Exploratory Field and Laboratory Investigation on the Use of Noncontact Digital Ski Sensor in the South Korea Expressway Network

the


Introduction
For road users to enjoy a satisfying driving experience, good road condition is essential [1][2][3][4][5][6][7][8].Smooth and flat road surfaces are advantageous since they lead to lower energy consumption and minimize maintenance and emissions [4][5][6].Also, a smoother and even asphalt pavement surface may be safer, reducing the accident rate and consequent social costs.In addition, this can result in an infrastructure that is less prone to distress, which leads to higher costs due to more frequent maintenance activity throughout the pavement service life [2,5,6,[8][9][10][11][12][13].
The smoothness of a road is frequently assessed using the international roughness index (IRI, m/km) as operated by many pavement management agencies [4-6, 14, 15].The IRI provides crucial information on pavement roughness; this parameter is conventionally derived based on the pavement profile measurements and mathematically schematized as a quarter-car vehicle with a simulated constant vehicle speed of 80 km/hr [16][17][18].Higher IRI values indicate poor and negative pavement smoothness; therefore, specific limiting thresholds are set and then defined in many pavement agencies.For the expressway system (South Korea's primary highway network), for instance, an IRI of 1.6 m/km is necessary for the quality control process [19,20].Different research has investigated pavement smoothness by investigating newer material solutions, improved pavement structure, and specific construction strategies [4][5][6][21][22][23][24].More sophisticated research was devoted to studying computational solutions for estimating and predicting pavement smoothness [8,9,14,[25][26][27][28], leaving little and less attention to the construction practice and technical solutions, which may ultimately result in better surface characteristics.Such a limited interest in the innovation and development of asphalt paving technology (e.g., paving equipment) for improving smoothness can be partially associated with a poor understanding and communication between academia and the asphalt pavement construction industry.Beyond the research environment, contractors recognized the necessity for more advanced tools as a critical component in enhancing infrastructure quality [4][5][6].The long-range surface contact ski (LSCS) system, also identified as long ski (LS), is commonly combined with the string line (SL) to define the smoothness in Korean Expressway construction and many other Asian countries [4][5][6].However, this system presents limitations that constrain further pavement surface improvement (e.g., resulting in lower IRI) [4][5][6].Many machine industries developed novel and advanced devices that may enable better surface smoothness of asphalt pavements.Multisonic sensors provide a solid alternative as they rely on the acoustic emissions [29][30][31][32] resulting in the noncontact digital ski (NCDS) system for asphalt pavement applications [31,32].Contrary to earlier traditional approaches, the NCDS does not call for direct physical contact between the machine and the road surface.A noncontact distance measuring technique that uses readings from several sensors automatically determines the design thickness of the layer to be paved [4][5][6][30][31][32].
In comparison to the traditional pavement surface smoothing technique (e.g., LS: long ski + SL: string line), the NCDS system, based on this technology, may be able to deliver consistent and enhanced pavement smoothness, also for curved sections and long ranges.Figure 1 schematizes how the LS + SL and NCDS systems vary from one another.

Methodology
This study compares the typical South Korean approach of combining the LS and SL techniques (LS + SL) for building composite pavement in the expressway network with the impact of using NCDS for increasing road smoothness [15][16][17].The IRI values from the LS + SL and NCDS were visually and statistically analyzed.In order to assess the cracking resistance of asphalt mixture at low temperatures, bending beam rheometer (BBR) was used to carry out creep test field material from sections paved with the NCDS and LS + SL methods [33][34][35][36][37][38].In assessing asphalt material lowtemperature performance evaluation, indirect test (IDT), semicircular bending (SCB), or dynamic modulus (DM) tests were widely used [36,[39][40][41].One of the drawbacks of these testing methods is associated with the expensive cost of the testing device and the large testing size of the specimens making the field testing cores obtained from thin surface layers hard to perform.Several studies experimentally demonstrated that the BBR mixture creep test can provide good estimation of the low-temperature response of paving mixtures as the conventional IDT testing approach [35][36][37] with an inexpensive sample preparation procedure, lower testing device costs, and the same sample size used to binder testing [33,36,42].The BBR creep test was thus used for this work.A visual and statistical evaluation was used analogously to the one adopted for the IRI data.The key findings and conclusion are then discussed.Figure 2 illustrates the adopted methodology.

Background Information
3.1.String Line (SL): Surface Smoothing Method Based on Physical Contact.The string line method consists of positioning steel sticks every 5-10 m on the ground along the road that will be covered with asphalt.After that, a steel wire is continually attached to the already-installed steel sticks and connected to them.The goal pavement thickness is defined by the level supplied by the steel line.If no significant roughness or fluctuations in the evenness of the ground or underlying pavement layer are present, the asphalt paving machine continuously discharges asphalt materials based on the preinstalled steel line.The string line method is widely adopted in the pavement industry due to its simplicity and reasonable smoothness quality that is achievable during construction (both for county road and expressway construction) [4][5][6]23].Nevertheless, this approach presents a series of limitations associated with the high-labor-intensive procedures that demand significantly skilled workers.Therefore, theLS equipment is used in conjunction with the SL.An example of an SL system adopted in a pavement construction site on the Korean expressway network is shown in Figure 3.

Long-Ski (LS): Surface Smoothing Method Based on Physical
Contact.In this method, to further enhance the pavement smoothness, a series of steel plates/rollers were in contact with the paving surface.A single beam that was between 10 and 15 m long and resembling a long ski was created by the connecting plates or rollers.For this peculiar characteristic, this smoothing  Advances in Civil Engineering solution is known as LSCS or, in short, LS.Similar to the SL method, the LS approach provides a simple tool to achieve a satisfactory smoothing of the asphalt pavement.This solution is coupled with moderately limited operation costs compared to other methods, such as sonic sensors.Two significant limits may be found, though.The smoothness of the final surface layer may be severely impacted by irregularities on the existing surface since this technology first requires a lower reference surface where the asphalt material will be laid.Second, the LS method presents limited flexibility when a curved section needs to be paved.Figure 4 illustrates an example of an LS paving system.Therefore, advanced and more flexible techniques are required to enhance the degree of pavement smoothness.

Noncontact Digital Ski (NCDS): Surface Smoothing Method
Based on Nonphysical Contact.The NCDS apparatus uses 3-4 sets of multisonic sensors (also known as cartridges) that were each attached to a long beam that were typically    Advances in Civil Engineering between 7 and 13 m long.Five separate, acoustic emissionbased sensors make up each multisonic sensor cartridge.An enormous beam with several multisensor cartridges was fastened to one side of the asphalt paver during pavement building.The NCDS system was installed on both sides of the asphalt paver when improved pavement smoothness is desired.With three or four sensor arrays, each thickness measurement offered 15-20 single data points on either side of the asphalt paver.Out of this data set, each cartridge's highest and lower values were discarded; the remaining data were averaged to determine the asphalt layer thickness.The distance from the surface was measured, the thickness of the paving layer was calculated, and this information was concurrently given to the asphalt paver as they work.The amount of asphalt material released up to the particular paving level may be quickly determined using this approach.Consequently, the NCDS approach offered the following benefits: no direct physical contact with the sublayer beneath the pavement is necessary, which allows for greater flexibility in the paving process; this system also ensures a higher degree of reliability in paving the desired surface layer thickness.However, before the actual paving session, the system had to be calibrated and tested using actual pavement work.
A schematic concept of NCDS system and its usage on the Korean expressway network are shown in Figures 5 and 6, respectively.

Actual Pavement Field Evaluation Sections
The Korean Expressway network's two most recent construction segments were chosen (Table 1 for further details).
These field tracks were utilized to compare the NCDS smoothing capabilities against that of the traditional LS + SL method.About 4,500 km of the Korean expressway system are now under development, with a 90% completion rate [43,44].A situation like this is changing the road paving activity from new construction to maintenance techniques heavily focused on asphalt overlay, as for the chosen aged concrete pavement test sections.When current jointed concrete pavements (JCP) reach the end of their useful life, this technique involves adding an asphalt layer on top of them to create a composite pavement system.The advantage of the carefully chosen test section is that NCDS and LS + SL techniques were implemented, enabling a simultaneous comparison.One of the field sections from the building sites used for this experiment is shown in Figure 7.
To evaluate the effectiveness of the NCDS system, a sensor array was attached on each side of the asphalt paver (i.e., left and right side) with a set of three cartridges consisting of five sensors on each side.After removing the extreme (e.g., largest and smallest) thickness values measured from each cartridge of sensors, a total of 18 measurements (i.e., nine values from each side) were acquired for the specific overlaying process.The paving machine's three separate multisonic sensor cartridges were installed at the front, middle, and rear so that, if needed, the sonic waves could be directed both to the already-paved area and to the road shoulder.The pictures in Figure 8 visually presents the position of the sensors of the NCDS system on the asphalt paver.

Field and Laboratory Testing
5.1.Asphalt Pavement Surface Smoothness Measurement.Figure 9 shows the results of three measurements of the IRI value performed at each testing site [45].It is well known that lower IRI results (and/or trends) indicate smoother pavement surface conditions, providing better car-riding experiences to drivers.The schematic information of IRI measuring equipment is shown in Table 2.
The average values were then obtained, and comparison was performed after removing potential outliers according to the following rules [46] and schematic in Figure 10: The commercial pavement smoothness analysis application PROVAL (version 3.6) was then used to examine the field data [45,47].For each pavement site in the current study, more than 5,000 data were employed for IRI analysis since IRI findings were obtained every 20 cm.    1 for additional information).Due to the extreme cold on those sites during the winter, a straightforward low-temperature creep test utilizing the BBR was used to assess the material behavior [33][34][35][36][37][38]48].
The small size of the BBR samples (i.e., 102.0 × 12.7 × 6.25 mm), which is advantageous for relatively thin layers of asphalt mixture (e.g., t = 10 cm), as in the case of an overlay, is the reason for choosing such a test in place of the more popular IDT [33-38, 48, 49].Therefore, three asphalt mixture cores were collected at each construction site for each paving smoothing technology for a total of 30 asphalt mixture cores.Small beam specimens were cut from this series of samples according to a procedure detailed elsewhere [35][36][37][38].Similar to the testing specification for asphalt binder [33], creep stiffness, S(t), and corresponding m-value, m(t), can be easily computed from the results of the midspan deflection, δ(t) when considering the imposed constant load applied to the specimen (approximately 4,000-6,000 mN) (Equation (1)).Given the higher stiffness commonly experienced on mixtures compared to asphalt binders, the testing duration was extended to 1,000 s, so sufficient time was allocated for observing a meaningful evaluation of the midspan deflection [35][36][37][38].The stone mastic asphalt (SMA) used for the field sections presented a nominal maximum aggregate size (NMAS) of 13 mm and an asphalt binder having performance grade PG 76-22 [34].Therefore, two temperatures, low (PG + 10°C) and low (PG + 10°C) −12°C (i.e., −22 + 10 = −12°C and −22 + 10 − 12 = −24°C), were adopted for testing and thermal stress estimation, σ(T°C, MPa) [35][36][37][38].For each construction site, six replicates were tested per each adopted pavement-smoothing system (i.e., in total 12 BBR beam specimens per each test site: Site A to E).The design of the particular SMA combination adheres to the regulations set out by the South Korean Ministry of Land Infrastructure and Transportation [18].After the top surface of the pavement was removed, the asphalt mixture samples are shown in Figure 11, and the BBR testing apparatus used in this study is shown in Figure 12.

Computation of Thermal Stress with Advanced Laplace Transformation Approach
Conventionally thermal stress can be computed with complex numerical analysis steps along with Gauss quadrature theory [35][36][37][38].In this paper, an alternative computation method: Laplace transform approach, was selected due to its simplicity with one-step computation process [48,[50][51][52].This method consists of a series of steps as detailed below: (1) Computed the shift factor, a T , from the creep compliance, D(t), at the reference temperature:   Advances in Civil Engineering (2) Based on the experimental results, obtain the master curve of creep compliance in the reduced time domain, D ξ ð Þ, as follows [48,[50][51][52]: where A, B, C, D, and E are fitting function parameters.And theses fitting parameters are computed based on experimental asphalt mixture creep compliance: D(t), data fitting process.
(3) Apply the Laplace transformation to the conventional equation relating thermal stress to strain: where thermal stress and the coefficient of thermal expansion/contraction, α, are given by: (4) After performing the inverse Laplace transformation process with the Stephest algorithm [52], σ(ξ) can be approximated with a simple power-law function and next converted to the time domain as follows: Based on this process, σ(T°C) can be obtained within the desired temperature range and a 1 mm/s cooling rate.Further details on the computational procedures can be found elsewhere [48,[50][51][52].3. It must be mentioned that IRI measurements were performed three times per each test section (i.e., from Site A to E).

Field and Laboratory Testing
Figures 13-17 report the IRI limit of 1.6 m/km (i.e., solid line parallel to the x-axis), which is a quality threshold according to the guidelines of Ministry of Land, Infrastructure and Transport [19,20].Lower and uniform IRI (m/km) result trends than 1.6 m/km at each measuring point were observed for the NCDS technology applied section compared to the section paved with LS + SL.Based on the IRI data generation trends, remarkable differences between the two methods can be visually and numerically observed.Approximately 33%-55% higher values of IRI were detected when traditional methods were used (see Table 3).Moreover, when the NCDS system was used, relatively (or remarkably) lower (and/or better) IRI results (e.g., 0.68-1.04m/km) were observed, suggesting substantial improvements in the asphalt pavement smoothness.
To further support the results of the plots by visual inspection, statistics were used: a hypothesis analysis, t-test with a 5% significance level [46] (i.e., p-value threshold below which the two smoothing systems can be considered as statistically different).Two assumptions, data normality and constant variance, were imposed [46].In this paper, a hypothesis test was set as follows: In a hypothesis test, Equation (7) means null hypothesis (i.e., mean values: μ of NCDS and LS + SL, are not significantly different), and Equation ( 8) presents the alternative hypothesis (i.e., mean values: μ of NCDS and LS + SL, are significantly different).From Equations ( 7) and ( 8), the pooled standard deviation: S P , can be derived as follows:     10 Advances in Civil Engineering In Equation ( 9), S A and S B are calculated standard deviation of IRI (m/km) from Group A (i.e., NCDS) and Group B (i.e., LS + SL), respectively.Moreover, n A and n B present numbers of IRI (m/km) measurement and evaluation on each test section (e.g., Site A to E).
The results of t-static and corresponding degree of freedom: d f , can be derived from Equations (7) to (9) as Equations (10) and (11):

Advances in Civil Engineering
More details on how to conduct this analysis can be found in the literature [46].The final output of the statistical analysis is summarized in Table 4 in terms of p-value with 5% of significant level (i.e., α = 0.05).If the results of p-value are higher than 0.05, the null hypothesis is accepted (i.e., Equation ( 7)); otherwise, the alternative hypothesis (i.e., Equation ( 8)) is accepted.
It is possible to notice significant variations in IRI (m/km) between NCDS and LS + SL.When the LS + SL system was used, greater IRI (m/km) values were discovered on all of the tested sections, which might be the reason for a possibly worse driving experience.This finding implies that the NCDS method may significantly improve pavement evenness and smoothness, even when weighing the advantages against the time needed for initial calibration.
7.2.Low-Temperature Cracking Resistance Comparison (NCDS vs. LS + SL).In this section, thermal stress: σ (T°C), results of cored asphalt mixture specimens were compared through visual and statistical analysis approaches.Thermal stress: σ (T°C), results were computed based on the previous section (Section 6, Equations ( 2)-( 6)).The calculated thermal stress results are presented in Figures 18-20.Moreover, the statistical comparison process was also performed for σ (T°C) with a t-test; the results were incorporated in the same series of plots.The statistical analysis (i.e., hypothesis test) was performed based on Equations ( 7)- (10) with values of n A and n B corresponding to 6, respectively.
The thermal stress plots demonstrate that for all five test sites, the NCDS technique was shown to provide relatively lower thermal stress levels than the traditional SL + LS smoothing technology.Below −20°C, which is near the PG limit of the asphalt binder employed in the SMA mix design formula, statistical analysis shows no significant differences between the SL + LS and NCDS techniques.But when the temperature dropped below −20°C (Sites C, D, and E) or −30°C (Sites A and B), clear differences could be seen.This suggests that the smoothing system had a moderate impact on the asphalt material's rheological and mechanical response in addition to its surface characteristics, with the use of the NCDS device showing potentially positive behavior.

Summary and Conclusions
In this study, the viability of using the NCDS, a nextgeneration asphalt pavement surface smoothing technology, was assessed and contrasted with the traditional approach used to build South Korea's primary expressway asphalt road network.The nonphysical interaction between the smoothing apparatus and the pavement layer and the real-time optimized pavement thickness calculating system are the NCDS system's main advantages over conventional techniques.The following two keypoints can be highlighted:  12 Advances in Civil Engineering (1) It was discovered that applying the NCDS technique led to noticeably decreased IRI as supported by the measurements at different filed sites.(2) When NCDS was used, relatively lower thermal stress values from the field mixture were detected, perhaps indicating greater performance against low-temperature cracking resistance with added advantages.
Based on the findings, remarkable improvements in smoothing asphalt pavement surfaces can reasonably be expected utilizing NCDS technology application.However, it needs to be remarked that equipment calibration (i.e., in the paving construction site or on a test section) is essentially required before the actual pavement construction can be performed with the NCDS tool.

Recommendations
In this work, just a short pavement portion was taken into account.To further confirm the incredibly optimistic results of this article, longer pavement test sections (for example, more than 4 km long) with different asphalt compositions are required.The first findings of this combined field and laboratory study are encouraging and support further study into the application of the NCDS system for paving.This is a component of ongoing research that also includes a nondestructive pavement performance assessment, a larger laboratory testing campaign, and an expanded selection of combinations for asphalt overlay.

( 1 )
Determine the values of the inter quartile range (IQR) and set Q1 (upper quartile) and Q3 (lower quartile) values.(2) Calculate the maximum and minimum data range by computing the values of Q3-1.5 * IQR (M3 range) and Q1 + 1.5 * IQR (M1 range).(3) Identify any measure as an outlier if the measured IRI results are smaller than M3 or higher than M1 values, respectively.

FIGURE 7 :
FIGURE 7: Picture of the testing construction site-Jun-Nam area (conditions at the end of the paving work for Lane 1) Ho-Nam expressway, South Korea.

FIGURE 10 :FIGURE 11 :
FIGURE 10: Schematic for the identification of outlier in the IRI (m/km) measurements.

7. 1 .
Asphalt Pavement Surface Smoothness Measurement.All the IRI results are shown in Figures 13-17 and summarized in Table

FIGURE 19 :
FIGURE 19: Thermal stress (Site and Site D).

TABLE 3 :
IRI (m/km) of the five test sections.

TABLE 4 :
Statistical analysis results of IRI (m/km) for the different pavement sites.