Analysis of moisture susceptibility of hot mix asphalt (HMA)

Indonesia has been extensively expanding its road networks to support the economic and the population growth. It is very important to ensure that the road networks perform well and last to their designed life. However, it is very common to find roads with asphaltic surfaces with some pavement failures not long after the construction finished. The most commonly pavement failures found are potholes and stripping, which are related to water intrusion to the hot mix asphalt. Indonesia is a tropical country and its soil has a high groundwater level in general. Therefore, it is very important to study the effect of moisture in asphaltic surfaces. This research aims to provide a preliminary analysis on the moisture susceptibility of hot mix asphalt. It was found that there were a few of Marshall test results that did not satisfy the requirements, which could be caused by human variation and the angularity of aggregates. The Cantabro Loss test results show that the HMA specimens lost around 11% and 15% of material after being exposed to water and abraded. From the research results, it was found that there is a strong relationship between the flow and the Cantabro Loss of the prepared HMA specimens.


Introduction
Indonesia is one of the countries with a high population density and a high population growth rate. Therefore, the infrastructure, especially the road and highway networks, must be continuously improved and maintained. It is crucial to correctly design the road pavement from the beginning to ensure that the asset lasts. However, it is very common to find pavement failures not long after the roads were constructed, especially after the monsoon season [1].
The presence of moisture or water in hot mix asphalt pavement could lead to the loss of strength and durability of the pavement because the adhesion bond between the asphalt and aggregate is broken and the cohesive bond in the bitumen (asphalt) is lost [2]. The most commonly pavement failures in flexible pavement (with asphaltic road surface) due to the presence of water are stripping, potholes, and rutting [3]. There are a number of factors that affect the susceptibility of hot mix asphalt (HMA) surfaces, such as the aggregates properties (minerals, source of aggregate, angularity, dust, and moisture content), the asphalt binder properties (stiffness, chemical composition, and refining process) [4], [5].
Usually the moisture damage starts at the bottom of the asphaltic surface or at the interface between asphalt layers, continued to become localized potholes, and then, cracking developed [6], [7]. The combination between water intrusion to this localized pavement failure and traffic loading will lead to further degradation of the pavement layer [7]. Ravelling or loss of aggregate can occur, especially in chip seal surfaces and binder from pavement could be flushed or turned into surface bleeding [4].
There are a number of research projects worldwide hat have studied the effect of moisture in hot mix asphalt (HMA) [8], [9]. Indonesia is a tropical country and most places have a high groundwater level, and hence, it is very important to study the effect of moisture in the HMA layers to ensure that the pavement layers could serve their functions throughout their service age. This research aims to provide a preliminary analysis on the moisture susceptibility of HMA as designed according to Indonesian standard.

Materials
HMA consists of aggregates, binder (asphalt), and filler (optional). In this research, a natural aggregate was used, as shown in Figure 1. These aggregates were crushed into four sizes, as shown in Table 1. The coarsest aggregates are denoted as Aggregate I and the finest aggregates are denoted as Aggregate IV. A Laser Induced Breakdown Spectroscopy was performed on this aggregate to determine the dominant chemical elements of this aggregate.  There were two types of asphalt used herein. They were both 60/70 asphalt (as stated in the product specifications) but sourced from different companies. There was no filler used in the HMA used for this research. The asphalt samples were then constructed by using the asphalt mix design used in [10] with 5.5% asphalt content, and the details are shown in Table 2. There were four samples constructed for each variation. Two samples were used for the Marshall tests and the other two samples were used for the Cantabro Loss tests.

Aggregate and Asphalt Preliminary Check
To ensure the suitability of the materials to be used in HMA design, there were several standard tests for coarse and fine aggregates and asphalt, as listed in Table 3, before the HMA samples were prepared.

Marshall Tests
The Marshall tests were conducted for the prepared asphalt samples according to Standar Nasional Indonesia (SNI) 06-2489-1991 [11] to determine the characteristics of the asphalt mixes. From the test results, the stability and the flow parameters of the samples could be determined. There were also other parameters that were calculated, such as Void in Mix (VIM), Void in Mineral Aggregate (VMA), Void Filled with Asphalt (VFA), and density. There were two samples constructed for each HMA variation.

Cantabro Loss Tests
The Cantabro Loss test is a test used to determine the loss of abrasion of compacted HMA samples as stated in [12]. This test involves measuring the breakdown of the prepared HMA specimens with Los Angeles Abrasion machine. The Cantabro Loss was calculated by finding the difference between initial and final weight of the HMA specimens after the specimen being rotated in Los Angeles machine for 300 revolutions at a speed of 30 revolutions per minute (rpm). The result was presented in percentage and it shows the durability and the quality of the asphalt binder.
For this research, due to the equipment limitation, there was procedure for this test was slightly modified. The HMA specimens were prepared with a diameter of 10.2 cm and a height of 6.35 cm. This is smaller than the specimen size specified in [12], but it does not hinder in achieving the goal of this test as this test is looking at the percentage lost after the specimens being abraded. Moreover, the prepared specimen was immersed in water bath at 60 o C for one hour to simulate the HMA surfaces being flooded, and hence, the ability of the specimen to resist disintegration action after contact with water can be assessed. This procedure is similar to the research project described in [13], although they immersed the specimen for 24 hours in the water at 60 o C. Table 4 shows the results of the tests conducted on the coarse and fine aggregates. It was found that the natural aggregate used in this research satisfy the requirements. Moreover, the mineral analysis was also performed on the aggregate and it was found that the aggregate is dominated by Calcite, Manganese, Sodium, and Aluminium, as seen in Figure 2.   Table 5 shows the results of the tests conducted on the asphalt. It can be seen that the binder passed all the requirements, except for the flash and fire point tests for Asphalt 1. In general, Asphalt 1 seemed to be more rigid than the Asphalt 2 as it has lower ductility value and penetration test. Both asphalts were used as the binder for the HMA designed in this research as both are commercially available and have been used in a number of road construction projects.  Table 6 shows the parameters of the HMA samples tested, which were obtained from the Marshall test and further calculations. The results for every samples and the average values of the two samples were shown. It can be seen that the HMA designed satisfy most of the requirements and the values that do not satisfy the requirement were given an asterisk mark (*) next to them. The flow of HMA samples made by using the Asphalt 2 exceeded the requirement, which suggests that the HMA samples were too soft. Moreover, both the VIM and density values of both samples did not fall between the allowable values. These values show that the samples were not compact enough. This could be due to the HMA sample compaction method, which was done manually, and hence,   Table 7 shows the results of Cantabro Loss tests, which were conducted to see the effect of water on the ability of the prepared HMA specimens to resist abrasion. It can be seen from Table 7 that the HMA samples prepared with Asphalt 1 has lower percentage of material lost than the HMA samples prepared with Asphalt 2. It seems that the HMA samples prepared with Asphalt 2 are slightly more sensitive to abrasion than the HMA samples prepared with Asphalt 1. However, it is important to note that there were only two samples prepared and there were a number of factors that could cause variation on the results. For Sample 1 prepared with Asphalt 2, the percentage of material loss was higher than its pair. This could be due to many factors, such as uneven compaction force and the angularity of coarse aggregates. If the data for Sample 1 prepared with Asphalt 2 is removed, it can be seen that the percentages of Cantabro Loss between Asphalt 1 and Asphalt 2 are similar, which is approximately 11%. These results suggest that there might not be any relationship between the asphalt and the resistance to abrasion of the prepared samples  Figure 3 shows the relationship between the Cantabro Loss and stability, flow, VIM, VMA, VFA,and density as listed in Table 6 and Table 7. The dashed black lines show the allowable limit for each parameter. It was found that there is a strong relationship between the Cantabro Loss and the flow of the prepared HMA samples with coefficient of determination (R 2 ) of 0.92, while there is almost no relationship between the Cantabro Loss and the other parameters.