Professional Documents
Culture Documents
Measuring The Influence of Aggregate Coating On The Workability and Moisture Susceptibility
Measuring The Influence of Aggregate Coating On The Workability and Moisture Susceptibility
1 ABSTRACT
2 The purportedly higher moisture susceptibility of emulsified asphalt mixtures relative to
3 traditional Hot Mix Asphalt (HMA) is a primary concern limiting their more widespread usage.
4 Of the factors contributing to moisture susceptibility in Cold Mix Asphalt (CMA) that can be
5 controlled by mix designers, aggregate coating is one of the most obvious physical means by
6 which to reduce this distress. More complete aggregate coating is expected to help limit moisture
7 susceptibility of CMA by reducing the amount of water that can be absorbed into exposed
8 aggregate. This paper refines a recently developed method to quantify aggregate coating in CMA
9 using digital imaging analysis and applies the procedure to several aggregate-emulsion systems
10 to isolate factors most directly affecting aggregate coating. After identifying the most significant
11 factors, a regression analysis of the coating matrix is used to develop quantitative models to
12 predict coating of aggregates in CMA as a function of mix design parameters. Using models to
13 predict several representative levels of aggregate coating, candidate mixtures were compacted in
14 laboratory and workability and moisture susceptibility were assessed using a compactability
15 parameter and modified Tensile Strength Ratio test, respectively. Results show that the
16 workability and moisture susceptibility of CMA are highly dependent on the level of aggregate
17 coating predicted by the quantitative models, suggesting that not only can the imaging procedure
18 be used to reliably predict aggregate coating in CMA, but also could be used to develop
19 practical, performance-based limits on plant-produced CMA mixtures in terms of aggregate
20 coating.
21
22 INTRODUCTION
23 Cold Mix Asphalt offers a number of advantages over traditional Hot Mix Asphalt and/or Warm
24 Mix Asphalt (WMA) that make it an attractive alternative for a wide range of paving
25 applications from preventative maintenance and repair, to full-scale pavement construction.
26 Namely, CMA does not require heating the asphalt binder or full drying of the aggregate
27 material, saving significant amounts of energy in mix production. CMA can be stockpiled and
28 has a longer working life, meaning it can be transported longer distances and placed in locations
29 generally inaccessible or impractical for more traditional methods (1).
30 Despite the potential benefits of using CMA, the lack of a performance-based mix design
31 procedure prohibits many agencies from promoting its use. Similar to most current asphalt
32 emulsion specifications, existing design/characterization procedures for CMA are generally
33 empirical in nature or do not represent field performance. To satisfy the need for performance-
34 based design procedure, the literature identifies three major design challenges: (a) coating of
35 aggregate material by the asphalt emulsion, (b) workability during placement and compaction,
36 and (c) curing and performance (2).
37 Aggregate coating in HMA is achieved by reducing the asphalt binder viscosity with the
38 application of heat energy and mechanical spreading. To overcome the need for heating the
39 asphalt binder in CMA, asphalt emulsions are used. Asphalt emulsions exhibit a relatively low
40 ambient temperature viscosity, allowing them to be spread over aggregates with a similar
41 mechanical effort as HMA but without heat (3). However, since the emulsifiers and additives
42 used to create asphalt emulsions induce a net charge (cationic or anionic) on the emulsion,
43 compatibility with the project aggregates is a concern. The ionic compatibility between an
44 emulsion and aggregate source can affect both coating and performance as the adhesive bond
45 between the aggregate and emulsion residue may suffer if the system is incompatible (4).
2.02
3.76
0.58
0.79
1.08
1.47
2.75
3.12
4.26
Sieve Size0.45 [mm]
1
2
3 FIGURE 1 Trial Gradations Used in Coating Study
4
5 The three levels of residual asphalt content, calculated based on the asphalt content in the
6 base emulsion, were selected considering the respective aggregate gradation specific surface
7 area. An average acceptable asphalt film thickness of 8-10 microns (the effective thickness of the
8 residual asphalt binder film on the aggregate particles) was assumed and the Duriez Method (9)
9 was used for determining the optimum residual asphalt binder content in terms of satisfying the
10 selected film thickness. The residual asphalt contents for the respective aggregate gradations
11 were averaged for the aggregate sources and are shown in Table 2. Since the estimated residual
12 asphalt contents are between approximately 4.5-5%, values of 3.5, 4.5 and 5.5% residual asphalt
13 content were selected to provide reasonable variation in the coating results.
14
15 TABLE 2 Residual Asphalt Content Based on Minimum Film Thickness
1 As CMA is stockpiled and trucked to the project site (sometimes up to 24 hours after the
2 initial mixing), significant re-mixing, or agitation, is expected. However, literature reports that
3 over-mixing emulsified asphalt mixtures can actually lead to aggregate stripping and loss of
4 coating for some emulsion-aggregate systems (7). As such, the effect of agitation was considered
5 for the mixtures in this study to identify the possibility of stripping due to the over-mixing of
6 CMA systems. To simulate this effect, mixes were placed back into the bucket mixer for an
7 additional 2.5 minutes of mechanical mixing 24 hours after the initial mixing process.
8
9 TEST METHODS
10
11 Aggregate Coating
12 Previous research by Swiertz et al. (5) demonstrated the utility of using digital imaging software
13 to quantify aggregate coating in CMA, both by reducing subjectivity and by exhibiting
14 sensitivity to several mix design factors (gradation, residual asphalt content, moisture content,
15 etc.). The study utilized image processing software in which images of the mixture are converted
16 to grayscale and individual pixels are identified as coated or uncoated based on their intensity.
17 This study followed the same methodology with minor adjustments to the mixture preparation
18 and analysis procedure.
19 It was found that the 1,500 g aggregate sample recommended by Swiertz et al. (5) did not
20 produce representative coating results for some combinations of gradation and residual asphalt
21 content, especially when considering the lowest levels of residual asphalt content. It was thought
22 that when insufficient emulsion is added to the sample, the amount of waste emulsion on the
23 sides of the bucket and mixing utensils exacerbates the lack of coating. Since the amount of
24 waste is generally independent of sample size if the same utensils are used, the sample size was
25 increased to 3,000 g for this study to be more representative of the coating experienced in full-
26 size mixtures.
27 The digital imaging software marks individual pixels (fractions of aggregates) as coated
28 or uncoated based on the grayscale intensity of the pixel; the user defines this ‘threshold’ pixel
29 intensity depending on the quality of picture, ambient lighting when the picture was taken,
30 aggregate source, aggregate moisture, and personal judgment. To mitigate subjectivity in this
31 study, all images were created with a standard office scanner to eliminate ambient lighting
32 interference. Similarly, a common threshold pixel intensity was determined to assess whether a
33 pixel is counted as coated or uncoated; bare aggregates without emulsion were scanned and the
34 average pixel intensity of the bare aggregate was used to represent the uncoated state. The
35 threshold pixel intensity was determined for aggregates in the dry and moist state, and the more
36 conservative threshold value was selected. Each aggregate blend (limestone/natural sand and
37 granite/blend sand) had a unique pixel threshold that was held constant throughout all of the
38 analysis.
39
40 Workability
41 Mixtures in this study were compacted using a modified gyratory mold which allows water to
42 drain from the samples as necessary during the compaction process. Samples were extracted
43 from the molds immediately after compaction and moved to a curing oven; both bulk specific
44 gravity (Gmb) and theoretical maximum specific gravity (G mm) samples were prepared in this
45 manner. Workability and compaction resistance was monitored using two metrics. An arbitrary
46 ‘N87’ parameter was introduced to monitor the number of gyrations (N) required to achieve
1 13±1% air voids (i.e., 87% Gmm) in the mixture. Literature (2) suggests that field CMA mixtures
2 are commonly in the range of 10-15% air voids directly after compaction, so the N87 was chosen
3 to represent the relative workability of mixtures in this study. A large N87 parameter may
4 indicate the mixture is too harsh, while a low N87 parameter may indicate a tender mix.
5 The internal friction resistance of the mixtures was also monitored in this study using a
6 Gyratory Load-Cell Plate Assembly (GLPA), which is a device that measures the effective
7 moment required to shear a mixture during compaction. Using the measured moment during
8 loading, the internal friction (shear) resistance of the mixture can be calculated. The friction
9 resistance is an indicator of mixture stability during compaction (10).
10
11 Moisture Susceptibility – Modified Tensile Strength Ratio
12 The TSR test (ASTM D4867-09) (11) commonly used for HMA and WMA must be modified
13 slightly for use with CMA mixtures due to the presence of water in the mixtures and the
14 relatively high air voids after compaction reported in the literature (12). A modified TSR was
15 suggested in the literature for Cold In-Place Recycling and was adapted for this study (13):
16 1. Samples were compacted to N = 75 gyrations with the modified gyratory compactor
17 according to ASTM D7229 (14). Theoretical maximum specific gravity (Gmm) samples
18 were also prepared.
19 2. Samples were cured in an environmental chamber at a temperature of 30±2°C and
20 humidity of 35±5% to a constant mass (defined here as less than 0.1% mass loss in 24
21 hours). It was thought that by controlling the humidity, samples would cure faster, which
22 is beneficial to expedite the testing process.
23 3. The Gmm and Gmb of the cured specimens were measured, and the air voids at N = 75 was
24 calculated;
25 4. Four samples, two replicates for unconditioned and two replicates for moisture
26 conditioning, were compacted to 13±1% air voids. The samples were cured following the
27 same conditions as specified above.
28 5. Moisture conditioned samples were saturated by applying a vacuum of 13-67 kPa
29 absolute pressure for five minutes, removing the vacuum, and measuring Gmb of each
30 sample. The degree of saturation was calculated with an allowable range of 55% to 75%.
31 If the saturation was successful, the samples were conditioned in a 25±1°C water bath for
32 24±1 hours prior to the indirect tensile testing. Note this is a major change from ASTM
33 D4867 (11) (explained below).
34 6. Samples were tested for indirect tensile strength with the load rate of 50 mm/min for the
35 unconditioned and conditioned samples and the maximum load and tensile strength ratio
36 were computed. Two replicates were run for each of the dry and conditioned samples. No
37 freeze-thaw testing was conducted.
38
39 According to Piratheepan (12) the moisture conditioning temperature (i.e., 60°C) for
40 samples based on the ASTM D4867 (11) standard may be too extreme for CMA samples with a
41 high air void content. A lower conditioning temperature of 25°C is instead suggested in some
42 provisional standards, and was adopted in this study (13).
43
44
45
46
1 agitation. Table 6 presents the results of the analysis with the chosen subset outline. The subset
2 was chosen based on a combination of a relatively high adjusted coefficient of determination
3 R2ADJ value, and a Mallows’ Cp value near the number of variables in the chosen subset plus one
4 (16). Higher order interaction effects were not considered as interpretation of such factors is
5 difficult for this research, and including interaction factors did not significantly strengthen (in
6 terms of R2 ADJ) the regression models. Based on Table 4, the factors selected by best subset are
7 congruent with the ANOVA table showing the most significant factors.
8
9 TABLE 4 Factors Selection by Best Subset Regression
Aggregate Residual
Mallows’ Gradation, Gradation,
Variables R2ADJ Moisture Agitation Asphalt
Cp κ λ
Content Content
1 48.3 255.2 X
1 27.4 412.5 X
2 75.3 52 X X
2 74.7 56.2 X X
3 81.8 4.1 X X X
3 81.1 9.2 X X X
4 81.8 5 X X X X
4 81.8 5 X X X X
5 81.8 6 X X X X X
10
11
Using the subset outlined in Table 6, a quantitative model can be derived, as shown in
12
Equation 2:
13
14 Coating 56.7 51.9 4.07 Agitation 13.0Content, 2
Radj 81.8% (2)
15
16 where, Coating = Estimated percent coated aggregates by total area
17 κ = Gradation scale parameter
18 Agitation = +1 mix was agitated, -1 mix was not agitated
19 Content = Residual asphalt content by mass
20
21 It is important to note that the range of applicable values for the parameters included in
22 Equation 2 is limited to the range used in this study (i.e., a value of zero cannot be input into for
23 Content).
24 A statistical analysis for the other three emulsion-aggregate systems was conducted in a
25 similar manner. In order to include the effect of aggregate-emulsion compatibility, aggregate
26 mineralogy was included as a factor in a final subset. A value of ‘-1’ was used to represent the
27 limestone/natural sand system and ‘+1’ was used to represent granite/blend sand system. The
28 final regression models for the anionic and cationic emulsions and the coefficients of the
29 significant factors for each model are listed in Table 5. A blank space indicates insignificant
30 factors.
31 The results presented in Table 5 indicate that by changing the emulsion charge (from
32 anionic to cationic or vice-versa) the individual factors that influence aggregate coating may also
33 change. Similarly, the effect of changing individual factors from one level to another may impact
1 one emulsion-aggregate system differently than another. For example, aggregate moisture did
2 not have a significant effect on aggregate coating for the anionic system, whereas adding
3 moisture to the cationic system improved coating. Aggregate mineralogy (compatibility) was
4 found to significantly influence aggregate coating for both systems; however, using granite with
5 the anionic emulsion is expected to decrease the amount of coating, while the opposite is true for
6 the cationic system, as expected based on charge compatibility. The relative effect of changing a
7 significant factor also differs between systems. Increasing the residual asphalt content increases
8 coating for both systems; however, coating in the anionic system was much more affected (the
9 coefficient is over six times higher for the anionic system).
10 Interestingly, the aggregate gradation was not found to be significant for the cationic
11 system. This may be partially explained by the relatively low R 2 ADJ value of this model. It was
12 observed that the cationic emulsion coated nearly all aggregate sizes and types completely; most
13 coating percentages for this system were higher than 90%. Very little variability in the coating
14 was observed when changing the levels of the factors, which served to reduce the R2ADJ. A
15 practical prediction model for the cationic system was therefore difficult to derive.
16
17 TABLE 5 Summary of Coefficients of Best Subset Regression Linear Models
Item Details
Emulsion Anionic E-150
Aggregate Limestone
Gradation Coarse
Pre-mix Moisture SSD-2%
Agitation Schedule Non-agitated
Residual Asphalt Content 3.5%, 5.0%
29
30 The final anionic coating model from Table 7 was used to predict the coating of the two
31 mixtures outlined in Table 8. The mixtures were produced and compacted to N = 75 gyrations.
32 The actual level of coating achieved in the mixtures is presented in Table 7. It is found that the
1 model can reasonably predict the coating within approximately 10% of the actual value,
2 validating the model for use in producing mixtures in the laboratory.
3
4 TABLE 7 Comparison between Measured and Predicted Coating
90.00%
Gmm (%)
85.00%
80.00%
3.5%_Coarse
75.00% 5.0%_Coarse
70.00%
1 10 100
Gyration Number
21
22
23 FIGURE 2 Densification Curves
24
25 Effect of Coating on Frictional Resistance during Compaction
26 The internal friction resistance of the mixtures calculated using the GLPA is presented in Figure
27 3. As it can be seen the effect of aggregate coating on mixture stability is minimal for this study.
1 This finding is in agreement with the locking point definition presented above; neither mixture
2 was classified as ‘tender’ (reaching a locking point below 30 gyrations). Figure 3 also suggests
3 that although the degree of coating may influence the rate of densification, the internal stability
4 (resistance to moment) may not be greatly affected. For fewer than 100 gyrations, the frictional
5 resistance from the GLPA depends mostly on aggregate physical properties such as angularity
6 and gradation, and to a lesser extent, emulsion content (10). The frictional resistance of mixtures
7 with the same aggregate physical properties is expected to be similar. It is expected, however,
8 that the internal aggregate structure after compaction may be significantly influenced by the
9 amount of coating.
10
200
160
Friction Resistance (kPa)
120
3.5%_C
80
5.0%_C
40
0
0 20 40 60 80
Gyration Number
11
12
13 FIGURE 3 Friction Resistance Curve
14
15 Effect of Coating on Moisture Susceptibility
16 Samples at both levels of coating were compacted to 13±1% air voids and tested in the modified
17 TSR procedure as specified earlier. Results of the modified TSR are presented in Table 8.
18 The results in Table 8 confirm the increased moisture susceptibility due to a lack of
19 coating in the 3.5% residual asphalt content samples, despite uniformly low tensile strengths (dry
20 and wet) between both mixtures.
21
22 TABLE 8 Modified TSR Results
Residual Degree of
Measured N87 (Gyrations Dry Tensile Wet Tensile
Asphalt Saturation TSR (%)
Coating to 13% voids) Strength (psi) Strength (psi)
Content (%)
3.5% 59.0% 35 57% 6.980 5.031 72.0%
5.0% 73.7% 14 62% 4.815 4.055 84.3%
23
24 The dry and wet tensile strengths obtained using the modified TSR procedure are very
25 low, and are most likely the result of inadequate sample curing. Although samples were cured to
1 a constant mass at 30°C, it was noted that they were internally very soft and ‘tacky’ when split
2 during the test. This is an indication that the curing temperature was not sufficiently high (or the
3 curing time was not sufficiently long enough) to remove all moisture from the sample. This
4 clearly demonstrates the importance of a consistent laboratory curing procedure during the mix
5 design process. The authors have experimented with curing in confined molds at 60°C for 90
6 hours, and it was found that much higher tensile strength can be obtained under these conditions
7 compared to the conditions used in this paper. It was also noted that the samples cured at 60°C
8 did not exhibit significant internal ‘tackiness’ that is characteristic of partially cured samples.
9 More work is needed to identify the most efficient and representative method of curing, however.
10
11 SUMMARY AND CONCLUSIONS
12 This paper investigates the applicability of using digital imaging software to measure the amount
13 of coating in CMA. Using the imaging software, significant factors affecting coating of CMA
14 aggregate were identified for aggregate-emulsion systems and quantitative models for predicting
15 coating were developed. The effect of coating on mixture workability and moisture susceptibility
16 was then assessed. The following remarks summarize the main findings:
17 1. The models developed to predict aggregate coating for anionic and cationic emulsion
18 systems were sensitive to aggregate moisture content, aggregate gradation, residual
19 asphalt content, agitation, and aggregate mineralogy. The relative importance and effect
20 of each factor was found to be different for each system, which is in line with existing
21 literature on the subject. The Cationic emulsion was found to be very good for coating
22 regardless of the factors.
23 2. By compacting mixtures with different levels of coating, it was clearly demonstrated that
24 mixtures with a higher level of coating were more workable based on the densification
25 parameter N87. The locking point of the mixtures with a higher level of coating was
26 earlier compared to the mixtures with a lower level of coating, validating the N87
27 parameter. However, the frictional resistance of both mixtures did not show a sensitivity
28 to coating, suggesting coating influences the rate of compaction for a given compaction
29 effort, but not necessary stability of the mixture.
30 3. A modified TSR test was conducted on mixtures with significantly different levels of
31 coating. The results indicate that coating is highly correlated to the moisture
32 susceptibility of CMA; mixtures with a higher level of coating demonstrated a higher
33 modified TSR value. The modified TSR testing highlighted the need for a consistent
34 curing procedure for CMA specimens.
35 As part of a more comprehensive CMA performance-based mix design procedure, this
36 study outlined the need for strict control over the mix design factors that affect aggregate
37 coating. Aggregate coating affects not only mixture workability, but also short term
38 performance. A significant knowledge gap that also needs to be addressed is a uniform and
39 practical compaction and curing procedure for CMA.
40
41 ACKNOWLEDGEMENTS
42 The materials support provided by Heritage Research Group is gratefully acknowledged. This
43 study is part of the Asphalt Research Consortium project and support provided by the FHWA
44 and WRI is acknowledged.
45
46
1 REFERENCES
2 1. Jarrett, P. M, A. N. S. Beaty, and A. S. Wojcik. Cold-Mix Asphalt Technology at
3 Temperatures Below 10°C. In Journal of the Association of Asphalt Paving
4 Technologists (AAPT), pp. 50-97, 1984.
5 2. Eckmann, B., J. J. Potti, M. Bourrel, P. Verlhac, C. Such, F. L. Calderon, and J. S. Polo.
6 Cold Mix Technology: The Contribution. Report OPTEL Project, pp. 6-15, 2001.
7 3. Salomon, D. R. Asphalt Emulsion Technology. Transportation Research Circular,
8 Number E-C102, . Transportation Research Board, Washington, DC, 2006.
9 4. Tarrer, A.R., and V. Wagh. The Effect of the Physical and Chemical Characteristics of
10 the Aggregate on Bonding. SHRP-A/UIR-91-507. Strategic Highway Research Program,
11 National Research Council, Washington, D.C., 1991.
12 5. Swiertz, D., P. Johannes, L. Tashman, and H. U. Bahia. Analysis of Laboratory Coating
13 and Compaction Procedures for Cold Mix Asphalt. In Journal of the Association of
14 Asphalt Paving Technologists (AAPT), Vol. 81, 2012, pp. TBD.
15 6. Blades, C., and E. Kearney. Asphalt Paving Principles. Publication CLRP #04–3. Cornell
16 Local Roads Program, New York Ltap Center, 2004.
17 7. Darter, M.I., S. R. Ahfield, P. L. Wilkey, and R. G. Wasill. Development of Emulsified
18 Asphalt-Aggregate Cold Mix Design Procedure. Project IHR-505, Illinois Cooperative
19 Highway Research Program, February, 1978.
20 8. Vavrik, W., G. Huber, W. Pine, S. Carpenter, and R. Bailey. Bailey Method for
21 Gradation Selection in HMA Mix Design. Transportation Research Circular Number E-
22 C044, Transportation Research Board, 2004.
23 9. Ferrari, P., and F. Giannini. Ingegneria Stradale Volume II – Corpo Stradale e
24 Pavimentazioni. ISEDI, Italy, 2007, pp. 217-223.
25 10. Guler, M., H. U. Bahia, P. Bosscher, P., and M. Plesha. Development of a Device for
26 Measuring Shear Resistance of HMA in the Gyratory Compactor. In Transportation
27 Research Record: Journal of the Transportation Research Board, No. 1723, pp. 116-124,
28 2000.
29 11. Standard Test Method for Effect of Moisture on Asphalt Concrete Paving Mixtures.
30 ASTM Standard D4867-09. ASTM International, West Conshohocken, Pa. 2009.
31 12. Piratheepan, M. Designing Cold Mix Asphalt (CMA) and Cold-In-Place Recycling (CIR)
32 Using SUPERPAVE Gyratory Compactor. MS Thesis, University of Nevada, Reno,
33 December, 2011.
34 13. Cold-In-Place Recycling (CIPR) Draft Guidelines for Laboratory Mix Design. Pacific
35 Coast Conference on Asphalt Specifications (PCCAS), February, 2008.
36 14. Standard Test Method for Preparation and Determination of Bulk Specific Gravity of
37 Dense-Graded Cold Mix Asphalt (CMA) Specimens by Means of Superpave Gyratory
38 Compactor. ASTM Standard D7229-08. ASTM International, West Conshohocken, Pa.
39 2008.
40 15. Venables, W. N., D. M. Smith, and the R Development Core Team. An Introduction to R.
41 Version 2.9.2, 2009.
42 16. Wu, J., and M. Hamada. Experiments: Planning, Analysis, and Optimization. Wiley,
43 2009.
44 17. Hemsley, M. Cold Mix Design. Presented at Asphalt Emulsion Manufacturers
45 Association, AEMA Annual Conference, Bermuda, 2002.