Construction and Building Materials 140 (2017) 301–309

Contents lists available at ScienceDirect

Construction and Building Materials

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

Development of an asphalt concrete mixture for Asphalt Core RockfillDam

http://dx.doi.org/10.1016/j.conbuildmat.2017.02.1000950-0618/� 2017 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.E-mail addresses: sjwcpt@kunsan.ac.kr (J.-W. Seo), dpark@kunsan.ac.kr

(D.-W. Park), lehominhtri92@gmail.com (T.H.M. Le).

Jung-Woo Seo, Dae-Wook Park ⇑, Tri Ho Minh LeDept. of Civil Engineering, Kunsan National University, 558 Daehak ro, Kunsan, Jeonbuk 54150, Republic of Korea

h i g h l i g h t s

� Asphalt concrete mixtures designed for rockfill dam require higher bitumen content than that for road pavement.� The strength behavior and moisture resistant of test specimens were impacted considerably by filler content.� Confining pressure has a remarkable effect on triaxial compressive strength of asphalt concrete specimens.� A small change in air void content of asphalt concrete mix could results in significant increase in permeability property.

a r t i c l e i n f o

Article history:Received 28 November 2016Received in revised form 21 January 2017Accepted 19 February 2017

Keywords:Asphalt Core Rockfill DamHot mix asphaltTriaxial strengthIndirect tensile strengthPermeability

a b s t r a c t

The main objective of this paper is to develop asphalt concrete mixture and conduct performance tests forAsphalt Core Rockfill Dam (ACRD). Three conditions of mineral filler content have been used: 10, 12 and14%, namely: F10, F12 and F14 respectively. The amount of optimum bitumen for each filler conditionwere determined by Marshall mix design method. The stress strain properties of the asphalt concrete(AC) were studied through Unconfined Compression Strength (UCS) Test, Indirect Tensile Test (IDT)and Triaxial Tests. Moisture susceptibility and water penetration were evaluated using Indirect TensileTest and Permeability Test respectively. The result exhibits that the increase of filler content from 10to 14% has significant effect on optimum asphalt content and properties of asphalt concrete mixture.Among all mixtures, mix F10 yielded highest strength behavior and moisture resistance. Result also sug-gests that asphalt concrete specimen was impervious to water (k value <10�8 mm/s) when the mixturewas produced at air void content no larger than 4%, 3.4% and 3.6% in mix F10, F12 and F14 respectively.

� 2017 Elsevier Ltd. All rights reserved.

1. Introduction

The embankment dam using asphalt core was first developed inGermany in the 1960s. Since then, more than 150 ACRD dams havebeen built in many regions around the world but mostly in north-ern Europe, including several in Norway. The first dam with anasphalt core was completed in Norway in 1978, and nearly all largeNorwegian embankment dams have been built using this method.Nowadays, many countries around the world have recognized thehuge advantages of asphalt core dam and several asphalt core damprojects have been constructed. The Chinese developed theirknowledge of the structures and built their first asphalt core damsin the 1970s. To date 13 dams of this type have been completed inChina. In North America, Canada is the first country completing

construction a dam of this type. After that, the biggest energycompany in Canada Hydro Quebec has decided to construct severalmore dams of this design.

Over the past few years detailed cost comparisons have beenmade between asphalt concrete core dams and their alternativesat the design stage of projects. For the Urar dam, completed in1997 in Norway, tenders were submitted for a Roller-Compactedconcrete (RCC) dam and a rockfill dam with asphalt core. Whenonly considering contractor costs and additional spillwayexpenses, the asphalt core alternative turned out to be approxi-mately 10% cheaper than the RCC option. For the Greater Ceresdam, completed in 1998 in South Africa, three alternatives werecompared at the design stage: RCC, concrete faced and asphalt con-crete core dams. The latter was chosen due to cost and because thedam was located in an earthquake region on a poor sandstonefoundation. It seems quite probable that embankment dams withasphalt concrete cores are likely to find a more prominent placein future dam construction.

Table 1Particle size distribution of aggregate.

Sieve size Percent passing (%)

19 mm 10013 mm 9910 mm 84#4 57#8 43#30 29#50 23#100 18#200 10/12/14*

* Three different ratios of filler content: 10, 12 and 14%.

302 J.-W. Seo et al. / Construction and Building Materials 140 (2017) 301–309

In spite of the few economic data, this type of solution seems

competitive, especially for locations where fine materials for theconstruction of a traditional core (clay) are scarce. Also, theincreased use of asphalt concrete rather than earth core is partlydue to the profession’s increased concern about internal erosionof earth core. By adjusting the bitumen content or the viscositypenetration, the viscoelastic-plastic properties can be tailored tolocal conditions and climate which makes asphalt core dams espe-cially suited to seismically active areas, and on compressible foun-dations where stiffer structures like Concrete Face Rockfills Dam(CFRD) and RCC dam may not be suitable. The reservoir can befilled during construction, which is not feasible for an upstreamfacing alternative. Furthermore, overtopping of an asphalt coreduring construction will not have the dramatic consequences asfor a clay core or an upstream facing solution.

Vlad [12] presented that asphalt core embankment dams builton good foundation have outstanding record with no seepageproblems or required maintenance. Wang and Höeg [13] suggestedthat asphalt concrete core not only offers virtually impervious andflexible characteristic but also resists to erosion and ageing. Espe-cially, the self-healing ability can be provided by the viscoelastic-plastic and ductile properties of asphalt core. This unique abilitycould prevent cracks develop in the core wall due to differentialdisplacements (shear distortions) caused by severe earthquakeloading. Asphalt concrete is also a very ‘‘forgiving” material in itsbehavior relieving itself of stress concentrations. The use of softerbitumen than in previous construction increases the self-sealingquality and allows lower operating temperatures and energy inputduring material production, transportation and core placement.Besides, some researches present another ideal advantage ofasphalt core dam named joint-less core construction which dra-matically enhance the dam quality. Furthermore, the core may beconstructed in cold and rainy weather without construction delays.Therefore, asphalt core is a competitive and economically viablealternative in comparison with other traditional ones [11].

The mix design of asphalt concrete used in impervious facingsand cores in embankment dam originated from road asphalt con-crete experience [13]. However, there are significant differencesbetween a road pavement and an interior dam core with respectto loading and environment conditions. The asphalt which is usedon roads and airfields, where deterioration becomes evident in pot-holes etc, has a different composition to the asphalt used in dams.Inside a dam the asphalt is kept under virtually ideal conditions.Fairly constant temperatures without exposure to the sun andthe rich, dense asphalt mix means that oxidation or hardening doesnot occur over time. Typically, the road asphalt concrete mixesconsist of 4–6% bitumen by mineral weight, 4–8% filler material(<0.075 mm), and 20–40% fine aggregates (0.075–2.36 mm). Theair voids content for road asphalt concrete after compaction is inthe range of 3–10%. In the other hand, the asphalt concrete usedin dam is required to be impervious and flexible and it consistsof more fine aggregates, filler and bitumen than the asphalt con-crete used in pavements. In previous designs of asphalt core dam(e.g. Höeg [13], Vlad [12]), these authors suggested the optimumasphalt content about 6.5 to 7.3% by mineral weight, <15% fillermineral (<0.075 mm), 35–52% fine aggregates (0.075–2.36 mm),33–55% coarse aggregates (2.36–19 mm) and laboratory air voidcontent no larger than 2% which will provide a virtually imperviousasphalt concrete (permeability <10–8 mm/s).

Particularly in recent years, construction of this type of dam inthe seismic areas of the world such as Japan, China and Iran hascaused researchers to focus on the dynamic behavior of asphaltcores. However, few researchers have investigated the stress-strain behavior of asphalt concrete used as watertight elementsfor dams by laboratory tests. Especially, very little has beenreported on the research of filler content suitability for hydraulic

dense grade asphalt concrete with air porosity less than 3%. Withparameters derived from the carried-out tests, the acceptable levelof the rheological behavior and stress train properties of bitumi-nous concrete mixes can be revealed. This research data can bevery useful for geotechnical engineers to conduct numerical analy-ses and predict field performance of asphalt core dams for differentsituations.

2. Research objectives and scope

The purpose of this paper is to systematically develop asphaltconcrete mixture and conduct performance tests for ACRD. Prelim-inary studies have indicated the specific content of filler for ACRD,but lack of researches has clearly justified the effect of this mineralcontent to the properties of asphalt concrete mixture. Therefore,three filler content: 10, 12 and 14% were selected to evaluate inthis study based on trial testing and previous experiences. TheMarshall mix design method was conducted to determine the opti-mum asphalt binder content in each condition. The stress strainproperties of the asphalt concrete were documented throughUnconfined Compression Test, Indirect Tensile Test and TriaxialTests. The ductility of the specimen after reaching peak strengthand any strain weakening behavior of the mix was also evaluatedto assure that the asphalt concrete exhibits flexible and ductile(not strain-softening) behavior required to adjust to dam deforma-tions caused by static and dynamic loads and differential founda-tion settlements. Indirect Tensile Test and Permeability Test werecarried out to determine the moisture susceptibility and waterpenetration respectively. Air void content also plays a very impor-tant role in the performance of asphalt concrete, especially the per-meability. Air void content in asphalt concrete mixture wasmodified from approximately 2–8.5% to determine the influenceof this content to the water penetration factor.

3. Material and methodology

3.1. Material

HMA is a mixture of mineral aggregate and bituminous binder.Asphalt binder PG58-22 was used in this study as it is the mostwidely used in Korea. The specific density of coarse and fine aggre-gate is 2.647 and 2.671 respectively following the AASHTO T 84-10[1] and AASHTO T 85-10 [2] standard. The nominal maximum sizeof 13 mm gradation was selected to enhance the design of imper-vious asphalt concrete. As mentioned before, there are three condi-tions of filler in this study: 10, 12, 14%. It is a combination of 50%pan and 50% mineral limestone which passing sieve #200. Basedon preliminary asphalt core dam studies and experiences, the par-ticle size distribution of aggregate designed for impervious core isshown in Table 1.

Table 2Marshall mix design requirements.

Properties Criteria limit

Optimum AC (%) At 2%*

Marshall stability (N) >4900 NFlow (0.01 cm) 20–50VMA (%) >13%VFA (%) �80

* Percentage of air voids.

Prepared specimens

Air voids ( )

Dry IDT test Wet IDT test

Water (25 5 ; 2h)

Vacuum (13-67 kPa) Saturation degree (55-80%)

IDT test Freezing

(-18 ; 16h)

Thawing (60 ; 24h)

Water (25 ; 2h)

IDT test

Tensile strength ratio

Fig. 2. Wet and dry IDT test diagram.

p

Fig. 1. Mohr-Coulomb circle.

J.-W. Seo et al. / Construction and Building Materials 140 (2017) 301–309 303

3.2. Asphalt mixture design by Marshall method

Wang and Höeg [13] concluded that the primary function ofasphalt concrete used in a dam core is to create an imperviouswater-barrier, flexible and sufficiently ductile to accommodatethe deformations imposed by the embankment, reservoir andfoundation, without cracking during construction, impounding,operation, and earthquake loading. Therefore, hydraulic asphaltconcrete is designed with a higher bitumen content than that inthe mix used for road and airfield pavements that are subjectedto very different loads and environmental conditions.

Based on trial testing and previous experiences, the initialdesign binder content (Pb) for mix F10, F12 and F14 was 6, 6 and6.3% respectively. Then, asphalt concrete samples were producedat Pb, Pb ± 0.5%, ±1% to determine optimum asphalt content. There-fore, there were 5 asphalt binder contents in each filler conditionswith three replicates per asphalt content.

Table 2 illustrates the Marshall mix design requirements in thisresearch. According to Marshall mix design method, aggregate wasfirst placed in the oven for more than 4 h, and less than one hourfor asphalt binder at mixing temperature. After mixing, each sam-ple was placed in the oven for 2 h at compaction temperature andcompacted with a Marshall hammer. The specimens were stored atroom temperature for 24 h and extruded from the mold prior totesting. The percent air voids, VMA, VFA, Marshall stability andflow values of the specimen were determined. Maximum specificgravity Gmm [5] and bulk specific gravity Gmb [3] of the mixturewere also figured out. Table 2 illustrates the Marshall mix designrequirements in this research.

The air-void content (porosity) of the asphalt concrete in thedam core after compaction is required to be less than 3% to ensurea permeability coefficient less than about 10�10 mm/s (almostimpervious condition) [15]. However, laboratory specimens ofthe specified design mix should be required to show an air voidcontent �2% to account for the difference in degree of compactionachieved in the laboratory and in the field [10]. Regards to this 2%air-void, optimum asphalt content was figured out in each filler

condition. Therefore, the optimum asphalt content was 5.9%, 5.9%and 6.2% for mixes F10, F12 and F14 respectively.

3.3. Unconfined compressive strength test

The test was designed with an attempt to investigate how largecompressive strength the asphalt concrete specimen could under-take under different filler content ratio. Unconfined compressivestrength (UCS) test of asphalt concrete is standardized in AASHTOT167 [4]. Test specimens were 100 mm in diameter and 150 mm inlength prepared by gyratory compactor. Testing was conductedusing a hydraulic-based universal testing machine. A strain-controlled at 2 mm/min rate of loading was applied along the axialdirection of the asphalt mix cylinder and the test was conducted atambient room temperature of 25 �C.

3.4. Indirect tension test (IDT test)

In this research, the IDT test method is used to quantify the ten-sile strength of a cylindrical specimen in accordance with ASTMD6931-12 [7] and researches of Lubinda [17], [18], [19]. Two repli-cate samples were produced from each condition. The specimengeometry was 100 mm in diameter and 60 mm in length preparedby gyratory compactor. The testing mechanism of the IDT entails acompressive loading to be applied across the diameter of a testspecimen using a Universal Testing Machine. This compressiveload produces tensile stresses in the test specimen in a directionperpendicular to the vertical applied load with a constant deforma-tion rate of 50.8 mm/min (2 in./min). The samples were tested at25 �C in an environmental-control chamber.

0.635mm thick lateral membrane was pressurized at 68.9 ± 3.4 kPa to ensure water would not flow laterally

500ml

0ml

h1

h2

Pressure gauge

Upper timing mark

Lower timing mark

Graduated cylinder I.D. = 31.75 mm (1.25 in.)

Hose bard fitting

Quick connect

Clamp assembly

Cap assembly

Test specimen Height: 60mmDiameter: 150mm

Outlet pipe

An air pump capable of applying 103.42 kPa (15psi)

Fig. 3. Water Permeability Testing Apparatus [8].

304 J.-W. Seo et al. / Construction and Building Materials 140 (2017) 301–309

3.5. Triaxial test

Triaxial compression tests should be carried out under differentconfining stresses to assure that the asphalt concrete exhibits flex-ible behavior required to adjust to dam deformations caused bystatic and dynamic loads [14]. Three compacted asphalt mixesnamed F10, F12 and F14 were used to conduct the triaxial com-pressive strength test. Two cylindrical replicates were used foreach mix. The specimen geometry was 151 mm in length and100 mm in diameter using gyratory compactor. Testing was con-ducted at three confining pressure, 35, 69 and 138 kPa in ambienttemperature (25 �C). A strain-controlled at 2 mm/min rate of load-ing was applied along the axial direction of the asphalt mix cylin-der. Testing was conducted using a hydraulic-based universaltesting machine. The loading time, displacement and uniaxialstress were collected throughout the test. Testing was terminatedat 12% strain to ensure that failure was occurred.

Triaxial compression equations base on Mohr-Coulomb theory:

r1 ¼ r3tan2ð45� þu2þ 2ctanð45� þu

2Þ ð2Þ

p ¼ r1 þ r3

2ð3Þ

q ¼ r1 � r3

2ð4Þ

u ¼ sin�1ðtanaÞ ð5Þ

c ¼ acosu

ð6Þ

where r1 is maximum principle stress, r3 minimum principlestress, C is cohesion, u is friction angle, a is the angle that the mod-ified failure envelope makes with the horizontal, q and p is the ver-tical and horizontal stress coordinate.

3.6. Indirect tensile test for moisture susceptibility

Moisture causes the loss of adhesion between the asphalt bin-der and the aggregate surface, and accelerates deterioration inthe form of potholes and cracking. In the construction of AsphaltCore Rockfill Dam, this characteristic becomes more critical. There-fore, moisture susceptibility was investigated in this study usingthe AASHTO T 283–03 [6]. The damage due to moisture is con-trolled by the specific limits of the tensile strength ratios. All spec-imens were fabricated using the gyratory compactor and had airvoid contents of 7 ± 0.5%. Six replicates from each filler condition

(a) (b)

(c)

70

75

80

85

90

95

F10 F12 F14

VFA

(%

)

12

12.5

13

13.5

14

14.5

F10 F12 F14

VMA

(%)

0

10

20

30

40

50

60

02000400060008000

10000120001400016000

F10 F12 F14

Flow

(0.

01cm

)

Stab

ility

(N

)

Stability (kN) Flow (mm)

Fig. 4. VMA(a), VFA(b), Marshall stability and flow(c) test results.

Table 3Summary of stability test results.

Mix Stability

Average (N) Standard Deviation C.O.V. (%)

F10 14168 467.5 3.3F12 13344 116.1 0.87F14 11673 122.6 1.05

J.-W. Seo et al. / Construction and Building Materials 140 (2017) 301–309 305

were produced and separated into two subsets, one subset for dryIDT test and the other for wet IDT test. The maximum indirecttensile force was recorded and the corresponding IDT strength ofthe asphalt concrete mixture was calculated. The tensile strengthratio (TSR) is determined from the dry and wet IDT test results[16]. A flowchart summarizing the experimental study was givenin Fig. 2.

3.7. Permeability test

In this study, the Florida Test Method (FM 5-565) [9] was fol-lowed to determine the permeability values. The permeability(hydraulic conductivity) of the test specimen was estimatedusing the Karol-Warner Asphalt permeameter. The testing is basedon the falling head principle to estimate the water flow ratethrough the asphalt concrete specimen (Fig. 3). Base on Darcy’slaw, the hydraulic conductivity can be estimated using equationbelow:

k ¼ aLAt

lnh1

h2

� �� tc ð7Þ

where: k is coefficient of permeability (cm/s), a is inside cross-sectional area of the buret, (cm2), L is average thickness of thetest specimen (cm), A is average cross-sectional area of the test

specimen (cm2), t is elapsed time between h1 and h2 (s), h1 is initialhead across the test specimen (cm), h2 is final head across the testspecimen (cm) and tc is temperature correction for viscosity ofwater (cm).

Three replicates (150 mm in diameter and 60 mm in height) ofeach mix were fabricated using gyratory compactor. Before perme-ability testing, superpave specimens were soaked in water to reachsaturation. A sealing tube with a flexible latex membrane0.635 mm (0.025 in.) thick was prepared to be capable of confiningasphalt concrete specimens. An air pump capable of applying103.42 kPa (15 psi) pressure was connected to apply vacuum toevacuate the air from the sealing tube/membrane cavity. Waterfrom a graduated cylinder was allowed to flow through a saturatedasphalt sample and time taken to reach a known change in headwas recorded. The coefficient of permeability k was calculatedusing Eq. (7). For each replicate, permeability tests were conductedthree times to report an average permeability value. Permeabilityvalues for the first and third test did not vary by more than 4%[8,16]. In this research, sample is considered to be imperviouswhen the testing time was approaching 30 min with no sign ofwater level moving from the upper timing mark.

4. Results and discussion

4.1. Analysis of mix design properties

The VMA, VFA, Marshall stability and flow test results were pre-sented in Fig. 4. Generally, using optimum asphalt content at 2% airvoid, the Marshall mix design properties show acceptable values.All test specimens meet the requirement value in Marshall Stabil-ity with a minimum value of 4900 N and the flow value is in therange from 38 to 54 mm. As can be seen from Fig. 4a, 4b, therewas an increase in VFA and VMA when the filler content wasincreased from 10 to 14%. Whereas, adding more filler content will

(a) (b)

(c) (d)

0

0.5

1

1.5

2

2.5

3

3.5

0 2 4 6 8

Stre

ss (M

Pa)

Strain (%)

F10-1 F10-2

0

0.5

1

1.5

2

2.5

3

0 2 4 6 8

Stre

ss (M

Pa)

Strain (%)

F12-1 F12-2

0

0.5

1

1.5

2

2.5

0 2 4 6 8

Stre

ss (M

Pa)

Strain (%)

F14-1 F14-2

0

0.5

1

1.5

2

2.5

3

F10 F12 F14

Stre

ss (M

Pa)

Fig. 5. UCS test result of mix: F10 (a), F12(b), F14(c) and peak stress comparison (d).

Table 4Summary of average stress results.

Mix Stress

Average (MPa) Standard Deviation C.O.V. (%)

F10 2.77 0.141 5.1F12 2.54 0.101 4F14 2.23 0.019 0.87

306 J.-W. Seo et al. / Construction and Building Materials 140 (2017) 301–309

decrease the Marshall stability and flow value of asphalt concretesamples (Fig. 4c). Table 3 presents the average stability valuesalong with standard deviation and COV. The test results showedlow standard deviations. The COV values varied between 0.87and 3.3% with an average of 1.74%. It can be concluded that themeasurements were repeatable.

4.2. Strength behavior of asphalt concrete mixture

Fig. 5 presents the unconfined compressive test results. Overall,the finding illustrates that there was a slight reduction in strengthof asphalt concrete specimens when the filler was raised from 10 to14%. The highest UCS of mixes F10, F12 and F14 were 2.87, 2.62and 2.24 MPa respectively. As can be seen from Fig. 5(a)–(c), thestress-strain relationship was found to be almost similar in all testconditions. The compressive stress of asphalt concrete specimensincreased with an increase in axial strain up to a certain peak.The maximum UCS of test specimens occurred at an axial strainapproximately 3%. After reaching the peak strength, the UCSdecreased with increasing axial strain. By examining stress-straincurves from Fig. 5a-c, all asphalt concrete samples yielded ductilebehavior after reaching the peak stress. The average UCS valuesalong with standard deviation and COV are exhibited in Table 4.The test results showed low standard deviations. With COV values

ranging from 0.87 to 5.1% and an average of 3.23%, the measure-ments were reasonably acceptable.

The IDT test results is presented in Fig. 6. The general trendshowed that higher filler content will results in lower tensilestrength of test specimens (Fig. 6d). As can be seen from Fig. 6d,mix F10 yielded the highest IDT peak load of 7.01 kN, whereasthe lowest IDT peak load of 6.14 kN was obtained by mix F14.However, all test specimens exhibited a relatively similar trendin the load-displacement curves (Fig. 6a–c). Result shows thatIDT load increased with an increase in displacement up to a certainpeak. After reaching the peak load, it decreased with increasingdisplacement and strain softening occurred. As can be seen fromTable 5, the IDT load showed very low standard deviations. Thealternation of COV values were between 0.47 and 3.3% with anaverage of 1.77%. It can be concluded that the measurements wererepeatable.

Fig. 7a presents the summary of triaxial test results of all mixes.It can be concluded that confining pressure has a significant impacton triaxial compressive strength of asphalt concrete mixture. Atlow level confining pressure of 35 kPa, triaxial test samples yieldedrelatively similar results with peak load of approximately 18.5 kN.When the confining pressure was enhanced to medium stress of69 kPa, all filler combinations obtained a slight increase in strengthwith average peak load value of 19.3 kN. At highest confining pres-sure of 138 kPa, it is noticed that all mix achieved the highest com-pression strength with average peak load value of 20.6 kN. Also atthis confining pressure, the strength between different filler com-bination started to vary obviously with mix F14 and F10 havingthe least and highest strength among all mixes, respectively.Results shows that the compressive strength of mix F10 increasedup to 115% when confining pressure was enhanced from 35 to138 kPa. However, the strength was slightly comparable in mixF14 regardless of confining pressure.

(a) (b)

(c)(d)

0

2

4

6

8

0 2 4 6 8

Loa

d (k

N)

Displacement (mm)

F10-1 F10-2

0

2

4

6

8

0 2 4 6 8

Loa

d (k

N)

Displacement (mm)

F12-1 F12-2

0

2

4

6

8

0 2 4 6 8

Loa

d (k

N)

Displacement (mm)

F14-1 F14-2

0

2

4

6

8

F10 F12 F14

Loa

d (k

N)

Fig. 6. IDT test result of mix: F10 (a), F12(b), F14(c) and peak load comparison (d).

Table 5Summary of average IDT load results.

Mix Load

Average (kN) Standard Deviation C.O.V. (%)

F10 7.1 0.234 3.3F12 6.224 0.029 0.47F14 6.148 0.095 1.54

J.-W. Seo et al. / Construction and Building Materials 140 (2017) 301–309 307

The Mohr-Coulomb failure theory was used to obtaincohesion and friction angle value (Fig. 1). The presence ofdifferent filler content results in considerable variation in thecohesion C and friction angle u values of asphalt concretespecimens. It is observed that the cohesion C increased from528 to 660 kPa when the filler changed from 10 to 14%(Fig. 7b). Fig. 7c represents the friction angle result of testspecimens, the general trend showed that lower internal frictionangle was obtained at higher filler content. The friction anglevalue of mix F10, F12 and F14 was 40, 32 and 29� respectively.Analysis revealed from triaxial test shows that the lower fillercontent provided mix F10 higher interlock and strengthcompared to mix F12 and F14.

Table 6 presents the average unconfined compressive strengthvalues along with standard deviation and COV. The COV valuesaltered between 0.1 and 11% with an average of 4.45%, whichreveals that the measurements were repeatable.

As can be seen from Fig. 8, the influence of filler content onmoisture resistance of test samples was obviously pronounced. Itshowed that adding more filler could result in the decrease ofIDT stress and TSR value of all asphalt concrete samples. Compar-isons of TSR values in Fig. 8 clearly indicates that mixture F10achieved the highest moisture resistance due to higher strengthacquired. Whereas mix F14 was severely affected by moisture with

33% reduction in IDT stress. Base on the result, mix F10, F12 andF14 has TSR values of 75%, 63%, and 67%, respectively. Accordingto AASHTO T 283-03, only mix F10 meets the criteria of a minimumTSR value of 75%.

The average IDT stress values along with standard deviation andCOV are illustrated in Table 7. As can be seen from table 7, the COVvalues ranged from 1.9 to 12.3% with an average of 6%, whichshows that the measurements were acceptable.

4.3. The effect of air void content on permeability property

The relationship between permeability, filler and air-void con-tent was evaluated shown in Table 8. The finding suggests thatair void content has a significant impact on the water resistanceof test specimens. It is noticeable that no water penetration wasrecorded in the test specimens when the air void content was ran-ged from 2 to nearly 3.5% in all filler combination. It can be con-cluded that those conditions are impervious to water.

The permeability seemed to be relatively low at low air voidscontent, and it increased more dramatically with higher air voidsin the mix. At 5.4 to 6% percent air void, the k value was4.5 � 10�5 mm/s, 3.02 � 10�4 mm/s and 1.5 � 10�4 mm/s in themix F10, F12 and F14 respectively. At approximately 8% air voids,a small change in this content could results in a remarkableincrease in permeability. For examples, the permeability of mixF10 produced at 8.6% air void was 22 times higher than that at6% percent air voids (9.9 � 10�4 mm/s compared to4.5 � 10�5 mm/s). Based on preliminary researches recommenda-tion (i.e. Vlad [12], Höeg [13]), asphalt concrete mixtures can beconsidered impervious and applied as water barrier when the per-meability coefficient k is no larger than 10�8 mm/s. In this study,those mixtures produced at air void content higher than 5% donot meet this requirement regardless of the filler content.

(a)

(b) (c)

0

5

10

15

20

25

35 69 138L

oad

(kN

) Confining pressure

F10

F12

F14

0

100

200

300

400

500

600

700

F10 F12 F14

Coh

essi

on (k

Pa)

0

10

20

30

40

F10 F12 F14

Fric

tion

Ang

le

(deg

.)

Fig. 7. Triaxial test results: (a) Peak axial load and confining pressure relationship; (b) Cohesion; (c) Friction Angle.

Table 6Summary of average triaxial load results.

Mix Load

Average (kN) Standard deviation C.O.V. (%)

F10 F10-35* 18.74 1.893 10F10-69 19.95 0.028 0.1F10-138 21.71 0.166 1

F12 F12-35 18.61 1.223 7F12-69 19.22 0.729 4F12-138 20.44 0.113 1

F14 F14-35 18.27 2.019 11F14-69 18.5 0.589 3F14-138 19.72 0.603 3

* F10a-35b: (a): filler content; (b): confining pressure.

0

20

40

60

80

100

5

6

7

8

9

10

11

12

F10 F12 F14

TSR

(%)

IDT

str

ess (

kN)

DRY WET TSR

Fig. 8. TSR test result of mix F10, F12 and F14.

Table 7Summary of average IDT stress results.

Mix IDT stress

Average (kN) Standard Deviation C.O.V. (%)

Dry specimenF10 11.5 0.80 6.9F12 11.2 0.52 4.7F14 10.4 0.23 2.3

Wet specimenF10 8.6 0.70 8.2F12 6.9 0.13 1.9F14 7 0.86 12.3

308 J.-W. Seo et al. / Construction and Building Materials 140 (2017) 301–309

Table 8Summary of Permeability test.

Filler content Air void content (%) Permeability coefficient k (mm/s)

F10 2.05 <10�8

2.9 <10�8

4 <10�8

6 4.5 � 10�5

8.6 9.9 � 10�4

F12 2.01 <10�8

2.5 <10�8

3.4 <10�8

5.4 3.02 � 10�4

8.4 5.8 � 10�4

F13 2.03 <10�8

2.9 <10�8

3.6 <10�8

5.4 1.5 � 10�4

8.4 5.8 � 10�4

J.-W. Seo et al. / Construction and Building Materials 140 (2017) 301–309 309

5. Summary and conclusions

This study aims to develop asphalt concrete mixture and con-duct performance test designed for ACRD. Primary findings aresummarized below:

(1) Using optimum asphalt content at 2% air void, the Marshallmix design criteria, VMA, VFA, Marshall stability, flowresults are acceptable.

(2) By varying filler content from 10 to 14%, analysis shows thatthis mineral content has a significant effect on optimumasphalt content and properties of asphalt concrete mixture.The increase in filler content leads to the increase in opti-mum asphalt content requirement. When the filler contentwas raised from 10 to 14%, the required optimum asphaltcontent was altered from 5.9 to 6.2%.

(3) In terms of UCS and IDT test, there was a slight reduction instrength when filler content was enhanced from 10 to 14%.However, all mixtures showed to follow the similar trendin stress-strain and load-displacement curve respectively.The stress (load) of asphalt concrete specimens increasedwith an increase in axial strain (displacement) up to a cer-tain peak. After reaching the peak, the stress (load)decreased with increasing axial strain (displacement). Also,all test specimens showed relatively ductile plastic behaviorafter the peak strength level had been reached.

(4) From triaxial test result, higher confining pressure resultedin higher compressive strength of asphalt concrete mixture.Also at higher confining pressure, the reduction in strengthof test specimens when adding more filler content was moreconsiderable.

(5) Regards to cohesion C, the general trend showed that highercohesion C was obtained at higher filler content. Meanwhile,the friction angle results exhibited the reverse trend.

(6) For moisture susceptibility, the result suggests asphalt con-crete mixture with 10% filler content showed highest resis-tance value. Moreover, only mix F10 meets therequirement of 75% of minimum TSR value.

(7) The result base on permeability test shows that air void con-tent has a remarkable influence on water penetration factorof asphalt concrete. A small change in this content could

results in a considerable increase in coefficient k. Asphaltconcrete specimens are considered to be impervious (k value<10�8 mm/s) when the mixture was produced at air voidcontent no larger than 4%, 3.4% and 3.6% in mix F10, F12and F14 respectively. In this study, permeability test speci-mens produced at higher 5% air void content do not reachthe permeability coefficient criteria of 10�8 mm/s base onprevious researches recommendation.

(8) When designing asphalt concrete as an impervious core inrockfill dam, mix design with 10% filler content and air voidbelow 3% is recommended to acquire desired permeabilityk-value of lower than 10�8 mm/s. Besides, based on thisstudy, desired ACRD mixture should acquire the followingproperty values: UCS of higher than 2.5 MPa; TSR value ofhigher than 75%; C value ranging from 500 to 600 kPa; fric-tion angle value varying from 30 to 40� and Triaxial stress ofhigher than 18, 20 and 21.5 kN at confining pressure of 35,69 and 138 kPa respectively.

Acknowledgement

This research was supported by K-Water, Republic of Korea.

References

[1] AASHTO T 84-10, Specific Gravity and Absorption of Fine Aggregate, AASHTO,Washington, DC, 2010.

[2] AASHTO T 85-10, Specific Gravity and Absorption of Coarse Aggregate,AASHTO, Washington, DC, 2010.

[3] AASHTO T 166-10, Bulk Specific Gravity of Compacted Hot Mix Asphalt UsingSaturated Surface-Dry Specimens, AASHTO, Washington, DC, 2010.

[4] AASHTO T 167, Standard Method of Test for Compressive Strength of Hot MixAsphalt, AASHTO, Washington, DC, 2010.

[5] AASHTO T 209-05, Theoretical Maximum Specific Gravity and Density of Hot-Mix Asphalt Paving Mixtures, AASHTO, Washington, DC, 2005.

[6] AASHTO T 283-03, Standard Method of Test for Resistance of CompactedAsphalt Mixtures to Moisture-Induced Damage, AASHTO, Washington, DC,2003.

[7] ASTM D6931, Standard test method for indirect tensile (IDT) strength ofbituminous mixtures, 2012.

[8] Designation: FM 5–565. Florida method of test for measurement of waterpermeability of compacted asphalt paving mixtures.

[9] K. Höeg, Asphaltic Concrete Cores for Embankment Dams, Experience andPractice, Norway, 1993.

[10] C.L., Monismith, J.P., Creegan, Asphalt and asphalt mixes: Characterization foruse in embankment dams. Seventeenth Annual USCOLD Lecture Series, 1997.

[11] Strabag. Asphaltic Concrete for hydraulic structures. Germany, 1990.[12] A. Vlad, P.T. Jean, V. Pierre, Design and Construction of Nemiscau-1 Dam, the

First Asphalt Core Rockfill Dam in North America, Canada Dam Association,2010.

[13] W. Wang, K. Höeg, The asphalt core embankment dam: a very competitivealternative. The 1st International Symposium on Rockfill Dams, 2009.

[14] W. Wang, K. Höeg, Y. Zhang, Design and performance of the yele Asphalt CoreRockfill Dam, Can. Geotech. J. 47 (12) (2010) 1365–1381.

[15] W. Wang, K. Höeg, Simplified material model for analysis of asphalt core inembankment dams, Constr. Build. Mater. 124 (2016) 199–207.

[16] S.H. Lee, D.W. Park, H.V. Vo, D. Samer, Asphalt mixture for the first asphaltconcrete directly fastened track in Korea, Adv. Mater. Sci. Eng. (2015).

[17] J. Lubinda, D. Brandon, S. Gautam, M. Thomas, R. Amy, M. Dale, Magdy, Searchfor a laboratory test to evaluate crack resistance of hot-mix asphalt, Transp.Res. Rec. 220 (2012) 73–80.

[18] Lubinda F, Walubita, Abu N F, Gautam D, Hossain A, Tanvir, Jun Z and Tom S.The Overlay Tester: A Sensitivity Study to Improve Repeatability & MinimizeVariability in the Test Results. Technical Research Report#. FHWA/TX-12/0-6607-1, TTI- Texas A&M University System, CS, TX, USA, 2013.

[19] Lubinda F, Walubita, Abu N F, Yasser K, Rong L, Tom S, Robert L L. The OverlayTester: Comparison with Other Crack Test Methods and Recommendations forSurrogate Crack Tests. Technical Research Report#. FHWA/TX-12/0-6607-2,TTI- Texas A&M University System, College Station, Texas, USA, 2013.