Advanced ultrasonic non-destructive testing for damage detection on thick and curved composite elements for constructions

Abstract

In this work, the problem of non-destructive testing on composite components with complex shapes for civil constructions and transport infrastructures is analyzed. In such applications at the state of the art main challenges are related with the inspection of thick sandwiches with low density cores (below 80 kg/m³) and curved panels. After a review of suitable non-destructive testing techniques, an original set-up for low frequency (100 kHz) ultrasonic inspection is proposed, which combines different solutions in through-transmission mode. The set-up is based on a hybrid configuration coupling a contact emitting probe with a non-contact air-coupled receiver. The use of a contact probe in emission is necessary to have enough energy to analyze thick components with low density core. The contact between probe and surface is made small (spot of 1 mm) and smooth using a spherical cap to increase lateral resolution at low frequency and to allow scan on irregular surfaces sometimes present in curved parts. To improve understanding this cap has been tested here also with a single probe in pulse echo mode. The non-contact probe in reception allows a better inspection flexibility on curved and thick components, where pulse echo is not feasible at all. The system is mainly developed for inspection after production in an industrialized production process, where through-transmission testing is possible. The analysis of results on two different samples (one thick sandwich with low density 40 kg/m³, 50 mm thick PUR core and one curved laminate panel) shows that the proposed methods can efficiently inspect construction composites of complex shape with satisfactory signal-to-noise ratio (usually SNR > 15 dB) and lateral resolution (2–3 mm).

Keywords

Non-destructive testing composite material curved laminate thick sandwich civil structures air-coupled ultrasonic

Introduction

Several research (e.g. HP FUTURE-Bridge [1], Safefloor [2] and MEgawind [3]) on polymer reinforced fibers for construction applications have demonstrated that the use of these components is very promising regarding requirements, quality, technical and economical feasibility and the favorable impact of using this kind of composite-based components in terms of sustainability, safety and quality of life.

Polymer composites, also known as fibre-reinforced polymers (FRPs), consist of two or more distinct constituent materials intentionally combined to a desired set of properties. These are commonly used for strengthening existing structures in concrete and steel in civil and building construction, but also new construction structural FRP components such as pultruded shapes and sandwiches are in high demand.

The increase of FRP applications, and the need to better meet economical and societal constraints, require today a new industrialized, flexible production process of composite infrastructures components covering the whole supply chain. An European Project denominated New Industrialised Construction Process for Transport Infrastructures based on Polymer Composite Components, 2009-2013 (TRANS-IND) is working on this topic. Industrialization will allow to overcome many of the limitations that in the last years limited the spread of such materials in the construction field (e.g. cost). The basic idea is to have an off-site industrial production of components, which will be then assembled in very short time in the construction site.

Within this project, attention is put also on how to detect defects in new composite structures with innovative flexible and efficient methods. FRP composite components can develop subsurface defects (e.g., cracks, voids, delaminations, disbonds) during the manufacturing process, transportation, construction and/or during the service life, therefore a reliable inspection method is necessary to guarantee a secure application. For in-service evaluation of structural integrity, the current trend is to use embedded sensors (e.g. Fiber Bragg Grating) for Structural Health Monitoring (SHM) [4,5], but for the inspections after production and on the assembly site a non-destructive testing (NDT) technique must be developed and used.

Different techniques have been developed and applied to inspect FRP components in the aeronautic [6], automotive and wind turbine fields, as thermography, shearography, ultrasonic phased array, air-coupled ultrasonic etc. The next paragraph analyses advanced NDT solutions in this field. However, the application of many of these techniques on civil composite structures can be difficult, because of the relevant thicknesses and peculiar shapes. In this paper, the focus will be on the inspection of structural elements made of composites with complex shapes, e.g. sandwich panels with thick low-density (40 kg/m³) core or FRP-curved panels.

Analysis of NDT advanced solutions for composites

Over the past 20 years, several techniques have been developed for NDT on composites. Optical methods gradually appeared and are now being applied more and more to NDT. Infrared thermography and shearography are widely applied, as they are fast in use and analysis (as they are full-field methods), non-contact and can be exploited with different excitation techniques [6]. These methods have the potential of application on complex shapes, if the surface temperature or shear gradients due to the defects can be well distinguished from effects due to the complex sample shape like curvatures or edges (e.g. Reference [7]). However, an important limit of these techniques is the capability of investigating thick panels, as the surface effect of deep defects is no longer visible. For thermography, for example, this is connected to the effect of thermal penetration depth (that is function of the thermal conductivity, volumetric heat-capacity, density, thermal load) on the thermal contrast C, which is proportional to the temperature difference between defected and non-defected areas and has a trend inversely proportional to the cube of the depth z:

C \approx 1 / z^{3}

(1)

The thermal contrast is a complex parameter that is dependent on several variables. A full description of the measurement of C is given in Reference [8]. Several experiences (e.g. References [9 –11]) showed that thermography and shearography are usually able to inspect monolithic composite panels up to a depth of about 7–8 mm. This is not enough for several civil structures.

Another promising optical technique for damage detection on composites is based on Laser Vibrometry [12], but also in this case the inspection depth is quite limited because of the required excitation energy, making this technique mostly suitable for regular panels of aeronautic or automotive use. Laser Ultrasonics [13] have better performances and flexibility for industrial use, but available systems are extremely expensive.

Acoustic emission (AE) testing shows an attitude for real-time monitoring of complex structures particularly for precise damage location and assessment. There are studies that analyzed damage mechanisms in CFRP composites and also in pultruded materials [14]. This technique detects the stress waves generated in the materials due to deformation, crack initiation and growth, crack opening and closure, fiber breakage and delamination in composite materials. However, the structure can be inspected under working conditions or under an externally applied fatigue load, which is not the best solution for control in an industrial production cycle. In addition, damages must have features suitable to generate AEs (e.g. a limited delamination between skin and foam core could not be detected).

Ultrasonic techniques have a great potential to inspect very different materials and elements, using different configurations and frequencies. However, for CFRP-GFRP (panels with a mix of layers in carbon fiber reinforced polymer and layers with glass fiber reinforced polymer) or sandwich civil structures, thickness and shape complexity present problems also for these techniques.

To quickly and reliably inspect large areas and curved shapes of composite elements, ultrasonic phased arrays have been developed (e.g. References [15,16]). Usually, commercial phased arrays have a frequency over 400–500 kHz, as smaller frequencies require an extension of the size of each element of the array. The bigger size of the element implicates a decrease of the lateral resolution, a difficulty to realize particular shape of the probe and a difficulty to move on raw (not smooth and planar) surfaces, so the application of phased array on the considered civil structures is not easy.

In order to improve penetration at medium frequencies (as 400–500 kHz) the component should be inspected in a water tank, which is not convenient for these new industrialized production processes of large components. In fact, the inspection systems for large components (e.g. bridge beams) are technologically complex, expensive and usually not portable (need of big basin to submerge the element; robot to move the probes; hermetic probes, etc.). There are some very promising portable systems that work with light curved shapes (e.g. wheel probes [17], probes coupled with a moving water container to inspect wind turbine blades [18]), but low thickness layer are analyzed, so medium-high frequency (from 400 kHz to 10 MHz) are used. In addition, the scan is performed on slightly curved parts, so irregular surfaces or highly curved panels are difficult to be inspected.

Air-coupled ultrasonic [19] is a very attractive method because it avoids the disadvantages of the coupling media and is suitable to inspect complex profiles for their easiness of movement and orientation, even though precise alignment and positioning are required by dedicated robotic manipulators (e.g. Reference [20]). Furthermore, the large gap of acoustic impedance between solid and air produces a very strong signal amplitude loss (which can be very critical for a thick sandwich, as it will be discussed in Paragraph 4). In most cases, air-coupled ultrasounds have been used in Pitch&Catch mode for inspection of thin panels [21] or honeycomb sandwiches often for aeronautic applications [22,23]. However, very interesting applications have been presented in literature also on foam structures [24] and concrete structures [25], thus showing potentials for investigation on thick porous construction materials. In References [26,27] sandwich structures with core thickness up to 60 mm (with 80 kg/m³ density) and 120 mm (with 200 kg/m³ density) are inspected using 50 kHz and 120 kHz air-coupled probes, but it is reported that due to high attenuation the technique has much lower accuracy for the 80 kg/m³ core, which can be inspected only at lower thickness and at lower frequency with reduced lateral resolution. These non-contact applications are mostly feasible in through-transmission mode, so they can be difficult for on-site inspection (i.e. once the component is installed) but can be suitable for inspection at the production site.

Contact ultrasonic is potentially the better technique (pulse echo and trough transmission modes with liquid or gel coupling) for accurate inspections of high-thickness components [28]. Low-frequency probes are usually employed to inspect such elements, thus providing a limited spatial lateral resolution (>6–8 mm at 50 kHz): for small defects in complex panels the detection can be challenging and the damage extension reconstruction may not be accurate, so a compromise between penetration and resolution must be found to identify the proper frequency and set-up.

The proposed method

In order to exploit the high penetration of a contact method and the inspection flexibility of a non-contact approach, an innovative combination between contact and non-contact probes (hybrid configuration) is here proposed, Figure 1. The emitter contact probe allows to have a high ultrasound energy injected through the material and the non-contact receiver probe allows to have more degrees of freedom to follow up the panel profile, usable with a low weight receive-side manipulator without the need for a squirter water supply, heavy pneumatics or rollers for the contact.

The basic idea was derived from the configuration proposed in Reference [29], where a dry coupled 250 kHz roller-probe transmitter (see also Reference [17]) was coupled with a non-contact 250 kHz air-coupled receiver to inspect a 25-mm thick graphite-epoxy composite section. However, there was the need to improve this set-up to be specifically applicable to sandwich structures with very thick low-density cores (as those produced in the Trans-Ind industrialized process) and curved shapes presenting sometimes irregular surfaces on the curve part.

The dedicated measurement chain here proposed is composed of the following equipment:

Ultran-Group pulser and receiver for transmission mode:

Type of emitted ultrasonic signal: Burst, duration of 30 µs;

Max excitation voltage: 475 V pk-pk;

DPR300 pulser and receiver for pulse-echo mode:

Type of emitted ultrasonic signal: spike, duration 10–70 ns;

Max excitation voltage: 900 V pk;

Digitizer NI PCI 5122 acquisition board:

2 channels simultaneously sampled at 14-bit resolution;

100 MS/s real-time;

100 MHz bandwidth;

Piezoelectric 100 kHz contact probes (emission).

Piezoelectric 100 kHz non-contact probes (reception).

Scanning system (custom made).

The piezoelectric 100 kHz contact probe (38 mm active diameter) is an untuned transducer that provides heavily damped broadband (about 60 kHz) performance, so it is suitable for tests that require improved signal-to-noise in attenuating or scattering materials. The piezoelectric 100 kHz non-contact probe has a frequency of 100 kHz (25 mm active diameter) and, according to the manufacturer specifications [30], is characterized by a high transduction efficiency in air/gases in a band of resonance of about 40–50 kHz. It can also acquire very small signals (fluctuations in the order of few tenths of Pa in air [31]), however, to increase the signal-to-noise ratio (SNR) the application of pass-band filters for the received signal is here adopted. The probe was installed at about 25 mm from the sample surface.

The contact emitting probe has been modified to make a punctual contact with the surface of the inspected sample. In this way, it is possible to improve the contact between the transducer and the sample in curved areas and also to increase the spatial resolution. Starting from the analysis of state-of-the-art solutions for coupling devices with narrow contact area having cone [32], ball shape [33] or rubber dry-coupled [34] for flexible and fast inspection, a simple and light dedicated set-up was here developed suitable for a continuous scan by applying a spherical cap (Figure 2) in front of the probe. In order to increase ultrasonic wave transmission, the empty space of the cap was filled with water, thus avoiding squirter water. For this preliminary prototype, a thin (about 1 mm) plastic element was used to realize the cap, the thin element allowing small interference on the transmitted wave. The contact between cap and surface can be dry or better coupled with an oil film.

The proposed system, thanks to the simple equipment, is suitable for application both on the component production site and on the assembly site, if a proper robotic manipulator is used. On the production site, where the components are not installed, an inspection in through-transmission mode can be readily performed. Once the components are installed, pulse echo mode is generally easier, but not enough sensitive in this case. So the air-coupled probes allow a better flexibility for the through-transmission mode.

Concerning the scanning speed, pulse repetition frequency (PRF) is clearly reduced when air-coupled techniques are used due to the much lower ultrasonic speed in air (340 m/s) compared e.g. to water (1500 m/s). In this case, the PRF is 100 Hz, which allows to reach a scanning speed of 100 mm/s with 1 mm step (with no averages on the signal). This is relatively slow, but since it can be automated, it is still compatible with the production cycle.

The proposed hybrid method has been firstly tested on a calibration composite sample (Figure 3(b)). The sample material is representative of real construction components, it is realized with a mix of CFRP and GFRP with size 270 × 470 × 30 mm. Simulated defects at different depths have been designed and realized using 4 Teflon inserts (Figure 3(a)), later extracted to create air voids. An area of 100 × 370 mm has been inspected (yellow dotted line in Figure 3(b)).

The map in Figure 3(c) shows that in pulse echo mode without spherical cap it is possible to inspect all defects, their size and depth, respectively, at 5, 10, 15 and 20 mm from the surface using time of flight (TOF) measurement. This map can be used as reference indication of the actual defects. The C-scan performed with the hybrid configuration and spherical cap of Figure 3(d) confirms the accurate detection of the defects and of their size in the calibration test.

Test samples and measurement set-up

The investigated test samples are fully representative of real structural elements used in civil infrastructures. The tests performed on two of them are here reported, each of them having a specific inspection challenge.

The sample #1 in Figure 4 is a sandwich element with a very low density (40 kg/m³) of PUR core 50 mm thick and two skins of CFRP and GFRP with about 8 mm of thickness each. The core thickness decreases until the two skins curve and join in a monolithic panel, with a rough and irregular surface at the internal side of the curvature. In this case, the attention is focused on the inspection of the sandwich element and the core at its maximum thickness, which is impossible with techniques as thermography or shearography. This element is very difficult to be inspected also with ultrasonic techniques (in particular to detect delaminations between skins and core) because of the very high thickness, attenuation at the interfaces and irregular surfaces (here surface oscillations can reach some millimeters). The sample #2 is a laminate curve element with a mix of CFRP and GFRP fibers, the thickness is about 24 mm. In practice, this sample is used to evaluate if the proposed hybrid set-up for sample #1 can be used for all parts of the same complex component.

Theoretical sensitivity evaluation

The measured signal intensity I transmitted through a layer of material is related to the incident intensity I₀ according to the inverse exponential power law that is usually referred to as Beer–Lambert law:

I = I_{0} e^{- α x}

(2)

where x denotes the path length (i.e. the thickness) and α the attenuation coefficient (which is proportional to frequency and porosity [32]). In the PUR core layer (density 40 kg/m³) of sample #1, α > 80 dB/m (increasing with frequency) and x = 50 mm. Therefore, the choice of the emitting probe has to be done with the objective to maximize I₀ at low frequencies. As previously discussed in literature (e.g. Reference [25]), even at 50 kHz air-coupled probes have problems in measuring cores with 80 kg/m³ density. This would be feasible with contact probes, but these probes do not guarantee applicability on the curved irregular surfaces of the analysed structures.

A simple finite element analysis based on Comsol Multiphysics software was performed to evaluate the signal amplitude reaching the air-coupled probe after passing through the whole sample #1, so as to evaluate the required sensitivity. The used material parameters are reported in Table 1. The properties for the air domain were taken from literature, while those of the PUR core material were provided by the manufacturer. The values for the composite material were experimentally determined: the velocity values by means of lab ultrasound test with contact and the density by means of the buoyancy method. For the solution on the discrete FE grid 12 degrees of freedom are guaranteed on the critical wave (i.e. in air) for the propagation direction. Results (Figure 5) show that very tiny % of the input pressure is able to reach the receiving air-coupled probe.

Table 1.

Characteristic parameters for sandwich material

Material	Ultrasound velocity (m/s)	Density (kg/m³)	Impedance (kg/m²•s)
Air	340	1.29	415
Skin	2855	1803	5147128
PUR core	678	40	27120

In fact, if an ultrasonic wave is propagating from a medium with acoustic impedance Z₁ to another medium with acoustic impedance Z₂, the reflected signal pressure will be proportional to the difference between Z₁ and Z₂. This is quantified by the coefficients of reflection R and transmission T of the sound pressure, defined as:

R = \frac{P_{r}}{P_{i}} = (\frac{Z_{2} - Z_{1}}{Z_{2} + Z_{1}})

(3)

T = \frac{P_{t}}{P_{i}} = (\frac{2 Z_{2}}{Z_{2} + Z_{1}})

(4)

where P_t, P_r and P_i are sound pressures of the transmitted, reflected and incident waves, respectively. In the present case, the signal reaches the receiving air-coupled probe after three interfaces: skin-core, core-skin and skin-air. The transmitted amplitude pressure is highly attenuated between skin-core (about 1% passes through) and between skin-air (about 0.016% passes through), as a consequence only about 0.00034% of the input pressure reaches the receiving air-coupled probe. This model can be used to estimate, as a function of the material parameters, the needed sensitivity in reception or (once the receiving sensitivity and the desired SNR are defined) the necessary input pressure (or intensity). In the present case, the input pressure was estimated from velocity Laser Doppler Virbometry measurements on the emitting transducer membrane. Using this input, it was found that only a few tenths of Pascal of pressure fluctuation arrives at the non-contact probe, this posing severe constraints to the required sensitivity, which is 3 mV/Pa for the air-coupled probe here used in reception. Due to the low signal amplitude, an important role is also played by the hardware instrumentation for acquisition, as low noise and high gain are required. A quantitative analysis of the pressure field by the FE model is useful for determining in complex problems the real applicability of air-coupled probes (e.g. the non-contact probe in reception in the hybrid set-up) unpredictable by means of analytical formulas and also for the simulation of the entire inspection process.

Simulated defect in the sample

In the sample #1, two simulated defects have been produced by drilling, one defect on the skin (10 × 30 × 2.5 mm) and one defect on the core (10 × 30 × 3 mm), Figure 6. The skin defect starts at 6 mm depth from the planar surface and finish at 2 mm depth and the core defect is 22.5 mm depth from the planar surface.

In sample #2, a defect with size 11 × 28 × 2 mm has been realized in the curved part of the panel at about 12 mm from the top surface, Figure 7.

It is important to underline that the defects searched in composite civil structures are usually larger than the defects searched on aeronautic structures (from 1 mm down to irregular porosities). Civil structures are heavier and thicker and have to bear different loading cycles with different defect propagation histories. According to end-users specifications in the Trans-Ind project, minimum relevant dimension in the order of 10 mm have to be detected (also on the core), but the shape of such panels is often very difficult to be inspected. However, a lateral resolution of few millimeters is required to have a good description of the defect shape so as to allow local repair to be performed and checked. This makes difficult the use of probes at frequencies well below 100 kHz. The defects simulated on the samples here analyzed are coherent with these specifications.

Measurement set-up

On the sandwich element #1, different configurations were tested (Figure 8): pulse echo and through-transmission mode, with spherical cap and not. Then the configuration with spherical cap in through-transmission mode was tested also on sample # 2 to verify applicability on the curved shape

Analysis of results on the thick sandwich structures (Sample #1)

Sandwich elements are frequently used for high-dimension components as they allow combining low weight, high strength and good dynamic properties. NDT inspection in these cases is very challenging but important to ensure continued structural integrity. The defects that are normally found on these elements are delamination, inclusion, crack in the skin, but also void in the core.

In this section, the results achieved on the composite sandwich #1, using the proposed configurations as in Figure 8, will be discussed. The ultrasound inspection was realized with a custom-made linear scanning system. The Cartesian robot moves automatically the ultrasound probes on the samples with a velocity of 100 mm/s. The inspection area, on the samples #1, was 50 × 74 mm (see Figure 9). Measurements are taken with 1 mm resolution.

Tests with two 100 kHz air-coupled probes in transmission mode showed that the transmitted energy was not enough to inspect the component. Comparative tests were also performed with active pulse Infrared Thermography (S40 Flir Camera and four 1 kW Infrared lamp) and air-coupled Pitch & Catch ultrasound configuration (100 kHz air-coupled probes, same scanning system as above). The results in Figure 10(b) and (c) show that only the skin defect until about 4 mm deep from the surface can be detected for this sample.

Pulse echo mode

The C-scan maps obtained in pulse echo mode using the contact transducer with and without spherical cap applied on the upper skin are compared in Figure 10. The maps report the root mean square (RMS) signal amplitude, which is thus proportional to attenuation. A gain of 70 dB is used for the acquisition with and without spherical cap.

In both the maps, the presence of the defect on the skin (white area) is easily detectable; in fact, the signal amplitude is maximum in the defected area. It is interesting to note how in the right map of Figure 11 the defect is better defined with dimensions closer to real ones. This improvement is due to the addition of the spherical cap which allows to have narrow measurement area (<1 mm) with increased spatial resolution.

The artificial defects have been realized with a drill bit of 2.5 mm of diameter. Actually, an array of holes have been made (30 mm deep) to have a 10-mm size defect. The map with spherical cap allows to see that the direction of the holes is not the same. This shows that the achievable lateral resolution with spherical cap is at least in the order of 2.5 mm, as details of such extension can be described. The average SNR measured on the ultrasonic signal is about 20 dB without cap and 17 dB with the cap, this showing that with the cap, even if the lateral resolution is improved, some energy is lost and therefore the focusing device can be further improved.

However, Figure 11 shows that the defect in the skin is easily detectable, but the defect in the core is not visible at all.

Through-transmission mode

In Figure 12, the maps obtained on the same panel with the hybrid contact/non-contact through-transmission configuration are compared for the set-ups with and without the spherical cap (gain 70 dB).

In this configuration, where the signal amplitude at the receiver is attenuated in correspondence to the defects, both damages (core defect and skin defect) have been found with a fast and flexible scan. The size of the two defects are very similar between the maps with and without spherical cap. There is however a clear improvement in the map with spherical cap, where the edge effects are much less evident and artefacts are reduced, thanks to the small contact area in the emitting probe. Also, in this case it is confirmed that the SNR decreases with the spherical cap (from 19 dB without cap to 13 dB with cap), but it is still good enough to detect defects both in the skin and the core.

Analysis of results on the laminate curved panel (Sample #2)

In this section, the results achieved on the laminate curved panel #2, using the proposed configuration as in Figure 7, are shown. In new bridge construction, the FRP composite panels (laminate) may be used in the entire structure, or they could be used as structural or reinforcing components. It is possible to easily realize elements with different shapes combining different material and thicknesses to have the required properties. Comparative inspections by Infrared thermography and shearography were not able to detect the defects, as positioned too deep (about 12 mm) with respect to the surface.

The inspection was performed in through-transmission mode with a step resolution of 1 mm; the inspected area was 30 × 40 mm. This component could be easily inspected with roller dry-coupled or phased-array probes, but the aim is to assess suitability for application on curved components with the same approach used for the thick sandwich, so as to have a unique flexible inspection system.

The scans were performed in semiautomatic mode since the used system allows only two degrees of freedom in a plane (along x and y axis, see Figures 10(e) and 13) so, after each scan line along the y axis, the transducer was manually moved along z and rotated to keep perpendicularity between the probe and the sample surface. A calibrated pre-loaded spring on the probe seat can be easily used to keep the contact pressure constant. Figure 14 shows the plot of both RMS amplitude and TOF, where the defect is clearly well detected, with the TOF the defect seems even more definite with improved lateral resolution. In this case, the average SNR increases up to about 25 dB.

Conclusion

In this work, the problem of inspecting structural integrity of composite components with complex shapes for civil construction and infrastructures has been analyzed. For complex shape it refers to sandwiches with very thick and low-density core (down to 40 kg/m³) and the curved panels. These elements are very challenging for NDT at the state of the art.

After a review of available techniques, an original set-up for low frequency (100 kHz) ultrasonic inspection is proposed, which is based on a through-transmission hybrid configuration coupling a contact emitting probe with a non-contact air-coupled receiver. The use of a contact probe in emission is necessary to have enough energy to analyze thick components. The contact between probe and surface has been reduced and made smooth using a spherical cap, which was here tested also with a single probe in pulse echo mode. The non-contact probe in reception allows a better inspection flexibility on curved components.

The analysis of results, summarized in Table 2, shows that the proposed method can efficiently inspect construction composites of complex shape with satisfactory SNR and lateral resolution (2–3 mm).

Table 2.

Summary of achieved results (SNR measured on ultrasonic signal)

	Pulse echo mode	Pulse echo mode with spherical cap	Through-transmission mode	Through- transmission mode with spherical cap
Thick sandwich #1	Skin defect visible, poor reconstruction SNR = 20 dB	Skin defect visible with 2–3 mm lateral resolution SNR = 17 dB	Both skin and core defects visible with artifacts SNR = 19 dB	Both skin and core defects visible with very limited artifacts and 2–3 mm lateral resolution SNR = 13 dB
Curved panel #2	Not possible	Defect visible SNR = 18 dB	Not possible	Defect visible, increased lateral resolution SNR = 25 dB

SNR: signal-to-noise ratio.

The system is mainly developed for inspection after production in an industrialized production process, where through-transmission testing is feasible. The results put the basis for the future development of automatic inspection procedures for large composite parts with complex shapes in an industrial environment. The scanning system can be automated through the use of an anthropomorphic robot.

Footnotes

Funding

This work is partially funded by the FP7-NMP-2008-LARGE-2 European Project Trans-IND (New Industrialised Construction Process for Transport Infrastructures Based on Polymer Composite Components 2009–2013).

Acknowledgement

The authors acknowledge the partner ACCIONA Infrastructures for having provided the measurement samples.

References

HP FUTURE-Bridge. High-performance (cost competitive, long life and low maintenance) composite bridges for rapid infrastructure renewal. European 6th RTD framework programme. October 2006–September 2009. Project reference: FP6 TST5-CT-2006-031522.

SafeFloor. Low risk and totally recyclable structural buildings. European 5th RTD framework programme. February 2001–September 2004. Project reference: FP5 EVK4-CT 2000-00020.

MEgawind. Development of a MW scale wind turbine for high wind complex terrain sites. European 5th RTD framework programme. January 2001–June 2005. Project reference: FP5 ENK5-CT-2000-00328.

Guo

Xiao

Mrad

. Fiber optic sensors for structural health monitoring of air platforms. Sensors 2011; 11: 3687–3705.

Chang C. Fiber-optic smart structures for monitoring and managing the health of transportation infrastructures, final report METRANS Project 09-13. California State University, Long Beach, 2010.

Garnier

Pastor

Eyma

. The detection of aeronautical defects in situ on composite structures using non-destructive testing. Compos Struct 2011; 93: 1328–1336.

Mayr

Dietermayr

Hendorfer

. Characterization of defects in curved CFRP samples using pulsed thermography and 3D finite element simulation. 2008. Quirt 2008, 9th international conference on quantitative infra red thermography. Krakow, Poland, 2–5 July. p. 8.

Maldague

XPV

. Theory and practice of infrared technology for non-destructive testing, New York, NY: John Wiley, 2001.

Krell

Brandenburg

Lauterborn

. Comparative investigation of pulse thermographic and shearography testing of composite materials. ECNDT 2006. Berlin. 25–29 September 2001. Tu4.6.1.

10.

Hofmann

Pandarese

Revel

. Optimization of the excitation and measurement procedures in non-destructive testing using shearography. Rev Sci Instrum 2008; 79(11): 115105–115105.

11.

Revel

Rocchi

. Defect detection in ceramic material by quantitative infrared thermography. 2006. QIRT 2006. Padova, Italy, 28–30 June. idn 003.

12.

Castellini

Revel

. Damage detection by laser vibration measurements. 2000. 15th world conference on non-destructive testing. Roma, Italy, 15–21 October. idn 709.

13.

Scruby

Drain

. Laser ultrasonics, Bristol, Philadelphia, PA, and New York, NY: Adam Hilger, 1990.

14.

Crivelli

Guagliano

Marangoni

. Fatigue damage analysis of pultruded glass fiber reinforced materials with acoustic emission methods. 2010. EWGAE 2010. Vienna, 8–10 September. pp. 1–8.

15.

Meier R, Becker J and Rehfeld T. Method for the ultrasound testing of a workpiece whiting a curved region of its surface and device suitable for the execution of the process. United States Patent no US 7,516,664 B2, USA, 2009.

16.

Habermehl

Lamarre

. Ultrasonic phased array tools for composite inspection during maintenance and manufacturing. 17th world conference on nondestructive testing. Shanghai, China. 25–28 October 2008. pp. 832–839.

17.

Sonatest website. Wheel probes, http://www.sonatest.com/products/range/transducers/wheelprobes/ (accessed 3 August 2012)..

18.

Jasiuniene

Raišutis

Šliteris

. Ultrasonic NDT of wind turbine blades using contact pulse-echo immersion testing with moving water container. Ultragarsas (Ultrasound) 2008; 63(3): 7–11.

19.

Hillger

Meier

Henrich

. Inspection of CFRP components by ultrasonics imaging with air coupling. Vol. 7. 2002. 8th ECNDT. Barcelona, 17–21 June.

20.

Cotter

Michaels

. Path controlled contour following for ultrasonic imaging of complex shaped composite components. 2000. 15th world conference on non-destructive testing. Roma, Italy, 15–21 October. Idn 216.

21.

Ramadas

Balasubramaniam

Joshi

. Characterisation of rectangular type delaminations in composite laminates through B- and D-scan images generated using Lamb waves. NDT&E Int 2011; 44: 281–289.

22.

Raijsutis

Kazjys

Zukauskas

. Ultrasonic air-coupled testing of square-shape CFRP composite rods by means of guided waves. NDT&E Int 2011; 44(7): 645–654.

23.

Pietroni

Revel

. Non contact ultrasonic techniques for composite material diagnostics in aeronautics applications. ECNDT 2006. Berlin, 2006. We.1.6.2.

24.

Strycek

Loertscher

. Ultrasonic air-coupled inspection of advanced material. e-Journal of Nondestructive Testing & Ultrasonics 1999; 4(11).

25.

Hillger

. Imaging of defects in concrete components with non-contact ultrasonic testing. 2004. 16th world conference on NDT. Montreal, Canada, 30 August–3 September. idn 720.

26.

Borum

Evaluation of sandwich materials using ultrasonic air-coupled scanning technique. In: Thomsen

Bozhevolnaya

Lyckegaard

(eds). Sandwich structures 7: advancing with sandwich structures and materials. Springer, Dordrecht (Netherland), 2005; 805–813.

27.

Borum

Berggreen

. Examination of sandwich materials using ultrasonic air-coupled scanning technique. 2004. 16th world conference on NDT. Montreal, Canada, 30 August–3 September. idn 710.

28.

Thomas

Automated ultrasonic inspection of large-scale sandwich structures. In: Thomsen

Bozhevolnaya

Lyckegaard

(eds). Sandwich structures 7: advancing with sandwich structures and materials. Springer, Dordrecht (Netherland), 2005; 795–804.

29.

Cotter

Michaels

. Squiter, roller probe, and air-coupled ultrasonic transducer techniques for low frequency inspection of advanced composite materials. 15th world conference on non-destructive testing. Roma, Italy. 15–21 October 2000, Idn:215.

30.

Ultran Group website, http://www.ultrangroup.com/Transducers/Catalogs/NCU.pdf (accessed 3 August 2012).

31.

Pietroni

Revel

. Development of a methodology for the characterisation of air-coupled ultrasound probes. 9th international conference on vibration measurements by laser techniques. Ancona. 22–25 June 2010. pp. 106–115.

32.

Sandoz

. Ultrasonic solid wood evaluation in industrial applications. 1996. 10th international symposium on nondestructive testing of wood. Lausanne, CH, 26–28 September. p. 147.

33.

Ginzel

. The ballprobe. e-Journal of Nondestructive Testing & Ultrasonics 1996; 1(2).

34.

Robinsona

Drinkwater

Allin

. Dry-coupled low-frequency ultrasonic wheel probes: application to adhesive bond inspection. NDT&E Int 2003; 36: 27–36.