Engineering characterisation of epoxidized natural rubber-modified hot-mix asphalt

Road distress results in high maintenance costs. However, increased understandings of asphalt behaviour and properties coupled with technological developments have allowed paving technologists to examine the benefits of introducing additives and modifiers. As a result, polymers have become extremely popular as modifiers to improve the performance of the asphalt mix. This study investigates the performance characteristics of epoxidized natural rubber (ENR)-modified hot-mix asphalt. Tests were conducted using ENR–asphalt mixes prepared using the wet process. Mechanical testing on the ENR–asphalt mixes showed that the resilient modulus of the mixes was greatly affected by testing temperature and frequency. On the other hand, although rutting performance decreased at high temperatures because of the increased elasticity of the ENR–asphalt mixes, fatigue performance improved at intermediate temperatures as compared to the base mix. However, durability tests indicated that the ENR–asphalt mixes were slightly susceptible to the presence of moisture. In conclusion, the performance of asphalt pavement can be enhanced by incorporating ENR as a modifier to counter major road distress.


Introduction
Asphalt is one of the most widely used materials in road construction because of its mechanical, viscoelastic, adhesive, and waterproofing properties [1]. The largest use of asphalt is in the production of hot-mix asphalt (HMA), which is primarily used in the construction of flexible pavements. Durable pavements with long service life and low maintenance and rehabilitation costs are preferred; hence, the demand for high-quality asphalt is growing. Consequently, new materials with increased performance such as polymer-modified asphalt (PMA) are being researched and developed. On the other hand, the need for suitable testing methods that can forecast the performance of materials not only during laying but also over the entire working life of the pavement is becoming increasingly important. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 (HMA-12) ENR (as a percentage of asphalt by weight) in terms of indirect tensile resilient modulus, dynamic creep, rutting, fatigue, and its susceptibility to moisture.

Materials and methods Materials
The asphalt used in this study was 80/100 penetration grade supplied by the asphalt factory at Port Klang, Malaysia. The ENR was obtained from the Malaysian Rubber Board under the trade name of ENR 50, with 53% epoxidation. It was passed through a 2.36 mm mesh sieve (before shearing). The physical properties of the asphalt and ENR used are shown in Table 1. The aggregate used for preparing the mixed samples was obtained from Negeri Roadstone Sdn. Bhd., Malaysia.
Preparation of the binder ENR-modified asphalts were produced by mixing 3, 6, 9, and 12% ENR (by weight of asphalt) with the base asphalt using a high shear mixer at 160˚C (±1˚C) under 4000 rpm for 60, 62, 66, and 66 min, respectively [21]. ENR was added to the asphalt after temperature stabilization at 160˚C.

Asphalt mix preparation
The asphalt mix design was based on the Superpave volumetric mix design. The design equivalent single-axle loads (ESALs) for this study were assumed to be less than 10 7 . This categorises the design in the 3 × 10 6 to 1 × 10 7 ESALs (80 KN/ESAL) category or traffic level 4 [22]. Traffic levels are used to determine design requirements, such as the number of design gyrations for compaction, the physical properties of the aggregate, and the volumetric requirements of the mix. The traffic level also determines the level of mix design required. For traffic levels of less than 10 7 ESALs, a level 2 mix design is recommended [22]. The mix in this study had a nominal maximum particle size aggregate of 19.0 mm. However, six stockpiles of the aggregate consisted of three coarse and three fine aggregates.

Asphalt mix performance tests
Performance tests were used to relate the laboratory mix design to actual field performance. The following section focuses on laboratory tests done on asphalt mixes.
Indirect tensile resilient modulus test. Sample preparation and testing procedures for the indirect tensile resilient modulus test were based on the ASTM-D4123 [23] standard and were carried out using a universal testing machine (UTM) (S1 Fig). The resilient modulus (M R ) is one of the most fundamental and important parameters used among researchers to determine the mechanical properties of asphalt mixes, which is then used in the mechanistic design of pavement structures. It is a non-destructive test and can be defined as the ratio of applied stress to recoverable strain at a particular temperature for a given load.
In this study, three replicate samples (100 mm in diameter and 63.5 ± 2.5 mm in height) for each type of mix were compacted at 4% air voids using a gyratory compactor. Prior to testing, all the samples and test equipment were conditioned at 5, 25, and 40˚C in an environmental chamber for a minimum of 2 h. The preconditioned samples were set with a haversine wave pattern with five conditioning pulses followed by five loading pulses during testing, after which the data were recorded.
In addition, the pulse width of the load was set at 100 ms with a pulse repetition period ranging from 300 to 1000 ms. With respect to the pavement, Tayfur et al. [24] suggested a pulse repetition period of 1000, 2000, and 3000 ms in order to simulate high, intermediate, and low traffic volume, respectively.
Dynamic creep test. The dynamic creep test, which can also be performed using the UTM (S2 Fig), determines the permanent deformation of the asphalt mix in a laboratory setting. In this study, the dynamic creep test under a haversine load pulse was conducted for the base asphalt mix and the ENR-modified mixes according to the protocols developed by the NCHRP 9-19 Superpave Model [25]. Each category of ENR mix consisted of three replicate samples, which were compacted at 4% air voids, with a diameter of 100 mm and a height of 150 mm. The samples were preconditioned in the environmental chamber at 40˚C for at least 4 h before testing commenced. During the initial stages of testing, samples were pre-loaded with a conditioning stress of 10 kPa for 120 s to ensure the platen was flat against the sample. Subsequently, a haversine wave load cycle was applied, which consisted of a 100 kPa stress pulse with a 100 ms pulse width, followed by a 900 ms rest period. The test was terminated when the accumulative strain reached 10,000 micro-strains or until 10,000 cycles were completed, whichever came first.
Rutting test. This test simulates wheel passes on pavement. The procedure followed is as described in BS-598-110 [26]. A Wessex wheel-tracking device was used to perform this test on a laboratory scale. Wheel-tracking test samples were produced for each mix once the respective design aggregate and optimum binder content (OBC) was known (S3 Fig). In this study, one mould was used to hold the rut sample in the wheel-tracking machine. The height of the mould exactly matched the original slab mould of the Wessex wheel-tracking device. Approximately 3,700 gm of mix was compacted to 7 ± 0.5% air voids to obtain a sample with an area of 300 × 300 mm and a height of 65 ± 1 mm. The samples were allowed to cool for 24 h at room temperature after compaction before calculating the air voids content as per test requirements. The rut test was conducted at 50˚C. Before testing, the samples were conditioned for at least 4 h at the test temperature. The test was conducted only in dry conditions. The samples were subjected to simulated traffic with a simple harmonic motion by applying 520 N loads for 1 h.
Flexural fatigue test. The flexural fatigue test is conducted to evaluate the fatigue characteristics of the mixes. However, fatigue cracking of pavement is considered to be more of a structural problem than simply a material problem. A repeated load flexure testing machine was developed to perform controlled strain flexure fatigue tests on asphalt mix beams. Symmetrical four-point loading produced a constant bending moment over the middle third of a 375-mm-long beam sample that was 50 mm deep and 62.5 mm wide (S4 Fig). The test was conducted in accordance with AASHTO [27]. Particularly, repeated haversine loads were applied to the third point of the sample, which was placed inside the environmental chamber to maintain the temperature at 20˚C during the test. In the constant strain beam fatigue test, a constant strain was maintained on the sample, but the magnitude of the load was allowed to decrease with increasing load cycles. With this method, the samples did not break during the test. Instead, failure was defined as 50% loss in initial beam stiffness. Moisture susceptibility. It is very important to evaluate the moisture sensitivity of the design mix. This step is accomplished by performing the AASHTO method on the design aggregate structure to determine the design asphalt binder content [28]. Base asphalt and ENR-modified asphalt samples with diameters of 100 mm and heights of 63.5 ± 2.5 mm were compacted to approximately 7% air voids. One subset of the three samples was defined as the control subset and the rest were defined as the conditioned subset. The conditioned subset was subjected to partial vacuum saturation followed by an optional freeze cycle and a 24 h thaw cycle at 60˚C. All samples were tested to determine their indirect tensile strength. The moisture sensitivity was defined as the ratio of the tensile strength of the conditioned subset to the tensile strength of the dry (control) subset. The minimum criteria for tensile strength ratio was 80% [28].

Results and discussion
Performance testing is an area of Superpave that has yet to be implemented widely. The performance tests discussed in this study have been used by various researchers. However, as with asphalt binder characterisation, the challenge in HMA performance testing is to develop physical tests that can satisfactorily characterise key HMA performance parameters and determine how these parameters will change throughout the life of a pavement. The key parameters to be considered are deformation resistance, fatigue life, stiffness, and moisture susceptibility [29]. However, the physical characterisations of the asphalt mix are described in this study. These descriptions are based on typical laboratory characterisation procedures. The effect of 3, 6, 9, and 12% ENR content (as a percentage of the base asphalt by weight) on the mechanical performance of the modified mixes was investigated and compared to that of HMA-0.
The design aggregate structure The nominal maximum particle size was chosen to be 19.0 mm. However, the six stockpiles of materials consisted of three coarse and three fine particle sizes. Aggregates were washed, oven dried and sieved to separate sizes before adding them to the mix to ensure that the gradation of the Superpave mix design was strictly followed. The following sieve sizes were used in the gradation: 19.00, 9.50, 4.75, 2.36, 0.30, and 0.075 mm sieves.
To begin with, the bulk specific gravity (G sb ) of the coarse and fine aggregates was calculated for all sieve sizes and the results are shown in Table 2. The apparent specific gravity (G sa ) and effective specific gravity (G se ) were also determined from these tests. Further analysis of the Superpave system consensus and source aggregate properties was carried out as this was critical to achieving high-performance HMA pavement. The results confirmed that the aggregates from the quarry complied with the standard specification criteria. Table 2 shows the results of the aggregate property tests. In general, the aggregate fulfilled the Superpave mix design requirements and was found suitable for use as a pavement material. Fig 1 shows the aggregate gradation used in this study. Aggregate gradation was prepared using 19.0 mm as the nominal maximum size in order to comply with the Superpave gradation limits.

Binder properties
Material characterisation based on standard and Superpave tests is presented in Table 3 [1,30]. These results complied with the specifications set by the standard criteria under binder PG 76.

Volumetric properties of mixes
Asphalt mixes were successfully developed in accordance with the procedures described in the Superpave mix design standard [22]. In this study, mixes with five different asphalt contents (4, 4.5, 5, 5.5, and 6%) were prepared (N = 3 for all the samples) for each base and modified HMA. Specimens were prepared by blending mineral aggregates in increments of 0.5% binder (i.e., 4.0, 4.5, 5.0, 5.5 and 6.0%) by weight of aggregate for the base and modified HMA.
The optimum binder content (OBC), which ensured acceptable volumetric properties as compared to the established mix criteria, was based on the Superpave gyratory compactor (SGC) specimens with 4% air voids. The volumetric properties consisted of OBC, effective binder content (P be ), voids in mineral aggregate (VMA), voids filled with asphalt (VFA), and the ratio of dust to the effective binder content (P0.075/P be ). The major factors that determine the stability and durability of Superpave HMA mixes are VMA, VFA, air voids, and dust proportion. Table 4 summarises the volumetric properties of the design mixes corresponding to

Resilient modulus test results
Resilient modulus is an important variable used to measure pavement response in terms of dynamic stresses corresponding to strains. However, this test can be used to evaluate low-temperature cracking, fatigue, and permanent deformation of asphalt mixes [31][32][33]. When elastic-layered system theory is used to design the asphalt mix, the modulus of asphalt is considered a basic design parameter [34].
During testing, horizontal deformation is measured from both sides of the specimen and the resilient modulus is calculated accordingly. In this test, the base and modified HMA at 5, 25, and 40˚C and at 0.33, 0.5, and 1 Hz were analysed, compared, and characterised. In general, it was observed that the resilient modulus for all mixes increased inversely with temperature, regardless of the availability of the modifier. In contrast, some studies have reported a decrease in the resilient modulus; however, numerous polymers were used as modifiers [24].
At low temperature (5˚C), the resilient modulus of the base HMA-0 mix was higher than that of the modified mixes. This indicates that the elasticity of the mix decreased after the addition of the polymer modifier, as shown in Fig 2. On the other hand, as the pulse repetition frequency during loading was increased from 0.333 Hz to 1 Hz, the resilient modulus values decreased. However, at 5˚C, HMA-0 showed the highest resilient modulus for low-frequency loading. At 25 and 40˚C, the resilient modulus values increased with increasing ENR content. This indicates that the elasticity increased after adding the polymer modifier, as shown in Figs 3 and 4.
Significantly, for all temperatures, the resilient modulus decreased with increasing loading frequency. Finally, these results indicated that HMA-0 was more susceptible to cracking at low temperature because of the increase in the resilient modulus and to fatigue cracking and permanent deformation at intermediate and high temperatures, respectively, because of the decrease in the resilient modulus. Tayfur et al. [24] modified various types of polymers and reported that the base asphalt had a higher resilient modulus as compared to PMA mixes at 5˚C, which meant that the base asphalt had a higher elasticity modulus (stiffness) and hence a lowest cracking resistance. Loading frequency increased the resilient modulus, especially at 25˚C and 40˚C. Finally, the results indicated that the resilient modulus values of the base mixes, especially at 5˚C, were higher than those of the PMA mixes, but at higher temperatures (25˚C, 40˚C), the values tended to decrease.

Dynamic creep test results
The dynamic creep test is used to determine the strength of HMA mixes to plastic deformation. Fig 5 shows the comparison between the accumulated strain of HMA-0 and the modified mixes at 40˚C and a stress of 100 kPa. The dynamic creep curve consists of three parts: primary, secondary, and tertiary. The accumulated strain was recorded at each load cycle and the test was terminated when the numbers of cycles reached 10,000. Because the loading period was terminated at 10,000 cycles, not all specimens failed before reaching the maximum Extreme permanent deformation was observed in the HMA-0 mix as compared to all modified mixes, as shown in Fig 5. This behaviour can be attributed to the influence of ENR on the binder and thus on the mix. The mechanical properties of all modified mixes rely on the properties of the binder at high temperatures, especially for permanent deformation. Therefore, it is expected that the resistance to permanent deformation would increase in the modified mix. Finally, adding ENR to asphalt mixes significantly decreased the asphalt's susceptibility to permanent deformation.
Similar polymer behaviour has been reported [35]; however, in that case, different percentages of styrene butadiene styrene (SBS) were used as the modifier. Khodaii and Mehrara found that adding SBS to the asphalt increased the asphalt mix's resistance to permanent deformation [35]. Fernando and Guirguis [36] indicated that the addition of 4% rubber decreases the creep compliance of asphalt mix by 42% at 45˚C. Eaton et al. [37] reported that the total creep was higher for the rubber mixes, pointing out the benefits of their performance at lower temperatures, that is, greater elasticity and better resistance to thermal cracking.

Rutting test results
The rutting test is a simulative test, also known as the torture test, conducted using a dry wheel-tracking device. A test temperature of 50˚C was chosen to simulate extreme environmental conditions for the HMA mixes. The wheel passed over the HMA specimens and was terminated after 45 min. In general, all modified mixes exhibited good rutting resistance as compared to HMA-0, as shown in Fig 6. Although the HMA-12 mix exhibited the lowest rutting value, HMA-0 exhibited the highest rutting value. Clearly, this indicates the high resistance of the modified mixes to rutting as compared to HMA-0.
The performance improvement of a base asphalt mix with addition of SBS has been reported [38], which was mainly attributed to the enhanced resistance to rutting imparted to the pavement by SBS. According to Shih et al. [39], addition of crumb rubber and SBR increased the rutting resistance of asphalt mixes. The results from laboratory study showed that the crumb rubber-modified and SBR-modified asphalt had higher stiffness at 60˚C than the base mixes.
Flexural fatigue test It is obvious that HMA-0 predominantly affected the initial stiffness of the mix as compared to the modified mixes, although the results from the fatigue life tests show a different aspect, i.e., HMA-0 failed after 18,030 cycles, whereas HMA-12 failed after 41,453 cycles. This indicates that the fatigue life improved after adding the modifier to the base asphalt mix. On the other hand, a decrease in the initial stiffness of the modified mix indicates a decrease in the elasticity of the mix. A previous study of an asphalt mix that used SBS as a modifier under a strain of 600 με found that SBS also increased the fatigue life of the asphalt mix [40]. Raad and Saboundjian [41] investigated the fatigue behaviour of the different mixes using controlledstrain third-point flexural beam tests. Results indicated that the crumb rubber modified asphalt mixes can enhance their fatigue resistance. The magnitude of improvement appears to depend on the degree and type of rubber modification. Lundy et al. [42] presented the three case histories in which ground rubber modification was used in the construction of asphalt mix. The results showed that the modified material has improved fatigue characteristics compared to the base mix as a result of the thicker asphalt films.

Moisture susceptibility
The indirect tensile strength (ITS) results of each mix under the dry and conditioned cases were averaged based on the results of three specimens [28]. All the ITS results are shown in Figs 9 and 10. All mixes tested were prepared such that their air void content was 7 ± 0.5%. The results showed that the ITS values of all the dry specimens decreased when they were conditioned, implying deterioration in the mixes, which affected the strength of the HMA mix.
The tensile strength ratio (TSR) indicates the susceptibility of the HMA mix to moisture damage. All mixes met the required minimum 80% TSR value as specified in AASHTO T283. The indirect tensile strength test showed that the modified mixes exhibited higher tensile strength as compared to HMA-0 before and after conditioning, which proved that HMA-0 was more susceptible to moisture damage. A previous study that used SBS (HMA-SBS)-modified asphalt also reported a reduction in the susceptibility of HMA to moisture damage [40]. It was found that the SBS-HMA asphalt samples achieved a higher TSR value (88.9%) than the base HMA samples (76.7%). On the other side, it was found that adding Elvaloy to the asphalt mix increased the susceptibility to moisture damage [43]. Coating smooth, rounded, siliceous gravel aggregates with cement plus SBR latex for use in asphalt mix increased the resistance to moisture damage [44].

Conclusions
The primary objective of this study was to investigate the performance characteristics of ENRmodified asphalt mixes through indirect tensile resilient modulus, dynamic creep, rutting, and fatigue tests and to investigate the durability of the ENR-modified asphalt mixes when exposed to moisture. The most important conclusions derived from this study are as follows: • In terms of the load-bearing capacity, adding ENR decreased the resilient modulus at low temperatures but increased it at intermediate and high temperatures.
• The permanent deformation of ENR-HMA at 40˚C and 50˚C was dominated by the ENR content, where ENR-HMA produced less deformation as compared to base-HMA. This is attributed to the presence of highly elastic rubber in HMAs at high temperature.
• The ENR modifier improved the HMA in terms of fatigue life, where ENR-HMA showed greater resistance to fatigue cracking as compared to base-HMA. However, the initial stiffness increased after adding ENR, indicating a decrease in the elasticity of ENR-HMA.
• Moisture damage was reduced after modifying the mixes with ENR, indicating that ENR-HMA was less susceptible to moisture damage as compared to base-HMA. Although this improvement was not significant, the results remain within the standard requirements.