- Email: [email protected]
Microstructural investigation on air void properties of porous asphalt using virtual cut section
Construction and Building Materials 155 (2017) 485–494
Contents lists available at ScienceDirect
Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat
Microstructural investigation on air void properties of porous asphalt using virtual cut section Mohd Zul Hanif Mahmud, Norhidayah Abdul Hassan ⇑, Mohd Rosli Hainin, Che Ros Ismail Faculty of Civil Engineering, Universiti Teknologi Malaysia, 81310 UTM Skudai, Johor, Malaysia
h i g h l i g h t s A research framework for virtual cut section analysis on porous asphalt mixture. Polar diagram is used to complement the conventional cross-section analysis. Air voids shape in porous asphalt are predominantly elongated. Recommend Fmax parameter in determining air void size of porous asphalt.
a r t i c l e
i n f o
Article history: Received 20 June 2017 Received in revised form 16 August 2017 Accepted 17 August 2017
Keywords: Porous asphalt mixtures Virtual cut section Air voids distribution Air voids properties Image analysis technique X-ray CT
a b s t r a c t The microstructure properties of asphalt mixtures reflect the structural and functional performance of asphalt mixtures. Thus, numerous imaging methods are developed including destructive and nondestructive methods to uncover the internal structure properties of asphalt mixtures. However, accurate analysis of internal microstructure properties is challenging. This study used a non-destructive method employing virtual cut section (imaging technique) to analyze the air void properties of laboratoryfabricated porous asphalt. Conventionally, the investigation of internal structure properties particularly on vertical section requires the sample to be cut (destructive method) into several sections. This method causes defects to the sample and limits the amount of data that can be extracted from the image. The imaging technique designed in this study provides comprehensive understanding of the distribution and properties of air voids from different angles within the compacted sample of porous asphalt. The void properties in porous asphalt should be assessed accurately to ensure effectiveness of water permutation through the porous structure and acoustic function of the mixture. The compacted samples were X-rayed and virtually analyzed into various cut sections. The void properties were characterized in terms of the content, number, shape, and size of voids. Results indicate that air voids within the porous asphalt are homogeneously distributed with an elongated shape; this characteristic indicates high void connectivity. The vertical cross-section analysis justifies the characteristics of air voids in the conventional analysis of horizontal cross section. Ó 2017 Published by Elsevier Ltd.
1. Introduction Porous asphalt is typically used as a wearing course that allows water to permeate into the porous surface. The process is possible because porous asphalt is designed on the basis of open graded gradation [1–3]. Open graded gradation consists of a large quantity of coarse aggregates and a small quantity of fine aggregates [4–6]. This method allows for the mobility of aggregates during compaction, thereby resulting in high formation of interconnected voids within the layer. Air void content (interconnected voids) in ⇑ Corresponding author. E-mail address: [email protected] (N.A. Hassan). http://dx.doi.org/10.1016/j.conbuildmat.2017.08.103 0950-0618/Ó 2017 Published by Elsevier Ltd.
porous asphalt mixture is the main parameter that influences the structural and functional performance as well as service life [7–10]. The common parameters used to describe the connectivity of air voids are the content, number, shape, and size of voids [1,11–15]. This pavement type needs to maintain sufficient amount of air void content to ensure effective permutation of the pavement and prolong the benefits of the structure to the road users. Therefore, a detailed analysis on the air void formation within the porous asphalt is significant to enhance understanding of its functional performance. The content or properties of air voids can be quantified using various methods, including vacuum method, dimensional analysis, water displacement concept, and imaging technique [16–23].
486
M.Z.H. Mahmud et al. / Construction and Building Materials 155 (2017) 485–494
Numerous efforts have been made in the recent years to investigate the microstructure properties (particularly air voids) of asphalt mixture using imaging techniques [9,24–27]. Using these techniques, air voids can be characterized either in a destructive or a non-destructive manner. Current non-destructive imaging technique, i.e., X-ray computed tomography (CT) scanning, is preferred as the sample can be used for subsequent testing after the scanning process. Another option to investigate air void properties is through destructive methods. A destructive method for imaging the internal structure properties of asphalt mixture was developed in early 1990s [28]. Although destructive imaging techniques are cost effective and do not require sophisticated equipment (digital camera) to obtain the cross-sectional view of the sample, the process of cutting requires the use of a special cutter, thereby limiting the number of permissible cross sections or image slices to be obtained. Furthermore, these techniques cause misleading analysis on the properties and distributions of air voids. The cutting process triggers air void distortion in addition to causing severe aggregate loss especially at the edge of the slices. As porous asphalt is designed to possess low structural strength and limited amount of fine aggregates, the cutting process will result in deformed samples and excessive aggregate loss. Thus, void properties captured using destructive methods in terms of quantity, size, and shape may differ from those obtained using non-destructive methods. Non-destructive techniques allow the internal structures of the asphalt samples to be analyzed without cutting or disrupting the sample [5,29,30]. Such techniques provide detail overview of the internal structure as the images can be captured throughout the samples and analyzed from various angles [13,31,32]. However, considering only the horizontal cross section of the X-ray CT images for the analysis of air void distribution cannot fully describe the exact void connectivity as the water flows from the top to the bottom section of the porous asphalt. Voids within the porous asphalt layer act as a series of micropipes that connect to one another to channel the water away from the surface [5,14]. Therefore, the horizontal cut section of the captured images will only provide the transverse section of the pipes. Compared with the vertical cross section, the horizontal one will provide the longitudinal section of the pipes to reflect the void connectivity. Accordingly, better views of the hydraulic conductivity from top to bottom within the porous asphalt layer can be provided. In this study, air void properties of porous asphalt were nondestructively characterized using the virtual cut section (vertical cross-sectional image) of X-ray CT images. A complete procedure of the image analysis technique was introduced including image acquisition, image processing, and image analysis. Although this study applies the procedure on porous asphalt, the concept is relevant to any potential materials. The void analysis covers a few basic parameters (i.e., content, number, size, and shape) measured from the images in horizontal and vertical cut sections and relates to the air void connectivity.
graded gradation enveloped with the nominal maximum aggregate size of 13.2 mm as shown in Fig. 1 [37]. Polymer modified bitumen (PG 76) was added to the sample as a binder at designed bitumen content of 5.25% (obtained from mixture design). The samples were mixed at 195 °C first and then compacted at 175 °C. The properties of the materials used are provided in Table 1.
2.2. Microstructure investigation This section emphasizes on the characterization of air void properties of porous asphalt. Air voids play an important role in influencing the permeability, acoustic characteristics, and frictional properties of porous asphalt. Thus, accurate data of air void properties should be obtained because they reflect the performance of porous asphalt. The porous asphalt samples were scanned using X-ray CT scanner to capture the horizontal images. The horizontal images were then processed and analyzed using the ImageJ software. Fig. 2 shows the detailed framework of the study. The framework was divided into two main stages. Stage I focuses on image processing wherein several imaging procedures were performed on the images, such as cropping, image scaling, and quality enhancement. Stage II involves the development of three-dimensional (3D) images from horizontal images using Volume Viewer. The vertical cut sections were obtained from the 3D images and then stacked for subsequent analysis. The air voids were then thresholded, analyzed, and verified to the experimental value. Details on the image acquisition and processing methods are provided in the following sections.
2.2.1. X-ray CT scan and image acquisitions This study used a non-destructive image acquisition technique to capture the cross-sectional view of porous asphalt. The samples were scanned using X-ray CT scanner (inspeXio smx-225 CT) with the maximum output voltage of 225 kV. This scanner used the cone beam method to reconstruct 3D images as shown in Fig. 3a. During the scanning process, the sample was rotated 360° at an image slice of 0.1 mm thickness. The scanning parameter of the X-ray CT scanner was set to capture 1200 views with the average count of 15 times at each angle of rotation. By increasing the number of views and average count, the data obtained from the scanning procedure would be accurate (detailed information) but would increase the scanning duration. These image slices were combined into a stack at the interval of approximately 0.1 mm as illustrated in Fig. 3b. The output voltage used to scan the sample was selected to be 190 kV with the ampere of 100 mA for good-quality images. The resolution obtained from the X-ray CT scan was 0.106 mm/pixel. These images were saved in Tagged Image File Format (TIFF) as 16 bits. TIFF digital file format was selected to avoid data compression and degraded image quality during image conversion and processing.
2.2.2. Image enhancement Considerable imaging software is available for image processing and image analysis. In this study, the images were processed using ImageJ. ImageJ is an open source imaging software initially designed for medical purposes. TIFF images were converted into 8 bits for further analysis. Notably, 8-bit image comprises of 256 gray levels of the dark regions (low-density material, i.e., air voids) and the brightest regions (high-density material, i.e., aggregate). Images obtained from the X-ray scanning were analyzed as stack formation. A few image slices at the top and bottom sections were excluded from the analysis due to image distortion during the scanning process. These images were cropped in accordance with the exact dimension of the sample (100 mm diameter and 50 mm thickness). Images obtained from X-ray CT scanning are often not in perfect conditions and thus require image enhancement and correction owing to excessive noise, poor contrast, imbalance color representation, and disproportional illumination [13,38,39]. Thus, image treatments using several mathematical functions, such as contrast enhancement and noise elimination (Gaussian filter or Median filter), were conducted to improve the appearance of the image as shown in Fig. 4. This procedure was conducted to visualize the region of interest for accurate image analysis.
2. Material and methods
Laboratory-fabricated samples of porous asphalt mixture were prepared for Xray CT scanning. The samples were prepared in accordance to Malaysian road specifications [33]. The samples were compacted using Superpave gyratory compactor with 50 gyrations in accordance with ASTM D 7064 specification [34]. Superpave gyratory compactor is able to simulate the actual field conditions [35]. Furthermore, the Superpave gyratory compactor method can produce lower coarse aggregate breakdown [36]. Each of these samples was compacted at the vertical pressure of 600 kPa and gyration rate of 30 rpm to achieve the desired air void content of approximately 20%. Due to the limitations of the X-ray scanner (inspeXio smx225 CT) in capturing the microstructure properties of the asphalt mixture at large scale, the samples were prepared in a 100 mm diameter mold with thickness of 50 mm. The granite type aggregates were sieved to obtain an Australian open
Percentage Passing. (%)
2.1. Materials and sample preparation
100
Upper Limit
Lower Limit
Middle
80 60 40 20 0 0.01
0.1
1
10
100
Sieve Size, (mm) Fig. 1. Australian open graded gradation for porous asphalt mixture, (AAPA, 2004).
M.Z.H. Mahmud et al. / Construction and Building Materials 155 (2017) 485–494 Table 1 Materials Properties. Materials
Physical Properties
Coarse Aggregate
Bulk Specific Gravity Water Absorption Aggregate Impact Value
2.601 0.9% 26.0%
Fine Aggregate
Bulk Specific Gravity Water Absorption
2.427 2.0%
PG 76 bitumen
Specific gravity at 25 °C Penetration test at 25 °C Softening point test Viscosity at 135 °C Viscosity at 165 °C G*/Sin d, kPa at 76 °C
1.030 44 PEN 75 °C 1.04 Pa.s 0.4 Pa.s 1.6 kPa
(2). For circularity, the value of 1 indicates a perfect circle and the value approaching 0 means otherwise. Meanwhile, high value of aspect ratio indicates the degree of elongation in the major axis is higher than in the minor axis. Most of the void shape in porous asphalt mixtures is elongated, indicating high air void connectivity. The size of the voids can be calculated using an equivalent projection area of a circle, DEQPC, as given in Eq. (3) [38]. DEQPC is described as the diameter of a circle that has an identical area as the selected air void regions. The size of air voids can be described using Feret’s dimensions, i.e., Maximum Feret’s diameter, Fmax, and minimum Feret, Fmin. Fmax and Fmin are given by the maximum and minimum distances between two pixels at the air void boundary as illustrated in Fig. 8. Other parameters such as number and percentage of air voids can be obtained from the ImageJ software.
Circularity ¼ 4
Aspect Ratio ¼ 2.3. Virtual cut section method The virtual cut sections were conducted in vertical and horizontal axes using the ImageJ plugin called Volume Viewer. Volume Viewer is a custom visual algorithm that offers excellent solution in interacting with image volume, meshes, and volume rendering in 3D visualization [40]. In this study, 18 vertical crosssectional images were obtained with an increment of 10° orientation angle as illustrated in Fig. 5. These cross-sectional views were obtained to evaluate the consistency of air void distribution within the sample. Fig. 6a shows the 3D reconstructed vertical cut section made by stacking the horizontal images. The dimension was 50 mm 100 mm as shown in Fig. 6b. The air voids were then thresholded and segmented for further analysis as shown in Fig. 7. Fig. 7a shows an X-ray image of vertical section obtained from the stack of images. The air voids were thresholded and verified in accordance with ASTM D 3203 [41] as shown in Fig. 7b. The image was then converted into binary images (black and white images) as presented in Fig. 7c for subsequent analysis on air void characteristics.
2.4. Air voids properties Air voids were analyzed in terms of content, number, shape, and size. The shape of the air voids can indicate the interconnectivity among them. The shape of air voids was described using circularity and aspect ratio as shown in Eqs. (1) and
Scan sample using X-ray CT scanner Image processing using ImageJ
DEQPC ¼
½Area ½Perimeter
2
½Length of major axis ½Length of minor axis
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 4 Area
p
ð1Þ
ð2Þ
ð3Þ
3. Result and discussion 3.1. Voids distribution (Percentage and Number) Void content and void number present in the porous asphalt mixture significantly influence the functional performance of pavement such as permeability, acoustic characteristics, and frictional properties. To analyze the void distribution, the air void content was initially determined in accordance with ASTM D3203 specification [41]. The total void content was obtained to be 19.5%. On the basis of the void content, the image threshold value was selected and later used to determine the void distribution throughout the sample. The analyses were compared between two different cross sections, i.e., horizontal and vertical cross sections.
Convert scanned images into vertical images Scanned images that are in the form of horizontal slices are converted into vertical images using volume viewer (ImageJ) at every 10° interval.
Combining stack images Images from the X-ray CT scan are combined in form of multiple stack (1024×1024 pixels, 16 bits).
Crop and stack images Images of the sample are cropped and stacked together.
Convert image Images that are in the form of 16 bits are converted to 8 bits.
Estimate air voids content Image histogram of the air voids area are summed and later compared to the actual experimental sample.
Crop and rescale images Images of the sample are cropped and rescaled to the actual dimension of the sample that are in mm.
Selecting air voids area (T0-T1) Air voids properties are selected upon the summation of threshold value.
Image enhancement Apply contrast enhancement and Gaussian filter (noise elimination function).
487
Proceed to analysis Air voids properties such as percentage, size, number, and shape are ready to be analysed.
Fig. 2. Framework on the vertical cut section analysis.
488
M.Z.H. Mahmud et al. / Construction and Building Materials 155 (2017) 485–494
Fig. 3. Image acquisition form horizontal cross sectional view (a) 3D reconstruction (b) stack of 2D images.
Low Contrast
(a)
Better Contrast
(b)
Fig. 4. X-ray images of porous asphalt (a) Original (b) After enhancement.
Fig. 5. Illustration of orientation angles on horizontal cross section.
Figs. 9 and 10 show the analysis of void content and void number on the horizontal and vertical cross sections, respectively. The void content and void number in Fig. 9 were plotted throughout the sample height at the interval of 1 mm height, and the vertical cross sections as shown in the polar diagram (Fig. 10) were
analyzed at the angle interval of 10°. The polar diagram shows the top view of the scanned sample with detailed vertical cross section. For example, the 0°–180° line represents an image of vertical cross section (virtual cut section). A total of 18 vertical cross sections were analyzed at the interval of 10° polar angle. Fig. 9 highlights the distributions of the void content and void number based on horizontal cross sectional analysis. The analysis provides the void distribution from top to bottom of the sample. Analyzing the material properties using this perspective is a common method used by many researchers in describing the distributions of void properties in asphalt mixtures [14,42–44]. The void content analysis in Fig. 9a shows that the mean air void content is 19.5% with the standard deviation of 1.38. The calculated mean number of voids is 11.7 with the standard deviation of 0.91 as shown in Fig. 9b. In general, the distributions of voids based on horizontal perspective shows that void content and void number are homogeneously distributed with slight deviation at the top and bottom of the sample because of the presence of high air void content. High percentage of voids can be observed at the top and bottom sections of the sample as a result of ‘‘confinement effect” during the compaction process. As shown in Fig. 9b, the void number is homogeneously distributed but is slightly reduced at the middle (at heights from 30 mm to 40 mm) of the sample. This finding is due to the presence of a large amount of coarse aggregates at the middle of the sample, thereby creating a large area of voids as a result of compaction. The second part of the analysis focuses on the radial distribution of voids in porous asphalt mixture presented in polar diagram as shown in Fig. 10. This method measures the distributions of voids within the sample using vertical cross sections at every 10°. Fig. 10a shows that the calculated mean air void content is
M.Z.H. Mahmud et al. / Construction and Building Materials 155 (2017) 485–494
489
Fig. 6. Vertical Cross Section View (a) 3D reconstruction (b) 2D Illustration.
Fig. 7. Image threshold process for the vertical cross section (a) X-ray image (b) thresholding (c) binary image.
3.2. Air voids shape properties
Fig. 8. Feret’s diameter and minimum Feret of an air voids.
19.48% with the standard deviation of 4.92. Fig. 10b shows that the mean number of voids distributed radially is 124 and the standard deviation is 10.84. In general, the void properties are homogeneously distributed within the sample. Comparing the two polar diagrams in Fig. 10a and b shows that, when the number of air voids present in the mixture is high, the percentage of air voids produced is small. These findings suggest that void content and void number exhibit an inverse relationship. For example, at plane 120°–300°, the percentage of air voids is 22.4 %, which is slightly higher than that of the mean air voids of 19.48%. On the same plane, the void number in Fig. 10b appears to experience depression with the value of 103 compared with the mean of 124. The reason is that large air voids converge to one another, thereby creating large size of air voids and thus producing less number of voids. Accordingly, the void connectivity increases.
Shape properties of air voids are significant parameters as they reflect the connectivity of the air voids [5,45]. Two main parameters were selected in this study, i.e., circularity and aspect ratio. ‘‘Circularity” describes whether the voids are in circular formation, while aspect ratio describes the elongation magnitude of the voids [5,38,42]. The formation of void connectivity in the porous asphalt can be described with high elongated voids and less circular shape, and the analyses were conducted for vertical and horizontal cut sections. Fig. 11 shows the distribution of void circularity and aspect ratio throughout the sample based on horizontal cut section analysis. Elongated air voids are observed at the bottom section of the sample unlike the top section. The above-mentioned parameters indicate that the size of the air voids at the bottom section is nearly three times the ratio of the minor axis detected. On the basis of the obtained mean value, the circularity and aspect ratio values are found to be 0.6 with standard deviations of 0.03 and 2.38 and 0.215, respectively, indicating void connectivity throughout the sample. Fig. 12 shows the analysis of void shape on the vertical cut sections and plotted at 10° interval. The polar diagram is an indicator for radial distribution of air void shape. Fig. 12 shows that the shape of air voids in terms of circularity and aspect ratio are generally well distributed with slight deviation. Fig. 12a shows a peak of 0.67 at 10°–190° plane of cross-section image. The mean circularity is 0.64 with the standard deviation of 0.015, and the mean value of aspect ratio is 2.13 with the standard deviation of 0.156. Therefore, the length of the major axis is approximately two times the length of the minor axis. Both shape parameters confirm that air voids within the porous asphalt mixture are prone to be elongated, thereby justifying the air void connectivity. Similarly, a study by Hassan et al. [5] also reveals that air voids in porous asphalt mixture are predominantly in elongated formation. From
490
M.Z.H. Mahmud et al. / Construction and Building Materials 155 (2017) 485–494
Voids, % 10
20
30
Number 40
5
50
5
5
10
10
Number
15
Mean
Percentage
15
Mean
Height, mm
Height, mm
10
15
0
0
20 25 30
20 25 30
35
35
40
40
45
45
50
50
(a)
(b)
Fig. 9. Analysis of (a) Voids contents (b) Voids number for horizontal cross section.
Percentage
Mean
Number
0̊ 340 ̊ 330 ̊ 320 ̊ 310 ̊
350 ̊ 25
10 ̊
0̊ 20 ̊ 30 ̊
330 ̊ 320 ̊ 310 ̊
40 ̊
20
50 ̊ 15
300 ̊
70 ̊
290 ̊
280 ̊
5
80 ̊
280 ̊
270 ̊
0
90 ̊
270 ̊
100 ̊
260 ̊
260 ̊ 250 ̊
110 ̊
240 ̊
120 ̊
230 ̊ 220 ̊ 210 ̊
130 ̊ 140 ̊ 150 ̊ 200 ̊
190 ̊
170 ̊
340 ̊
300 ̊
60 ̊
10
290 ̊
Mean
160 ̊
180 ̊
(a)
350 ̊ 160 140 120 100 80 60 40 20 0
10 ̊
20 ̊
30 ̊ 40 ̊ 50 ̊ 60 ̊ 70 ̊ 80 ̊ 90 ̊ 100 ̊
250 ̊
110 ̊
240 ̊ 230 ̊ 220 ̊ 210 ̊
120 ̊
200 ̊
190 ̊
170 ̊
160 ̊
130 ̊ 140 ̊ 150 ̊
180 ̊
(b)
Fig. 10. Analysis of (a) Voids contents (b) Voids number for vertical cross section.
Figs. 11 and 12, the voids are found to be connected vertically and horizontally. In addition to the vertical cross-section analysis, the vertical cut section was equally divided into top, middle, and bottom in assessing the void shape properties within the said regions. This procedure was conducted to assess the concentration of void connectivity throughout the sample and the homogeneous distribution of voids within the sample. The image of vertical cut section divided into top, middle, and bottom regions is shown in Fig. 13. Fig. 14 shows the detailed distribution of circularity and aspect ratio values for the top, middle, and bottom regions. For comparison, the distribution within the different regions was compared with the ‘‘total” as presented in the polar diagram. In general, the shape of air voids using circularity and aspect ratio parameters is evenly distributed at different heights of the sample. Fig. 14a shows the distribution of air voids on the basis of the circular parameter. As the value approaches 1, the air voids are in perfect circular shape. Based on the result, it was found that approximately about 34% of the air voids is in the form of perfect circle
that falls in the range of 0.8–1.0. On the other hand, the presence of perfect elongated air voids (0–0.2) was found approximately 6%. Although the result in Fig. 14a shows that only 34% air voids are in the form of a perfect circle, the remaining air voids shape is not in a form of a perfect circular formation. Fig. 14b shows the analysis on the aspect ratio, which reflects the magnitude of void elongation. More than 50% of the air voids are in the range of 1–2. Therefore, most size of the air voids is approximately two times the size of the identified minor axis. The remaining 42% of air voids present an aspect ratio of greater than 2, indicating the void connectivity within the compacted sample. These finding shows the air voids are predominantly elongated and evenly distributed radially and vertically. One of the main reasons that caused the formations of the air voids to be predominately elongated is due to the aggregates interlocking process that occurs within the aggregate skeleton and later, influenced the shaped of the air voids. Masad and Button [12] discussed on the selection of aggregate gradation based on Bailey method. When large aggregates (coarse aggregates) interlocks with other large aggregate
491
M.Z.H. Mahmud et al. / Construction and Building Materials 155 (2017) 485–494
Circularity 0.45
0.5
0.55
Aspect Rao
0.6
0.65
1
0.7
0
0
5
5
10
15
Height, mm
Height, mm
3
4
10
Top
15 20 25 30
Top
20 25 30 35
35 40
2
40
Bottom
Circularity
Aspect rao
Bottom
45
45 Mean
50
Mean
50
(a)
(b)
Fig. 11. Sample height diagram on voids shape properties (a) Circularity (b) Aspect Ratio.
Circularity
Aspect Rao
Mean
340 ̊ 330 ̊ 320 ̊ 310 ̊
350 ̊ 0.70
10 ̊
40 ̊ 50 ̊
70 ̊
0.50 0.45
270 ̊
0.40
260 ̊ 250 ̊
230 ̊ 220 ̊ 210 ̊ 200 ̊
190 ̊
170 ̊
90 ̊
270 ̊
0.0
100 ̊
260 ̊
160 ̊
30 ̊ 40 ̊ 50 ̊ 60 ̊ 70 ̊
1.0 0.5
130 ̊ 140 ̊ 150 ̊
20 ̊
1.5
290 ̊ 280 ̊
120 ̊
10 ̊
2.0
80 ̊
110 ̊
240 ̊
350 ̊ 3.0 2.5
300 ̊
60 ̊
0.55
280 ̊
330 ̊ 320 ̊ 310 ̊
30 ̊
0.60
290 ̊
340 ̊
20 ̊
0.65
300 ̊
Mean 0̊
0̊
80 ̊ 90 ̊ 100 ̊
250 ̊
110 ̊
240 ̊
120 ̊
230 ̊ 220 ̊ 210 ̊
130 ̊ 140 ̊ 150 ̊ 200 ̊
190 ̊
170 ̊
160 ̊
180 ̊
180 ̊
(a)
(b)
Fig. 12. Polar diagram of the vertical cut sections on voids shape properties (a) Circularity (b) Aspect Ratio.
3.3. Voids diameter
Top Middle Bottom Fig. 13. Vertical cut section divided into top, middle and bottom regions.
particles, it will create voids where these voids will be filled by fine aggregates. For porous asphalt, large voids present due to lack of fine aggregate in the asphalt mixture gradation. In this study, these parameters (i.e., circularity and aspect ratio) show that the air voids shape are predominantly elongated.
Void diameter was analyzed using DEQPC and Fmax. The DEQPC parameter measures the equivalent projection area of a circle. While, Fmax parameter measures the maximum distance between two pixels at the boundary of the air voids. Hassan et al. [38] used these two parameters in estimating the dimension of the aggregate gradations analysis and successfully described the dimensions of the coarse aggregate particles. This study applies these parameters in describing the air voids dimension. Fig. 15 compares both parameters analyzed using horizontal and vertical cut sections. As shown in Fig. 15a, the means of DEQPC and Fmax are approximately 1.5 and 2.5 mm, respectively. Fig. 15b shows the means of DEQPC and Fmax from horizontal cut sections are approximately 3.1 and 4.1 mm, respectively. Comparing both findings suggests that the diameter analyzed using vertical cut sections provides a
492
M.Z.H. Mahmud et al. / Construction and Building Materials 155 (2017) 485–494
Total
Top
Middle
Boom
34.64
35
Percentage. %
30 22.59
25 19.78
20
16.83
15 10
6.16
5 0 0.0-0.2
0.4-0.6
0.2-0.4
0.6-0.8
0.8-1.0
Circularity
(a) Total
Percentage. %
60
Top
Middle
Boom
54.78
50 40 28.75
30 20 10
9.20 4.73
2.54
0 0-1
1-2
2-3
3-4
>4
Aspect Rao
(b) Fig. 14. Distribution of voids shape properties in different regions based on vertical cut sections (a) Circularity (b) Aspect ratio.
F max
Diameter, mm 0
1
2
3
4
Mean F max
D EQPC
5
0
Mean D EQPC 0̊
340 ̊
350 ̊ 6
330 ̊ 320 ̊ 310 ̊
5 10
Height, mm
25 30
40 ̊ 50 ̊ 60 ̊
3
70 ̊
2
F max
280 ̊
1
Mean F max
270 ̊
0
D EQPC
30 ̊
4
290 ̊
20
20 ̊
5
300 ̊
15
10 ̊
80 ̊ 90 ̊
260 ̊
100 ̊
250 ̊
35 Mean D EQPC 40
110 ̊
240 ̊
120 ̊
230 ̊ 220 ̊ 210 ̊
45
130 ̊ 140 ̊ 150 ̊ 200 ̊
50
190 ̊
170 ̊
160 ̊
180 ̊
(a)
(b)
Fig. 15. Voids diameter using Feret’s diameter and Diameter EQPC method analyzed on (a) horizontal (b) vertical cut sections.
1 mm larger mean diameter than the one analyzed using horizontal cut sections. Therefore, the vertical cut sections complement the conventional analysis using horizontal cut sections. Similar pattern of distribution (homogeneous) in radial direction is shown in Fig. 15b for DEQPC and Fmax. Large air void diameter is visible at a plane of 120°–300° with the Fmax size of more than 5.1 mm. However, Fig. 15a shows that Fmax is more sensitive than DEQPC in
detecting changes in air void diameter and the connectivity of voids. For detailed size distribution, Fig. 16 categorizes void diameter (DEPC and Fmax) in different ranges according to the standard sieve size. The analysis was conducted for the top, middle, and bottom regions. As discussed earlier, Fmax is a good parameter for detecting changes in void size; this deduction is supported by the results
493
M.Z.H. Mahmud et al. / Construction and Building Materials 155 (2017) 485–494
Diameter, DEQPC
Total
Top
Middle
Boom
Percentage. %
30 25 20 15 10 5 0 0.15-0.30
0.3-0.60
0.60-1.18
1.18-2.36
2.36-5.00
5-10
10-14
14-20
Size, mm
(a) 30
Feret's Diameter, Fmax
Percentage. %
25 20 15 10 5 0 0.15-0.30 0.3-0.60 0.60-1.18 1.18-2.36 2.36-5.00
5-10
10-14
14-20
20-28
> 28
Size, mm
(b) Fig. 16. Distributions of voids diameter in different regions based on vertical cut sections (a) Diameter EQPC (b) Feret’s Diameter.
shown in Fig. 16. Fmax detects a larger range of air void size of up to 28 mm diameter than DEQPC with only up to 20 mm diameter. The analysis performed with DEQPC also shows that the majority of the air void sizes (approximately 28%) are in the range of 1.18– 2.36 mm. Notably, the void size is evenly distributed from the top to bottom of the sample. However, the analysis using Fmax shows large void size with the majority of air voids detected in the range of 1.18–5.00 mm. A small void size, i.e., 0.15–0.30 mm, appears to be significantly lesser when analyzed using Fmax than using DEQPC. The finding shows the analysis of voids at the top, middle, and bottom of the sample indicates that most of the small voids (<1.18 mm) are located at the top section and large voids (>2.36 mm) are slightly higher at the bottom of the sample than at the other sections. This is because the influence of aggregate particle rearrangement in the formation of air voids particularly in the process of compacting asphalt mixture. In this process, blended coarse aggregates tend to settle into the gyratory compaction mold due to gravitational force and the small aggregates tends to stick on the surface of the mixing mold [39]. As a result, the formation of large size air voids are at the bottom of the sample where the smaller size air voids are located at the top of the sample.
4. Conclusions This study introduced a research framework for performing virtual cut section analysis on porous asphalt mixture using X-ray CT images. By using virtual cut sections, this method is able to explore the internal structure of the asphalt mixture by regenerating new multiple cross-sectional images at various angles without disrupting the physical properties of the sample. For this study, the outcome of using virtual cut section in porous asphalt mixture enables the radial distribution of air voids properties to be analyzed. This analysis is presented in the form of polar diagram. This
diagram is used to complement the conventional cross-section analysis to provide a complete view (vertical and horizontal views) of the distribution of air void properties. As a result, more data can be extracted from the analysis particularly from the radial distribution perspective. Based on the analysis performed on the air voids properties, it shows that this method is effective in analyzing the radial distributions of void connectivity of the porous asphalt mixture sample and complements the conventional cross-section analysis (horizontal). The findings indicate that both of the cross-section images show that the air voids are homogeneously distributed. The result also shows that percentage of air voids content and number of voids have an inversed relationship. Study on circularity and aspect ratio confirms that the shape of the air voids in the porous asphalt mixture is predominantly elongated. By using Fmax parameter in analyzing the dimension of the sample, the measurement is considered more accurate whereby it gives approximately 1 mm larger compared to DEQPC parameter. Hence, this study recommends the use of Fmax parameter in analyzing the dimension of air voids in porous asphalt mixture. Acknowledgments Universiti Teknologi Malaysia and Malaysia Ministry of Education are sincerely acknowledged for providing the research funds (Vote no. Q.J130000.2522.11H76 and R.J130000.7822.4F867) and laboratory facilities. References [1] A.E. Alvarez, A.E. Martin, C. Estakhri, A review of mix design and evaluation research for permeable friction course mixtures, Constr. Build. Mater. 25 (3) (2011) 1159–1166. [2] R.B. Mallick, P. Kandhal, L.A. Cooley, D. Watson, Design, construction, and performance of new-generation open-graded friction courses, Asphalt Paving Technol. 69 (2000) 391–423.
494
M.Z.H. Mahmud et al. / Construction and Building Materials 155 (2017) 485–494
[3] B.J. Putman, L.C. Kline, Comparison of mix design methods for porous asphalt mixtures, J. Mater. Civ. Eng. 24 (11) (2012) 1359–1367. [4] G. Huber, Performance Survey on Open-Graded Friction Course Mixes, Synthesis of Highway Practice 284, National Cooperative Highway Research Program, Transportation Research Board, National Research Council, Washington, DC, 2000. [5] N.A. Hassan, M.Z.H. Mahmud, R.P. Jaya, Air void characterisation in porous asphalt using X-ray computed tomography, Adv. Mater. Res. 2014 (911) (2014) 443–448. [6] N.A. Hassan, N.A.M. Abdullah, N.A.M. Shukry, M.Z.H. Mahmud, N.Z.M. Yunus, R. P. Jaya, M.R. Hainin, H. Yaacob, Laboratory evaluation on the effect of clogging on permeability of porous asphalt mixtures, J. Teknol. 76 (14) (2015) 77–84. [7] A.E. Alvarez, A.E. Martin, C.K. Estakhri, J.W. Button, C.J. Glover, S.H. Jung, Synthesis of Current Practice on the Design, Construction, and Maintenance of Porous Friction Courses (Report No. FHWA/TX-06/0-5262-1), U.S. Department of Transportation, Federal Highway Administration, 2006. [8] T. Fwa, S. Tan, Y. Guwe, Laboratory evaluation of clogging potential of porous asphalt mixtures, Transp. Res. Rec. 1681 (1999) 43–49. [9] W. Jiang, A. Sha, J. Xiao, Experimental study on relationships among composition, microscopic void features, and performance of porous asphalt concrete, J. Mater. Civ. Eng. 27 (11) (2015). [10] P.J. Vardanega, State of the art: permeability of asphalt concrete, J. Mater. Civ. Eng. 26 (1) (2012) 54–64. [11] E. Arambula, E. Masad, A.E. Martin, Influence of air void distribution on the moisture susceptibility of asphalt mixes, J. Mater. Civ. Eng. 19 (8) (2007) 655– 664. [12] E. Masad, J. Button, Implications of experimental measurements and analyses of the internal structure of hot-mix asphalt, Transp. Res. Rec. 1891 (2004) 212– 220. [13] N.A. Hassan, G.D. Airey, M.R. Hainin, Characterisation of micro-structural damage in asphalt mixtures using image analysis, Constr. Build. Mater. 54 (2014) 27–38. [14] N.A. Hassan, M.Z.H. Mahmud, K.A. Ahmad, M.R. Hainin, R.P. Jaya, N. Mashros, Air voids characterisation and permeability of porous asphalt gradations used in different countries, ARPN J. Eng. Appl. Sci. 11 (24) (2016) 14043–14047. [15] N. Neithalath, M.S. Sumanasooriya, O. Deo, Characterizing pore volume, sizes, and connectivity in pervious concretes for permeability prediction, Mater. Charact. 61 (8) (2010) 802–813. [16] A.E. Alvarez, A.E. Martin, C. Estakhri, R. Izzo, Determination of volumetric properties for permeable friction course mixtures, J. Test. Eval. 37 (1) (2009) 1. [17] E. Brown, M.R. Hainin, L.A. Cooley Jr, Determining minimum lift thickness for hot mix asphalt (HMA) mixtures (with discussion), J. Assoc. Asphalt Paving Technol. 74 (2005). [18] L.A. Cooley Jr, B.D. Prowell, M.R. Hainin, Comparison of the saturated surfacedry and vacuum sealing methods for determining the bulk specific gravity of compacted HMA (with discussion and closure), J. Assoc. Asphalt Paving Technol. 72 (2003). [19] O. Deo, M. Sumanasooriya, N. Neithalath, Permeability reduction in pervious concretes due to clogging: experiments and modeling, J. Mater. Civ. Eng. 22 (7) (2010) 741–751. [20] M.R. Hainin, L.A. Cooley Jr, B.D. Prowell, An investigation of factors influencing permeability of superpave mixes, Int. J. Pavements 2 (2) (2003) 41–52. [21] M.R. Hainin, N.I.M. Yusoff, M.K.I.M. Satar, E.R. Brown, The effect of lift thickness on permeability and the time available for compaction of hot mix asphalt pavement under tropical climate condition, Constr. Build. Mater. 48 (2013) 315–324. [22] M.R.M. Hasan, M.O. Hamzah, M. van de Ven, J. Voskuilen, Disruption of air voids continuity based on permeability loss due to mortar creep, Constr. Build. Mater. 116 (2016) 347–354.
[23] D. Watson, K. Moore, K. Williams, L. Cooley, Refinement of new-generation open-graded friction course mix design, Transp. Res. Rec. 1832 (2003) 78–85. [24] D. Chen, N. Roohi Sefidmazgi, H. Bahia, Exploring the feasibility of evaluating asphalt pavement surface macro-texture using image-based texture analysis method, Road Mater. Pavement Des. 16 (2) (2015) 405–420. [25] E. Masad, B. Muhunthan, N. Shashidhar, T. Harman, Internal structure characterization of asphalt concrete using image analysis, J. Comput. Civ. Eng. 13 (2) (1999) 88–95. [26] E. Masad, V.K. Jandhyala, N. Dasgupta, N. Somadevan, N. Shashidhar, Characterization of air void distribution in asphalt mixes using X-ray computed tomography, J. Mater. Civ. Eng. 14 (2) (2002) 122–129. [27] L. Tashman, E. Masad, J. D’Angelo, J. Bukowski, T. Harman, X-ray tomography to characterize air void distribution in superpave gyratory compacted specimens, Int. J. Pavement Eng. 3 (1) (2012) 19–28. [28] Z.Q. Yue, I. Morin, Digital image processing for aggregate orientation in asphalt concrete mixtures, Can. J. Civ. Eng. 23 (1996) 480–489. [29] A.E. Alvarez, A.E. Martin, C. Estakhri, Internal structure of compacted permeable friction course mixtures, Constr. Build. Mater. 24 (6) (2010) 1027–1035. [30] J.B. Król, R. Khan, A.C. Collop, The study of the effect of internal structure on permeability of porous asphalt, Road Mater. Pavement Des. (2017) 1–17. [31] A. Bozorgzad, Consistent distribution of air voids and asphalt and random orientation of aggregates by flipping specimens during gyratory compaction process, Constr. Build. Mater. 132 (2017) 376–382. [32] Z. You, S. Adhikari, M.E. Kutay, Dynamic modulus simulation of the asphalt concrete using the X-ray computed tomography images, Mater. Struct. 42 (5) (2009) 617–630. [33] P.W.D. Malaysia, Standard Specification for Road Works, Section 4, Flexible Pavement, Jabatan Kerja Raya Malaysia, Kuala Lumpur, 2008. [34] ASTM D 7064, Standard Practice for Open-Graded Friction Course (OGFC) Mix Design, ASTM International, West Conshochocken, PA, 2013. [35] M. Buchanan, E. Brown, Effect of superpave gyratory compactor type on compacted hot-mix asphalt density, Transp. Res. Rec. 1761 (2001) 50–60. [36] E. Brown, J. Haddock, Method to ensure stone-on-stone contact in stone matrix asphalt paving mixtures, Transp. Res. Rec. 1583 (1997) 11–18. [37] Australian Asphalt Pavement Association (AAPA), National Asphalt Specification, Australian Asphalt Association, 2004. [38] N.A. Hassan, Microstructural Characterisation of Rubber Modified Asphalt Mixtures (Doctoral dissertation), University of Nottingham, UK, 2012. [39] N.A. Hassan, G.D. Airey, R. Khan, A.C. Collop, Nondestructive characterisation of the effect of asphalt mixture compaction on aggregate orientation and segregation using X-ray computed tomography, Int. J. Pavement Res. Technol. 5 (2) (2012) 84–92. [40] K.U. Barthel, 3D-data representation with ImageJ, in: ImageJ Conference, 2006. [41] ASTM D 3203, Standard Test Method for Percent Air Voids in Compacted Dense and Open Bituminous Paving Mixtures, ASTM International, West Conshohocken, PA, 2011. [42] M. Aboufoul, A. Garcia, Influence of air voids characteristics on the hydraulic conductivity of asphalt mixture, Road Mater. Pavement Des. (2017) 1–11. [43] W.D. Martin, B.J. Putman, N.B. Kaye, Using image analysis to measure the porosity distribution of a porous pavement, Constr. Build. Mater. 48 (2013) 210–217. [44] D. Lo Presti, R. Khan, N.A. Hassan, G. Airey, A. Collop, Laboratory mix design of asphalt mixture containing reclaimed material, Adv. Mater. Sci. Eng. 2014 (2014). [45] W.S. Abdullah, M.T. Obaidat, N.M. Abu-Sa’da, Influence of aggregate type and gradation on voids of asphalt concrete pavements, J. Mater. Civ. Eng. 10 (2) (1998) 76–85.
Contents lists available at ScienceDirect
Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat
Microstructural investigation on air void properties of porous asphalt using virtual cut section Mohd Zul Hanif Mahmud, Norhidayah Abdul Hassan ⇑, Mohd Rosli Hainin, Che Ros Ismail Faculty of Civil Engineering, Universiti Teknologi Malaysia, 81310 UTM Skudai, Johor, Malaysia
h i g h l i g h t s A research framework for virtual cut section analysis on porous asphalt mixture. Polar diagram is used to complement the conventional cross-section analysis. Air voids shape in porous asphalt are predominantly elongated. Recommend Fmax parameter in determining air void size of porous asphalt.
a r t i c l e
i n f o
Article history: Received 20 June 2017 Received in revised form 16 August 2017 Accepted 17 August 2017
Keywords: Porous asphalt mixtures Virtual cut section Air voids distribution Air voids properties Image analysis technique X-ray CT
a b s t r a c t The microstructure properties of asphalt mixtures reflect the structural and functional performance of asphalt mixtures. Thus, numerous imaging methods are developed including destructive and nondestructive methods to uncover the internal structure properties of asphalt mixtures. However, accurate analysis of internal microstructure properties is challenging. This study used a non-destructive method employing virtual cut section (imaging technique) to analyze the air void properties of laboratoryfabricated porous asphalt. Conventionally, the investigation of internal structure properties particularly on vertical section requires the sample to be cut (destructive method) into several sections. This method causes defects to the sample and limits the amount of data that can be extracted from the image. The imaging technique designed in this study provides comprehensive understanding of the distribution and properties of air voids from different angles within the compacted sample of porous asphalt. The void properties in porous asphalt should be assessed accurately to ensure effectiveness of water permutation through the porous structure and acoustic function of the mixture. The compacted samples were X-rayed and virtually analyzed into various cut sections. The void properties were characterized in terms of the content, number, shape, and size of voids. Results indicate that air voids within the porous asphalt are homogeneously distributed with an elongated shape; this characteristic indicates high void connectivity. The vertical cross-section analysis justifies the characteristics of air voids in the conventional analysis of horizontal cross section. Ó 2017 Published by Elsevier Ltd.
1. Introduction Porous asphalt is typically used as a wearing course that allows water to permeate into the porous surface. The process is possible because porous asphalt is designed on the basis of open graded gradation [1–3]. Open graded gradation consists of a large quantity of coarse aggregates and a small quantity of fine aggregates [4–6]. This method allows for the mobility of aggregates during compaction, thereby resulting in high formation of interconnected voids within the layer. Air void content (interconnected voids) in ⇑ Corresponding author. E-mail address: [email protected] (N.A. Hassan). http://dx.doi.org/10.1016/j.conbuildmat.2017.08.103 0950-0618/Ó 2017 Published by Elsevier Ltd.
porous asphalt mixture is the main parameter that influences the structural and functional performance as well as service life [7–10]. The common parameters used to describe the connectivity of air voids are the content, number, shape, and size of voids [1,11–15]. This pavement type needs to maintain sufficient amount of air void content to ensure effective permutation of the pavement and prolong the benefits of the structure to the road users. Therefore, a detailed analysis on the air void formation within the porous asphalt is significant to enhance understanding of its functional performance. The content or properties of air voids can be quantified using various methods, including vacuum method, dimensional analysis, water displacement concept, and imaging technique [16–23].
486
M.Z.H. Mahmud et al. / Construction and Building Materials 155 (2017) 485–494
Numerous efforts have been made in the recent years to investigate the microstructure properties (particularly air voids) of asphalt mixture using imaging techniques [9,24–27]. Using these techniques, air voids can be characterized either in a destructive or a non-destructive manner. Current non-destructive imaging technique, i.e., X-ray computed tomography (CT) scanning, is preferred as the sample can be used for subsequent testing after the scanning process. Another option to investigate air void properties is through destructive methods. A destructive method for imaging the internal structure properties of asphalt mixture was developed in early 1990s [28]. Although destructive imaging techniques are cost effective and do not require sophisticated equipment (digital camera) to obtain the cross-sectional view of the sample, the process of cutting requires the use of a special cutter, thereby limiting the number of permissible cross sections or image slices to be obtained. Furthermore, these techniques cause misleading analysis on the properties and distributions of air voids. The cutting process triggers air void distortion in addition to causing severe aggregate loss especially at the edge of the slices. As porous asphalt is designed to possess low structural strength and limited amount of fine aggregates, the cutting process will result in deformed samples and excessive aggregate loss. Thus, void properties captured using destructive methods in terms of quantity, size, and shape may differ from those obtained using non-destructive methods. Non-destructive techniques allow the internal structures of the asphalt samples to be analyzed without cutting or disrupting the sample [5,29,30]. Such techniques provide detail overview of the internal structure as the images can be captured throughout the samples and analyzed from various angles [13,31,32]. However, considering only the horizontal cross section of the X-ray CT images for the analysis of air void distribution cannot fully describe the exact void connectivity as the water flows from the top to the bottom section of the porous asphalt. Voids within the porous asphalt layer act as a series of micropipes that connect to one another to channel the water away from the surface [5,14]. Therefore, the horizontal cut section of the captured images will only provide the transverse section of the pipes. Compared with the vertical cross section, the horizontal one will provide the longitudinal section of the pipes to reflect the void connectivity. Accordingly, better views of the hydraulic conductivity from top to bottom within the porous asphalt layer can be provided. In this study, air void properties of porous asphalt were nondestructively characterized using the virtual cut section (vertical cross-sectional image) of X-ray CT images. A complete procedure of the image analysis technique was introduced including image acquisition, image processing, and image analysis. Although this study applies the procedure on porous asphalt, the concept is relevant to any potential materials. The void analysis covers a few basic parameters (i.e., content, number, size, and shape) measured from the images in horizontal and vertical cut sections and relates to the air void connectivity.
graded gradation enveloped with the nominal maximum aggregate size of 13.2 mm as shown in Fig. 1 [37]. Polymer modified bitumen (PG 76) was added to the sample as a binder at designed bitumen content of 5.25% (obtained from mixture design). The samples were mixed at 195 °C first and then compacted at 175 °C. The properties of the materials used are provided in Table 1.
2.2. Microstructure investigation This section emphasizes on the characterization of air void properties of porous asphalt. Air voids play an important role in influencing the permeability, acoustic characteristics, and frictional properties of porous asphalt. Thus, accurate data of air void properties should be obtained because they reflect the performance of porous asphalt. The porous asphalt samples were scanned using X-ray CT scanner to capture the horizontal images. The horizontal images were then processed and analyzed using the ImageJ software. Fig. 2 shows the detailed framework of the study. The framework was divided into two main stages. Stage I focuses on image processing wherein several imaging procedures were performed on the images, such as cropping, image scaling, and quality enhancement. Stage II involves the development of three-dimensional (3D) images from horizontal images using Volume Viewer. The vertical cut sections were obtained from the 3D images and then stacked for subsequent analysis. The air voids were then thresholded, analyzed, and verified to the experimental value. Details on the image acquisition and processing methods are provided in the following sections.
2.2.1. X-ray CT scan and image acquisitions This study used a non-destructive image acquisition technique to capture the cross-sectional view of porous asphalt. The samples were scanned using X-ray CT scanner (inspeXio smx-225 CT) with the maximum output voltage of 225 kV. This scanner used the cone beam method to reconstruct 3D images as shown in Fig. 3a. During the scanning process, the sample was rotated 360° at an image slice of 0.1 mm thickness. The scanning parameter of the X-ray CT scanner was set to capture 1200 views with the average count of 15 times at each angle of rotation. By increasing the number of views and average count, the data obtained from the scanning procedure would be accurate (detailed information) but would increase the scanning duration. These image slices were combined into a stack at the interval of approximately 0.1 mm as illustrated in Fig. 3b. The output voltage used to scan the sample was selected to be 190 kV with the ampere of 100 mA for good-quality images. The resolution obtained from the X-ray CT scan was 0.106 mm/pixel. These images were saved in Tagged Image File Format (TIFF) as 16 bits. TIFF digital file format was selected to avoid data compression and degraded image quality during image conversion and processing.
2.2.2. Image enhancement Considerable imaging software is available for image processing and image analysis. In this study, the images were processed using ImageJ. ImageJ is an open source imaging software initially designed for medical purposes. TIFF images were converted into 8 bits for further analysis. Notably, 8-bit image comprises of 256 gray levels of the dark regions (low-density material, i.e., air voids) and the brightest regions (high-density material, i.e., aggregate). Images obtained from the X-ray scanning were analyzed as stack formation. A few image slices at the top and bottom sections were excluded from the analysis due to image distortion during the scanning process. These images were cropped in accordance with the exact dimension of the sample (100 mm diameter and 50 mm thickness). Images obtained from X-ray CT scanning are often not in perfect conditions and thus require image enhancement and correction owing to excessive noise, poor contrast, imbalance color representation, and disproportional illumination [13,38,39]. Thus, image treatments using several mathematical functions, such as contrast enhancement and noise elimination (Gaussian filter or Median filter), were conducted to improve the appearance of the image as shown in Fig. 4. This procedure was conducted to visualize the region of interest for accurate image analysis.
2. Material and methods
Laboratory-fabricated samples of porous asphalt mixture were prepared for Xray CT scanning. The samples were prepared in accordance to Malaysian road specifications [33]. The samples were compacted using Superpave gyratory compactor with 50 gyrations in accordance with ASTM D 7064 specification [34]. Superpave gyratory compactor is able to simulate the actual field conditions [35]. Furthermore, the Superpave gyratory compactor method can produce lower coarse aggregate breakdown [36]. Each of these samples was compacted at the vertical pressure of 600 kPa and gyration rate of 30 rpm to achieve the desired air void content of approximately 20%. Due to the limitations of the X-ray scanner (inspeXio smx225 CT) in capturing the microstructure properties of the asphalt mixture at large scale, the samples were prepared in a 100 mm diameter mold with thickness of 50 mm. The granite type aggregates were sieved to obtain an Australian open
Percentage Passing. (%)
2.1. Materials and sample preparation
100
Upper Limit
Lower Limit
Middle
80 60 40 20 0 0.01
0.1
1
10
100
Sieve Size, (mm) Fig. 1. Australian open graded gradation for porous asphalt mixture, (AAPA, 2004).
M.Z.H. Mahmud et al. / Construction and Building Materials 155 (2017) 485–494 Table 1 Materials Properties. Materials
Physical Properties
Coarse Aggregate
Bulk Specific Gravity Water Absorption Aggregate Impact Value
2.601 0.9% 26.0%
Fine Aggregate
Bulk Specific Gravity Water Absorption
2.427 2.0%
PG 76 bitumen
Specific gravity at 25 °C Penetration test at 25 °C Softening point test Viscosity at 135 °C Viscosity at 165 °C G*/Sin d, kPa at 76 °C
1.030 44 PEN 75 °C 1.04 Pa.s 0.4 Pa.s 1.6 kPa
(2). For circularity, the value of 1 indicates a perfect circle and the value approaching 0 means otherwise. Meanwhile, high value of aspect ratio indicates the degree of elongation in the major axis is higher than in the minor axis. Most of the void shape in porous asphalt mixtures is elongated, indicating high air void connectivity. The size of the voids can be calculated using an equivalent projection area of a circle, DEQPC, as given in Eq. (3) [38]. DEQPC is described as the diameter of a circle that has an identical area as the selected air void regions. The size of air voids can be described using Feret’s dimensions, i.e., Maximum Feret’s diameter, Fmax, and minimum Feret, Fmin. Fmax and Fmin are given by the maximum and minimum distances between two pixels at the air void boundary as illustrated in Fig. 8. Other parameters such as number and percentage of air voids can be obtained from the ImageJ software.
Circularity ¼ 4
Aspect Ratio ¼ 2.3. Virtual cut section method The virtual cut sections were conducted in vertical and horizontal axes using the ImageJ plugin called Volume Viewer. Volume Viewer is a custom visual algorithm that offers excellent solution in interacting with image volume, meshes, and volume rendering in 3D visualization [40]. In this study, 18 vertical crosssectional images were obtained with an increment of 10° orientation angle as illustrated in Fig. 5. These cross-sectional views were obtained to evaluate the consistency of air void distribution within the sample. Fig. 6a shows the 3D reconstructed vertical cut section made by stacking the horizontal images. The dimension was 50 mm 100 mm as shown in Fig. 6b. The air voids were then thresholded and segmented for further analysis as shown in Fig. 7. Fig. 7a shows an X-ray image of vertical section obtained from the stack of images. The air voids were thresholded and verified in accordance with ASTM D 3203 [41] as shown in Fig. 7b. The image was then converted into binary images (black and white images) as presented in Fig. 7c for subsequent analysis on air void characteristics.
2.4. Air voids properties Air voids were analyzed in terms of content, number, shape, and size. The shape of the air voids can indicate the interconnectivity among them. The shape of air voids was described using circularity and aspect ratio as shown in Eqs. (1) and
Scan sample using X-ray CT scanner Image processing using ImageJ
DEQPC ¼
½Area ½Perimeter
2
½Length of major axis ½Length of minor axis
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 4 Area
p
ð1Þ
ð2Þ
ð3Þ
3. Result and discussion 3.1. Voids distribution (Percentage and Number) Void content and void number present in the porous asphalt mixture significantly influence the functional performance of pavement such as permeability, acoustic characteristics, and frictional properties. To analyze the void distribution, the air void content was initially determined in accordance with ASTM D3203 specification [41]. The total void content was obtained to be 19.5%. On the basis of the void content, the image threshold value was selected and later used to determine the void distribution throughout the sample. The analyses were compared between two different cross sections, i.e., horizontal and vertical cross sections.
Convert scanned images into vertical images Scanned images that are in the form of horizontal slices are converted into vertical images using volume viewer (ImageJ) at every 10° interval.
Combining stack images Images from the X-ray CT scan are combined in form of multiple stack (1024×1024 pixels, 16 bits).
Crop and stack images Images of the sample are cropped and stacked together.
Convert image Images that are in the form of 16 bits are converted to 8 bits.
Estimate air voids content Image histogram of the air voids area are summed and later compared to the actual experimental sample.
Crop and rescale images Images of the sample are cropped and rescaled to the actual dimension of the sample that are in mm.
Selecting air voids area (T0-T1) Air voids properties are selected upon the summation of threshold value.
Image enhancement Apply contrast enhancement and Gaussian filter (noise elimination function).
487
Proceed to analysis Air voids properties such as percentage, size, number, and shape are ready to be analysed.
Fig. 2. Framework on the vertical cut section analysis.
488
M.Z.H. Mahmud et al. / Construction and Building Materials 155 (2017) 485–494
Fig. 3. Image acquisition form horizontal cross sectional view (a) 3D reconstruction (b) stack of 2D images.
Low Contrast
(a)
Better Contrast
(b)
Fig. 4. X-ray images of porous asphalt (a) Original (b) After enhancement.
Fig. 5. Illustration of orientation angles on horizontal cross section.
Figs. 9 and 10 show the analysis of void content and void number on the horizontal and vertical cross sections, respectively. The void content and void number in Fig. 9 were plotted throughout the sample height at the interval of 1 mm height, and the vertical cross sections as shown in the polar diagram (Fig. 10) were
analyzed at the angle interval of 10°. The polar diagram shows the top view of the scanned sample with detailed vertical cross section. For example, the 0°–180° line represents an image of vertical cross section (virtual cut section). A total of 18 vertical cross sections were analyzed at the interval of 10° polar angle. Fig. 9 highlights the distributions of the void content and void number based on horizontal cross sectional analysis. The analysis provides the void distribution from top to bottom of the sample. Analyzing the material properties using this perspective is a common method used by many researchers in describing the distributions of void properties in asphalt mixtures [14,42–44]. The void content analysis in Fig. 9a shows that the mean air void content is 19.5% with the standard deviation of 1.38. The calculated mean number of voids is 11.7 with the standard deviation of 0.91 as shown in Fig. 9b. In general, the distributions of voids based on horizontal perspective shows that void content and void number are homogeneously distributed with slight deviation at the top and bottom of the sample because of the presence of high air void content. High percentage of voids can be observed at the top and bottom sections of the sample as a result of ‘‘confinement effect” during the compaction process. As shown in Fig. 9b, the void number is homogeneously distributed but is slightly reduced at the middle (at heights from 30 mm to 40 mm) of the sample. This finding is due to the presence of a large amount of coarse aggregates at the middle of the sample, thereby creating a large area of voids as a result of compaction. The second part of the analysis focuses on the radial distribution of voids in porous asphalt mixture presented in polar diagram as shown in Fig. 10. This method measures the distributions of voids within the sample using vertical cross sections at every 10°. Fig. 10a shows that the calculated mean air void content is
M.Z.H. Mahmud et al. / Construction and Building Materials 155 (2017) 485–494
489
Fig. 6. Vertical Cross Section View (a) 3D reconstruction (b) 2D Illustration.
Fig. 7. Image threshold process for the vertical cross section (a) X-ray image (b) thresholding (c) binary image.
3.2. Air voids shape properties
Fig. 8. Feret’s diameter and minimum Feret of an air voids.
19.48% with the standard deviation of 4.92. Fig. 10b shows that the mean number of voids distributed radially is 124 and the standard deviation is 10.84. In general, the void properties are homogeneously distributed within the sample. Comparing the two polar diagrams in Fig. 10a and b shows that, when the number of air voids present in the mixture is high, the percentage of air voids produced is small. These findings suggest that void content and void number exhibit an inverse relationship. For example, at plane 120°–300°, the percentage of air voids is 22.4 %, which is slightly higher than that of the mean air voids of 19.48%. On the same plane, the void number in Fig. 10b appears to experience depression with the value of 103 compared with the mean of 124. The reason is that large air voids converge to one another, thereby creating large size of air voids and thus producing less number of voids. Accordingly, the void connectivity increases.
Shape properties of air voids are significant parameters as they reflect the connectivity of the air voids [5,45]. Two main parameters were selected in this study, i.e., circularity and aspect ratio. ‘‘Circularity” describes whether the voids are in circular formation, while aspect ratio describes the elongation magnitude of the voids [5,38,42]. The formation of void connectivity in the porous asphalt can be described with high elongated voids and less circular shape, and the analyses were conducted for vertical and horizontal cut sections. Fig. 11 shows the distribution of void circularity and aspect ratio throughout the sample based on horizontal cut section analysis. Elongated air voids are observed at the bottom section of the sample unlike the top section. The above-mentioned parameters indicate that the size of the air voids at the bottom section is nearly three times the ratio of the minor axis detected. On the basis of the obtained mean value, the circularity and aspect ratio values are found to be 0.6 with standard deviations of 0.03 and 2.38 and 0.215, respectively, indicating void connectivity throughout the sample. Fig. 12 shows the analysis of void shape on the vertical cut sections and plotted at 10° interval. The polar diagram is an indicator for radial distribution of air void shape. Fig. 12 shows that the shape of air voids in terms of circularity and aspect ratio are generally well distributed with slight deviation. Fig. 12a shows a peak of 0.67 at 10°–190° plane of cross-section image. The mean circularity is 0.64 with the standard deviation of 0.015, and the mean value of aspect ratio is 2.13 with the standard deviation of 0.156. Therefore, the length of the major axis is approximately two times the length of the minor axis. Both shape parameters confirm that air voids within the porous asphalt mixture are prone to be elongated, thereby justifying the air void connectivity. Similarly, a study by Hassan et al. [5] also reveals that air voids in porous asphalt mixture are predominantly in elongated formation. From
490
M.Z.H. Mahmud et al. / Construction and Building Materials 155 (2017) 485–494
Voids, % 10
20
30
Number 40
5
50
5
5
10
10
Number
15
Mean
Percentage
15
Mean
Height, mm
Height, mm
10
15
0
0
20 25 30
20 25 30
35
35
40
40
45
45
50
50
(a)
(b)
Fig. 9. Analysis of (a) Voids contents (b) Voids number for horizontal cross section.
Percentage
Mean
Number
0̊ 340 ̊ 330 ̊ 320 ̊ 310 ̊
350 ̊ 25
10 ̊
0̊ 20 ̊ 30 ̊
330 ̊ 320 ̊ 310 ̊
40 ̊
20
50 ̊ 15
300 ̊
70 ̊
290 ̊
280 ̊
5
80 ̊
280 ̊
270 ̊
0
90 ̊
270 ̊
100 ̊
260 ̊
260 ̊ 250 ̊
110 ̊
240 ̊
120 ̊
230 ̊ 220 ̊ 210 ̊
130 ̊ 140 ̊ 150 ̊ 200 ̊
190 ̊
170 ̊
340 ̊
300 ̊
60 ̊
10
290 ̊
Mean
160 ̊
180 ̊
(a)
350 ̊ 160 140 120 100 80 60 40 20 0
10 ̊
20 ̊
30 ̊ 40 ̊ 50 ̊ 60 ̊ 70 ̊ 80 ̊ 90 ̊ 100 ̊
250 ̊
110 ̊
240 ̊ 230 ̊ 220 ̊ 210 ̊
120 ̊
200 ̊
190 ̊
170 ̊
160 ̊
130 ̊ 140 ̊ 150 ̊
180 ̊
(b)
Fig. 10. Analysis of (a) Voids contents (b) Voids number for vertical cross section.
Figs. 11 and 12, the voids are found to be connected vertically and horizontally. In addition to the vertical cross-section analysis, the vertical cut section was equally divided into top, middle, and bottom in assessing the void shape properties within the said regions. This procedure was conducted to assess the concentration of void connectivity throughout the sample and the homogeneous distribution of voids within the sample. The image of vertical cut section divided into top, middle, and bottom regions is shown in Fig. 13. Fig. 14 shows the detailed distribution of circularity and aspect ratio values for the top, middle, and bottom regions. For comparison, the distribution within the different regions was compared with the ‘‘total” as presented in the polar diagram. In general, the shape of air voids using circularity and aspect ratio parameters is evenly distributed at different heights of the sample. Fig. 14a shows the distribution of air voids on the basis of the circular parameter. As the value approaches 1, the air voids are in perfect circular shape. Based on the result, it was found that approximately about 34% of the air voids is in the form of perfect circle
that falls in the range of 0.8–1.0. On the other hand, the presence of perfect elongated air voids (0–0.2) was found approximately 6%. Although the result in Fig. 14a shows that only 34% air voids are in the form of a perfect circle, the remaining air voids shape is not in a form of a perfect circular formation. Fig. 14b shows the analysis on the aspect ratio, which reflects the magnitude of void elongation. More than 50% of the air voids are in the range of 1–2. Therefore, most size of the air voids is approximately two times the size of the identified minor axis. The remaining 42% of air voids present an aspect ratio of greater than 2, indicating the void connectivity within the compacted sample. These finding shows the air voids are predominantly elongated and evenly distributed radially and vertically. One of the main reasons that caused the formations of the air voids to be predominately elongated is due to the aggregates interlocking process that occurs within the aggregate skeleton and later, influenced the shaped of the air voids. Masad and Button [12] discussed on the selection of aggregate gradation based on Bailey method. When large aggregates (coarse aggregates) interlocks with other large aggregate
491
M.Z.H. Mahmud et al. / Construction and Building Materials 155 (2017) 485–494
Circularity 0.45
0.5
0.55
Aspect Rao
0.6
0.65
1
0.7
0
0
5
5
10
15
Height, mm
Height, mm
3
4
10
Top
15 20 25 30
Top
20 25 30 35
35 40
2
40
Bottom
Circularity
Aspect rao
Bottom
45
45 Mean
50
Mean
50
(a)
(b)
Fig. 11. Sample height diagram on voids shape properties (a) Circularity (b) Aspect Ratio.
Circularity
Aspect Rao
Mean
340 ̊ 330 ̊ 320 ̊ 310 ̊
350 ̊ 0.70
10 ̊
40 ̊ 50 ̊
70 ̊
0.50 0.45
270 ̊
0.40
260 ̊ 250 ̊
230 ̊ 220 ̊ 210 ̊ 200 ̊
190 ̊
170 ̊
90 ̊
270 ̊
0.0
100 ̊
260 ̊
160 ̊
30 ̊ 40 ̊ 50 ̊ 60 ̊ 70 ̊
1.0 0.5
130 ̊ 140 ̊ 150 ̊
20 ̊
1.5
290 ̊ 280 ̊
120 ̊
10 ̊
2.0
80 ̊
110 ̊
240 ̊
350 ̊ 3.0 2.5
300 ̊
60 ̊
0.55
280 ̊
330 ̊ 320 ̊ 310 ̊
30 ̊
0.60
290 ̊
340 ̊
20 ̊
0.65
300 ̊
Mean 0̊
0̊
80 ̊ 90 ̊ 100 ̊
250 ̊
110 ̊
240 ̊
120 ̊
230 ̊ 220 ̊ 210 ̊
130 ̊ 140 ̊ 150 ̊ 200 ̊
190 ̊
170 ̊
160 ̊
180 ̊
180 ̊
(a)
(b)
Fig. 12. Polar diagram of the vertical cut sections on voids shape properties (a) Circularity (b) Aspect Ratio.
3.3. Voids diameter
Top Middle Bottom Fig. 13. Vertical cut section divided into top, middle and bottom regions.
particles, it will create voids where these voids will be filled by fine aggregates. For porous asphalt, large voids present due to lack of fine aggregate in the asphalt mixture gradation. In this study, these parameters (i.e., circularity and aspect ratio) show that the air voids shape are predominantly elongated.
Void diameter was analyzed using DEQPC and Fmax. The DEQPC parameter measures the equivalent projection area of a circle. While, Fmax parameter measures the maximum distance between two pixels at the boundary of the air voids. Hassan et al. [38] used these two parameters in estimating the dimension of the aggregate gradations analysis and successfully described the dimensions of the coarse aggregate particles. This study applies these parameters in describing the air voids dimension. Fig. 15 compares both parameters analyzed using horizontal and vertical cut sections. As shown in Fig. 15a, the means of DEQPC and Fmax are approximately 1.5 and 2.5 mm, respectively. Fig. 15b shows the means of DEQPC and Fmax from horizontal cut sections are approximately 3.1 and 4.1 mm, respectively. Comparing both findings suggests that the diameter analyzed using vertical cut sections provides a
492
M.Z.H. Mahmud et al. / Construction and Building Materials 155 (2017) 485–494
Total
Top
Middle
Boom
34.64
35
Percentage. %
30 22.59
25 19.78
20
16.83
15 10
6.16
5 0 0.0-0.2
0.4-0.6
0.2-0.4
0.6-0.8
0.8-1.0
Circularity
(a) Total
Percentage. %
60
Top
Middle
Boom
54.78
50 40 28.75
30 20 10
9.20 4.73
2.54
0 0-1
1-2
2-3
3-4
>4
Aspect Rao
(b) Fig. 14. Distribution of voids shape properties in different regions based on vertical cut sections (a) Circularity (b) Aspect ratio.
F max
Diameter, mm 0
1
2
3
4
Mean F max
D EQPC
5
0
Mean D EQPC 0̊
340 ̊
350 ̊ 6
330 ̊ 320 ̊ 310 ̊
5 10
Height, mm
25 30
40 ̊ 50 ̊ 60 ̊
3
70 ̊
2
F max
280 ̊
1
Mean F max
270 ̊
0
D EQPC
30 ̊
4
290 ̊
20
20 ̊
5
300 ̊
15
10 ̊
80 ̊ 90 ̊
260 ̊
100 ̊
250 ̊
35 Mean D EQPC 40
110 ̊
240 ̊
120 ̊
230 ̊ 220 ̊ 210 ̊
45
130 ̊ 140 ̊ 150 ̊ 200 ̊
50
190 ̊
170 ̊
160 ̊
180 ̊
(a)
(b)
Fig. 15. Voids diameter using Feret’s diameter and Diameter EQPC method analyzed on (a) horizontal (b) vertical cut sections.
1 mm larger mean diameter than the one analyzed using horizontal cut sections. Therefore, the vertical cut sections complement the conventional analysis using horizontal cut sections. Similar pattern of distribution (homogeneous) in radial direction is shown in Fig. 15b for DEQPC and Fmax. Large air void diameter is visible at a plane of 120°–300° with the Fmax size of more than 5.1 mm. However, Fig. 15a shows that Fmax is more sensitive than DEQPC in
detecting changes in air void diameter and the connectivity of voids. For detailed size distribution, Fig. 16 categorizes void diameter (DEPC and Fmax) in different ranges according to the standard sieve size. The analysis was conducted for the top, middle, and bottom regions. As discussed earlier, Fmax is a good parameter for detecting changes in void size; this deduction is supported by the results
493
M.Z.H. Mahmud et al. / Construction and Building Materials 155 (2017) 485–494
Diameter, DEQPC
Total
Top
Middle
Boom
Percentage. %
30 25 20 15 10 5 0 0.15-0.30
0.3-0.60
0.60-1.18
1.18-2.36
2.36-5.00
5-10
10-14
14-20
Size, mm
(a) 30
Feret's Diameter, Fmax
Percentage. %
25 20 15 10 5 0 0.15-0.30 0.3-0.60 0.60-1.18 1.18-2.36 2.36-5.00
5-10
10-14
14-20
20-28
> 28
Size, mm
(b) Fig. 16. Distributions of voids diameter in different regions based on vertical cut sections (a) Diameter EQPC (b) Feret’s Diameter.
shown in Fig. 16. Fmax detects a larger range of air void size of up to 28 mm diameter than DEQPC with only up to 20 mm diameter. The analysis performed with DEQPC also shows that the majority of the air void sizes (approximately 28%) are in the range of 1.18– 2.36 mm. Notably, the void size is evenly distributed from the top to bottom of the sample. However, the analysis using Fmax shows large void size with the majority of air voids detected in the range of 1.18–5.00 mm. A small void size, i.e., 0.15–0.30 mm, appears to be significantly lesser when analyzed using Fmax than using DEQPC. The finding shows the analysis of voids at the top, middle, and bottom of the sample indicates that most of the small voids (<1.18 mm) are located at the top section and large voids (>2.36 mm) are slightly higher at the bottom of the sample than at the other sections. This is because the influence of aggregate particle rearrangement in the formation of air voids particularly in the process of compacting asphalt mixture. In this process, blended coarse aggregates tend to settle into the gyratory compaction mold due to gravitational force and the small aggregates tends to stick on the surface of the mixing mold [39]. As a result, the formation of large size air voids are at the bottom of the sample where the smaller size air voids are located at the top of the sample.
4. Conclusions This study introduced a research framework for performing virtual cut section analysis on porous asphalt mixture using X-ray CT images. By using virtual cut sections, this method is able to explore the internal structure of the asphalt mixture by regenerating new multiple cross-sectional images at various angles without disrupting the physical properties of the sample. For this study, the outcome of using virtual cut section in porous asphalt mixture enables the radial distribution of air voids properties to be analyzed. This analysis is presented in the form of polar diagram. This
diagram is used to complement the conventional cross-section analysis to provide a complete view (vertical and horizontal views) of the distribution of air void properties. As a result, more data can be extracted from the analysis particularly from the radial distribution perspective. Based on the analysis performed on the air voids properties, it shows that this method is effective in analyzing the radial distributions of void connectivity of the porous asphalt mixture sample and complements the conventional cross-section analysis (horizontal). The findings indicate that both of the cross-section images show that the air voids are homogeneously distributed. The result also shows that percentage of air voids content and number of voids have an inversed relationship. Study on circularity and aspect ratio confirms that the shape of the air voids in the porous asphalt mixture is predominantly elongated. By using Fmax parameter in analyzing the dimension of the sample, the measurement is considered more accurate whereby it gives approximately 1 mm larger compared to DEQPC parameter. Hence, this study recommends the use of Fmax parameter in analyzing the dimension of air voids in porous asphalt mixture. Acknowledgments Universiti Teknologi Malaysia and Malaysia Ministry of Education are sincerely acknowledged for providing the research funds (Vote no. Q.J130000.2522.11H76 and R.J130000.7822.4F867) and laboratory facilities. References [1] A.E. Alvarez, A.E. Martin, C. Estakhri, A review of mix design and evaluation research for permeable friction course mixtures, Constr. Build. Mater. 25 (3) (2011) 1159–1166. [2] R.B. Mallick, P. Kandhal, L.A. Cooley, D. Watson, Design, construction, and performance of new-generation open-graded friction courses, Asphalt Paving Technol. 69 (2000) 391–423.
494
M.Z.H. Mahmud et al. / Construction and Building Materials 155 (2017) 485–494
[3] B.J. Putman, L.C. Kline, Comparison of mix design methods for porous asphalt mixtures, J. Mater. Civ. Eng. 24 (11) (2012) 1359–1367. [4] G. Huber, Performance Survey on Open-Graded Friction Course Mixes, Synthesis of Highway Practice 284, National Cooperative Highway Research Program, Transportation Research Board, National Research Council, Washington, DC, 2000. [5] N.A. Hassan, M.Z.H. Mahmud, R.P. Jaya, Air void characterisation in porous asphalt using X-ray computed tomography, Adv. Mater. Res. 2014 (911) (2014) 443–448. [6] N.A. Hassan, N.A.M. Abdullah, N.A.M. Shukry, M.Z.H. Mahmud, N.Z.M. Yunus, R. P. Jaya, M.R. Hainin, H. Yaacob, Laboratory evaluation on the effect of clogging on permeability of porous asphalt mixtures, J. Teknol. 76 (14) (2015) 77–84. [7] A.E. Alvarez, A.E. Martin, C.K. Estakhri, J.W. Button, C.J. Glover, S.H. Jung, Synthesis of Current Practice on the Design, Construction, and Maintenance of Porous Friction Courses (Report No. FHWA/TX-06/0-5262-1), U.S. Department of Transportation, Federal Highway Administration, 2006. [8] T. Fwa, S. Tan, Y. Guwe, Laboratory evaluation of clogging potential of porous asphalt mixtures, Transp. Res. Rec. 1681 (1999) 43–49. [9] W. Jiang, A. Sha, J. Xiao, Experimental study on relationships among composition, microscopic void features, and performance of porous asphalt concrete, J. Mater. Civ. Eng. 27 (11) (2015). [10] P.J. Vardanega, State of the art: permeability of asphalt concrete, J. Mater. Civ. Eng. 26 (1) (2012) 54–64. [11] E. Arambula, E. Masad, A.E. Martin, Influence of air void distribution on the moisture susceptibility of asphalt mixes, J. Mater. Civ. Eng. 19 (8) (2007) 655– 664. [12] E. Masad, J. Button, Implications of experimental measurements and analyses of the internal structure of hot-mix asphalt, Transp. Res. Rec. 1891 (2004) 212– 220. [13] N.A. Hassan, G.D. Airey, M.R. Hainin, Characterisation of micro-structural damage in asphalt mixtures using image analysis, Constr. Build. Mater. 54 (2014) 27–38. [14] N.A. Hassan, M.Z.H. Mahmud, K.A. Ahmad, M.R. Hainin, R.P. Jaya, N. Mashros, Air voids characterisation and permeability of porous asphalt gradations used in different countries, ARPN J. Eng. Appl. Sci. 11 (24) (2016) 14043–14047. [15] N. Neithalath, M.S. Sumanasooriya, O. Deo, Characterizing pore volume, sizes, and connectivity in pervious concretes for permeability prediction, Mater. Charact. 61 (8) (2010) 802–813. [16] A.E. Alvarez, A.E. Martin, C. Estakhri, R. Izzo, Determination of volumetric properties for permeable friction course mixtures, J. Test. Eval. 37 (1) (2009) 1. [17] E. Brown, M.R. Hainin, L.A. Cooley Jr, Determining minimum lift thickness for hot mix asphalt (HMA) mixtures (with discussion), J. Assoc. Asphalt Paving Technol. 74 (2005). [18] L.A. Cooley Jr, B.D. Prowell, M.R. Hainin, Comparison of the saturated surfacedry and vacuum sealing methods for determining the bulk specific gravity of compacted HMA (with discussion and closure), J. Assoc. Asphalt Paving Technol. 72 (2003). [19] O. Deo, M. Sumanasooriya, N. Neithalath, Permeability reduction in pervious concretes due to clogging: experiments and modeling, J. Mater. Civ. Eng. 22 (7) (2010) 741–751. [20] M.R. Hainin, L.A. Cooley Jr, B.D. Prowell, An investigation of factors influencing permeability of superpave mixes, Int. J. Pavements 2 (2) (2003) 41–52. [21] M.R. Hainin, N.I.M. Yusoff, M.K.I.M. Satar, E.R. Brown, The effect of lift thickness on permeability and the time available for compaction of hot mix asphalt pavement under tropical climate condition, Constr. Build. Mater. 48 (2013) 315–324. [22] M.R.M. Hasan, M.O. Hamzah, M. van de Ven, J. Voskuilen, Disruption of air voids continuity based on permeability loss due to mortar creep, Constr. Build. Mater. 116 (2016) 347–354.
[23] D. Watson, K. Moore, K. Williams, L. Cooley, Refinement of new-generation open-graded friction course mix design, Transp. Res. Rec. 1832 (2003) 78–85. [24] D. Chen, N. Roohi Sefidmazgi, H. Bahia, Exploring the feasibility of evaluating asphalt pavement surface macro-texture using image-based texture analysis method, Road Mater. Pavement Des. 16 (2) (2015) 405–420. [25] E. Masad, B. Muhunthan, N. Shashidhar, T. Harman, Internal structure characterization of asphalt concrete using image analysis, J. Comput. Civ. Eng. 13 (2) (1999) 88–95. [26] E. Masad, V.K. Jandhyala, N. Dasgupta, N. Somadevan, N. Shashidhar, Characterization of air void distribution in asphalt mixes using X-ray computed tomography, J. Mater. Civ. Eng. 14 (2) (2002) 122–129. [27] L. Tashman, E. Masad, J. D’Angelo, J. Bukowski, T. Harman, X-ray tomography to characterize air void distribution in superpave gyratory compacted specimens, Int. J. Pavement Eng. 3 (1) (2012) 19–28. [28] Z.Q. Yue, I. Morin, Digital image processing for aggregate orientation in asphalt concrete mixtures, Can. J. Civ. Eng. 23 (1996) 480–489. [29] A.E. Alvarez, A.E. Martin, C. Estakhri, Internal structure of compacted permeable friction course mixtures, Constr. Build. Mater. 24 (6) (2010) 1027–1035. [30] J.B. Król, R. Khan, A.C. Collop, The study of the effect of internal structure on permeability of porous asphalt, Road Mater. Pavement Des. (2017) 1–17. [31] A. Bozorgzad, Consistent distribution of air voids and asphalt and random orientation of aggregates by flipping specimens during gyratory compaction process, Constr. Build. Mater. 132 (2017) 376–382. [32] Z. You, S. Adhikari, M.E. Kutay, Dynamic modulus simulation of the asphalt concrete using the X-ray computed tomography images, Mater. Struct. 42 (5) (2009) 617–630. [33] P.W.D. Malaysia, Standard Specification for Road Works, Section 4, Flexible Pavement, Jabatan Kerja Raya Malaysia, Kuala Lumpur, 2008. [34] ASTM D 7064, Standard Practice for Open-Graded Friction Course (OGFC) Mix Design, ASTM International, West Conshochocken, PA, 2013. [35] M. Buchanan, E. Brown, Effect of superpave gyratory compactor type on compacted hot-mix asphalt density, Transp. Res. Rec. 1761 (2001) 50–60. [36] E. Brown, J. Haddock, Method to ensure stone-on-stone contact in stone matrix asphalt paving mixtures, Transp. Res. Rec. 1583 (1997) 11–18. [37] Australian Asphalt Pavement Association (AAPA), National Asphalt Specification, Australian Asphalt Association, 2004. [38] N.A. Hassan, Microstructural Characterisation of Rubber Modified Asphalt Mixtures (Doctoral dissertation), University of Nottingham, UK, 2012. [39] N.A. Hassan, G.D. Airey, R. Khan, A.C. Collop, Nondestructive characterisation of the effect of asphalt mixture compaction on aggregate orientation and segregation using X-ray computed tomography, Int. J. Pavement Res. Technol. 5 (2) (2012) 84–92. [40] K.U. Barthel, 3D-data representation with ImageJ, in: ImageJ Conference, 2006. [41] ASTM D 3203, Standard Test Method for Percent Air Voids in Compacted Dense and Open Bituminous Paving Mixtures, ASTM International, West Conshohocken, PA, 2011. [42] M. Aboufoul, A. Garcia, Influence of air voids characteristics on the hydraulic conductivity of asphalt mixture, Road Mater. Pavement Des. (2017) 1–11. [43] W.D. Martin, B.J. Putman, N.B. Kaye, Using image analysis to measure the porosity distribution of a porous pavement, Constr. Build. Mater. 48 (2013) 210–217. [44] D. Lo Presti, R. Khan, N.A. Hassan, G. Airey, A. Collop, Laboratory mix design of asphalt mixture containing reclaimed material, Adv. Mater. Sci. Eng. 2014 (2014). [45] W.S. Abdullah, M.T. Obaidat, N.M. Abu-Sa’da, Influence of aggregate type and gradation on voids of asphalt concrete pavements, J. Mater. Civ. Eng. 10 (2) (1998) 76–85.