Evaluation of Bleeding Resistance in Chip Seal and Asphalt Emulsion Residue Rheology

coatings

Article Evaluation of Bleeding Resistance in Chip Seal and Asphalt Emulsion Residue Rheology

Preeda Chaturabong Faculty of Engineering, King Mongkut’s Institute of Technology Ladkrabang, Bangkok 10520, Thailand; [email protected]

Received: 10 September 2019; Accepted: 9 October 2019; Published: 16 October 2019

Abstract: Chip seal bleeding is influenced by many factors, including design inputs, material properties, and project-specific conditions. It reduces the surface texture of the pavement and thus compromises the safety of the traveling public. Even though factors that bring about premature bleeding are known, currently, no laboratory test methods for evaluating bleeding in chip seals have been specified. The objective of this paper is to present the results of an investigation of the influence factors of asphalt emulsion residue properties measured by the ASTM D7405 multiple stress creep and recovery (MSCR) test, as well as other factors related to chip seal bleeding resistance as measured by the modified loaded wheel test (MLWT). In this study, the MSCR test was used as a tool for evaluating the performance of asphalt emulsions because it has been identified as a potential test related to bleeding in the field. In addition, MLWT was selected as a tool for evaluating chip seal bleeding performance in the laboratory. The results of the MLWT showed that the emulsion application rate (EAR), aggregate gradation, and emulsion properties were significant factors affecting bleeding. The MSCR test was found to be a promising tool for the performance evaluation of asphalt emulsion residue, as the test was able to differentiate between emulsion chemistries and modifications in terms of sensitivity to both temperature and stress. In relation to chip seal bleeding resistance, only the creep compliance (Jnr) obtained from the MSCR test results was identified as a significant property affecting potential for bleeding.

Keywords: asphalt emulsion; chip seal; modiﬁed loaded wheel test; multiple stress creep and recovery; bleeding

1. Introduction Chip seals are placed on a variety of roadways, including those with low and high traffic volumes, where low volume is defined as those with average daily traffic (ADT) less than 5000 and high volume is defined as those with more than 20,000 ADT. Polymer-modified asphalt emulsions are recommended for roads with a high traffic volume [1,2]. Also, inverted seals have successfully been used on high-traffic-volume (30,000 ADT) Australian pavements [3]. Since most of the traffic is composed of passenger cars and small trucks, it is important to select an appropriate range of stress levels for laboratory evaluation of bleeding. Bleeding is an important failure, since it reduces the surface texture of the pavement and, hence, compromises the safety of the traveling public, particularly during wet seasons and at intersections. The bleeding performance of chip seals relies on many factors, including climatic conditions, traffic volume and type, aggregate properties, asphalt emulsion properties, emulsion application rate (EAR), and existing pavement surface. Aggregate properties, including size, shape, gradation, and toughness, influence bleeding performance. Regarding material properties, both aggregates and asphalt emulsions contribute to bleeding resistance. For example, aggregate size and nonuniform aggregate gradations increase the potential for

Coatings 2019, 9, 670; doi:10.3390/coatings9100670 www.mdpi.com/journal/coatings Coatings 2019, 9, 670 2 of 13 bleeding; however, aggregate size can be accounted for by adjusting the EAR to provide an equivalent embedment percentage to smaller aggregates [4,5]. To resist bleeding, ideal asphalt emulsion residue properties include resistance to softening at increased temperatures and/or stresses. The performance of chip seals is largely dependent on the asphalt emulsion, as it is the binding component between the aggregates and the existing surface. Soft asphalt emulsion can result in bleeding in hot weather, since it can allow the movement of aggregates to the residue film on the existing surface while forcing the residue to move to the surface of the seal covering the aggregates. Therefore, stiffening an asphalt emulsion by modification could favorably influence the bleeding performance. However, since the cost of modified asphalt is higher than that of unmodified asphalt, the selection of a suitable asphalt emulsion for each location should be based on critical factors. Climate is one necessary factor that requires consideration regarding bleeding, because in hot weather, bleeding tends to occur more often due to the softening of the asphalt binder. This phenomenon allows chips to penetrate into the underlying binder, leaving excess asphalt binder on the surface [6]. Therefore, the asphalt emulsion type needs to be selected depending on the regional climate. In addition to climate, the condition of the existing pavement surface and application rates also must be considered. Prior to selecting the target EAR, the existing pavement is surveyed, and an application rate correction factor is applied to account for the existing pavement surface condition. The application rate must be correct during construction in order to achieve optimum performance of the chip seal; if the emulsion application is excessively low, it will not retain chips in place under traffic and cause raveling, while if it is excessively high, bleeding in hot weather will occur, thus resulting in a loss of friction. Generally, the EARs are not considered as much with respect to the effect of repeated loadings for high traffic volumes [7]. This is because the heavy traffic will continue to embed the aggregates into the underlying surface after the road is open to traffic. Bleeding generally occurs at high temperatures with high traffic stress and volume. Surface treatment specifications, while having been studied by many researchers, have not been finalized. Hanz et al. (2012) stated that the multiple stress creep and recovery (MSCR) test is capable of discriminating between asphalt emulsion type and the effect of asphalt emulsion modification; however, these efforts did not include comparisons to chip seal performance [8]. Kim et al. (2017) employed the MSCR test and G*/sinδ to assess the bleeding resistance of residual asphalt emulsions at high-temperature performance grades. It was found that the correlation between creep compliance (Jnr) values and high-temperature mixture performance for chip seals is good. Also, the MSCR test better captures the elastic response of the polymer network than the Dynamic Shear Rheometer (DSR) test employed to measure G*/sinδ [9]. In earlier research, the bleeding resistance of chip seal mixtures was evaluated through simulated loading using a third-scale model mobile loading simulator [10]. Although the results of the study are promising, the simulator equipment is costly and not widely available. Chaturabong et al. (2015) applied the modified loaded wheel test (MLWT) to evaluate bleeding in chip seals. This method was successful at measuring chip seal bleeding resistance because it eliminated raveling and better represented chip seals in the field [11]. In this study, we investigated the influence factors of asphalt emulsion residue properties, and other factors related to chip seal bleeding resistance were measured by the MLWT.

2. Materials and Testing Procedure

2.1. Materials Five asphalt emulsions and two aggregate types were used in this study. The selected asphalt emulsions included cationic and anionic emulsions in order to account for the eﬀects of emulsion chemistry, and polymer- and latex-modiﬁed asphalt emulsions to test for resistance to bleeding. The asphalt emulsions employed in this study were labeled CRS-2, CRS-2L, CRS-2P, HFRS-2, and HFRS-2P. All asphalt emulsions were combined with both granite and limestone aggregate Coatings 2019, 9, 670 3 of 13

Coatings 2019, 9, x FOR PEER REVIEW 3 of 13 sources to study the eﬀects of aggregate source properties. The properties of the aggregates used are given in Table1. These properties met the speciﬁcations of AASHTO T96 [ 12] and ASTM D5821 [13] These properties met the specifications of AASHTO T96 [12] and ASTM D5821 [13] for percent Los for percent Los Angeles (LA) abrasion and percent fractured faces accordingly. Angeles (LA) abrasion and percent fractured faces accordingly.

Table 1. Aggregate types and properties used. Table 1. Aggregate types and properties used.

AggregateAggregate Properties Properties Specification Speciﬁcation Granite Limestone Limestone %% LA LA abrasion abrasion Limit Limit 35% 35% 18.1% 23% 23% % Fracture Faces 90–100 95% 90% % Fracture Faces 90–100 95% 90%

2.1.1. Aggregate Gradation Chip sealsseals areare generallygenerally constructedconstructed withwith uniformuniform gradations.gradations. The eeffectsffects of gradationgradation werewere evaluatedevaluated by considering two gradations, namely, coarsecoarse andand fine,fine, basedbased onon thethe maximum aggregate size.size. The The coarse coarse gradation gradation was was made made up up of of 50% 50% passing passing the the 9.5 9.5 mm mm size size sieve sieve and and 0% 0% passing passing the the6.5 6.5mm mm size size sieve. sieve. The Thefine finegradation gradation had had50% 50%passing passing the 6.5 the mm 6.5 mmsize sizesieve sieve and and0% passing 0% passing the 4.75 the 4.75mm mmsize sieve, size sieve, as shown as shown in Figure in Figure 1. In addition,1. In addition, all aggregates all aggregates were washed were washed to remove to removedust coatings dust coatingsfrom the fromlarger the particles. larger particles.

FigureFigure 1.1. Gradation chart. 2.1.2. Material Application Rates 2.1.2. Material Application Rates Material application rates for chip seal samples were determined using the equations of the Material application rates for chip seal samples were determined using the equations of the modified McLeod’s method [14]. In order to account the effect of EAR, two binder application rates modified McLeod’s method [14]. In order to account the effect of EAR, two binder application rates of 35% and 70% air void filled were selected. Table2 presents the material application rates used in of 35% and 70% air void filled were selected. Table 2 presents the material application rates used in this study. this study.

Table 2. Emulsion and aggregate application rate. Table 2. Emulsion and aggregate application rate. EAR (L/m2) EAR (L/m2) Aggregate Application Aggregate Asphalt Emulsion Type Aggregate 2Application % Void Filled Asphalt Emulsion Type Rate (kg/m ) % Void Filled AggregateGradation GradationCRS-2 CRS-2L CRS-2P Rate (kg/m2) CRS-2 CRS-2L CRS-2P Fine 1.02 0.95 0.97 8.13 35% CoarseFine 1.511.02 1.420.95 1.440.97 11.89 8.13 35% CoarseFine 2.031.51 1.911.42 1.931.44 8.13 11.89 70% CoarseFine 3.022.03 2.841.91 2.871.93 11.89 8.13 70% Coarse 3.02 2.84 2.87 11.89

2.2. Asphalt Emulsion Residue Performance Evaluation

Coatings 2019, 9, 670 4 of 13

2.2. Asphalt Emulsion Residue Performance Evaluation Coatings 2019, 9, x FOR PEER REVIEW 4 of 13 The overall objective of the testing design was to investigate the influence factors of asphalt emulsionThe residue overall properties, objective of as the well testing as other design factors was related to investigate to chip sealthe bleedinginfluence resistancefactors of measuredasphalt emulsion residue properties, as well as other factors related to chip seal bleeding resistance measured by the MLWT. The MSCR test, as specified in ASTM D 7405 [15] and AASHTO TP 70 [16] was used to by the MLWT. The MSCR test, as specified in ASTM D 7405 [15] and AASHTO TP 70 [16] was used evaluate if the Jnr values correlate with bleeding resistance at high temperatures. to evaluate if the Jnr values correlate with bleeding resistance at high temperatures. Test conditions were varied to assess both sensitivity to temperature and stress. AASHTO M Test conditions were varied to assess both sensitivity to temperature and stress. AASHTO M 332 332 [17] provides specification criteria based on the Jnr value for different traffic loading conditions. [17] provides specification criteria based on the Jnr value for different traffic loading conditions. In the In the specification, the maximum allowable value for standard traffic (S) is equal to 4.5 kPa 1; specification, the maximum allowable value for standard traffic (S) is equal to 4.5 kPa−1; for− forsubsequent subsequent traffic traffi ccondit conditionsions associated associated with with heavy heavy (H), (H), very very heavy heavy (V), (V), and and extremely extremely heavy heavy (E) (E) tratraffic,ffic, the the Jnr Jnrthreshold threshold isis reduced to to adjust adjust for for increasing increasing traffic traffi levels.c levels. In this In study, this study, the range the rangeof test of testtemperatures temperatures was was selected selected to represent to represent possible possible service service temperatures temperatures experienced experienced on roadways on roadways and andto toobtain obtain Jnr values Jnr values above above and below and below the S-grade the S-grade threshold. threshold. To further To assess further stress assess sensitivity, stress sensitivity, a third a thirdstress stress level levelof 10 of kPa 10 kPawas was included included in the in the test test procedure. procedure. All All asphalt asphalt emulsion emulsion residues residues were were recoveredrecovered using using the the low-temperature low-temperature evaporative evaporative recovery recovery procedure procedure specified specified in in AASHTO AASHTO PP PP 72 72 [18 ]. This[18]. recovery This recovery method method involves involves drawing drawing the asphalt the as emulsionphalt emulsion down todown a film to thicknessa film thickness of 381 ofµm 381 on a siliconeµm on mat a silicone and curing mat and it for curing 6 h at it 60for◦ 6C h in at a 60 forced °C in drafta forced oven. draft oven.

2.3.2.3. Modified Modified Loaded Loaded Wheel Wheel Test Test to to Assess Assess Chip Chip SealSeal BleedingBleeding PreviousPrevious research research [ 8[8]] has has established established thatthat thethe MSCRMSCR test is is capable capable of of discriminating discriminating between between asphaltasphalt emulsion emulsion type type andand the effect effect of of asphalt asphalt em emulsionulsion modification; modification; however, however, these these efforts e ffdidorts not did notinclude include comparisons comparisons to tochip chip seal seal performance. performance. As Asa resu a result,lt, a test a test is isneeded needed to tovalidate validate that that the the nr diffdifferenceserences observed observed in in asphalt asphalt emulsion emulsion residue residue Jnr J values values are are related relatedto tochip chip sealseal performanceperformance and to applyto apply this relationshipthis relationship to propose to propose performance performance limits limi forts for bleeding. bleeding. InIn this this study, study, applying applying the the MLWT MLWT to evaluateto evaluate bleeding bleeding in chipin chip seal seal modifications modifications was was done done using a standardusing a standard device that device included that included the addition the addition of a temperature of a temperature control device,control adevice, base plate a base to secureplate to the secure the chip seal sample, a rubber mat on the top of the sample to protect tire wear, and neoprene chip seal sample, a rubber mat on the top of the sample to protect tire wear, and neoprene foam under foam under the sample to represent existing flexible pavement. This method, which eliminates the sample to represent existing flexible pavement. This method, which eliminates raveling and better raveling and better represents chip seals in the field, is capable of measuring chip seal bleeding represents chip seals in the field, is capable of measuring chip seal bleeding resistance [11]. The clamps resistance [11]. The clamps used to hold the sample in place were also modified, as shown in Figure used to hold the sample in place were also modified, as shown in Figure2. 2.

FigureFigure 2. 2.Diagram Diagram ofof thethe modiﬁedmodified loaded wheel wheel test test (MLWT). (MLWT).

Coatings 2019, 9, 670x FOR PEER REVIEW 55 of of 13

Samples consisting of asphalt emulsion/aggregate combinations were prepared according to the Samples consisting of asphalt emulsion/aggregate combinations were prepared according to the guidance provided in ASTM D7000 [19]. After curing, samples were placed on the MLWT support guidance provided in ASTM D7000 [19]. After curing, samples were placed on the MLWT support and heated to the specified testing temperature. Testing included two replicates for each sample by and heated to the speciﬁed testing temperature. Testing included two replicates for each sample by rotating the sample 90° to the wheel path. Once the test was completed, the sample was scanned, and rotating the sample 90 to the wheel path. Once the test was completed, the sample was scanned, and a digital image was taken◦ and processed to compare the initial image with the same after trafficking. a digital image was taken and processed to compare the initial image with the same after traﬃcking. To conduct this analysis, IPAS2 software, developed by UW-Madison, was used to convert the To conduct this analysis, IPAS2 software, developed by UW-Madison, was used to convert the captured captured image to a binary (black and white) image using well-established image threshold image to a binary (black and white) image using well-established image threshold techniques [20]. techniques [20]. An example of the application of image analysis to quantify bleeding is provided in An example of the application of image analysis to quantify bleeding is provided in Figure3. Figure 3.

(a) (b)

Figure 3. PreparedPrepared sample sample and and resulting resulting black-and-white image ( a) prior to bleeding test and ( b) after after bleeding test.

The extent of bleeding was then calculated using Equation (1): Bleeding (%) = [(𝐴𝐴𝑖𝑛𝑖𝑡𝑖𝑎𝑙 − 𝐴𝐴 )/𝐴𝐴 ] × 100 (1) Bleeding (%) = [(AAinitial AAx)/AA] 100 (1) − initial × where where AAx = area fraction of aggregate (white color) after x cycles, and AAx = area fraction of aggregate (white color) after x cycles, and

AAinitialAA=initialarea = area fraction fraction of aggregate of aggregate (white (white color) color) prior prior to testing. to testing.

Since both images were processe processedd using the same filtering filtering threshold, the difference difference could be assumed to be the effect effect of bleeding. The The smaller smaller th thee white white area area (aggregates) (aggregates) after x cycles, cycles, the the greater greater the bleeding percent. 2.4. Experimental Design 2.4. Experimental Design The experimental plan is given in Table3. This experiment was carried out with granite aggregate The experimental plan is given in Table 3. This experiment was carried out with granite at a constant stress level of 299 kPa (43.4 psi) for 1 day of curing with the loading cycle kept at 200 aggregate at a constant stress level of 299 kPa (43.4 psi) for 1 day of curing with the loading cycle kept cycles. The test procedure used to evaluate residue properties is described in subsequent sections. at 200 cycles. The test procedure used to evaluate residue properties is described in subsequent In the experimental design, there were three asphalt emulsion types and one aggregate type for sections. conducting the test to quantify the factors influencing bleeding of chip seals. To verify if the significant In the experimental design, there were three asphalt emulsion types and one aggregate type for factors from the main experimental design are valid, an additional “null” experiment was conducted to conducting the test to quantify the factors influencing bleeding of chip seals. To verify if the verify that the measured and estimated bleeding based on the model derived from the main experiment significant factors from the main experimental design are valid, an additional “null” experiment was was well correlated. For the null experiment, different asphalt emulsions and aggregates from those conducted to verify that the measured and estimated bleeding based on the model derived from the used in the main experiment were selected. Specifically, the null experiment used limestone coarse main experiment was well correlated. For the null experiment, different asphalt emulsions and gradation with a stress of 299 kPa (43.4 psi), 1 day of curing time, and 200 loading cycles. The null aggregates from those used in the main experiment were selected. Specifically, the null experiment experiment matrix is shown in Table4. used limestone coarse gradation with a stress of 299 kPa (43.4 psi), 1 day of curing time, and 200 loading cycles. The null experiment matrix is shown in Table 4.

Coatings 2019, 9, 670 6 of 13

Table 3. Experimental design for assessing the eﬀects of design factors on bleeding resistance and the relationship between creep compliance (Jnr) and bleeding resistance with the MLWT.

Factors Levels Description Parameter Measured Asphalt Emulsion Type 3 CRS-2, CRS-2P, and CRS-2L − Aggregate Gradation 2 Fine and coarse − Emulsion Application % Area with bleeding 2 30% and 70% of voids ﬁlled Rate (EAR) (by imaging) J @ 3.2 kPa—0.5, 2.0, and 5.0 (adjust Temperature 3 nr temperature to keep Jnr constant) − Replicate 2 2 − Total Tests 72 − −

Table 4. Null experiment design for verifying the best subsets for bleeding resistance with the MLWT.

Factors Levels Description Parameter Measured Asphalt Emulsion 2 HFRS-2 and HFRS-2P EAR 2 %void ﬁlled = 70% % Area with bleeding Temperature 2 [email protected] kPa—2.0 and 5.0 Replicate 2 2 − Total Tests 16 − −

1 As shown in Table3, three levels of J nr (0.5, 2, and 5 kPa− ) were selected to represent low, medium, and high bleeding potential, respectively. Testing was carried out on diﬀerent asphalt emulsion residues to determine the test temperature at which the Jnr at the stress level of 3.2 kPa was equal to 1 0.5, 2.0, and 5 kPa− . The temperatures determined for diﬀerent emulsions are presented in Table5.

Table 5. Test temperatures at which diﬀerent asphalt emulsions met the speciﬁed Jnr values for a stress level of 3.2 kPa.

Asphalt Emulsion Jnr at 3.2 kPa Temperature (◦C) CRS-2 45 CRS-2L0.5 46 CRS-2P 53 CRS-2 51 CRS-2L2.0 53 CRS-2P 60 CRS-2 59.5 CRS-2L5.0 61 CRS-2P 67

3. Experimental Results Chip seal samples were prepared and tested with the MLWT at the test temperatures corresponding 1 to asphalt emulsion Jnr values of 0.5, 2, and 5 kPa− , as listed in Table5. The results for the EAR of 35% air void filled are presented in Figure4 for the fine and coarse gradations. The results for both gradations show a positive relationship between Jnr and percent bleeding of chip seals. As the Jnr increased, the amount of bleeding was also observed to increase for all three asphalt emulsions and both aggregate gradations. 1 The percent bleeding almost doubled when the Jnr increased from 0.5 to 5 kPa− . This indicates that a Jnr at 3.2 kPa could be a good indicator of potential asphalt emulsion bleeding. The results, however, showed minimal sensitivity to asphalt emulsion type, irrespective of aggregate gradation, 1 except for the Jnr equal 5 kPa− for the fine gradation. This implies that, under these test conditions, latex- and polymer-modified asphalt emulsions, as well as unmodified asphalt emulsions, will provide similar resistance to bleeding if they have equal Jnr values at the testing temperature. Coatings 2019, 9, x FOR PEER REVIEW 7 of 13 the Jnr increased, the amount of bleeding was also observed to increase for all three asphalt emulsions and both aggregate gradations. The percent bleeding almost doubled when the Jnr increased from 0.5 to 5 kPa−1. This indicates that a Jnr at 3.2 kPa could be a good indicator of potential asphalt emulsion bleeding. The results, however, showed minimal sensitivity to asphalt emulsion type, irrespective of aggregate gradation, except for the Jnr equal 5 kPa−1 for the fine gradation. This implies that, under these test conditions, latex- and polymer-modified asphalt emulsions, as well as unmodified asphalt emulsions, will Coatings 2019, 9, 670 7 of 13 provide similar resistance to bleeding if they have equal Jnr values at the testing temperature.

Fine Gradation with 35% Air Void Filled 60% 50% 40% 30% CRS-2 20%

% Bleeding CRS-2L 10% CRS-2P 0% 0.5 2 5 -1 Jnr(kPa )

(a) Coarse Gradation with 35% Air Void Filled 80% 70% 60% 50% 40% CRS-2 30% CRS-2L % Bleeding 20% 10% CRS-2P 0% 0.5 2 5

-1 Jnr(kPa )

(b)

FigureFigure 4. 4. RelationshipRelationship between between J Jnrnr andand percent percent chip chip seal seal bleeding bleeding for for ( (aa)) fine ﬁne gradation gradation with with low low EAR EAR andand ( (bb)) coarse coarse gradation gradation with with low low EAR. EAR.

TheThe results results for for the the high high EAR EAR (70% (70% air air voids voids filled) filled) are are shown shown in in Figure Figure 55 forfor thethe finefine andand coarsecoarse gradations.gradations. Similar Similar to to the the trend trend observed observed for for the the low low EAR, EAR, a a positive relationship between between percent percent bleedingbleeding on chip seals and J nr cancan be be noted. noted. Percent Percent bleeding bleeding also also doubled doubled when when the the J Jnr increasedincreased from from 1 0.50.5 to to 5 5 kPa −1, ,indicating indicating that that a a residue residue with with a a high high J Jnrnr maymay be be prone prone to to bleeding. bleeding. For For fine fine gradation, gradation, thethe percent percent bleeding rangedranged betweenbetween 30% 30% and and 38%, 38%, 41% 41% and and 49%, 49%, and and 59% 59% and and 65% 65% for for the the Jnr Jvaluesnr values of 1 of0.5, 0.5, 2.0, 2.0, and and 5.0 kPa5.0 −kPa, respectively,−1, respectively, for all for three all three asphalt asphalt emulsion emulsion types. Similarly,types. Similarly, the percent the bleedingpercent bleedingfor coarse for gradation coarse increasedgradation at increased higher Jnr atvalues. higher The Jnr di values.fference The of percent difference bleeding of percent between bleeding fine and betweencoarse gradations fine and coarse for the gradations high emulsion for the rate hi variedgh emulsion depending rate varied on the depending Jnr value. The on the percent Jnr value. bleeding The percentvalues forbleeding coarse values gradation for coarse were greatergradation than were those greater for fine than aggregates those for byfine 17%, aggregates 24%, and by 18%17%, for 24%, Jnr 1 andvalues 18% of for 0.5, Jnr 2.0, values and 5of kPa 0.5,− 2.0,, respectively. and 5 kPa−1, respectively.

Coatings 2019, 9, 670 8 of 13 Coatings 2019, 9, x FOR PEER REVIEW 8 of 13

Fine Gradation and High EAR (70% void filled) 80% 70% 60% 50% 40% CRS-2 30% CRS-2L % Bleeding 20% CRS-2P 10% 0% 0.5 2 5 -1 Jnr(kPa )

(a) Coarse Gradation and High EAR (70% void filled) 90% 80% 70% 60% 50% CRS-2 40% CRS-2L % Bleeding 30% 20% CRS-2P 10% 0% 0.5 2 5 -1 Jnr(kPa )

(b)

Figure 5. RelationshipRelationship between between J Jnrnr andand percent percent chip chip seal seal bleeding bleeding for for ( (aa)) fine ﬁne gradation gradation with with low low EAR EAR and ( b)) coarse coarse gradation gradation with high EAR.

The results for thethe highhigh EAREAR alsoalso showed showed less less sensitivity sensitivity to to asphalt asphalt emulsion emulsion modification modification types types (P (Pand and L) thanL) than to J nrtovalues. Jnr values. There There was was a noticeable a noticeable difference difference between between modified modified and unmodified and unmodified asphalt asphaltemulsions emulsions in terms in of terms percent of bleedingpercent bleeding at all values at all of values Jnr. These of J resultsnr. These indicate results that indicate Jnr is relatedthat Jnr tois relatedpercent to bleeding percent forbleeding both conventional for both conventional and modified and modified asphalt emulsion asphalt emulsion types and types that and the ethatffect the of effectmodification of modification is marginal is marginal when compared when compared at the same at Jthenr value. same RecallJnr value. that Recall the experiment that the experiment controlled controlledthe Jnr value the rather Jnr value than rather the temperaturethan the temperature in the MLWT in the comparisons. MLWT comparisons. As a result, As a modified result, modified asphalt 1 asphaltemulsions emulsions achieved achieved a Jnr value a J ofnr 5.0value kPa of− 5.0attemperatures kPa−1 at temperatures of 1.5–6.0 ◦ofC higher1.5–6.0 than°C higher the control. than the control.Furthermore, the results showed that coarse gradation provided more percent bleeding than fine aggregatesFurthermore, in all conditions. the results showed Since the that aggregate coarse grad shapeation of provided the coarse more aggregates percent wasbleeding more than angular, fine aggregatesthis resulted in inall nonuniformity conditions. Since of the the chip aggregate spread. sh Thisape canof the lead coarse to more aggregates stress concentration was more angular, on the thiscontact resulted area, in and nonuniformity this assumption of the can chip be verified spread. by This the can results lead shown to more above. stress concentration on the contactFigure area,6 andshows this a assumption composite summary can be verified of previously by the results presented shown results. above. The labels for this figure indicateFigure emulsion–gradation 6 shows a composite (F is fine,summary C is Coarse) of previously Jnr at 3.2 presented kPa. All highresults. EARs The showed labels higherfor this percent figure indicatebleeding emulsion–gradation than low EARs. Chip (Fseals is fine, with C lowis Coarse) EARshad Jnr at percent 3.2 kPa. bleeding All high in theEARs range showed of 21%–70%, higher percentwhile chip bleeding seals withthan highlow EARs. emulsion Chip rates seals had with percent low EARs bleeding had percent in the range bleeding of 30%–81%. in the range of 21%– 70%, while chip seals with high emulsion rates had percent bleeding in the range of 30%–81%.

Coatings 2019, 9, 670 9 of 13 Coatings 2019, 9, x FOR PEER REVIEW 9 of 13

90% 80% 70% 60% 50% 40% 35% 30% % Bleeding 70% 20% 10% 0% CRS-2 - C -… CRS-2L - F -… CRS-2L - C -… CRS-2P - F -… CRS-2P - C -… - CRS-2P CRS-2 - F 2 CRS-2 - F 5 CRS-2 - C 2 CRS-2 - C 5 CRS-2L - F - 2 CRS-2L - F - 5 CRS-2L - C - 2 CRS-2L - C - 5 CRS-2P - F - 2 CRS-2P - F - 5 CRS-2P - C - 2 C - - CRS-2P CRS-2P - C - 5 C - - CRS-2P CRS-2 - F 0.5 FigureFigure 6. EffectEffect of of EAR EAR on on bleeding bleeding resistance. 3.1. Factors that Influence Bleeding Resistance 3.1. Factors that Influence Bleeding Resistance The information presented in the previous figures identify many factors that have a potential The information presented in the previous figures identify many factors that have a potential impact on bleeding resistance, including aggregate gradation, EAR, and Jnr at 3.2 kPa. Statistical impact on bleeding resistance, including aggregate gradation, EAR, and Jnr at 3.2 kPa. Statistical analysis was used to quantify the significance of these factors and assess their relative contribution to analysis was used to quantify the significance of these factors and assess their relative contribution bleeding resistance. Prior to conducting the analysis, factors were added or modified to best represent to bleeding resistance. Prior to conducting the analysis, factors were added or modified to best the materials used. The material properties considered for the analysis included those obtained from represent the materials used. The material properties considered for the analysis included those MSCR testing (percent recovery at 3.2 kPa). Moreover, aggregate gradation for statistical analysis obtained from MSCR testing (percent recovery at 3.2 kPa). Moreover, aggregate gradation for was quantified by fitting the gradation curve to a cumulative Weibull distribution; parameters κ and statistical analysis was quantified by fitting the gradation curve to a cumulative Weibull distribution; λ denote the shape (fineness or coarseness) and the scale (dense or open/gap-graded) of aggregate parameters κ and λ denote the shape (fineness or coarseness) and the scale (dense or open/gap- type, respectively. graded) of aggregate type, respectively. The Weibull distribution was fitted to the gradation curves in order to calculate the gradation The Weibull distribution was fitted to the gradation curves in order to calculate the gradation shape parameters. The parameters calculated were κ and λ, which denote the shape (fineness or shape parameters. The parameters calculated were κ and λ, which denote the shape (fineness or coarseness) and the scale (dense or open/gap-graded) of aggregate type, respectively. These parameters coarseness) and the scale (dense or open/gap-graded) of aggregate type, respectively. These were determined using Equation (2). These parameters were used in the statistical analysis to evaluate parameters were determined using Equation (2). These parameters were used in the statistical the effects of gradation: analysis to evaluate the effects of gradation: ( x )κ P(x) = 1 e− λ (2) − ( ) 𝑃(𝑥) = 1 −𝑒 (2) where where P(x) = percent finer than sieve x,

x = sieveP(x) size in= percent millimeters finer tothan the sieve 0.45 power,x, and

λ, κ =x =curve sieve ﬁt size parameters in millimeters of shape to the and 0.45 scale, power, respectively. and

Analysis of varianceλ, κ (ANOVA)= curve fit wasparameters conducted of shape at a significance and scale, levelrespectively. of 0.1 to screen the significant factorsAnalysis affecting of bleeding.variance For(ANOVA) the analysis, was conducted the software at useda significance was R-project level [21 of]. The0.1 to definitions screen the of significanteach factor factors are as follows:affecting Gradation_bleeding. Forλ: Weibullthe anal distributionysis, the software parameter used which was R-project denotes the[21]. shape The definitions(fineness or of coarseness), each factor are where as follows: a high λGradation_value means λ: Weibull greater distribution coarseness; parameter EAR: EAR; which Jnr_3.2: denotes Jnr at 3.2 kPa; R_3.2: percent recovery at 3.2 kPa; Rep: replicates. The ANOVA results are shown in Table6. the shape (fineness or coarseness), where a high λ value means greater coarseness; EAR: EAR; Jnr_3.2: Jnr at 3.2 kPa; R_3.2: percent recovery at 3.2 kPa; Rep: replicates. The ANOVA results are shown in Table 6.

Coatings 2019, 9, 670 10 of 13

Table 6. ANOVA table for bleeding test.

Variables F-Value Pr(>F) Signiﬁcance Gradation_ λ 108.43 <0.001 *** EAR 104.00 <0.001 *** Jnr_3.2 139.74 <0.001 *** R_3.2 2.97 0.090 . Rep 0.51 0.479 Note: Signiﬁcance codes—0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1.

The results presented in Table6 indicate that Gradation_ λ, EAR, Jnr_3.2, and R_3.2 were statistically significant factors affecting bleeding resistance. The significant factors can be ordered by the F-value, and the F-values of Jnr_3.2, Gradation_ λ, EAR, and R_3.2 are the ordered significant factors related to bleeding resistance. Moreover, the results show that the replicate was not significant for the bleeding resistance, indicating that the results are reliable and the test method is repeatable.

3.2. Best Subsets Regression As the study selected the constant traﬃc stress, volume, and existing pavement, the factors emphasized included asphalt emulsion rheology, EAR, and aggregate gradation. Therefore, the measured bleeding was dependent on ﬁve factors, as stated in the previous section. However, to design the model for bleeding resistance, only four factors were considered when conducting the regression analysis. A best subsets regression was used to identify factors to include in a prediction model. The independent variables considered in this analysis included Gradation_ λ, EAR, Jnr_3.2, and R_3.2. Table7 shows the results of the best subsets analysis performed in the statistical analysis program 2 Minitab. The methodology used to choose the best subsets was based on a high Radj value and the close value of a low Mallows’s Cp and the number of variables in the chosen subset.

Table 7. Factor selection by best subset regression.

2 Variables Radj Mallows’s Cp Gradation_ λ EAR Jnr_3.2 R_3.2 1 49.2 272.4 X 2 68.5 142.3 X X 2 67.7 147.7 X X 3 87.5 17.7 X X X

Using the subset outlined in Table7, a quantitative prediction model was deﬁned, as shown in Equation (3):

% bleeding = 0.124 + 0.0316 Gradation_ λ + 0.410 EAR + 0.0628 J _3.2, R2 = 87.5% (3) − nr adj where % Bleeding = Estimated percent coated aggregates by total area,

Gradation_ λ = Weibull distribution parameter describing the shape (ﬁneness or coarseness),

EAR = Emulsion application rate, and

Jnr_3.2 = Jnr at 3.2 kPa. The model in Equation (3) was used to select the optimum asphalt emulsion, EAR, and aggregate gradation considering the constant existing pavement, traﬃc volume, traﬃc stress, and construction quality. In the best subset, the parameters which were included in the regression equation were Gradation_λ, EAR, and Jnr_3.2. There was no need to include percent recovery since there was a high Coatings 2019, 9, x FOR PEER REVIEW 11 of 13

Coatings 2019, 9, 670 11 of 13 correlation between Jnr and percent recovery. Also, it would be very difficult to control Jnr and percent recovery independently. correlationThe selected between model Jnr and (Equation percent recovery. (3)) showed Also, that it would bleeding be very is high diffi cultwhen to controlGradation_ Jnr andλ is percent coarse recovery(high) and independently. EAR and Jnr_3.2 are high. The coefficients for each factor showed that the most significant factorThe to selectedcause bleeding model (Equation is EAR, (3))followed showed by that Jnr_3.2 bleeding and isGradation_ high whenλ. Gradation_ The best λwayis coarse to use (high) this andequation EAR andis to J nrinput_3.2 arethe high. value The of coetheffi requiredcients for EAR each factorand the showed gradation that the(λ)most value significant and select factor the tomaximum cause bleeding value of is EAR,Jnr at the followed specific by climate Jnr_3.2 conditions and Gradation_ (pavemλ. Theent besttemperature) way to use that this will equation lead to is the to inputmaximum the value percent of the bleeding required allowed. EAR and The the equation gradation indicates (λ) value that and finer select aggregate the maximum gradation value and of lower Jnr at theEAR specific are also climate favorable. conditions (pavement temperature) that will lead to the maximum percent bleeding allowed.As stated The equation earlier, indicatesto verify thatthe finerequation aggregate for the gradation bleeding and resistance, lower EAR a new are also(null) favorable. experiment neededAs statedto be carried earlier, out to verify in the the MLWT. equation For forthe the null bleeding experiment, resistance, the asphalt a new emulsions (null) experiment and aggregate needed toselected be carried were out different in the MLWT.from the For main the nullexperiment experiment,. The theaggregate asphalt for emulsions this experiment and aggregate was limestone selected werewith di23%fferent LA abrasion from the and main 90% experiment. fractured Theface. aggregate for this experiment was limestone with 23% LA abrasionThe results and shown 90% fractured in Figure face. 7 indicate that the values from the equation were consistent with the valueThe results from the shown imaging in Figure analysis.7 indicate The thatlabel the in valuesthe plot from indicates the equation asphalt were emulsion, consistent Jnr at with 3.2 kPa, the valueand EAR. from theThis imaging indicates analysis. that all The factors label inin the the plot regression indicates analysis asphalt emulsion, significantly Jnr at affected 3.2 kPa, andbleeding EAR. Thisresistance. indicates that all factors in the regression analysis significantly affected bleeding resistance.

100%

90%

80%

70%

60% % Bleeding 50%

40%

30% Imaging analysis Regression Analysis

HFRS-2, 2 kPa-1, 35% HFRS-2, 2 kPa-1, 70% HFRS-2, 5 kPa-1, 35% HFRS-2, 5 kPa-1, 70%

HFRS-2P, 2 kPa-1, 35% HFRS-2P, 2 kPa-1, 70% HFRS-2P, 5 kPa-1, 35% HFRS-2P, 5 kPa-1, 70%

Figure 7. Comparison of measured percent bleeding (imaging analysis) and estimated percent bleeding Figure 7. Comparison of measured percent bleeding (imaging analysis) and estimated percent (regression analysis). bleeding (regression analysis). 4. Findings and Conclusions 4. Findings and Conclusions This study evaluated the relationship between asphalt emulsion residue Jnr values measured with theThis MSCR study standard evaluated protocol the relationship and chip seal between bleeding asphalt measured emulsion in the residue laboratory Jnr values using themeasured MLWT device.with the The MSCR following standard points protocol summarize and chip the seal findings: bleeding measured in the laboratory using the MLWT device. The following points summarize the findings: 1. The MSCR test is a promising evaluation tool for determining the effect of asphalt emulsion 1. The MSCR test is a promising evaluation tool for determining the effect of asphalt emulsion residue properties on the bleeding performance of chip seals. It is clear that a Jnr at 3.2 kPa value residue properties on the bleeding performance of chip seals. It is clear that a Jnr at 3.2 kPa value can differentiate between the effects of asphalt emulsions in terms of sensitivity to bleeding under can differentiate between the effects of asphalt emulsions in terms of sensitivity to bleeding different temperatures and cycles. As a result, it has the potential to be used to identify materials under different temperatures and cycles. As a result, it has the potential to be used to identify more prone to bleeding due to softening related to temperature. materials more prone to bleeding due to softening related to temperature. 2. The results indicate that the EAR, aggregate gradation, and Jnr at 3.2 kPa contribute to the bleeding 2. The results indicate that the EAR, aggregate gradation, and Jnr at 3.2 kPa contribute to the of chip seals. As expected, the highest impact observed was for the EAR when it was changed bleeding of chip seals. As expected, the highest impact observed was for the EAR when it was from 35% to 70%. However, for each EAR, the Jnr value of the asphalt emulsion residue at the test changed from 35% to 70%. However, for each EAR, the Jnr value of the asphalt emulsion residue temperature had the second largest effect. Thus, the laboratory results indicate that for a given at the test temperature had the second largest effect. Thus, the laboratory results indicate that EAR, bleeding performance can be controlled through proper evaluation of asphalt emulsions at for a given EAR, bleeding performance can be controlled through proper evaluation of asphalt

Coatings 2019, 9, 670 12 of 13

the project climate using MSCR testing. However, the existing pavement, traﬃc volume, traﬃc stress, and construction quality need to be taken into consideration for future work.

3. Asphalt emulsions modiﬁed with polymers have lower Jnr values and thus have less of a tendency to allow bleeding at the same temperature.

4. MSCR Jnr at a 3.2 kPa value is a property of asphalt emulsion residue that is only one of the factors inﬂuencing bleeding resistance measured by MLWT. 5. Temperature control can be improved by creating a chamber for controlling the temperature. 6. The limitation of this study is that only a single chip seal was evaluated for bleeding. Many diﬀerent multilayer seal treatments are currently used in practice, so further research is needed to evaluate if the concepts presented here can be applied to other seal systems. 7. Future research with additional technologies (i.e., emulsions and traditional testing) is needed for improving the validity of the testing.

Funding: This research received no external funding. Conﬂicts of Interest: No conﬂict of interest.

References

1. Shuler, S. Chip Seals for High Traffic Pavements. In Transportation Research Record 1259; Transportation Research Board, National Research Council: Washington, DC, USA, 1990; pp. 24–34. 2. Wegman, S. Design and Construction of Seal Coats; Final Report; Minnesota Department of Transportation, Mendota Heights: Mendota Heights, MN, USA, 1991; p. 34. 3. Gransberg, D.; James, D.M.B. NCHRP Synthesis 342: Chip Seal Best Practices; Transportation Research Board, National Research Council: Washington, DC, USA, 2005. 4. Gransberg, D.D.; Senadheera, S.; Karaca, I. Analysis of Statewide Seal Coat Constructability Review; Texas Department of Transportation; Research Report TX-98/0-1787-1R; Texas Tech University: Lubbock, TX, USA, 1998. 5. Shuler, S.; Lord, A.; Epps-Martin, A. Manual for Emulsion-Based Chip Seals for Pavement Preservation; NCHRP Report 680; Transportation Research Board, National Research Council: Washington, DC, USA, 2005. 6. Senadheera, S.P.; Khan, J.R. Standardized Seal Coat Terminology Pamphlet; TxDOT Research Implementation Study Number 5-1787; Center for Multidisciplinary Research in Transportation, Texas Tech University: Lubbock, TX, USA, 2001; p. 4. 7. Michael, C.; Lee, P.E. Seal Coat and Surface Treatment Manual; Texas Department of Transportation: Austin, TX, USA, 2003. 8. Hanz, A.; Johannes, P.; Bahia, H. Development of Emulsion Residue Testing Framework for Improved Chip Seal Performance. In Transportation Research Record: Journal of the Transportation Research Board, No. 2293; Transportation Research Board of the National Academies: Washington, DC, USA, 2012; pp. 106–113. 9. Kim, Y.R.; Adams, J.; Castorena, C.; Ilias, M.; Im, J.H.; Bahia, H.; Chaturabong, P.; Hanz, A.; Johannes, P. Performance-Related Specifications for Emulsified Asphaltic Binders Used in Preservation Surface Treatments; National Cooperative Highway Research Program (NCHRP) Research Report 837; The National Academies Press: Washington, DC, USA, 2017. [CrossRef] 10. Kim, Y.R.; Lee, S. Performance-Based Analysis of Polymer-Modified Emulsions in Asphalt Surface Treatments; Publication FHWA-NC-07-06; North Carolina State University. Department of Civil, Construction, and Environmental Engineering: Raleigh, NC, USA, 2009. 11. Chaturabong, P.; Hanz, A.; Bahia, H. Development of the Loaded Wheel Test (LWT) for Evaluating Bleeding in Chip Seals. In Transportation Research Record Journal of the Transportion Research Board; No. 2481; Transportation Research Board of the National Academies: Washington, DC, USA, 2015. 12. AASHTO T96. Standard Method of Test for Resistance to Degradation of Small-Size Coarse Aggregate by Abrasion and Impact in the Los Angeles Machine; American Association of State Highway and Transportation Officials (AASHTO): Washington, DC, USA, 2019. 13. ASTM D5821. Standard Test Method for Determining the Percentage of Fractured Particles in Coarse Aggregate; ASTM International: West Conshohocken, PA, USA, 2017. Coatings 2019, 9, 670 13 of 13

14. Johannes, P.; Mahmoud, E.; Bahia, H.U. Sensitivity of ASTM D7000 Sweep Test to EAR and Aggregate Gradation. In Transportation Research Record: Journal of the Transportation Research Board, No. 2235; Transportation Research Board of the National Academies: Washington, DC, USA, 2011; pp. 95–102. 15. ASTM D 7405. Standard Test Method for Multiple Stress Creep and Recovery (MSCR) of Asphalt Binder Using a Dynamic Shear Rheometer; ASTM International: West Conshohocken, PA, USA, 2015. 16. AASHTO TP 70. Standard Method of Test for Multiple Stress Creep Recovery (MSCR) Test of Asphalt Binder Using a Dynamic Shear Rheometer (DSR); American Association of State Highway and Transportation Officials (AASHTO): Washington, DC, USA, 2013. 17. AASHTO M 332. Specification for Performance-Graded Asphalt Binder Using Multiple Stress Creep Recovery (MSCR) Test; American Association of State Highway and Transportation Officials (AASHTO): Washington, DC, USA, 2014. 18. AASHTO PP 72. Standard Practice for Recovering Residue from Emulsified Asphalt Using Low-Temperature Evaporative Techniques; American Association of State Highway and Transportation Officials (AASHTO): Washington, DC, USA, 2011. 19. ASTM D7000. Standard Test Method for Sweep Test of Bituminous Emulsion Surface Treatment Samples; ASTM International: West Conshohocken, PA, USA, 2017. 20. Roohi, N.; Tashman, L.; Bahia, H.U. Internal Structure Characterization of Asphalt Mixtures for Rutting Performance Using Imaging Analysis. Road Mater. Pavement Des. 2012, 13, 21–37. 21. Venables, W.N.; Smith, D.M.; R Development Core Team. An Introduction to R; Version 2.9.2; R Foundation for Statistical Computing: Vienna, Austria, 2009.

© 2019 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).