In this study, two type of bituminous mixtures were employed as sub-ballast layer: A hot mix asphalt (HMA) manufactured at 160°C, and a warm mix asphalt (WMA) similar to traditional HMA in which a chemical additive has been used to reduce the manufacturing temperature to 25°C.Both asphalt mixtures studied as sub-ballast were a dense-graded mix type AC 22 S (UNE-EN 13108-1:2007). Limestone aggregates were used for the different fractions 0/6, 6/12, 12/18 and 18/25, presenting appropriate properties for their use in asphalt mixes in accordance with Spanish PG-3 regulations. The characteristics of the aggregates are shown in Table 1.

The filler used was Portland cement, CEM II/B-L 32.5 N (UNE-EN 197-1:2011). 95% of this filler has a particle size smaller than 0.063 mm with an apparent density (UNE-EN 1097-3:1999) of 0.7 Mg/m³. The same mineral skeleton (Figure 1) and bitumen type B 50/70 with a binder content of 4.25% and 54 dmm of penetration (UNE-EN 1426:2007), 54.4°C of softening point (UNE-EN 1427:2007), were used to manufacture the mixes.

The WMA was manufactured at 135°C and compacted at 125°C, through the use of a surfactant chemical additive that reduces the surface tension of asphalt binder and improves the moisture resistance of pavements by serving as an antistrip (21). The additive was introduced at 0.5% by weight (additive density: 1.01 gm/cc) and it had previously been mixed with the bitumen at a temperature of 160°C by using a conventional blender, set at a speed of 350 rpm for a period of 10 minutes, in accord with (22).

In order to compare the mechanical performance of the WMA with that presented by conventional HMA, a number of laboratory tests were carried out to characterize their behaviour and to evaluate their response under different failure modes that can take place during the real service life of these materials in railway tracks.

Table 2 summarizes the tests performed, the properties tested, and the main parameters investigated in this study.

Furthermore, the influence of temperature was evaluated, since the bituminous mixtures have high temperature susceptibility, becoming more viscous as their temperature decreases, and showing more rigid behaviour, all of which implies greater susceptibility to premature cracking induced by fatigue. Conversely, the materials become less viscous as their temperature increases, behaving more flexibly and showing higher permanent deformations under loading (23, 24). This evaluation was conducted to determine the possibility of using the sub-ballast under a range of different temperatures (both standard and adverse) that are expected in railway tracks.

Marshall test (EN 12697-34:06), water sensitivity test (UNE-EN 12697-12:2009) and indirect tensile stiffness modulus test (UNE-EN 12697-26:2012, Annex C) were developed to characterize both types of bituminous sub-ballast. The cyclic triaxial test (EN 12697-25, Method B), and the punching test were conducted to evaluate the resistance to plastic deformations of both types of sub-ballast. The tests were run both at temperature of service for this layer in railway tracks, and at higher temperatures, where plastic deformations are more likely to develop in asphalt mixtures. The plate bearing test (UNE-EN 103807:2008), was developed to measure the bearing capacity of the layer as well as its capacity to dissipate the stresses transmitted to the sub-grade, at both standard and high temperatures (25 and 40ºC) that can modify the endurance of mix asphalt. Finally, the permeameter test (UNE-EN 103403:1999) was conducted to assess its capacity to waterproof the remainder of the substructure, since permeability is scarcely affected by temperature.

The Marshall test was conducted to determine the percentage of voids in the mixes, along with the bulk density, deformation, and stability for the mixtures. The test was carried out on 3 cylindrical specimens (of approximately 60 mm in height and 101.6 mm in diameter, compacted with 75 blows on each side by a Marshall hammer) that were immersed in water at a temperature of 60°C for 50±5 min.

For the water sensitivity test, six cylindrical specimens of the Marshall type were manufactured, divided into two groups of three. This protocol involves conducting the indirect tensile strength test on a first set of specimens, treated with 67–72 h of immersion in a thermostatic bath at 40°C, and on a second set left at room temperature for the same length of time, at a test temperature of 15°C. The results are expressed in terms of the retained strength of the test cylinders after dividing the mean strength of the wet cylinders by the mean strength of the dry cylinders (ITSR, %).

To determine the stiffness modulus of the mixes, the indirect tensile stiffness modulus test (ITSM) was conducted in accord with UNE-EN 12697-26, Annex C. The test temperature was 25°C for two cylindrical specimens (for the two types of mixtures) compacted with 75 blows on each side by a Marshall hammer. During the test, 15 semi-sinusoidal test pulses were applied along the vertical diameter of the specimens with a total duration of 3±1 s, consisting of a rise time of 124±4 ms in a regime of deformation control (5 μm). The test procedure was repeated in a perpendicular diameter. The stiffness modulus of the mixtures was recorded by taking the average stiffness of the five pulses registered in the two diameters.

The first test for evaluating the resistance of the mixes to plastic deformations was the cyclic triaxial test (Figure 2a) (UNE-EN 12697-25:2006, Method B). This test requires two cylindrical test specimens manufactured by an impact compactor (75 blows per side) (UNE-EN 12697-30:2013) with a height of approximately 60 mm and a diameter of 101.6 mm. The specimens were tested at the standard expected temperature for sub-ballast in railway tracks (25°C) and extreme temperatures of 40°C and 60°C, which could occur in warm climates, and where plastic deformations are more likely to develop in asphalt mixtures. A conditioning period of 2 h was implemented before beginning the test. The experiment involved the application of a confinement load of 80 kPa and another cyclic sinusoidal out-of-phase axial loading of 200 kPa at a frequency of 3 Hz for 10,000 load cycles (25). At the end of the test, measures were recorded (average of two specimens) for creep modulus, creep slope at the final 5,000 loading cycles, and permanent deformation.

In addition, resistance to permanent deformations was evaluated by a punching test (Figure 2b), which is able to simulate the contact between the ballast and the asphalt layer (Figure 2c). The test was conducted at 25°C and 40°C and required a pair of test specimens of 300 mm x 300 mm x 120 mm, for each temperature and for both types of mix (HMA and WMA). The test consisted of applying a stress of 200 kPa for 50,000 load cycles (regarded as an unfavourable loading condition that can occur over the sub-ballast layer) (10) by means of a ballast plate with dimensions of 150 mm x 150 mm (Figure 2d) on the top of the sub-ballast specimen. The calculated parameter represents the evolution of the permanent deformation caused by the punching of ballast particles.

To evaluate the bearing capacity of the mixtures, as well as their capacity to dissipate the stress transmitted to sub-grade, a plate bearing test (Figure 3) (UNE-EN 103807:2008) was carried out. This test involved the manufacture of a prismatic test slab of 408 mm x 512 mm for both types of mixture. A roller compactor with a smooth steel roller of 120 mm in thickness was employed for compaction. This test was conducted at 25°C and 40°C, expected and high temperatures to modify the endurance of bituminous sub-ballast (26), under static and dynamic loads. For the static test, the load was applied at various stress levels: 20, 50, 150 and 250 kPa, across two charge cycles, while the dynamic test was carried out by applying a cyclic sinusoidal load of 200 kPa at a frequency of 4 Hz for 50000 load cycles. Moreover, to analyse the capacity to distribute loads, a pressure cell was collocated at the bottom of each mixture.

To evaluate the permeability of the two types of sub ballast (WMA-HMA), a permeameter (Figure 4) of square section (a surface of 300 mm x 300 mm, and 500 mm in height with a hole in the bottom to collect the water drained by the sub-ballast through a pipe of the same diameter) was used. The constant-head method was applied in accordance with the Standard UNE-EN 103403:1999 that is usually used to determine the permeability of soils by Darcy’s law. The parameter calculated the evolution of permeability in each type of sub-ballast, in accord with the methodology applied by other authors (7, 27) for the study of permeability of bituminous layers in railways.

The Marshall Test results indicate that the conventional hot mix asphalt and the warm mix have very similar values of density and air voids. This implies that use of the additive could improve the workability of the WMA, despite a 25ºC reduction in the manufacturing temperature (13–16). It is also clear that in spite of the values of Marshall Flow being very similar for both mixtures (3.4 mm), there was a decrease of Marshall Stability in the warm mix sub-ballast.

The same occurred during the Water sensitivity test, where indirect tensile strength of WMA decreased with respect to that recorded for the conventional HMA. However, there was an improvement in the indirect tensile strength ratio (ITSR %). This is probably due to the lower compaction temperature that could increase the potential for moisture damage as established in the studies carried out by Hurley (14, 28), since lower mixing and compaction temperatures can result in incomplete drying of the aggregate.

Stiffness (stress/strain ratio) generally decreases as the manufacturing temperature decreases (29, 30). The present results revealed a slight decrease of the modulus in the warm mix sub-ballast due to the decrease in manufacturing temperature that reduces the aging of the mixtures, along with its stiffening, which is in accord with other studies (29). Table 4 lists the main properties of these materials.

The results of the triaxial test (Figure 5) show that the values of permanent deformation, slope of deformation, and creep modulus recorded for the warm mix sub-ballast are very similar to those presented by hot mix sub-ballast at both 25 and 40°C. This fact indicates the possibility that warm mix could be used as a sub-ballast without increasing plastic deformations, since it has been shown that under the standard temperatures of service expected for sub-ballast in railway tracks (around 25°C), conventional sub-ballast shows higher resistance to plastic deformations than that of the conventional granular sub-ballast (7, 25). However, increasing the temperature to 60°C (as expected in extreme climatic conditions) led to a decrease in resistance to permanent deformation for the WMA, with values of final deformation and creep modulus being lower than those obtained for the HMA. This implies that WMA could be more vulnerable to deformations than the granular layer (7). However, it should be noted that it is rare to achieve such extreme temperatures in real-life service, since the ballast layer usually protects the sub-ballast from high temperatures.

The punching test was carried out to evaluate the resistance to plastic deformations under more realistic load conditions. Starting from the deformation curves obtained during the test at 25 and 40°C for both bituminous sub-ballast (HMA and WMA), it was possible to determine the initial and final deformation and the slope of the creep curve of the stable phase (7). The initial deformation (primarily associated with the recompaction and reorganization of particles) was determined at around 700 cycles, the point at which the deformation response tends to stabilize. This point until the end of the test (50000 cycles), may be regarded as the stable phase to determine the slope of the curve. In addition, the final deformation was determined at 50000 cycles.

The results (Figure 6) show that the warm mix sub-ballast presented values of initial and final permanent deformation equal to or very similar to those obtained for the hot mix sub-ballast at both temperatures. The slope of the creep curve increased slightly in the case of WMA, however in the order of very similar values compared to those recorded for the HMA. Thus, under routine in-service temperatures (25°C) and at higher temperatures (particularly 40°C), the effects due to ballast punching were very similar for both mixtures in both the short and long-term. Further, and in accord with other studies, the results obtained are lower in comparison with the traditional granular sub-ballast (7). This indicates the possibility of using warm mix asphalt as sub-ballast in railway tracks without increasing the appearance of plastic deformations.

Regarding the bearing capacity of both sub-ballast types, Figure 7 displays the static and dynanmic modulus obtained during the Plate bearing test at 200 KPa, which was designed to reproduce unfavorable loading conditions expected in railway tracks for this layer. It is clear that both bituminous sub-ballast recorded similar values of resistance to loads transmitted by trains, even at the higher temperature (40°C) in both static and dynamic conditions. According to earlier studies, the application of hot mix asphalt as sub-ballast was able to improve the bearing of the layer compared to that obtained by the more traditional granular configuration (7, 25). For this reason, the present results indicate that WMA could be appropriate for application in railway tracks, since it has presented a similar performance to that measured for conventional bituminous sub-ballast.

The results regarding bearing capacity were in line with values of pressure recorded under the sub-ballast layer, by using pressure cells under static conditions (Figure 8) and dynamic conditions (Figure 9), for temperatures of 25 and 40ºC. It is clear that the pressure reaching down through the sub-ballast layer is quite similar for the two mixtures (HMA and WMA). This means that the decrease in the manufacturing temperature of the mixture has little effect on its capacity to dissipate the loads transmitted by the passage of trains. It can also be observed that an increase of the test temperature generally decreases the capacity of the mixtures to dissipate the stress. Despite this decrease, the values recorded for the warm mix sub-ballast are very similar to those recorded for the conventional bituminous sub-ballast. Furthermore, it is worth noting that the values of pressure registered are significantly lower (by up to 50%) than those measured under the layer of the granular sub-ballast (7, 25, 31). This implies that, in comparison with granular sub-ballast, bituminous sub-ballast is able to provide a more durable system, reducing maintenance and rehabilitation costs regardless of the temperature of manufacture.

As mentioned previously, in addition to bearing capacity, another fundamental parameter for this layer is its capacity to waterproof the remainder of the substructure. This parameter is specified by the Spanish standard, which indicates that the vertical permeability coefficient of the sub-ballast (K) must be less than 10–6 m/s (32). Figure 10 shows the evolution of the drained flow for both types of sub-ballast used in this study (with a thickness of 12cm), analyzed using a constant-head permeameter. The HMA presents slightly lower values of flow compared to the those obtained by WMA. However, both mixtures have presented an appropriate coefficient of permeability for application in railway tracks according to the Spanish Standard for sub-ballast. The present results suggest that both types of sub-ballast have the capacity to prevent the possible contamination of the sub-structure by the vertical hydraulic transport of mud and fines, offering an impermeable uniform drainage layer that reduces the effects of freeze/thaw action and leads to a lower variation in the moisture content with respect to the granular configuration (7, 9, 33).