Low temperature properties of hot mix asphalts prepared with different polymer modified binders

The low temperature crack propagation resistance of pure and polymer modified mixtures is analysed in the paper by means of linear and nonlinear fracture mechanics. The influence of two bitumen-modification binders is considered: styrene-butadienestyrene (SBS) and ethylene-vinyl-acetate (EVA). The experiments demonstrated that the fracture toughness values increase, and the maximum vertical strain values decrease, with a decrease in temperature. It was established by both techniques used in the paper that the lowest fracture toughness was obtained using the EVA binder.


Introduction
Many pavement distresses, described as rutting, moistureinduced damage, low temperature cracking etc., may appear on asphalt pavements that have been in use for an extended period of time.Thermal fatigue cracking is one of the major failure modes for asphalt pavements that are subjected to alternating heating and cooling, while low-temperature cracking is a serious distress mode in very cold regions [1][2][3].Failure by low temperature cracking occurs as a result of thermal stress buildup in the pavement.Microcracks appear on the surface of the pavement in cases when this stress becomes equal to or greater than the tensile strength of pavement.Propagation of cracks through the pavement is likely to occur as a result of continuing low temperature cycles [4].When water fills these cracks during winter, it freezes and ice lenses as well as frost heave may develop.This condition results in the loss of fines and formation of voids under the pavement, leading to reduction in load bearing capacity of the pavement.It is important to take appropriate measures during design of pavements to be used in cold regions.Otherwise cracking may cause very serious problems, such as poor ride quality, reduced service life, and high maintenance costs [5].Polymeric materials can be used as modifiers in bitumen and hot mix asphalt (HMA) mixtures so that a wider performance range can be achieved [6][7][8].When plastomers are incorporated into bitumen, the resultant structure is a rigid three dimensional network that resists deformation.However, structures with elastomeric features, i.e., those resisting permanent deformation and recovering the original shape after loading, are obtained by using elastomeric modifiers in bitumen [9].Styrene-butadiene-styrene (SBS) and ethylene-vinyl-acetate (EVA) are two materials that are most frequently used for bitumen-modification purposes [10].EVA has been revealed to be a good modifier that improves resistance to permanent deformation and thermal cracking of HMAs [11,12].EVA is a copolymer consisting of 5 to 50 wt.% of vinyl acetate (VA).Acetate groups in the ethylene chain decrease the crystallinity of the copolymer whose properties are controlled by the amount of VA present in the structure.As the VA content increases, the increase in flexibility accompanies the decrease in crystallinity of the copolymer.A flexible copolymer exhibits low melting points and heat seal temperatures, as well as reduced stiffness, tensile strength and hardness [13].SBS, a copolymer of styrene and butadiene, has a rubbery structure with a three dimensional network of physical cross-links.In the copolymer, while polystyrene (PS) blocks impart strength to the resin, soft polybutadiene (PB) blocks provide elasticity [14].The structure of PB block in the SBS copolymer can be adjusted by means of special catalysts.This involves partial transfer of the double bonds on the PB blocks to the side chains.It was found that this modification has many advantages for the PMBs, such as low viscosity and better compatibility at equivalent molecular weight, resistance to oxidation, and thermal stability [15,16].
According to the loading conditions, i.e., speed and temperature, fracture mechanics is used to study the fracture mechanism of hot mix asphalts.Principles of linear-elastic fracture mechanics (LEFM) are applicable to HMAs for the temperatures of 1 °C or less, while elastic-plastic fracture mechanics (EPFM) principles are applied at convenient loading speeds for the temperatures of 25ºC and above [17].Fracture toughness is the ability of a material with a crack to resist fracture.When a stress intensity factor (K 1 ) exceeds fracture toughness (K IC ), fracture occurs according to LEFM principles.It is obvious that in order to meet requirements for a satisfactory service life, asphalt pavements should have high fracture toughness values so that they can withstand loads when exposed to an extended range of temperatures.Formation of localized asphalt damage can be observed when the temperature drops under a certain level.This condition, along with the change in the micro-structural stress mechanism, determines the fracture resistance behaviour of asphalt concrete with respect to temperature [18].However, experimental studies revealed that LEFM is not valid for quasi-brittle materials such as concrete, rock and asphalt concrete because K Ic depends on the size and geometry.The inapplicability of LEFM is due to existence of an inelastic zone, the fracture process zone (FPZ), in front of a crack in quasi-brittle materials.Several non-linear fracture mechanics models have been developed to characterize the FPZ.These models can be classified as the cohesive crack models and the effective crack models.Cohesive crack models simulate the FPZ by a closing pressure, which diminishes near the crack tip, while effective crack models simulate the FPZ by an effective crack length.The main aim of any approach is to determine the critical crack extension (size of FPZ) at peak load.A long term cyclic fracture, due to cycling loading or change in temperature, and short term cyclic fracture, due to an abrupt drop of temperature along with contraction stresses, are two typical fracture models.Present study mainly deals with the short term cyclic fracture which is a one time fracture failure problem.The crack propagation resistance of pure and polymer modified mixtures is examined in this study by means of two different experimental techniques, and in accordance with the both LEFM and non-linear elastic fracture mechanics, as considered in the above discussion.

Materials and sample preparation
The materials used in the experimental study were PG 58-34 bituminous binder, supplied by TUPRAS Batman Refinery; two SBS modifiers, i.e., Kraton D 1101 and Kraton MD 243, produced by Shell Chemical Co.; and Evatane ® 2805, a type of EVA modifier, produced by Arkema.For the production of modified bitumens, pure bitumen was mixed with a selected modifier for 60 minutes at 180 °C.The speed of the mixer was fixed to 1000 rpm throughout the mixing operation.
Low temperature properties of hot mix asphalts prepared with different polymer modified binders Malatya, a city in the eastern part of Turkey, was selected in the scope of this study as the area for application of the final binder.After taking traffic and climate conditions of the city into account, PG 70-22 was selected as the desired binder performance grade.The binder and hot mix asphalt design were made in this study according to Superpave method.It was reported in previous studies that 3-7 % SBS or 2-6 % EVA could be added to pure bitumen so that the permanent polymer phase can be obtained [19].In order to compare the effects of additives on the HMA properties, and based on the information given in literature, the amount of additive was kept constant at 4 wt.% in all mixtures.Table 1 shows results of the dynamic shear rheometer (DSR) and bending beam rheometer (BBR) tests for pure bitumen as well as mixtures modified with 4 % SBS D1101 (MB SBS-D ), SBS MD243 (MB SBS-M ) and Evatane ® 2805 (MB EVA ).As can be seen in the table, performance grades of MB SBS-D and MB SBS-M are the same, i.e., PG 70-34.However, the low temperature performance grade of mixture with MB EVA was one level lower, namely PG 70-28.Thus, it can be stated that the performances of MB SBS-D and MB SBS-M are better than MB EVA at low temperatures. .It should be noted that all modified bitumens comply with the desired binder performance grade, PG 70-22.In order to determine the mixing and compaction temperatures of hot mix asphalts, rotational viscometer tests were also conducted for the unaged pure and modified binders at 135 °C and 165 °C, respectively.Using the collected viscosity data, a temperature-viscosity graph with the corresponding trend line was drawn to interpret the temperature dependency of the viscosity.The plot was used to determine the mixing and compaction temperatures of the mixtures.According to literature, bitumen binders should possess viscosities of about 170 ± 20 cP for mixing and 280 ± 30 cP for compaction [20], (Note: 1 centipoise [cP] = 0.001 pascal second [Pa•s]).Based on this information and the temperatureviscosity graph, the mixing and compaction temperatures were identified for the corresponding viscosity values.Meanwhile, as can be seen in Table 2, the binder fulfilled the workability requirement stating that the viscosity value at 135 °C should not exceed 3 Pa .s (3000 cP), and so that workability could be maintained [21].It should be noted that the viscosity of the binder increased with the use of the binders, in which case the mixing and compaction temperatures also increased.Physical properties of the crushed limestone aggregate used in the mixture are summarized in Table 3; the gradation is presented in Figure 1.

Temperature
[°C] Taner Alataş, Mesude Yilmaz °C for mixtures prepared with MB SBS-M ) using a special mixer.Uncompacted samples were placed on plates in an amount of 21-22 kg/m 2 .Then, they were placed into an oven preheated to 135 °C, where they were short-term aged for 4 h.After ageing, the samples were compacted at the rate of 30 rpm by gyratory compactor having compaction angle of 1.25 o .In the compaction process, vertical compression of 600 kPa was used for 100 rotations.Design binder contents (DBC) of compacted specimens were determined by means of their volumetric properties.As can be seen in Table 4, which shows volumetric properties and Superpave specification limits of the pure and polymer-modified mixtures, DBC values increase with the use of modified binder.The table demonstrates that all mixtures prepared within the scope of the study meet the Superpave specification criteria.

Semi-circular bending (SCB) test
The experimental procedure described in EN 12697-44 was used to measure the resistance of HMA specimens to crack propagation based on the concept of LEFM principles [22].A gyratory compactor was used for preparation of specimens measuring 150 mm in diameter and 120 mm in thickness.Thus, the specimens contained 4 vol.% of air voids at the end of compaction.The compacted specimens were sliced into two equal semi-circular pieces.Then, the resulting semi-circular specimens were cut into two equal slices, each 50 mm in thickness.A single notch about 10 mm in depth and 1.5 mm in width was carved in the middle of the specimens.At the three-point loading configuration, the deformation was performed at the rate of 5.0 mm/min, see Figure 2 for the loading configuration details.Three different temperatures, i.e., 0 o C, -10 o C and -20 o C, were selected for the SCB test.Specimens were kept at the test temperature for 6 h before the testing.The load utilized and the corresponding deformation values were recorded throughout the SCB experiments.The fracture toughness (K IC , N/mm 3/2 ) and maximum vertical strain values (ε max , %) were calculated using the experimentally determined parameters, i.e., maximum stress value at failure (s max , N/mm 2 ), maximum force value (F max , N), and deformation value at maximum force (ΔW, mm).For this purpose, the following equations were used: Low temperature properties of hot mix asphalts prepared with different polymer modified binders (3) where, D, W and t are the diameter, thickness and height of specimens in mm, respectively.The variation of the fracture toughness (K IC ) value with the type of additive and temperature is given in Figure 3.As shown in Figure 3, the fracture toughness value increases with a decrease in temperature.At all temperatures, the lowest K IC value was obtained in MB EVA mixtures.The highest value, on the other hand, was obtained in MB SBS-D at 0 °C and MB SBS-M at -10 °C and -20 °C.At the temperature of 0 °C, the K IC value of the control mixture (prepared with PG 58-34) was by 24.9 % higher than that of MB EVA .Similarly, at the same temperature, the K IC value of the mixture with MB SBS-D was found to be by 26.5 % higher than that of the mixture with MB EVA and by 20.9 % higher than that of mixture with MB SBS-M .At the temperatures of -10 °C and -20 °C, the mixture with MB SBS-M , which had the highest K IC value overall, had the K IC value by 17.9 % and 15.7 % higher than that of the mixture with MB EVA , respectively.It was detected that K IC values increase significantly if the temperature drops from 0 °C to -10 °C.In such a case, the increase in the K IC value was 22.9 %, 18.4 %, 34.1 % and 37.5 % for the mixture with pure bitumen (PG 58-34), MB SBS-D , MB SBS-M and MB EVA , respectively.However, the increase was less significant when the temperature was changed from -10 °C to -20 °C.This was acquired if the increase in K IC value was calculated for temperature drop from 0 °C to -20 °C.For this condition, it was detected that the increase was 27.8 %, 28.7 %, 35.9 % and 41.9 % for mixture with PG 58-34, MB SBS-D , MB SBS-M and MB EVA , respectively.This also shows that the mixture with MB EVA was affected the most by temperature drop.Variations of maximum strain (ε max ) values with temperature and the type of additive are given in Figure 4.As shown in Figure 4, the mixture with MB EVA had the smallest ε max value at all temperatures.In contrast to that the highest value was obtained for MB SBS-M at 0 °C as well as at -20 °C and for MB SBS-D at -10 °C.At the temperatures of 0 °C and -20 °C, the ε max value of the mixture with MB SBS-M was by 19.8 % and 12.6 % higher than that of mixture with MB EVA , respectively.It was found that the difference was 14.7 % at -10 °C.
It was observed that maximum strain values decrease with the decrease in temperature.When the temperature decreased from 0 °C to -10 °C, the decrease in ε max value was 13.9 %, 3.5 %, 14.5 % and 5.6 % for the mixture with pure binder, MB SBS-D , MB SBS-M and MB EVA , respectively.In case of temperature reduction from 0 °C to -20 °C, the decrease in K IC value was 25.3 %, 14.5 %, 20.6 % and 13.4 % for the mixture with pure binder, MB SBS-D , MB SBS-M and MB EVA , respectively.It can be seen that the ε max value of the MB EVA mixtures was affected the least from temperature drop, as the ε max value of this mixture was rather small at all temperatures.
According to semi-circular bending tests, it can be stated that the mixtures with maximum resistance to crack propagation at 0 °C would be MB SBS-D , while at -10 °C and -20 °C, such mixtures would be MB SBS-M .In contrast to those, the resistance to crack propagation at low temperatures is fairly low for MB EVA mixtures.Meanwhile, the experiments pointed out that with the addition of the elastomeric type of additives (Kraton D 1101 and Kraton MD 243), the ε max value of the binder can be increased at lower temperatures, while the ε max value decreases with the use of plastomeric type of additives (Evatane® 2805).

Single edge notched beam test
The single-edge notched beam (SE(B)) test geometry has been frequently applied [23,24] for determining the fracture toughness value of asphalt concrete.Compared to other proposed geometries, this geometry yields more reliable results in the determination of fracture toughness, by providing simple loading configurations, by minimizing the edge effect by large dimensions and stable crack propagation in Mode I [25].In fact, a standard was not previously set for determining fracture properties of prism shaped asphalt concrete.For this reason, dimensions used by Kim and Hussein (1997) were selected in this study as specimen dimensions [18].To this end, first slab-shaped specimens measuring 30.5 × 30.5 × 5.0 cm were compacted by roller compactor to 4 vol.% of air voids.The amount of HMA necessary for slab specimens was calculated using the following equation: After the complete cooling of the specimen, it was cut to the required dimensions as given above.First, an insert positioned at the bottom centre of the beam mould was used to form a notch on the specimens having initial notch-depth (a 0 ) of 21 mm and initial notch-depth to beam-depth (W) ratio of 0.3.By using the three-point bending configuration, the specimens were deformed at a constant cross-head deformation rate of 3.24 mm/min.The configuration is shown in Figure 5.The loading span/notch depth ratio (S/W) of 4 was adopted in all deformation experiments.The effective crack model (ECM) was applied to determine the fracture behaviour of the specimens.By using P i as P max /2 [N], δ i as deformation at P i [mm]; B, W and S as width, depth and length of specimens [mm]; w as weight per unit length of beam, the Young's modulus (E) of the mixtures was calculated using the following equation [26]: Here, the value of F 2 (α 0 ) is determined as follows: (6) By using the known values ofα 0 = a 0 / W and S/W, F 1 (β) can be re-written as (7) process zone ahead of a visible crack are two main reasons for the reduction of beam stiffness.However, as it is rather difficult to distinguish between them, it is assumed that the critical notch depth (a e ) could be calculated by introducing a fictitious beam containing a notch a e ,.It should be noted that, here, the notch has an unchanged stiffness that would be equal to the reduced stiffness of the real beam containing a notch of depth a 0 , i.e. (8) where (9) It should be noted that the values of α e is a e /W and F 1 (β) is directly obtained by using Equation 7.Then, the following equation can be applied to calculate the critical stress intensity factor (K IC ) of the specimens; (10) where (11) Variations of maximum load values (P max ) of the mixtures with the type of additive and temperature are given in Figure 6.As shown in Figure 6, the maximum load value increases with the decrease in temperature.At all temperatures, the lowest value of P max was obtained in MB EVA .The highest value, on the other hand, was obtained in the mixture with MB SBS-D at temperatures Low temperature properties of hot mix asphalts prepared with different polymer modified binders of 0 °C and -10 °C.At -20 °C, the highest value was obtained in the mixture with MB SBS-M .It was found that the P max value of the mixture with MB SBS-D was by 18.9 % and 15.4 % higher than those with MB EVA at 0 °C and -10 °C, respectively.At -20 °C, the highest P max value, obtained from the mixture with MB SBS-M , was by 25 % higher than the lowest P max value, which was obtained from the mixture with MB EVA .

Figure 6. Variation of P max values with type of additive and temperature
There was a significant increase in the P max value when the temperature was decreased from -10 °C to -20 °C.In this condition, P max values of the mixture with pure binder, MB SBS-D , MB SBS-M and MB EVA increased by 31.2 %, 26.0 %, 28.3 % and 17.9 %, respectively.Similarly, when the temperature was decreased from 0 °C to -10 °C, P max values of the mixture with pure binder, MB SBS-D , MB SBS-M and MB EVA , increased by 12.6 %, 6.5 %, 8.9 % and 9.7 %, respectively.These results show that the mixture with pure binder was affected the most by temperature change in terms of P max value.The variation of deformation (δ max ) value at maximum load with temperature is given in Figure 7.As shown in Figure 7, the highest δ max value was obtained in the mixture with MB SBS-D at the temperature of 0 °C.At the temperatures of -10 °C and -20 °C, the highest value was registered for MB SBS-D and MB SBS-M , respectively.It was identified that at -10 °C, δ max values of MB SBS-D and MB SBS-M were by 4.0 % and 3.3 % higher than that of the mixture, respectively.δ max values decreased regularly with the decrease in temperature.The variation of fracture toughness values with the type of additive and temperature is given in Figure 8. Figure 8 shows that the K IC value increases at all temperatures with the use of SBS in mixtures, while it decreases with the use of EVA.At the temperatures of 0 °C and -20 °C, the highest value was obtained in MB SBS-D .It was found that at -10 °C, the highest value was obtained for the mixture with MB SBS-M .At the temperatures of 0 °C and -20 °C, the K IC value of the mixture with MB SBS-D was by 15.8 % and 18.6 % higher compared to the one with MB EVA , respectively.At -10 °C, the K IC value of the mixture with MB SBS-M was by 18.5 % higher compared to the one with MB EVA , respectively.K IC values increased regularly with the decrease in temperature.
When the temperature was decreased from 0 °C to -10 °C, K IC values of the mixture with pure binder, MB SBS-D , MB SBS-M and MB EVA , increased by 12.5 %, 9.1 %, 14.6 % and 10.0 %, respectively.Similarly, when the temperature was decreased from 0 °C to -20 °C, K IC values increased by 38.0 %, 34.8 %, 35.4 % and 31.6 %, respectively.

Comparison of results of fracture mechanics tests
When compared with Figure 3 based on LEFM and Figure 8 based on the non-linear fracture mechanics, K Ic values obtained from beam tests (approximately 30 Nmm 3/2 ) were greater than those of semi-circular bending tests (approximately 20 Nmm 3/2 ).This points to the existence of an inelastic zone (FPZ) in front of a crack in asphalt materials in even lower temperature ranges.The ECM based on nonlinear fracture mechanics can also be employed to determine whether the quasi-brittle material behaviour is ductile or brittle.According to ECM, the relative length of FPZ (α eα 0 ) gives a good indication of brittleness.For instance, the length of FPZ is zero for perfectly brittle materials and LEFM is applicable.As shown in Table 5, 6 and 7, the relative effective crack length (α e ) slightly decreases with the decrease in temperature.Consequently, it may be concluded that the beam specimens exhibit a more brittle behaviour with the decrease in temperature.However, it is well known that the tensile strength of asphalt materials, which is the most important parameter in cohesive crack approaches, increases with Taner Alataş, Mesude Yilmaz  the decrease in temperature.As clearly indicated in Table 5 and Table 6, the nominal strength values of beams highly increased with the decrease in temperature.Finally, it was concluded based on this discussion that the critical stress intensity factor values of the beam specimens increase with the decrease in temperature.
Low temperature properties of hot mix asphalts prepared with different polymer modified binders

Comparison of experimental procedures of fracture mechanics
Fracture mechanics tests were performed in the scope of the study on both semi-circular and rectangular prism shaped samples using the principles of fracture mechanics.In both experiments, fracture toughness values were calculated at different temperatures.SE(B) test results are given in Table 5, 6 and 7. SCB test results are given in Table 8, 9 and 10.In order to elucidate the linear relationship and its strength between Correlation coefficients (r) for K IC values obtained from SCB and SE(B) samples at 0 ºC, -10ºC and -20ºC were calculated.It was determined that these correlation coefficient values were 0.930, 0.938 and 0.990, respectively.These values demonstrated a positive correlation between K IC values.Furthermore, the analysis of correlation strength demonstrated that r values were greater than 0.90 at all three temperatures (0 ºC, -10ºC and -20ºC).Therefore, there was a strong correlation between the results obtained with both experimental methods.Furthermore, it was identified that the strength of this correlation increased with the decrease in temperature.

Conclusions
The low-temperature crack propagation resistance of control and polymer modified mixtures was examined in this study using two distinct experimental techniques.Within this scope, the design of binder and mixture was conducted using the Superpave method.
According to semi-circular bending tests performed, fracture toughness values increased with the decrease in temperature, while the maximum vertical strain values decreased.At all temperatures, the lowest K IC value was obtained in MB EVA .The highest value, on the other hand, was obtained in MB SBS-D at 0 °C and MB SBS-M at -10 °C and -20 °C.Single edge notched beam experiments revealed that fracture toughness values based on ECM increase, while the maximum vertical deformation value decreases with the decrease in temperature.At all temperatures, the lowest K IC value was obtained in MB EVA .The highest value, on the other hand, was obtained in MB SBS-D at 0 °C and -20 °C, MB SBS-M at -10 °C.
When the effects of binders are compared, the experiments show that the low-temperature performance of MB SBS-D and MB SBS-M is higher compared to MB EVA .The experiments in which mixtures were compared with each other showed that, at low temperature, the resistance to crack propagation increases with the addition of SBS type elastomers to pure mixtures.For EVA type plastomers, the resistance is lost to some extent.When the experimental procedures are compared, it can be stated that two techniques are coherent and approximately linear to one another.

Figure 4 .
Figure 4. Maximum vertical strain (ε max ) values of mixtures weight of the specimen [kg] L -the inner length of the mould [mm] l -the inner width of the mould [mm] e -the final height of the specimen [mm] ρ m -the maximum density of the bituminous mixture [kg/m 3 ] v -the air voids of the specimen [%].

Table 7 . SCB test results at 0°C and -10°CTable 8 . SCB test results at -20°C
It was revealed by analysis results that K IC values for SCB samples could be explained by K IC values of SE(B) samples at the rate of 86.41 % and at 0 °C, whereas the same rate was 88.07 % and 98.06 % at -10 and -20 °C, respectively.