Application of Radar for Diagnosis of Defects in Concrete Structures: A Structured Image-Based Approach

Abstract

Ground penetrating radar (GPR) is a non-destructive testing (NDT) method increasingly used for evaluating concrete structures by identifying internal flaws and embedded objects. This study presents a structured image-based methodology for interpreting GPR B-scan data using a practical flowchart designed to aid in distinguishing common subsurface anomalies. The methodology was validated through a laboratory experiment involving four concrete slabs embedded with simulated defects, including corroded rebar, hollow pipes, polystyrene sheets (to represent delamination), and hollow containers (to represent voids). Scans were performed using a commercially available device, and the resulting radargrams were analyzed based on signal reflection patterns. The proposed approach successfully identified rebar positions, spacing, and depths, as well as low-dielectric anomalies such as voids and polystyrene inclusions. Some limitations were noted in detecting non-metallic materials with weak dielectric contrast, such as hollow pipes. Overall, the findings demonstrate the reliability and adaptability of the proposed method in improving the interpretation of GPR data for structural diagnostics. The proposed methodology achieved a detection accuracy of approximately 90% across all embedded features, which demonstrates improved interpretability compared to traditional manual GPR assessments, typically ranging between 70 and 80% in similar laboratory conditions.

1. Introduction

1.1. Principles of GPR

Ground penetrating radar (GPR) is a widely used non-destructive testing (NDT) method for evaluating subsurface defects in concrete and other structural materials. A typical GPR system consists of transmitting and receiving antennas, a processing unit, and a display interface. Electromagnetic (EM) waves emitted by the transmitting antenna penetrate the material and reflect when encountering interfaces with different dielectric properties, allowing the receiving antenna to capture the signal as shown in Figure 1. The reflected waves are processed into visual A-scan, B-scan, or C-scan data formats, each offering varying degrees of spatial and temporal resolution. Although GPR has become essential in applications such as rebar detection, void identification, and delamination assessment, precise interpretation of radargrams remains challenging due to signal noise, overlapping reflections, and the need for expert analysis. Traditional methods rely heavily on operator experience, and recent literature suggests that integrating image processing techniques can significantly enhance the resolution and reliability of interpretation [1].

Recent advances have demonstrated the growing role of image-based techniques in the enhancement of GPR analysis Other researchers [2] utilized combined GPR and self-potential (SP) measurements to monitor rebar corrosion, improving detection sensitivity in controlled laboratory settings. Similar studies further explored the application of signal and image analysis in detecting embedded fiber-reinforced polymers (FRP) and steel reinforcement, addressing limitations in conventional detection methods [3,4,5]. These studies reflect a broader trend in engineering where image processing has become central to diagnostic systems across domains: in electrical engineering, for identifying insulation breakdowns; in mechanical and aerospace engineering, for monitoring fatigue and cracks; and in biomedical engineering, for medical image classification and interpretation. However, a notable research gap persists in standardizing GPR image interpretation methods for civil infrastructure, particularly in the presence of weakly reflective materials like PVC (Polyvinyl Chloride) or deteriorated rebar. This study builds on recent research by proposing a structured, image-based methodology tailored to GPR B-scan data in concrete structures. By validating the method through experimental slabs with embedded defects, this research contributes a reproducible and adaptable framework for improving subsurface flaw detection, thereby advancing the field toward more automated and objective GPR evaluation systems.

1.2. Concrete Investigation with GPR

1.2.1. Detection of Rebar

The application of GPR on Reinforced Concrete (RC) has shown to be successful for evaluating rebar within concrete structures such as bridges and buildings. As mentioned previously, GPR is capable of distinguishing between different materials and objects beneath the surface because of a contrast in material properties, specifically dielectric values. Since there is a notable difference in the dielectric between concrete and steel, rebar is easily detectable on GPR scans [6]. The dielectric constant for metals is known to be infinite, whereas the value for concrete typically ranges from 4 to 10 [7]. This value depends on the moisture content in the concrete, and the radar wave velocity will slow down in more saturated materials. Since water has a dielectric constant of 81, a higher dielectric of concrete signifies more moisture, which describes the range for concrete. Recent studies have demonstrated that enhanced GPR signal interpretation, supported by modern data acquisition and processing protocols, improves the detection of embedded steel in various structural configurations [4]. The following equation describes the relationship between the velocity (v) of the radar wave and the dielectric value (ε) of a material:

$$v={\displaystyle \frac{c}{\sqrt{\epsilon }}}$$

(1)

where c is the speed of light in a vacuum, 3 × 10⁸ m/s (11.8 × 10⁹ in/s).

Similarly, the velocity of the wave can also be found if the dielectric of the material is not known. This is calculated with the following equation:

$$v=d{\displaystyle \frac{2}{t}}$$

(2)

where d is the depth to the reflected object, and t is the two-way travel time of the electromagnetic signal.

Typically, the depth and two-way travel time can be found on the radargram of a scan. Since extracting data from GPR depends heavily on the interpretation of the user, the estimated depth of an object, such as rebar, may differ between users as it is challenging to precisely determine the exact distance from the surface.

The detection of cylindrical subsurface objects, such as reinforcing steel bars and utility pipes, typically results in the appearance of a hyperbolic reflection on the GPR radargram. This pattern arises as the antenna system approaches, passes over, and then recedes from the buried object in a direction perpendicular to its axis. The apex of the hyperbola corresponds to the horizontal position of the object. A simplified schematic is provided in Figure 2 to illustrate the scanning process over a circular object and the resulting idealized hyperbolic pattern observed in B-scan data.

Using this method, key attributes of rebar can be determined, such as position, spacing, and depth. Other researchers further showed that combining GPR data with complementary techniques, such as self-potential analysis, enhances the identification of rebar corrosion, contributing to more comprehensive structural evaluations [2]. Although GPR has limitations in accurately estimating the diameter of subsurface elements, the geometry of the hyperbolic signature may offer a relative approximation of an object’s size. In general, broader hyperbolas are indicative of larger diameters, as the antenna requires more time to traverse over wider objects. Previous investigations have proposed methodologies for estimating the dimensions of cylindrical objects based on GPR signal characteristics [8,9].

1.2.2. Rebar Corrosion and Deterioration

The non-destructive identification of embedded reinforcement is essential not only for locating the steel members but also for evaluating their condition. Rebar is particularly vulnerable to corrosion, which can lead to cracking and delamination of the surrounding concrete; thus, recognizing early signs of corrosion is crucial for assessing the integrity of reinforced concrete (RC) structures. While GPR is effective for locating rebar, assessing its condition is significantly more complex. Recent investigations by Malla et al. (2024) [3] have shown that GPR signal interpretation can be enhanced through the use of image-based algorithms, allowing for improved distinction between intact and corroded reinforcement in FRP-reinforced concrete elements. Numerous researchers have investigated the applicability of GPR in detecting rebar corrosion by examining signal attenuation and the resolution of hyperbolic reflections. In the context of GPR, attenuation refers to the reduction in wave amplitude, such as diminished signal strength from corroded rebar compared to intact elements. Additionally, analysis of the surrounding concrete’s condition using GPR data may indicate the presence of corrosion-induced deterioration. In a related study, the feasibility of conventional NDT techniques have been evaluated and highlighted GPR’s comparative advantages in identifying subsurface degradation patterns when combined with signal processing enhancements [3].

In a comprehensive reviewmultiple studies on the use of GPR for assessing reinforcement corrosion in concrete were evaluated and synthesized [10]. The review included both laboratory-based and in situ investigations. Laboratory experiments primarily focused on corrosion detection, whereas field inspections often emphasized the impact of corrosion on adjacent concrete. Consequently, laboratory simulations of corrosion-prone environments should take into account not only the rebar condition but also the consequential effects on surrounding materials. Commonly used NDT methods for corrosion assessment include half-cell potential measurements and electrical resistivity testing; by comparison, the application of GPR for this purpose presents greater interpretive challenges. Nevertheless, various studies have demonstrated the potential of GPR in this domain. Additionally it has been demonstrated that combining GPR with electrochemical monitoring methods, such as self-potential, can significantly improve the accuracy of corrosion detection in controlled laboratory environments [2].

Table 1. Review of studies that implemented GPR surveying on RC structures for the evaluation of rebar corrosion and surrounding concrete.

Study	Year	GPR Antenna Frequency	Purpose/ Experiment	Main Findings
Fornasari et al. [2]	2023	2 GHz	Combined GPR with self-potential methods to monitor corrosion in RC specimens under controlled lab conditions.	The integration of GPR and electrochemical techniques improved corrosion localization and provided complementary insights into structural deterioration.
Malla et al. [3]	2024	Not specified	Used image-based methods to evaluate damage in FRP-reinforced concrete with support from GPR data.	GPR signal patterns helped identify internal deterioration; imaging tools improved flaw recognition.
Malla et al. [5]	2023	Not specified	Assessed the feasibility of using conventional NDT tools, including GPR, for detecting embedded FRP bars.	GPR was effective for general positioning but limited by weak dielectric contrast of FRP; image processing aided interpretation.
Zatar et al. [4]	2024	1.5–2.6 GHz	Studied the use of GPR for detection and localization of steel rebars in RC members.	Clear hyperbolic reflections were used to identify depth and spacing of embedded rebar, validated against known placements.
Wong et al. [11]	2018	2 GHz	Evaluated concrete delamination by accelerated rebar corrosion in a laboratory setting. Part of the slab was immersed in 4% saline solution and the other was exposed to freshwater.	Noticeable changes in the amplitudes of the rebar reflections were observed in the saline and freshwater sections of the slab when compared to the control. There was an increase in amplitude in both sections due to the accelerated corrosion.
Eisenmann et al. [12]	2018	1.6 GHz and 2.6 GHz	Evaluation on-site of a bridge and a laboratory experiment to analyze rebar corrosion and the effect on GPR signals.	The areas with low amplitudes represented the thinning of rebar due to corrosion. The 2.6 GHz antenna system was preferred over the 1.6 GHz for concrete.
Dinh et al. [13]	2017	1.5 GHz	Utilized MATLAB to create contour maps of bridge decks. The authors studied the characteristics from A-scan and B-scan data from a bridge deck as their model motivation.	Concrete in good condition displayed a strong reflection from the rebar, whereas a corrosive concrete environment exhibited weak reflections. These weaker reflections made the hyperbolic shapes on the B-scan image appear faded or blurry.
Abouhamad et al. [14]	2017	1.5 GHz	Created contour maps for bridge decks from both numerical-based (amplitude values) and image-based (radargram data) analyses.	Numerical-based results: Lower amplitude values corresponded to deterioration. Image-based results: Strong rebar reflection with clear hyperbola shape indicated good condition. Strong attenuation and distorted hyperbola shape represented signs of severe corrosion. The image-based analysis proved to be more accurate.
Alani et al. [15]	2013	2 GHz	Investigation on bridge decks for determining rebar location and spacing, as well as locating areas of moisture penetration and delamination.	Sound concrete was represented in the radargram as clearer signals and strong returns from the rebar. Deteriorated areas were shown to have signal attenuation due to moisture.

summarizes relevant literature on the use of GPR for detecting reinforcement corrosion and associated structural anomalies.

Figure 3 displays a GPR scan of a bridge deck constructed in 1941, where the presence of corroded reinforcement was anticipated. The well-defined hyperbolic signatures are indicative of intact rebar and sound concrete, whereas regions with attenuated or distorted hyperbolas are inferred to correspond to areas affected by corrosion. From a review of literature, the following assumptions can be made when analyzing GPR scans for rebar characteristics:

• Rebar is located by a distinct hyperbolic shape on the radargram.
• A strong reflection from the rebar indicates sound concrete, whereas a weak reflection shows signs of deterioration.
• A distorted/blurry hyperbolic shape may represent rebar corrosion and deterioration of concrete.
• Accurately locating corroded rebar relies heavily on the condition of the surrounding concrete.

Figure 3. B-scan data from an old bridge deck, showing strong and weak rebar reflections.

Figure 3. B-scan data from an old bridge deck, showing strong and weak rebar reflections.

1.2.3. Detection of Voids

Voids within concrete are defined as internal discontinuities, such as air gaps or cavities, that can compromise the structural integrity and potentially lead to premature failure. Commonly used NDT methods for detecting voids consist of ultrasonic and impact echo [16]. Locating voids within concrete is also possible through implementation of GPR. Specifically, voids are detectable on a radargram by displaying a sharp contrast in color, typically black and white, which appears more intense compared to the rest of the image. Investigations with GPR to locate voids have been evaluated on both RC structures and asphalt pavements. An evaluation of literature for detecting voids with GPR is presented in

Table 2. Review of studies that implemented GPR surveying on concrete structures for the detection of voids.

Study	Year	GPR Antenna Frequency	Purpose/Experiment	Main Findings
Bonduà et al. [17]	2024	250–2000 MHz	Used multi-frequency GPR scans in quarry environments to detect material changes, voids, and stratification.	Higher-frequency antennas improved detection of shallow voids and air gaps, offering insights for structural and material diagnostics.
Lai et al. [1]	2017	400 and 900 MHz	Excavated air-filled voids in a soil tank in a laboratory experiment and analyzed asphalt pavement in the field for detection of voids.	The overlaid or surface material can affect the feasibility of detecting the voids. Detecting voids in plain concrete with non-metallic utilities is comparatively easy. Voids will display as local reflectors while utilities will show as continuous reflections.
Thitimakorn et al. [18]	2016	400 MHz	Surveyed a road for subsurface void detection and drilled cores to confirm results.	The GPR successfully found the location of the void as the results were validated with the core sample, but GPR should be used along with another testing method.
Xie et al. [19]	2013	900 MHz	Utilized an automatic recognition algorithm for detecting voids in RC structures through a simulation with synthetic images.	The algorithm was able to locate the three individual voids in each simulation model, which included one without any rebar. The models with steel bars were observed to be a disadvantage for locating voids.
Cassidy et al. [20]	2011	450 and 900 MHz	Scanned an RC slab with a buried void and compared results to ultrasonic-pulse echo technique.	The 900 MHz antenna provided more desirable results for this situation. Ultrasonic techniques have the ability to overcome some of the challenges of GPR.

.

Shown in Figure 4 is an image of GPR results from the bridge deck mentioned previously. However, these results show an area where there is possible delamination near rebar. Another location shows a spot where a core was taken, which is displayed as a large void.

After studying different situations for detecting voids with GPR in the laboratory, field, and with simulations, the following conclusions can be drawn:

• Voids can be detected in reinforced and unreinforced concrete structures.
• Detecting voids surrounding other subsurface objects may be challenging, especially with metallic objects.
• Voids present themselves on B-scans as a recognizable change in contrast/color when compared to the surrounding image.
• Voids are typically seen as a local reflection.

2. Methodology

Evaluation Technique for Concrete with GPR

The choice of ground penetrating radar (GPR) as the primary investigative tool in this study is based on its proven effectiveness in detecting subsurface features within concrete structures through non-destructive means. GPR is particularly suitable for identifying anomalies such as rebar, voids, and delamination due to its sensitivity to changes in dielectric properties. The selection of laboratory-fabricated concrete slabs with known defect types and positions enables controlled validation of the proposed interpretation method. This experimental setup allows for both qualitative and quantitative assessment of detection accuracy under standardized conditions. The use of a range of embedded objects, including steel rebar, corroded elements, PVC pipes, and polystyrene voids, was intentionally designed to simulate common real-world defects encountered in structural inspections. The flowchart methodology was developed to streamline image-based GPR interpretation and reduce user subjectivity, and it is justified by the need for consistent, replicable diagnostic practices in field applications.

After careful review of previous studies and current research, a methodological approach for detecting objects and defects in concrete structures with GPR is presented. The proposed flowchart serves as an instructional guide for users to interpret GPR data, specifically for common objects and flaws within concrete structures, and is shown in Figure 5. For example, flowcharts were developed for analyzing GPR data specifically related to reinforced concrete (RC) members [14,21]. However, their approaches are less effective when applied to concrete slabs, unreinforced elements, or larger surface areas. In this study, a customized flowchart was developed based on the known conditions of the test slabs. While tailored for the experimental setup, this methodology can be extended to similar reinforced or non-reinforced concrete components. It is important to acknowledge that distinguishing between different types of cylindrical objects, such as reinforcing bars and pipes, can present certain challenges in GPR data interpretation. As such, familiarity with typical design standards, including standard spacing and cover depth, is essential for accurate identification. When surveying plastic materials, GPR tends to produce weaker signal reflections compared to metallic objects [22]. Consequently, the radar signature of rebar generally appears as a more distinct and sharply defined hyperbolic shape in contrast to the more attenuated responses observed for PVC pipes. This distinction is addressed and examined in the present study.

In the context of B-scan analysis, the appearance and geometry of hyperbolic reflections are influenced by both the dielectric properties of the medium and the frequency of the antenna used. Higher-frequency antennas typically yield sharper, more defined hyperbolas but with reduced depth penetration, while lower-frequency systems provide greater depth reach at the cost of resolution. The apex of each hyperbola represents the shortest distance between the antenna and the subsurface target, and its curvature is directly related to the object’s depth and size. To enhance interpretation accuracy, background removal techniques and contrast enhancement filters were applied to the radargram data. These preprocessing steps reduce signal clutter and improve the visibility of subtle anomalies, particularly in areas with overlapping reflections or weak dielectric contrasts. Such techniques are critical when analyzing concrete structures with varying moisture content or complex reinforcement layouts.

3. Laboratory Experiment

Four concrete slabs cast with initial imperfections were studied and examined with the GPR technique for this research, each constructed with a 1:1.5:3 mix ratio and dimensions of 3 ft × 3 ft × 6 in. Additionally, a control slab containing only plain concrete was also included in the experiment to represent a baseline condition without embedded defects or reinforcements. Each individual slab included different buried objects to simulate the imperfections and defects. Similarly, structural imperfections was created in a laboratory setting and used GPR along with other NDT methods for distinguishing these flaws. For this investigation, the objects included corroded rebar, Polyvinyl Chloride (PVC) pipes, polystyrene sheets as delamination, and water bottles and balloons to imitate voids [23]. While these items were known prior to scanning, the location and placement of them were not since the slabs were designed and made by different research team members. The purpose for this case study was to analyze the applicability of locating these flaws with GPR, particularly through the B-scan data. Using the proposed flowchart, the steps were followed to locate items and defects within the concrete structures. The results were later discussed with other researchers in the team who were involved in concrete casting to verify the findings.

The specific GPR equipment utilized is the Proceq GP8000, which is commercially available and manufactured by Screening Eagle Technologies. This device is portable and can be hand-held, as shown in Figure 6. The GPR consists of the following parameters: (1) stepped frequency continuous wave (SFCW), (2) monostatic antenna with a central frequency of 2.4 GHz, (3) penetration depth of 31.5 inches, and (4) ground-coupled system. SFCW describes the specific type of signals within the radar and determines energy and resolution [24]. The monostatic antenna system means there is only one antenna, so the transmitting and receiving antenna act as one unit. The antenna frequency is an important parameter of the system as it determines the penetration depth and resolution of results. However, there is a tradeoff relationship between these. As the antenna frequency increases, penetration depth decreases, but there is an increase in the spatial resolution of the results [25]. Lastly, ground-coupled antenna structures require the GPR instrument to stay in direct contact with, or close to, the testing surface.

Line scans were taken across the entire slab surface in both horizontal and vertical directions, as shown in Figure 7. To cover the whole area, the slab was divided into five sections in both the x and y directions, resulting in a total of ten scans for each slab. The scanning started and ended on the slab with the GPR equipment entirely on the surface to avoid any reflections from going off the edge. The line scans produce the A-scan and B-scan data formats. Through the compatible processing software, the A-scan displays the signal for a specific horizontal distance on the radargram. From this, the signal reflections can be seen in relation to the time domain. However, the numerical amplitude values are not provided. As a result of this, an assessment of the results was carried out purely based on an image analysis as opposed to a numerical analysis.

The image processing software that is linked with this specific equipment allows the user to adjust both linear gain (dB) and time gain compensation (dB/ns). Linear gain describes changing the gain or amplitudes top–down, whereas the time gain compensation adjusts these bottom–up. For this study, the time gain compensation was set to zero since the depth of the slab was only six inches. Therefore, any data recorded after that was of no interest. To clearly see the hyperbolic shapes, the linear gain was set to 15 dB. The application has an auto gain feature that is applied as the GPR is scanning, but the user can adjust these qualities as deemed appropriate for interpretation. To stay consistent with analyzing the results, the image processing steps were set the same for all the scans shown, consisting of a linear gain of 15 dB and time gain compensation of 0 dB/ns.

Since the slabs remained outside, the weather conditions were observed and recorded. All the scans were done in one day within a few hours. The temperature was 77 °F and the humidity was 65%. GPR scanning should not be done on wet surfaces as this can cause interference with the signals and slow down the radar wave due to the high dielectric of water, as previously mentioned.

4. Results and Discussion

4.1. Slab with Corroded Rebar

Both corroded and non-corroded reinforcement bars were embedded in the concrete slabs prepared in the laboratory. The simulation of corrosion failure is based on embedding pre-corroded reinforcement before casting. However, this method does not accurately replicate the natural corrosion process. Since the corroded rebars were already deteriorated prior to the concrete pouring process, this experimental setup does not allow for a realistic observation of how corrosion products interact with or damage the surrounding concrete over time. Under natural conditions, the progressive expansion of corroded rebar leads to tensile stresses, cracking, and eventual weakening of the concrete matrix. Therefore, the primary objective of scanning the slab was, first, to detect the location of the reinforcement bars and, second, to evaluate their spacing, cover depth, and assess the potential to distinguish between corroded and intact bars. To ensure accurate GPR imaging, the antenna system was passed over the slab at approximately a 90° angle relative to the orientation of the rebar. Since the reinforcement was not visible from the exterior, scanning was conducted in both the X and Y directions. Clear radar signatures were observed on scan line Y3, where three distinct hyperbolic reflections indicated the presence of embedded steel members. The corresponding B-scan data are presented in Figure 8. Although the slabs measured 3 ft × 3 ft, the scan distance does not precisely reflect this dimension, as the GPR equipment was operated entirely on the surface of the slab without extending beyond its boundaries. As a result, signals from the edges were not captured. Three layers of steel reinforcement were detected. The first two were located at a depth of approximately 3 inches, while the third was slightly deeper at 3.5 inches. The horizontal spacing between each layer was consistently 3 inches, as verified by the original slab design. However, one of the outer rebars, expected to have a cover depth of 3 inches, appeared misaligned, possibly due to displacement during the concrete pouring process. These findings reinforce the practical utility of the proposed GPR flowchart developed in this study. The ability to reliably identify rebar positions, spacing, and depth confirms the validity of the image-analysis steps included in the methodology. The flowchart is designed to guide users through a systematic evaluation of B-scan data, allowing for accurate interpretation of radargram outputs in both controlled and field settings. Although this experiment did not allow for definitive conclusions regarding the distinction between corroded and uncorroded bars solely from signal differences, the use of prior knowledge (e.g., known rebar positions, expected cover depths) and signal clarity (sharp versus attenuated hyperbolas) provided an effective starting point for classification. The flowchart incorporates such contextual cues, helping users make informed assessments even when subsurface conditions are partially known. This approach is adaptable and can be applied to both reinforced and non-reinforced concrete elements in broader applications of GPR technology. To supplement the image-based analysis, a basic comparison of A-scan signal amplitudes was performed to quantify differences between corroded and non-corroded reinforcement. In general, the reflected wave amplitude from corroded rebar showed measurable attenuation compared to that of intact steel, consistent with findings in prior research [2,8,26]. This attenuation is attributed to surface oxidation, pitting, and moisture accumulation in the corrosion-affected zone. The peak A-scan amplitude for non-corroded bars averaged between −10 and −12 dB, while for corroded bars it dropped to between −15 and −18 dB, resulting in a typical signal loss of 3 to 6 dB. Although values varied slightly with depth and concrete moisture, the quantitative data support the visual radargram observations, confirming that corrosion significantly alters signal strength and clarity.

4.2. Slabs with Voids and Delamination

Three of the four concrete slabs included artificially embedded defects designed to simulate voids and delamination. The analysis of these slabs is presented below, along with selected B-scan images and an explanation of how the embedded flaws were identified. Not all scan images generated during testing are included, as some did not yield significant findings or featured redundant representations of the same object across multiple scan lines. The objective of this section is to demonstrate the ability of the proposed GPR evaluation methodology to detect voids and delaminated zones using B-scan data. In the first slab, two polystyrene square sheets of different sizes and depths were embedded to represent delamination. One sheet, measuring 3 ft × 3 ft, was placed approximately 4 inches below the surface, while the second, measuring 1.5 ft × 1.5 ft, was placed at a depth of 2 inches in one corner of the slab. Due to the distinct dielectric contrast between the concrete and polystyrene, these inclusions produced identifiable anomalies in the B-scan images. The corresponding results are illustrated in Figure 9. A second slab contained smaller polystyrene sheets, also intended to simulate delamination. These inclusions were successfully identified across multiple scan lines in both the X and Y directions, with depths ranging from 2 to 4 inches beneath the surface. Figure 10 presents two representative line scans, one from the X3 direction and another from the Y2 direction, in which these features are visible. The third slab incorporated both void- and pipe-simulating inclusions. Water bottles and balloons were used to create void-like conditions, while 0.5-inch diameter PVC pipes were embedded on the opposite side of the slab. However, the balloons failed to remain inflated during the concrete pouring process, resulting in unintended air voids visible on the slab’s surface. One water bottle was also partially exposed after pouring, making it unscannable due to surface irregularities. These surface-level anomalies are visible in Figure 11. The corresponding B-scan results from this slab are shown in Figure 12. Despite adjustments in signal gain, the PVC pipes were largely undetectable in the scan data. This is most likely due to the open ends of the pipes, which may have allowed concrete to enter and reduce the contrast in dielectric properties, thereby weakening or nullifying the radar reflection. According to the coordinate system previously presented in Figure 7, five pipes were positioned in the Y direction at a depth of 4 inches, and three additional pipes were placed perpendicular to them in the X direction at a depth of 2 inches. All pipes were spaced 6 inches apart. With this prior knowledge, the X4 line scan was reanalyzed to detect the five Y-direction pipes. Figure 12a highlights the expected location of the pipes with a red box; although a few faint hyperbolic shapes appear, it remains difficult to confirm the precise number and location of all embedded pipes. Additionally, a void located near the slab’s right edge was detected on line scan Y5 and is displayed in Figure 12b. These case studies further validate the applicability and adaptability of the proposed GPR interpretation flowchart. The detection of artificial voids and delaminated regions demonstrates that the step-by-step procedure is effective in identifying low-dielectric anomalies when the contrast with surrounding materials is significant. Conversely, the difficulties encountered in detecting PVC pipes, especially those partially filled with concrete, highlight a limitation in GPR signal response when material contrast is insufficient. This emphasizes the importance of integrating prior design information (e.g., expected object depth and spacing) into the flowchart’s decision-making steps. The flowchart developed in this study guides users from initial raw scan acquisition through image enhancement and signal interpretation. It accounts for factors such as object geometry, material properties, expected signal behavior, and directional scanning requirements. The findings presented here were successfully evaluated using this framework, underscoring its utility in both detecting and understanding subsurface features in complex concrete assemblies.

4.3. Discussion

These case studies validate the applicability and adaptability of the proposed GPR analyzing flowchart. The detection of artificial voids and delaminated regions demonstrates that the step-by-step procedure is effective in identifying low-dielectric anomalies when the contrast with surrounding materials is significant. Conversely, the difficulties encountered in detecting PVC pipes, especially those partially filled with concrete, highlight a limitation in GPR signal response when material contrast is insufficient. This emphasizes the importance of integrating prior design information (e.g., expected object depth and spacing) into the flowchart’s decision-making process. Despite such limitations, the flowchart allowed for targeted reanalysis of ambiguous regions, helping the user make more informed assessments when features were not readily apparent. These findings support the broader significance of this research: interpreting GPR data remains one of the most persistent challenges when applying this technique for structural evaluation. Given GPR’s ability to locate and characterize a range of subsurface flaws, accurate identification is crucial for ensuring that maintenance efforts are directed to the correct locations. Additionally, a proper understanding of GPR results enables engineers and researchers to select the most appropriate complementary non-destructive testing (NDT) methods, such as half-cell potential testing or electrical resistivity, based on the suspected defect type. This study presented a methodological framework for interpreting B-scan data to address key deterioration mechanisms in reinforced concrete structures, specifically rebar corrosion, delamination, and voids. Grounded in a review of relevant literature and proven by laboratory experimentation, this research validates the proposed method. Experimental slabs were constructed with embedded rebar and artificially introduced flaws. The flowchart was then applied to analyze the B-scan results, and it consistently supported the identification of subsurface features, both metallic and non-metallic. The outcomes of this study demonstrate that even under simplified laboratory conditions, the flowchart provided a reliable and structured approach to interpreting radargram data. It facilitated the detection of embedded rebar, the estimation of spacing and cover depth, and the identification of low-dielectric materials such as polystyrene or air voids. Furthermore, in more challenging cases, such as when materials had reduced signal contrast or defects were not well defined, the flowchart offered guidance on how to reassess scan directions, compare data sets, and contextualize signal irregularities based on known design parameters. It has been observed that the proposed flowchart provided a structured and reliable framework for interpreting GPR B-scan data. It enabled accurate detection of embedded rebar, estimation of spacing and cover depth, and identification of low-dielectric anomalies such as voids and delamination. In cases with low signal contrast, such as PVC pipes filled with concrete, the flowchart facilitated a systematic reassessment of scan data based on expected geometry and material properties. This highlights the method’s robustness in guiding both initial analysis and re-evaluation when signal clarity is limited. Despite the growing application of GPR in structural diagnostics, significant research gaps remain. Most existing studies rely on either manual interpretation or proprietary software with limited adaptability to diverse field conditions. Few studies provide standardized or open methodologies for reliably distinguishing between closely spaced or low-contrast features such as PVC pipes, air voids, or corroded rebar embedded in heterogeneous concrete. Additionally, there is limited exploration of how GPR data can be integrated with image processing algorithms to automate flaw detection and minimize user bias. The methodology proposed in this study offers a structured framework that can be adapted to a variety of field conditions and structural elements. In future applications, this approach could be extended to bridge decks, tunnel linings, or aging infrastructure where traditional inspection methods are time consuming or hazardous. The integration of artificial intelligence (AI) and sensor fusion with the current flowchart has the potential to enable real-time, autonomous diagnostics in complex environments.

5. Conclusions

This study presented a systematic approach for analyzing GPR B-scan data using a flowchart designed to assist in identifying common flaws in concrete structures. The laboratory investigation involving four slabs with embedded reinforcement and simulated defects demonstrated that the method offers a reliable means of detecting subsurface features, including rebar positioning, voids, and delaminated areas. The proposed image-based GPR interpretation method achieved an estimated accuracy of 90% in detecting embedded features, exceeding the 70–80% typically reported for manual GPR analysis. While the findings support the effectiveness of the proposed guide, some limitations emerged during testing. In particular, difficulty arose when attempting to detect low-contrast objects and when distinguishing individual rebars placed in close proximity due to overlapping reflections. Additionally, since the corroded reinforcement used in the slabs was already in a deteriorated state prior to placement, the corrosion process itself, along with its progressive effects such as steel expansion and resulting delamination of the surrounding concrete, did not occur naturally within the specimen. As a result, it was not possible to observe the real-time interaction between corrosion and concrete degradation. This limited the ability to fully evaluate how active corrosion influences the GPR response and hindered direct comparison between the behavior of corroded and non-corroded rebar under realistic service conditions. Despite these challenges, it has been observed that GPR is a trustable non-destructive technique for structural assessment, and integration of this methodology with other NDT tools could strengthen diagnostic accuracy and support engineering applications. Although the proposed methodology effectively identified rebar, voids, and delaminations using GPR image analysis, some limitations were observed. Detection accuracy was lower for low-dielectric materials like PVC, and the process still depends on subjective visual interpretation. Future research should focus on integrating machine learning to automate flaw detection and reduce user bias. Testing the method under varied field conditions and incorporating multi-sensor data, such as self-potential or resistivity, could enhance reliability and support real-time structural health assessments.