Construction and Building Materials xxx (2017) xxx–xxx
Contents lists available at ScienceDirect
Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat
Laboratory evaluation of electromagnetic density gauges for hot-mix asphalt mixture density measurement Zhen Leng a,⇑, Zeyu Zhang a, Yuan Zhang a, Yangyang Wang a, Huayang Yu a, Tianqing Ling b a b
Department of Civil and Environmental Engineering, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong Faculty of Architecture and Urban Planning, Chongqing Jiaotong University, Nan’an, Chongqing, China
h i g h l i g h t s Performance of common asphalt pavement EM density gauges was investigated. Effects of measurement direction, asphalt content, and air void content are insignificant. Effects of moisture, mixture gradation and gauge calibration method are significant. PaveTracker with mix calibration provides best accuracy.
a r t i c l e
i n f o
Article history: Received 15 June 2017 Received in revised form 24 September 2017 Accepted 27 September 2017 Available online xxxx Keywords: Asphalt mixture Density Electromagnetic density gauge Evaluation
a b s t r a c t This paper presents a laboratory study aiming to evaluate the performance of electromagnetic (EM) density gauges as a non-destructive tool for hot mix asphalt (HMA) density measurement. In total, 36 HMA testing slabs with different compositions were prepared in laboratory. EM density gauge data were collected from these slabs using two common types of gauges, i.e., PQI 301 and PaveTracker 2701B, and compared with their bulk densities measured by the standard saturated surface dry (SSD) method. It was found that measurement direction, asphalt binder content, mixture air void content, and thickness of the testing slabs do not affect the accuracy of the EM density gauges. However, EM density gauge measurements are affected by the presence of moisture, gradation of asphalt mixture and gauge calibration method. It was also concluded that the accuracy of PaveTracker with mix calibration is comparable to that of the conventional standard method. Ó 2017 Published by Elsevier Ltd.
1. Introduction In-situ density is one of the key factors affecting the durability of asphalt pavement, because density that is either too high or too low may lead to early pavement failures [1,2]. For instance, insufficient density could increase the risk of oxidation [3,4], water damage [5–7], cracking, and ravelling [8,9], while rutting, shoving, and bleeding may occur if the asphalt mixture is over-compacted [10]. Conventionally, two methods have been commonly used to estimate the in-situ density of asphalt mixture: coring and using a nuclear density gauge. The coring method has been widely used worldwide because it provides accurate measurement [11], while the nuclear density gauge method has been commonly used in the US, as it is a non-destructive method and able to provide ⇑ Corresponding author. E-mail addresses:
[email protected] (Z. Leng),
[email protected] (Z. Zhang),
[email protected] (Y. Zhang),
[email protected] (Y. Wang),
[email protected] (H. Yu),
[email protected] (T. Ling).
reasonably accurate estimation of the in-situ asphalt mixture density [10]. However, both these methods have some limitations. While coring provides accurate measurement, it damages pavement. The coring locations often become weak spots where early distresses occur, and these localized distresses may quickly extend to a large area, causing extra cost in maintenance and rehabilitation [1]. In addition, the coring and filling process is timeconsuming. It may delay the traffic opening time and cause long traffic interruption. Although nuclear density gauge is a nondestructive tool, it carries potential safety risks due to the use of radiative materials. As a result, special care is needed when storing and transporting the gauges [12], and their operation must be conducted by licensed operators [10]. EM density gauges, also known as non-nuclear density gauges, have gained increasing interest recently as an alternative to the two conventional methods. Since their operation is based on sending and receiving electromagnetic waves [13–16], they have the advantages of completely bypassing the pavement damage and safety concerns. To investigate the performance of this new type
https://doi.org/10.1016/j.conbuildmat.2017.09.186 0950-0618/Ó 2017 Published by Elsevier Ltd.
Please cite this article in press as: Z. Leng et al., Laboratory evaluation of electromagnetic density gauges for hot-mix asphalt mixture density measurement, Constr. Build. Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.186
2
Z. Leng et al. / Construction and Building Materials xxx (2017) xxx–xxx
of density gauges, various studies have been conducted. A field study conducted by Romero concluded that the calibrated EM density gauges are accurate for the quality control applications [17]. Similarly, it was recommended by Allen et al. that non-nuclear density gauges can be used as quality control tools for hot mixture asphalt pavement construction [18]. Zhuang et al. conducted a series of field and laboratory tests and concluded that the EM density gauge they used provided acceptable density measurement accuracy [19]. However, some other studies raised the concerns on the potential effects of various factors, such as mat temperature, moisture condition and presence of paint on the accuracy of EM density gauge measurement [11,20,21], and so far there has been no study reporting that EM density gauges are reliable enough to be used as a quality assurance tool [22]. Williams et al. characterized the effects of several factors on the accuracy of the EM density gauges, and found that temperature does not affect the measurement accuracy of EM density gauge [11]. In addition, they recommended to increase the reference sample numbers to improve the EM density gauge density measurement accuracy. The earlier EM density gauge models, such as PQI 300, were even not recommended as a quality control tool by some researchers [23,24]. Because of the inconclusive findings on the accuracy of EM density gauges, the current application of these gauges in practice is still limited despite their advantages compared with the conventional methods, and a better understanding on the effects of various parameters on the density measurement accuracy using EM density gauge is desired. Correspondingly, this study aims to characterize the effects of various factors, including density gauge operation (measurement direction and gauge calibration method), asphalt mixture composition (asphalt binder content, mixture air void content, and gradation of asphalt mixture), asphalt layer thickness, and testing environment (presence of moisture and marking paint) on the measurement accuracy of EM density gauges through comprehensive laboratory testing and statistical analysis. 2. Background on EM density gauges 2.1. Operation principle EM density gauges refer to those gauges which use EM waves to measure the in-situ density of asphalt pavement. PQI and PaveTracker are two common types of EM density gauges, which are commercially available by Trans-Tech System Inc. and Troxler Electronics Lab, respectively. The application of both PQI and PaveTracker is based on measuring the dielectric constant of the asphalt mixture, which is a function of its bulk density. As Fig. 1 illustrates, the measurement
Fig. 1. Operation principle schematic of EM density gauges [11].
is conducted using the toroidal electrical sensing field established by a sensing plate [11,25]. If a dielectric (i.e., non-conductor), for instance, bituminous mixture, is introduced into the electrical field, the field will be affected. Then, the dielectric constant of the paving materials will be estimated by measuring the change of the electrical field and converted into material density.
2.2. Operation procedures The operation procedures of different EM density gauges are in general similar, which include three major steps: calibration, parameter input and data collection. Calibration is the first step of density measurement using an EM density gauge. It helps improve the accuracy of the EM density gauges by using the laboratory or nuclear measured density as reference. Then, appropriate parameters (such as measurement mode and nominal maximum aggregate size of the asphalt mixture) will be selected or input. Finally, the density data of the asphalt mat will be estimated based on the electrical field response and displayed on the screen [8,14,25]. Normally, calibration programs come with the EM density gauges. The bulk density of the asphalt mixture measured by nuclear density gauge or coring method is needed as input for the calibration program. Then the calibration can be conducted using different methods, which can be grouped into two categories: slope function method (i.e., mix calibration method) and simple plus-minus method (i.e., offset method). The slope function method builds a function between the raw data collected by the gauge and the entered bulk density. Then, the calibrated data will be calculated and generated automatically, utilizing the built slope function (Fig. 2(a)). The offset method, in comparison with the mix calibration method, is easier and simpler. A constant offset value, which is determined by subtracting the raw data from input bulk density, will be added into the raw value (Fig. 2(b)). Nevertheless, there are also some differences between the two density gauges. Compared with PQI, PaveTracker recommends an extra procedure, namely reference, for the purpose of selfadjusting, which is conducted by placing the device on a standard plate fixed inside the gauge carrying container. Each gauge pairs to one standard plate with known density. After placing gauge on the corresponding standard plate, the density read by the gauge is compared to the known density of the plate. The results of this comparison are used to automatically adjust gauge measurement.
3. Experimental program and research methods To explore the effects of variable factors on the density gauge measurement accuracy, the saturated surface dry (SSD) bulk density measured in accordance with the ASTM standard D2726/ D2726M-14[26] and the gauge measured raw data of the laboratory prepared testing slabs were collected under various conditions. Then, those data were statistically analysed using the statistical analysis software IBM SPSS. In this study, six factors, including measurement direction, calibration method, gradation of asphalt mixture, asphalt binder content, moisture condition, and the thickness of the asphalt mat, were considered. Table 1 shows the factors and the levels corresponding to each factor. It is worth noting that, the effect of aggregate type was not taken into consideration in this paper, since researchers have reported that the aggregate type does not have statistically significant effect on the density measurement of EM density gauges [21]. In addition, only granite-type aggregate is used in Hong Kong for pavement construction.
Please cite this article in press as: Z. Leng et al., Laboratory evaluation of electromagnetic density gauges for hot-mix asphalt mixture density measurement, Constr. Build. Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.186
3
Z. Leng et al. / Construction and Building Materials xxx (2017) xxx–xxx
Real Value F(x)
F(x1)
x1
F(x2)
x2
F(x3)
x3
F(xn)
xn
Raw Data x
F(x1)
x1
F(x2)
x2
F(x3)
x3
F(xn)
xn
F(xn)
xn
Function Determination Raw Data xn+1
F(x)=A (x)+B Raw Data xn+1
Offset value
Calibrated Data
Calibrated Data
(a)
(b)
Fig. 2. Calibration principle schematic: (a) mix calibration method, (b) offset method.
‘‘Continuous Reading Mode”, typically used for quality control measurements, were selected when collecting data using PaveTracker and PQI, respectively.
Table 1 Factors and levels table. Factors
Levels
EM Density Gauge
Measurement Direction
1 2
Calibration Method
Composition and Thickness of Testing Slab
Gradation
Asphalt Binder Type Asphalt Binder Content
Testing Environment
Moisture Condition
1
Perpendicular to the compaction direction Parallel to the compaction direction Reference Only^
2 3
Reference and Offset^^ Reference and Mix Calibration^^^
1 2 3 1
Dense-graded Gap-graded Open-graded 60/70
2 1
PG 76 4.2%–5.5%
2 3
6.0%–6.2% 6.2%–8.2%
1
2.1 (H2O index)
2 3
4.1 (H2O index) 8.1 (H2O index)
^, ^^ For PQI, ^ and ^^ means without calibration and offset, respectively. ^^^Mix calibration method is not available on PQI.
3.1. Laboratory sample preparation In this study, two types of asphalt binder commonly used in Hong Kong, i.e., asphalt binder with a penetration grade of 60/70 (Pen 60/70) and asphalt binder with a Superpave performance grade of 76–16, were used to prepare the testing slabs. Granitetype of aggregate and mineral fillers were used. The optimum binder contents of different asphalt mixtures were determined according to the Marshall asphalt mixture design method and were named normal binder contents. In total, 36 asphalt mixture testing slabs with the dimension of 30 cm in width by 30 cm in length were prepared in laboratory. The details of the prepared testing slabs and their corresponding gradations are shown in Table 2 and Table 3, respectively.
3.2. EM density gauges Two EM density gauges, i.e., PaveTracker 2701B (Fig. 3(a)) and PQI 301 (Fig. 3 (b)), were applied in this study. ‘‘Continuous” and
3.3. Research methods The main principle of experimental procedure is manufacturing asphalt mixture slabs of known density and comparing the known density to the gauge-measured density. EM density gauge data of the slabs were collected by two common types of gauges. Two testing condition factors were considered, moisture condition and measurement direction. As discussed in Section 2.2, before measure density using EM density gauges, calibration procedure should be conducted. While, in this study, according to the manufactures’ suggestions, a modified measurement and analysis procedure was performed. Raw data were collected by uncalibrated EM density gauges followed by calibrating them in laboratory. Raw data calibrated by offset and mix calibration are labelled as offset and mix calibration, respectively. Measurements for the bulk density of prepared testing slabs were conducted in accordance with the ASTM standard D2726/ D2726M-14 [26]. A customized apparatus was designed and prepared for suspending the slabs in water. 4. Results and analysis 4.1. Measurement direction In field data collection, density gauge might be placed in a random direction selected by the operator. Thus, it is necessary and important to know the effect of measurement direction. To quantify such effect, the density data of the 6 testing slabs with normal binder content were collected by using PQI and PaveTracker at two perpendicular directions (Fig. 4), i.e., parallel to the compaction direction and perpendicular to the compaction direction. Fig. 5(a) and (b) show the raw data collected at two directions by using PQI and PaveTracker, respectively. It can be seen that for both gauges, the measured densities at two directions are very close to each other. Student-t tests were further conducted to check the statistical difference in the measurement between the two directions, and the results show that at 95% confidence level, the differences caused by variation of measurement direction are not statistically significant. Thus, it can be concluded that measurement direction does not significantly affect the measured results of both gauges. This finding is expectable, because both density
Please cite this article in press as: Z. Leng et al., Laboratory evaluation of electromagnetic density gauges for hot-mix asphalt mixture density measurement, Constr. Build. Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.186
4
Z. Leng et al. / Construction and Building Materials xxx (2017) xxx–xxx
Table 2 Details of asphalt mixture testing slabs. Asphalt Binder Type
Gradation Type
Aggregate Type
Thickness(cm)
Binder Content
Amount
Pen 60/70 PG 76-16
Dense-graded Wearing Course Gap-graded SMA* Gap-graded SMA Gap-graded SMA Gap-graded SMA Open-graded Friction Course
Granite
5 5 10 5 5 5
Normal^ High^^ Normal Normal Low^^^ Normal
6 6 6 6 6 6
^, ^^, ^^^ Normal binder content is determined according to the standard of the Highway Department of Hong Kong, high and low binder content means 2% higher and lower than the normal binder content, respectively. * SMA represents the Stone Mastic Asphalt.
Table 3 Asphalt mixture gradations. Gradation Type Binder Content (%)
B.S. Sieve(mm) 14 10 5 2.36 1.18 0.60 0.30 0.15 0.075
Low Normal High
Dense-graded Wearing Course
Gap-graded SMA
Open-graded Friction Course
– 6 –
4.2 6.2 8.2
– 5.5 –
% Passing 100 96 70 49 35 25 17 11 5.2
% Passing 100 93 32 27 – – – – 10.4 (Including 2% hydrated lime)
% Passing 100 88 20 11 – – – – 4 (Including 2% hydrated lime)
Fig. 3. EM density gauges for density measurement: (a) PaveTracker; (b) PQI.
gauges have a transmitter in the middle surrounded by a circular receiver. Such axis-symmetric structure makes the gauges insensitive to the measurement direction.
4.2. Calibration method As aforementioned, calibration is necessary for all EM density gauges. To identify the effects of different calibration methods on the accuracy of density measurement, the raw data of each measurement and their corresponding calibrated data were compared.
Fig. 6 presents the raw gauge data, calibrated gauge data and the laboratory measured bulk densities of 6 slabs prepared with SMA with normal binder content, as an example. The bulk density in this figure refers to the density measured by the SSD method according to ASTM, and this density was treated as the reference to calculate the prediction errors of the EM density gauges in this study. The measurements from the slabs prepared with other mixtures show similar trends. The average error percentages, which are calculated by Equation 1, are presented in Fig. 6c. From Fig. 6a and b, it can be seen that regardless the gauge type, there are considerable differences between the raw data and the SSD
Please cite this article in press as: Z. Leng et al., Laboratory evaluation of electromagnetic density gauges for hot-mix asphalt mixture density measurement, Constr. Build. Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.186
5
Z. Leng et al. / Construction and Building Materials xxx (2017) xxx–xxx
Perpendicular Compaction Direction
Parallel Fig. 4. Measurement Directions.
2.00 1.90
Raw Data (g/cm3)
Raw Data (g/cm3)
1.90
1.80
Perpendicular Parallel
1.70 1
2
3
4
5
6
1.80 1.70 1.60 1.50
Perpendicular Parallel
1.40 1
2
3
4
5
Test Slab No.
Test Slab No.
(a) PQI collected raw data
(b) PaveTracker collected raw data
6
Fig. 5. Raw data collected at two directions.
bulk density. After calibration, the accuracies of the density measurement using both density gauges were significantly improved. As Fig. 6a shows, compared with the uncalibrated data, PQI measured raw data after offset calibration are closer to the SSD bulk densities. Similarly, offset is also a feasible calibration method to increase the accuracy of PaveTracker, as Fig. 6b shows. From the data presented in Fig. 6c, it can be found that ‘‘offset” calibration can effectively decrease the error percentages of both PQI and PaveTracker, but the error percentage of PQI is smaller. Among all data, PaveTracker with the ‘‘mix calibration” provides the most accurate density prediction, as evidenced by the very small prediction error of 0.2%.
Error Percentage ¼
Measured Value SSDBulk Density 100% SSDBulk Density ð1Þ
To statistically evaluate the effect of calibration method on density measurement using EM density gauges, Analysis of Variance (ANOVA) was performed on the raw data and calibrated data. As expected, significant differences at the confidence level of 95% were concluded between the SSD bulk density and raw data, indicating that the raw data of EM density gauges cannot be directly used. Paired-t tests were further performed on the calibrated data to verify the effectiveness of various calibration methods. As Table 4 shows, regardless of the gauge type and calibration method, all P-values are larger than 0.05. This indicates that the null hypothesis that there are no statistical differences between the calibrated data and bulk density can be accepted. Based on the data presented in Table 4, it was found that calibration can significantly improve the measurement accuracies of the EM density gauges.
Please cite this article in press as: Z. Leng et al., Laboratory evaluation of electromagnetic density gauges for hot-mix asphalt mixture density measurement, Constr. Build. Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.186
6
Z. Leng et al. / Construction and Building Materials xxx (2017) xxx–xxx
2.20 2.10 2.00 1.90 1.80 1.70 2.18
2.40 2.30 2.20 2.10 2.00 1.90 1.80 1.70 1.60 1.50 2.180
Measured Value (g/cm3)
Measured Value (g/cm3)
2.30
2.19
2.19
2.20
2.20
2.21
2.185
Bulk Density (g/cm3) Raw Data
Offset
2.190
2.195
2.200
2.205
Bulk Density (g/cm3)
Bulk Density
Raw Data
(a) Changes in the PQIm easured data
Offset
Mix Calibration
Bulk Density
(b) Changes in the PaveTracker measured data
Error Percent (%)
30 25
PQI
PaveTracker
19.1
20 15.7 15 10
4.0
5
1.4
0.2
0 Uncalibrated
Offset
Mix Calibration
Calibration Method
(c) Error Percent of Different Calibration Methods Fig. 6. Effect of calibration method on the accuracy of EM density gauges.
Table 4 Results of the Paired-t test for calibration methods. PaveTracker
Wearing Course SMA
Lower Binder Content Normal Binder Content
Dry Surface H2O index 4.1^ H2O index 8.1
Higher Binder Content Friction Course ^
PQI
Offset P-Value
Mix Calibration P-Value
Offset P-Value
0.312 0.235 0.545 0.316 0.113 0.362 0.695
0.303 0.262 0.696 0.605 0.256 0.990 0.938
0.292 0.339 0.693 0.795 0.484 0.125 0.580
H2O index was measured by PQI.
25
8.5
12 PQI Air Void Content
8 15 6 10 4 5 0
4.20%
6.20%
8.20%
6.3
6.5 6
6.1
5.4
5.5 5
0
4.5
Fig. 7. Relationship between the prediction error, air void content and the binder content.
6.9
7
2
Asphalt Binder Content
7.6
7.5
H2O Index
20
7.9
8
10
Air Void Conent (%)
Error Percentage (%)
PaveTracker
4
4.7
1
3
5 7 Passing Number of Compactor
9
11
Fig. 8. Relationship between passing number of compactor and H2O index.
4.3. Asphalt binder content Fig. 7 shows the air void contents and the measurement error percentages of PMSMA test slabs with different asphalt binder contents. It can be observed that the air void contents of the slabs
decreased with the increase of binder content. Meanwhile, the error percentages of both PQI and PaveTracker varied among different asphalt contents, but showed no clear variation trends. ttests were further performed to analyze the differences of error
Please cite this article in press as: Z. Leng et al., Laboratory evaluation of electromagnetic density gauges for hot-mix asphalt mixture density measurement, Constr. Build. Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.186
7
Z. Leng et al. / Construction and Building Materials xxx (2017) xxx–xxx
percentage, and the results indicated that these differences are significant at the 95% confidence level. However, it cannot be concluded that asphalt content affects the accuracy of EM density gauges, because the change of asphalt content also causes change to air void content, which may also be a factor affecting EM density measurement.
2.4
Measured Density (g/cm3)
2.3 2.2 2.1 2
4.4. Moisture condition
1.9 PQI-Uncalibrated
1.8
PQI-Offset 1.7
PaveTracker-Uncalibrated
1.6
PaveTracker-Offset
1.5
PaveTracker-Mix Calibration 2
3
4
5
6
7
H2O Index Fig. 9. Relationship between H2O index and measured density.
Fig. 10. Reference location determination.
d = 3 cm
8
During the construction process, water is often sprayed on the asphalt pavement surface to prevent asphalt mixture from sticking on the compactor rollers. The presence of moisture may result in change in the EM density gauge data since the dielectric constant of water (approximately 81) is much higher than that of asphalt mixture (approximately 4–7). To explore the relationship between the moisture level and the EM density gauge data, different amounts of water were manually sprayed onto the testing slab surface before testing. PQI has the function of reporting H2O index as an indicator of moisture level, while PaveTracker does not. As it is a very challenging task to accurately control the moisture level on testing slab surface in the lab, this study used the H2O index provided by PQI as a quantitative index of the moisture level. It is worth noting that the exact relationship between the moisture content and the H2O index is not provided by the manufacture, but it is known that a higher H2O index corresponds to a higher moisture content [23]. To determine the H2O index range to be considered in the laboratory testing, the H2O index range during the field construction process was first determined by collecting PQI data at the Po Shek Wu Road in Sheung Shui, Hong Kong. Fig. 8 presents the relationship between the passing number of the compactor and the H2O index during the construction process, which indicates that the H2O indexes of the pavement during compaction process were from 4.7 to 7.9. Based on this observation, density data at three different H2O index levels (2.1, 4.1, and 8.1) were collected to investigate the influence of moisture on density measurement. These H2O index levels were selected to cover the moisture content range in field construction. Fig. 9 illustrates the relationship between H2O index and the raw and calibrated data of PQI and PaveTracker. It is clear that both the raw and calibrated PQI data stay constant within the H2O index range of 2.1 to 8.1, implying that within the testing moisture range, the moisture’s effect on PQI can be ignored. However, it can be found that the raw density data and the offset calibrated data of PaveTracker increase with the increasing H2O index. But when the mix calibration is applied, the calibrated data of PaveTracker become relatively constant. Therefore, it is recommended that mix calibration should be applied to PaveTracker to remove the effect of moisture.
d = 5 cm
d = 7 cm
Fig. 11. Circular painting area with three diameters.
Please cite this article in press as: Z. Leng et al., Laboratory evaluation of electromagnetic density gauges for hot-mix asphalt mixture density measurement, Constr. Build. Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.186
2.50
(a)
PT Measured Data: 0 min
2.40 2.30 2.20 2.10
d=0 d=3 d=5 d=7
2.00 1.90 1.80
1
2
3
4
5
6
Measured Density (g/cm3)
Z. Leng et al. / Construction and Building Materials xxx (2017) xxx–xxx
Measured Density (g/cm3)
8
2.10 1.90 d=0 d=3 d=5 d=7
1.70 1.50
1
2
(c)
PT Measured Data: 30 min
2.10 1.90 d=0 d=3 d=5 d=7
1.70 1.50
1
2
3
4
3
4
5
6
Testing Slab No.
5
6
Measured Density (g/cm3)
Measured Density (g/cm3)
Testing Slab No. 2.30
(b)
PQI Measured Data: 0 min
2.30
(d)
PQI Measured Data: 30 min
2.30 2.10 1.90
d=0 d=3 d=5 d=7
1.70 1.50
1
2
Testing Slab No.
3
4
5
6
Testing Slab No.
Fig. 12. Effects of paint diameter and curing time on EM density gauge data.
(S)
SMA
Friction Course
(SW)
(SF)
(WS)
(WF)
Wearing Course
Cement Concrete
(W)
Fig. 13. Description of different structures.
4.5. Marking paint During the construction process, the locations for density measurement are often marked by spraying paint at the pavement surface for quality control purpose (Fig. 10). EM density gauge data are then collected at these locations to adjust the passing number of the compactor. However, the presence of paint may affect the accuracy of density estimation. In this study, circular painting with three different diameters (3 cm, 5 cm and 7 cm) were sprayed onto the surface of the wearing course testing slabs and SMA testing slabs to explore the relationship between paint diameter and EM density gauge data (Fig. 11). In addition, EM density gauge data were collected right after paining and 30 min after painting to evaluate the effect of paint curing on density measurement. Fig. 12 presents the EM density gauge data collected under different paint diameter and paint curing time conditions. As Fig. 12
(b) and Fig. 12(d) show, regardless of curing time, PQI measured data stay constant within the dimeter range of 0 cm to 7 cm. On contrary, raw density data of PaveTracker highly depend on the curing time and paint dimater. PaveTracker collected raw data increases with the increasing of painting area when the painting area was still wet. But after 30 min’ curing, there are not significantly differences among different paint diameters. Therefore, it is recommended that PaveTracker data should be collected after the painting area becomes dry if marking paint is applied. 4.6. Underlying layer It is necessary to evaluate the effect of underlying layer on density estimation using EM density gauge since pavement has a multiple-layer structure. Correspondingly, this study characterized
Please cite this article in press as: Z. Leng et al., Laboratory evaluation of electromagnetic density gauges for hot-mix asphalt mixture density measurement, Constr. Build. Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.186
9
2.50
PT Measured Data
2.30 2.10 1.90 1.70
S SW SF
1.50 1.30
1
2
3
4
5
6
Measured Density (g/cm3)
Measured Density (g/cm3)
Z. Leng et al. / Construction and Building Materials xxx (2017) xxx–xxx
PQI Measured Data
2.30 2.10 1.90
S SW SF
1.70 1.50
1
2
2.50
PT Measured Data
2.40 2.30 2.20 2.10
W WS WF
2.00 1.90 1.80
1
2
3
4
3
4
5
6
Testing Slab No.
5
6
Measured Density (g/cm3)
Measured Density (g/cm3)
Testing Slab No.
PQI Measured Data
2.30 2.10 1.90
W WS WF
1.70 1.50
1
2
Testing Slab No.
3
4
5
6
Testing Slab No.
Fig. 14. Effect of underlying layer on EM density gauge data: (a) PT collected raw data from S, SW, and SF, (b) PQI collected raw data from S, SW, and SF, (c) PT collected raw data from W, WS, and WF, and (d) PQI collected raw data from W, WS, and WF. Table 5 General linear model analysis results.
PQI PQI-Offset PaveTracker PaveTracker-Offset PaveTracker-Mix Calibration
Gradation
Binder Type
Binder Content
Air Void Content
Moisture Condition
Thickness
Y Y N N N
N/A N/A N/A N/A N/A
N N N N N
N N N N N
N N Y Y N
N N N N N
Y = Significantly different. N = Not significantly different.
4.7. General linear model analysis General linear model was applied to determine the effects of gradation and thickness of testing slabs on EM density gauge measurement. The SPSS software package was used to perform the general linear model analysis. In this model, asphalt binder type, moisture condition, asphalt binder content, air void content, and thickness of the testing slabs were input as independent variables, and error percentages of PQI and PaveTracker measurements were selected as depended variables. The analysis results are shown in Table 5. It can be observed that the accuracy of PQI measurement
can be affected by the changes in gradation of asphalt mixture. In order to exclude the effect of the presence of moisture, PaveTracker need to be mix calibrated.
0.8
Coefficient of Variation (%)
the performance of EM density gauges on six pavement structures as shown in Fig. 13. S, SW and SF represent 5 cm SMA with underlying cement concrete, 5 cm SMA with 5 cm underlying wear course, and 5 cm SMA with 5 cm underlying friction course, respectively, while W, WS, and WF represent 5 cm wear course with underlying cement concrete, 5 cm wearing course with 5 cm underlying SMA, and 5 cm wearing course with 5 cm underlying friction course, respectively. Fig. 14 shows the EM density gauge data collected from different structures with three different underlying layers. As can be seen, regardless of the structure composition and gauge type, EM density gauge data stay constant for structures with the same surface layer and different underlying layers. It can be concluded that the underlying layer of the measured layer may not affect the accuracy of the density measurement using EM density gauges. It should be noticed though that the thickness of the surface layer in this study is 5 cm, and when the thickness of the surface layer is smaller than 5 cm, this conclusion might not be valid any more.
0.6
0.4
0.2
0.0
PaveTracker
PQI
Wearing Course
PaveTracker
PQI
SMA
Fig. 15. The coefficient of variation of the EM density gauge data.
Please cite this article in press as: Z. Leng et al., Laboratory evaluation of electromagnetic density gauges for hot-mix asphalt mixture density measurement, Constr. Build. Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.186
10
Z. Leng et al. / Construction and Building Materials xxx (2017) xxx–xxx
4.8. Repeatability To verify the repeatability of the two EM density gauges, the data collection process was repeated 10 times at the same location. In each collection process, EM density gauges were rose around 30 cm followed by placing on the same location. Fig. 15 presents the coefficient of variation of the EM density gauge measured raw data. It can be seen that the coefficient of variations of PaveTracker collected raw data are approximately 0.4%, while those of PQI are only around 0.04%. This indicates that PQI provides better repeatability than PaveTracker. 5. Summary and findings This study investigated the accuracies of two common types of EM density gauges, PQI and PaveTracker, for asphalt pavement density measurement under different conditions. The following summarizes the major findings of this study: The effect of gauge measurement direction on EM density gauge measurement is insignificant. The accuracy of EM density gauge measurement can be considerably improved by calibration. For the test samples prepared in this study, the error percentages of PQI and PaveTracker were reduced from 15% to 1.5% and from 19% to 0.2%, respectively. After mix calibration, the effect of the presence of moisture on measurement accuracy is insignificant for both PQI and PaveTracker. But the accuracy of PaveTracker can be affected by moisture if the gauge is not calibrated or calibrated by the offset method. There is no clear relationship between asphalt content and accuracy of EM density gauge measurement. Statistical analysis showed that the effect of asphalt mixture gradation on density measurement using PQI cannot be ignored. Currently, follow-up field studies are being conducted to validate the on-site performances of the two gauges, and the corresponding results will be published when they become available. Acknowledgments The authors sincerely acknowledge the funding support from the Construction Industry Council (CIC) of Hong Kong (Project Number: K-ZJK6). The support from the Welcome Construction Co., Ltd. by providing the electromagnetic density gauges used in this study is also highly appreciated. References
[2] J. Hu et al., Investigation on fatigue damage of asphalt mixture with different air-voids using microstructural analysis, Constr. Build. Mater. 125 (2016) 936– 945. [3] S. Caro et al., A micromechanical model to evaluate the impact of air void content and connectivity in the oxidation of asphalt mixtures, Constr. Build. Mater. 61 (2014) 181–190. [4] P. Kandhal, S. Chakraborty, Effect of asphalt film thickness on short-and longterm aging of asphalt paving mixtures, Trans. Res. Rec. J. Trans. Res. Board 1535 (1996) 83–90. [5] S. Xu et al., Moisture characteristics of mixtures with warm mix asphalt technologies–a review, Constr. Build. Mater. 142 (2017) 148–161. [6] F. Xiao, V. Punith, B.J. Putman, Effect of compaction temperature on rutting and moisture resistance of foamed warm-mix-asphalt mixtures, J. Mater. Civ. Eng. 25 (9) (2012) 1344–1352. [7] A. Varveri et al., Influence of air void content on moisture damage susceptibility of asphalt mixtures: computational study, Trans. Res. Rec. J. Trans. Res. Board 2446 (2014) 8–16. [8] B. Killingsworth, Quality characteristics for use with performance related specifications for hot mix asphalt, Res. Results Digest 291 (2004). [9] X. Luo, F. Gu, R.L. Lytton, Prediction of field aging gradient in asphalt pavements, Trans. Res. Rec. J. Trans. Res. Board 2507 (2015) 19–28. [10] Z. Leng et al., Field application of ground-penetrating radar for measurement of asphalt mixture density: case study of illinois route 72 overlay, Trans. Res. Rec. J. Trans. Res. Board 2304 (2012) 133–141. [11] S.G. Williams, Non-nuclear Methods for HMA Density Measurements, MackBlackwell Rural Transportation Center, 2008. [12] IAEA, Manual on Nuclear Gauges, 1996, Vienna. [13] P. Romero, Evaluation of non-nuclear gauges to measure density of hot-mix asphalt pavements. Pooled Fund Study Final Report, The University of Utah, Department of Civil and Environmental Engineering, 2002. [14] Troxler Electronic Laboratories Inc., Manual of Operation and Instruction 2012, NC, USA. [15] Z. Leng, I.L. Al-Qadi, An innovative method for measuring pavement dielectric constant using the extended CMP method with two air-coupled GPR systems, NDT E Int. 66 (2014) 90–98. [16] Z. Leng, I. Al-Qadi, Railroad ballast evaluation using ground-penetrating radar: laboratory investigation and field validation, Trans. Res. Rec. J. Trans. Res. Board 2159 (2010) 110–117. [17] P. Romero, F. Kuhnow, Evaluation of new nonnuclear pavement density gauges with data from field projects, Trans. Res. Rec. J. Trans. Res. Board 1813 (2002) 47–54. [18] D.L. Allen, D. B. Schultz Jr, D.A. Willett, Evaluation of non-nuclear density gauges, 2003. [19] Z. Zhuang, Effectiveness Study of Non-nuclear Gauge for Hot mix Asphalt (hma) Pavement Construction, 2011. [20] S. Sebesta, M. Zeig, T. Scullion, Evaluation of non-nuclear density gauges for HMAC: Year 1 report. 2003, Texas Transportation Institute, Texas A & M University System. [21] A. Timm, et al. Evaluation of Non-nuclear density gauges for measuring inplace density of hot mix asphalt. in: Transportation Research Board 92nd Annual Meeting, 2013. [22] G. Hurley, B. Prowell, L. Allen Cooley Jr, Evaluating nonnuclear measurement devices to determine in-place pavement density, Trans. Res. Rec. J. Trans. Res. Board 1900 (2004) 56–64. [23] J.W. Henault, Field evaluation of a non-nuclear density pavement quality indicator, 2001. [24] M. Rose, H. Wen, S. Sharma, Evaluation of non-nuclear-density gauges for measuring in-place density of soils and base materials, in: Transportation Research Board 93rd Annual Meeting, 2014. [25] TransTech Systems Inc., Pavement Quality Indictor Model 301 Operator’s Handbook, 2003. [26] Standard Test Method for Bulk Specific Gravity and Density of Non-Absorptive Compacted Bituminous Mixtures, 2014.
[1] Z. Leng, I.L. Al-Qadi, S. Lahouar, Development and validation for in situ asphalt mixture density prediction models, NDT E Int. 44 (4) (2011) 369–375.
Please cite this article in press as: Z. Leng et al., Laboratory evaluation of electromagnetic density gauges for hot-mix asphalt mixture density measurement, Constr. Build. Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.186