STUDY ON THE QUANTIFICATION AND CORRELATION OF GRADATIONAL SEGREGATION IN ASPHALT MIXTURES

Transcription

1 STUDY ON THE QUANTIFICATION AND CORRELATION OF GRADATIONAL SEGREGATION IN ASPHALT MIXTURES XU Ke *, ZHANG Xiao-Ning, LI Zhi and WANG Duan-yi Road Research Institute, South China University of Technology, Guangzhou, P. R. of China * E-ail: drxuke@gail.co. Tel: ABSTRACT In order to evaluate the gradational segregation in asphalt ixtures, Superpave Gyratory Copaction speciens were copacted according to 21 coon gradations. Sectional iages of the speciens were acquired using the digital iage processing technique. FeretDiaeter was used to distinguish the coarse aggregates. The stereological deterination technology that was used to evaluate the anisotropy of the structures was iproved. The coarse aggregate structure factor, J f, was used to evaluate the anisotropy of the asphalt ixtures. Linear regression was done between J f and the subtotal triage of coarse aggregates, and the correlation coefficient achieved was For the gradations with the sae axiu noinal grain size, grey correlation was used to analyze the results. It was concluded that, except for the 9.5 axiu noinal grain size gradation, J f was affected ost by the subtotal triage of coarse aggregates in the thirdlargest sieve size. 1. INTRODUCTION The unifority of asphalt ixtures is one of the ost iportant ways of assuring a long lifespan for an asphalt paveent. Segregation of asphalt ixtures has been found to be related to the preature daage of asphalt paveents. The segregation of asphalt ixtures can be divided into gradational segregation and teperature segregation. Gradational segregation refers to the non-unifor distribution of coarse and fine aggregates, which eans that the actual gradation no longer accords with the designed gradation. The air voids will be uch larger where the coarse aggregates congregate. At the sae tie, the air voids will be uch saller where the fine aggregates congregate. This is the reason for peranent deforation and bleeding (Sha Qing-Li, 2001). At present, the study on the gradational segregation of asphalt ixtures still needs further work. Muhunthan (2000), Masad and co-workers (2002) and Li Xiao-jun (2004) have all undertaken research into the distribution of air voids in asphalt ixture speciens. In evaluating the anisotropy of asphalt ixtures, these researchers did not take into account the distribution of aggregates, which would have ade the work uch ore coplex. Li Zhi (2002) assessed the distribution characteristics of aggregates in annular and zonal anners but did not achieve a satisfying result. Peng Yong et al. (2004) evaluated the unifority of asphalt ixtures directly according to the position of the aggregates. However, because he did not use the fabric sensor syste, it is difficult to translate his results into structural easureents. Proceedings of the 25 th Southern African Transport Conference (SATC 2006) July 2006 ISBN Nuber: Pretoria, South Africa Produced by: Docuent Transforation Technologies cc Conference organised by: Conference Planners

2 2. RESEARCH METHOD BASED ON THE IMAGE PROCESSING TECHNIQUE Six-inch (150 ) asphalt ixture speciens were copacted using an SGC (Superpave Gyration Copacter). The gradations of the asphalt ixture are shown in Table 1. Three speciens were copacted according to one gradation. Table 1. Gradations of SGC speciens. Gradations Mass proportion (%) through the following sieves () AC-25I AC-25II AM AC-20I AC-20II AM AC-16I AC-16II AM AK SMA Gradations Mass proportion (%) through the following sieves () AC-13I AC-13II AM AK-13A AK-13B SMA AC-10I AC-10II AM SMA The speciens were cut using a double-edged saw, and eight sectional iages could be acquired fro one specien. The cutting ethod is shown in Figure 1.

3 Fig. 1. Sketch ap of specien cutting. Fig. 2. Sectional iage of SMA-16 specien. Fig. 3. 2D iage of SMA-16 specien. The sectional iages of the asphalt ixture speciens were acquired using a CCD digital caera, and those iages are shown in Figure 2. After enhanceent and noise eliination, we used the double-peak ethod to segent the iages in order to acquire the inforation on the aggregates. The original iages were converted into twodiensional iages, as shown in Figure 3. To be able to evaluate the distribution of the coarse aggregates, those belonging to the different sieve sizes needed to be distinguished. Masad et al. (1999) used the FeretDiaeter to distinguish between the coarse aggregates. Li Zhi (2002) presented the equivalent inor axis as the shape paraeter for separating the coarse and fine aggregates. Yang Yu-liang (2003) regarded the sectional area of aggregates as a suitable index. Through coparison, it was found that the FeretDiaeter is a uch ore suitable paraeter for distinguishing between the coarse aggregates. The FeretDiaeter can be calculated fro equation 1: FeretDiaeter = 2 Area / π (1) where FeretDiaeter = equivalent diaeter and Area = sectional area of the aggregates. The coarse aggregates are segregated according to the sieve size after the fine aggregates have been eliinated. The results of the SMA-16 specien segregation are shown in Figures 2 and 3. Referring to the iages of the coarse aggregates, the structure factor, J f, is used to evaluate the anisotropy of the internal structure of the asphalt ixture. 3. STEREOLOGICAL DETERMINATION TECHNOLOGY FOR EVALUATING THE STRUCTURE ANISOTROPY The internal structure of an asphalt ixure can be evaluated through the distribution of the aggregates, the ajor axis orientation of the aggregates, and so on. Hillard (1968) presented stereological deterination technology to easure the voids ratio. Muhunthan and Chaeau (1997) developed this ethod further. This ethod has been iproved since this study is trying to evaluate the gradational segregation of asphalt ixtures. Firstly,

4 a sectional iage is selected as a typical unit circle. A series of easuring lines are picked up to intersect the coarse aggregates (Figure 4). After the intersecting length between the easuring line and the coarse aggregates has been easured, the ratio of the coarse aggregates' length on the easuring line versus the whole easuring line can be calculated. Figure 4. A typical unit circle. According to this ratio, the function of the directional area ratio of the coarse aggregates can be calculated as follows: n l) = n (1 + F l l ) (2) ( ij i j where n F ij = the average area ratio of the coarse aggregates, = the second-order fabric tensor of the coarse aggregates and l = the unit vector. For the two-diensional proble, the ethod of deterining the directional area ratio of the coarse aggregates is as follows: 1) Represent the sectional iages with typical unit circles. In the circle, draw a series of lines through the centre of circle. The angles between the lines are defined as π θ =, = 0,1,2,3, N 1. Measure the proportion of coarse aggregates, L l ( θ ), N in every line (as shown in Figure 5). 2) Calculate the average area ratio of the coarse aggregates, as shown in equation 3. N = 1 Ll ( θ ) n (3) = 0 N 3) Calculate the fabric tensor of the coarse aggregates, as shown in equations 4 and 5. A2 B2 F ij = (4) B2 A2 cos(2π / N) θ Ll ( θ ) (5) N 1 N 1 2 = 2 Ll ( ) B2 = 0 sin(2π / N) = 0 A

5 4) Calculate the function of the directional area ratio of the coarse aggregates. With regard to the two-diensional proble, it can be calculated fro equation n( θ ) = n (1 + A cos θ + 2B sinθ cosθ A sin θ ) (6) 1 2 k 2 k 2 k k 2 k Define J f = ( Fij Fji ) as the structure factor, J f, which is the invariant of the fabric tensor, F ij. For the two-diensional proble, J f can be calculated fro equation 7. J = F + 2F = A + 2B (7) f J f As the anisotropy grows, becoes larger. In this paper it is accepted that A 2 = B 2 = J 0, f = 0. The structure factor, Jf, is used to evaluate the gradational segregation of the asphalt ixture speciens. 4. DATA ANALYSIS The average structure factor, J f, of every specien was calculated. Using the subtotal triage of coarse aggregates as weighting paraeters, the J f of every gradation can be deterined through average processing, as shown in Table 2. Table 2. Structure factor (J f ) of every gradation axiu noinal grain AC-25I AC-25II AM-25 Coarse aggregates structure factor J f axiu noinal grain AC-20I AC-20II AM-20 Coarse aggregates structure factor J f axiu noinal grain AC-16I AC-16II AM-16 SMA-16 AK-16 Coarse aggregates structure factor J f axiu noinal grain AC-13I AC-13II AM-13 SMA-13 AK-13A AK-13B Coarse aggregates structure factor J f axiu noinal grain AC-10I AC-10II AM-10 SMA-10 Coarse aggregates structure factor J f After analysing the data the following is observed: 1. As the axiu grain size increases in the gradation, the gradational segregation of an asphalt ixture becoes a atter for greater concern, which accords with experience. 2. With the subtotal triage of coarse aggregates set as an independent variable, and the structure factor, J f, set as a dependent variable, a linear regression analysis was carried out. The results of the significance test are shown in Table 3.

6 Table 3. Results of significance test. Sieve size Structure factor J f H H H H Note:H eans there is a highly significant relation between variables and results(p<0.01), a blank eans there is no significant relation between variables and results. The linear correlation result is shown in equation 8, and the correlation coefficient, R, was 0.96: y = P P P P 4.75 (8) where P 26.5, P 16, P 13.2, P 4.75 = subtotal triage of the coarse aggregates of the corresponding sieve size (%). 3. According to the usual experience, when the axiu noinal grain size is no larger than 13.2, there is little gradational segregation. So we can extrapolate that, when the structure factor, J f, is no larger than 000, the asphalt ixture can be regarded as having little segregation. For the sae axiu noinal grain, grey correlation was used to analyse the data. 5. GREY CORRELATION ANALYSIS "Grey syste" theory was developed by Professor Deng Ju-long in 1982, when it was a new interdisciplinary subject. The research objective of grey syste theory is the syste of uncertainty. By analyzing the partly known inforation and extracting the valuable inforation, the theory helps to ipleent efficient control of the syste. Since the syste of uncertainty exists everywhere, grey syste theory has a proising future (Liu Si-feng et al., 1999). The basic concept of grey correlation is to judge the correlation relationship according to the siilarity of curves. The ore siilarity there is between the two curves, the ore correlative they are. The grey correlation analysis ethod akes up for the disadvantages of the statistical ethod, which is very convenient for calculation. We shall take the 16.0 axiu grain as an exaple to show the calculation procedure of grey correlation analysis. The variant and correlation factors are shown in Table 4. Table 4. Variant and correlation factors. Gradation AC-16I AC-16II AM-16 AK-16 SMA-16 Structure factor J f Subtotal triage of 16.0 sieve (%) Subtotal triage of 13.2 sieve (%) Subtotal triage of 9.5 sieve (%) Subtotal triage of 4.75 sieve (%)

7 1) Calculate the absolute degree of association. Q X = ( x (1) x (1), x (2) x (1), x (3) x (1), x (4) x (1), x (5) x (1)) 0 i i i i i i i i i i i = ( xi (1), xi (2), xi (3), xi (4), xi (5)); i= 0,1, 2,3, 4 0 i X = ( x (1), x (2), x (3), x (4), x (5)) = (0,0.065,-0.024,0.073,0.008) = ( x (1), x (2), x (3), x (4), x (5)) = (0,2.5,2.5,2.5,2.5,) = ( x (1), x (2), x (3), x (4), x (5)) = (0,5.0,7.5,9.0,5.0) = ( x (1), x (2), x (3), x (4), x (5)) = (0,,1.5,-1.0,5.5) = ( x (1), x (2), x (3), x (4), x (5)) = (0,4.5,11.0,7.0,13.5) Q s = x ( k) + x (5) ; i= 0,1, 2,3, i i i k= 2 2 s = 0.118; s = 8.750; s = ; s = 3.750; s = Q s s = ( x ( k) x ( k)) + ( x (5) x (0)); i= 1,2,3, i 0 i 0 i 0 k = 2 2 s s = ; s s = ; s s = ; s s = Q ε 1+ s + s 0 i 0i = i = 1+ s0 + si + si s0 ; 1,2,3,4 ε = 334; ε = 856; ε = 878; ε = ) Calculate the relative degree of association. ' ' ' ' ' ' xi(1) xi(2) xi(3) xi(4) xi(5) Q Xi = ( xi(1), xi(2), xi(3), xi(4), xi(5)) = (,,,, ); i = 1, 2,3, 4,5 x (2) x (2) x (2) x (2) x (2) ' ' ' ' ' ' i ' ' ' ' ' ' ' ' ' ' i i i i i X = ( x (1), x (2), x (3), x (4), x (5)) = (1.00,1.1039,0.9624,1.1158,1.0131) = ( x (1), x (2), x (3), x (4), x (5)) = (1.00,2.00,2.00,2.00,2.00) = ( x (1), x (2), x (3), x (4), x (5)) = (1.00, ,1.50,1.60,1.3333) = ( x (1), x (2), x (3), x (4), x (5)) = (1.00,1.0345,1.1034,0.9310,1.3793) ' ' ' ' ' = ( x (1), x (2), x (3), x (4), x (5)) = (1.00,1.2903,1.7097,1.4516,1.8710) ' ' ' ' ' A siilar procedure is just like calculating the absolute degree of association. γ = 861; γ = 121; γ = ; γ = ) Calculate the coprehensive degree of association. Qθ = ρ0i = θε 0i + (1 θ ) γ 0i; i = 1, 2,3, 4, ρ 01 ρ 02 ρ 03 ρ 04 = 597, = 989, = , = 940 4) Results analysis: Q ρ03 > ρ02 > ρ04 > ρ01 Therefore the subtotal triage of the 9.5 sieve size > the subtotal triage of the 13.2 sieve size > the subtotal triage of the 4.75 sieve size > the subtotal triage of the 16.0 sieve size.

8 For the sae axiu noinal grain, the structure factor, ost by the subtotal triage of the 9.5 sieve size. J f, is affected The grey correlation analysis results of the sae axiu noinal grain are shown in Table 5. Table 5. Grey correlation analysis results of the sae axiu noinal grain. Subtotal triage of coarse aggregates 26.5 axiunoinal grain 19.0 axiu noinal grain 16.0 axiu noinal grain 13.2 axiu noinal grain 9.5 axiu noinal grain (%) Ab Rel Co Ab Rel Co Ab Rel Co Ab Rel Co Ab Rel Co 26.5 sieve size sieve size sieve size sieve size sieve size 4.75 sieve size CONCLUSION In this paper, an index for evaluating the anisotropy of an asphalt ixture the coarse aggregate structure factor J f was presented. It is based on an iage-processing technique and stereological deterination technology that focuses on the two-diensional distribution of coarse aggregates. The gradational segregation in 21 coon gradations was analysed with this index. However, cutting of the speciens with a two-edged saw leads to a destructive easureent. The authors have not been able to acquire enough sectional iages of asphalt ixture speciens. With the developent of CT (coputerised toography) techniques, the authors expect to be able to depict the internal structure of asphalt ixture speciens three-diensionally in the near future. 7. REFERENCES [1] Hilliard, JE Measureent of Volue in Volue Quantitative Microscopy. McGraw-Hill Book Co. Inc., New York, USA. [2] Li Xiao-juan Siulation and discriination of daage process of asphalt ixture. Post-PhD research report of the South China University of Technology. [3] Li Zhi Analysis of volue constituent characteristics for asphalt ixture based on digital iage processing. Doctoral dissertation, Harbin Institute of Technology. [4] Liu Si-feng, Guo Tian-bang and Dang Yao-guo The theory and application of grey syste. Second Edition, Science Press. [5] Masad, E and Soadevan, N Microstructural finite-eleent analysis of influence of localised strain distribution on asphalt ix properties. Journal of Engineering Mechanics, Aerican Society of Civil Engineers, 128(10), October, pp

9 [6] Masad, E, Muhunthan, B, Shashidhar, N, et al Internal structure characterization of asphalt concrete using iage analysis. Journal of Coputing in Civil Engineering, 13(2), April, pp [7] Muhunthan, B and Chaeau, JL Void fabric tensor and ultiate state surface of soils. Journal of Geotechnical and Geo-environental Engineering, 123(2), February, pp [8] Muhunthan, B, Masad, E and Assaads, A Measureent of unifority and anisotropy in granular aterials. Geotechnical Testing Journal, GTJODJ, 23(4), Deceber, pp [9] Peng Yong, Sun Li-jun and Dong Rui-kun Discussion on new ethod for evaluating the hoogeneity of asphalt ixture. Journal of Tongji University (Natural Science Edition), 33(2), pp [10] Sha Qing-lin, Preature Daage and Precaution in Asphalt Paveent of Expressway. China Counication Press. [11] Yang Yu-liang Study of sub-icrostructure analysis syste for asphalt concrete. Doctoral dissertation, Tongji University.