Abstract

Carbon fiber-reinforced polymer (CFRP) laminates have been successfully used as externally bonded reinforcements for retrofitting, strengthening, and confinement of concrete structures. The adequacy of the CFRP-concrete bonding largely depends on the bond quality and integrity. The bond quality may be compromised during the CFRP installation process due to various factors. In this study, the effect of four such construction-related factors was assessed through nondestructive evaluation (NDE) methods, and quantification of the levels of CFRP debonding was achieved. The factors were surface cleanliness, surface wetness, upward vs. downward application, and surface voids. A common unidirectional CFRP was applied to small-scale concrete samples with factorial combinations. Ground-penetrating radar and thermography NDE methods were applied to detect possible disbonds at CFRP-concrete interfaces. Thermography was found to clearly detect all four factors, while the GPR was only effective for detecting the surface voids only. The thermal images overpredicted the amount of debonded CFRP areas by about 25%, possibly due to scaling errors between the thermograph and the sample surface. The maximum debonded CFRP area in any sample was about two percent of the total CFRP area. This is a negligible amount of debonding, showing that the factors considered are unlikely to significantly affect the laminate performance or any CFRP contribution to the concrete member strength or confinement.

1. Introduction

The carbon fiber-reinforced polymer (CFRP) external laminate is widely used to strengthen and confine concrete structures for strength and durability. The adequacy of the externally bonded FRP-strengthening mechanism depends on the quality of the concrete-CFRP bond. Ideally, the FRP laminate that is perfectly bonded to the concrete surface without any debonding, delamination, or air bubbles is desirable. However, during the installation process, various factors may cause such defects and result in bond loss. The quality of the interfacial bond ensures satisfactory performance of the FRP-strengthened structure.

Nondestructive evaluation (NDE) methods are becoming increasingly popular for evaluation of both new and old structures. They are used for quality control, quality assurance, and assessing the quality of material and structural integrity. Several prior studies utilized NDE methods for assessing the concrete-FRP bond. NDE methods such as thermography, ultrasonic C-scan, acousto-ultrasonic, impact-echo, microwave, ground-penetrating radar, eddy current, and laser shearography were used to scan and image three precracked reinforced concrete (RC) beams with externally applied CFRP laminates [1]. Artificial delamination was made when the epoxy was freshly applied. The beams were found to have 9% less flexural strength and 6% more deflection than the beams without delamination. The locations and sizes of the defects were successfully detected, and microwave NDE was found to be the most accurate, detecting delaminations as small as 100 mm2.

Infrared thermographic inspection is a noncontact, full-field, fast, accurate, and reliable NDE procedure. Thermography senses the emission of thermal radiation from the material surface over time and gives the rate of cooling for the material. Thermography was used for the inspection of CFRP installation in a bridge near Besancon, France [2]. A simple setup using the uncooled infrared camera coupled with a handheld thermal excitation device was used. Wrapping and gluing defects in the installation were detected that resulted in debonding. Another study detected artificial defects inserted at the concrete-CFRP interface in small beams and roughly estimated the defect sizes [3]. Thermography is particularly effective in identifying actual or potentially weak or missing bond areas due to construction defects or imperfections. It has been used to monitor CFRP-strengthened reinforced concrete bridge columns, bridge decks, and reinforced concrete beams [46]. Gailetti et al. [7] located the predetermined voids and delaminations using an infrared camera. However, their method of creating voids and delaminations is by inserting foils of Teflon underneath the surface. These foils of Teflon might affect the results and the reading of the infrared camera. Therefore, there was a need to investigate the capability of the infrared camera to detect voids similar to the actual field condition.

Ground-penetrating radar (GPR) can be used to assess internal damage in structures through the generation of pulses to obtain information of the scanned surface. The information can include detail about different media, such as soil, rock, structural materials, water, and pavement. GPR can be air-coupled or ground-coupled depending on the antenna type. The former does not need direct contact with the surface, while the latter requires full contact with the surface. Previously, ground-coupled GPR with a 400 MHz antenna was used to evaluate the subsurface condition of roadway pavements [8]. The investigation showed that the approach can be successfully used in finding anomalies and voids hidden under the pavement surface. Another study detected rebars in glass FRP- (GFRP-) retrofitted cylinders using the 8–12 GHz GPR frequency range. Debonding using the far-field airborne radar technique integrating inverse synthetic aperture radar measurements and backprojection algorithm for the condition assessment were utilized [9].

Premature delamination of the FRP-concrete bond due to construction-related factors or errors can decrease any CFRP-strengthening contribution to the concrete structure that is accounted for in design (such as flexural, shear, and axial capacities) and also decrease any confining effects. Designers typically assume a perfect bond between the CFRP and the concrete substrate in capacity calculations. Detection and quantification of such construction-related factors will allow proper quality-control decisions on the CFRP application process and also allow the consideration of the debonding extent in structural calculations. No study has been conducted to date on the quantitative detection of any such detrimental factors through the NDE means and any quantitative detrimental effect of these factors on the FRP-concrete bond.

In the study reported herein, four common and likely construction-related factors were considered in developing quantitative NDE detection procedures, as follows:(1)Concrete surface cleanliness: this is one of the possible critical parameters that may affect the quality of the interfacial bond. Even though the surface may be roughened according to the manufacturer’s specifications, follow-up thorough cleaning to remove all loose materials could be an important factor in proper bonding.(2)Concrete surface wetness: the wet concrete surface due to rain or water leakage may result in weak bonding or bubbles, and evaporation of this water may cause localized debonding.(3)Upward vs. downward CFRP application: during flexural and shear strengthening of overhead concrete members (such as bridge girders), the FRP is normally applied upward (against gravity). However, most of the previous studies [10, 11] were conducted on laboratory specimens where the FRP was applied downward on the concrete surface. The overhead application may result in inadequate bonding because of the gravity effect.(4)Surface voids: voids can be present on or near the concrete surface due to air bubbles in the wet concrete and formwork imperfections. If the voids are left unpatched prior to the FRP application, some interfacial bonding may be lost.

2. Material Properties and Methods

2.1. Sample Preparation

The study utilized an experimental approach with small unreinforced concrete beams, application of CFRP on the concrete surface, and detection of any debonding through NDE methods. A total of 11 plain concrete beams of dimension 200 × 200 × 920 mm were cast using plywood forms. A 200 × 200 mm cardboard piece was inserted in each fresh concrete sample to form a notch at the midspan. The notch was not filled with the epoxy, and the purpose of the notch was to simulate a vertical concrete crack in the samples. Ready-mix concrete supplied by a local concrete plant was used to make sure that all beams have the same properties. The samples were moist-cured for 28 days to achieve a target 28-day compressive strength of 20.7 MPa and a slump of 114 mm. The mix design is presented in Table 1.

Figures 1 and 2 show the dimensions and the finished beam samples, respectively.

A common unidirectional CFRP laminate was applied at an ambient temperature of 16°C, in the absence of direct sunlight, to prevent any undesirable problems with the epoxy. The one-layer CFRP application completely covered one horizontal face of each sample, simulating a flexural strengthening in a beam-type member. The manufacturer-specified values for the CFRP laminate are presented in Table 2. Figure 3 shows the application process of the CFRP laminate on the concrete beams.

The selected type 1 epoxy, which was compatible with the selected CFRP and from the same manufacturer, was used to seal the surface irregularities on the concrete surface introduced during casting. It was a high-strength, high-modulus, and moisture-tolerant impregnating resin. The manufacturer-specified epoxy properties are presented in Table 3.

Type 2 epoxy was the major gluing agent, applied to both sides of the CFRP and also on the concrete surface. A Hi-Mod Gel, 2-component, 100% solid, solvent-free, moisture-tolerant, high-modulus, high-strength, and structural epoxy paste adhesive was used, conforming to ASTM C-881 [12] and AASHTO M-235 [13] specifications. Type 2 epoxy seals cracks and blends with the type 1 epoxy. The manufacturer-specified epoxy properties are presented in Table 4.

The study involved a parametric combination and two sample replications for all parameters except the ones with surface voids and unclean surface, as shown in Table 5. As stated in ACI 440 [13], the concrete surface was roughened by sandblasting to a CSP3 profile [15] to facilitate good-quality CFRP bonding, as shown in Figure 4.

2.1.1. Parameter 1: Concrete Surface Cleanliness

After sandblasting, the concrete surfaces were cleaned well to remove all loose materials or dust through air blasting, for samples 1–10. For sample 11, the surface was not cleaned, and some additional dust was deliberately placed on the dry surface before the CFRP application to determine the effect of an unclean surface [13]. This represents cases where the CFRP is not applied immediately after the sandblasting operation or improper cleaning after sandblasting.

2.1.2. Parameter 2: Surface Wetness

The surfaces of samples 7 and 8 were saturated with sprinkled water just prior to the application of epoxy and CFRP. For other samples, no water was applied, and the surface was kept dry before the CFRP application.

2.1.3. Parameter 3: Upward vs. Downward CFRP Application

Samples 5 and 6 were supported on both ends, allowing access below the samples, as shown in Figure 5. The CFRP was then applied from below as an upward application, typical for application on the bottom flange of girders and underneath of slabs. For the other samples, the CFRP was applied from the top, as shown in Figure 3.

2.1.4. Parameter 4: Surface Voids

To investigate the effect of air voids under the CFRP, a few small to large surface voids were placed on the fresh concrete for samples 3–6 using foam cubes (Figure 6). The void dimensions were selected to result in a possible decrease in bond areas. The allowable void size under CFRP, according to ACI 440.2R [14], is less than 1300 mm2 each. The void sizes were 10.2 × 10.2 × 10.2, 20.3 × 20.3 × 20.3, and 30.5 × 30.5 × 30.5 mm, respectively. No voids were planted on the surfaces of the remaining samples.

2.2. NDE Experimental Procedure

The NDE was performed seven months after the FRP application. This age effect can be considered as negligible because the FRP curing time is about eight hours. The concrete would continue to strengthen slightly with age. However, the NDE procedures detected only the reflected waves emitted from the entrapped air between the FRP and the concrete surface, and the concrete strength was not a factor herein. The NDE experimental approach used herein consisted of scanning the surface of the applied CFRP on concrete with two different NDE equipment, the GPR and thermal imaging. The ground-coupled antenna used in the GPR scans was of 2.6 GHz frequency, allowing for high-resolution radar images up to 250 mm depth. This scanning depth worked well with the very small thickness of the CFRP laminate. Figure 7 shows the cart-mounted GPR assembly that was used in this study.

Each sample was scanned through a handheld GPR scanner. With the normally polarized antenna orientation, two-line scans were captured along the length of the beam. The adjacent sides were also scanned. The 2.6 GHz antenna frequency produced an initial insensitivity of 60 mm, also known as the blind zone or depth. To overcome the blind depth, thin separation boards made of wood, foam, and concrete were placed on top of the CFRP surface, and scans from each were reviewed. The 65 mm thick concrete separator plank was found to work well and adopted herein. The wave profile, frequency, and amplitude were recorded using the GPR. During the investigation of the parameters, the scans could detect the presence of voids. Other parameters could not be identified by the GPR. Figure 8 shows the GPR test setup schematic, and Figure 9 presents a photograph of an actual GPR scan in progress.

The infrared (IR) thermography camera used in this study can detect temperatures in the range of −20°C to 120°C (4°F to 248°F) with an accuracy of ±2%. The main components of the IR camera are a lens, a detector in the form of a focal plane array, possibly a cooler for the detector, and the electronics and software for processing and displaying images. The detector type used in the camera is an uncooled microbolometer with 19,200 pixels. The infrared energy emitted from the object is converted to an apparent temperature, and the result is displayed as an infrared image. To check the bond quality between concrete and CFRP, the surface was heated with a heat lamp for 30 sec, and images were captured before and after the heating process. The lamp was of 250 watt incandescent 120 volt type. By moving the heat source over the sample from a fixed distance resulted in heat evenly distributed over the surface. The temperature of the heat lamp was allowed to reach a fixed value before taking measurements. Any anomalies and disbonds would result in a temperature difference and were recorded as a thermal image. Figure 10 shows the test setup and process.

3. Results and Discussion

3.1. Ground-Penetrating Radar

Figure 11(a) shows the GPR radargram for control sample 1. The graph shows a significant change in amplitude and a horizontal black area, which is the interface of the additional concrete separator and the beam surface. Except that, no other anomalies are noticed in this scan. The filtered scan for sample 5 (Figure 11(b)) shows a small parabola that indicates the clear presence of voids that were prefabricated in this sample. For multiple voids, multiple parabolas were detected, as seen in Figures 11(c) and 11(d). Presence of voids is indicated by a phase inversion that occurs at a concrete-air interface because of the lower dielectric constant of air. A corresponding change in amplitude could be observed but difficult to identify if the change was due to the CFRP-concrete interface (at the additional layer) or presence of voids. No significant change in the GPR amplitude or radargram was visible in the presence of surface wetness, upward CFRP application, and unclean surface (Figures 11(e)11(g)). Thus, the GPR was effective in detecting only the surface voids at the concrete-CFRP interface.

The major drawback of the GPR procedure was the difficulty in placing an additional concrete surface layer while scanning overhead. If the scanning is done without an additional surface layer, voids and defects present near the laminate surface could be ignored due to the initial insensitivity of the antenna. Scanning with GPR could be possible for cases where the CFRP is applied vertically (such as in column retrofitting or a beam web) or downward applications (such as the top of a slab or deck). The additional concrete layer can be easily placed in the downward applications. Application of the additional concrete layer could also be possible for vertical CFRP applications, if it can be adequately supported, and made sure that there is no gap between the additional layer and the CFRP surface. The GPR handheld antenna can then be conveniently employed on top of the additional layer.

3.2. Infrared Thermography

The thermal camera could capture both surface and subsurface defects and ultimately help determine the quality of the bond. The scans were very similar to each other for identical samples; so, only one of these scans is presented herein.

Figure 12(a) shows the thermograph of one half of sample 1 (control). The sample surface looks uniform due to absence of any uneven heating. The presence of the simulated concrete vertical crack is clearly detected as a solid line. The uneven heating on the surface due to the voids present is clearly visible as bright spots for samples 3, 4, 5, and 6, as shown in Figures 12(b)12(e). Also, the relative sizes of voids under the CFRP and the number of voids are clearly distinguishable in the thermograph images. The presence of surface wetness is also clearly visible in Figure 12(f), as compared to the control sample scan. Figure 12(g) shows that the upward CFRP application made a noticeable difference in the surface thermal profiles. The bright spots visible in the thermograph are the epoxy pockets that formed due to the gravity effect. The bright spots visible in Figure 12(h) are the increased temperatures due to the presence of dust on the surface before CFRP application. The size of voids could be roughly estimated by quantifying the bright zones in the thermograph. Thus, it is apparent that the thermography technique was successful in detecting all parameters that were considered herein.

3.3. Quantification of CFRP Delamination

The thermal images from samples with preinserted voids, shown in Figures 12(b)12(e), were used herein to quantify apparent CFRP-delaminated areas due to the presence of various surface defects. The associated thermal images were analyzed herein via export to AutoCAD [16]. The images were scaled to the exact sample dimensions. Only images from the camera vertically directly over the samples were used herein, as follows: (1) the exact plan view of the sample was drawn in AutoCAD; (2) the thermal images were then imported to AutoCAD and scaled to match exactly the plan view of the beam; (3) the debonded areas (with brighter thermal images) were then approximately estimated with the aid of the AutoCAD area tool. The known voided areas showed brighter images; thus, it was reasonable to assume that areas that exhibited a similar color scale were also delaminated; and (4) adding the known voided areas if any (clearly visible in the thermal images) and the estimated delaminated areas yielded the total delaminated areas, as shown in Table 6. The actual void sizes are also presented in this table. Control sample 1 did not have any surface defects; however, the thermograph showed a small amount of delaminated area, possibly due to some inadvertent CFRP separation induced during application.

It is also noticed that estimated delamination areas are larger than the actual voided areas by around 24% on average. This could be due to the relative scaling of beam surfaces in CAD and also the relative square nature of the thermograph in relation to the rectangular beam surface.

The percentage delamination of the total CFRP-applied area (the sample plan dimensions 91,400 mm2) is shown in Table 6. It is clear that the delaminations are minor compared to the total CFRP area. For possible distributed defects such as overhead application, wet concrete surface, and unclean concrete surface (samples 7–11), the delamination is limited to a maximum of 2.1% of the CFRP area. ACI 440 [14] states that small delaminations less than 1300 mm2 each are permissible in wet FRP layup systems as long as the delaminated area is less than 5% of the total lamination area and there are no more than 10 such delaminations per 1 m2. Therefore, these defects are unlikely to cause any significant loss of CFRP-concrete bonding and flexural capacity contribution from the CFRP.

It is possible to detect the depth of the voids by heating and scanning the adjacent sides of the samples, but this was outside the scope of this study. This thermography method is economic and can be conveniently used in the field. The heating time will increase with an increase in the CFRP area to be examined.

4. Conclusions

The nondestructive evaluation (NDE) approach can be used effectively to detect and quantitatively evaluate the carbon fiber-reinforced polymer- (CFRP-) concrete interfacial bond quality and any debonding extent. The approach can assist in identifying induced defects during the CFRP installation process and environmental effects and could ultimately help in preventing premature debonding.(1)The effect of surface voids, surface wetness, surface cleanliness, and upward CFRP application on concrete surfaces can be conclusively detected through the infrared thermography process. Quantification of the resulting CFRP debonded areas can be conveniently achieved through simple digitization of the thermal images. The CFRP and epoxy types used in this study are quite common and represent several popular types available from other manufacturers. The study results may not be applicable to other CFRP/epoxy combinations, in wet layup systems and masonry structures.(2)Thermography may overpredict the amount of debonded CFRP areas by about 25% on average, possibly due to scaling errors between the thermograph and the sample surface.(3)The overall debonded CFRP areas due to the factors considered were negligible, in relation to the total applied CFRP areas. The maximum debonded CFRP area in any sample was about two percent of the total CFRP area. Therefore, the factors considered are unlikely to significantly affect the laminate performance or any CFRP contribution to the concrete member strength or confinement.(4)Narrow vertical surface concrete cracks can be clearly identified with the thermography procedure.(5)The ground-penetrating radar (GPR) approach may not be able to effectively detect any CFRP debonding due to surface wetness, surface cleanliness, and upward CFRP application. Only the surface voids were effectively detected herein. The major drawback of using GPR is that it requires skilled labor and the GPR system and equipment are expensive.(6)While both thermography and GPR approaches need skilled personnel to operate the equipment and interpret the results, the GPR equipment is much more expensive than the thermal imaging equipment. The need of using an additional concrete board over the CFRP during scanning can make the GPR approach undesirable, especially in an upward CFRP application situation. A combination of the two approaches may also be considered.

Data Availability

The data used to support the findings of this study are included within the article.

Conflicts of Interest

The authors hereby declare no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Acknowledgments

Support from the Civil Engineering Department in the University of Texas at Arlington for laboratory facilities and equipment is gratefully acknowledged. The study was performed under a contract from the Texas Department of Transportation.