Construction and Building Materials 145 (2017) 88–97

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

Accurate detection method for compaction uniformity of asphalt

pavement
Zhengqi Zhang a,⇑, Shuolei Huang a, Ke Zhang b
a
Key Laboratory for Special Area Highway Engineering of Ministry of Education, Chang’an University, Xi’an, Shaanxi 710064, China
b
Fuyang Normal University, College of Information Engineering, Fuyang, Anhui 236041, China

h i g h l i g h t s

The reasonable detection interval of PQI is determined.

The detection points in the road can be divided into high, medium and low density regions where the core samples are took according to proportion.

The void distribution characteristic on sectional image of core sample is analyzed by image processing technology.

The coefficients UCN and UCR are put forward to evaluate quantitatively compaction uniformity of asphalt pavement from the microscopic structure.

a r t i c l e i n f o a b s t r a c t

Article history: To evaluate the compaction uniformity effectively, firstly, the feasibility of using density and void distri-
Received 23 August 2016 bution as compaction uniformity evaluation index was demonstrated by laboratory test. Secondly, rely-
Received in revised form 27 March 2017 ing on the project of Tong-Xun Highway, reasonable detection interval of non-nuclear density gauge PQI
Accepted 29 March 2017
was determined, and quantitative evaluation index of density distribution uniformity was also proposed.
Available online 7 April 2017
Next, core samples at representative density point were drilled and of which the voids distribution char-
acteristic on sectional image of core sample was analyzed by image processing technology, and further
Keywords:
on, the evaluation index of void distribution uniformity was put forward. Finally, the accurate detection
Road engineering
Asphalt pavement
method of compaction uniformity based on density and void distribution was exampled relying on enti-
Compaction uniformity tative engineering. The results show that it is feasible to evaluate the compaction uniformity of asphalt
Detection method pavement with its density and void distribution. The reasonable lateral detection interval of PQI is 1.5 m,
Density distribution and the reasonable longitudinal detection space is 50 m. Density statistical index such as Sr, SR, k and h
Void distribution can reflect the density distribution uniformity on detection section. Voids number uniform coefficient
UCN and voids ratio uniform coefficient UCR of core samples can reflect the compaction uniformity of
asphalt pavement from the aspect of micro structure.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction causes of asphalt mixture segregation and improving

Ensuring quality uniformity and stability is the technical
requirements which any structures and materials must meet, so
the asphalt pavement must meet them to prevent the asphalt
pavement from non-uniformity phenomenon, such as gradation
segregation, temperature segregation, uneven compaction etc. At
present, the studies on non-uniformity of asphalt pavement mostly
concern about the mixture design and production stage, and most
study focused on the segregation effects on the performance of
asphalt mixture using the laboratory simulation method, the
⇑ Corresponding author.
E-mail addresses: z_zhengqi@126.com (Z. Zhang),
(S. Huang), zhangke_365@126.com (K. Zhang).
segregation-resistance of the mixture and evaluating segregation
of the mixture are also the research hotspots. China also lacks
the evaluation standard of pavement construction uniformity in
the quality inspection and acceptance of asphalt pavement con-
struction. Khedaywi found that gradation segregation caused sig-
nificant decrease of indirect tensile strength, rutting resistance
and water sensitivity of asphalt mixture [1,2]. Sivilevičius simu-
lated randomly the influence of aggregate gradation and the con-
tent of fine aggregate on the uniformity of asphalt mixture [3].
Liu studied the way to relieve the segregation in asphalt pavement
compaction, and found that the compaction segregation can be
controlled by some technical measures such as improving the
operating characteristics of stalls paver and roller machine [4].
h_shuolei@126.com
Yang studied the relativity between envelope area of segregation

http://dx.doi.org/10.1016/j.conbuildmat.2017.03.227
0950-0618/Ó 2017 Elsevier Ltd. All rights reserved.
 Z. Zhang et al. / Construction and Building Materials 145 (2017) 88–97
grading curve and design gradation curve and mixture perfor-
mance deviation, and proposed that the envelope area of grading
curve can be used to assess quantitatively the segregation of
asphalt mixture [5].
Some scholars have tried to use advanced equipment and image
processing technology, such as non-nuclear density meter PQI,
laser texture meter, etc to evaluate the uniformity of asphalt pave-
ment construction. Zhang analyzed the feasibility of detecting
asphalt pavement by using non-nuclear density meter PQI [6].
Chang proposed that the compaction uniformity is controlled by
Intelligent Compaction and IC technology measures ICMV (Intelli-
gent Compaction Measure Value) and gives optimal pass count
[7,8]. Bao proved the feasibility of using texture depth as the judg-
ment index of asphalt pavement segregation, and established eval-
uation standard of the different levels of segregation [9].Wang
used image processing technology to measure the texture depth
of asphalt pavement, and found that the testing method of texture
depth based on image processing was convenient and accurate
[10].Song handled image of asphalt pavement, and evaluated the
uniformity of texture distribution of asphalt pavement by calculat-
ing the variation coefficient of the concave area fraction under the
road surface [11]. On the basis of previous studies, some scholars
have tried to combine the density (compactness) with the texture
depth to evaluate the construction uniformity of asphalt pavement
[12,13]. However, above researches which evaluate construction
quality of asphalt pavement based on the macroscopic indexes
such as density and texture depth cannot fully reflect the unifor-
mity of pavement construction, especially the compaction
uniformity.
Compaction degree is very important to the asphalt mixture,
which influence the performance and duration of the asphalt
pavement.
For the asphalt mixture, With the increase of run times of com-
paction, mixture particles will become closer and closer, so, in a
certain volume, the solid phase increase while air phase decrease
according to the solid–liquid-air phase system theory, if the pave-
ment is detected, the two detection parameters will show such a
trend that the density of the pavement increases, void ratio
decreases. So density and void ratio can be used to reflect the com-
paction degree of asphalt pavement.
Density can reflect the overall construction quality of asphalt
pavement in certain areas, and void composition and distribution
could reflect the fine structure of a certain road area. The two
parameters characterized asphalt pavement construction quality
integrally and partially respectively. From the density and the void
distribution, precise method can be established to detect the com-
paction uniformity of asphalt pavement. The feasibility of using
density and void distribution as compaction uniformity evaluation
index was demonstrated by laboratory test in the paper. The non-
nuclear density gauge PQI, industrial CT system combined with
image processing technique were used to detect density distribu-
tion and internal void distribution of cored from asphalt pavement,
and finally an accurate detection method of compaction uniformity
were put forward based on density and void distribution. The
research results could provide a reference for the testing and eval-
uation of asphalt pavement compaction uniformity.
2. Experimental apparatus and method
Industrial CT: Industrial CT is the abbreviation of industrial
computed tomography. The utility model can display the inner
structure, the composition, the material and the defect condition
of the detected object clearly, accurately and visually in the form
of a two-dimensional fault image or a three-dimensional image
under the Nondestructive condition to the detected object.
89
IPP software: IPP software was commonly used in medical and
biological fields, and it had powerful function in the image size
measurement and counting, classification statistics and analysis .
The IPP software was used to process the CT fault image and ana-
lyze the voids in two-dimensional section, therefore, parameters
such as voids number and section void ratio in fault image of core
sample were obtained.
Experimental method: Voids number and section void fraction
of core can be processed by Methods of mathematical statistics
analysis.
3. Analysis on the evaluation index of compaction uniformity
3.1. Density distribution
AC-13 and AC-20 asphalt mixture with different gradation and
segregation degree were designed in the laboratory. Uneven com-
paction of asphalt pavement was simulated by different com-
paction temperature and wheel roller. AC-13 and AC-20
gradation composition of segregated mixture are shown in Table 1.
SK90# base asphalt is used.
Ran 24 times by wheel mill, non-segregation slab of asphalt
mixture of AC-13 and AC-20 were formed at a temperature of
155 °C, 140 °C, 125 C and 110 °C. The bulk volume density and void
ratio of each sample were measured in laboratory were shown in
Fig 1.
Fig 1 showed that under the same compaction condition, the
bulk density of the slab specimen increased with the forming tem-
perature. In the range of 110 °C–140 °C, the growth rate of bulk
volume density was relatively fast, however, the growth rate of
the bulk volume density became slow when the temperature
exceeded 140 °C. In contrast to the change trend of the volume
density, the void ratio of the specimen decreased rapidly with
the forming temperature, and decreased slowly when the temper-
ature exceeded 140 °C. When the temperature surpassed 140 °C,
the void ratio of the slab specimen could reach the design require-
ment, which indicated that rising temperature over a certain tem-
perature had slight effect on compaction characteristics of the
mixture. It was because that as temperature exceeded 140 °C, the
mixture was at stable condition, and the compaction of asphalt
pavement would not be affected by different temperatures in this
rang.
Asphalt mixture slab specimens with various segregation
degrees were formed by different wheel run times 12, 18, 24, 30,
36 and 42 times respectively. The cylindrical samples were drilled
from the slab specimen, and then the bulk volume density of each
specimen was measured by submerged weight in water method, as
shown in Fig 2.
It can be seen from Fig 2, for AC-13 and AC-20mixture, the den-
sity of asphalt mixture of different degrees of segregation showed
the same trend with the rolling compaction times. With the
increase of rolling times, asphalt mixture density first increased
rapidly, then slowly increased to the peak, and then with the con-
tinued rolling times, the density of the mixture decreased slightly.
It was because that, in a certain range, with the increase of rolling
compacted times, voids of asphalt mixture decreased gradually
which shown up as a mass increase in a certain volume of asphalt
mixture. However, when the rolling times exceeded a certain
times, the density would decrease.
In summary, the variation of compaction processes such as
compaction temperature and rolling compaction times would sig-
nificantly affect the asphalt pavement density. The asphalt pave-
ment density could be used to characterize the construction
quality of asphalt mixture in general, and it was feasible to evalu-
ate the compaction uniformity of asphalt pavement with the den-
sity distribution.
90 Z. Zhang et al. / Construction and Building Materials 145 (2017) 88–97

Table 1
Gradation composition of AC-13 segregation mixture.

Mixture type Segregation degree Pass rate (%)

26.5 19 16 13.2 9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.075
AC-13 N NA NA 100 95.1 78.7 50.2 34.8 26.5 17.0 11.6 9.3 5.2
L NA NA 100 93.9 70.1 46.3 38.9 25.6 18.4 11.8 8.7 5.4
H NA NA 100 92.7 64.4 39.9 32.0 22.0 16.1 10.9 8.2 4.9
M NA NA 100 90.2 52.2 31.1 23.4 17.1 13.2 9.8 7.6 4.2
F NA NA 100 95.3 77.2 50.4 42.8 27.9 19.7 12.3 9.0 5.7
AC-20 N 100 98.2 89.2 76 62.2 37 26.7 18.3 12.6 8.1 5.9 5
L 100 99 87.2 71.2 56.3 32.8 24 18.2 14.6 9.8 6.2 4.3
H 100 98.8 84.6 65.2 47.4 27.7 21.3 17 13.9 9.7 6 4.1
M 100 98.5 80.9 56.9 35 21.5 17.9 15.4 12.9 9.5 5.7 3.8
F 100 99.3 90.9 79.5 68.7 41.3 28.1 19.8 15.5 10 6.5 4.6

Note: NA = Not Applicable. N: no segregation L: light segregation H: high segregation

M: medium segregation F: fine aggregate segregation

2.50 10
Density-AC-13 characteristics of internal voids in the core samples A and B were
Density-AC-20 9 shown in Table 2 and Fig 3.
It can be seen from Table 2 and Fig3:
2.45 8 The calculated voids of two core samples A and B was similar,
Void ratio (%)
Density (g/cm3 )

Void ratio-AC-13 but the differences were very significant in internal voids number,
7
Void ratio-AC-20 the total volume and the equivalent void diameter between the
2.40 6 two samples. Core sample A had more total void number than
the core sample B by 12.6%. But core sample B had more average
5
single void volume and the equivalent void diameter than core
2.35 4 Sample A by137%、111% times respectively. It manifested there
was no necessary relationship between cores number and the
3
cores size. This results shown that the voids ratio of core samples
2.30 2 could not differentiate the internal voids among different mixtures.
110 120 130 140 150 160 170 In the core sample A, the minimum void volume is 0.01 mm3,
Molding temperature (°C) the maximum void volume is 374.41 mm3,the void volume less
than 1 mm3 accounts for 92.66% of the total voids, the void volume
Fig. 1. The volume parameters of the specimen under different molding
among 1–10 mm3 accounts for 6.5% of the total voids. In the core
temperature.
sample B, the minimum void volume is 0.01 mm3, the maximum
void volume is 503.25 mm3,the void volume less than 1 mm3
3.2. Void distribution accounts for 88.8% of the total voids, the void volume among 1–
10 mm3 accounts for 9.7% of the total voids. Obviously, the voids
Core samples A and B were drilled at corresponding site from in the core sample of the asphalt pavement are micro voids that
the asphalt pavement of Tong-Xun highway which had the similar the volume is smaller than1 mm3.
density detected by PQI [7,8]. Then the specimen were scanned by Void fraction is one of the factors that affect the density, so the
CT and parameters such as the three-dimensional coordinates, the results show that at the macro level, as density increases, the void
void number and its volume in core sample were calculated with fraction drops, but that they does not have the absolute connection
the defect analysis mode in VG software [14]. The composition in the micro level.

2.50 2.55
AC-13-N AC-13-L AC-20-N AC-20-L
AC-13-H AC-13-M AC-20-M AC-20-H
AC-13-F 2.50 AC-20-F
2.45
Density (g/cm3)

Density (g/cm3)

2.45
2.40
2.40

2.35
2.35

2.30 2.30
12 16 20 24 28 32 36 40 44 12 16 20 24 28 32 36 40 44
Rolling frequency(time) Rolling frequency (time)
(a) AC-13 Specimen (b) AC-20 Specimen
Fig. 2. The Relationship between the Density of Asphalt Mixture and the Rolling Times.
 Z. Zhang et al. / Construction and Building Materials 145 (2017) 88–97 91

Table 2
The composition characteristics of internal voids in the core samples.

Number Void number (ind) Total void volume (mm3) Average single void volume (mm3) Void equivalent diameter (mm) Calculated porosity (%)
A 10789 27367.36 2.203 1.615 6.95
B 9580 28828.78 3.010 1.792 6.90

50 45

40
40
35
Void Ratio (%)

30

Void Radio (%)

30
25

20
20
15

10 10

0 0

0
0. 0.1

0
0. 0.2
0. 0.3
0. 0.4
0. 0.5
0. 0.6
0. 0.7
0. 0.8
0. 0.9
1. 1.0
1. 1.5
2. 2.0
2. 2.5
3. 3.0
4. 4.0
0~ 0
10 .0
0
0. 0.1

0
0. 0.2
0. 0.3
0. 0.4
0. 0.5
0. 0.6
0. 0.7
0. 0.8
0. 0.9
1. 1.0
1. 1.5
2. 2.0
2. 2.5
3. 3.0
4. 4.0
0~ 0
10 .0

~2
>2
5. ~5.
~2
>2
5. ~5.

10
10

<
1~
2~
3~
4~
5~
6~
7~
8~
9~
0~
5~
0~
5~
0~
<
1~
2~
3~
4~
5~
6~
7~
8~
9~
0~
5~
0~
5~
0~

0
0

The rang of void (mm 3 ) The rang of void (mm 3)

(a) Core Sample A (b) Core Sample B
Fig. 3. Distribution Ratio of Internal Voids in Pavement Core Samples.

Obviously, the voids distribution of asphalt pavements varied
greatly, even for the asphalt pavement with roughly similar void
ratio calculated from core samples. Therefore, according to the
voids distribution in core samples, the difference for the com-
paction uniformity of asphalt pavement can be evaluated from
the microstructure perspective.
4. The detection method for density uniformity of asphalt
pavement
4.1. Determination for the reasonable detection distance of PQI
Referring to the literature [13], Currently the PQI detection
interval is suggested1 m  1 m. Relying on the mid-surface of
Tong-Xun Highway, density was detected at a certain interval in
the whole width of the road surface. In the first detection, the lateral
interval was 0.75 m, and the longitudinal spacing was 1 m. The
density data, mean value and the variance of each measurement
point of the test section K211 + 220–K211 + 230 were shown in
table 3.
According to the mathematical statistics principle, lateral detec-
tion interval of asphalt pavement increased, namely the cross-
sectional density data in the part of the road were selected from
sample population as the sub samples. When the mean and vari-
ance of sub sample data did not show significant variability com-
pared with the sample population, it could be considered that
increasing transverse space of road surface did not affect the detec-
tion results of the asphalt pavement. Otherwise, it was appropriate
to reduce the operation frequency of PQI. Based on Table 3, the test
samples were adjusted, and the sub samples of road surface were
got after increasing the detection interval. The average value and
variance of sub samples were calculated as shown in Table 4.
The Table 4 shows that, the indexes of the sub samples did not
change significantly compared with the total sample when lateral
detection space of asphalt pavement was increased from 0.75 m to
1.5 m.At this time, it was reasonable for lateral detection interval
to be set as1.5 m when the density of asphalt pavement was
detected by PQI. Similarly, the lateral interval was increased to
2.1 m, namely the 5 detection points were taken from each section.
The variance test was carried out and it manifested that the sub
sample data statistics showed significant variation compared with
sample population. That meant that the lateral spacing of 2.1 m
was too large. Therefore, the reasonable lateral interval of 1.5 m
was recommended when the asphalt pavement density was
detected by PQI.
Similarly, the longitudinal space of pavement could be adjusted
and determined in accordance with the way mentioned above. The
longitudinal detection space was increased to 2 m and 4 m, and no
significant variation was found in the detection results. Then, the
longitudinal detection space was increased to 10 m, 20 m and
50 m, and no significant variation was found in the detection
results. Then, the longitudinal detection identified that reducing
the detection frequency of the longitudinal density appropriately
would not lead to a large deviation for the statistical results of
the overall sample. Considering the construction length of one
vehicle of asphalt mixture was about 30–60 m and the compaction
unit length was about 50 m in the construction process, therefore,
it was suggested that the longitudinal space was 50 m when PQI
was used.
In practical engineering, PQI could be used to detect the density
distribution of asphalt pavement with the frequency lateral space
1.5 m, longitudinal space 50 m. Then according to the test results,
the compaction uniformity of asphalt pavement was evaluated,
and representative points of the asphalt pavement density were
determined.
92 Z. Zhang et al. / Construction and Building Materials 145 (2017) 88–97

Table 3
Density data of the test section (kg/m3).

Longitudinal lateral 220 221 222 223 224 225 226 227 228 229 230 Average value Standard deviation
1 2465 2465 2474 2464 2430 2430 2411 2422 2418 2421 2410 2437 25
2 2477 2478 2489 2481 2454 2419 2406 2404 2403 2409 2418 2440 36
3 2469 2461 2484 2429 2419 2417 2419 2420 2408 2411 2397 2430 28
4 2496 2478 2493 2431 2414 2408 2394 2407 2406 2404 2410 2431 39
5 2512 2503 2484 2338 2425 2414 2416 2397 2408 2420 2424 2440 40
6 2494 2505 2485 2447 2442 2419 2426 2434 2429 2437 2434 2450 30
7 2509 2461 2445 2482 2411 2410 2398 2418 2400 2425 2382 2422 38
8 2431 2448 2418 2412 2436 2424 2405 2416 2412 2438 2380 2420 19
9 2505 2482 2456 2422 2496 2433 2430 2410 2400 2413 2411 2433 35
10 2473 2485 2413 2404 2406 2402 2404 2422 2404 2406 2410 2421 29
11 2473 2470 2512 2428 2415 2430 2436 2421 2427 2426 2436 2443 29
12 2498 2504 2443 2433 2431 2438 2443 2432 2428 2445 2432 2448 27
13 2506 2483 2492 2425 2438 2392 2411 2416 2406 2408 2415 2436 39
14 2483 2477 2485 2474 2461 2439 2434 2420 2418 2416 2411 2447 29
Average value 2485 2479 2470 2434 2434 2420 2417 2417 2412 2420 2412 ND ND
Standard deviation 22 17 30 37 24 14 15 10 10 13 17 ND ND

Note: ND = No Data.

Table 4
Sub sample variance test of detection section.

Longitudinal Sample Population Sub Sample Variance Test Mean Test

Average Standard Average Standard ðn  1Þ r2 S2 Range 1.237– x  lo j
j
Less Test

Thant2a ðn  1Þ pS ffiffin
o
value deviation value deviation 14.449 result

220 2485 22 2491 21 5.5 NSV 6 21 NSV

221 2479 17 2475 15 4.7 NSV 4 16 NSV
222 2470 30 2478 22 3.2 NSV 8 28 NSV
223 2434 37 2427 45 8.9 NSV 7 25 NSV
224 2434 24 2433 29 8.8 NSV 1 17 NSV
225 2420 14 2418 15 6.9 NSV 2 13 NSV
226 2417 15 2417 13 4.5 NSV 0 14 NSV
227 2417 10 2415 9 4.9 NSV 2 9 NSV
228 2412 10 2410 10 6.0 NSV 2 10 NSV
229 2420 13 2418 7 1.7 NSV 2 12 NSV
230 2412 17 2411 18 6.7 NSV 1 16 NSV

Note: NSV = No significant variation.

4.2. The evaluation for density uniformity of asphalt pavement and overall compaction uniformity of asphalt pavement by intro-
ducing density repeatability standard deviation of in detection
The density of middle surface layer in asphalt pavement was track Sr, density repeatability standard deviation among detection
detected by PQI in the K212 + 020–K212 + 970 section in the track SR and density uniformity coefficient of detection point CU.
Tong-Xun Highway, and the PQI was calibrated before the detec- (1) Density repeatability standard deviation in detection track
tion, and the detection space for lateral and longitudinal interval Sr:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
of asphalt pavement was1.5 m and 50 m respectively. The test u p
uX
results of PQI on the test section were shown in Table 5. Sr ¼ t S2 =p ð1Þ
The statistical analysis of test data in Table 5 shows that there 1
were different variation degrees in the compaction uniformity of
asphalt pavement. At different lateral positions, the density range where, S is density standard deviation in detection track; P is the
of longitudinal detection points was above 160 kg/m3. Especially number of detection track.
at some position such as 2.0 m, 3.5 m, 6.5 m away from separator, (2) Density repeatability standard deviation among detection
the longitudinal detection range even exceeded 200 kg/m3. It man- tracks SR:
ifested that there was a variation in the longitudinal compaction rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðn  1Þ
quality at some points of asphalt pavement. In different longitudi- SR ¼ ðSxav e Þ2 þ ðSr Þ2  ð2Þ
nal positions, there was a significant difference in the lateral den- n
sity of detected points. Density of road surface of strip 0.5 m away where, Sxave is density standard deviation among detection tracks;
from separator was less than other lateral positions, because it lied Sr is density repeatability standard deviation among detection
in the working edge of the spreading machine and roller, and the tracks; n is the number of test points in detection tracks. The Sxave
compaction work was inadequate. can be calculated by formula followed:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
The above qualitatively analyzed the compaction uniformity of u p
uX
¼ t ðxi  xÞ =ðp  1Þ
asphalt pavement, but it could not quantitatively analyze the 2
Sxave ð3Þ
severity of the non-uniform phenomenon. Detection direction 1
could be divided laterally and longitudinally when using PQI to
detect the construction uniformity of asphalt pavement. Referenc- where, x. idensity mean of all detection points in detection section;
ing ASTM (American Institute of materials and testing) contrast xi . idensity average value in detection track i; p is the number of
test analysis [15,16] to quantitatively evaluate longitudinal, lateral detection sections.
 Z. Zhang et al. / Construction and Building Materials 145 (2017) 88–97 93

Table 5
Test results of PQI on the test section (kg/m3).

Stake number Distance to separator zone

0.5 (1) 2.0 (2) 3.5 (3) 5.0 (4) 6.5 (5) 8.0 (6) 9.5 (7)
K212 + 020 2421 2441 2459 2424 2432 2427 2412
K212 + 070 2406 2427 2421 2438 2417 2378 2521
K212 + 120 2410 2417 2448 2412 2403 2379 2408
K212 + 170 2392 2404 2437 2403 2379 2411 2445
K212 + 220 2386 2417 2395 2381 2369 2398 2442
K212 + 270 2367 2420 2411 2377 2448 2430 2458
K212 + 320 2369 2425 2408 2391 2401 2402 2452
K212 + 370 2359 2449 2414 2388 2446 2429 2459
K212 + 420 2414 2421 2407 2419 2461 2522 2529
K212 + 470 2383 2410 2426 2395 2533 2518 2524
K212 + 520 2370 2400 2395 2429 2446 2392 2350
K212 + 570 2336 2326 2323 2398 2394 2359 2349
K212 + 620 2447 2451 2530 2500 2493 2475 2506
K212 + 670 2452 2525 2438 2397 2434 2538 2423
K212 + 720 2427 2436 2399 2401 2394 2424 2414
K212 + 770 2340 2395 2372 2343 2334 2395 2365
K212 + 820 2428 2440 2361 2522 2401 2379 2468
K212 + 870 2505 2523 2482 2500 2540 2463 2470
K212 + 920 2427 2429 2466 2412 2460 2448 2426
K212 + 970 2459 2531 2463 2429 2438 2406 2420
Minimum value 2336 2326 2323 2343 2334 2359 2349
Maximum value 2505 2531 2530 2522 2540 2538 2529
Range 169 205 207 179 206 179 180
Standard deviation 42.81 47.64 45.90 44.25 51.37 51.13 53.18
Average value 2404.9 2434.4 2422.8 2418.0 2431.2 2428.7 2442.1
Variation coefficient/% 1.780 1.957 1.887 1.830 2.113 2.105 2.178

(3) Density uniformity coefficient of detection points CU:
0 Xm 1 standard deviation among detection tracks.
jh  hj
i¼1 i
CU ¼ @1  X m
A  100%
h i
i¼1

density average value of
where, hi . the density of detected point i; h.
all detection points in detection section; m is total number of points
in the detection tracks.
Meanwhile, in order to quantitatively evaluate longitudinal, lat-
eral and overall compaction uniformity of asphalt pavement, refer-
encing consistency statistics of in laboratory data and consistency
statistics among laboratory data, density consistency statistics in
detection track k and density consistency statistics among detec-
tion track h were introduced .
(1) Density consistency statistics in detection track k:
k ¼ S=Sr
where, S is density standard deviation in detection track; Sr is den-
sity repeatability standard deviation in detection track.
The acceptable level for the longitudinal compaction uniformity
of detection track was determined through giving the critical value
of density consistency test kcrit.
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
p detection tracks could meet the requirements basically except
kcrit ¼
1 þ ðp1Þ
F detection track 1 which manifested that the lateral compaction
where, p is the number of detection tracks; F is F Distribution and
its value was obtained from F Distribution Table according to the
first freedom degrees v1and the second freedom degrees v2 which
were calculated by v1 = m  1,v2 = (p  1)(m  1) and m is the
number of detection points in each detection track. General signif-
icance level was 5%.
(2) Density consistency statistics among detection track h:
h ¼ d=Sxave
where, d is deviation among detection tracks; Sxave is the density
The acceptable level for the lateral compaction uniformity of
ð4Þ
detection tracks was determined through giving the critical value
of density consistency test hcrit.
ðp  1Þt
hcrit ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð8Þ
pðt 2 þ p  2Þ
where, p is the number of detection tracks; t is t distribution and its
value was obtained from t Distribution Table according to the free-
dom degrees v which were calculated by v = p  2.General signifi-
cance level was 5%.
Density consistency statistics in detection track k and density
consistency statistics among detection tracks h were calculated
ð5Þ by mathematical statistics method according to the test data in
Table 5. The results were shown in Table 6.
Table 6 showed that consistency statistics k in different detec-
tion tracks of test section were less than critical value kcrit therefore,
there was no great variations in density among different detection
tracks which manifested that the test section do not appear obvious
longitudinal segregation and the longitudinal compaction of
asphalt pavement was fine. Consistency statistics h among different
ð6Þ
quality is non-uniform in the detection track 1.This was consistent
with the previous conclusion analyzed through the detection data.
Consistency statistics k and h certificated that lateral and longitudi-
nal compaction in test section was fine in general.
Above all, the uniformity condition longitudinal and lateral den-
sity distribution in different test tracks could be directly compared
by density repeatability standard deviation in detection track Sr
and density repeatability standard deviation among detection
track SR. And it could be judged whether the density distribution
ð7Þ uniformity of longitudinal and lateral in test section uniformity
94 Z. Zhang et al. / Construction and Building Materials 145 (2017) 88–97
Table 6
The consistency statistic of density data in test section.
Detection track Test point S d
1 20 42.8 -22.5
2 20 47.6 6.95
3 20 45.9 5.35
4 20 44.2 -9.45
5 20 51.4 3.75
6 20 51.1 1.25
7 20 53.2 14.65
could meet the demand with density consistency statistics in
detection track k and density consistency statistics among detec-
tion track h.
When asphalt pavement density was detected by PQI, the
above indexes could be used to evaluate whether the longitudi-
nal and lateral compaction of the road section meet the require-
ments and judge the compaction uniformity in different test
sections.
5. Uniformity detection methods of asphalt pavement void
distribution
5.1. Determination of voids detection point in asphalt pavement
At home and abroad, the asphalt pavement was generally
required that the asphalt pavement void ratio was in the range
of [17] 3%–8%, that meant that the density of asphalt pavement
was among the 92%–97% of the maximum theoretical density of
asphalt mixture. If the pavement density was less 92% than the
maximum theoretical density of asphalt mixture, which meant
the in-situ air void ratio of the pavement was too large. While, if
it was more than 97%, the in-situ air void ratio of the pavement
was small. It was suitable for in-situ air void to be in rang 3%–
8%. Early damage in asphalt pavement occurs whenever the voids
were too large or small. Therefore, according to the maximum the-
oretical density of the mixture, the asphalt pavement density could
be divided into three representative regions: high, medium and
low density which represented three different sections with large,
suitable and small in-situ air void respectively.
Relevant data from laboratory at construction site were got and
listed in the Table 5,the maximum theoretical density of asphalt
mixture in test section was 2592 kg/m3.According to the reason-
able range of compaction degree 92%–97%, the asphalt pavement
density in different points were divided into high density region:
>2514 kg/m3, medium density region: 2385 kg/m3–2514 kg/
m3and low density region: <2385 kg/m3.
PQI test data in statistical Table 5 showed that there are 13
points in the high density region which account for 9.3% of
detected point number; there were 103 points in the middle
density region which accounted for 73.6% of detected point
number; 24 points in the low density region which accounted
for 17.1% of detected point number.10 points were chose as rep-
resentative point of density distribution in test section from
1 km long test section. According to the ratio of point position
in the high, medium and low density region, it was necessary
to select 1 point in high density region, 7 in middle density
region and 2 in low density region to ensure that density distri-
bution of whole section can be represented by chosen points.
Point position in each density region was chosen randomly and
the position of each point correspond to stake number in Table 5
which was underlined and in bold. Void distribution uniformity
in test section was characterized by void distribution in these
representative cores.
Sxave Sr SR k kcrit h hcrit
12.3 48.2 48.6 0.888 1.237 -1.829 ± 1.648
0.988 0.565
0.952 0.435
0.917 -0.768
1.066 0.305
1.060 0.102
1.104 1.191
5.2. Evaluation of asphalt pavement void distribution uniformity
The composition and distribution of voids in core samples were
further discussed by drilling core samples from representative
point and carrying out industrial CT tomography scanning. With
the core sample A and B which had basically equal void ratio as
the example, voids distribution uniformity in asphalt pavement
structure was analyzed by industrial CT tomography scanning
and image processing technology. IPP software was commonly
used in medical and biological fields, and had powerful function
in the image size measurement and counting, classification statis-
tics and analysis [18].The IPP software was used to process the CT
fault image and analyze the voids in two-dimensional section,
therefore, parameters such as voids number and section void ratio
in fault image of core sample were obtained.
In order to eliminate the influence from incompleteness at the
two ends of the slice, CT scanning tomography within 5 cm–
55 cm range were selected as study object in this paper. In differ-
ent sections of core sample A and B, changes of voids number
and voids ratio along the core sample height were shown in Figs. 4
and 5.
Fig. 4(a) showed that voids number in core samples was not
uniformly distributed along core sample height, but it showed a
trend that it was large at both ends and small in the middle part.
Fig. 4(b) showed that section void ratio in core sample was large
at both ends and small in the middle part. Section void ratio at
the bottom of the core sample was smaller than top. In the height
range 20 cm–40 cm, the void ratio of core sample was similar and
the minimum section void ratio was also located in this height
range. The variety of core sample voids distribution mainly pre-
sented in the upper and lower ends.
Generally, it was ideal that the voids number should maintain
the same section from top section to bottom section. In this paper,
a parameter, uniform coefficient of section void number UCN was
put forward to evaluate the voids distribution number in pave-
ment. The parameter was defined as variation coefficient of voids
number and calculated using variation of 11 sections at an interval
5 cm within the 5–55 m height range. Calculating formula was
followed.
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi X
sX
n 2, n
ðVN i  VNÞ VN i
UC N ¼ i¼1 i¼1
ð9Þ
n1 n
Where, V N i . voids number in fault image i; VN. average value of
voids number in fault image of core samples; n is fault images
number in core sample, which is 11 in medium surface layer of
core sample.
Uniform coefficient of section void number (U CN .) represents
variation degree of void number in different sections along the core
sample height, and it can be used to evaluate compaction unifor-
mity of asphalt pavement. The lager the U CN . i the worse the uni-
formity of voids number distribution is. The U CN . core sample A
 Z. Zhang et al. / Construction and Building Materials 145 (2017) 88–97 95

Void Number void number

60 80 100 120 140 160 180 200
60 70 80 90 100 110 120 130 140 0
0

(A) (B)
10

Core Sample Height (mm)

10
Core Sample Height (mm)

20
20

30 30

40 40

50 50

60 60

(a) Changes of voids number along core sample height

Section Void Ratio (%) Section void void/%

0 3 6 9 12 15 0 3 6 9 12 15
0 0

(A) Core Sample Heihgt (mm) (b)

10 10
Core Sample Height (mm)

20 20

30 30

40 40

50 50

60 60

(b) Changes of voids ratio along core sample height

Fig. 4. Changes of voids number and voids ratio along core sample height.

and core sample B is 0.182, 0.247 respectively that suggested uni-
formity of voids distribution in core sample A is better than B.
Section void ratio of core samples are large at the two ends, and
smaller in the middle parts than two ends, so void ratio distribu-
tion in different sections of core samples can be represented by
section void ratio of two ends and middle section. This paper pro-
poses uniform coefficient of section void ratio (UCR), which is
defined as radio of section void ratio differentia between two ends
of core sample (5 cm and 55 cm positions) to section void ratio of
sample medium (30 cm position). Calculating formula is followed:
VRtop  VRbottom
UC R ¼
VRmedium
where, VRtop is void ratio of the top section of core sample (Section at
the top 5 cm); VRbottom is void ratio of the bottom section of core
sample (Section at the bottom 5 cm); VRmedium is void ratio of sam-
ple medium.
The U C R . n reflect uniformity of the void ratio in the void distri-
bution from different section of core samples. The larger the U C R .,
the worse the uniformity of voids ratio distribution in core samples
is. The U CR . core sample A and core sample B is 0.523, 0.725
respectively so uniformity of voids distribution in core sample A
is better than B in accordance to the UCR.
In summary, uniformity of void distribution inside asphalt
pavement could be reflected by UCN and UCR from void number
and void ratio of tomography image. The differentia of compaction
uniformity of asphalt pavement could be evaluated quantitatively
by using UCN and UCR of the representative cores samples from
detection section.
6. Accurate detection and evaluation for compaction uniformity
of asphalt pavement
In construction site of Tong-xun highway, two sections of mid-
dle surface layer 1000 m long were chosen as detection sections
which are called section A and B here in after. The density of differ-
ð10Þ ent regions was detected by PQI and representative cores samples
drilled from two sections were scanned by industry CT and used to
evaluate compaction uniformity of detection section from density
distribution uniformity and void distribution uniformity
respectively.
In the detection sections A and B, the density of different
regions was detected by PQI with lateral interval 1.5 m, longitudi-
nal interval 50 m. Density data of A and B were statistically
analyzed and the results were shown in Table 7.
It can be seen from table 7:
Density data statistics Sr and SR in 7 detected tracks have signif-
icant difference. Density repeatability standard deviation of in
detection track(Sr .)f section A is smaller than B, which indicated
that longitudinal compaction uniformity of A is better than B;
96 Z. Zhang et al. / Construction and Building Materials 145 (2017) 88–97

Table 7
Density data statistics of detection section A and B.

Detected section Detected track Test point S d Sxave Sr SR k kcrit h hcrit CU

A 1 20 32.6 81.0 44.3 32.0 54.2 1.020 1.237 1.829 ±1.648 90.4%
2 20 33.6 0.1 1.051 0.002
3 20 40.2 34.0 1.257 0.767
4 20 25.7 47.1 1.007 1.062
5 20 20.2 35.1 0.630 0.791
6 20 40.2 16.0 1.257 0.362
7 20 25.7 18.9 1.007 0.427
B 1 20 42.7 13.8 18.0 38.5 41.6 1.110 1.237 0.769 ±1.648 95.6%
2 20 57.2 15.9 1.487 0.881
3 20 31.9 9.5 0.829 0.527
4 20 46.7 11.7 1.212 0.650
5 20 17.6 7.9 0.456 0.437
6 20 22.8 17.8 0.593 0.987
7 20 35.0 29.8 0.908 1.658

however, density repeatability standard deviation among detec- Table 8

tion track ðSR Þ. section A is larger than B, which suggested that lat-
eral compaction uniformity of A is worse than B. In addition,
density uniformity coefficient of detection point (CU) of section A
is smaller than B, which manifested that density variation of differ-
ent detection points in section A is larger than B, so overall com-
paction uniformity of B is better than A.
In section A, consistency statistics k of track 3 and 6 is larger
than critical value kcrit, which manifested that longitudinal com-
paction uniformity is poor and cannot meet the requirement. K
of other tracks is smaller than critical value kcrit, which indicated
that longitudinal compaction uniformity in detection track is bet-
ter. In section B, k of track 1, 3, 5, 6 and 7 is smaller than critical
value kcrit, which indicated that there is not obvious longitudinal
segregation phenomenon in asphalt pavement of detection track;
but k of track 2 surpasses critical value kcrit which suggested that
longitudinal construction segregation is more serious. However, k
of track 4 approaches critical value kcrit which manifested that
although longitudinal construction quality can meet requirements,
the compaction uniformity is unsatisfactory which bring hidden
dangers to performance of asphalt pavement.
In section A, consistency statistics h of track 1 is larger than crit-
ical value hcrit which indicated that lateral construction segregation
is more serious and lateral compaction uniformity cannot meet
requirements. H of other tracks is smaller than critical value hcrit
which manifested that there is not obvious lateral segregation phe-
nomenon and lateral compaction uniformity of pavement can meet
requirements. In section B, except h of track 7 is larger than critical
value hcrit, h of other tracks is smaller than critical value hcrit which
manifested that lateral segregation is more serious and lateral com-
paction uniformity is worse in track 7, however, there is no obvious
lateral segregation, so lateral compaction uniformity is good.
Referring to the relevant data from laboratory at construction
site, maximum theoretical density of asphalt mixture of section A
and B were determined. According to PQI density data of different
regions from section A and B, the ratio of high, medium and low
density region were shown in Table 8.
In the detection sections of 1 km long 10 points were chosen as
representative point of asphalt pavement density distribution.
According to the proportion of detection points of high density
region, medium density region and low density region in section
A and B, points are chosen as that 1 point in high density region,
7 points in medium density region, 2 points in low density region
which ensures that density distribution of whole section can be
represented by the point of access.core samples were drilled from
representative point of section A and B, and the cores were scanned
by industry CT. The void distribution of core samples was calcu-
lated and used to evaluate the void distribution uniformity of
The ratio of different density region in detection section%.
Detection section Region type
high medium low
A 8.3 69.8 21.9
B 7.9 73.4 18.7
asphalt pavement in detection section. Void distribution unifor-
mity index in representative core samples of section A and B was
shown in Table 9.
Table 9 shows that the differentia is large between maximum
section void ratio and minimum section void ratio of representa-
tive core samples in section A and B. Generally, the position of
the maximum section void ratio is at the end of core samples while
minimum section void ratio is in the middle of core samples. There
was no obvious boundary between U CN . d U CR the high and low
density region, but both UCN and UCR vary in value among differ-
ent core samples in 3 region. There is no regular correlation
between the two indexes and the calculated void ratio of core sam-
ples, namely that the distribution uniformity of section void num-
ber and void ratio do not become better or worse along the core
sample height as the calculated void ratio become larger. There
is no direct connection between void distribution uniformity and
in-situ void ratio of asphalt pavement, which indicates that it is
not comprehensive to evaluate the compaction uniformity of
asphalt pavement only by the void ratio index. So, it is recom-
mended to evaluate compaction uniformity of asphalt pavement
from the microscopic structure by using the uniform coefficient
of section void number and void ratio according to the differences
of void distribution in pavement.
According to the data in Table 9, the uniform coefficient of sec-
tion void number (UCN) and void ratio (UCR) of representative core
samples from section A and B are calculated and shown in Table 10.
Table 10 shows that there is certain variability between uniform
coefficient of section voids number (UCN) and void ratio (UCR)
among different core samples in section A and B. Comparing with
section A, the UCN, the average value and variation coefficient of
the UCR) are smaller in section B which manifests that void distri-
bution uniformity in section B is better than A. Therefore, the com-
paction uniformity of section B is better according to void
distribution in pavement structure.
Section A and B were detected and evaluated from the density
distribution of asphalt pavement and the voids distribution in
pavement structure respectively above. Combined with the density
distribution of section A and B and the evaluation results of the
voids distribution, it is known that the compaction uniformity of
the section B is better than section A.
 Z. Zhang et al. / Construction and Building Materials 145 (2017) 88–97 97

Table 9
Void distribution uniformity index in representative core samples of section A and B.

Index Region type

High Medium Low
Calculated porosity /% A 1.86 5.64 6.83 4.92 6.71 4.59 5.26 6.33 10.24 9.35
B 2.63 6.46 5.97 6.21 6.32 5.33 5.71 4.45 8.73 9.15
Maximum section void ratio/% A 8.57 12.52 13.29 13.56 12.25 13.21 12.13 13.72 13.95 14.67
B 6.36 11.92 14.15 13.83 12.61 12.5 12.94 12.79 14.17 13.44
Minimum section void ratio /% A 1.61 3.55 2.88 2.44 3.67 3.01 3.12 3.35 3.93 3.55
B 1.67 3.1 3.34 3.13 3.15 3.62 3.53 2.41 3.48 3.25
UCN A 0.147 0.237 0.153 0.216 0.232 0.263 0.098 0.305 0.193 0.278
B 0.165 0.126 0.147 0.175 0.124 0.106 0.187 0.229 0.166 0.094
UCR A 1.950 0.927 2.698 2.369 0.853 1.256 1.571 0.964 1.623 1.794
B 1.186 0.894 0.907 1.080 0.968 2.171 1.603 1.606 1.707 0.975

Table 10
Statistic of voids distribution uniformity in detection section.

Index Section Average value Standard deviation Variation coefficient coefficient

UCN A 0.212 0.065 0.305
B 0.152 0.041 0.269
UCR A 1.600 0.623 0.389
B 1.310 0.435 0.332

7. Conclusion References

1. The changes of compaction parameter such as compaction tem-
perature and run times can significantly affect the density of
asphalt pavement, so it is feasible to evaluate the compaction
uniformity of asphalt pavement on the whole by using the den-
sity distribution condition. The differentia of compaction uni-
formity of asphalt pavement can be reflected by voids
distribution in core samples from microscopic structure.
2. In practical engineering, when the PQI is used to detect the den-
sity distribution in different regions of asphalt pavement, the
reasonable lateral interval is 1.5 m and longitudinal interval is
50 m.
3. The quality of density distribution uniformity in lateral and lon-
gitudinal of different sections can be compared directly by den-
sity repeatability standard deviation of detection tracks Sr, SR.
Density consistency statistic in detection sections k、h can
reflect whether the density distribution uniformity of detection
sections in lateral and longitudinal can meet requirements.
4. According to the reasonable in-situ air void ratio of asphalt
pavement, the detection points are divided into three density
region, high density, medium and low density. Combined with
the required number of core samples, and core samples are
drilled randomly from each density region in accordance to its
proportion as representative cores of detection sections.
5. In view of the distribution characteristics of the voids distribu-
tion in core sample of the pavement, uniform coefficient of sec-
tion voids number (UCN) and void ratio (UCR) index are put
forward to evaluate quantitatively compaction uniformity of
asphalt pavement from the point of the microscopic structure.
The larger the UCN and UCR are, the worse the voids distribution
uniformity in core samples is.
Acknowledgments
This work was supported by Research Project by Department of
Transport of Shaanxi Province (2014-01 K), the National Natural
Science Foundation of China (51008031) and the Key Laboratory
of the Ministry of education. The authors gratefully acknowledged
their financial support.
[1] T.S. Khedaywi, T.D. White, Effect of segregation on fatigue performance of
asphalt paving mixtures, Transp. Res. Rec. 1543 (1996) 63–70.
[2] T.S. Khedaywi, T.D. White, Development and analysis of laboratory techniques
for simulating segregation, Transp. Res. Rec. 1492 (1995) 36–45.
[3] H. Sivilevičius, K. Vislavičius, Stochastic simulation of the influence of variation
of mineral material grading and dose weight on the homogeneity of hot-mix
asphalt, Constr. Build. Mater. 22 (9) (2008) 2007–2014.
[4] Liu Honghai, Yin Ran, Wu Shaopeng, Reducing the compaction segregation of
hot mix asphalt, J. Wuhan Univ. Technol. Mater. Sci. Ed. 22 (1) (2007) 132–135.
[5] Yang. Bo, Zheng. Jianlong, Zhang. Rui, Research and discussion on the
quantitative evaluation of asphalt mixture aggregate segregation division
standard, J. Build. Mater. 18 (6) (2015) 1099–1105.
[6] Zhengqi Zhang, Shuangquan Jiang, Quantitative detection and evaluation of
non-uniformity of asphalt concrete pavement by using non-nuclear density
meter, Road 4 (2010) 158–163.
[7] G. Chang et al., Accelerated Implementation of Intelligent Compaction
Technology for Embankment Subgrade Soils, Aggregate Base, And Asphalt
Pavement Materials. s.l.: FHWA, 2011. Publication FHWA-IF-12-002.
[8] G. Chang, Q. Xu, J. Rutledge, S. Garber, A Study on Intelligent Compaction and
In-Place Asphalt Density. s.l.: No. FHWA-HIF-14-017, 2014.
[9] Xiuning Bao, Xiaoning Zhang, Duanyi Wang, Characterization of segregation of
asphalt pavement structure by texture depth, Highway Traffic Sci. Technol. 23
(09) (2006) 20–22.
[10] Duanyi Wang, Weijie Li, Xiaoning Zhang, Evaluating and measuring the
texture depth of asphalt pavement structure with digital image technology, J.
South China Univ. Technol. (Natural Science Edition) 32 (02) (2004) 42–45.
[11] Song. Yongzhao, Yan. Gongxi, Sui. Yongqin, Texture structure distribution of
asphalt pavement surface based on digital image processing technology, J.
Central South Univ. (Natural Science Edition) 45 (11) (2014) 4075–4080.
[12] Yao. Ke, Zhang. Xiaoning, Li. Zhi, Experimental study on evaluation method of
asphalt pavement evenness, J. Shenyang Constr. Univer. (Natural Science
Edition) 23 (06) (2007) 930–935.
[13] Xiaoning Fu, Study on evaluation of asphalt pavement paving segregation
method, Highway Traffic Sci. Technol. 25 (04) (2008) 37–41.
[14] Zhang. Ke, Zhang. Zhengqi, Effect of high moisture content on properties and
internal shape of asphalt mixture, J. Wuhan Univ. Technol. 36 (09) (2014) 48–
53.
[15] ASTM E177–14. Standard practice for use of the terms precision and bias in
ASTM test method [S]. Philadelphia: American Society for Testing and
Material, 1992.
[16] ASTM E691–15. Conducting an inter laboratory study to determine the
precision of a test method [S]. Philadelphia: American Society for Testing and
Material, 1992.
[17] JTG F40–2004. Technical Specification for Construction of Highway Asphalt
Pavement [S]. Beijing: People’s Communication Press, 2004.
[18] Zhang. Ke, Zhang. Zhengqi, Quantitative evaluation of adhesion between
asphalt and aggregate, J. Hefei Univ. Technol. (Natural Science Edition) 38 (6)
(2015) 810–815.