the conducted research the core samples that occurred to have the lowest stiffness in −2 ◦C (9661 MPa) are characterized by the lowest compaction index (94.3%), the highest porosity (8.3%), and the lowest frost resistance (ITSR = 82.4%).

The results clearly indicate that the sti ffness modules tend to decrease as the voids increase, which is the expected relation [9,56].
