Testing and interpreting permeability of asphalt mixes

This paper presents a field study showing the distribution of air voids in asphalt layers, and the potential effect of water on pavement performance. The paper also discusses the applicability of the European test method and equipment for assessing permeability of asphalt layers. A series of laboratory permeability tests were conducted on in-situ core samples and laboratory samples. It was found that mixes with smaller particle sizes, the air void content results in lower permeability compared to mixes with larger particle sizes.


Introduction
The presence of water in road pavement layers is evident for unbound granular layers such as subgrades and subbase layers; however, its presence in and between bound layers, especially asphalt layers, is still considered a novelty to the wider professional community, despite numerous publications have been available in this field since the 1960s [1]. Several cases of water-induced damage to asphalt layers have been published in international and national literature [2,3].
Since the 1970s, a number of case studies have also depicted clear relations between the compaction rate and interlayer bonding of asphalt layers and the presence and movement of water within and between asphalt layers, respectively [4][5][6][7]. Despite such findings, no specifications have been made in a number of countries, including Hungary, about permeability of asphalt layers. Such specifications normally require only testing of the air voids content of the placed layer(s). The authors have encountered several cases of water permeability issues within asphalt layers and asphalt pavements when conducting research and consulting works at the Asphalt and Railway Laboratory of the Budapest University of Technology and Economics. The asset owner of a major Hungarian motorway contacted the Laboratory to analyse possible causes of pavement failure, involving symptoms related to water movement within asphalt layer(s). The issue was investigated based on international literature, and a permeability testing device, conforming to the relevant European Standard [8], was acquired. Main findings relying on these simple test methods and computed tomography (CT) are presented in the paper, and attempts are made to establish further research goals for additional assessment of such problems.

Test plan
Ground-penetrating radar (GPR) measurements were carried out on a total of five selected sections of the motorway. According to the test plan, in-situ core samples were extracted from all traffic lanes and the emergency lane (two samples from each lane), equalling to a total of 30 samples. Almost all samples showed the presence of water at first sight as parts of the samples were wet after several hours or being on the surface, showing a relatively higher void content -filled with water from the wet cutting process as shown in Figure 1 as an example. According to international literature, permeability of asphalt pavements is primarily related to air void content (rate of compaction) in the placed layers; accordingly, the test plan consisted of permeability measurements and determination of air void content. For the first time in Hungary, computed tomography (CT) was also applied as an experimental technique.

Asphalt permeability tests
Several test methods for determining permeability of laboratory prepared asphalt samples and placed layers are currently available. The results of field tests are convincing as, in these cases, the permeability of pavement cracks is also taken into account, in addition to material permeability of pavement layers. Moreover, in this case, the permeability of the whole pavement is tested i.e. the water capable of reaching the subgrade is inferred [9][10][11][12]. Laboratory tests according to the European standard [8] consist of a water column of constant height in a cone making the permeation of water through the sample possible. Figure  2 shows the test setup for two standard methods, the vertical permeability, where the water permeates into and out of the sample on the horizontal plane, and the horizontal permeability measurement, where the water permeates into the sample on the horizontal plane, but permeates out at the vertical cylinder. In both cases, the water column is kept at constant height resulting in constant water pressure, and the mass of water exfiltrating the sample is measured over time. Permeability is given as calculated in eqn. (1) and (2). (1) (2)  The air void content of the samples was determined using the standard test method EN 12697-8. The test method is assumed to be commonly known and will not be discussed here.

Computed tomography
Computed tomography (CT) can be used to analyse composition of a given material, via assessment of frequently made sectional images, leading to a whole new level of material analysis [13]. These images are created using X-ray measurements from different angles, detecting the rays through the samples and processing the data with special software (see Figure 3). There can be several cases for using CT in the research of asphalt mixes. Most of the research at hand focuses on methods for evaluating microstructure of asphalt specimens via various procedures [13][14][15][16].
There are also papers that investigate the asphalt microstructure obtained via CT and the estimated mechanical or structural performance [17,18] to complete the well established fundamental description of mix behaviour [19,20].
Using this technology, the proportions of aggregate, binder, filler, and air voids within a given sample can be estimated with some precision, which is mainly dependant on the CT device and the algorithms used for data assessment. CT tests were conducted on the samples with the focus on the distribution of the air voids with depth. The results are shown in Figure 4, where the distance between the slices is 1,5 mm. The interlayers are also shown in the figure.

Test results
As shown in a number of previous research papers, there is a clear correlation between the air void content and permeability. However, the permeability of asphalt samples is not only affected by the air void content, maximum particle size, sample dimensions, and measurement method, but also by hydraulic characteristics of water [21,22]. These findings are quite similar to the ones presented in the research on permeability of cement concrete [23,24]. There are also research findings suggesting correlation between the normalised air void content and permeability. Permeability of different magnitudes can be assessed according to  Table 2 are assumed to be impermeable, [26]. The construction tolerance is +3 %, although a given layer may be accepted within an additional +2 % air void content at a discounted price. A given pavement layer is rejected at the air void content given in the last row. As previous research indicates, the correlation between the air void content and permeability shows a third-degree polynomial relation in the form of K = A•exp(B•V). Parameters A and B were determined using Solver, for air void-permeability of samples, for each particle size separately. The coefficient of determination was found to be high for all individual cases, and ranges from 0.85 to as much as 0.98, as shown in Figures 5,  6, and 7. Parameters and regression coefficients are shown in Table 3. A strong correlation was established between the permeability and air void content and, at that, the aggregate particle size was also taken into consideration. The results are based on 11, 9, and 24 test results, respectively, as some samples were damaged during processing or coring.    As can be seen, low-to-average correlation values were obtained for linear regression with an intersection at 0,0. Trends show that in cases of layer 3 (were the maximum aggregate size is 22mm) the values are close to the 1:1 ratio, with a relatively high regression coefficient, at both average and median values, which may suggest that one or both methods are more accurate at larger aggregate sizes. This conclusion is also supported by the trends of layer 1 (maximum aggregate size of 11mm), having the lowest nominal aggregate size and the lowest level of correlation between the air void contents. A distinction between aggregate sizes of the samples should be made to determine the possibility of conversion between the methods [27].

Conclusions
An investigation was carried out on core samples extracted from a major Hungarian motorway, and it revealed failures due to the presence of water in asphalt pavement. Where the performance of the pavement structure was compromised and water detected by the GPR, cores were extracted and tested using the European test method for permeability. Core sample test results show a clear correlation between the air void content and the horizontal and vertical permeability of the samples; the findings are in line with relevant international experience.
The results showed that horizontal permeability is always higher than vertical permeability, which shows a difference in test method, and indicates that the two methods should be distinguished from one another. It was found that in case of mixes with smaller particle sizes, the air void content results in lower permeability compared to mixes with bigger particle sizes. This also indicates that results should be analysed considering not only the air void content, but also the particle size distribution. Compaction technology obviously has a high impact on the permeability, but it is not the only parameter [28,29].
With regard to the air void content derived from CT measurements and obtained using the EN 12697-8 standardised method, it can be concluded that there is a need for further research, in which not only a logical correlation but also a mathematical conversion should be established. In the latter case the nominal aggregate size must be taken into account [30,31]. Based on the air void content versus permeability, asphalt layers meeting air voids limits according to the specifications may be in fact permeable, which may lead to a variety of pavement failure types and deterioration of lower pavement layers. Based on the results presented in this paper, a well-established relation may be determined between permeability and air void content. This could result in incorporation of layer waterproofness in the specifications, which is presently poorly addressed. The introduction of permeability limits in relevant specifications would require establishing precision of the test method, which is currently unknown.