Integrated Assessment Methodology for Asphalt Pavement Integrity Under Accelerated Loading Conditions and GPR ^†

Abstract

Ensuring the integrity of pavement structures necessitates a thorough evaluation of both surface-level damage and subsurface mechanical performance. This study proposes an integrated, non-destructive assessment framework tailored for semi-rigid base asphalt pavements subjected to repeated vehicular loading via MLS66 full-scale accelerated testing equipment. The proposed methodology integrates ground-penetrating radar (GPR) using the CO4080 system and dynamic response measurements from a falling weight deflectometer (FWD) to characterize structural conditions across multiple depths. Comparative analysis between pre-loading and post-loading data revealed significant deterioration trends in the surface layers, with stiffness loss closely associated with increasing load repetitions. In contrast, the underlying base layers exhibited stable deformation characteristics, with variations in deflection basin indices remaining within ±5%. Subgrade dielectric properties derived from GPR data confirmed consistent compaction quality throughout the test site. Statistical analysis further validated the synergy between GPR and FWD results, demonstrating that the combined application enhances diagnostic accuracy. The dual-method approach improved overall evaluation reliability by approximately 22–35% compared to using individual techniques alone under accelerated pavement testing scenarios. These findings support broader implementation of integrated sensing systems and highlight the potential for application across varied pavement types and loading conditions.

1. Introduction

With asphalt pavement constituting 78% of China’s expressway network, its structural resilience faces multifaceted challenges [1,2,3]. Field investigations reveal that 54% of premature failures (service age < 8 years) originate from composite degradation mechanisms where dynamic load spectra exceeding 100 kN (AASHTO T320) interact with seasonal temperature fluctuations (ΔT = 62 °C) and hydraulic erosion (permeability > 0.03 cm/s) [4,5,6,7]. This deterioration mechanism progressively diminishes pavement serviceability [8,9]. Such multi-physics interactions necessitate integrated condition assessment protocols combining NDT methods [10,11].

Ground Penetrating Radar (GPR), as a novel approach within the domain of pavement non-destructive assessment, enables the rapid and efficient detection of deterioration across different structural layers of the pavement system [12,13,14]. By sending high-frequency electromagnetic pulses below the pavement surface and recording the returning reflections, GPR provides a continuous profile of internal pavement conditions without causing any physical damage to the structure [15,16]. This capability is especially valuable for identifying hidden defects such as moisture ingress, delamination, voids, and layer debonding that are often difficult to detect using traditional methods [17,18,19,20].

In addition to two-dimensional scans, GPR also facilitates the acquisition of three-dimensional (3D) radar datasets representing the volumetric structure of pavement layers [21,22]. These 3D representations allow for more intuitive and comprehensive visual interpretation, while also enabling advanced analytical techniques to assess the spatial distribution and severity of subsurface pavement anomalies [23]. This enhanced interpretability greatly aids engineers and maintenance planners in understanding the true condition of the pavement, even in complex or multilayered structures [24,25]. Complementary to GPR, the drop hammer bending instrument is utilized to simulate dynamic loading conditions like those imposed by actual vehicular traffic. This method evaluates pavement surface responses in real-time [26], offering insights into stiffness characteristics, deflection behavior, and the mechanical performance of individual layers [27]. Such dynamic assessments are essential for diagnosing early-stage fatigue or rutting potential that may not be evident in static or surface-level evaluations. Furthermore, the integration of artificial intelligence, particularly deep learning algorithms [28,29,30,31], has opened new avenues for automated and intelligent interpretation of GPR data [32,33,34,35]. These techniques enhance the detection accuracy of internal deterioration, such as micro-cracking or material segregation, and support the quantification of mechanical integrity across pavement depths [36,37,38]. Overall, the synergistic application of GPR, dynamic testing, and AI-driven interpretation contributes to a more precise and holistic evaluation of pavement condition. This integrated approach not only minimizes uncertainty in diagnostic results but also supports data-informed decision-making for optimized maintenance scheduling, cost-effective rehabilitation planning, and the extension of pavement service life [39,40].

Moreover, highway pavement structures often suffer degradation due to the combined effects of traffic loading and environmental conditions [41,42], such as temperature fluctuations [43] and moisture infiltration [44]. To assess the real-world performance of pavement structures or materials over their design service period, traditional approaches require long-term monitoring through field projects or test roads, which are typically time-intensive and labor-demanding, with results confined to specific sites [45,46]. To address this, researchers have introduced the Accelerated Pavement Testing (APT) method for asphalt pavements [47]. Utilizing the comprehensive APT setup [48,49], the accelerated full-scale loading procedure can simulate the cumulative impacts of actual traffic loads within a condensed timeframe, offering a more efficient means to evaluate pavement durability [50,51,52]. This methodology enables comparative studies across various pavement structures and materials under differing loading scenarios [53], while also supporting the validation of pavement response models and material behavior theories.

Therefore, to assess the behavior of semi-rigid pavement under APT loading, with possible reference to other pavement structures, the MLS66 system was employed to apply repeated loads to the test road. Core specimens were then extracted from both trafficked and untrafficked areas, and the structural response to loading was evaluated through FWD measurements, backcalculated layer moduli, and GPR inspection.

2. APT Design

In this study, the test pavement section was constructed with a multi-layered structure designed to replicate typical expressway pavement compositions. The structure comprised the following layers from top to bottom [54]: 7 cm of SMA-13 surface course, 6 cm of SUP-20, 8 cm of SUP-25, and a 36 cm-thick CSM1 layer with a cement content of 4.5%. Beneath this, a 20 cm-thick CSM2 layer containing 3.0% cement was placed, as illustrated in Figure 1. These layers were selected to reflect both standard asphalt pavement configurations and semi-rigid base conditions commonly found in service. To assess the pavement response behavior under repeated traffic loading, APT was employed. Testing was conducted in a controlled environment with ambient air temperature maintained at 20 °C. The loading device applied a single-wheel load of 50 kN, simulating a standard truck tire load, and operated at a constant speed of 22 km/h. The MLS66 full-scale loading system was utilized to impose one million repetitions of dynamic loading on the pavement structure. This was done to investigate the progressive deformation characteristics and long-term performance degradation of each structural layer. The experimental process enabled continuous tracking of pavement condition changes, thereby providing valuable insight into the structural resilience and fatigue resistance of the layered system over extended service simulation.

3. Structural Properties Evaluation

3.1. Pavement Modulus Analysis Based on FWD

FWD test was conducted on a full-scale test track to describe pavement deflection under various loading durations and assess the impact of load repetition on structural bearing performance. Due to the significant influence of temperature on surface bending behavior, a temperature compensation factor was derived through calibration in unloaded areas and subsequently used to adjust the measured deflection in the wheel path zone. This correction minimized temperature-related deviations. The layout of deflection measurements under varying loading durations is illustrated in Figure 2.

To describe the curve shape of the sink basin of asphalt pavement structure as realistically as possible, nine sensors were deployed during test. With the increase in distance from the load center, the spacing of sensors increased appropriately, and the specific spacing of sensors was 0, 20 cm, 30 cm, 45 cm, 60 cm, 90 cm, 120 cm, 150 cm, and 180 cm, respectively.

3.1.1. Representative Deflection

To minimize the thermal influence on pavement deflection measurements, the center point deflection values recorded under various loading repetitions were standardized by converting them to an equivalent value at the reference temperature of 20 °C. This temperature normalization process ensures that the results are directly comparable across different loading conditions. The correction coefficient used for temperature adjustment was derived based on deflection measurements taken from unloaded areas within the accelerated pavement testing zone, where external thermal effects could be isolated. This approach allowed for the establishment of a reliable baseline unaffected by loading-induced heat accumulation. The effectiveness of the temperature correction method is demonstrated in Figure 3, which compares center point deflection values before and after adjustment under loading scenarios of 580,000 and 810,000 repetitions. The temperature-corrected data provides a more accurate representation of the true structural response of the pavement, improving the precision of performance evaluation under accelerated testing.

After correcting temperature effects, representative center deflection values were calculated, as shown in

Table 1. Central deflection values under various loading.

Loading Times (10 Thousand)	Average Deflection (mm)	Standard Deviation (0.1 mm)	Coefficient of Variation (%)	Representative Deflection (mm)
9	4.2	0.3	6.4	4.6
58	4.8	0.9	19.8	6.4
81	5.1	1.2	23.5	7.1

. When subjected to cumulative loadings of 90,000, 580,000, and 810,000 cycles, the corresponding deflections were recorded as 4.0 mm, 6.4 mm, and 7.1 mm. Even under one million load applications, the pavement maintained minimal deformation and demonstrated stable structural integrity. In addition, a rise in the coefficient of variation was noted with increasing load cycles, indicating a progressive decline in the pavement’s structural capacity. When loading number reached 1.03 million, the deflection at the measurement location was reduced from 7.1 mm to 6.2 mm after applying a 4 cm SMA-13 overlay, highlighting the overlay’s effectiveness in improving load resistance.

3.1.2. Modulus Back Calculation

The Modulus 6.0 software was employed to perform inverse analysis of the stiffness values for each pavement structural layer, based on the measured sink basin data under varying load repetitions. Detailed results of this analysis are shown in

Table 2. Inverse analysis results of layer modulus under varying load applications.

Loading Times (10 Thousand)	Modulus of Structural Layers (MPa)
	Surface	Base	Subbase	Subgrade
9	7963 ± 1018	15,578 ± 4675	8749 ± 3529	154 ± 11
58	5435 ± 1277	12,479 ± 5541	9517 ± 1005	163 ± 9
81	5019 ± 225	10,178 ± 3549	9862 ± 667	166 ± 7
103	5541 ± 298	11,652 ± 3651	11,567 ± 1457	176 ± 12

. The stiffness values for the base, subbase, and subgrade layers demonstrated varying patterns under different loading conditions. As the cumulative number of load repetitions increased, the modulus of the surface layer progressively decreased from 7963 MPa to 5019 MPa. After a 4 cm SMA-13 overlay was applied, the modulus value rose again to 5541 MPa. These observations were consistent with changes in the sink basin curve and its corresponding geometric indices from D₀ to D₂₀. In Modulus 6.0, layer moduli were back calculated using a multilayer linear-elastic model by iteratively adjusting the moduli to best fit the measured FWD deflection basin (RMS/least-squares criterion). Poisson’s ratios were assumed as ν = 0.35 for asphalt layers, ν = 0.25 for cement-stabilized macadam layers, and ν = 0.40 for the subgrade.

3.2. GPR Investigation

To assess the presence of potential internal defects within the asphalt pavement layers, the base, intermediate, and surface courses of the test road were examined using GPR before and after loading operations. The scanning was conducted in both transverse and longitudinal directions. A CO4080 model GPR device (ImpulseRadar Sweden AB, Skolgatan 22, SE-939 31 Malå, Sweden), was employed for the data acquisition. Details of the GPR setup and detection parameters are listed below [55,56]: The CO4080 GPR detection configuration uses a center frequency of 800 MHz with an operating bandwidth of 500 to 1300 MHz. The horizontal GPR trace spacing is set to 1 cm, with 600 vertical sampling points and a 26 ns time window. A relative permittivity of 6 is assumed, corresponding to an equivalent propagation velocity of 0.102 m/ns and an equivalent detection depth of approximately 2.4 m. Distance (range) estimation is performed using a distance-measuring wheel, with a wheel diameter/circumference of 287/900 mm. While the layout of the scan directions is illustrated in Figure 4.

3.2.1. Structural Conditions Before APT

The detected image was filtered, and the obtained image data was used for disease recognition. Results show that the reflected electromagnetic wave signal within the internal pavement layers did not exhibit significant amplitude enhancement features. Instead, the waveform distribution remained relatively stable, with consistent layer interfaces and uniform reflection intensity across the inspection line. As indicated in Figure 5, the road subgrade in this section was generally in favorable condition, suggesting adequate compaction, minimal moisture intrusion, and well-bonded structural layers. The absence of anomalous signal characteristics implies that no significant structural deficiencies or voids are present within the surveyed segment.

In contrast to the uniform signals observed in other sections, the electromagnetic wave reflections within the red-marked region of Figure 6 were characterized by enhanced continuity and intensity, deviating from the baseline pattern. These anomalies may indicate the presence of structural irregularities or inconsistencies in material composition. Specifically, the increased amplitude and continuity of the reflected signals suggest potential zones of partial voids or delamination between structural layers. Such features are typically attributed to construction defects, such as insufficient layer bonding, inadequate compaction, or segregation of materials during placement [57].

In this case, the irregular reflections most likely originate from the interfaces between the lower base and the subbase layers, as well as between the subbase and subgrade. These discontinuities may result from a variety of in-service factors, including moisture infiltration, differential settlement, or load-induced fatigue. Additionally, poor drainage or repeated freeze–thaw cycles can exacerbate the formation of voids, leading to progressive deterioration if left unaddressed [58,59]. Therefore, the observed anomalies warrant further investigation through complementary non-destructive testing or core sampling and should be prioritized for targeted maintenance or rehabilitation measures to ensure pavement integrity and long-term performance.

3.2.2. Structural Conditions After APT

According to the interpretation of Figure 7, apart from the parabolic electromagnetic reflections caused by the embedded sensors (indicated in red), the signals obtained from both the longitudinal (Figure 7a) and transverse (Figure 7b) GPR profiles showed no significant increase in amplitude or disruption in continuity. These hyperbolic signatures are typical of localized point reflectors such as embedded objects, and do not indicate structural damage [60].

The remaining reflected waveforms across the scanned sections remained stable and continuous, without evidence of diffraction patterns, abrupt amplitude anomalies, or phase shifts that are typically associated with material defects, layer separation, or voids. This consistency suggests that the internal structure in this area was minimally affected by the APT loading process, and that no substantial deterioration occurred within the radar’s effective detection depth.

Furthermore, the absence of signal scattering or loss of coherence supports the conclusion that the pavement layers maintained good interfacial bonding and mechanical integrity under applied loading. The well-preserved stratification in both directions (longitudinal and transverse) indicates that the construction quality and material properties were sufficient to resist deformation or delamination. This confirms that the test section performed adequately under simulated traffic conditions, and no immediate rehabilitation measures are required at this stage. Continuous monitoring, however, is recommended to detect potential long-term damage initiation that may evolve under extended loading cycles or environmental exposure.

The combined FWD–GPR results indicate that the dominant deterioration at the investigated loading stage is concentrated in the upper asphalt layers. The back calculated surface-layer modulus shows a clear reduction with increasing repetitions, whereas the GPR profiles remain largely stable in deeper layers and do not suggest pronounced debonding or void development. This pattern is consistent with the commonly reported mechanism for semi-rigid base asphalt pavements, where early damage tends to initiate near the surface as fatigue microcracking and stiffness loss before evolving into macroscopic defects detectable by radar. The observed improvement after the SMA-13 overlay further supports the near-surface control of structural response, highlighting the complementarity of FWD (mechanical degradation quantification) and GPR (subsurface condition verification).

4. Conclusions

The results obtained from APT demonstrated that the surface layer exhibited a progressive reduction in its structural bearing capacity as the number of load repetitions increased. In contrast, the base layer maintained relatively stable mechanical performance throughout the loading cycles. GPR assessments further indicated that the roadbed in the examined section remained in satisfactory condition, indicating minimal structural deterioration in the deeper layers. These combined findings underscore the effectiveness of applying both FWD and GPR methods for evaluating pavement conditions from a multi-depth perspective. The integrated approach allows for more comprehensive monitoring of both surface and subsurface layer performance, enabling timely detection of potential weaknesses. This methodology provides valuable insights for pavement design optimization, preventive maintenance planning, and service life prediction, while future application to varied pavement types and loading spectra will enhance its generalizability. The combined interpretation indicates that the observed degradation is primarily surface-dominated at the investigated loading stage, while no pronounced deep-layer defects or interfacial failures were identified by GPR, demonstrating the complementarity of mechanical (FWD) and electromagnetic (GPR) diagnostics.