Construction and Building Materials 38 (2013) 766–775

Contents lists available at SciVerse ScienceDirect

Construction and Building Materials

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

A study on volumetric versus surface properties of wearing courses

F.G. Praticò a, R. Vaiana b,⇑a University Mediterranea, Reggio Calabria, Italyb University of Calabria, Arcavacata Campus, Cosenza, Italy

h i g h l i g h t s

" The volumetric and surface characteristics of HMA specimens were analyzed." Gyratory compactor, roller compactor and laser profilometer were used." A model for predicting surface texture and volumetrics was proposed." The model is mainly a function of compaction effort and process.

a r t i c l e i n f o

Article history:Received 8 May 2012Received in revised form 11 August 2012Accepted 20 September 2012

Keywords:Surface characteristicsVolumetric characteristicsLaser profilometerGyratory compactorSlab roller compactor

0950-0618/$ - see front matter � 2012 Elsevier Ltd. Ahttp://dx.doi.org/10.1016/j.conbuildmat.2012.09.021

⇑ Corresponding author. Tel.: +39 0984 496786; faxE-mail address: vaiana@unical.it (R. Vaiana).

a b s t r a c t

The main purpose of this study was to analyze the volumetric and surface characteristics of hot mixasphalt (HMA) specimens as a function of compaction process. Specimens were produced in the labora-tory by two different compaction devices, a gyratory compactor and a roller compactor. The volumetricand surface characteristics (air void content, bulk specific gravity) of these specimens, as well as the rela-tionships among surface texture, volumetrics and compaction, were investigated. Analysis of theseresults may allow determinations of how material movements under compaction determine volumetricsdistribution and variations and surface properties. A tentative theoretical framework for synergisticallypursuing texture and volumetric targets was formulated. Outcomes of this study are expected to benefitboth practitioners and researchers.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Pavement texture and volumetrics of hot mix asphalts (HMAs)play an important role in pavement management, expectedpavement life, road safety, sustainability [1,2] and mechanisticperformance [3]. The surface properties of sustainability (noisepollution reduction [4] and environmental impact [5,6]), efficiency(consumption reduction) and user safety are strongly influenced bythe ‘‘texture’’ of the upper layer of wearing courses [7,8]. Indeed,texture, defined by ISO Standards 13473-1 as ‘‘the deviation of apavement surface from a true planar surface’’, is related to theaccident ratio [9,10]. In particular, surface texture has been foundto affect friction and its evolution over time [11–15]. At high wave-lengths, texture is related to roughness [16,17], which affects usercomfort and has an impact on the general cost of transportation.Furthermore, texture affects drainability [18] and thereforeacceptance procedures [19]. Further research, however, is neededon the evolution of texture properties as a function of time

ll rights reserved.

: +39 0984 496787.

and/or compaction energy and type, both in the laboratory andon site [20].

HMA volumetric properties have been shown to depend on therelationship between Gmm (maximum theoretical specific gravity)and Gmb (bulk specific gravity). The specific gravity is the ratio ofthe weight in air of a volume of material at 25� C to the weightin air of an equal volume of water. HMA maximum specific gravity(Gmm or the corresponding theoretical maximum density, termed‘‘Rice’’ density) can be derived from the ratio of the weight of aloose sample to the weight of an equal volume of water at a stan-dard temperature of 25 �C [21]. Gmm is dependent on Gse (effectiveaggregate specific gravity), Pb (asphalt binder content), and Gb (spe-cific gravity of the asphalt binder). Air voids (AVs) are determinedfrom Gmm and the bulk specific gravity (Gmb) of the compactedmixture.

Both in situ and laboratory compaction processes have beenshown to affect the mechanical, volumetric and surface proper-ties of HMAs. Among the important parameters involved inand affected by compaction processes are reduction of air voids;the internal structure of samples [22–25]; transport of asphaltbinders; re-orientation and segregation of aggregates [26,27];

F.G. Praticò, R. Vaiana / Construction and Building Materials 38 (2013) 766–775 767

re-organization of aggregate-bitumen matrices; specimen sur-face, shape and morphology; and, consequently, surface texture,friction properties and other surface performance parameters,including drainability and noise emissions. Laboratory proce-dures are important in the determination of volumetrics. In con-trast, on site procedures and methods, even if very useful from amanagement standpoint, do not present a comparable level ofreliability and accuracy [28,29]. The same concept can be ex-tended to the measure of surface texture [20].

Although there have been many studies investigating the volu-metric and surface properties of HMAs, many issues require furtherresearch, including those related to the relationships among com-paction level, texture spectrum and specific gravity. A unifying the-oretical framework would be of great interest for enhancing ourunderstanding of how to improve mix composition and compac-tion to optimize both surface and volumetric properties. Due tosample inhomogeneity and boundary factors, it has become moredifficult to formulate a comprehensive theoretical model able toexplain texture and volumetrics variations.

2. Research objectives and scope

The main purpose of this research was to analyze the volumet-ric and surface characteristics of HMA specimens as a function ofthe compaction process. Based on the relevance of these character-istics over the entire boundary surface of a specimen and to deter-mine sample inhomogeneity, both the upper and lower surfaceswere investigated.

HMA specimens were produced in the laboratory using two dif-ferent compaction devices, a gyratory compactor (GC) and a Unicalslab roller compactor (USRC).

This paper is organized as follows:

– The above introduction describes the importance of pavementbulk and surface performance.

– The experimental program is described in the next section.– Results are presented and discussed and several equations are

proposed.– Finally conclusions are drawn.

3. Experiments and results

3.1. Experimental program

The experimental program was developed at the Department of Territorial Plan-ning, University of Calabria (Italy), and is summarized in Fig. 1.

Two compaction devices were used, a GC (Table 1) and a USRC (Table 2). TheUSRC device is a mechanical, self-propelled smooth steel roller with forward-reverse control, designed according to the standard UNI-EN 12697-33. The slabswere compacted at different pressures and numbers of passes until the desiredheight of the slab was reached. The compaction parameters chosen for both devicesare summarized in Tables 1 and 2.

Initially, only gyratory samples were produced, at three different levels of com-paction, corresponding 10, 60, and 200 gyrations, respectively. In the second phase,the specific gravity of the gyratory specimens was derived for each level of compac-tion. Starting from these values, 3 cm- and 6 cm-thick slabs were produced by vary-ing the weight of the material. A schematic illustration of the experimental programis shown in Fig. 1.

Overall, 36 samples were produced (see Table 3):

– 12 Gyratory compacted specimens (6 for the bituminous mix herein termed MAand 6 for the bituminous mix herein termed MB, see Section 4.1);

– 24 Roller compacted specimens (12 for MA and 12 for MB).

At the end, the actual density of each slab sample was calculated to verifywhether the fixed %Gmm at each level of compaction was reached.

Each GC specimen was cut with a wet saw into three parts (bottom, center, top)to separate the top and bottom parts (see Fig. 2). The slabs (30.5 � 40.5 � h cm)were cut and divided into 9 sectors. The central zone measured 15 cm � 15 cm

and only the top surface of this compacted zone was analyzed (see Fig. 2). The cen-tral part of the 6 cm-thick slabs was divided into two parts (top and bottom, each�3 cm thick).

Table 4 summarizes the tests performed (see also Fig. 2).In reference to texture measurements, MPD (mean profile depth), ETD (esti-

mated texture depth), and RMS (root mean square) are indicators referring to sur-face texture (for microtexture wavelengths of 0.5–50 mm), regardless of texturewavelength. In contrast, Lt is the texture level for a given wavelength k. Note thatthe lowest wavelength that can be measured depends on the minimum step ofthe device and that the latter interacts with the diameter of the spot of the laser de-vice. Therefore, it is relevant that the highest frequency that can be represented cor-rectly by a sampled signal (f) is half the sampling frequency (fs) and that, if thesignal contains frequency components above the Nyquist frequency (i.e., abovefs/2), these will be misinterpreted as lower frequencies in the spectrum of the sam-pled signal, a phenomenon known as ‘‘aliasing’’. Furthermore, the required evalua-tion length depends on the frequency analysis to be performed. For one-third-octave bands, the evaluation length (l) must be at least (5 to) 15 times kmax, wherekmax is the longest (one-third-) octave-band-center wavelength used in spectralanalyses. These requirements imply that the octave-band levels, or one-third-oc-tave band levels, determined under these evaluation lengths will be within a95%-confidence interval of approximately ±3 dB of the true band levels (ISO/TS13473-4:2008(E)).

By referring to volumetrics, and hypothesizing that AV = 0, Gmm will representthe specific gravity of a mixture. Gmb would therefore represent the actual specificgravity (6Gmm) corresponding to the actual air void content, AV (P0). Despite Gmm,the definition of which is independent of the state of matter of components (bitu-men � liquid; aggregates � solid; air � gas), Gmb and AV will be affected by the ac-tual characteristics of the bituminous mixture. Thus, for a given aggregate gradationand bitumen content, the lowest AV can be appreciably greater than 0 and the high-est Gmb much lower than Gmm, regardless of the level of compaction. Furthermore, Pb

indicates the content in terms of liquid phase, while Gsb provides a measure ofaggregate density, with both affecting compaction performance.

3.2. Materials

The first step was to select two different asphalt mixes and to produce gyratorycompacted samples and slab specimens in the laboratory at different energy levelsand at two different thicknesses. Table 5 shows aggregate gradation and the mainaggregate and asphalt binder properties for both of the mixes studied.

The two mixes had the same gradation (BRZ, below the restricted zone), butwere composed of different types of aggregate, limestone for MA and lime-stone + basalt for MB. The aggregate blend used for MB contained 30% by mass ofbasaltic material (aggregate size > 5 mm). In contrast, only limestone aggregatewas used in the MA mix, with a nominal maximum aggregate size of 9.5 mm.

3.3. Gyratory compacted specimens

Figs. 3–7 and Table 6 summarize results and analyses performed on the gyra-tory compacted specimens. Eqs. (1)–(5) refer to the proposed model.

The volumetric properties of GC specimens are summarized in Fig. 3. In the toppanels, the x-axes indicate air void content and the y-axes indicate the number ofgyrations (N). In the bottom panels, the x-axes indicate the number of gyrationsand the y-axes indicate %Gmm.

These results indicate that:

– Increased compaction energy (number of gyrations) resulted in increased %Gmm

and decreased air void content.– The center of each specimen was always denser than its top and bottom. How-

ever, the cutting process may have caused appreciable changes in the top andbottom surfaces of the central part of the specimens [28]. Furthermore, X-raycomputed tomography and image analysis techniques showed similar reduc-tions in air void content [23]. A study of air void homogeneity using gamma-ray measurements indicated that the maximum compaction level was obtainedapproximately 5 cm from the surface (i.e., at the center of the specimens) [31].

– The air void content was always higher at the top than at the bottom ofspecimens.

– The air void content of the MA mix was similar to that of the MB mix.

Fig. 4 illustrates texture levels (LT(k), or LTOP � LBOT, y-axis) as a function ofwavelength (k, mm, x-axis).

An increase of around 20 dB was observed for transitions around wavelengthsof 0.5–5 mm, regardless of the number of gyrations and of the position at the topor bottom (see Fig. 4a–d).

Moreover, regardless of mix, a higher number of gyrations yielded lower levelsof texture, especially in the domain of macrotexture (wavelengths in the range 0.5–50 mm, see Fig. 4a–d).

Top surfaces usually had higher texture levels than bottom surfaces, for eachwavelength in the macrotexture domain (Fig. 4e). In this domain, the differenceLTOP � LBOT ranged from 1 to 5.

GC specimens USRC slabs

%Gmm@200%Gmm@60 %Gmm@10

Cutting into 3 parts

BOTCENTER TOP

%Gmm@200%Gmm@60 %Gmm@10

6 cm 3 cm

Cutting into 2 parts

BOTTOP

Top (2.5 cm)

Center

Bottom(2.5cm)

GC specimens

Top surface, TS

Bottom surface, BS

Top (3 cm)

Bottom (3 cm)

USRC slabs

6 cm Top surface, TS

Bottom surface, BS

(3 cm)

3 cm Top surface

Texture and volumetric tests

Legend GC: gyratory compactor; USRC: Unical Slab Roller Compactor; %Gmm@N: level of compaction at a given number of gyrations. Top, center, bot(tom): the three parts of a cylindrical specimen compacted by a GC; USRC slabs: 3cm- or 6cm-thick slabs, compacted by the roller compactor USRC.

Fig. 1. Schematic diagram of the experimental program for the two mixtures.

Table 1Compaction parameters for the gyratory compactor (GC).

Gyratory compactor (GC) Parameters Values Reference

Compaction temperature 165 �C UNI EN 12697-31Sample height >114 mmRotation speed 29.8 rpmHand punch diameter 150 mmInclination 0.84�Pressure 600 kPa

768 F.G. Praticò, R. Vaiana / Construction and Building Materials 38 (2013) 766–775

Finally, Fig. 4g (below, right) compares our results with those reported previ-ously [32–34] on gyratory samples, on laboratory compacted slabs and on site,for the same nominal maximum aggregate size (open-graded mixes not included).Dotted lines refer to the minimum and maximum value for each wavelength. Atwavelengths of 0.3–10 mm, the texture levels of GC specimens had first derivativesaround 8 and intercepts around 24 (natural logarithm regression curves). The var-iability of data was represented by the error bars, which indicate the percent uncer-tainty in the reported measurements. The graph shows that GC values were fairlylow at lower wavelengths, but at higher wavelengths were within the minimumto maximum value reported in the literature. Previous studies have reported firstderivatives around 4–6 (i.e., smaller) in the same range of wavelengths. The texturelevel for a wavelength of 5 mm was found to be well correlated with the noise levelmeasured by the cpx-method (close proximity method, see also [32]).

Fig. 5 shows the relationship between AV and macrotexture, in terms ofMPDiso, MPDpiarc, ETD, and RMS. The mean profile depth, MPD (and, as a conse-quence, the estimated texture depth, ETD) was derived according to both PIARCand ISO methods. The PIARC method is based on the difference between peakand average levels (average z), whereas in the ISO method the profile is dividedinto two parts and the average peak level is determined. An analysis of textureindicators showed that:

– Regardless of mix (MA or MB) and specimen vertical position (i.e., top or bot-tom), macrotexture increased when AV increased.

– GC compacted specimens usually had rougher top than bottom surfaces. Fur-thermore, lower first derivatives were obtained for MA mixes (withoutbasalt).

Table 2Compaction parameters for the Unical slab roller compactor (USRC).

Unical slab roller compactor (USRC) Parameters Values Reference

Compaction temperature 165 �C UNI EN 12697-33Sample height 30 mm or 60 mmHorizontal dimensions of samples 405 mm � 305 mmPassage series 1–7thPressure 2.5–7.5 MPaNumber of passes 2–10

Table 3Summary of samples produced by the GC and USRC devices in experimental survey.

Mixes MA MB

Devices GC USRC GC USRC

Specimens £15 cm 3 cm-thick Slab 6 cm-thick slab £15 cm 3 cm-thick slab 6 cm-thick slabLevels of compaction %Gmm@10 2 2 2 2 2 2

%Gmm@60 2 2 2 2 2 2%Gmm@200 2 2 2 2 2 2

Fig. 2. Gyratory compactor specimens (above) and USRC slab samples (below).

F.G. Praticò, R. Vaiana / Construction and Building Materials 38 (2013) 766–775 769

– For both mixes, the macrotexture depth was lower for the bottom than for thetop surface. This finding was also observed during texture spectral analysis.Indeed, the difference in texture level between top and bottom was generallypositive and higher in the range of macrotexture wavelengths (k > 0.5 mm),see Fig. 4e.

– The macrotexture depth was generally higher for MA than MB. This differencewas minimal only at high air void contents (lower level of compaction @10gyrations).

Based on these findings, our volumetric analysis showed that the internalstructure of the gyratory compacted specimens changed with the depth of thesample (see also [15,22,24]). The vertical distribution of air void content con-firmed the above differences when top, center and bottom parts wereconsidered.

Regardless of the type of mix, position (top or bottom) and wavelength, a lowerlevel of compaction (e.g., the number of gyrations) was associated with a highertexture level.

Furthermore, (a) a greater number of cycles/gyrations was associated with ahigher %Gmm; (b) a greater level of compaction was associated with a lower macro-texture; and (c) the method of compaction greatly affected the level of compaction.In other words, even if Gmm expresses an intrinsic property of the mix (Pb, Gb, PA,and Gse), %Gmm and its derivatives were method-dependent.

In analyzing and understanding the above results, it is relevant to discuss thetheoretical and practical thresholds of Gmm, AV and macrotexture aggregate indica-tors (e.g., MPD).

Gmm is dependent on asphalt binder content (Pb, by weight of mixture), its spe-cific gravity (Gb), aggregate content (1 � Pb, by weight of mixture), and effectivespecific gravity (Gse). Therefore, a theoretic interpretation of Gmm is independent

Table 4Summary of tests.

Laser profilometer scanningMethod/devices/standards: Three profiles, i, j, k, were measured for the top and bottom surfaces of the GC samples, each 120 degrees from the other. Five profiles(1–5) were measured on the top of the USRC slab central samples along the same direction as compaction. From profile analyses, aggregate and disaggregate textureindicators were derived (MPDiso, MPDpiarc, ETDiso, RMS, Lt, see [11]; ISO 13473-1; ISO/CD TS 13473-4). Profiles were determined as (x,z) coordinates, where zrepresents profile depths. A laser profilometer based on conoscopic holography was used. The device has the following characteristics (ISO 13473-3): (i) Mobility:stationary, slow; (ii) Texture wavelength range: Range covered BD class 0.20 � 50 mm; (iii) Pavement contact: Contactless devices; (iv) Principle of operation: laserprofilometer; (v) Objective focal length: 100 mm; (vi) Maximum vertical measuring range: 35 mm; (vii) Vertical resolution for class 0.003 � 0.03 mm: 0.012 mm;(viii) Stand-off distance: 90 mm; (ix) Minimum horizontal resolution Dx (sampling interval) BD for class 0.05 � 1 mm: 0.01 mm; (x) Angle coverage: 170�

Maximum specific gravity of mixture, Gmm

Method/devices/standards: Corelok device, Weighing Station, AASHTO T209 and Corelok methodBulk specific gravity, Gmb, and air void content, AV

Method/devices/standards: ASTM D 6752-09/AASHTO T 331-08(2008); AASHTO T 269-97(2007); ASTM D 3203-05. Note that Gmb, Gmm% � Gmb/Gmm ⁄ 100Asphalt binder content (Pb) and aggregate bulk specific gravity (Gsb)

Method/devices/standards: Pb: UNI EN 12697-1. Gsb: MoDOT TM 81-AASHTO T85/T85Aggregate gradations and aggregate shape angularity

Method/devices/standards: SC, FC, EC, UNI EN 933-3/4

Table 5Aggregate and asphalt binder properties for MA and MB mixes.

Aggregate size d > 5 mm Limestone Basalt References

Shape coefficient (SC) 2.23 2.39 UNI EN 933-3/4Flakiness coefficient (FC) 1.40 1.59Elongation coefficient (EC) 1.62 1.54

Binder (hard modified) MA MB References

Asphalt Binder content of mixture - Pb (%) 5.80% 6.06% UNI EN 12697-1Binder content on aggregates-Pb’ (%) 6.15% 6.46%Softening point (�C) 74 UNI EN 1427Penetration (0.1 mm) 50 UNI EN 1426Penetration index 0.44 [30]

770 F.G. Praticò, R. Vaiana / Construction and Building Materials 38 (2013) 766–775

of the aggregates being composed of solid (and not liquid) particles. In contrast, Gmb,AV and MPD% are dependent on the actual characteristics and internal structure,including aggregate gradation.

AV is related to VMA (voids in mineral aggregate), with VMA depending onmany factors [35], including aggregate gradation (dense gradations decreaseVMA), aggregate shape (more rounded aggregates decrease VMA), aggregate tex-ture (smooth or polished aggregates decrease VMA), asphalt absorption (increasedasphalt absorption results in lower effective asphalt content and lower VMA, for thesame level of compaction), dust content (higher dust contents increase surface area,decrease film thickness, and tend to lower VMA), baghouse fines/generation of dust(increased fines and dust increase surface area, decrease film thickness, and tend tolower VMA), plant production temperature (higher plant production temperaturesdecrease asphalt binder viscosity, resulting in more asphalt absorption, lower

effective asphalt binder and lower VMA), temperature of HMA during paving (high-er temperatures during paving create softer mixtures, reduce air voids, and lowerVMA), hauling time (longer hauling times allow for increased asphalt absorption,lower effective asphalt content and lower VMA), and aggregate handling (moresteps in aggregate handling increase the potential for aggregate degradation, result-ing in an increase in fines, and lower VMA). If VMA is expressed as a function of as-phalt binder contents (i.e., the aggregate gradation remaining constant), it has aminimum around 10–15 for asphalt binder contents around 4.5–6%. Superpave re-quires a minimum allowable VMA of 11–15 as a function of the nominal maximumaggregate size (in the range 9.5–37.5 mm).

These findings indicate that a theoretical range 0, 1 cannot be applied to air voidcontent (where AV is expressed in decimal form). The same concept can be ex-tended to macrotexture aggregate indicators. Indeed, for a given composition, and

Fig. 3. Differences in AV and %Gmm contents among gyratory specimens.

Fig. 4. Texture spectra and compaction effort.

Fig. 5. Surface texture versus AV for MA and MB mixes, Top and bottom specimens.

F.G. Praticò, R. Vaiana / Construction and Building Materials 38 (2013) 766–775 771

772 F.G. Praticò, R. Vaiana / Construction and Building Materials 38 (2013) 766–775

regardless of the energy of compaction, the lowest AV will be mix-specific and, ingeneral, different from 0. The upper limit of 1 %Gmm will therefore be independentof both mix compaction and aggregate gradation.

For each mix composition (Pb, Gse, and Gb) and set of boundary conditions (tem-perature and compaction level), it seems reasonable to assume a minimum MPDand a minimum AV different from 0, as the number of gyrations tends to becomeinfinite.

Based on the above hypotheses, we have formulated Eqs. (1)–(5). Air void con-tent AV can be expressed as:

AV100

¼ VMA100

� Vbe

VT¼ VMA

100� PbeGmb

100 � Gbð1Þ

In Eq. (1), Vbe is the effective binder volume, i.e., the volume of bituminous bin-der external to the aggregate particles and not absorbed into the aggregate, while VT

is the total volume, i.e., the bulk aggregate volume and the effective binder volume.VMA (voids in mineral aggregate) and Gmb (mix bulk specific gravity) will be af-fected by the level of compaction, with Pbe referring to effective bitumen contentand Gb to bitumen specific gravity.

Thus, for a given Pb (and Pbe), it is possible to define a minimum, mix-specific airvoid content, AV, herein termed AVmms, as:

AVmms ¼ limN!1

Va ¼ VMAmms �PbeGmb

Gb–0; ð2Þ

where VMAmms is the corresponding value of VMA.Similarly, a minimum AV will be associated with a minimum MPD, herein

termed MPDmms.

Fig. 6. Volumetrics of GC specimens as a

Eqs. (3)–(5) were formulated to represent and describe the dependence of%Gmm, %MPD and %AV on the number of gyrations:

%Gmm

100¼ Gmb

Gmm¼ a� b � e�N

s

aþ b � e�Ns

ð3Þ

%MPDmms

100¼ MPDmb

MPDmms¼ aþ b � e�N

s

a� b � e�Ns

ð4Þ

%AVmms

100¼ AVmb

AVmms¼ aþ b � e�N

s

a� b � e�Ns

ð5Þ

where N is number of gyrations; a, b, and s are regression coefficients (with a > b ands > 0); e is Napier’s constant; MPDmms is the minimum mix-specific profile depth;AVmms is the minimum mix-specific air void content; and Gmm is the maximum the-oretical specific gravity. MDPmb and AVmb can be easily derived from Eqs. (4) and (5),respectively (see also Fig. 6), with both referring to the level of compaction pertain-ing to Gmb.

Eqs. (3)–(5) were applied to our data, where the values for MA and MB wereaveraged. Table 6 summarizes the best-fit regression coefficients. The theoreticalranges of the variations in Table 6 were derived from the above hypotheses (see alsoFig. 6).

Fig. 6 illustrates how %Gmm (6a and b), AV (6c and d, i.e., AVmb as in Eq. (5)) andMPD (6e and f, i.e., MPDmb as in Eq. (4)) vary as a function of the number of gyra-tions. Note that Fig. 6 compares experimental data and models (Eqs. (3)–(5)). Thequotient max/min, a measure of compaction susceptibility, ranged from 1.1(%Gmm) to 1.5 (%MPDmms) to 8.45 (AVmms). From a practical standpoint, these find-ings indicate that small variations in MPDs correspond to appreciable variations in

function of the number of gyrations.

Fig. 7. Densities of compacted slabs.

Table 6Best fit coefficients for Eqs. (3)–(5).

Independent variable Top Bottom

a b s MPDmms/AVmms Range a b s MPDmms/AVmms Range

%Gmm 0.867 0.04 215 – 0.91–1.00 0.85 0.04 183 – 0.91–1.00%MPDmms 2.148 0.42 20.8 0.9 1.00–1.49 2.75 0.33 50.5 0.7 1.00–1.27%AVmms 1.231 0.97 539 1.12 1.00–8.45 1.23 0.97 539 1.03 1.00–8.45

Notes: Dependent variable = N; %Gmm, %MPDmms, %AVmms are expressed in decimal form.

F.G. Praticò, R. Vaiana / Construction and Building Materials 38 (2013) 766–775 773

AVs. Based on MPD and AV target values, the above equations can be used to predictthe two corresponding optimal levels of compaction and to determine whetherthere is a suitable intermediate compaction level.

3.4. Slab roller compacted specimens

Figs. 7–10 summarize the results for HMA slabs.Fig. 7 shows AV and %Gmm as a function of compaction level. Two different types

of mixes (MA and MB) and three different types of slabs (3 cm, 6 cm top, 6 cm bot-tom, see Figs. 1 and 8) were considered.

We found that.– Regardless of mix (MA or MB), the initial air void distribution (@10) was fairly

even. In general the maximum difference was �1%.– Again, regardless of mix (MA or MB), when compaction energy was increased

(from @60 to @200), the 3-cm slabs and 6-cm bottom slabs maintained differ-ences of �1% relative to target value (AV). In contrast, the 6-cm top slabsshowed an increase in the difference to �2.5%, suggesting that the distributionof the air void content across the slab was uneven. The bottom of the middlepart was always more compacted because the forward-reverse system of passespushed the material downwards.

Fig. 8. Compacted slabs (after sawing).

– Based on a survey of the slabs, it is likely that the material was shoved sidewaysunder the roller passes: at the end of compaction the slab was more compactedin the middle than in the lateral zones.

– The bottoms of the 6-cm-thick slabs were denser than their tops, regardless ofthe number of cycles.

Fig. 9 shows AV versus macrotexture plots.Regardless of mix (MA or MB) and slab thicknesses, the macrotexture increased

when AV increased (i.e., from right to left in Fig. 8).By comparing the surface textures of 3-cm and 6-cm thick slabs, we found that

the MPD (AV) curves were always steeper for the 6-cm thick slabs (i.e., the firstderivative was higher).

These findings illustrate a condition in which thicker slabs undergo higher MPDvariations for the same AV variations.

Fig. 10 illustrates the texture spectra for the two mixes under consideration(MA, MB) and for the two thicknesses (3 cm and 6 cm).

Different levels of compaction were considered.Higher levels of compaction yielded lower texture spectral levels, regardless of

mix type (A or B), texture wavelength and slab thickness (3 or 6 cm).In more detail, the effect of compaction on texture levels increased at higher

wavelengths, at lower thicknesses and for the mix MA.On average, for a given wavelength in the macrotexture range, the following

relationship can be derived:

Lf ¼ Li �kth

ð6Þ

At a given texture wavelength, k, Lf indicates the final level of texture (e.g.,%Gmm@200), Li refers to the corresponding initial texture level (e.g., %Gmm@10), kand h are coefficients, and t refers to the specimen thickness (cm). These parametersdepend on HMA and compaction characteristics. For the cases under consideration,h ffi 2 and k ffi 75.

Finally, Fig. 10g compares our results with previous findings [32–34]. The dot-ted lines refer to the minimum and maximum values for each wavelength, whilethe solid line refers to the averages at the final compaction level (@200). At wave-lengths of 0.3–10 mm, the texture levels of slabs had first derivatives around 7 andintercepts around 23 (natural logarithm regression curves). The texture levels of theslabs were thus fairly low over the entire spectrum of wavelengths in the consid-ered range.

Fig. 10. Compacted slabs: texture level versus compaction level.

Fig. 9. Compacted slabs: macrotexture versus AV or%Gmm.

774 F.G. Praticò, R. Vaiana / Construction and Building Materials 38 (2013) 766–775

4. Conclusions

Previous studies have shown the importance of understandinghow laboratory compaction techniques can determine the organi-zation of the internal structure of samples. Indeed, compaction af-fects pavement management, expected life, road safety. Fewstudies to date have analyzed the effects of compaction methodson volumetrics and surface properties, and further research isneeded on the terms of relationships among compaction level, tex-ture spectrum and specific gravity. As a consequence, in this studythe volumetric and surface characteristics of HMA specimens as afunction of compaction process were examined.

Two different laboratory compaction devices, a GC and a USRCwere used, and the degree to which material movements undercompaction determine volumetrics distribution and variationsand surface properties was investigated.

Several key-factors emerged by the analysis and interpretationof our data. At a given level of compaction, the thickness of thesample affects volumetrics and texture spectrum and their relatedmajor consequences, including the expected life of the pavement,safety, and pavement management. In jointly considering volumet-rics and surface texture as the main issues in mix design, it is use-ful and logical to use regression curves in which the upper andlower limits take into account the relevant role of mixture compo-sition (aggregate gradation, etc.).

Even if the transferability of inferences from laboratory testsand methods to on site tests and methods may be limited, our the-oretical approach provides a tentative framework of reference forpractical applications. The results of this study can be used in the

design, construction and quality control of asphalt mixtures, tar-geting, for example, surface macrotexture and air void content.

Several issues emerged that call for further research. Nonethe-less, based on the present results, it seems imperative that specificguidance for the construction of surface courses be provided tominimize the potential for differences in surface and volumetricperformance. Further enhancement of the spectral analysis of mi-cro- and macrotexture partitions on the upper and bottom surfacesis recommended. The outcomes of this study are expected to ben-efit both practitioners and researchers.

Acknowledgments

The authors wish to acknowledge Eng. Francesco De Masi (RoadMaterial Laboratory Technician, University of Calabria), Eng. TeresaIuele and Eng. Vincenzo Gallelli for their contributions duringexperimental stages.

References

[1] Romanoschi SA, Metcalf JB. The evaluation of the probability distributionfunction for the life of pavement structures. J Transport Res Board2000;1730:91–8. http://dx.doi.org/10.3141/1730-11.

[2] Gubler R, Partl MN, Canestrari F, Grilli A. Influence of water and temperatureon mechanical properties of selected asphalt pavements. Mater Struct/Mater etConstr 2005;38(279):523–32.

[3] Canestrari F, Santagata E. Temperature effects on the shear behaviour of tackcoat emulsions used in flexible pavements. Int J Pavement Eng2005;6(1):39–46.

[4] Anfosso-Ledee F, Do MT. Geometric descriptors of road surface texture inrelation to tire-road noise. Transportation Research Record: Journal of the

F.G. Praticò, R. Vaiana / Construction and Building Materials 38 (2013) 766–775 775

Transportation Research Board, No. 1806, TRB, National Research Council,Washington, DC; 2002. p 160–7.

[5] Lieplapa L, Blumberga D. Assessing methods of PM10 and NOx emission for EIAof roads. Manage Environ Qual 2012;23(2):163–72.

[6] Praticò FG, Ammendola R, Moro A. Factors affecting the environmental impactof pavement wear. Transport Res Part D: Transport Environ 2010;15(3):127–33. http://dx.doi.org/10.1016/j.trd.2009.12.002. Elsevier Science Ltd.

[7] Praticò FG, Vaiana R, Moro A, Iuele T. HMA composition versus surfacecharacteristics: issues and perspectives to optimize road asset management.In: Second international conference on transport infrastructures, San Paolo,Brazil, August; 2010.

[8] Praticò FG, Vaiana R. Improving infrastructure sustainability in suburban andurban areas: is Porous asphalt the right answer? And how? In: 18thInternational conference on urban transport and the environment, 15–17May, A Coruña, Spain; 2012. p. 673–84.

[9] Cairney P. Macrotexture and crash risk. Some recent findings from Australiaand their implications. In: 22nd ARRB conference – research into practice,Canberra, Australia; 2006.

[10] Mayora P, Pina J. Effects of pavement friction improvement on crash rates onSpanish two lane rural raods. Washington, DC: Transportation Research Boardof the National Academies; 2008.

[11] Boscaino G, Praticò FG. A classification and inventory of pavement surfacingtexture indicators. Bulletin de Liaison des Laboratories des Ponts et ChaussesNo. 234, Sept–Oct; 2001. p. 17–34+123+125+127.

[12] Davies R, Cenek P, Henderson RJ. The effect of skid resistance and texture oncrash risk. In: International conference for surface friction for roads andrunways, Christchurch, New Zeland, May; 2005. doi: 10.1016/j.conbuildmat.2009.11.015; 2005.

[13] Viner H, Sinhal R, Parry T. Review of UK skid resistance policy. In: Fifthinternational symposium on pavement surface characteristics road andairports, Toronto, Ontario, Canada; 2004.

[14] Gothie M. Influence de l’adherence sur la securite routier. Bulletin de Liaisondes Laboratories des Ponts et Chausses, No. 185; 1993. p. 27–32.

[15] Goodman SN, Hassan Y, Abd El Halim AO. Preliminary estimation of asphaltpavement frictional properties from superpave gyratory specimens and mixparameters. In Transportation Research Record: Journal of the TransportationResearch Board, No.1949, Transportation Research Board of the NationalAcademies, Washington, DC; 2006. p. 173–80.

[16] Mucka P. Properties of random component of longitudinal road profileinfluenced by local obstacles. International Journal of Vehicle SystemsModelling and Testing 2009 – Vol. 4, No.4, Inderscience; 2009. p. 256–76.

[17] Kropác O. Non-uniqueness of single-number Indicators of longitudinal roadprofiles. Road Mater Pavement Des; 2009.

[18] Praticò FG, Moro A, Permeability and volumetrics of porous asphalt concrete: atheoretical and experimental investigation, Road Mater Pavement Des 2007;8/4:799–817.

[19] Praticò FG. Quality and timeliness in highway construction contracts: a newacceptance model based on both mechanical and surface performance offlexible pavements. Int J Constr Manage Econ 2007;25(3):305–13.

[20] Boscaino G, Celauro B, Celauro C, Amadore A. Evaluation of the laboratoryprediction of surface properties of bituminous mixtures. Constr Build Mater2009;23(2):943–52.

[21] Pratico’ FG, Moro A, Ammendola R. Modeling HMA bulk specific gravities: atheoretical and experimental investigation. Int J Pavement Res Technol2009;2(3):115–22. ISSN: 1997-1400 (on-line) 1996-6814 (print).

[22] Masad E, Muhunthan B, Shashidhar N, Harman T. Quantifying laboratorycompaction effects on the internal structure of asphalt concrete. In:Transportation Research Record: Journal of the Transportation ResearchBoard, No. 1681, Transportation Research Board of the National Academies,Washington, DC; 1999. p. 179–85.

[23] Alvarez AE, Martin AE, Estakhri C. Internal structures of compacted permeablefriction course mixtures. Constr Build Mater 2010;24:1027–1035. doi:10.1016/j.conbuildmat.2009.11.015.

[24] Jonsson M, Partl M, Flisch A. Comparison of different compaction methodsusing X-ray computer tomography. EMPA, Report No.113/12; 2002.

[25] Masad E, Button J. Implications of experimental measurements and analyses ofthe internal structure of HM. Washington, DC: Transportation Research Boardof the National Academies; 2004.

[26] Hunter AE, Airey GD, Collop AC. Effect of asphalt mixture compaction onaggregate orientation and mechanical performance. In: Proceedings of the 8thconference on asphalt pavements for Southern Africa (CAPSA’04), 12–16September; 2004.

[27] Iwama M. Influence of specimen size and orientation on the mechanicalproperties of laboratory compacted asphalt specimens. The University ofNottingham, Department of Civil Engineering, Nottingham TransportationEngineering Centre – Thesis submitted for the degree of Master of Philosophy;July 2009.

[28] Praticò FG, Moro A. Measurement of air void content in hot mix asphalts:method and core diameter dependence. Constr Build Mater 2012;26(1):344–349. doi:10.1016/J.Conbuildmat.2011.06.032.

[29] Praticò FG, Moro A, Ammendola R, Factors affecting variance and bias of non-nuclear density gauges for PEM and DGFC. Baltic J Road Bridge Eng 1822–427X, 1822–4288 on line 2009;4(3):99–107.

[30] Pfeiffer JH, Van Doormal PM. The rheological properties of asphaltic bitumens.J Inst Petrol 1936;22:414–40.

[31] Dubois V, De La Roche C, Burban O. Influence of the compaction process on theair void homogeneity of asphalt mixture samples. Constr Build Mater2010;24:885–897. doi: 10.1016/j.conbuildmat.2009.12.004.

[32] Storeheier SÅ. Road surface texture and rolling noise. Bergen, Norway BNAM2010, May 10–12, http://www.kongress.no/sitefiles/13/bilder/Kongress2010/Akkustisk/BNAM2010/Paper/Storeheier.pdf – last access; 2012.

[33] U.Sandberg, P. Anund – Measurements of macrotexture by the VTI mobile laserprofilometer in the PIARC international experiment 1992 – VTI n. TF 52-20,1992.

[34] Boscaino G, Praticò FG, Vaiana R, Wearing courses intrinsic and extrinsicproperties: experimental and theoretical research, Rivista Le strade; 2003.ISSN: 0373–2916 [special issue].

[35] Chadbourn BA, Skok EL, Newcomb DE, Crow BL, Spindle S. The effect of voids inmineral aggregate (VMA) on hot-mix asphalt pavements. University ofMinnesota, Report No. MN/RC, September 1999, Department of CivilEngineering, 500 Pillsbury Drive, SE Minneapolis, MN 55455–0116; 2000.