Construction and Building Materials 49 (2013) 772–780

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier .com/locate /conbui ldmat

Laboratory evaluation of stabilizing methods for porous asphaltmixtures

0950-0618/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.conbuildmat.2013.08.076

⇑ Corresponding author. Tel.: +1 864 656 0374.E-mail addresses: krlyons@g.clemson.edu (K.R. Lyons), putman@clemson.edu

(B.J. Putman).

Kimberly R. Lyons, Bradley J. Putman ⇑Glenn Department of Civil Engineering, Clemson University, 109 Lowry Hall, Clemson, SC 29634, United States

h i g h l i g h t s

� Stabilizing additives should be incorporated into porous asphalt mixtures.� Cellulose fibers and crumb rubber were the most effective means of reducing draindown.� Cellulose fibers and crumb rubber improve the long-term draindown resistance.� Crumb rubber improved the abrasion resistance of the porous asphalt mixture.

a r t i c l e i n f o

Article history:Received 15 May 2013Received in revised form 20 August 2013Accepted 29 August 2013Available online 27 September 2013

Keywords:Open graded friction coursePorous asphaltRubber modified asphaltPavementPolymerFiberStabilizer

a b s t r a c t

The objective of this study was to compare the performance of different stabilizing additives (cellulosefibers, styrene–butadiene–styrene [SBS], and crumb rubber modifier [CRM]) in a porous asphalt mixture.The mix evaluation was based on draindown, permeability, abrasion resistance, moisture susceptibility,and rutting. The results emphasized the importance of stabilizing additives in porous asphalt mixtures.Further, the results indicated that the addition of fibers or CRM were the most effective at minimizingdraindown; CRM or the combination of fibers and SBS were the most effective at increasing the abrasionresistance; and the fibers had no effect on the mix strength.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Porous asphalt is an asphalt mixture with little or no fine aggre-gate. The reduced amount of fines creates interconnected, stableair pockets in the asphalt mix that allow water to flow throughthe mix. Porous asphalt is used across the world for two mainpavement applications: Wearing courses on high-speed roadwaysand for porous pavements for stormwater management. As a wear-ing course, a thin layer of porous asphalt ranging from 19 to 50 mmthick is placed over a conventional impermeable pavement surface[1]. This porous overlay, referred to as an open graded frictioncourse (OGFC) or porous friction course (PFC) has been shown toimprove roadway safety by allowing water to drain into the porouslayer and then flow laterally within the porous layer until it exitsthe pavement through a daylighted edge [2–6]. When used for

stormwater management purposes, a thicker porous asphalt layer(50–100 mm thick) is placed over an open graded aggregate basecourse that acts as a reservoir for stormwater before it infiltratesinto the underlying soil [7]. In these applications, the quantity ofstormwater runoff to be accommodated by conventional storm-water management infrastructure is significantly reduced due toinfiltration and the porous pavement structure acts as a filter dur-ing the process [8,9].

With all of the positive attributes of porous asphalt mixtures forspecific applications, there are two common problems that havecaused inconsistency in porous asphalt performance, specificallyin OGFC applications: Raveling and binder draindown [1,5,10].

Raveling is a pavement distress resulting from the loss of indi-vidual aggregate particles at the surface of a pavement due to a lossof adhesion between the binder and the aggregate or due to poorcohesion within the binder mastic coating the aggregate. In porousasphalt, this problem is thought to be caused by the open voidstructure typical of such a mix which increases the exposure ofthe binder film to air and the elements compared to traditional

Table 1Aggregate properties.

Property Value

Bulk Specific Gravity 2.60Bulk Specific Gravity (SSD) 2.62Apparent Specific Gravity 2.65Absorption 0.8%LA Abrasion Loss (C Grading) 29%

Table 2Aggregate gradation evaluated in this study.

Sieve size (mm) % Passing

19.0 10012.5 94.09.50 69.04.75 19.02.36 6.00.60 4.00.15 2.30.075 1.0

K.R. Lyons, B.J. Putman / Construction and Building Materials 49 (2013) 772–780 773

dense graded mixes. This increased exposure can lead to prema-ture oxidation of the binder, thus making it brittle and leading toraveling [11]. In addition to aging, other factors can lead to ravelingof OGFC mixtures including: Stripping, low binder content, dustcoated aggregates, aggregate gradation, compaction temperatureand effort, traffic frequency, studded tires, and others.

Draindown can be seen as excess asphalt binder that drains outof a porous asphalt mixture at temperatures typically used for pro-duction and construction and it is attributed to the lack of fines topromote permeability and the increased asphalt binder content toincrease durability [12]. In addition to production temperatures,draindown can also occur over time at high service temperatures(hot summer days) if the mix design does not include materialsto prevent draindown. In this case, the high pavement temperaturecauses the binder to soften and gradually flow downward throughthe void structure due to gravity until it reaches a cooler portion ofthe pavement (approximately 12.5 mm below the surface) where itthen stops flowing. This long-term draindown can reduce the per-meability of the porous asphalt [13].

If the binder content of a porous asphalt mixture is arbitrarilyincreased to improve durability without some other adjustmentto the mix design, a portion of the added binder will be lost dueto draindown. To design porous asphalt mixtures with higher bin-der contents, additives are typically incorporated to stabilize themix and prevent draindown. The stabilizing additives employedmost commonly in porous asphalt mixtures are polymers whichstiffen the asphalt binder and fibers which absorb the additionalbinder creating a thicker mastic around the aggregate particles thatis less susceptible to draindown [1,5].

Several studies have been conducted on stabilizing additives formixtures that are susceptible to draindown, such as porous asphaltand stone matrix asphalt (SMA). However, the majority have inves-tigated only one category of stabilizers, either polymer modifiers orfibers, but not both [14,15]. Brown et al. evaluated two differenttypes of polymers and two types of fibers in the same study anddetermined that the fibers and polymers were effective in reducinghigh temperature draindown of SMA mixtures, but the celluloseand rock wool fibers were significantly more effective stabilizersthan the styrene–butadiene–styrene (SBS) and polyolefin polymermodifiers used in the study [16]. Similar results were found in astudy by Stuart and Malmquist [17]. In addition to traditional poly-mers, crumb rubber derived from scrap tires has been used as abinder modifier resulting in stiffer binders having increased viscos-ity at high temperatures [18]. The stiffening effect of crumb rubberon asphalt binders can help to minimize or prevent draindownwhile increasing the durability of the mix [19,20].

1.1. Research objective

The primary objective of this study was to investigate the influ-ence of different stabilizing additives on porous asphalt mixtures.This was accomplished by comparing porous asphalt (or OGFC)mixtures made with three different types of stabilizing additives(SBS polymer, crumb rubber, and cellulose fibers). The study fo-cused on both the stabilizing effect of the additives as well as theinfluence on the performance properties of the mixtures.

2. Experimental materials and methods

To satisfy the objectives of this research, six different mixtures were preparedand evaluated using one aggregate source and gradation and three different stabi-lizing additives (SBS, crumb rubber, and cellulose fibers). Each mixture was evalu-ated to determine the influence of the stabilizing additive (or combination) ondraindown before measuring the performance properties of each mix.

2.1. Materials

2.1.1. AggregateFor this study, one crushed granite aggregate source was used for the compar-

ative investigation as granite is a common aggregate type used for asphalt mixturesin the Southeastern US. The properties of the aggregate used in this study are sum-marized in Table 1. The aggregate was sampled from the quarry and transported tothe lab where it was dried in an oven at 110 �C, then mechanically sieved into theindividual size fractions needed to produce the mix gradation in Table 2.

2.1.2. BinderIn total, four different binders were used in this study: A neat binder, a modified

binder made by adding 3% SBS to a PG 64-22 binder, a modified binder made byadding 5% crumb rubber to the neat binder (CRM5%), and a modified binder madeby adding 12% crumb rubber to the neat binder (CRM12%). The neat binder had aperformance grade of PG 64-22 and was obtained from an asphalt terminal. Thisneat binder was used by itself as one of the binder treatments in this study andit was also used as the base binder to produce the crumb rubber modified (CRM)binders. The SBS modified binder was obtained from the same terminal, however,it could not be confirmed if the source of the base binder used to produce the mod-ified binder was the same as the PG 64-22 neat binder used in this study. This SBSmodified binder is marketed as a PG 76-22 binder. The properties of all of the bind-ers used in the study are summarized in Table 3.

2.1.3. Crumb rubberThe crumb rubber modifier (CRM) used to produce the CRM binders was man-

ufactured by processing scrap passenger automobile tires using an ambient shred-ding operation at a tire processing facility. The crumb rubber was classified as a 30Mesh crumb rubber (Class 30-1) in accordance with ASTM D5603 [21]. This desig-nation means that the rubber has no more than 10% cumulative retained on the No.30 (0.6 mm) sieve and is a Grade 1 rubber (processed from whole passenger car,truck, and/or bus tires from which the fiber and metal have been removed). The gra-dation of the rubber as determined in accordance with ASTM D5644 is included inTable 4 [22].

2.1.4. Cellulose fibersCellulose fibers were included in this study as they are commonly used as a sta-

bilizing additive to minimize, or prevent draindown in porous asphalt mixtures. Thefibers were derived from post-consumer paper and shredded to a maximum fiberlength of 6 mm. When included in an asphalt mixture, the fibers were added at arate of 0.3% by total mixture weight [23].

2.1.5. Hydrated limeAll of the mixtures evaluated in this study incorporated hydrated lime as an

anti-stripping additive as required by many transportation agencies. The hydratedlime was added at a rate of 1% by weight of the aggregate [23]. After mixing withthe dried aggregate, water was added (5% of the aggregate weight) to activate thehydrated lime.

Table 3Properties of the binders used in this study.

Property Binder

PG 64-22 PG 76-22 (SBS) CRM5% CRM12%

Original binderViscosity @ 135 �C, Pa s 0.40 0.87 0.62 0.74Viscosity @ 165 �C, Pa s 0.25 0.42 0.26 0.54G⁄/sind kPa (test temp, �C) 1.35 (64) 1.86 (76) 1.72 (64) 1.30 (76)d, � (test temp, �C) 86.1 (64) 70.3 (76) 84.7 (64) 80.1 (76)Failure temperature, �Cc 66.4 82.9 68.7 78.6

RTFO Aged Residuea

G⁄/sind kPa (test temp, �C) 3.80 (64) 4.27 (76) 3.62 (64) 2.34 (76)d, � (test temp, �C) 80.5 (64) 63.8 (76) 65.6 (64) 69.0 (76)Failure Temperature, �Cc 68.3 83.5 72.4 76.5Mass Change, % 0.03 0.01 0.01 0.00

PAV Aged Residueb

G⁄sind kPa (test temp, �C) 4603 (25) 1299 (31) 2490 (25) 922 (31)d, � (test temp, �C) 47.8 (25) 45.9 (31) 49.8 (25) 42.9 (31)Stiffness @ �12 �C, MPa 97.4 132 108 104m-value @ �12 �C 0.354 0.360 0.352 0.324

Mixing temperature, �C 157 163 163 163Compaction temperature, �C 152 157 157 157

a Rolling Thin Film Oven.b Pressure Aging Vessel.c Temperature at which G⁄/sind was equal to 1.0 kPa for original aged binder or 2.2 kPa for RTFO residue.

Table 4Gradation of the 30 Mesh crumb rubber used to produce CRM binders.

Sieve size (mm) % Passing

1.19 1000.60 93.20.30 26.00.15 2.4

Table 5Mix designations and parameters used for draindown testing.

Mix Description Binder contentrange (%)

Test temperature(�C)

C PG 64-22 withoutfibers

4.5–6.5 157

SBS SBS modified withoutfibers

5.5–7.5 163

C-F PG 64-22 with fibers 4.5–6.5 157SBS-F SBS modified with

fibers5.5–7.5 163

CRM5% 5% CRM without fibers 5.0–7.0 163CRM12% 12% CRM without

fibers5.5–7.5 163

774 K.R. Lyons, B.J. Putman / Construction and Building Materials 49 (2013) 772–780

2.2. Experimental methods

2.2.1. CRM binder preparationTo prepare the two CRM binders used in this study the procedure developed by

Putman et al. was followed [24]. The procedure started by portioning out 600 g ofPG 64-22 base binder into metal cans having a volume of approximately 950 mL.Prior to adding the crumb rubber, the binder was heated to 182 �C and then placedon a sand bath on top of a hot plate. Then, a mechanical mixer equipped with a highshear radial flow impeller having a diameter of 50.8 mm began mixing the binder ata speed of 700 rpm for 30 min. Once the mixing started, the appropriate amount ofcrumb rubber was continuously added to the binder within approximately twominutes. Mixing continued for 30 min and the temperature was maintained at177 �C after the addition of the rubber.

2.2.2. Mixture preparationThe steps involved in preparing each specimen included: Aggregate batching,

mixing, and compaction. The aggregate batch for each specimen was prepared bycombining the appropriate quantities of each individual size fraction of aggregateto meet the mixture gradation included in Table 2, including 1% hydrated lime.Water was then added at a rate of 5% by weight of the aggregate batch and thenthe batch was thoroughly mixed before placing it in an oven heated to 177 �C over-night. This heating process ensured that the aggregate batch was completely driedand heated to promote adequate mixing.

After heating, the aggregate was removed from the oven and placed in the mix-ing bowl. If the mixture required fibers, the appropriate amount of fibers was addedand distributed in the aggregate by hand mixing with a spoon. The required amountof binder, heated to the appropriate mixing temperature (Table 3), was then addedand mixed either by hand or by mechanical means, depending on the type of spec-imen being prepared. For small specimens, including those used for draindown test-ing and determination of optimum binder content, mixing was completed by handand for larger specimens that were compacted for performance testing, a bucketmixer was used for mixing.

After mixing, the loose mix was conditioned in an oven heated to the appropriatecompaction temperature (Table 3) for two hours prior to compaction. After condi-tioning, 150 mm diameter specimens having a height of 115 ± 5 mm were com-pacted using 50 gyrations of a Superpave gyratory compactor. Once the specimens

were compacted, the mold was placed in front of a fan to cool for approximately25 min before removing the compacted specimen. This was done to prevent thespecimens from collapsing after extraction. A total of twelve compacted specimenswere made for each mix at the respective optimum binder content.

The conditioning/compaction process was only used for the compacted speci-mens and not the loose specimens used for draindown testing and optimum bindercontent (OBC) determination.

2.2.3. DraindownDraindown testing was performed on all the mixes in accordance with AASHTO

T305 with the exception that the tests were only performed at the mixing temper-ature for each mix (Tables 3 and 5) [25]. This testing procedure entailed measuringthe binder lost from the loose mix placed in a draindown basket (No. 4 mesh) andconditioned at the mixing temperature for 3 h with the draindown being measuredevery hour. Two draindown specimens were tested per binder content over a bindercontent range (Table 5). This test showed the amount of binder draindown relativeto the total weight of the mix. A maximum draindown of 0.3% by weight of total mixis typically the maximum value for draindown of a porous asphalt mix [5,26].

2.2.4. Optimum binder content determinationOnce the draindown testing was completed, the OBC for each mixture was

determined in accordance with SC-T-91, the South Carolina DOT (SCDOT) testmethod for designing OGFC mixtures [27]. The loose asphalt mixture was placedinto a 229 mm by 330 mm clear Pyrex dish having a depth of 50 mm. The mixturewas placed inside an oven at the appropriate temperature (Table 5) for two hours.After two hours, the OBC was determined by judging the appearance of the asphaltthrough the Pyrex dish. The OBC for a porous asphalt mix design is that at whichthere is neither excessive binder draindown, nor filling of uncompacted voids inthe mix, nor so little binder that the binder film is too thin.

Fig. 1. Schematic of the permeability test setup.

K.R. Lyons, B.J. Putman / Construction and Building Materials 49 (2013) 772–780 775

2.2.5. PorosityThe porosity of each of the compacted specimens was determined using the

procedure originally outlined by Montes et al. [28]. In this study, the effectiveporosity was calculated using Eq. (1) and used for the basis of comparison. Theeffective porosity only accounts for the water accessible air voids while the totalporosity (or air void content) accounts for all the air voids within a specimen.The effective porosity was used because only the accessible voids will contributeto the permeability of the specimens. In this procedure the volume of the specimenis determined by measuring the diameter and height in at least three different loca-tions to determine the average diameter and height. These dimensions were thenused to calculate the total volume (VT) of the cylindrical specimen. Next, the drymass of the specimen was recorded (Mdry) and the specimen was then submergedin a 25 �C water bath for 30 min. After 30 min, the specimen was inverted 180�and tapped on the bottom of the water bath five times to release air entrapped inthe specimen. Tapping was done in such a manner that it was firm enough to re-lease entrapped air while being careful not to damage the specimen. After tapping,the specimen was inverted 180� again and the submerged mass was recorded(Msub). The porosity of each specimen was calculated using Eq. (1), where qw isthe density of water.

Porosityð%Þ ¼ 1�ðMdry�MsubÞ

qw

VT

24

35� 100 ð1Þ

Upon completion of the porosity testing, the twelve specimens from each mixdesign were divided into four groups for further testing (3 specimens for aged abra-sion resistance, 4 specimens for indirect tensile strength (ITS), 3 specimens for ruttesting, and 2 specimens for long-term draindown). The porosity data was used togroup the specimens to ensure that each group was representative of the overallmix design properties. To verify that the porosity of the four test groups was similar,an analysis of variance (ANOVA) was performed using a = 0.05.

2.2.6. PermeabilityThe permeability of all of the specimens was measured using the falling-head

procedure previously used to measure the permeability of porous asphalt mixtures[29,30]. The first step to preparing a specimen was to wrap the specimen in plasticwrap around the sides to force the water to exit through the bottom of the speci-men instead of out the perimeter of the specimen. A piece of clear tape was thenplaced along the top of the specimen and folded over with the sticky side facingout, so that once the specimen was in the stand pipe, water could not flow betweenthe specimen and standpipe. The specimen was then placed into the standpipe(Fig. 1) and plumbers putty was applied to the outer edge of the tape to prevent

any water leakage between the standpipe and the specimen. The water outletwas located at the same elevation as the top of the specimen and the permeametersetup was leveled.

After the specimen was secured in the standpipe, the specimen was initially sat-urated with water by filling the outlet pipe. The standpipe was then filled withwater to approximately 375 mm above the top of the specimen, and the valve atthe bottom of the specimen was opened to allow the water to flow through thespecimen. The time required for the water to fall from a level of 300 mm abovethe specimen (h1) to 75 mm above the specimen (h2) was recorded using a stop-watch and repeated three times per specimen. The average time (t) was then usedto calculate the permeability of each specimen using Eq. (2), where A is the cross-sectional area of the specimen, a is the cross-sectional area of the stand pipe, andL is the height of the specimen.

Permeability ¼ aLAt

lnh1

h2

� �ð2Þ

2.2.7. Abrasion resistanceThe abrasion resistance of each mixture was measured using the Cantabro abra-

sion test after aging for 7 days in an oven at 60 �C. Three compacted specimens wereused for this test. The test was conducted by recording the initial mass (Mi) of aspecimen then placing it in the Los Angeles abrasion drum for 300 revolutions with-out the steel charge at 25 �C. After 300 revolutions, the specimen was removed,brushed off, and the final mass (Mf) was recorded. The percent mass loss was thencalculated by dividing the mass loss (Mi � Mf) by the initial mass (Mi) of thespecimen.

2.2.8. Indirect tensile strength and moisture susceptibilityThe indirect tensile strength (ITS) of each mixture was measured after dry and

wet conditioning. Two compacted specimens from each mixture were conditionedin air at 25 �C for 24-h prior to testing the ITS (ITSdry) and two specimens were sub-merged in 60 �C water for 24-h followed by 1-h in 25 �C water before testing the ITS(ITSwet). Because of the porous nature of these specimens, it was not possible to at-tain the level of saturation required by typical test procedures similar to AASHTOT283 [31]. Therefore, the saturation step was omitted with the reasoning that thespecimen would easily become saturated just by soaking in the water bath. TheITS was measured by loading the specimens across the diametral plane and record-ing the peak load, then calculating the ITS.

In addition to assessing the strength of the mixtures, the potential for moistureinduced damage was also determined based on the tensile strength ratio (TSR). TheTSR is calculated as the ratio of the wet ITS to the dry ITS (ITSwet/ITSdry � 100%). Theminimum value for the TSR of porous asphalt mixtures varies by agency, but is typ-ically required to be greater than or equal to 80% [5,26].

2.2.9. Rutting resistanceThe Asphalt Pavement Analyzer (APA) was used to evaluate the potential for

rutting of the porous asphalt mixes in accordance with ASSHTO T340 using threespecimens from each mix [32]. Because the specimens were compacted based onthe number of gyrations not height, the specimens were cut to the required heightof 75 ± 2 mm. APA testing was conducted at a test temperature of 64 �C and thespecimens were conditioned in the APA chamber for 4 h prior to testing. The hosepressure was maintained at 689 kPa and a wheel load of 445 N. The test was run for8000 cycles after which the rut depths were measured manually.

2.2.10. Long-term draindownThe potential for binder draindown due to exposure to elevated temperatures

over long periods of time was evaluated using specimen porosity and permeabilityas a basis for comparison. Two specimens from each mixture were conditioned in a60 �C chamber for 56-days and the porosity and permeability of each specimen wasmeasured at intervals (7, 14, 28, 42, and 56 days) to determine the occurrence ofbinder draindown through the specimen, which would reduce the permeability.

During conditioning, each specimen was wrapped using a galvanized steelhardware cloth to prevent deformation or collapse due to the high temperatureconditioning. Additionally, the top of the specimen remained the top for the entire56-day conditioning period, so the binder would only be able to flow in onedirection.

3. Results and discussion

3.1. Draindown testing

Draindown is typically used as an indication of the effectivenessof stabilizing additives and most specifications limit the draindownof a porous asphalt mixture to a maximum of 0.3% [26]. The drain-down of each mixture was measured in one hour increments over athree hour period. This was done to determine how the draindown

(b)(a)

(d)(c)

(f)(e)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Dra

indo

wn

(%)

Binder Content (%)

1 Hour

2 Hours

3 Hours

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Dra

indo

wn

(%)

Binder Content (%)

1 Hour

2 Hours

3 Hours

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Dra

indo

wn

(%)

Binder Content (%)

1 Hour

2 Hours

3 Hours

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Dra

indo

wn

(%)

Binder Content (%)

1 Hour

2 Hours

3 Hours

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Dra

indo

wn

(%)

Binder Content (%)

1 Hour

2 Hours

3 Hours

0.0

0.5

1.0

1.5

2.0

2.5

3.0

4 5 6 7 8 4 5 6 7 8

4 5 6 7 8 4 5 6 7 8

4 5 6 7 8 4 5 6 7 8

Dra

indo

wn

(%)

Binder Content (%)

1 Hour

2 Hours

3 Hours

Fig. 2. Draindown results for (a) Mix C, (b) Mix C-F, (c) Mix SBS, (d) Mix SBS-F, (e) Mix CRM5%, and (f) Mix CRM12%. The dashed line shows the maximum limit of 0.3%draindown.

5.05.5 5.5

7.0

5.5

7.0

0

1

2

3

4

5

6

7

C C-F SBS SBS-F CRM5% CRM12%

Opt

imum

Bin

der

Con

tent

(%

)

776 K.R. Lyons, B.J. Putman / Construction and Building Materials 49 (2013) 772–780

rate changed over time. Draindown is typically only measuredafter one hour of conditioning.

The draindown curves are included in Fig. 2. The results showthat modifying porous asphalt mixes by adding cellulose fibers orpolymers (SBS or crumb rubber) does help to reduce binder drain-down. While mixes SBS and CRM5% were polymer modified (withSBS or CRM, respectively) the level of modification alone was notsufficient to completely prevent draindown at higher binder con-tents over longer durations. The addition of cellulose fibers (C-Fand SBS-F) and the higher dosage of crumb rubber (CRM12%) werethe most effective in eliminating draindown at the binder contentstested even after three hours at the respective mixing temperature.

Mix Type

Fig. 3. Optimum binder content of porous asphalt mixtures.

3.2. Optimum binder content

The optimum binder content of any asphalt mixture is animportant factor in the ultimate mixture performance and this isespecially true for porous asphalt mixtures. If the binder contentis too low for a given mix, then it will be susceptible to prematureraveling and cracking. If it is too high, then excess binder coulddraindown during storage and hauling prior to placement, or aftercooling, the binder could draindown over time, which can sealvoids and reduce the permeability of the mix. The optimum bindercontents of the mixes evaluated in this study are included in Fig. 3.The results demonstrate the effect that the stabilizing additives

(SBS, crumb rubber, and fiber) have on increasing the OBC and,therefore, the binder film-thickness around the aggregate structurein the porous asphalt mixes. Because the aggregate gradation is thesame for each mix, the higher binder content provides an indirectindication of a thicker binder film. Based on the results, it is evidentthat the addition of fiber, SBS, or 5% CRM to the control mixtureindividually slightly increases the binder content (and binder filmthickness) without excessive draindown as seen in Fig. 3. However,when the fibers were combined with the SBS or the crumb rubber

16.8

13.116.2

11.6

16.714.4

0

2

4

6

8

10

12

14

16

18

20

C C-F SBS SBS-F CRM5% CRM12%

Por

osit

y (%

)

Mix Type

A D A BA C

Fig. 4. Average porosity of the porous asphalt mixtures. Mixes with a commonletter are not significantly different based on the test for least significant differencefor a = 0.05. Error bars indicate one standard deviation from the mean.

33

1117

10

0

5

10

15

20

25

30

35

40

45

C C-F SBS SBS-F CRM5% CRM12%

Age

d A

bras

ion

Los

s (%

)

Mix Type

A A A B

32 32

B B

Fig. 6. Average abrasion loss for aged porous asphalt mixtures. Mixes with acommon letter are not significantly different based on the test for least significantdifference for a = 0.05. Error bars indicate one standard deviation from the mean.

K.R. Lyons, B.J. Putman / Construction and Building Materials 49 (2013) 772–780 777

content was increased to 12%, the binder film increased signifi-cantly as indicated by the 2% increase in OBC compared to mix Cthat had no stabilizing additives. Due to the addition of the stabi-lizers, these mixtures did not exhibit any draindown even thoughthe binder content was significantly greater than the other mixes.

3.3. Porosity

Fig. 4 shows the porosity results for each of the mixtures. Theporosity of mixtures C, SBS, and CRM5% was statistically similar.As the mixes became more stable from a draindown perspectivedue to additional stabilizing additive, the porosity of the com-pacted mixture decreased. The addition of cellulose fiber (0.3% inthis study) had the most significant impact on the porosity asmixes C-F and SBS-F had porosity values that were 22% and 28%lower than their fiberless counterparts (mixes C and SBS), respec-tively. The CRM12% mix had a porosity reduction of approximately14% compared to the control mix C.

These results indicate that the addition of stabilizers to thepoint where the draindown is negligible over a reasonable rangeof binder contents will reduce the porosity of the mix to differentdegrees depending on the stabilizing additive(s). This reduction inporosity is due to the increased thickness of the binder film (bindercontent), and thus the reduction in pore size and/or continuity ofchannels connecting voids within the mix. Increasing the CRM con-tent to 12% reduced the porosity to a lesser degree than adding cel-lulose fibers alone and the combination of the more viscous SBSmodified binder with the cellulose fibers resulted in the greatestreduction in porosity.

764 713828

411589

421

0

200

400

600

800

1000

1200

C C-F SBS SBS-F CRM5% CRM12%

Pea

mea

bilit

y (m

/day

)

Mix Type

D DAB B A C

Fig. 5. Average permeability of porous asphalt mixtures. Mixes with a commonletter are not significantly different based on the test for least significant differencefor a = 0.05. Error bars indicate one standard deviation from the mean.

3.4. Permeability

The permeability of each mix design was evaluated using thefalling head test method and the results are included in Fig. 5.The addition of both fibers and crumb rubber had a tendency to re-duce the mix permeability. The results show that the mixes withthe highest binder content (mixes SBS-F and CRM12%) had thelowest permeability values and the results were statistically simi-lar. While the remaining four mixtures had relatively similar bin-der contents, there were differences in permeability. The additionof fibers in the C-F mixture did not have as significant of an effecton reducing the permeability as the addition in the SBS-F mixture.This is potentially due to the increased viscosity of the SBS modi-fied PG 76-22 binder compared to the PG 64-22 binder, as wellas the difference in binder content. Finally, the addition of crumbrubber also resulted in a permeability reduction of the porous as-phalt mixes and as the rubber content increased, the permeabilitydecreased. This was also likely due to the increase in viscosityresulting from the addition of the crumb rubber, which led to anincrease in optimum binder content. As the binder content in-creases, the binder film thickness increases, which reduces thepore size and restricts the flow channels connecting pores withinthe mix.

3.5. Abrasion resistance

The raveling resistance of the porous asphalt mixes was evalu-ated using the Cantabro abrasion test. It is recommended that theabrasion loss from this test should not exceed 20% for unaged spec-imens and 30% for aged specimens [5]. In this study, the specimenswere aged for 7 days in a 60 �C oven before testing and the resultsare summarized in Fig. 6.

The results indicate that the additives, or combination of addi-tives have a significant influence on the abrasion resistance of themixes. Three of the mixtures (C, C-F, and SBS) had average abrasionloss values that were statistically similar to each other and greaterthan the maximum recommended limit of 30%. The remainingmixtures (SBS-F, CRM5%, and CRM12%) were also statistically sim-ilar to each other, but the abrasion loss values were approximately50–70% lower than the other three. These results suggest that neat(unmodified) binders may not produce porous asphalt mixturesthat are resistant to raveling, even with cellulose fibers. The mixmade with SBS modified binder also did not perform to the recom-mended standard without fibers. However, when cellulose fiberswere added to the SBS mix (SBS-F), the abrasion loss was reducedby approximately 70% compared to the SBS mixture. This couldpotentially be due to the 1.5% increase in binder content of theSBS-F mix, which results in a thicker, more durable film around

279328

412486

340

512

282 297416 462

215 240

0

100

200

300

400

500

600

700

C C-F SBS SBS-F CRM5% CRM12%

Indi

rect

Ten

sile

Str

engt

h (k

Pa)

Mix Type

Conditioned Dry

C B C B ABC A AB A B C A B

Fig. 7. Average indirect tensile strength (ITS) of dry and conditioned specimens.Mixes with a common letter in either the dry or conditioned state are notsignificantly different based on the test for least significant difference for a = 0.05.Error bars indicate one standard deviation from the mean.

8.6

5.66.7

4.8

7.9 8.2

0

2

4

6

8

10

12

C C-F SBS SBS-F CRM5% CRM12%

Rut

Dep

th (

mm

)

Mix Type

A AB AB B AB AB

Fig. 9. Comparison of average rut depth values of porous asphalt mixtures. Mixeswith a common letter are not significantly different based on the test for leastsignificant difference for a = 0.05. Error bars indicate one standard deviation fromthe mean.

778 K.R. Lyons, B.J. Putman / Construction and Building Materials 49 (2013) 772–780

the aggregate particles that increases the cohesion within the mix.The addition of crumb rubber to the porous asphalt mixtures alsosignificantly improved the abrasion resistance. The binder contentof the CRM12% mix was the same as the SBS-F mix (7.0%), which isone likely reason for the increased abrasion resistance. However,the CRM5% mix also had a significantly lower abrasion resistancecompared to mixes C, C-F, and SBS even though the binder contentwas similar (5.5%). This indicates that the CRM binder potentiallyenhances the mix cohesion to a greater extent than the SBS binderin this study.

3.6. Indirect tensile strength and moisture susceptibility

The moisture susceptibility for each of the six mix designs wasbased on the retained indirect tensile strength, or tensile strengthratio (TSR), in accordance with SC-T-70 [27]. For agencies having aspecification for TSR of porous asphalt, the minimum value is typ-ically 80% [26]. This test was used not only to determine whethereach mix was susceptible to moisture induced damage, but alsoto compare the indirect tensile strengths of each mixture, whichcould be an indication of the cracking resistance of the porous as-phalt mixes.

Fig. 7 compares the indirect tensile strength (ITS) of the dry andconditioned specimens. The results indicate that the different sta-bilizing additives had different effects on the ITS. The cellulose fi-bers did not have any influence on the ITS for either the dry orconditioned specimens. This is evident by the fact that the ITS val-ues for Mixes C and C-F were statistically similar to each other andMixes SBS and SBS-F were statistically similar to each other. Thepolymer modifiers (SBS and CRM) did have a significant effect on

99 110 99 105

158

213

0

50

100

150

200

250

C C-F SBS SBS-F CRM5% CRM12%

Tens

ile S

tren

gth

Rat

io (

%)

Mix Type

Fig. 8. Tensile strength ratio (TSR) of porous asphalt mixtures.

the ITS as the SBS, CRM5%, and CRM12% exhibited higher ITS valuesthan the control, Mix C. The results also indicated that the dry ITSvalues for the mixtures containing crumb rubber (CRM5% andCRM12%) were significantly lower than the conditioned ITS andwere closer to mixes C and C-F. In addition, the amount of CRMadded to the binders influenced the ITS. As the crumb rubber con-tent increased from 5% to 12%, the ITS values increased signifi-cantly and the CRM12% mix had a similar wet conditioned ITScompared to the SBS mixtures.

Fig. 8 compares the TSR values of the different mix designs. Theresults of this study show that the TSR values of all of the mixtureswere well above the typical minimum value of 80%, which meansthat these mixes are expected to be resistant to moisture induceddamage. The TSR values for mixes C, C-F, SBS, and SBS-F were closeto 100% indicating that the conditioning regime used in this studydid not have a major effect on the ITS. The TSR values of the CRMmixes were well above 100%, which indicates that the conditioningactually improved the strength of the mixes and the effect wasmore pronounced as the crumb rubber content increased from0% (mix C) to 5% (mix CRM5%) to 12% (mix CRM12%). In this case,the addition of crumb rubber to the mix had more effect on theconditioned ITS and TSR than the addition of cellulose fibers, whichis evident by comparing the results of mixes CRM5% and CRM12%to those of mix C-F.

3.7. Rutting resistance

Rutting resistance is a potentially important characteristic forporous asphalt pavements because of the high air void content ofthe mixture. The higher air void content in porous asphalt allowsaggregate particles to shift their location when subjected to vehic-ular loading causing the porous asphalt to rut. For OGFC pave-ments, however, rutting resistance is not as much a factor due tothe thin lifts (19–38 mm) typically used in such applications.Fig. 9 summarizes the average rut depths of the mixtures evaluatedin this study after 8000 cycles of the Asphalt Pavement Analyzer(APA). The results showed a relatively high degree of variabilityas indicated by the error bars that show one standard deviation.Because of this variability, the rut depths of all of mixes were sta-tistically similar except for mixes C and SBS-F. Ignoring the resultsof the statistical analysis, the fibers generally had a largest impacton increasing the resistance to rutting as mixes C-F and SBS-Fexhibited the lowest rut depths in this laboratory study. Addition-ally, the CRM mixes did not have a significant effect on the ruttingresistance as measured by the APA.

0

5

10

15

20

25

0 14 28 42 56

Por

osit

y (%

)

Time (Days)

C C-F SBS SBS-F CRM5% CRM12%

Fig. 10. Average porosity measured at intervals during long-term conditioning at60 �C.

0

100

200

300

400

500

600

0 14 28 42 56

Per

mea

bilit

y (m

/day

)

Time (Days)

C C-F SBS SBS-F CRM5% CRM12%

Fig. 11. Average permeability measured at intervals during long-term conditioningat 60 �C.

K.R. Lyons, B.J. Putman / Construction and Building Materials 49 (2013) 772–780 779

3.8. Long-term draindown

Two specimens from each mix design were conditioned at 60 �Cfor 56 days to evaluate the susceptibility of each mix to binderdraindown when exposed to high temperatures while in serviceover an extended period of time. The long-term aging could poten-tially decrease the porosity and permeability of the specimen dueto binder draindown clogging the pores of the structure over time.The porosity and permeability for each specimen were measuredafter conditioning intervals of 7, 14, 28, 42, and 56 days.

The results of the porosity testing of the long-term conditioningprocedure are plotted with respect to time in Fig. 10. The resultsshow that the porosity of the mixes actually increased to some de-gree over time when exposed to high temperatures and the great-est change in porosity generally occurs in the first 7 days. Therewas no noticeable trend for the change in porosity with respectto the stabilizing additive.

In Fig. 11, the average permeability with respect to the condi-tioning duration is presented for each mixture. It was expectedthat the permeability would decrease to some extent when ex-posed to high temperatures for a long duration of time becausethe binder may drain downward within the specimen cloggingthe voids in the bottom portion of the specimen, thus reducingthe permeability. The results indicate that the addition of the dif-ferent modifiers used in this study did have some effect on thelong-term binder draindown as measured by permeability. Mix C(virgin PG 64-22 binder without fibers) had the greatest reductionin permeability over time as the permeability decreased 54.5% onaverage over the 56 day duration. The SBS modified binder resultedin a slight improvement as mix SBS showed a 48.5% reduction overthe duration. The cellulose fiber did show a benefit in reducing thepermeability loss, but was more effective with the neat binder (mixC-F) than the SBS modified binder (mix SBS-F) as mix C-F showed a

31.3% reduction in permeability and mix SBS-F reduced 43.3%. Theaddition of crumb rubber to the binder had the most significant ef-fect on minimizing the long-term draindown as the permeability ofmixes CRM5% and CRM12% showed a permeability reduction of35.9% and 11.6% on average, respectively. Additionally, whenstudying the curves in Fig. 11, the mixes without the fibers orcrumb rubber (Mixes C and SBS) showed a relatively high rate ofpermeability reduction after 14 days (approximately 20 m/day/day) compared to the mixes containing fibers or crumb rubber(Mixes C-F, SBS-F, CRM5%, and CRM12%), which showed a perme-ability reduction rate between 1 and 2 m/day/day during the sameperiod. After 28 days, the rate of permeability reduction becamemore consistent for each mix through the end of the 56 days. Ahigh rate of permeability reduction could be detrimental to thelong-term functionality of a porous pavement.

The results of the long-term draindown testing indicate that theporosity slightly increases over time when exposed to high temper-atures, which would lead one to expect the permeability to increaseas well. However, the permeability of all mixtures did show somereduction in permeability despite an increase in porosity. The rea-son for this is likely related to the test procedures. When measuringthe porosity, water was able to enter the specimen from surfacevoids around the entire exterior of the specimen. However, to mea-sure the permeability, water is only able to enter the specimenthrough the top surface of the specimen. Therefore, the porositymeasurement does not capture the effect of binder migrating down-ward in some of the mixes that are susceptible to long-term drain-down. It is hypothesized that as the binder drains downward withinthe specimen over time, it causes a porosity reduction in a lowerportion of the specimen, which restricts the vertical movement ofwater through the specimen, thus reducing the permeability.

4. Conclusions

The objective of this study was to evaluate the effects of differ-ent stabilizing additives on the behavior of porous asphalt mix-tures. Based on the results of this study, it can be concluded thatthe stabilizing additives evaluated (cellulose fibers, SBS, and crumbrubber) affected the porous asphalt mixture to different degrees.When considering the two main characteristics of interest withporous asphalt mixtures, draindown and raveling, it is evident thatstabilizing additives should be incorporated into porous asphaltmixtures. While the addition of fiber and/or polymer modifiers inthis study resulted in a reduction in porosity and permeability,the reduction was offset by the gains in other properties (drain-down resistance, abrasion resistance, and strength).

With respect to mitigating draindown at the high temperaturesexperienced during production and construction, all of the stabiliz-ing additives evaluated in this study were effective in maintainingdraindown levels below the recommended maximum of 0.3% atthe optimum binder content (OBC) for one hour. However, theaddition of cellulose fibers and crumb rubber were the most effec-tive means of reducing draindown, even at binder contents abovethe OBC and up to three hours at the mixing temperature. Increas-ing the crumb rubber content from 5% to 12% resulted in a reduc-tion in draindown.

The draindown of each mixture was also evaluated at elevatedpavement service temperature (60 �C) and the results indicatedthat the SBS alone was not sufficient to significantly reduce thelong-term draindown for the conditioning regime. The use of cellu-lose fiber and crumb rubber, however, did significantly improvethe long-term draindown resistance compared to the control andSBS mixes. The results indicate that the crumb rubber modifierwas the most effective of the stabilizing additives evaluated atminimizing long-term draindown.

780 K.R. Lyons, B.J. Putman / Construction and Building Materials 49 (2013) 772–780

The use of crumb rubber was effective in enhancing the abra-sion resistance of the porous asphalt mixture indicating that itwas more resistant to raveling than the mixture containing theSBS modifier alone. However, when the SBS modified binder wascombined with the cellulose fiber, the combination exhibited asubstantial increase in abrasion resistance that was similar to themixes containing crumb rubber. The cellulose fibers alone (withoutpolymer modified binder) were not effective in improving theabrasion resistance of the porous asphalt mixture.

The addition of cellulose fibers did not have any influence onthe indirect tensile strength of the porous asphalt mixtures in-cluded in this study. The polymer modifiers (SBS and CRM), how-ever, did significantly improve the strength of the mixes asmeasured by the ITS. The crumb rubber content did affect theamount of strength gain as strength increased with crumb rubbercontent from 0% to 12%.

Acknowledgement

The authors wish to acknowledge to the Asphalt Rubber Tech-nology Service (ARTS) at Clemson University for the financial sup-port of this study.

References

[1] Cooley LA, Brumfield JW, Mallick RB, Mogawer WS, Partl M, Poulikakos L, et al.Construction and maintenance practices for permeable friction courses. ReportNumber 640. National Cooperative Highway Research Program. WashingtonD.C.; 2009.

[2] Brunner RJ. Open-graded asphalt friction course. Pennsylvania DOT ResearchProject 74-8; 1975.

[3] Maupin GW. Virginia’s experience with open-graded surface mix. In: AnnualMeeting of the Transportation Research Board. Transportation Research Boardof the National Academies, Washington, D.C.; 1976.

[4] Chaignon F. Porous asphalt in France. The Asphalt Year Book. The Institute ofAsphalt Technology, UK; 1993.

[5] Kandhal P. Design, construction, and maintenance of open-graded asphaltfriction courses. Information Series 115. National Asphalt PavementAssociation. Lanham, MD; 2002.

[6] Poulikakos LD, Partl MN. Evaluation of moisture susceptibility of porousasphalt concrete using water submersion fatigue tests. Constr Build Mater2009;23:3475–84.

[7] Jackson N. Design, construction and maintenance guide for porous asphaltpavements. Information Series 131. National Asphalt Pavement Association;2003.

[8] Brattebo BO, Booth DB. Long-term stormwater quantity and qualityperformance of permeable pavement systems. Water Resour2003;37(18):4369–76.

[9] Sansalone JJ, Buchberger SG. An infiltration device for a best managementpractices for immobilizing heavy metals in urban highway runoff. Water SciTechnol 1995;32(1):119–25.

[10] Alvarez AE, Epps Martin A, Estakhri C, Izzo R. Evaluation of durability tests forpermeable friction course mixtures. Int J Pave Eng 2010;11(1):49–60.

[11] Kandhal P, Mallick R. Open-graded friction course: state of the practice. ReportNumber E-C005. Transportation Research Circular; 1998.

[12] Huber G. Performance survey on open-graded friction course mixes. NCHRPSynthesis of Highway Practice 284. Transportation Research Board, NationalResearch Council. Washington, D.C.; 2000.

[13] Ferguson BK. Porous pavements. CRC Press; 2005.[14] Cooley LA, Brown ER, Watson DE. Evaluation of OGFC mixtures containing

cellulose fibers. Report No. 2000-05. National Center for Asphalt Technology,Auburn, AL; 2000.

[15] Putman BJ, Amirkhanian SN. Utilization of waste fibers in stone matrix asphaltmixtures. Resour Conserv Recycl 2004;42:265–74.

[16] Brown ER, Haddock JE, Mallick RB, Lynn TA. Development of a mixture designprocedure for stone matrix asphalt (SMA). J Assoc Asph Pav Technol1997;66:1–30.

[17] Stuart KD, Malmquist P. Evaluation of using different stabilizers in the USRoute 15 (Maryland) stone matrix asphalt. transportation research record:Journal of the Transportation Research Board No. 1454. TransportationResearch Board of the National Academies; 1994.

[18] Thodesen CC. Development of prediction models of high temperature crumbrubber modified binders, Dissertation. Clemson University: Clemson (SC);2008.

[19] Smith JR. Asphalt rubber open graded friction course. Asphalt Rubber 2000Villamoura, Portugal; 2000.

[20] Fernandes Jr JL, Neves Filho CLD, Bernucci LLB. Evaluation of SMA mixturesproduced with asphalt-rubber binder in a LCPC wheel tester. In: Proceedings ofthe 10th International Conference on Asphalt Pavements. Quebec City(Quebec, Canada); 2006.

[21] ASTM. Standard classification for rubber compounding materials—recycledvulcanizate particulate rubber. D5603. West Conshohocken (PA); 2008.

[22] ASTM. Standard test methods for rubber compounding materials—determination of particle size distribution of recycled vulcanizate particulaterubber. D5644. West Conshohocken (PA); 2008.

[23] South Carolina Department of Transportation (SCDOT). Standard specificationsfor highway construction. Columbia (SC); 2007.

[24] Putman BJ, Thompson JU, Amirkhanian SN. High temperature properties ofcrumb rubber modified binders. Mairepav4 – Fourth International Conferenceon Maintenance and Rehabilitation of Pavements and Technological Control.Belfast (Ireland); 2005.

[25] AASHTO. Standard method of test for determination of draindowncharacteristics in uncompacted asphalt mixtures. T305. Washington, D.C.;2009.

[26] Putman BJ, Kline LC. Comparison of mix design methods for porous asphaltmixtures. J Mater Civil Eng 2012;24(11):1359–67.

[27] South Carolina Department of Transportation (SCDOT). Laboratory ProcedureManual. Office of Materials and Research. Columbia (SC); 2010.

[28] Montes F, Valavala S, Haselbach LM. A new test method for porositymeasurements of Portland cement pervious concrete. J ASTM Int 2006;2(1).

[29] Wurst JE, Putman BJ. Laboratory evaluation of warm-mix open graded frictioncourse mixtures. J Mater Civil Eng 2013;25(3):403–10.

[30] Mansour TN, Putman BJ. Influence of aggregate gradation on the performanceproperties of porous asphalt mixtures. J Mater Civil Eng 2013;25(2):281–8.

[31] AASHTO. standard method of test for resistance of compacted hot mix asphalt(HMA) to moisture-induced damage. T283. Washington, D.C.; 2007.

[32] AASHTO. Standard method of test for determining the rutting susceptibility ofhot mix asphalt (HMA) using the asphalt pavement analyzer (APA). T340.Washington, D.C.; 2007.