Field Evaluation and Application of Intelligent Quality Control Systems

Abstract

During road construction, the accuracy of compaction work is critical for the structural stability and maintenance of the road. Although the plate load test (PLT) is commonly used for quality inspections, it is impractical to test every section due to time and cost constraints. Therefore, simpler testing methods are being extensively developed. This study compared quality inspection results using the commonly used PLT, the relatively simple dynamic cone penetrometer test (DCPT), the lightweight deflectometer (LWD) test, and an intelligent quality control system equipped with accelerometer and global positioning system (GPS) sensors in intelligent compaction (IC) rollers. The results showed a strong correlation between the conventional tests (PLT, DCPT, and LWD) and the values obtained from the intelligent quality control system. The correlation analysis between the intelligent quality control system and PLT, LWDT, and DCPT yielded R-square values of 0.69, 0.91, and 0.95, respectively, indicating significantly high correlations. The implementation of intelligent quality management systems in earthwork construction projects will facilitate a thorough verification of the compaction quality throughout all construction segments, ensuring consistent compaction across the project. By enabling real-time data acquisition and analysis, these systems differ markedly from traditional methods, reducing the frequency and necessity of manual inspections. This approach not only streamlines construction processes, but also enhances operational efficiency. As a result, integrating these intelligent systems is anticipated to significantly increase productivity by optimizing the workflow and resource utilization in earthwork construction.

1. Introduction

Compacting the subgrade and roadbed at every stage of road construction is critical for ensuring road quality and lifespan. Therefore, quality during construction is very important. Repetitive compaction work is essential to achieve the required compaction level after filling. Currently, the plate load test (PLT) and field density test (FDT) are the most common methods for assessing post-compaction quality. Firstly, performing on-site PLT and FDT measurements requires considerable effort, as it necessitates engineers to conduct each test individually. Secondly, only a few measurements are taken at specific locations, which are then used to evaluate the compaction quality across the entire road construction site. This method raises concerns about the accuracy of these assessments due to the sparse distribution of their measurements. Thirdly, only a limited number of inspectors are available, who are typically called in to perform these measurements only after the compaction process has been completed. The absence of immediate feedback and the inability to adjust processes in real time indicate missed opportunities to enhance the quality of compaction and expedite projects by ensuring optimal compaction on the first attempt, thus preventing the need for any rework [1].

Moreover, the repeated construction and inspection of road constructions and estate developments underscore the need for technological developments to resolve inefficiencies. With quality verification methods relying on manpower, ensuring uniform quality across all sections is impossible, leading to decreased accuracy and excessive time consumption. Thus, a technology capable of automating quality monitoring is required. Currently, the sand cone test and PLT are typically used for compaction quality inspection in civil engineering. However, these methods depend on the tester’s judgment regarding test location selection, and their accuracy varies depending on the tester’s expertise. Therefore, current testing methods cannot quantitatively verify compaction quality, and cost and time constraints limit their inspection coverage. To overcome these limitations, intelligent quality control technologies have been developed for use during compaction work [2,3,4]. Following the development of IT technologies and the Federal Highway Administration (FHWA)’s 2004 IC Roadmap, the development and field application of intelligent compaction (IC) technologies have accelerated. Original equipment manufacturers (OEMs) initially led the development of compaction rollers, resulting in different assessment metrics across equipment. Recently, a commercial retrofit kit using the compaction meter value (CMV) was developed, allowing for module usage irrespective of the OEM and its application in IC rollers, broadening the technology’s adoption. The continuous compaction control (CCC) method allowed for operators to assess the compaction quality on-site simultaneously with compaction work [5,6,7]. In the U.S., IC technology was first used in highway construction for Minnesota’s TH-64 project, which was recognized as the first U.S. project requiring IC technology [8]. Several states are now applying IC technology for quality control (QC) and quality assurance (QA) [9,10]. The CCC method involves fitting various sensors on compaction rollers to measure the ground stiffness through the relationship between roller vibrations and the ground’s reaction (rebound). Rollers equipped with CCC technology are called IC rollers, and many standard specifications have been established internationally [11]. In earthworks, applying new technologies requires reliable public standards, and the U.S. and a few European countries are leading in establishing such IC standards. Austria first introduced IC standards in 1990, followed by Germany, Sweden, and Switzerland. The International Society for Soil Mechanics and Geotechnical Engineering (ISSMGE) issued related guidelines in 2005 based on Austrian standards. These guidelines were integrated into the European technical standard CEN/TS 17006 “Earthworks-Continuous Compaction Control (CCC)” in 2018 [12]. The U.S. FHWA released generic IC specifications for soil applications, titled “Intelligent Compaction Technology for Soil Applications”, in 2014, and State Departments of Transportation (DOTs) have been progressively approving IC standards that incorporate local regulations. In South Korea, the Intelligent Compaction Specification [13] has been established, and the introduction of intelligent quality management in road construction is in its initial stages. Currently, road construction sites use location information to monitor the number of passes made by compaction rollers. Additionally, there is growing demand for intelligent quality management utilizing Compaction Meter Value (CMV) technology.

Although IC for soil applications is being adopted slower compared to IC for asphalt, its application is steadily increasing. Our study verified its field applicability by comparing and analyzing an intelligent QC system’s test results against traditional quality inspection tests.

2. Materials and Methods

Plate load tests are used to evaluate the load-bearing capacity of soil, primarily in large civil engineering projects. ASTM D1195 and ASTM D1196 cover the procedures for conducting plate load tests, but do not provide specific guidelines on the frequency of testing [14,15]. Generally, plate load tests are conducted 1–2 times per major section. Field density tests in road construction are performed every 200–500 m². Additionally, there are no specific regulations regarding the frequency of DCPT and LFWD tests.

2.1. Plate Load Test (PLT)

In PLT, a 300 mm diameter circular loading plate is placed on the compacted ground and a hydraulic jack is used for testing. The self-weight of the vibrating roller (10,555 kg) serves as the reaction force, and the modulus of elasticity (K30) is calculated from the vertical load–displacement curve. This method is described in ASTM D1195 [14]. The plate loading test is commonly used for evaluating the bearing capacity and stiffness characteristics of subgrades. To obtain the stress–settlement curve of a subgrade, a loading plate with a diameter of 152–762 mm is pushed by a hydraulic jack supported by a dead load, and the settlement of the loading plate is measured using dial gauges according to the load acting on the loading plate. The depth of the influence zone of the plate loading test corresponds to twice the diameter of the loading plate.

2.2. Field Density Test (FDT)

The FDT employs the sand replacement method to measure the in situ density of the ground and estimate the compaction. This method, detailed in ASTM D1556, is crucial for evaluating the soil compaction quality in construction projects [16]. The apparatus includes a sand cone with an aluminum funnel, a calibrated jar, a base plate, a scale for weighing soil and sand, and standard silica sand. The procedure begins with calibrating the apparatus by measuring the sand required to fill the cone and base plate. In the field, a hole is excavated, and the removed soil is weighed. The sand cone is then placed over the hole, and sand is allowed to flow into the hole. The remaining sand is weighed to determine the volume used. The volume of the hole is calculated using the weight and density of the sand. The in situ density of the soil is then determined by dividing the weight of the removed soil by the volume of the hole.

2.3. Dynamic Cone Penetrometer Test (DCPT)

In DCPT, an 8 kg hammer is dropped from a height of 575 mm to penetrate a 20 mm diameter cone into the ground, measuring the penetration depth of the cone per hammer strike to evaluate ground strength. The test is described in detail in ASTM D6951-03 and continues until the total number of strikes reaches 14 or the penetration depth exceeds 300 mm [17]. The test results are expressed as the dynamic cone penetration index (DCPI), which varies with the number of strikes. In this study, the value commonly employed in dynamic cone penetration tests (DCPT), calculated by dividing the total penetration depth by the number of strikes, as shown in Equation (1), was utilized for analysis.

$$DCPI={\displaystyle \frac{\mathrm{P}\mathrm{e}\mathrm{n}\mathrm{e}\mathrm{t}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}{\mathrm{C}\mathrm{u}\mathrm{m}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e} \mathrm{b}\mathrm{o}\mathrm{l}\mathrm{w}\mathrm{s}}}$$

(1)

2.4. Light Falling Weight Deflectometer (LFWD)

Our study adopted LFWD to calculate the ground’s modulus of elasticity. LFWD measures the surface deflection caused by the impact of a falling weight, enabling an accurate calculation of the ground’s modulus of elasticity. This testing method is described in detail in ASTM E2583-07 [18]. The experiments were conducted according to these standards. The data were measured 3–5 times per location, and any outliers with a deviation of 10 MPa or more were excluded from the analysis. The final modulus of elasticity was calculated as the average of the remaining measurements. LFWD equipment specifications and elasticity measurement methods vary by manufacturer and type, thus affecting the obtained results [19,20,21,22]. In this study, we utilized products from Dynatest, a world leader in testing equipment.

2.5. Intelligent QC System

IC is a technique that improves the accuracy and efficiency of compaction work on foundation soils. In intelligent QC, an integrated system with a vibratory roller that provides CCC is defined as an IC system [23,24]. Ensuring the desired compaction level of foundation soil during earthworks is critical. Conventional vibratory rollers can lead to localized under- or over-compaction, potentially causing uneven settlement or a reduced performance of foundation soil. This issue arises because roller operators struggle to assess the compaction area and soil levels in real time, making uniform compaction difficult to achieve. To improve compaction assessment, non-destructive and penetration testing methods (e.g., soil stiffness gauges, dynamic cone penetrometers, lightweight deflectometers, nuclear density gauges) are used to rapidly evaluate the physical and mechanical properties of soil. However, even rapid non-destructive tests, such as spot checks, can only cover approximately 1% of the total area, providing insufficient data to represent the overall compaction state of a site [3]. The concept of IC originated in Europe in the 1970s [25,26] and began with correlating compaction layer stiffness with vibration frequency, leading to the proposal of CMV. The first study on roller-based acceleration measurements was conducted in 1974 by the Swedish Road Administration using a Dynapac vibratory roller equipped with an accelerometer. Swedish Geodynamik was established in 1975, and CMV was developed and introduced in 1978. Swedish Dynapac began commercially offering a CMV-based compaction meter in 1980, followed by Swiss Ammann and American Caterpillar, among others. German Bomag developed Evib in the late 1990s to measure dynamic ground stiffness. In 1999, Ammann introduced ks, considering a two-degree-of-freedom spring-mass-dashpot system to measure soil compaction. These parameters (Evib and ks) are characteristic values of ground performance, unlike the traditional acceleration-based CMV. In 2003, Caterpillar introduced machine drive power (MDP) based on rolling resistance; in 2004, Japanese Sakai developed the compaction control value (CCV), i.e., a CMV-based measure using harmonic vibration components, which was applied to IC rollers [27]. Recently, various IC roller parameters are collectively represented by the intelligent compaction measurement value (ICMV), which is currently a standard metric in IC measurements.

To utilize the ICMV, the system includes a GNSS antenna and receiver, vibration sensors (accelerometers), connecting cables, and a monitoring (Figure 1) [4]. Each sensor is used to monitor precise location data and compaction levels. The GNSS antenna is mounted on top of the roller, the accelerometer is installed on the roller frame, and the tablet is placed in the operator’s cabin. The synchronization of compaction and location data is achieved through real-time data acquisition systems and wireless communication, with the initial data processing performed via edge computing. The collected data are stored in a central data logger and transmitted to cloud storage for further analysis and long-term storage. Real-time data are visualized on the tablet in the operator’s cabin, providing immediate feedback and generating detailed reports for quality control and performance evaluation. This system enables the management of the number of compactions and QC of the area before compaction.

From the method of continuous compaction evaluation, the CMV is obtained, calculated by analyzing the measured acceleration during the vibratory roller’s operation in terms of the ratio of the first harmonic amplitude to the fundamental frequency amplitude. In previous studies, CMV was expressed as shown in Equation (2) [24,25]

$$CMV=\mathrm{C}{\displaystyle \frac{A_1}{A_0}}$$

(2)

where $\mathrm{C}$ = constant (usually 300, related to ground characteristics), $A_0$ = first harmonic acceleration amplitude component, and $A_1$ = fundamental frequency acceleration amplitude component.

In this study, the constant $\mathrm{C}$ was typically 300; the position data and intelligent quality values were synchronized and recorded at a rate of one per second.

3. Results

3.1. Test Site Construction

To evaluate the applicability of the intelligent QC system, a full-scale site was constructed. The test location was at the future "National Performance Test Site for Road Infrastructure" project site within the Yeoncheon SOC Research Center of the Korea Institute of Construction Technology. The test method involved 7 cycles (14 one-way passes) of back-and-forth compaction (Figure 2). The compaction area was divided into seven lanes; each was approximately 17 m long, totaling approximately 20 m in length.

Table 1. Field sample properties.

USCS	PI	Gs	Gravel (%)	Sand (%)	Silt (%)	Clay (%)	γdmax (t/m3)	ωopt (%)
SP-SM	N.P	2.65	9.8	71.4	13.4	4.5	1.93	9.9

shows the results of the grain size analysis from laboratory testing on the fill material. The soil compaction test [28] indicated that the maximum dry density was 1.928 t/m³ and the optimum moisture content was 9.9%, meeting the quality standards for roadbed fill material (Figure 3 and Figure 4).

3.2. PLT

Due to the nature of the PLT, conducting tests in all areas and at every compaction pass is challenging. Therefore, at this test site, PLT was conducted in one lane during the second, fourth, and sixth compactions (4, 8, and 12 one-way passes). The results were obtained based on the 2.5 mm standard for asphalt pavement (subgrade), as shown in

Table 2. PLT test results.

Average Settlement	Load Strength (kN/m2)	Bearing Capacity Factor (K30)	Number of Compactions (Round Trips)
2.5 mm standard for AP pavement (subgrade)	185	74	2
	223	89	4
	245	98	6
	480	192	7

.

3.3. DCPT

The DCPT was performed in all lanes, with data obtained for each compaction count. As the number of compactions increased, the DCPI values generally decreased. The values from each lane were organized by the number of compactions. To determine the mean values after excluding outliers, a box plot was drawn, as shown in Figure 5.

3.4. LFWD

The LWDT was performed across all sections, with data obtained for each compaction count. The E values for each lane, based on the number of compactions, are shown in Figure 6, showing an increasing trend as the number of compactions increases.

3.5. CMV

The intelligent QC test using an accelerometer was also performed across all sections, with data obtained for each compaction count. The CMV values for each lane, according to the number of compactions, are presented in Figure 7, which also shows an increasing trend as the number of compactions increases.

3.6. Analysis

To analyze the correlation between the testing methods, the median (i.e., a commonly used data analysis measure) was chosen. The median represents the value that lies exactly in the middle of a data set when arranged in order. Unlike the mean, it is less affected by outliers, making it useful for indicating the central tendency. This study conducted a correlation analysis using the median values from the intelligent QC test (CMV) and the traditional QC tests (PLT, LWDT, and DCPT). The intelligent QC tests and PLT were conducted at compaction counts of two, four, six, and seven. Apart from the seventh compaction, the correlation with CMV was high, as shown in Figure 8. However, the R-squared value was relatively low compared to the other test values, likely due to the variability of the point-test method in PLT, while CMV evaluated the entire section. The correlation between the intelligent QC method and LWDT was very high, with a correlation coefficient of approximately 0.91 or higher, as shown in Figure 9. Additionally, the correlation analysis between the intelligent QC method and DCPT showed an exceptionally high correlation, exceeding 0.95, as shown in Figure 10.

The correlation analysis of the four testing methods showed that each method was statistically significant at the 0.01 level, as shown in

Table 3. Correlation between intelligent quality control methods and conventional quality control.

Correlation
		CMV	DPI	E	PLT
CMV	Pearson correlation	1	−0.977 **	0.956 **	0.829
	Significance probability (two-tailed)		0.000	0.001	0.171
	N	7	7	7	4
DPI	Pearson correlation	−0.977 **	1	−0.895 **	−0.749
	Significance probability (two-tailed)	0.000		0.006	0.251
	N	7	7	7	4
E	Pearson correlation	0.956 **	−0.895**	1	0.746
	Significance probability (two-tailed)	0.001	0.006		0.254
	N	7	7	7	4
PLT	Pearson correlation	0.829	−0.749	0.746	1
	Significance probability (two-tailed)	0.171	0.251	0.254
	N	4	4	4	4

**. The correlation is statistically significant at the 0.01 level (two-tailed).

. This outcome indicated that the correlation between the variables was simultaneously very strong and statistically significant. Additionally, the Pearson correlation coefficient was also very high, suggesting that the intelligent QC methods were feasible for field application.

4. Methods for Field Application of Intelligent QC System

For applying the intelligent QC system to field conditions or systems, four stages can be proposed, as shown in

Table 4. Levels of application of the intelligent quality control system: 4 stages.

Stage	Number of Compactions	IC Value	PLT	Quality Check
Current	-	-	Performed	Number of Compactions + PLT (Random Location)
Stage 1	Automatically Recorded	Selectively Recorded	Performed	Number of Compactions + PLT (Random Location)
Stage 2	Automatically Recorded	Automatically Recorded	Performed	Number of Compactions + PLT (Location with Lowest IC Value)
Stage 3	Automatically Recorded	Automatically Recorded	Performed	Number of Compactions + PLT
Stage 4	Automatically Recorded	Automatically Recorded	Performed	IC Value

. Currently, compaction counts are managed from memory according to the optimum number of compactions determined by the test construction, and PLTs are performed at random points after compaction. Stage 1 involves automatically recording the number of compactions and performing PLT at random points after compaction. This approach can reduce operator memory errors and directly identify under-compacted or over-compacted sections. Stage 2 involves automatically recording the number of compactions and selectively recording IC values, allowing operators to identify the locations with the lowest measured IC values and perform PLTs at these spots. Stage 3 involves automatically recording both the number of compactions and IC values and automatically specifying the locations for PLT. Finally, Stage 4 involves replacing traditional quality tests with intelligent quality test values. It is believed that this approach will allow intelligent quality control to be effectively applied to field conditions.

5. Conclusions

This study assessed the feasibility of applying intelligent QC systems in the field based on comparative tests with conventional QC methods. The field test was conducted at the SOC Demonstration Center of the Korea Institute of Construction Technology, using actual road fill materials to simulate road embankment sections. The same materials and systems were used for the field tests across all sections. Traditional tests, including PLT, DCPT, and LWDT, were conducted, and the results were compared and analyzed. After evaluating the field applicability of the intelligent QC system, four stages were proposed for practical field use, with the results as follows.

First, four types of tests were conducted: an intelligent QC test, PLT, LWDT, and DCPT. The intelligent QC test value (CMV) increased as the compaction level rose; PLT also showed a tendency for the K30 value to increase as the number of compactions increased. The LWDT confirmed that the modulus of elasticity (E value) increased with the number of compactions, while DCPT showed that the DPI value decreased as the number of compactions increased. The correlation analysis between the intelligent QC system and PLT, LWDT, and DCPT yielded R-square values of 0.69, 0.91, and 0.95, respectively, indicating significantly high correlations. Second, to apply the intelligent QC system in the field, comparative analyses were performed using results from the conventional QC tests, including PLT, LWDT, and DCPT. The Pearson correlation analysis showed very high correlation coefficients: 0.83 for the intelligent QC method and PLT, 0.96 for LWDT, and 0.98 for DCPT, suggesting that the intelligent QC system could be applied in the field. Third, four stages were proposed for implementing this intelligent QC system in the field. Stage 1 involves automatically recording the number of compactions to manage the site, while Stage 2 involves automatically recording the number of compactions and measuring the IC values at necessary locations, performing PLTs at points where the intelligent quality values are lowest. Stage 3 involves automatically recording both the number of compactions and IC values, using intelligent quality values to automatically specify the locations for PLT. Finally, Stage 4 involves managing the quality of road construction sites using intelligent quality values without conventional QC tests.

Overall, our study proposed stages for evaluating the applicability of intelligent QC systems on granitic weathered soil grounds and their field implementation. To apply intelligent quality values in actual field conditions, additional testing and analysis will be needed to perform tests under various ground conditions and scenarios. Further analysis will be required in various situations.