Evaluation of different digital image processing software for aggregates and hot mix asphalt characterizations

Evaluation of different digital image processing software for aggregates and hot mix asphalt characterizations

Construction and Building Materials 37 (2012) 370–378 Contents lists available at SciVerse ScienceDirect Construction and Building Materials journal...

1MB Sizes 0 Downloads 49 Views

Construction and Building Materials 37 (2012) 370–378

Contents lists available at SciVerse ScienceDirect

Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat

Evaluation of different digital image processing software for aggregates and hot mix asphalt characterizations Iuri S. Bessa ⇑, Verônica T.F. Castelo Branco, Jorge B. Soares Universidade Federal do Ceará (UFC), Laboratório de Mecânica dos Pavimentos (LMP), Campus do Pici S/N, Bloco 703, 60455-760 Fortaleza, Ceará, Brazil

h i g h l i g h t s " Different aggregates (natural and alternatives) were evaluated using digital image processing. " Different HMAs composed by those aggregates were also investigated. " Aggregates were analyzed in respect to roundness and flat and elongated particles percentage. " Mixtures were analyzed in respect to aggregates particles segregation, orientation and contact points. " The use of image software leads to more complete and precise results even for alternative aggregates.

a r t i c l e

i n f o

Article history: Received 1 December 2011 Received in revised form 13 July 2012 Accepted 23 July 2012 Available online 5 September 2012 Keywords: Digital image processing Aggregate shape Asphalt mixes internal structure Alternative aggregates

a b s t r a c t Asphalt pavement performance is related to design and construction of durable layers, which require adequate materials selection. Several studies have been developed on characterizing aggregates and hot mix asphalt (HMA) internal structure through digital image processing (DIP). This paper aimed to characterize three different aggregates: granitic, steel slag, and construction and demolition waste (CDW), and HMA internal structure composed by those aggregates with different gradations. The aggregates were also evaluated with respect to flat and elongated particles percentage and roundness. The mixes were analyzed with respect to the number of contact points between aggregates, particles orientation and segregation potential. The results for different software show that the use of DIP leads to more complete and accurate results. The alternative materials investigated performed well in terms of Superpave specifications and HMA mechanical characterization. Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction Hot mixes asphalt (HMA) is a composite constituted by binder, aggregates and air voids. As a heterogeneous material, HMA behavior is influenced by mechanical and shape properties of its aggregates, besides geometric properties such as position and orientation [1,2]. Since coarse aggregates play a fundamental role on pavement stability and load support capacity, these materials should be characterized with respect to: (i) form, (ii) angularity, and (iii) texture. Superpave methodology (e.g. the use of caliper for measuring the particles’ sizes, and the visual analysis of aggregates particles to measure their angularity) includes procedures for aggregates shape characterization, but these methods present disadvantages: (i) lack of connection between coarse and fine aggregates proper-

⇑ Corresponding author. Tel.: +55 853366 9488x249; fax: +55 853366 9488x250. E-mail addresses: [email protected] (I.S. Bessa), [email protected] (V.T.F. Castelo Branco), [email protected] (J.B. Soares). 0950-0618/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.conbuildmat.2012.07.051

ties, (ii) results influenced by more than one property, (iii) qualitative and time demanding tests, and (iv) aggregate properties obtained using indirect measurements [2,3]. HMA plays an important role on pavement’s resistance, in relation to its major distresses, i.e., permanent deformation, fatigue cracking, and thermal cracking [4–6]. The compaction methodology and effort dictate aggregate particles arrangement within an asphalt mix. The overall aggregate distribution is a net result of each aggregate distribution and the existing number of contact points among them [7]. Aggregate shape characteristics can be obtained using simple equipment, with each particle being individually evaluated. Aside from being time consuming, such tests results determine a global index based on averages, instead of using cumulative properties distributions [8]. With computers and software advances, images can be processed and analyzed. Digital image processing (DIP) techniques have been used to characterize several materials, including HMA internal structure. Recent equipment and software improvements have provided opportunities to significant advance on HMA design and characterization. There are several types of DIP

I.S. Bessa et al. / Construction and Building Materials 37 (2012) 370–378

methodologies, using equipment such as the aggregate measurement image system (AIMS) and the X-ray tomography. This work aims to analyze different aggregates (granitic, construction and demolition waste – CDW, and steel slag), using a planar DIP technique, in relation to some of its shape properties (percentage of flat and elongated particles, and angularity). The technique requires simple equipment, besides being fast. It also aims to characterize different HMAs (dense-graded and stone matrix asphalt – SMA) internal structure: (i) contact points; (ii) segregation potential, and (iii) particle orientation. Two different software were used: (i) ImageTool and (ii) Digital Image Analysis System. The dense-mixes were produced using the three distinct aggregates. The SMA mix was produced using only with granitic aggregates. The mixes were compacted using: the Superpave gyratory compactor (SGC), and a slab compactor, developed by the Institut français des sciences et technologies des transports, de l’aménagement et des réseaux (IFSTTAR). 2. Aggregate shape properties and HMA internal structure characterization 2.1. Aggregate shape properties Aggregates in asphalt mixes represent approximately 90–95%, by weight. According to Superpave specifications, their properties were classified as consensus (angularity for coarse and fine aggregates, amount of flat and elongated particles and plastic fines) and source (abrasion resistance, sanity and deleterious materials determination) [9]. In order to determine if the particles are flat or elongated (ASTM D 4791/99) two ratios need to be observed: (i) aggregate length to width, and (ii) aggregate width to thickness. This procedure requires a special caliper that provides the mentioned ratios for each particle. Flat and/or elongated aggregates in asphalt mixes can break decreasing workability during compaction, and under traffic. Superpave specifications suggest aggregate shape evaluation with respect to its flat and elongated particles using the ratio 5:1. Up to 10% of flat/elongated particles is admitted for traffic volume from medium to high (Equivalent Single Axle Loads (ESALs) > 3  106). Angularity is associated to the level of variations at the aggregates’ corners. This shape parameter is related to the superficial texture and can guarantee a high level of internal friction between aggregates, and influence HMA rutting resistance. Manual observation of each particle is required to verify the existence of fractured faces. However, this test does not evaluate angularity level (sharpness) or texture [10]. Superpave specifications indicate the minimum value for fractured faces according to the traffic or to the proximity of the layer to the surface. 2.2. HMA internal structure The occurrence of a contact point between two aggregates is established when they are separated from each other at a fixed maximum distance. Fig. 1a shows a scheme of a contact between two aggregates particles. There is a relationship between HMA strength resistance and the number of contact points. Tashman et al. [11] concluded that the number of contacts points between aggregates depends on the parameters (angle, pressure, temperature and height) used in the compaction. Sousa et al. [12] showed that samples produced using the kneading compactor have higher number of contact points and higher resistance to permanent deformation, when compared to those produced by other compaction methods. The orientation of an aggregate particle is the angle between the particle major axis and the horizontal line on a digitalized im-

371

age (Fig. 1b). This axis is defined as the largest distance between two points in the particle contour [1,11–14]. It is possible to quantify the aggregate directional distribution using the vector magnitude (D), according to Eq. (1).



  qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 100 2  ðR sin hk Þ þ ðR cos hk Þ2 N

ð1Þ

where D: vector magnitude; hk: angle between the aggregate particle and the horizontal line and N: number of aggregates particles analyzed in the image. The vector magnitude describes the aggregate particles preferential orientation. It varies from 0% to 100%. 0% represents a completely random particles distribution. 100% shows that aggregates are aligned exactly in the same direction [1,15]. According to Tashman et al. [16], anisotropy is caused by the microstructure constituents’ non-uniform distribution. Anisotropic materials have their mechanical properties depending on its particles direction. Masad et al. [1] verified that increasing the compaction effort (up to 100 gyrations), there is a tendency to form a more uniform orientation. For numbers of gyrations higher than 100, the aggregate structure presents a more random orientation. HMA internal structure for samples compacted using the SGC and extracted from the field were compared. Those authors concluded that, to achieve the same orientation level existing in the field, a higher number of gyrations was necessary. The air void content found for the field samples was obtained in the laboratory with 25 gyrations, but the vector magnitude was obtained with 100 gyrations. Hunter et al. [17] concluded that the particles orientation must be induced by the confining boundary conditions. Findings from the same study showed that samples produced using slab compactors lead to aggregates with a smaller orientation angle. This results in lower resistance to permanent deformation, when compared to other compaction methods, such as gyratory or vibratory. Zhang et al. [18] showed that SGC samples have a more uniform orientation, or, higher vector magnitude, in comparison to vibratory compactor samples. HMA segregation can be defined as a lack of homogeneity between its constituents in a certain level that can accelerate pavement distresses [19]. Fig. 1c shows a transversal specimen section, with three segregation groups. Ideally, a dense-mix must have a uniform aggregates distribution. Gyratory and vibratory compaction methods induce to a higher segregation level when compared to the plate fabrication method [17]. Masad et al. [1] evaluated aggregates segregation and verified that field samples are more prone to segregation if compared to laboratory ones. Tashman et al. [11] calculated aggregates lateral segregation, or radial distribution, and concluded that coarse aggregates tend to be located on the external area of the SGC specimen horizontal section. Hunter et al. [17] concluded that the vibratory compaction induces particles migration to the specimen boundaries, while slab compaction may induce to a lower segregation level.

2.3. Digital image processing (DIP) techniques The difficulties faced during aggregates and HMA internal structure characterizations have encouraged the DIP use. Different DIP methods have been used in the past few years. The aggregate imaging measurement system (AIMS), equipment distributed by Pine Instrument Company, can be used to capture and to analyze aggregates with different sizes. The X-ray tomography is a non-destructive technique for 3D analysis [20]. The simplest DIP technique involves the digitalization of a real planar image and the analysis of its components to obtain parameters related to aggregates shape

372

I.S. Bessa et al. / Construction and Building Materials 37 (2012) 370–378

Fig. 1. (a) Contact between aggregates, (b) aggregate particles orientation, and (c) segregation groups.

and orientation. Some researchers used this technique for pavement materials characterization [1,7,14,21,22]. There are several software such as: ImageTool (developed by the Department of Dental Diagnostic Science at the University of Texas Health Science Center), ImageJ (developed by the National Institutes of Health), the Digital Image Analysis System (developed by the University of Wisconsin-Madison) and Abaqus 2008, that can be used to perform image analyzes. The process using image software requires the digitalization of a real image and the analysis of its components to obtain parameters related to aggregates shape and orientation. There are four steps involved in DIP: (i) digitalization, which transforms a real image into a digital one, (ii) enhancement, when the aggregates get to be more clearly differentiated one from another detecting aggregates boundaries, (iii) restoration, performed to correct the imperfections that may appear during the first process, and (iv) segmentation, an identification procedure that labels the regions or objects inside an image, detecting its edges. The last step allows the image subdivision into smaller parts, so that they can be identified

and treated separately [23]. The subdivision level must be defined according to the objects’ colors supposed to be characterized [24]. If an image can be identified it is possible to determine properties such as: (i) area, (ii) perimeter, (iii) orientation, and (iv) shape. DIP allows easy access to aggregate shape characteristics, leading to more precise and realistic parameters determination [25]. Aggregate sphericity and orientation can be determined. Sphericity can be measured through a parameter called roundness (Eq. (2)) that ranges from 0 to 1. A value closer to 1 indicates a more circular object. If the result of the analyses is 1, the object would resemble a perfect circle.



4pA P2

ð2Þ

where r: roundness; A: surface area; and P: perimeter of an isolated aggregate particle. The aggregates’ arrangement, directional distributions, and contact points, forms the mineral skeleton. These distributions result

Fig. 2. Gradation curves and specifications limit.

373

I.S. Bessa et al. / Construction and Building Materials 37 (2012) 370–378 Table 1 Mix design parameters (100 gyrations). Parameters/mix

HMA

Binder (%) Voids (%) Gmm

SMA

Mix I (100% granitic)

Mix II (50% CDW)

Mix III (50% steel slag)

Mix IV [29] (100% granitic)

6.0 4.0 2.400

6.9 4.0 2.322

6.6 4.0 2.732

7.2 3.2 2.323

Table 2 Volumetric parameters: slab specimens. Parameters/ mix

HMA Mix I (100% granitic)

Mix II (50% CDW)

Mix III (50% steel slag)

Voids (%) CV (%)

7.4 5.2

2.7 11.4

3.6 17.1

from the materials proportions and the compaction method. HMA microstructure has been associated to its mechanical properties [2,7]. The Digital Image Analysis System can be used to determine aggregate orientation, segregation level, and contact points in an asphalt mix. The sample preparation procedure involves sawing a compacted specimen. The specimen faces are scanned in 2D. Image treatment and transformation processes are performed, leading to the parameters for the analysis [26]. Similar procedure can be use to determine aggregate orientation with the use of ImageTool.

3. Material and methods 3.1. Aggregates and binder The aggregates used were: (i) granitic; (ii) CDW; and (iii) steel slag. Aggregate characterization was performed using Brazilian standard procedures. One of the tests performed was the shape index (as described on the standard DNER-ME 086/94), which results into a value the ranges from 0 to 1. A value of 1 means that the aggregate particle is a perfect cube. The results were 0.66 for granitic, 0.85 for CDW and 0.80 for steel slag. The binder used was a conventional non-modified asphalt cement (AC) produced by Petrobras, classified by penetration as an AC 50/70 [27].

3.2. HMA design HMAs were designed following the Superpave methodology with 100 gyrations. Mixes were also compacted using the LCPC compactor. The SMA mix was compacted only in the SGC. The dense-graded mixes were compacted with different aggregates: (i) mix I – granitic aggregates, (ii) mix II – 50% of CDW and 50% of granitic aggregates, and (iii) mix III – 50% of steel slag and 50% of granitic aggregates. Mix IV, a ½’’ SMA, was composed only by granitic aggregates. Fig. 2 shows the mixes gradation curves. To produce the specimens, the binder was heated at 165 °C and the aggregates at 175 °C. The mixes were short term aged for 2 h at 150 °C. With Superpave design results, slabs (500  180  50 mm) were produced in the LCPC compactor. Slabs compaction was done following French specifications [28]. Table 1 shows the mixes design parameters. Cylindrical specimens were extracted from the central part of the slabs. Table 2 presents the volumetric parameters (mean average and Coefficient of Variation – CV) for 12 specimens, for each mix. Besides the volumetric parameters, HMA were characterized using resilient modulus (RM) and indirect

Fig. 3. Aggregates set to be analyzed: (a) granitic, (b) steel slag, and (c) construction and demolition waste (CDW).

tensile strength (ITS) tests. The procedures for these tests are described, respectively, on Brazilian standards, DNER-ME 133/94 and DNER-ME 138/94. Table 3 presents the results obtained for each mix.

Table 3 Mechanical test results. Properties/mix

Superpave giratory compactor (SGC)

LCPC compactor

HMA

RM (MPa) ITS (MPa)

SMA [29]

HMA

Mix I

Mix II

Mix III

Mix IV

Mix I

Mix II

Mix III

2224 0.83

2935 0.99

2241 0.84

3790 1.11

2683 0.94

2922 1.01

2876 1.05

374

I.S. Bessa et al. / Construction and Building Materials 37 (2012) 370–378

Fig. 4. Cuts on the specimens for the analyses – cut AA (horizontal) and cut BB (vertical): (a) scheme, and (b) pictures of the specimens. 3.3. Aggregates shape characterization The percentage of flat/elongated particles was determined for coarse aggregates [30]. According to Superpave, an aggregate particle is considered flat or elongated when its greater dimension is, at least, five times bigger than its smaller one. The angularity test was performed according to ASTM D 5821/01 [31]. 3.4. Digital image processing (DIP) procedures 3.4.1. Aggregates characterization Parameters related to aggregate shape (percentage of flat/elongated particles and angularity) were also obtained using DIP techniques. A picture of the aggregates disposed on a plane surface was taken by and analyzed. The process of taking the pictures was done by using a five mega pixels portable digital camera, and there was no specific lighting consideration through the process. The pictures were transformed into gray scale images and passed through: enhancement, restoration, segmentation, and border detection. The objects of the images were identified and analyzed using ImageTool software. Aggregates were disposed in a way such that the bigger and the smaller face of each particle could be obtained (Fig. 3). The particles number was about 77, approximately the same amount used during AIMS analysis. 3.4.2. HMA internal structure characterization HMA internal structure was analyzed in respect to: (i) contact points; (ii) particles orientation; and (iii) radial segregation. The specimens were sawed in two locations: (i) at one third from the top and (ii) at one third from the bottom, as shown in Fig. 4. Two compaction methods (IFSTTAR compactor and SGC) and directions (horizontal and vertical) were evaluated. After sawing, each specimen face was digitalized, and the DIP procedure was applied (Fig. 5). Two different software, ImageTool and Digital Image Analysis System, were used. To use the Digital Image Analysis System a few steps had to be followed according to [26]. First, the lower size of the visible aggregate on the image was identified. After that, some image filters were used to remove imperfections and to decrease the variation of the pixels intensity (providing a uniform grey scale image). As the third and fourth steps, the image passes through a boundary detection process, where aggregates could be defined inside the image, and it was transformed into a black and white picture, to differentiate aggregate from mastic. Results were automatically generated. For the contact points, the software provided the number of times that it could be identified any aggregate–aggregate contact – when two aggregates were separated by 0.5 mm or less. For aggregate orientation, the angles between each particle and the horizontal axis were plotted in frequency histograms. For the segregation, the number of aggregates particles found in each of the radial regions (Fig. 1c) was defined for each aggregate size. The use of ImageTool requires the same steps used for the other software, which was described above. The major difference is that, for this paper, there was only one parameter for the HMA internal structure analyzed by ImageTool, which was the aggregate orientation.

Fig. 5. Steps for DIP: (a) digitalized image, (b) image transformed on grey scale, and (c) borders’ detection.

4. Results and discussion 4.1. Aggregates shape characterization 4.1.1. Percentage of flat/elongated particles The percentages of flat/elongated particles for the aggregates were determined through images analysis. Fig. 6 presents the results obtained using ImageTool software. The manual test to determine the amount of flag/elongated particles provides realistic values due to the possibility to measure the length of any of the aggregate directions, but it demands too much time, because the particles must be analyzed one by one. There was a notable differ-

375

I.S. Bessa et al. / Construction and Building Materials 37 (2012) 370–378

ence between the values obtained in the laboratory, using the caliper, and the ones from the DIP. Aggregate pictures were taken in 2D. The third dimension (aggregate depth) cannot be seen and measured by ImageTool. Therefore, the use of AIMS would bring more realistic values for flat/elongated particles determination. Nevertheless, the methodology involving ImageTool has a very low cost and can provide real ratios values. The aggregates investigated match Superpave criteria in terms of flat and elongated particles (less than 10% above 1:5). The granitic aggregate tends to be more flat/elongated than the alternative ones. The frequency of granitic aggregates that have their dimension on the 1:2 proportion was, approximately, twice the other aggregates’ frequencies. The use of alternative aggregates can produce mixes with better resistance during compaction. 4.1.2. Angularity and roundness Angularity results provided the number of fractures faces for each aggregate (Fig. 7). The results were compared to roundness. The lower the number of fractured faces, the higher is the aggregate roundness. The aggregates met Superpave specification for every traffic/pavement thickness combination, except the granitic, which has almost 60% of its particles with one fractured face. Fig. 8 presents roundness values obtained using ImageTool software. The granitic aggregate has a high percentage (65%) of particles with sphericity lower than 0.6. The alternative aggregates have a high percentage of particles with sphericity higher than 0.6 (75% and 65% for CDW and steel slag, respectively). CDW has a great quantity (42%) of its particles with sphericity values higher than 0.7, the closest to a sphere shape among the aggregates investigated. The values obtained using DIP are more precise and based on statistic distributions, not only on average values.

nitic aggregates. The Digital Image Analysis System provided aggregates orientation, contact points, and aggregates segregation. ImageTool software provided only aggregates orientation. Only coarse aggregates were selected due to (i) software precision, which presents limitations on detecting fine aggregates and (ii) coarse aggregate influence on HMA mineral skeleton. 4.2.1. Contact points Table 4 presents the number of contact points for each mix. The number of contacts between aggregates particles was higher (12%) for the SMA. The specimens extracted from the slabs have a higher level of contact points (12%), comparative to the SGC ones (mix I). For the mixes composed by CDW (mix II), samples compacted using both methods presented similar results. This might have happened due to the alternative aggregate heterogeneity. The CV values indicate that the different compaction methods produce specimens with distinct characteristics. 4.2.2. Orientation For aggregate orientation, the vector magnitude was calculated. The values indicate the particles orientation uniformity within the mixes. Table 5 shows the results using ImageTool software for horizontal and vertical specimens sections. Table 6 presents the re-

4.2. HMA internal structure analysis With the two different software, an analysis of the internal structure was performed for: (i) HMA specimens produced using the CGS, one composed by granitic aggregates and another composed by CDW (50%) and granitic aggregates (50%), (ii) HMA specimens produced using the LCPC compactor, one composed by granitic aggregates and another composed by CDW (50%) and granitic aggregates (50%), and (iii) SMA specimens composed by gra-

Fig. 7. Number of fractured faces (laboratory test).

Fig. 6. Percentage of flat/elongated particles (DIP).

Fig. 8. Roundness values (DIP).

Table 4 Number of contact points. Contact points/mix

Superpave gyratory compactor (SGC) HMA

Average CV (%)

LCPC compactor SMA

HMA

Mix I (100% granitic)

Mix II (50% CDW)

Mix IV (100% granitic)

Mix I (100% granitic)

Mix II (50% CDW)

267 15.6

279 9.3

298 3.8

298 5.8

278 5.1

376

I.S. Bessa et al. / Construction and Building Materials 37 (2012) 370–378

Table 5 Vector magnitude values (horizontal and vertical sections) – ImageTool software. Vector magnitude/mix

Section

HMA Superpave gyratory compactor (SGC)

LCPC compactor

Mix I (100% granitic)

Mix II (50% CDW)

Mix I (100% granitic)

Mix II (50% CDW)

Average (%) CV (%)

Horizontal

55.4 4.7

48.2 3.8

53.2 15.7

52.3 5.3

Average (%) CV (%)

Vertical

32.0 3.0

33.8 6.5

52.8 4.1

51.6 8.0

Table 6 Vector magnitude values (horizontal section) – Digital Image Analysis System. Vector magnitude/mix

Superpave gyratory compactor (SGC) HMA

Average (%) CV (%)

LCPC compactor SMA

HMA

Mix I (100% granitic)

Mix II (50% CDW)

Mix IV (100% granitic)

Mix I (100% granitic)

Mix II (50% CDW)

20.9 12.8

23.6 27.7

24.2 22.3

22.1 15.1

24.0 13.4

sults obtained by Digital Image Analysis System for horizontal sections. This software only allows analysis on a circular image. It can be observed that the specimens extracted from the slabs presented similar characteristics in both directions. These materials tend to be isotropic, i.e., aggregate distributions are similar in both directions. For SGC specimen, there was no uniformity for the vector magnitude for both directions. Horizontal sections results were closer to the ones found for the slabs specimens. Yue et al. [21] and Vasconcelos et al. [7] concluded that SGC samples

presented a more homogeneous distribution when compared to Marshall ones. Mix II (containing CDW) presented results similar to mix I. CDW and granitic aggregate mixes presented similar results even when different compaction methods were used. The SMA mix (mix IV) presented a vector magnitude 16% higher than the dense-graded granitic mix (mix I). Some vector magnitude values obtained for the horizontal direction are different when comparing both software results. ImageTool analysis considers coarse and fine aggregates. This can

Fig. 9. Segregation for each mix (digital image analysis system): (a) mix I – 100% granitic, SGC; (b) mix II – 50% CDW, SGC; (c) mix I – 100% granitic, LCPC compactor; (d) mix II – 50% CDW, LCPC French compactor; and (e) mix IV – 100% granitic, SGC.

I.S. Bessa et al. / Construction and Building Materials 37 (2012) 370–378

cause the issue of assuming binder and fine aggregates together as being a single coarse aggregate particle. Also, the Digital Image Analysis System has a better sensitivity, being more precise and able to better mark out the edges of the aggregates to be considered. 4.2.3. Segregation Digital Image Analysis System was able to provide the amount of each aggregate size at each group: (i) group 1, related to the internal circle, with 1/3 of the specimen radius; (ii) group 2, related to the intermediate zone, with a radius equivalent to 2/3 of the specimen radius; and (iii) group 3, related to the external border, with the specimen full radius (Fig. 3). The software considered the aggregates retained on four sieves: No. 4, 3/800 , 1/200 and 3/400 . The results are presented in Fig. 9. It is possible to conclude that the SGC generated mixes with aggregates that are more segregated, i.e., distributed in a less uniform way along the three groups, especially for the SMA one. Comparing different compaction methods, LCPC compactor mixes presented lower segregation. The results in this research are similar to the ones found by Tashman et al. [11] and Hunter et al. [17], who concluded that there must be more segregation on SGC mixes. It is important to state that Masad et al. [1] found opposite results, which indicate that there is no clear tendency in relation to aggregates segregation. 5. Conclusions This paper aimed to characterize aggregates (conventional and alternatives) and their distribution within asphalt mixes using DIP techniques. Manual laboratory tests were also performed in order to compare the results with those obtained through DIP. It was shown that DIP characterization is easier and faster. Image analysis can provide results more precise than those obtained from lab tests, with respect to shape parameters. Laboratory tests present average values while images analyses present statistical distributions for each parameter. Roundness results obtained using DIP show the same tendency obtained in the laboratory. In respect to the used software, ImageTool was able to characterize aggregates particles shape and orientation, and the Digital Image Analysis System only captures the parameters that are related to HMA internal structural, which includes segregation, orientation and contact points. The granitic aggregate presented roundness values lower than the ones from steel slag and CDW. This indicates a better interlock when the natural aggregates are used. The SMA mix aggregate particles presented higher contact points than the conventional dense-graded mix, which is expected, since SMAs have a higher amount of coarse aggregates. Regarding aggregates orientation after compaction, the aggregates distribution and orientation were very similar for both dense-graded and SMA mixes. With respect to angular orientation, the compaction methods did not present relevant influence. When comparing mixes composed only by granitic aggregates with mixes composed also by CDW, it can be stated that this residue does not affect the HMA structure in terms of aggregates distribution. The SGC mixes segregation was greater compared to the LCPC compactor ones. For the specimens extracted from the slabs, lower segregation was obtained. Acknowledgements The authors thank CNPq (Conselho Nacional de Desenvolvimento Científico e Tecnológico) for supporting this research; LTP/USP (Laboratório de Tecnologia de Pavimentos/USP) for the slabs compaction; and Professors Hussain Bahia at the University of

377

Wisconsin-Madison and Enad Mahmoud at the Bradley University for providing references and software. References [1] Masad E, Muhunthan B, Shashidhar N, Harman T. Effect of compaction procedure on the aggregate structure in asphalt concrete. Manuscript No. 991052. Transportation Research Board; 1999. [2] Souza LT. Investigation of aggregate angularity effects on asphalt concrete mixture performance using experimental and virtual asphalt samples. M.Sc. thesis. Lincoln (NE): UNL; 2009. [3] Castelo Branco VTF, Masad E, Little DN, Soares JB, Motta LMG. Caracterização de forma, angularidade e textura de agregado de brita granítica e escórias de aciaria usando o Aggregate Imaging System (AIMS) In: XX Congresso de Pesquisa e Ensino em Transportes – ANPET, Brasília; 2006. p. 1303–14 [in Portuguese]. [4] Monismith CL. Summary report on permanent deformation in asphalt concrete. SHRP-A/lr-91-104. Washington (DC): National Research Council; 1991. [5] Vinson TS, Janoo VC, Haas RCG. Summary report low temperature and thermal fatigue cracking. SHRP-A/IR-90-001. Washington (DC): National Research Council; 1999. [6] Bernucci LB, Motta LM, Ceratti JAP, Soares JB. Pavimentação asfáltica: formação básica para engenheiros. 1st ed. Rio de Janeiro (RJ): Petrobras; 2007 [in Portuguese]. [7] Vasconcelos KL, Evangelista Jr F, Soares JB. Análise da estrutura interna de misturas asfálticas In: XVII Congresso Brasileiro de Pesquisa e Ensino em Transportes. Recife, PE; 2005 [in Portuguese]. [8] Masad E. X-ray computed tomography of aggregates and asphalt mixes. Mater Eval 2004:775–83. [9] Roberts FL, Kandhal PS, Kennedy TW. Hot mix asphalt materials, mixture design and construction. 2nd ed. Lanham (MD): National Center for Asphalt Technology; 1996. [10] Gouveia LT. Contribuições ao estudo da influência de propriedades de agregados no comportamento de misturas asfálticas densas. M.Sc. thesis. São Carlos, SP: USP; 2006 [in Portuguese]. [11] Tashman L, Masad E, Peterson B, Saleh H. Internal structure analysis of asphalt mixes to improve the simulation of superpave gyratory to field conditions. J Assoc Asphalt Paving Technologists 2001;70:605–55. [12] Sousa JB, Harvey J, Painter L, Deacon JA, Monismith CL. Evaluation of laboratory procedures for compacting asphalt – aggregate mixtures. Report No. SHRP-AUWP=91-523. Washington (DC): Strategic Highway Research Program, National Research Council; 1991 [13] Bessa IS, Castelo Branco VTF, Soares JB. Caracterização de agregados convencionais e alternativos utilizando técnicas de processamento digital de imagens In: XXIII Congresso de Pesquisa e Ensino em Transportes; ANPET, Vitória, ES; 2009 [in Portuguese]. [14] Lopes MM, Bessa IS, Castelo Branco VTF, Soares JB. Efeito do tipo de compactação nos parâmetros volumétricos e no comportamento mecânico de misturas asfálticas. In: 20° Encontro de Asfalto – Instituto Brasileiro de Petróleo, Gás e Biocombustíveis, Rio de Janeiro, RJ; 2010 [in Portuguese]. [15] Masad E, Kassem E, Chowdhury A. Application of imaging technology to improve laboratory and field compaction of HMA. The transportation research information services TRIS; 2009. p. 250. [16] Tashman L, Masad E, Little D, Hussein Z. A microstructure-based viscoplastic model for asphalt concrete. Int J Plast 2004:1659–85. [17] Hunter AE, Airey GD, Collop AC. Aggregate orientation and segregation in laboratory-compacted asphalt samples. Transport Res Record TRB 2004;1891:8–15. [18] Zhang L, Wang Z. Characterization of HMA internal structure using image analysis (1968–1975). American Society of Civil Engineers; 2008. [19] Stroup-Gardiner M, Brown ER. Segregation in hot mix asphalt pavements. Interim report. Study No. 9–11. Washington (DC): National Cooperative Highway Research Program, Transportation Research Board; 1998. [20] Masad E, Jandhyala VK, Dasgupta N, Somadevan N, Shashidhar N. Characterization of air void distribution in asphalt mixes using X-ray computed tomography. J Mater Civil Eng 2002:122–9. [21] Yue ZQ, Bekking W, Morin I. Application of digital image processing to quantitative study of asphalt concrete microstructure. Transportation research record 1492. Washington (DC):TRB, National Research Council; 1995. p. 53– 60. [22] Evangelista Jr F, Souza LT, Soares JB. Processamento digital de imagens aplicado à caracterização de agregados quanto à forma. In: XIX Congresso de Pesquisa e Ensino em Transportes, ANPET, Recife, PE; 2005 [in Portuguese]. [23] Souza LT, Evangelista Jr F, Soares JB. Processamento digital de imagens aplicado a análise micromecânica de misturas asfálticas. In: 18° Encontro de Asfalto – IBP; Rio de Janeiro, RJ; 2006 [in Portuguese]. [24] Gonzalez RC, Woods RE. Digital image processing. Addison–Wesley; 1993. [25] Masad E, Button JW. Unified imaging approach for measuring aggregate angularity and texture. Comput-Aid Civil Infrastruct Eng 2000: 273–80. [26] Mahmoud E, Kutay E, Bahia H. Digital image analysis system. In: Standard method for determining aggregate structure in asphalt mixes by means of planar imaging – draft procedure; 2010 .

378

I.S. Bessa et al. / Construction and Building Materials 37 (2012) 370–378

[27] ANP Resolução ANP No. 19. de 11.7.2005 DOU 12.7.2005 ‘‘Regulamento Técnico No. 3/2005’’. Agência Nacional do Petróleo; 2005 [in Portuguese]. [28] NF P 98-250-2. Essais relatifs aux chaussées – Préparation dês mélanges hydrocarbonés. Partie 2: Compactage des Plaques; 1991 [in French]. [29] Onofre FC, Lopes MM, Araújo Jr PC, Vale AC, Oliveira Filho CMS, Soares JB. Comportamento mecânico de misturas asfálticas do tipo SMA, concreto asfáltico e areia-asfalto In: XXII Congresso de Pesquisa e Ensino em Transportes, ANPET, Fortaleza, CE; 2008 [in Portuguese].

[30] ASTM D 4791. Standard test method for flat particles, elongated particles, or flat and elongated particles in coarse aggregate. American Society for Testing and Materials, ASTM; 1999. [31] ASTM D 5821. Standard test method for determining the percentage of fractured particles in coarse aggregate. American Society for Testing and Materials, ASTM; 2001.