Evaluation and Influencing Factors on Particle Agglomeration in RAP

TANG Wei, LI Ning*, ZHUANG Yuan, ZHAN He, YU Xin, WU Wenxiu, DING Gongying

(1.College of Civil and Transportation Engineering, Hohai University, Nanjing 210098, China; 2.Jiangsu Expressway Engineering Maintenance Technology Center, Nanjing 211106, China; 3.Jinhua Highway and Transportation Management Center, Jinhua 321000, China)

Abstract: Asphalt extraction test and scanning electron microscopy (SEM) were used for analysis of agglomerations of reclaimed asphalt pavement (RAP) particles.In order to quantify the agglomeration degree of RAP, the fineness modulus ratio (FMR) and the percentage loss index (PLI) were proposed.In addition, grey correlation analysis was conducted to discuss the relationship between particle agglomerations and RAP size,asphalt content (AC), and surface area.Two indexes indicate that the agglomeration degree increases in general as the RAP size reduces.This can be attributed to that particles are prone to agglomeration in the case of higher AC.Based on the SEM images and the material composition of RAP, the particle agglomeration in RAP can be classified into weak agglomeration and strong agglomeration.Grey correlation analysis shows that AC is the crucial factor affecting the agglomeration degree and RAP variability.In order to produce consistent and stable reclaimed mixtures, disposal measures of RAP are suggested to lower the AC of RAP.

Key words: RAP; particle agglomeration; grey correlation analysis; asphalt content

1 Introduction

Asphalt pavement maintenance contributes a great amount of solid waste[1,2].In China, nearly 970 million tons of reclaimed asphalt pavement (RAP) are produced per year[3].In light of environmental concerns and soaring cost of pavement materials, incorporating RAP into asphalt mixtures has become more dominant recently[4-6].However, it is reported that the average content of RAP in asphalt mixtures in China does not exceed 30%[7,8].That restricts reclaimed asphalt mixtures from providing more economic and environmental benefits[9].There are varying factors affecting the use of high percentage RAP, such as RAP variability,poor workability, compactibility and performances of reclaimed mixtures[1,10-12].Recent researches conducted by the authors lead to the confirmation that these factors are highly related to the agglomerations phenomenon of particles in reclaimed mixtures[13-15].The particle agglomerations could be classified into two types: new agglomerations and old agglomerations.The former one refers to the particle agglomerations that are newly formed during mixing process, and the other one refers to the particles agglomerations that already present in RAP before the mixing process[16,17].

Some studies have been conducted to investigate the new agglomerations.According to Bressiet alaged asphalt coating on RAP subjected to partial differential aging[16].During the mixing process, the over-aged and stiff asphalt on the external surface was removed due to the abrasion effect of sharp virgin aggregates, allowing the less aged and softer asphalt in the internal layer to emerge and act as glue.This could be the cause of agglomeration formation of particles in reclaimed mixtures.Furthermore, an index with respect to quantitative evaluation of agglomerations degree was proposed based on the hypothesis of adhesion of RAP particles to the virgin aggregates instead of an actual mobilization of RAP binder.It was found in Bressiet al’s another study that mixing temperature played a dominant role in the formation of agglomeration.There was a possible mixing temperature threshold, beyond which the particles would not form any obvious agglomerations[17].The agglomeration formation would also be influenced by the quality and the quantity of virgin aggregates[18].Navaroet alused microscopy image analysis to study the disappearance kinetics of the agglomerations in reclaimed mixtures[19].The reduction in the size of agglomerations followed with the combined effect of production temperature and mixing time.Nguyenet alrevealed that the reduction of RAP agglomerations during mixing process was attributed to the abrasion between RAP particles and virgin aggregates, breaking up the edge of RAP material[20].

The old agglomeration in RAP deserves more research effort than the new agglomerations, since part of new agglomerations in reclaimed mixtures may derive from the old agglomerations.In Zhuet al’s study[21],the asphalt extraction test, modified Los Angeles abrasion test and mixing test were respectively performed to describe the agglomeration characteristic of RAP particles.Based on these test results, the RAP particles could be divided into three categories: strong RAP,weak RAP and RAP aggregate.In order to assess the degree and stability of agglomerations of RAP particles, Xuet alproposed the quantitative index of the percentage loss rate and stability index based on the asphalt extraction tests and cantabro crushing test, respectively[13].Results showed that with the increase of RAP size, the agglomeration degree increased and the agglomeration stability differed.However, these test methods with regard to agglomerations of RAP particles (i e, old agglomerations) are mostly conducted at macroscale level.In addition, effects of RAP properties on the agglomeration degree still remain unknown.The objective of this study is to investigate the agglomeration phenomenon in RAP at multiscale level.The effects of RAP properties, including RAP size, asphalt content and surface area, on the agglomeration degree as well as RAP variability are taken into analysis.The obtained results are expected to provide practical guidance on the application of RAP.

2 Experimental

2.1 RAP materials

Two types of SMA-13 RAP materials were obtained from different segments of Suhuaiyan Expressway in Jiangsu, China.According to the technical specification JTG T5521-2019[22], sieving the RAP material into two fractions is required before using it.Two processing methods,viz, conventional crushing and screening and refined crushing and screening (named as CCS and RCS method), were used.The aim of CCS is to separate RAP materials into several different fractions depending on particle size.Not only for the separation, the aim of RCS is also to reduce the agglomeration of RAP particles.

The first step of CCS process is sun drying to reduce water content of RAP.After that, the RAP with particle size smaller than 22 mm was separated into three fractions by linear vibrating screener,viz, 0-5 mm(1#), 5-12 mm (2#), and 12-22 mm (3#).The RAP larger than 22 mm was first broken by roller crusher and then screened into three fractions mentioned above.In the process of RCS, the RAP was first sun-dried and then separated into two fractions: coarse RAP (＞5 mm)and fine RAP (＜5 mm).The coarse RAP subsequently entered into a centrifugal device, so asphalt mortar surrounding the coarse RAP partially peeled off.The RAP materials after centrifuging were screened into three groups: 0-5 mm (1#), 5-10 mm (2#), and 10-15 mm (3#).

Fig.1 presents the images of RAP materials obtained from the two methods.The samples of CCSRAP show obvious agglomeration, while the surface of RCS-RAP sample is smoother.In addition, with the purpose of analyzing the material composition and variability of RAP, the CCS-RAP and RCSRAP materials were further separated with individual sieve size 13.2, 9.5, 4.75, 2.36, and 1.18 mm in the laboratory.

Fig.1 RAP materials processed by: (a) CCS process; (b) RCS process

2.2 Methods

2.2.1 SEM test

The micromorphology of the coarse and fine RAP(9.5 and 2.36 mm) was acquired by a scanning electron microscope (SEM JSM-IT200, Japan).The RAP particle was cured by epoxy resin and then saw-cut to ensure a relatively flat surface.To observe the fracture surface, the RAP sample was fixed on an SEM stage.Before testing, the fracture surface was sputter-coated with gold film to make the material conductive.

2.2.2 Extraction test

The particle agglomeration in RAP was at macroscopic level analyzed by the centrifugal asphalt extraction test and the test method T0722-1993 (JTG E20-2019) was followed[23].This test was performed on the three graded RAP mentioned above.RAP was eluted with asphalt-solvent trichloroethylene and a centrifuge device was then employed to separate aggregates and asphalt solution.After that, sieve analysis was carried out on the RAP aggregates as well as RAP particles.Four parallel samples were prepared for each test.Asphalt content (AC) can be calculated according to the lost weight of RAP sample.As is known to all,the gradation of RAP aggregates is finer than that of RAP particles[21].It can be considered that the gradation refinement is due to the de-agglomeration of RAP particles.Therefore, the fineness modulus ratio (FMR)of RAP aggregates to RAP particles can reflect the agglomeration degree of graded RAP.This index is calculated using Eqs.(1) and (2).The higher the FMR is, the lower agglomeration degree of the RAP is.

where,fis the fineness modulus;fAGGandfRAPis the fineness modulus of the RAP aggregates and RAP particles, respectively;U1-U12is the cumulative percent of material retained on each sieve size from 26.5 to 0.075 mm respectively,%.

Asphalt extraction test was also conducted on the RAP with each size (13.2, 9.5, 4.75, 2.36, and 1.18 mm).Because the indicatorFMRis not applicable to each size of RAP material, percentage loss index (PLI)was proposed (Eq.(3)).The higherPLIvalue means the greater degree of agglomeration.

where,Piis the percent of RAP aggregates retained on theith sieve size,%.

3 Test results

3.1 Micromorphology of RAP with individual size

Fig.2 presents the SEM images of the fracture surface of the coarse RAP (9.5 mm).As shown in Fig.2(a),the aged asphalt is not in a film form to adhere to the CCS-RAP aggregates, but in the form of asphalt mortar.There are fine aggregates and powder in the aged asphalt mortar.By comparison of morphology in different areas, it can be seen that asphalt mortar distributes quite non-uniformly.The thickness of asphalt mortar varies from 268 μm to 2.1 mm, with a difference of nearly eight times.The great difference in thickness of asphalt mortar may result in the inhomogeneous asphalt film in reclaimed asphalt mixtures[24], which definitely affects the performance of asphalt mixtures.However,the thickness of asphalt mortar of RCS-RAP ranges between 100 and 300 μm, which can be observed in Fig.2(b).The asphalt mortar can be the cause of agglomeration phenomenon in coarse RAP.

Fig.2 SEM images of coarse RAP (9.5 mm): (a) CCS-RAP; (b)RCS-RAP

SEM images of the fine RAP (2.36 mm) are shown in Fig.3.The area outlined by the yellow line is the RAP and the outer area is epoxy resin.As seen from Fig.3(a), the area of CCS-RAP is mainly occupied by the aggregates with different sizes.Unlike CCSRAP, one aggregate that is significantly larger than other aggregates occupies the most area of the RCS-RAP,as shown in Fig.3(b).This structure is similar to that of coarse RAP.

Fig.3 SEM images of fine RAP (2.36 mm): (a) CCS-RAP; (b) RCSRAP

3.2 RAP material composition

The percentage of RAP aggregates retained on each sieve is shown in Fig.4.It can be found that the coarse RAP (＞ 4.75 mm) is primarily composed of aggregates with the corresponding size.The percentage of aggregate with other size is significantly lower.This indicates that the agglomeration phenomenon in coarse RAP is mainly caused by small aggregates agglomerating together and then adhering on these large aggregates.Also, this case is found in fine RCS-RAP(＜4.75 mm).While for CCS-RAP material, the fine RAP mainly consists of fine aggregates of the next one or two size.The test results correspond to the structures presented in Figs.2 and 3.

Fig.4 Residual percentage of RAP aggregates at each sieve size

Fig.5 (a) and (b) show the gradations of the graded RAP samples before and after asphalt extraction.As expected, the gradation of each graded RAP changes with finer materials separated from the agglomerated RAP particles.Compared to CCS-RAP, the change in gradation for RCS-RAP is significantly smaller, indicating the slighter agglomeration.

Fig.5 Gradation before and after asphalt extraction for: (a) CCS-RAP; (b) RCS-RAP

3.3 Agglomeration degree of RAP materials

Based on the results in Fig.4, thePLIvalue of RAP with each size is calculated and shown in Fig.6.Fig.6 shows that agglomeration exists on all sizes of RAP materials.For CCS-RAP, thePLIexhibits an overall increasing trend with the decrease of RAP size,in particular thePLIof 100% for 1.18 mm RAP.It indicates that the RAP with the smaller size corresponds to the higher degree of agglomeration.This can be explained by the fact that the asphalt content (AC) of RAP increases with the reduction in particle size.Particles are prone to agglomeration in the case of higher AC.In order to ensure the stability of reclaimed mixtures, the content of fine RAP in a mixture should be limited to some extent.

Fig.6 PLI of RAP with each particle sizes

Compared to CCS-RAP, all RCS-RAP with the same size show lowerPLIvalues.The RCS method aims to reduce the agglomeration degree of coarse RAP, as mentioned above.Only these particles are subjected to centrifugal effect.Partial coarse RAP is obtained directly by sieving unprocessed RAP and the rest is stripped from the RAP larger than 5 mm.Therefore, coarse RCS-RAP shows lowPLIvalues, ranging from 11.5% to 25.4%.PLIvalues of fine RCS-RAP(2.36 and 1.18 mm) equal to 29.2% and 40.7% respectively, remaining relatively higher.Besides, the CCSRAP and RCS-RAP with the size of 4.75 mm both have the lowestPLIvalues.It can be calculated from Fig.5 (a) and (b) that 4.75 mm RAP accounts for the highest content in unprocessed CCS-RAP (32.8%) and RCS-RAP (41.3%).RAP with this size is sufficiently crushed during processing, which results in the slightest agglomeration.

Fig.7 shows theFMRvalues before and after RAP processing.FMRincreases after CCS and RCS processing.The increase inFMRfor RCS-RAP is obviously greater than that for CCS-RAP, indicating that RCS is more conducive to the reduction of agglomeration and production of more stable and uniform RAP.RCS method is recommended for disposal of RAP before using it.Comparing theFMRof the three graded RAP, it can be seen that the agglomeration degree ranks as: 1#＞2#＞3#.This trend is consistent with the results indicated byPLI, which validates the feasibility of usingFMRto evaluate the agglomeration degree of graded RAP.

Fig.7 FMR values of graded RAP

3.4 Classification of RAP agglomeration

Based on the results and findings in the above tests, the particle agglomeration in RAP can be classified into two types: weak agglomeration and strong agglomeration.Weak agglomeration refers to the phenomena that large aggregate is surrounded by fines,which occurs in coarse and fine RAP.Strong agglomeration refers to the phenomena that small aggregates with different sizes are wrapped in asphalt mastic,which only occurs in the fine RAP.Fig.8 presents the illustration of the two types of agglomeration.The classification conduces to the improvement of consistency of RAP in mix design.

Fig.8 Two types of agglomeration in RAP

4 Analysis

4.1 Grey correlation analysis

Grey correlation analysis has been widely used to analyze the correlation degree between different sequences in a system[25].Reliable conclusions can be obtained from this analysis method, even if insufficient samples are used.Four primary steps for grey correlation analysis are as follows:

Step 1: Set one data series as the reference and comparison sequence:

Where,Xi(k) is the reference sequence wheniequals 0;Xi(k) are the comparison sequences wheniequals 1, 2,...,n;nis the number of factors to be considered, andmis the data length in a sequence.

Step 2: Get the sequences dimensionless.To make the sequences comparable, the data is processed by Eq.(5) for nondimensional alization.

where,Yi(k) are the dimensionless sequences.

Step 3: Calculate the grey relational coefficient.Upon completing the nondimensional alization, a grey relational coefficientξi(k) is defined as,

where, |Y0(k)-Yi(k)| is the absolute value of the difference betweenY0(k)andYi(k), andρis the discriminant coefficient (ρ?[0,1]), which can be adjusted on the basis of actual requirements.ρis generally 0.5.

Step 4: Determine the relational degree.Grey relational coefficients are averaged to obtain the relational degree, as shown in Eq.(7).If a particular comparison sequence is more important to the reference sequence than other comparison sequences, the relational degree of this comparison sequence would be higher.

where,γiis the relational degree betweenY0(k) andYi(k).

4.2 Factors infulencing particle agglomeration and RAP variability

The grey correlation analysis was adopted to determine the correlation degree of different factors to the RAP agglomeration, including AC, surface area(SA) and RAP size.Using the method provided in the Asphalt Institute Manual Series, the SA of extracted aggregates was calculated[26].The three factors were considered as comparison sequences, denoted asX1,X2andX3respectively.The reference sequenceX0denoted the agglomeration degree.Table 1 presents the original data.Following the steps mentioned above, the results are obtained and shown in Table 2.

Table 1 Original data for grey correlation analysis of RAP agglomeration

Table 2 Correlation degree between RAP agglomeration and other factors

As can be seen from Table 2, the correlation degree between AC and RAP agglomeration is the largest, followed by particle size and SA.In engineering practice, it is necessary to pay close attention to the agglomeration of stone mastic asphalt(SMA) mixture type, since relatively higher AC is used in SMA mixture than other mixture types, such as open graded friction course (OGFC) and dense-graded asphalt concrete (DAC).

The grey correlation analysis was also used to determine the correlation degree of AC, SA,PLIand RAP size (denoted asX1,X2,X3andX4) to the RAP variability (X0).RAP variability was reflected by coefficient of variation (COV) of AC, COV of percent of material passing two key sieve sizes,viz.4.75 and 2.36 mm.The data used for analysis is shown in Table 3.Tables 4 to 6 directly present the analysis results.

Table 3 Original data for grey correlation analysis of RAP variability

Table 4 Correlation degree between COV of AC and other factors

Table 5 Correlation degree between COV of percent of material passing 4.75 mm sieve and other factors

Table 6 Correlation degree between COV of percent of material passing 2.36 mm sieve and other factors

As seen from Tables 4 to 6, AC and the RAP agglomeration take the top two spots in the correlation degree to the RAP variability and SA and particle size are followed.This is the case among COV of AC, COV of percent of material passing 4.75 and 2.36 mm sieve.The findings obtained form the grey correlation analysis validate the statement in many studies that the particle agglomeration in RAP is one of the most important factors to influence the gradation variability[3,27].

Based on the analysis results, the relationship among AC, particle agglomeration and RAP variability is established and then depicted in Fig.9.It can beseen from Fig.9 that the asphalt content is the crucial factor contributing to particle agglomeration and RAP variability.Particle agglomeration and variability as two main material characteristics of RAP greatly affect the mix design and performance of reclaimed mixtures.Therefore, disposal measures of RAP are required to reduce the AC before its utilization.

Fig.9 The relationship among asphalt content, particle agglomeration and variability of RAP

4.3 Calculation of allowable RAP asphalt content

Fig.10 shows the correlation between asphalt content and COV of AC, COV of percent of material passing 4.75 and 2.36 mm sieve.Both COV of AC and COV of 4.75 mm passing percent show good positive linear correlation with the AC, withR2value of 0.633 4 and 0.707 3.It indicates that lowering AC of RAP is conducive to the reduction of RAP variability.

Fig.10 Correlation analysis: asphalt content vs COV of AC, COV of percent of material passing 4.75 and 2.36 mm sieve

If the RAP material is to be used in asphalt mixtures, the variability of RAP is supposed to be limited.The allowable standard deviation (SD) of AC and percent of RAP passing key sieve size is coveredin the Chinese Technical Specification (JTG T5521-2019)[22,28].Therefore, the allowable AC of RAP can be determined based on the relationship between AC and variability of RAP.Using the fitting equation in Fig.10, Table 7 presents an example (SMA-13 mixture type) of the calculation of allowable RAP AC.It can be seen that the AC of RAP is controlled by the percent of material passing 4.75 mm sieve, and the AC should not exceed the value of 2.11%.

Table 7 Example of calculation of allowable RAP AC

5 Conclusions

Particle agglomeration phenomenon in RAP is a challenge for the mix design and performances of reclaimed asphalt mixtures.In this study, the particle agglomeration was investigated by performing SEM test and asphalt extraction test.In addition,the relationship among agglomeration degree, RAP variability and asphalt content was discussed by grey correlation analysis.The major conclusions are drawn as follows.

a) Fine RCS-RAP (<4.75 mm) is mainly composed of the aggregate with corresponding sieve size, which is similar to coarse RAP (>4.75 mm).Fine CCS-RAP consists of the aggregates with different sizes.

b) Two proposed indexes, percentage loss index(PLI) and fineness modulus ratio (FMR), can be used to quantify the agglomeration of each size of RAP and graded RAP, respectively.They both show that the agglomeration degree typically increases with the decrease of RAP particle size.This can be attributed to the higher asphalt content of fine RAP, in the case fine material is prone to agglomeration.

c) The particle agglomeration in RAP is classified into weak agglomeration and strong agglomeration.Weak agglomeration occurs in coarse and fine RAP,which results in lower agglomeration degree.Strong agglomeration only occurs in fine RAP, which results in higher agglomeration degree.

d) Grey correlation analysis shows that asphalt content is the crucial factor affecting agglomeration degree and RAP variability.Disposal measures of RAP,such as RCS method in this study, are suggested to lower the asphalt content of RAP to produce consistent material.This paper also provides a calculation of allowable RAP asphalt content.

Conflict of interest

All authors declare that there are no competing interests.