Content uploaded by Nishantha Bandara

Author content

All content in this area was uploaded by Nishantha Bandara on Sep 04, 2015

Content may be subject to copyright.

46 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING / FEBRUARY 2000

C

ORRELATION OF

T

IRE

W

EAR AND

F

RICTION TO

T

EXTURE OF

C

ONCRETE

P

AVEMENTS

By M. Gunaratne,

1

N. Bandara,

2

J. Medzorian,

3

M. Chawla,

4

and P. Ulrich

5

A

BSTRACT

:Pavement surface texture signiﬁcantly contributes to tire wear and tire-pavement friction. Currently

available relationships quantify tire wear and tire-pavement friction simply in terms of empirical pavement

texture parameters. The objective of this paper is to present correlations with which the tire wear rate and tire-

pavement friction on smooth concrete pavements can be predicted using actual texture properties. Using a

laboratory tire wear simulator and an aircraft tread-rubber block, a number of wear and friction tests are per-

formed on pavement samples having different ﬁne aggregate sizes. Frequency characteristics of the texture of

the pavements are achieved by decomposing the proﬁlometer measurements using the fast Fourier transform

technique and constructing power spectral density plots of texture over surface spatial frequency. Then the tire

wear rates as well as dry friction and wet friction of the tire-pavement interface are correlated to microtexture

and macrotexture components of the texture power spectral density. The developed correlations indicate that

both tire wear and dry friction are signiﬁcantly affected by pavement microtexture. The developed wear corre-

lations can be also useful for predicting the wear index of a pavement based on conventional Mu-meter and

grease patch test results. This is illustrated by an example in which the wear index for a concrete runway

pavement at Luke Air Force Base, Ariz., is computed using the new relations.

INTRODUCTION

Vehicle maneuvers such as braking and cornering require

sufﬁcient skid resistance or tire-pavement friction to maintain

vehicle stability. On the other hand, these vehicle maneuvers

can also cause excessive tire wear. Hence, in designing a run-

way pavement in particular, it is essential to consider both the

level of skid resistance and the tire wear potential of the sur-

face course. Pavement texture plays a vital role in the devel-

opment of both pavement friction and tire wear. Therefore the

knowledge of the effect of texture parameters of friction and

tire wear will certainly assist pavement engineers in designing

pavements that improve tire life without compromising the all

important skid resistance.

Pavement texture can be grouped into two classes based on

ASTM E 867: (1) Pavement microtexture—the deviations of

a pavement surface with characteristic dimensions of wave-

length and amplitude <0.5 mm; and (2) pavement macrotex-

ture—the deviations of a pavement surface with characteristic

dimensions of wavelength and amplitude from 0.5 mm up to

a value that no longer affects tire-pavement interaction. Ob-

viously, the measurement of microtexture is more rigorous

than the measurement of macrotexture. However, using a very

sensitive proﬁlometer, proﬁle heights related to both microtex-

ture and macrotexture can be obtained.

Several standard pavement texture measuring methods are

available that can be broadly categorized into two groups: (1)

Direct measuring techniques such as the proﬁlometer method;

and (2) indirect methods such as the sand patch method, the

grease patch method and the British pendulum test. The sand

patch and the grease patch method provide an indirect measure

of macrotexture by the volume of grease or sand required to

1

Assoc. Prof., Dept. of Civ. and Envir. Engrg., Univ. of South Florida,

Tampa, FL 33620.

2

Res. Asst., Dept. of Civ. and Envir. Engrg., Univ. of South Florida,

Tampa, FL.

3

Mech. Engr., Wright-Patterson Air Force Base, Dayton, OH 45433.

4

Deceased 1996, formerly, Mech. Engr., Wright-Patterson Air Force

Base, Dayton, OH.

5

Chf. Mech. Engr., Wright-Patterson Air Force Base, Dayton, OH.

Note. Associate Editor: Jan Olek. Discussion open until July 1, 2000.

To extend the closing date one month, a written request must be ﬁled

with the ASCE Manager of Journals. The manuscript for this paper was

submitted for review and possible publication on July 7, 1997. This paper

is part of the Journal of Materials in Civil Engineering, Vol. 12, No. 1,

February, 2000. 䉷ASCE, ISSN 0899-1561/00/0001-0046–0054/$8.00 ⫹

$.50 per page. Paper No. 16150.

cover a known pavement area. Results of the tests are ex-

pressed using an empirical measure of the pavement macro-

texture called the mean texture depth (MTD). On the other

hand the British pendulum test provides an indirect measure

of microtexture. This tester is equipped with a standard rubber

slider that comes into a locked position when it is horizontal.

When released, the slider makes contact with the test surface

on its swing. A drag pointer is used to measure the swing in

terms of the British pendulum number (BPN). The greater the

friction between the slider and the test surface (due to the

microtexture), the more the swing is retarded and the larger

the BPN reading.

Previous Studies on Tire-Pavement Interaction

Many researchers have studied tire-pavement interaction in

the recent past and Kummer (1966) established a relationship

between friction components (adhesion and hysteresis) and

pavement texture properties. They related adhesion to molec-

ular texture and hysteresis to macrotexture properties. Another

set of experimental studies was carried out by Henry (1968)

to establish statistical relationships among skid numbers (SN)

at different speeds, BPN, and MTD of pavements. In these

relationships, the SN at zero speed is directly correlated to

microtexture properties, and the rate of change of the SN with

the speed is correlated to the macrotexture properties. Conse-

quently, SN at 63.36 km/h (40 mi/h) was expressed as

SN = k(BPN) ⫹k(MTD) ⫹k(1)

40 1 2 3

where k

1

,k

2

, and k

3

= constants.

Although scientiﬁc literature on tread wear versus pavement

texture is sparse, a signiﬁcant contribution was made by

Lowne (1971). The tread wear of passenger car tires was mea-

sured on a series of test pavements at the Transportation and

Road Research Laboratory in the United Kingdom. These tests

showed that the microtexture was the dominant factor in de-

termining tread wear. Lowne (1971) showed that tread wear

Wwas given by a multiple regression equation of the form

W=⫺9.2 ⫹90(S)⫹18T(2)

50

where S

50

= wet cornering friction coefﬁcient at 50 km/h for

a smooth no-tread pattern tire; and T= macrotexture depth

parameter (mm).

On the other hand, researchers have investigated highway

pavement roughness using spectral analysis. In this regard, an

J. Mater. Civ. Eng. 2000.12:46-54.

Downloaded from ascelibrary.org by Lawrence Technological University on 01/16/15. Copyright ASCE. For personal use only; all rights reserved.

JOURNAL OF MATERIALS IN CIVIL ENGINEERING / FEBRUARY 2000 / 47

FIG. 2. Sample Pavement Surface

FIG. 1. SURTRONIX 3ⴙProﬁlometer

attempt by Marcondes et al. (1991) to develop better power

spectral density (PSD) models for the prediction of truck-bed

acceleration level in relation to transportation hazards such as

shocks and vibrations is signiﬁcant. Proﬁle height measure-

ment for this study was conducted using the Michigan De-

partment of Transportation proﬁlometer. Proﬁlometer readings

were taken at 7.5 cm (3 in.) intervals over 300.96 m (0.19 mi)

sections, and the pavement elevation PSDs were computed up

to a maximum spatial frequency of 2.36 cycles/m (Marcondes

et al. 1988).

EXPERIMENTAL PROGRAM

Measurement of Surface Texture

Because the pavement skid resistance and tire wear cannot

be described by the macrotexture properties alone, it was de-

cided to use the proﬁlometer tracing technique to measure both

microtexture and macrotexture properties. The SURTRONIX

3⫹proﬁlometer (Fig. 1) manufactured by Rank Taylor Hob-

son Inc., Des Plaines, Ill., was selected for proﬁle measure-

ments. The relevant speciﬁcations for SURTRONIX 3⫹are

indicated below:

• Horizontal resolution = 1 m (40 in.)

• Vertical resolution = 0.001 m (0.04 in.)

• Traverse length = 25.4 mm (1 in.)

In addition, the grease patch test was also used in this ex-

perimental program to compare its results with macrotexture

portions of proﬁlometer measurements.

Friction/Wear Testing Machine

The essence of this experimental program was the measure-

ment of the friction coefﬁcient and tire wear rates on different

pavement sections. Because these measurements had to be car-

ried out at speciﬁc speeds, a machine with a belt-driven ro-

tating disk providing both horizontal as well as vertical load

measuring mechanisms was developed to achieve this objec-

tive. This 55-cm-diameter disk is furnished with a 5-cm-wide

ring to form concrete pavement samples to a depth of 5 cm

(Fig. 2). The disk can be spun around a vertical axis using a

speed reduction pulley arrangement. The spinning motion is

smoothened by ﬁxing the disk to a trailer hub. The pulley

arrangement consists of two multiradii pulleys with diameters

of 5, 7.5, 10, and 12.5 cm. While one pulley is driven by the

motor, the disk can be rotated at a speciﬁed speed by using

the other pulley as an idle wheel. The above described ar-

rangement is clearly shown in Figs. 2 and 3.

Preparation of Pavement Surfaces

Although the pavement strength is not a governing factor

in this testing program, test pavement surfaces were designed

according to the following American Concrete Institute (ACI)

speciﬁcation (ACI 211.1-91):

• Compressive strength = 20 MPa

• Slump = 75–25 mm

• Water-to-cement ratio = 0.62

The weight fraction of each component in the mixture is as

follows:

• Cement = 0.14

• Course aggregate = 0.43

• Fine aggregate = 0.35

• Water = 0.08

To obtain different texture characteristics, the particle size

of the ﬁne aggregate was varied while keeping the other at-

tributes the same. Selected particle sizes and labels assigned

for pavement surfaces are as follows:

• Pave 40—Fine aggregates passing 1.18 mm sieve and

retained on 0.425 mm sieve

• Pave 60—Fine aggregates passing 0.425 mm sieve and

retained on 0.25 mm sieve

• Pave 100—Fine aggregates passing 0.25 mm sieve and

retained on 0.15 mm sieve

• Pave 200—Fine aggregates passing 0.15 mm sieve and

retained on 0.075 mm sieve

Then, proﬁle measurements were performed on the above

sample pavement surfaces over a traverse length of 25 mm.

Measurement of Tread-Rubber Wear

The wear rate can be measured either as a weight loss or a

change of volume in the tread rubber. Of several distinct wear

criteria proposed by previous researchers (Moore 1972), the

gravimetric wear rate K

W

, was employed in this research

⌬W

K= (3)

W

LA

a

where ⌬W= weight loss of the rubber block; and Land A

a

=

distance slid along the pavement surface and the apparent area

of the rubber block, respectively.

Friction Measurements

During each wear test the dry rubber/concrete friction force

was measured using the load cell arrangement shown in Fig.

4. After each wear test, the wet rubber/concrete friction was

J. Mater. Civ. Eng. 2000.12:46-54.

Downloaded from ascelibrary.org by Lawrence Technological University on 01/16/15. Copyright ASCE. For personal use only; all rights reserved.

48 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING / FEBRUARY 2000

FIG. 3. Friction/Wear Testing

FIG. 4. Horizontal and Vertical Load Measuring Load Cells

also measured by ﬂooding the pavement using a steady water

supply applied in front of the rubber block. The sequence of

texture, friction, and rubber tread wear measurements for each

pavement is illustrated in Fig. 5.

ANALYSIS OF TEST RESULTS

Analysis of Proﬁle Measurements

Spectral analysis techniques such as the fast Fourier trans-

form (FFT) technique are commonly adopted to analyze pave-

ment proﬁles because their statistical characteristics resemble

those of random signals. If z(x) is the surface proﬁle height

expressed as a function of longitudinal distance x, the corre-

sponding ﬁnite-length Fourier transform can be written

L

Z(k)= z(x)exp(⫺ikx)dx (4)

冕

0

where k=2f

m

and f

m

= spatial frequency components of the

surface roughness. It can be assumed that the surface proﬁle

data set consists of Nvalues of z(x) that are measured at

equally spaced intervals ⌬xover a total length of L(=N⌬x).

If these discrete height data are adjusted to have a zero mean

value denoted by z(n), then the mean square roughness 

2

is

given by

N⫺1

1

22

=z(n) (5)

冘

N

n=0

The highest surface spatial frequency resolved in the mea-

surement process, which is the Nyquist frequency, is f

c

=

(2⌬x)

⫺1

. Hence, it can be seen that the measurement interval

⌬xlimits the bandwidth of frequencies.

Using standard FFT methods, the digital equivalent of (3)

yields the Fourier transform Z(m)as

ˆ

Z(m)=⌬xZ(m), ⫺N/2 ⱕmⱕN/2 (6)

where

N⫺1

ˆ

Z(m)= z(n)exp(⫺2imn/N), ⫺N/2 ⱕmⱕN/2 (7)

冘

n=0

Then, a plot of the real component of will clearly

ˆ

Z(m)

exhibit the wavelength (or frequency) spectrum of the surface

proﬁle.

On the other hand, the PSD of the proﬁle can be expressed

as

2

兩Z(k)兩

PSD = (8)

L

J. Mater. Civ. Eng. 2000.12:46-54.

Downloaded from ascelibrary.org by Lawrence Technological University on 01/16/15. Copyright ASCE. For personal use only; all rights reserved.

JOURNAL OF MATERIALS IN CIVIL ENGINEERING / FEBRUARY 2000 / 49

FIG. 5. Texture,Wear, and Friction Testing Methodology

The digital equivalent of (7) can be written

2

N⫺1

P(m)= z(n)exp(⫺2imn/N) (9)

N

冏冘 冏

n=0

Finally, (9) can be simpliﬁed to

⌬xˆˆ

P(m)= (Z(m)⫻(Z(m))*) (10)

N

N

where = complex conjugate of It is seen that

ˆˆ

(Z(m))* Z(m).

P

N

(m) is a combination of both real and imaginary components

of and it thus contains more information on the fre-

ˆ

Z(m),

quency spectrum than alone.

ˆ

Z(m)

Furthermore, one can integrate the PSD over the Nyquist

bandwidth limits of surface spatial frequency to obtain the area

under the PSD curve. Parseval’s theorem (Gunaratne and Ban-

dara 1997) states that this area is equal to the mean-square

roughness 

2

of the pavement. Therefore it can be seen that

the PSD plot of a surface proﬁle can be employed to describe

both frequency and roughness characteristics in terms of the

spatial frequencies and the area under the plot, respectively.

The other conventional texture parameters are the arithmetic

mean of the departures of the proﬁle from the mean line R

a

and root-mean-square parameter (RMS) expressed, respec-

tively, by

L

1

R=兩z(n)兩dx (11)

a

冕

L

0

L

1

2

RMS = (12)

z(n)dx

冑冕

L

0

As the horizontal resolution of the SURTRONIX 3⫹pro-

ﬁlometer is 1 m, each sample proﬁle consists of 25,000 data

points.

Then, the FFT technique was used to obtain the PSD of

proﬁles. This task was facilitated by a MATLAB program that

is also capable of generating PSD plots and determining other

associated parameters using FFT algorithms.

Fig. 6 shows a sample proﬁle plot obtained from one pave-

ment sample (Pave 60). Because the horizontal resolution of

the proﬁle measurement is 1 m in this testing program, the

Nyquist frequency, which is equal to (1/2⌬x)

⫺1

, is as high as

0.5 cycles/m. The FFT versus spatial frequency plot and the

corresponding PSD plot for Pave 60 are given in Figs. 7 and

8, respectively.

In Figs. 7 and 8, it is clear that the proﬁle is composed of

several sinusoidal waves with different frequencies (wave-

lengths) of which only a few are predominant. Furthermore,

the plots can also be used to differentiate the pavement mi-

crotexture and macrotexture at the demarcation wavelength of

0.5 mm, which corresponds to a frequency of 2 cycles/mm.

Then, using the microtexture/macrotexture demarcation, the

roughness contributions from pavement microtexture and

macrotexture can be easily deduced from the respective areas

under the PSD plot.

To account for the randomness along the given test pave-

ment sections, two locations were selected on each test pave-

ment, and two proﬁle measurements were taken at each lo-

cation. Thus, an average PSD curve was obtained for each test

pavement based on all four proﬁles.

Roughness Correlations

By performing regression analysis on the data, the following

relation was observed between the MTD and the macrotexture:

Macro(PSD) = ⫺69.78 = 11,058 ⫻MTD (13)

(⫺0.39) (8.45)

where Macro(PSD) = area under the macrotexture portion of

the PSD curve (m

2

). Values in the parentheses give the t

J. Mater. Civ. Eng. 2000.12:46-54.

50 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING / FEBRUARY 2000

FIG. 6. Surface Proﬁle for Pave 60

FIG. 7. FFT versus Spatial Frequency Plot (Pave 60)

statistics for the individual variables, and the Fstatistic for the

complete model is 71.41 with R

2

being 0.71. Fig. 9 shows the

predicted and observed area under the macrotexture portion of

the PSD curve versus MTD.

Correlation of Tread-Rubber Pavement Friction to

Pavement Texture Properties

The measured dry and wet friction coefﬁcients on different

pavement surfaces were also correlated to the corresponding

pavement texture properties. These dry and wet friction coef-

ﬁcients were determined by averaging the time history of the

friction coefﬁcients obtained in each 7.92 km/h (5 mi/h) test.

The two mechanisms in which tire-pavement friction is gen-

erated are adhesion and hysteresis. The adhesion component

of the friction depends on the pavement microtexture, whereas

the hysteresis component primarily depends on pavement

macrotexture. To test this dependency, multiple regression

models were developed for wet and dry friction coefﬁcients in

terms of texture properties. The developed models can be de-

scribed as follows:

= 1.015 ⫹0.251 Log(Micro) ⫺0.302 Log(Macro) (14)

wet

(33.36) (8.93) (27.36)

where

wet

= friction coefﬁcient at wet condition at 7.92

km/h (5 mi/h); Micro = area under the microtexture portion

of the average PSD curve (m

2

); and Macro = area under the

macrotexture portion of the average PSD curve (m

2

). The

values in the parentheses are Fstatistics for the individual

variables, and the Fstatistic for the model is 14.02 with R

2

being 0.609. The model can be simpliﬁed as shown below

J. Mater. Civ. Eng. 2000.12:46-54.

JOURNAL OF MATERIALS IN CIVIL ENGINEERING / FEBRUARY 2000 / 51

FIG. 8. PSD Plot (Pave 60)

FIG. 9. Predicted and Observed Area under MacrotexturePor-

tion of PSD Curve versus MTD FIG. 10. Plot of Predicted

dry

versus Observed

dry

FIG. 11. Plot of Predicted

wet

versus Observed

wet

0.251

10.35(Micro)

= Log (15)

wet

再冎

0.302

(Macro)

Also, for dry friction

= 2.618 ⫹0.481 Log(Micro) ⫺0.701 Log(Macro) (16)

dry

(38.75) (4.44) (23.74)

where

dry

= friction coefﬁcient at dry condition at 7.92

km/h (5 mi/h). The values in the parentheses are Fstatistics

for the individual variables, and the Fstatistic for the model

is 11.97 with R

2

being 0.52. The model can be simpliﬁed as

shown below

0.481

414.96(Micro)

= Log (17)

dry

再冎

0.701

(Macro)

As expected, (15) and (17) clearly show the stronger de-

pendency of the dry friction coefﬁcient on pavement micro-

texture compared to the wet friction coefﬁcient. This is be-

cause dry friction is mostly based on adhesion, which is

governed by microtexture. Figs. 10 and 11 show the plot of

predicted versus observed values for dry and wet friction co-

efﬁcients, respectively.

Variation of Tread-Rubber/Pavement Friction

with Speed

The variation of tread-rubber/pavement friction with speed

was also analyzed in this experimental program. The drytread-

rubber/pavement friction was measured at three different

J. Mater. Civ. Eng. 2000.12:46-54.

52 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING / FEBRUARY 2000

FIG. 12. Variation of Friction Coefﬁcient with Speed for Sam-

ple Pavements

FIG. 15. Plot of Predicted

dry

versus Observed

dry

for New

Pavement

FIG. 14. Comparison of R

a

/RMS Values of New Pavement with

Test Pavements

FIG. 13. Plot of Predicted K

W

versus Observed K

W

speeds: 7.92 km/h (5 mi/h), 15.84 km/h (10 mi/h), and 23.76

km/h (15 mi/h). It is a well-known fact that the friction co-

efﬁcient decreases with increasing speed. A similar trend is

seen in the current testing program as well. Fig. 12 shows the

variation of dry friction coefﬁcient with speed for the four

sample pavements.

Correlation of Tread-Rubber Wear Rates to Pavement

Texture Properties

Another objective of this research was to develop a corre-

lation between tread-rubber wear rates and pavement texture

properties. Extremely high wear rates were observed on newly

constructed pavements as new pavements are abrasive. Hence,

these initial observations for each pavement were discarded,

and the analysis was performed only on subsequent data. The

following regression equation was obtained:

Log(K)=⫺0.959 ⫹0.365 Log(Micro) ⫺0.486 Log(Macro)

W

(15.93) (10.89) (39.81)

(18)

where K

w

= gravimetric wear rate at 7.92 km/h (5 mi/h); Micro

= area under the microtexture portion of the PSD curve (m

2

);

and Macro = area under the macrotexture portion of the PSD

curve (m

2

). R

2

and the Fstatistic for the model are 0.72 and

20.29, respectively. The model can be rewritten

0.365

(Micro)

K= (19)

W0.486

69.375(Macro)

As seen in (19), the wear rate is directly proportional to the

microtexture and inversely proportional to the macrotexture.

The microtexture is caused by the roughness of the individual

aggregate particles, whereas the assembly of asperities in a

given surface form the macrotexture. Hence, when the macro-

texture increases on a given pavement, it reduces the contact

between the tire and individual aggregate particles, thus re-

ducing the wear. On the other hand, when microtexture in-

creases, the surface becomes rougher owing to the increased

contact between the tire and individual particles thus increas-

ing tire wear. Both these phenomena are amply illustrated by

(19). Fig. 13 shows the variation of predicted K

W

versus ob-

served K

W

.

VERIFICATION OF DEVELOPED RELATIONS

To verify the above results, another set of similar experi-

ments was conducted on a pavement with texture properties

different from those used for formulating the relations. This

pavement was constructed using a mixture of ﬁne aggregates

employed for the preparation of the previous test pavements.

Hence the roughness of the new pavement was within the

range of roughness encompassed by the previously tested

pavements. This is clearly seen by the R

a

/RMS values of the

new pavement (Fig. 14).

Using developed models, dry and wet friction coefﬁcients

and wear rates were determined. Figs. 15–17 show the re-

spective plots of the observed values of dry friction, wet fric-

J. Mater. Civ. Eng. 2000.12:46-54.

JOURNAL OF MATERIALS IN CIVIL ENGINEERING / FEBRUARY 2000 / 53

FIG. 16. Plot of Predicted

wet

versus Observed

wet

for New

Pavement

FIG. 17. Plot of Predicted K

W

versus Observed K

W

for New

Pavement

TABLE 1. Measured Friction and Texture Characteristics of

Luke Air Force Base Runway 03R/21L

Section

(1)

Length from

21L end

(m)

(2)

wet

(63.36 km/h)

(3)

wet

(95.04 km/h)

(4)

MTD

(mm)

(5)

A 300–600 0.40 0.36 0.308

B 600–1,200 0.35 0.26 0.285

C 2,400–2,700 0.37 0.26 0.272

TABLE 2. Estimated Wet FrictionCoefﬁcients at 7.92 km/h

Section

(1)

wet

at 7.92 km/h

(2)

A 0.481

B 0.589

C 0.686

TABLE 3. Estimated Wear Rates for Different Pavement Sec-

tions

Section

(1)

Area under

macrotexture portion

of PSD curve

(m

2

)

(2)

Area under

microtexture portion

of PSD curve

(m

2

)

(3)

Wear rate

K

W

(g/cm

2

)

(4)

A 3,336.08 129.29 0.1648

B 3,081.75 316.05 0.2368

C 2,937.99 727.99 0.3296

tion, and wear rate against those predicted by (15), (17) and

(19).

These plots show that (for smooth concrete pavements

within the tested roughness range) friction and wear can be

predicted to a reasonable accuracy using the developed rela-

tions.

APPLICATION

Finally, the developed relations were used along with avail-

able Mu-meter readings and grease patch measurements at

Luke Air Force Base, Ariz., to predict the wear rate of a run-

way. LukeAir Force Base possesses two runways: (1) Runway

03L/21R consisting of an asphalt surface; and (2) runway 03R/

21L consisting of a concrete surface. Prediction of the wear

rate is possible only for runway 03R/21L, because the above

developed models are applicable for concrete pavements.

Wet friction coefﬁcients had been measured on both run-

ways in an earlier study (HQ AFCESA 1992). Each runway

is divided into three sections, and the friction measurements

for each section are provided in Table 1.

To obtain the corresponding texture parameters for this run-

way from (15), one needs to extrapolate the value at 7.92

km/h (5 mi/h) from those at 63.36 km/h (40 mi/h) and 95.04

km/h (60 mi/h) (Table 1). The following equation, which ex-

presses the variation of wet friction with speed (Leu and Henry

1978), can be used for this purpose:

PNG

SN = SN exp ⫺V(20)

0

冋冉 冊册

100

where SN

0

= skid number (or ⫻100) at zero speed; PNG

= percent normalized gradient of the SN versus speed Vcurve;

and SN = skid number at speed V. Table 2 shows the accord-

ingly estimated wet friction coefﬁcients at 7.92 km/h (5 mi/h)

for above pavement sections. Then, the values in Table 2,

MTD values in Table 1, and the values of (13) and (15) can

be used to determine the respective areas under the macrotex-

ture portion and microtexture portion of the PSD curves, for

pavement sections A, B, and C. These results are shown in

Table 3. Finally, the above-estimated texture parameters can

be used to predict the wear rates K

w

of the above pavement

sections using (19), as indicated in Table 3.

CONCLUSIONS

This research study was initiated with the goal of establish-

ing reliable friction and tire wear prediction methods based on

actual pavement texture properties. A state-of-the-art electro-

mechanical proﬁlometer was used to record the pavement pro-

ﬁle heights up to an accuracy of 0.01 m ensuring that the

measured proﬁle data incorporates both microtexture and ma-

crotexture properties of the pavement surfaces. The FFT tech-

nique was used for the decomposition of pavement proﬁles

into individual frequency components and the subsequent con-

struction of the PSD plots. The PSD plots are used to obtain

the pavement texture properties related to microtexture and

macrotexture.

A laboratory tire wear simulator was developed to measure

tire wear as well as dry and wet friction levels on a number

of pavement samples with different ﬁne aggregate sizes. Dif-

ferent texture characteristics could be produced on the pave-

ment surface by changing the particle size of the ﬁne aggregate

portion. This simulator also facilitates the performance of

rubber wear tests and rubber-pavement friction measurements

under a range of normal loads and speeds.

Reasonably good correlations were obtained between the

tire-pavement friction and proﬁle texture parameters such as

the macrotexture and microtexture portions of the PSD plot.

J. Mater. Civ. Eng. 2000.12:46-54.

54 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING / FEBRUARY 2000

Remarkably, these relationships tend to explain the respective

roles of adhesion and hysteresis components of tire-pavement

friction. A strong correlation was also obtained between tire

wear rates and macrotexture and microtexture portions of

pavement roughness. Tire wear rates are seen to be directly

proportional to the microtexture and inversely proportional to

the macrotexture. On the other hand, neither the tire wear nor

the tire-pavement friction shows a signiﬁcant correlation to the

currently used texture parameter of MTD. Hence these results

indicate the importance of pavement texture proﬁle parameters

on the development of more meaningful tire wear and tire-

pavement friction relationships. Moreover, the developed re-

lationships will be useful in the design of pavements with ac-

ceptable tire wear rates with a sufﬁcient level of skid

resistance. This can be achieved by selecting suitable mix de-

signs that provide the optimum combination of microtexture

and macrotexture.

Furthermore, the newly developed correlations can also be

used to predict the tire wear rate for a given pavement based

on regular Mu-meter measurements and grease patch test data.

This is clearly demonstrated by predicting the tire wear rates

for different sections of a concrete runway at Luke Air Force

Base.

ACKNOWLEDGMENT

The writers wish to acknowledge the technical assistance of Dr. Gray

Mullins, Assistant Professor of Civil Engineering, University of South

Florida, Tampa, in assembling the tire-wear simulator.

APPENDIX. REFERENCES

Gunaratne, M., and Bandara, N. (1997). ‘‘Correlations of pavement tex-

ture characteristics to tire wear and tire-pavement friction.’’ Final Rep.

Submitted to Wright-Patterson AFB, Dayton, Ohio.

Henry, J. J. (1968). ‘‘Tire wet-pavement traction measurement:A state-of-

the-art review.’’ The tire pavement interface, ASTM STP 929,M.G.

Pottinger and T. J. Yager, eds., ASTM, West Conshohocken, Pa., 3–25.

HQ AFCESA. (1992). ‘‘Runway friction characteristics evaluation, Luke

AFB, AZ.’’ Rep. No. DSN 523-6429, Headquarters Air Force Civil

Engineering Support Agency, Pavement Surface Effects Team, Tyndale

Air Force Base, Fla.

Kummer, H. W. (1966). ‘‘Uniﬁed theory of rubber friction.’’ Engrg. Res.

Bull. B-94, Penn State University, State College, University Park, Pa.

Leu, S. J., and Henry, J. J. (1978). ‘‘Prediction of skid resistance as a

function of speed from pavement texture.’’ Transp. Res. Rec. 666,

Transportation Research Board, Washington, D.C., 38–43.

Lowne, R. W. (1971). ‘‘Effect of road surface texture on tire wear.’’

Rubber Chem. and Technol., 44(5), 1159–1160.

Marcondes, J., et al. (1991). ‘‘Spectral analysis of highway pavement

roughness.’’ J. Transp. Engrg., ASCE, 117(5), 540–549.

Moore, D. (1972). The friction and lubrication of elastomers. Pergamon,

New York.

J. Mater. Civ. Eng. 2000.12:46-54.