Evaluation of Construction Practices That Influence the

Bond Strength at the Interface between Pavement Layers

Laith Tashman, A.M.ASCE1; Kitae Nam2; Tom Papagiannakis, P.E., M.ASCE3; Kim Willoughby, P.E.4;
Linda Pierce, P.E.5; and Tom Baker, P.E.6

Abstract: This study investigated the influence of several construction practices on the bond strength at the interface between pavement
layers. These practices included the surface treatment, curing time, residual application rate, and equipment tracking. Three tests were
performed for estimating the bond strength between an existing hot mix asphalt 共HMA兲 and a newly constructed HMA overlay, namely
the Florida Dept. of Transportation shear tester, the University of Texas at El Paso pull off test, and the torque bond test. Testing involved
a CSS-1 type emulsion as the tack coat. The results from the three tests were statistically analyzed. Generally, milling provided a
significantly better bond at the interface between the existing surface and the new overlay. Curing time had a minimal effect on the bond
strength. The results indicated that the absence of tack coat did not significantly affect the bond strength at the interface for the milled
sections, whereas it severely decreased the strength for the nonmilled sections. The results also showed that increasing the residual rate of
tack coat did not generally affect the bond strength at the interface.
DOI: 10.1061/共ASCE兲0887-3828共2008兲22:3共154兲
CE Database subject headings: Bonding; Construction management; Statistics; Pavements; Interfaces.

Introduction A tack coat is considered as a simple, relatively inexpensive,

A tack coat is a light application of an asphaltic emulsion between
pavement lifts, most commonly used between an existing surface
and a newly constructed overlay. The role of a tack coat is to
provide adequate adhesive bond between pavement lifts so that
they behave as a monolithic structure. The inadequacy or failure
of this bond causes slippage between the pavement layers, which
results in a significant reduction in the shear strength of the pave-
ment structure, thus making it more susceptible to a variety of
distresses, such as cracking, rutting, and potholes 共WSDOT
2003兲.
1 It is generally accepted that the bond characteristics at the
Assistant Professor, Dept. of Civil and Environmental Engineering,
Univ. of Jordan, Amman 99164, Jordan 共corresponding author兲. E-mail:
ltashman@wsu.edu
2
WCAT Manager and Research Associate, Dept. of Civil and Envi-
ronmental Engineering, Washington State Univ., Pullman, WA 99164-
2910. E-mail: knam@wsu.edu
3
Professor and Chair, Dept. of Civil and Environmental Engineering,
Univ. of Texas-San Antonio, San Antonio, TX 78249. E-mail:
at.papagiannakis@utsa.edu
4 use of slow-setting type of emulsions. The most common among
Research Manager, Construction, Materials, Bridges and Structures,
and Maintenance, Washington State Dept. of Transportation, Olympia,
WA 98504-7372. E-mail: willouk@wsdot.wa.gov
5
State Pavement Engineer, Washington State Dept. of Transportation,
Tumwater, WA 98512-6951. E-mail: piercel@wsdot.wa.gov
6
State Materials Engineer, Washington State Dept. of Transportation,
Tumwater, WA 98512-6951. E-mail: bakert@wsdot.wa.gov
Note. Discussion open until November 1, 2008. Separate discussions
must be submitted for individual papers. To extend the closing date by
one month, a written request must be filed with the ASCE Managing
Editor. The manuscript for this paper was submitted for review and pos-
sible publication on March 15, 2007; approved on May 25, 2007. This
paper is part of the Journal of Performance of Constructed Facilities,
Vol. 22, No. 3, June 1, 2008. ©ASCE, ISSN 0887-3828/2008/3-154–161/
$25.00.
yet essential step in the pavement construction process. However,
there is currently a lack of unified guidelines on the construction
practices and quality control/acceptance of tack coats. In a recent
study conducted by Washington State Department of Transporta-
tion 共WSDOT兲, field cores were extracted to analyze the mode of
cracking failure. Despite the fact that WSDOT requires tack coat
on all hot mix asphalt 共HMA兲 paving surfaces, approximately one
third of the cores that exhibited “top-down cracking” debonded at
the interface between the existing pavement and the subsequent
overlay during extraction. These bond failures raise concerns
about the adequacy of the adhesive bond achieved under current
pavement construction practices.
interface between an existing pavement and an overlay are af-
fected by several factors including the tack coat type, residual
application rate, curing time, and surface condition. Existing lit-
erature reveals that various kinds of materials are used as tack
coats between HMA layers. These include asphalt emulsions, pav-
ing grade asphalt binders, and cutback asphalts. Paul and
Scherocman 共1998兲 indicated that most states have adopted the
them are SS-1, SS-1h, CSS-1, and CSS-1h.
An excessive amount of tack coat can cause slippage, whereas
too little may result in debonding problems. Therefore, it is im-
portant to estimate the optimum amount of tack coat that will
produce the best performance. Mohammad et al. 共2002兲 recom-
mended an optimum residual rate of 0.09 L / m2. Lavin 共2003兲
recommended application rates of 0.2– 1.0 L / m2 and suggested
that milled pavements may require application rates of 1.0 L / m2
or more due to a larger surface area caused by grooving. Sholar et
al. 共2004兲 developed a shear testing device to evaluate shear
strength of HMA overlays. Their study also involved the con-
struction of three field projects and the evaluation of several vari-
ables that could affect the bond strength between HMA layers.

154 / JOURNAL OF PERFORMANCE OF CONSTRUCTED FACILITIES © ASCE / MAY/JUNE 2008

Downloaded 13 Oct 2010 to 80.90.175.248. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
Their study showed that there was a slight effect of the residual
rates on the shear strength. They recommended a residual rate of
0.09 L / m2 as a minimum and 0.23 L / m2 as an optimum.
There is no unanimous agreement in the literature on the cur-
ing time of tack coats. Some research studies and guidelines sug-
gest that tack coat be cured before laying the new pavement layer
共AI 1998; AI 2001; AI 2001; Flexible Pavements of Ohio 2001;
TxDOT 2001; U.S. Army Corps of Engineers 1991兲. Sholar et al.
共2004兲 concluded that the shear strength slightly increased with
curing time. On the other hand, Lavin 共2003兲 reported that an
overlay could be placed either directly after the tack coat has been
applied or after it has broken. The bond between the layers will
still be created either way 共Lavin 2003兲.
The surface condition of an existing pavement is a very im-
portant factor that influences the bond strength at the interface. It
is generally recommended that the tack coat be applied to a clean
and dry surface. The recommended cleaning method is to sweep
the surface with a power broom 共AI 1989; AI 2001; Brown et al.
2000; Flexible Pavements of Ohio 2001; TxDOT 2001; U.S. Fig. 1. Layout of test sections. Approximate target residual rates are
Army Corps of Engineers 1991兲. Sholar et al. 共2004兲 found that listed in L / m2.
coarse graded HMA mixes created higher interface shear strength
compared to fine grade mixes and that as the surface roughness Exit 116 of northbound Interstate I-5 in Olympia, Wash. The
increased, the influence of the tack coat application rate dimin- weather condition during the experiment was sunny and clear
ished. Milling increased the shear strength at the interface and with an average high temperature of 23° C and a wind speed of
reduced the effect of the application rate. For milled sections, it 5 km/ h. Fig. 1 shows a layout of the test sections. The first seven
was noticed that using tack coat was not effective in increasing sections were milled and cleaned using a broom, whereas Sec-
the shear strength at the interface. West et al. 共2005兲 concluded tions 8–14 were not milled but cleaned 共nonmilled sections兲. A
that fine-graded mixtures generally had higher bond strengths nondiluted CSS-1 tack coat emulsion was applied to the test sec-
than the coarse-graded mixture when tested at 25° C. However, tions at four different target residual rates of 0.00 共no tack兲, 0.08,
they also reported that there were significant interactions of mix 0.22, and 0.32 L / m2. A new 50 mm overlay was placed using a
type with other variables 共application rate, materials used, and Superpave 12.5 mm nominal maximum aggregate size HMA mix-
testing temperature兲, which would reverse this trend in some ture. The odd sections 共except Section 7兲 were paved after the
cases. tack coat had broken, i.e., the tack coat had enough time to cure
and set 共approximately 2.5 h兲. The placement of HMA on the
even sections 共except Section 8兲 began approximately 3 min after
Objective the tack coat had been applied 共unbroken sections兲. Test Sections
Number 7 and 8 had no tack coat 共no tack sections兲.
The objective of this study was to investigate the influence of The residual tack coat rates were measured by WSDOT ac-
several tack coat construction practices on the adhesive bond at cording to ASTM D2995. One measurement was taken for each
the interface between pavement lifts. These practices included the section 共except the no tack Sections Number 7 and 8兲. It should
surface treatment, tack coat curing time, tack coat residual rate, be noted that WSDOT has modified the ASTM D2995 test to
and equipment tracking. Three tests were performed to estimate
the bond strength between an existing HMA and a newly con-
structed HMA overlay. Table 1. Target and Measured Residual Tack Coat Rates 共L / m2兲
Average
Target Measured measured
Experiment Test residual residual residual
section rate rate rate
The experiment began on September 13, 2005 under the supervi- 1 0.08 0.15
sion of WSDOT and in cooperation with the Washington Asphalt
2 0.08 0.21
Pavement Association, Woodworth & Company Inc., and Lake- 0.16
9 0.08 0.15
side Industries. It consisted of building 14 test sections in order to
10 0.08 0.15
investigate the influence of several construction factors on the
bond strength at the interface between the existing HMA surface 3 0.22 0.17
and the newly constructed HMA overlay. These factors included 4 0.22 0.19
0.18
the following: 11 0.22 0.16
• Surface treatment: milled versus nonmilled; 12 0.22 0.21
• Curing time: broken versus unbroken; 5 0.32 0.19
• Approximate target residual rate: 0.00, 0.08, 0.22, and 6 0.32 0.23
0.32 L / m2; and 0.21
13 0.32 0.20
• Equipment tracking: wheel path 共WP兲 versus middle of lane
14 0.32 0.20
共ML兲.
Correlation 0.62
The experiment took place at the Nisqually weigh station near

JOURNAL OF PERFORMANCE OF CONSTRUCTED FACILITIES © ASCE / MAY/JUNE 2008 / 155

Downloaded 13 Oct 2010 to 80.90.175.248. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
Fig. 2. Varying residual tack coat rates 共approximate兲 on existing
surface

measure the residual rate instead of the application rate. Table 1

shows the target and measured residual rates. The correlation be-
tween them was 0.62 indicating that, in general, high values of
target residual rates corresponded to high values of measured re-
sidual rates and low values of target residual rates corresponded
to low values of measured residual rates. Nonetheless, the test
overpredicted the target residual rates at lower rates and under-
predicted them at higher rates. This could be attributed to an
inaccuracy in measuring the target residual rates, lack of samples,
an inaccuracy in applying the target rates by the truck, or any
combination of these factors. However, it was noticed during the
placement of the tack coat that the variations in the residual rates
among the test sections were larger than the predicted values by
the ASTM test as visually shown in Fig. 2. Therefore, it was
decided to use the target residual rates in the subsequent statistical Fig. 3. Three tests performed in this study: 共a兲 FDOT shear tester; 共b兲
analyses. It is noteworthy that the statistical analyses were also torque bond; and 共c兲 UTEP pull-off 共after Sholar et al. 2004兲
performed using the measured values and the results were similar
to those of the target residual rates. This was anticipated given the
correlation between them. field core is conditioned at a temperature of 25° C for 2 h.
The experiment involved performing three tests, namely the The torque bond test was originally developed in Sweden for
FDOT shear tester, the torque bond test, and the UTEP pull-off the in situ assessment of bond conditions and has been adopted in
test. Fig. 3 shows pictures of these three tests. Details about these the United Kingdom as part of the approval system for thin sur-
tests are documented by Tashman et al. 共2006兲. In the FDOT facing systems. Recently, a laboratory-based torque test was de-
shear tester, the field core is placed between the shear plates so veloped allowing the test to be conducted in a more controlled
that the direction of traffic marked on the core is parallel to the environment 共Choi et al. 2005兲. The laboratory torque bond test
shear direction. The core is then deformed at a constant rate of was conducted at 20° C. Torque was applied to the plate until
50 mm/ min until failure occurs. Before performing the test, the failure of the bond occurs or a torque of 400 N m is exceeded

156 / JOURNAL OF PERFORMANCE OF CONSTRUCTED FACILITIES © ASCE / MAY/JUNE 2008

Downloaded 13 Oct 2010 to 80.90.175.248. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
共British Board of Agrément standards require 300 N m兲.
The pull-off test was developed at the University of Texas at
El Paso 共UTEP兲. It measures the tensile strength of the tack coat
before a new overlay is paved. After the tack coat is applied and
allowed to set 共typically less than 30 min兲, the device is placed on
the tack-coated surface. The torque wrench is rotated clockwise
until the contact plate is firmly set on the tack-coated surface. An
88 kg 共40 lb兲 load is placed on the weight key 共at the top of the
device兲 for 10 min prior to testing in order to set the contact plate.
The load is then removed and the torque wrench is rotated in the
counterclockwise direction to detach the contact plate from the
tack-coated surface. The torque required to detach the contact
plate is recorded and then converted to tensile strength using a
calibration factor.
The UTEP pull-off test was performed on the broken sections
before paving. Two devices were used simultaneously: one on the
milled sections and the other on the nonmilled sections. For the
milled sections, the test was performed in the following order: 1,
3, 5, 1, 3, 5 with approximately 12 min between each test. The
same order was followed for the nonmilled sections, i.e., 9, 11,
13, 9, 11, 13. Thus, for each section, the test was performed twice,
with the second test being performed approximately 45 min after
the first.
The day following paving and compaction, five nuclear den-
sity measurements were taken for each section. The average den-
sity for all the measurements was 92.9% and the coefficient of
variation was 2.1%. This was an indication of the uniformity in
the density, thus it was anticipated that it would not be a factor
that would significantly affect the analyses.
Twelve 150 mm cores were taken from each test section, of
which six were taken in the wheel path and six were taken in the
middle of the lane 共three replicates for each of the Florida Dept.
of Transportation 共FDOT兲 shear tester and the torque bond test兲. Fig. 4. Tracking as seen: 共a兲 on different WSDOT HMA project; 共b兲
This was to assess the potential loss of tack coat as it adheres to during this experiment
the tires of the paving and delivery equipment 共tracking兲. Equip-
ment tracking typically occurs in both wheel paths, which are the
most critical locations for the new surface to bond to the existing P value. The P value is the smallest level of significance at which
surface. During this experiment, the tack coat remained in place, the factor is considered significant in affecting the response. A
thus tracking was not an issue as shown in Fig. 4 共i.e., very slight level of significance equal to 0.05 was considered for the purpose
tracking occurred兲. of this study, i.e., 95% confidence. A P value of 0.05 or less
A total of 168 field cores were cored. However, all the cores indicates the factor is significant, whereas a P value greater than
taken from test Section Number 8 debonded at the interface dur- 0.05 indicates the factor is not.
ing sampling. These cores were not tested but were included in
the analyses 共i.e., zero bond strength兲. In addition, two cores from
test section Number 9, two from test section Number 14, and one FDOT Shear Tester
from test section Number 1, broke during sampling, shipping, or The number of samples analyzed for the purpose of this test was
testing. These cores were excluded from the analyses. Thus, a 80. These include the six samples from test section Number 8 that
total of 151 cores were tested. Of these, 74 cores were tested were cored but not tested because they debonded due to the lack
using the FDOT shear tester and 77 were tested using the torque of bond strength 共i.e., their shear strength was assumed to be
bond test. zero兲. The average shear strength of all 80 samples was 809 kPa
with a standard deviation of 477.5 kPa. The randomization
scheme of the experiment dictated the ANOVA. The randomiza-
Results and Analyses tion was done at various levels. At first, the road was divided into
two surface condition categories: “milled” and “nonmilled” sec-
Analysis of variance 共ANOVA兲 was used to determine the factors tions. Hence, surface condition was considered a main effect fac-
that significantly influenced the bond strength at the interface be- tor. Thereafter, within each category, the surface condition was
tween the existing surface and the newly constructed overlay. divided into three curing time categories: no tack, broken, and
ANOVA refers broadly to a collection of statistical procedures unbroken. Hence, the curing time was considered as “nested”
suited to quantitatively determine whether particular treatments within the surface condition. Target residual rates of 0.08, 0.22,
and their combinations affect a particular outcome, which is the and 0.32 L / m2 were applied within the broken and unbroken sec-
bond strength at the interface in this study. MINITAB statistical tions. As a result the residual rate was nested within the curing
software 共MINITAB Release 14.20兲 was used to perform the sta- time. The tracking effect 共ML versus WP兲 was over the entire
tistical analyses. The ANOVA table consists of a value called the road, hence it was considered as a main effect factor.

JOURNAL OF PERFORMANCE OF CONSTRUCTED FACILITIES © ASCE / MAY/JUNE 2008 / 157

Downloaded 13 Oct 2010 to 80.90.175.248. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
Table 2. General Linear Model for FDOT Shear Tester 共Response: Shear Strength兲
Factor Type Levels Values
Surface condition Fixed 2 Milled, nonmilled
-Curing time Fixed 6 Broken, no tack, unbroken, broken,
no tack, unbroken
-Residual rate Fixed 14 0.08, 0.22, 0.32, 0.00, 0.08, 0.22, 0.32,
0.08, 0.22, 0.32, 0.00, 0.08, 0.22, 0.32
Tracking Fixed 2 Middle of lane 共ML兲, wheel path 共WP兲
ANOVA for FDOT shear tester
Factor DFa SSb MSc Fd P value
e
Surface condition 1 280,358 280,358 507.10 0.000
-Curing time 4 21,847 5,462 9.88 0.000e
-Residual rate 8 20,642 2,580 4.67 0.000e
Tracking 1 1,714 1,714 3.10 0.083
Error 65 35,936 553 — —
Total 79 — — — —
Note: Tabbing indicates a nested factor.
a
Degrees of freedom.
b
Sum of squares.
c
Mean square, which is the SS divided by DF.
d
Ratio of mean squares. It is used to determine the P value.
e
Factor is significant 共P value⬍ 0.05兲.

Table 2 shows the ANOVA table for the FDOT shear tester
results. Except for the tracking factor, all other factors are statis-
tically significant. Fig. 5 shows the effect of the surface condition
and curing time on the FDOT shear strength. There is a significant
difference in the shear strength between the milled and nonmilled
sections: the milled sections had significantly higher shear
strength than the nonmilled sections. The Tukey test was con-
ducted in order to obtain the pairwise comparisons for the curing
time. For the milled sections, the pairwise comparisons showed
that there was no significant difference in the mean shear strength
between the three curing time categories. For the nonmilled sec-
tions, there was a significant difference between the no tack and
the broken sections as well as between the no tack and the unbro-
ken sections. There was no significant difference between the
broken and the unbroken sections for the two surface conditions.
Thus, according to the Tukey comparisons for the FDOT shear
tester results, curing time was an insignificant factor.
Fig. 6 shows the effect of the residual rate nested within the
curing time. The Tukey test was also conducted in order to obtain
the pairwise comparisons. For the milled sections, the pairwise
comparisons showed that the only significant difference was be-
tween 0.08-unbroken and the 0.32-unbroken. For the nonmilled
sections, the significant difference was between the no tack and
all other combinations except for the 0.08-broken. Accordingly,
the absence of tack coat did not significantly affect the shear
strength of the milled sections, whereas it significantly did so for
the nonmilled sections. Furthermore, increasing the residual rate
did not generally improve the shear strength either for the milled
sections or for the nonmilled ones. There was one exception to
that, which was between the 0.08-unbroken and the 0.32-
unbroken milled sections.
Tracking effect was found to be statistically insignificant. This
was attributed to the negligible tracking that occurred during this
experiment.
Torque Bond Test
The number of samples analyzed for the purpose of this test was
83. These include the six samples from test section Number 8 that

Fig. 5. Effect of curing time nested within surface condition on Fig. 6. Effect of residual rate nested within curing time on FDOT
FDOT shear strength shear strength

158 / JOURNAL OF PERFORMANCE OF CONSTRUCTED FACILITIES © ASCE / MAY/JUNE 2008

Downloaded 13 Oct 2010 to 80.90.175.248. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
Table 3. Generalized Linear Model for Torque Bond Test 共Response: Torque兲
Censoring information count 23 uncensored value, 60 right censored value
Estimation method Maximum likelihood
Distribution Normal
Regression table
Factor Coefficient Standard error Z P value
Surface condition 共milled versus nonmilled兲 −430.988 86.700 −4.97 0.000a
Curing time 共broken versus no tack兲 −280.832 94.550 −2.97 0.003a
Curing time 共no tack versus unbroken兲 324.839 97.194 3.34 0.001a
Curing time 共broken versus unbroken兲 44.007 56.544 0.78 0.436
Tracking 共WP versus ML兲 −46.516 49.997 −0.93 0.352
Residual rate 共0.08 versus 0.22, 0.08 versus 430.234 1,397.58 0.31 0.758
0.32, and 0.22 versus 0.32兲
Scale 155.395 25.833 — —
a
Factor is significant 共P value⬍ 0.05兲.

were cored but not tested because they debonded due to the lack
of bond strength 共i.e., their torque bond strength was assumed to
be zero兲. The average torque of all 83 samples was 342.1 N m
with a standard deviation of 115.0 N m. Although the torque bond
test was performed according to the British Board of Agrément
standards, which require a torque wrench of 300 N m capacity
共400 N m was used in this study兲, only 28% of the tested cores
failed at a torque value lower than 400 N m 共23 cores out of a
total of 83 including the ones from test section Number 8兲. In
order to take into account the high-censored nature of the torque
bond data, “regression with life data” was used. Censored obser-
vations are those for which an exact failure time or value is un-
known. Cores that did not fail before reaching the 400 N m torque
value were considered right censored, whereas cores that failed
before reaching the 400 N m torque value were considered left
censored.
The response variable 共torque兲 was regressed using the surface
condition, curing time, tracking, and residual rate as explanatory
variables. The results from regression are presented in Table 3.
Table 3 indicates that there is a significant difference between the
milled and the nonmilled sections. Milled sections yielded a
higher torque than the nonmilled sections as shown in Fig. 7. In
fact, all but one of the cores that failed at torque values less than
400 N m were cored from the nonmilled sections. Fig. 7 also
indicates that for the nonmilled sections, there was a significant
difference between the no tack sections 共zero torque bond兲 and
the broken ones as well as between the no tack sections and the
unbroken ones. No other factors were found to be significant.
It should be noted that test section Number 8, which recorded
zero torque with the torque bond test and zero shear stress with
the FDOT shear tester 共due to delamination of the cores during
removal兲, experienced minor shoving and cracking problems dur-
ing construction as shown in Fig. 8. This could be an indication
that the lack of bond at the interface manifests itself in a prema-
ture pavement failure.
UTEP Pull-Off Test
The number of samples analyzed for the purpose of this test was
12. The average tensile strength of all 12 samples was 12.5 kPa
with a standard deviation of 4.9 kPa. The randomization scheme
of the experiment also suggested that the ANOVA approach is
suitable for analyzing the UTEP pull-off test results. The random-
ization was done at various levels. At first, the road was divided
into two surface condition categories: milled and nonmilled sec-
tions. Hence, surface condition was considered a main effect fac-
tor. Thereafter, within each category, the surface condition was
divided into three residual rates: 0.08, 0.22, and 0.32 L / m2. As a
result, residual rate was nested within the surface condition. The
testing time 共A versus B, where Test B was performed 45 min
after Test A兲 effect was over the entire road and hence, was con-
sidered as a main effect factor.

Fig. 7. Effect of curing time on torque bond strength 共torque bond

test can measure torque up to 400 N m兲 Fig. 8. Cracking problems in test section Number 8

JOURNAL OF PERFORMANCE OF CONSTRUCTED FACILITIES © ASCE / MAY/JUNE 2008 / 159

Downloaded 13 Oct 2010 to 80.90.175.248. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
Table 4. General Linear Model for UTEP Pull-Off Test 共Response: Tensile Strength兲
Factor Type Levels Values
Surface condition Fixed 2 Milled, nonmilled
-Residual rate Fixed 6 0.08, 0.22, 0.32, 0.08, 0.22, 0.32
Testing time Fixed 2 A, B 共B is performed
45 min, after A兲
ANOVA for UTEP pull-off test
Factor DFa SSb MSc Fd P value
Surface condition 1 3.7297 3.7297 15.96 0.010e
-Residual rate 4 0.7323 0.1831 0.78 0.582
Testing time 1 0.8060 0.8060 3.45 0.122
Error 5 1.1685 0.2337 — —
Total 11 — — — —
Note: Tabbing indicates a nested factor.
a
Degrees of freedom.
b
Sum of squares.
c
Mean square, which is the SS divided by DF.
d
Ratio of mean squares. It is used to determine the P value.
e
Factor is significant 共P value⬍ 0.05兲.

Table 4 shows the ANOVA table for the UTEP pull-off test.
The table shows that the surface condition was the only signifi-
cant factor. Fig. 9 shows that the nonmilled sections had a higher
pull-off strength than the milled sections. The UTEP pull-off con-
tact plate did not adhere properly to the milled surface as well as
it did to the nonmilled one. Hence, the UTEP pull-off test results
indicated that the nonmilled sections had higher tensile strength
than the milled ones.
In general, the pull-off strength decreases with the residual
rate. The UTEP pull-off test was performed in chronological order
starting from the lowest rate and moving to higher rates, with
approximately 12 min between successive tests. This might have
influenced the results of the residual rates. However, this factor
was found to be insignificant.
The UTEP pull-off strength decreased with time. The average
tensile strength was 14.2 kPa for “A” samples and 10.8 kPa for
“B” samples, which is a decrease of 25%. However, this factor
was also found to be insignificant.
Summary
Three tack coat construction quality tests were performed to esti-
mate the bond strength at the interface between an exiting HMA
and a newly constructed HMA overlay. These tests were the
FDOT shear tester, torque bond test, and UTEP pull-off test.
These tests were used to study the influence of several construc-
tion factors on the bond strength provided by the tack coat at the
interface between pavement lifts. These factors included the sur-
face treatment, curing time, application rate, and equipment track-
ing. The results from each test are summarized subsequently.
FDOT Shear Tester
1. The milled sections had significantly higher shear strength
than the nonmilled sections;
2. Curing time was an insignificant factor;
3. The absence of tack coat did not affect the shear strength for
the milled sections, whereas it significantly did so for the
nonmilled sections;
4. Generally, increasing the residual rate beyond 0.08 L / m2 did
not significantly improve the shear strength for either the
milled sections or the nonmilled ones. However, the milled
sections were more sensitive to increasing the residual rate
compared to the nonmilled sections; and
5. The shear strength at the interface was not affected by the
slight equipment tracking.

Torque Bond Test

1. The milled sections had significantly higher strength than the

nonmilled ones;
2. Curing time had no effect on the strength at the interface; and
Fig. 9. Effect of residual rate nested within surface condition on 3. The absence of tack coat significantly decreased the strength
pull-off tensile strength at the interface for the nonmilled sections.

160 / JOURNAL OF PERFORMANCE OF CONSTRUCTED FACILITIES © ASCE / MAY/JUNE 2008

Downloaded 13 Oct 2010 to 80.90.175.248. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
UTEP Pull-Off Test
1. The nonmilled sections had a higher pull-off strength than
the milled sections. This was the only significant factor.
Conclusions
The three tests have different testing mechanisms. The FDOT
shear tester measures the bond strength of the interface between
the two lifts in shear, the torque bond test measures it in torsional
shear, whereas the UTEP pull-off test measures the tensile
strength of the tack coat. The FDOT shear tester seems to better
simulate the state of stress encountered in the field 共shear stress兲
that causes the debonding at the interface between pavement lay-
ers. The experimental data from the torque bond test were highly
censored, and thus provided limited observations. Nonetheless,
the limited observations from the torque bond test were consistent
with those from the FDOT shear tester. The results from the
UTEP pull-off test were generally different from the other two
tests.
Overall, milling provided a significantly better bond at the
interface between the existing surface and the new overlay. For
milled sections, the absence of tack coat did not significantly
affect the bond strength at the interface. This was not true for the
nonmilled sections, where the absence of tack coat severely de-
creased the bond strength 共there was no bond at all兲. Curing time
had minimal effect on the bond strength at the interface. Residual
rates in the range of 0.08– 0.32 L / m2 did not generally affect the
bond strength at the interface. Equipment tracking did not occur
to the extent expected during the experiment, hence its effect on
the bond strength was insignificant.
It should be noted that these conclusions were drawn based on
the initial results from this study, which only used one type of
tack coat and one HMA mixture for the existing surface as well as
the new overlay. Also, this study was conducted in 1 day where
the weather condition remained almost unchanged. The results
could differ if the experiment was conducted under different
weather conditions 共effect of moisture and/or temperature兲. Fur-
thermore, long-term performance data have not been collected at
this time.
Recommendations
It is recommended to take field cores from the experimental site
in the next several years and measure the bond strength using the
FDOT shear tester and the torque bond test. This will complement
the results from this study with performance data, which will
assist in pinpointing the factors that significantly affect the bond
strength at the interface from a long-term performance perspec-
tive. Furthermore, this may assist in determining critical quality
control/quality assurance values that should be achieved by either
test to ensure that the pavement will have adequate adhesive bond
at the interface between layers. It is also recommended to perform
further testing on the measurement of the residual rates of tack
coat to determine if the current ASTM test method is adequate for
capturing the actual residual rates or not.
JOURNAL OF PERFORMANCE OF CONSTRUCTED FACILITIES © ASCE / MAY/JUNE 2008 / 161
Downloaded 13 Oct 2010 to 80.90.175.248. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org
Acknowledgments
A number of individuals and agencies have contributed to the
successful completion of this study. This research was sponsored
by the Washington State DOT through a pooled-funded study
with contributions from Minnesota DOT, Texas DOT, and Florida
DOT. Special thanks go to John Grisham of Woodworth & Com-
pany Inc.; Dave Bell of Lakeside Industries; Tom Gaetz of Wash-
ington Asphalt Pavement Association; Vivek Tandon of the
University of Texas at El Paso; Gregory Sholar of the Florida
DOT, Mel Hitzke, Terry MacAuley, Dave Mayoh, Mark Wil-
loughby, Mark Rickert, and John Brooks of Olympic and North-
west Region personnel; Jeff Uhlmeyer and Jim Weston of
Headquarters Materials Laboratory personnel; Louay Mohammad
of Louisiana State University; and Joe Button of Texas A&M
University.
References
Asphalt Institute 共AI兲. 共1989兲. The asphalt handbook, Manual Series No.
4 共MS-4兲, Lexington, Ky.
Asphalt Institute 共AI兲. 共2001兲. A basic asphalt emulsion manual, 3rd Ed.,
Manual Series No. 19 共MS-19兲, Lexington, Ky.
Asphalt Institute 共AI兲. 共2001兲. Construction of hot mix asphalt pave-
ments, 2nd Ed., Manual Series No. 22 共MS-22兲, Lexington, Ky.
Brown, E. R., et al. 共2000兲. The hot-mix asphalt paving handbook 2000,
AC 150/5370-14A, U.S. Army Corps of Engineers, Washington, D.C.
Choi, Y., Collop, A., Airey, G., and Elliott, R. 共2005兲. “A comparison
between interface properties measured using the Leutner test and the
torque test.” Electron. J. Assoc. Asph. Paving Technol., 74E.
Flexible Pavements of Ohio. 共2001兲. “Proper tack coat application.” Tech-
nical Bulletin, Columbus, Ohio.
Lavin, P. G. 共2003兲. Asphalt pavements. E & FN Spon, New York.
Mohammad, L. N., Raqib, M. A., and Huang, B. 共2002兲. “Influence of
asphalt tack coat materials on interface shear strength.” Transporta-
tion Research Record. 1789, Transportation Research Board, Wash-
ington, D.C., 56–65.
Paul, H. R., and Scherocman, J. A. 共1998兲. “Friction testing of tack coat
surfaces.” Transportation Research Record. 1616, Transportation Re-
search Board, Washington, D.C., 6–12.
Sholar, G., Page, G., Musselman, J., Upshaw, P., and Moseley, H. 共2004兲.
“Preliminary investigation of a test method to evaluate bond strength
of bituminous tack coats.” Electron. J. Assoc. Asph. Paving Technol.,
73, 771–801.
Tashman, L., Nam, K., and Papagiannakis, T. 共2006兲. “Evaluation of the
influence of different tack coat field procedures on the bond strength
at the interface between pavement layers.” Rep. No. WA-RD 645.1,
Washington State Department of Transportation, Olympia, Wash.
Texas Department of Transportation 共TxDOT兲. 共2001兲. “Proper use of
tack coat.” Technical Advisory 2001-1, Construction and Bridge Divi-
sions, Austin, Tex.
U.S. Army Corps of Engineers 共USACE兲. 共1991兲. Guide specifications
for military construction, No. CEGS-02556, Washington, D.C.
Washington State Department of Transportation 共WSDOT兲. 共2003兲. “Tack
coat tech note.” 具http://training.ce.washington.edu/WSDOT/Modules/
07-construction/tack-coats.htm典.
West, R. C., Zhang, J., and Moore, J. 共2005兲. “Evaluation of bond
strength between pavement layers.” NCAT Rep. No. 05-08, National
Center for Asphalt Technology, Auburn, Ala.