Detection of delamination in laminated CFRP composites using eddy current testing: Simulation and experimental study

Abstract

The applicability of conventional eddy current (EC) testing to detect delamination in laminates of carbon fiber reinforced polymers (CFRP) was assessed. As delamination impedes the induced currents flowing across the layers, the EC technique offers the potential for delamination in CFRPs to be detected by sensing the change in probe output signals caused by distortions in the magnetic field. In a numerical study, ECs distribution in CFRPs alters considerably in multidirectional and cross-ply laminates and currents drop steeply when delamination is present. In experiments, the detectability of delamination at different depths was investigated using 24-layer laminates, each with different stacking sequences. The change in amplitude of the probe output signal obtained from data agrees qualitatively with the numerical analysis. The size and location of the delamination defect were compared with those obtained in ultrasonic scanning images. Experiments show that delamination at a maximum depth of 0.75 mm in cross-ply samples could be detected, whereas subsurface defects in the unidirectional plate went undetected.

Keywords

Conventional eddy current testing carbon fiber reinforced polymer delamination detection

1. Introduction

Advanced carbon fiber reinforced polymer (CFRP) materials are of interest to numerous industries because of their superior mechanical properties and tailoring for complex structural applications. Laminated CFRPs, however, are particularly susceptible to low-energy impacts in engineering applications. As a result, internal damage in the form of fiber fractures, resin matrix cracks, and delamination may be induced because of poor properties in the through-thickness direction of laminates. Among the defects, delamination, which is a particular interlaminar mode of fracture and refers to the local separation of adjacent laminae, is the most critical failure mode as the strength and stiffness of the CFRPs may fall suddenly, and eventually result in unexpected failure of the CFRP structure, even when the defect is undetectable under visual inspection from the impact side. Therefore, detecting delamination is imperative in maintaining structural reliability and integrity of practical laminated CFRP structures.

The eddy current (EC) technique is a well-established nondestructive testing method for defect detection in electrically conductive materials. In accordance with Faraday’s law, under testing using a driver coil, an EC is induced in a CFRP plate because of electrically conductive fibers. Defects in CFRPs cause local changes in the EC path that result in a distorted magnetic field generating changes in the output signal of the pickup coil. Attributes such as simplicity and ease of operation, rapid noncontact scanning, online inspection and real-time imaging, and low surface-cleaning maintenance have made the EC method a quick and easy testing candidate in inspecting actual CFRP structures [1, 2]. Different types of defect including fiber breakage and cracks [3], missing or misaligned fiber bundles [1, 4, 5], fibers waviness [6, 7, 8] and impact damages [9, 10] have been detected under EC testing. However, to the best of the authors’ knowledge, little attention has been paid to detect delamination of CFRPs; there has been only a few experimental results on monitoring delamination growth caused by tension loads [11]. Controversially, another study [12] has claimed that the conventional EC-based technique lacks the requisite sensitivity to detect internal delamination in laminated CFRP composites.

Therefore, under scrutiny here is whether the conventional EC technique is capable of detecting delamination in laminated CFRP composites. First, an optimal working frequency of the EC probe is determined such that the pickup coil produces its largest fractional change in voltage. Second, the detectability of delamination defects using the method is investigated in numerical simulations and experiments. We analyze the influence of delamination on ECs induced in laminated CFRPs and also changes in amplitude of the probe’s output signal. In the experiments, different laminated CFRP samples with artificially induced delamination are tested to verify the reliability of the EC detection approach. Finally, concluding remarks are drawn.

2. EC probe characterization

The probe is one of the more critical parts of EC technique; its sensor provides the high sensitivity when currents are strongly altered by discontinuities caused by cracks. Coil structures are the most widely used type of sensors in EC probe for detecting the wide variety of defects. In general, the skin effect limits the effective detection depth to the near surface. The research of Reimche et al. [13] and Mook et al. [14] showed that the non-axial transmit-receive (TR) probe (Fig. 1a) offers potential enhancements in inspection depth by optimizing the distance between its two coils. Moreover, in imaging applications, the point spread function (PSF), which describes the imaging response to a point input, of the EC probe is of great interest as it plays a critical role in EC imaging. Compared with coaxial EC probes, the PSF of TR-type probes has a Mexican-hat shape when it passes through a point-like defect (Fig. 1b and c) and is more suitable for imaging. For this reason, the TR-type probe is employed in this study.

Figure 1.

(a) Configuration of transmitter-receiver (TR) probe; (b) EC probe passing through a point-like defect; (c) PSF of coaxial and non-axial coils.

In practice, conventional EC testing operates either at a low frequency or at high frequency. However, the sensitivity of the probe output voltage at low frequency is not sufficient for CFRP testing because the electrical conductivity is low. At higher frequency, the probe is more susceptible to liftoff, tilt, and surface roughness, causing higher background noise. Thus exploring an effective alternative for maximizing the probe sensitivity is important. Reports [15, 16, 17] prove that shifting the electrical resonance frequency is potentially a complementary approach to maximize the inspection sensitivity without needing complex probe designs or data analysis.

According to circuit theory, an EC probe can be simply modeled as a singular lumped element with an effective capacitance and inductance; its inherent frequency ( $f_{0}$ ) is then simply expressed as [15, 16]

$\displaystyle f_{0}=\frac{1}{2\pi\sqrt{L_{0}C_{0}}},$ (1)

where $L_{0}$ and $C_{0}$ denote the equivalent inductance and capacitance, respectively. However, the probe couples electromagnetically to the ECs induced in the conductive non-ferromagnetic material when it is close to a conductor. Hence, the coupling changes the electrical properties of the probe including the resonance frequency of the system,

$\displaystyle f_{1}=\frac{1}{2\pi\sqrt{L_{0}C_{0}\left({1-\frac{k^{2}}{2}}% \right)}},$ (2)

where $k$ is the coupling coefficient between EC coil and conductor. While the resonance frequency $f_{1}$ of the transformer circuit clearly depends on $k$ , its value strongly depends on many parameters of the coupled system, for instance, electric conductivity, magnetic permeability, frequency, liftoff, and tilt. As a result, a slight change in any of the parameters values may lead to a change in coupling and shift the resonance frequency.

Using an impedance analyzer (Agilent 4294A), an experiment was performed to investigate both the inherent and resonance frequencies (i.e., $f_{0}$ and $f_{1}$ ) of the TR-type probe (Fig. 1a) in different environments, in which the resonance frequency was employed in subsequent experiments. An air-cored pancake coil was connected to the analyzer via a coaxial cable of 1-m length; the height, and inner and outer diameters of the coil are 0.8 mm, 1.2 mm, and 3.2 mm, respectively. The probe has a characteristic resistance of $R_{0}=$ 10.9 $\Omega$ , an inductance of $L_{0}=$ 32 $\pm$ 0.5 $\mu$ H, and the coaxial cable has capacitance of $C_{0}=$ 109 pF/m.

From the impedance-frequency curves ( $Z(f)$ ) of the probe (Fig. 2a) for both off (in air) and on the undamaged conductive materials (Aluminum and CFRP) show an obvious upward shift in the electrical resonance frequency occurring when the probe is brought close to the aluminum plate. However, a slight shift in resonance frequency is observed when the probe is brought near the CFRP sample. This is because of the lower electrical conductivity of the CFRP, which results in a much smaller coupling coefficient $k$ compared with the aluminum plate under the same condition. The inherent and resonance frequencies ( $f_{0}$ and $f_{1}$ ) for each $Z(f)$ profile are listed in Table 1. The inherent frequency $f_{0}$ is about 2.69 MHz, the resonance frequencies $f_{1}$ are about 2.91 MHz (above aluminum plate) and 2.7 MHz (above CFRP sample), respectively.

Table 1

Resonance frequency, signal peak frequency and voltage of the probe both off and on the undamaged conducting materials

Material	Resonance frequency	Signal peak frequency	Voltage signal at	Voltage signal at
	$f_{0}$ and $f_{1}$ ( $\pm$ 0.01 MHz)	$f_{\textit{NERSE}}$ ( $\pm$ 0.01 MHz)	$f_{0}$ and $f_{1}$ (mV)	$f_{\textit{NERSE}}$ (mV)
Air	2.69	2.46	48	172
Aluminum	2.91	2.65	22	63.2
CFRP	2.70	2.48	59.2	184

Figure 2.

(a) Impedance-frequency curves of the TR probe in air and above two different conductive materials; (b) Resonance-shift of the output voltage of the TR probe on CFRP (solid line) and Aluminum samples (dash line).

Figure 3.

Modeling of interior delamination: (a) electric insulation element as delamination; (b) section profile of the model.

To establish the appropriate frequency that improves the detection sensitivity of the conventional industrial EC technique, a frequency sweep method is applied by driving the probe with a sinusoidal voltage $\pm$ 0.5 V through frequencies from 0 to 4 MHz. The amplitude of the obtained voltage at each frequency was recorded and plotted (Fig. 2b). Here, impedance 1 and voltage 1 correspond to the probe being above the CFRP sample, and impedance 2 and voltage 2 correspond to the probe being on the aluminum plate. From the results, it is obvious that the output voltage signal of the probe reaches a local maximum at a frequency $f_{\textit{NERSE}}$ near the electrical resonance. The useful signal then is either submerged in the electrical background noise or suffers from higher levels of background noise when operating the probe at frequencies dissimilar from $f_{\textit{NERSE}}$ . Therefore, it is quite effective for enhancing inspection sensitivity of the conventional EC method by operating the probe at the frequencies approaching resonance. The peak frequencies of the probe signal obtained from different materials and their corresponding voltages are listed in Table 1. Clearly, the voltage at $f_{\textit{NERSE}}$ is larger than that at the impedance resonance, and the peak frequency of the probe signal above the CFRP plate is about 2.48 MHz.

3. Numerical analysis for delamination

The EC-based method has been successfully used to detect defects in laminated CFRP materials. Previous studies demonstrated that ECs flow along fibers and at fiber-fiber contact points pass from one to the other. Once delamination occurs, the network of fiber contacts between adjacent laminae breaks. Breakages results in discontinued currents and ultimately affects its conductive property. To investigate the influence of delamination, finite-element (FE) numerical simulations and analyses of the electrical properties of CFRP materials are first presented and detailed below. The FE model configured COMSOL Multiphysics is presented in Fig. 3.

3.1 Numerical model

Three plate models (100 mm $\times$ 100 mm $\times$ 1 mm), labeled M1–M3, composed of 8 thin layers with each of ply thickness of 0.125 mm, were configured; stacking sequences and delamination areas are listed in Table 2. For the probe, an air-cored pancake coil was configured with a 0.5 mm lift-off (from the top surface of the plate to the bottom surface of the coil). The height and the inner and outer diameters of the coil are 0.8 mm, 1.2 mm, and 3.2 mm, respectively, and the excitation frequency is 2.48 MHz.

Table 2
Stacking sequence of samples

Models	Stacking sequences	Delamination area
M1	0 ${}^{\circ}$ ${}_{8}$	20 mm $\times$ 20 mm
M2	$-$ 45 ${}^{\circ}$ /0 ${}^{\circ}$ /45 ${}^{\circ}$ /90 ${}^{\circ}$ / $-$ 45 ${}^{\circ}$ /0 ${}^{\circ}$ /45 ${}^{\circ}$ /90 ${}^{\circ}$
M3	0 ${}^{\circ}$ /90 ${}^{\circ}$ /0 ${}^{\circ}$ /90 ${}^{\circ}$ /0 ${}^{\circ}$ /90 ${}^{\circ}$ /0 ${}^{\circ}$ /90 ${}^{\circ}$

Table 3

Contact resistivity for various composites [20]

Contact resistivity $\rho_{co}$ ( $\Omega$ $\cdot$ m ${}^{2}$ )
$\rho_{co}$ in laminates [0/0]	9.28 $\times$ 10 ${}^{-5}$
$\rho_{co}$ in laminates [0/45]	5.60 $\times$ 10 ${}^{-5}$
$\rho_{co}$ in laminates [0/90]	3.52 $\times$ 10 ${}^{-5}$

Using the magnetic vector potential $\bm{A}$ and the electric scalar potential $\phi$ , the basic governing equation is written

$\displaystyle j\omega\sigma\bm{A}+\mu^{-1}\nabla\times\nabla\times\bm{A}+% \sigma\nabla\phi=\bm{J}_{e},$ (3)

where $\omega$ and $\sigma$ are the angular frequency and conductivity tensor, and $\varepsilon$ and $\mu$ are the permittivity and permeability. The vector $\bm{J}_{e}$ is the external current density driving the coil. With $\bm{n}$ the unit normal vector on the external boundary of the domain, the condition $\bm{n}\times\bm{A}=0$ is imposed. Each lamina of the FE model was simulated individually and regarded as a homogeneous and anisotropic sheet, and the conductivity tensor has only nonzero diagonal elements (29940, 4.8, 1.1) S/m. Also, when the fibers are oriented at some arbitrary angle with respect to the $x$ -axis, a rotation matrix is used [18] to obtain the corresponding conductivity tensor.

Generally, delamination – the separation of two adjacent lamina – is a planar defect occurring within the CFRPs. Hence, to model the delamination, an electric insulation element is inserted between the adjacent layers with the proviso that

$\displaystyle\bm{n}\cdot\bm{J}=0,$ (4)

where $\bm{n}$ is the unit normal vector, and $\bm{J}$ is the EC density. The remaining interfaces are modeled using a surface contact resistivity [19]. The values of the contact resistivity, which are related to the fiber directions on the two sides of the interfaces, are given in Table 3 [20].

Figure 4.

Numerical results of the $x$ , $y$ and $z$ -directional components of currents, ( $J_{x}$ , $J_{y}$ , $J_{z}$ ), distributed in the 2nd interface. (a)–(c): on the top surface of the 3rd ply in sample M1 without delamination; (d)–(f): on the top surface of the 3rd ply in sample M1 with delamination.

Figure 5.

As for Fig. 4 but with sample M2.

Figure 6.

As for Fig. 4 but with sample M3.

3.2 Results and discussion

Numerical results of the EC densities obtained from plates M1–M3 are shown in Figs 4–6; the solid boxes mark the site of delamination. All current densities presented in the figures are from the 2nd interface between the 2nd and 3rd laminae. Note that the EC density distributions from the three samples are very different from each other.

The directional components of the current $J_{x}$ , $J_{y}$ and $J_{z}$ in sample M1 without and with delamination (Fig. 4a–f, respectively) clearly show almost no difference between current densities $J_{x}$ and $J_{y}$ , where the two components are stretched in the fiber direction and compressed in the transverse direction. The current distribution path appears as a narrow ellipse because the ratio $\sigma_{0}/\sigma_{90}$ is higher. A noticeable square hole of similar size as the delamination site is observed (Fig. 4f), indicating that the delamination impedes the current flowing across the interface. The lower value of $J_{z}$ corresponds to the lower value of $\sigma_{cp}$ , and undergoes dual variations in the $x$ and $y$ directions. Additionally, little change is found in the amplitude of the current densities, which results in little influence on the electrical conductivity of M1, as clearly observed from the voltage curves (Fig. 8).

Distributions of $J_{x}$ , $J_{y}$ and $J_{z}$ for samples M2 and M3 are presented (Figs 5 and 6, respectively). With no delamination, the simulation results demonstrate that the current distribution is sensitive to the difference in direction of the fibers on the opposing sides of the interface. The currents are stretched either in the two directions, 0 ${}^{\circ}$ and 45 ${}^{\circ}$ or in 0 ${}^{\circ}$ and 90 ${}^{\circ}$ (Figs 5a–c and 6a–c). With delamination, the interlaminar damage causes a current repartition and its domain depends on the shape of the delamination area (Figs 5d–f and 6d–f; marked by open squares). The current density components $J_{x}$ and $J_{y}$ in the delamination area are stretched in only one direction around the interface, and not in two directions as found with no delamination (Figs 5a–c and 6a–c).

Current repartition in the multidirectional composites M2 and M3 supports the notion that interlaminar damage has a significant influence on current densities near the interface between two laminae with different fiber directions. This is because the fibers of such laminae tend to press against each other during curing, causing a high concentration of current density around the interfaces (see Fig. 7a). In contrast, at the layer-layer interface of sample M1, fibers from one simply sank into the fibers of the other, and hence little current is found near the interface. In other words, the number of contact points between fibers at an interface is less for 0 ${}^{\circ}$ /0 ${}^{\circ}$ -ply than for 0 ${}^{\circ}$ /45 ${}^{\circ}$ -ply or 0 ${}^{\circ}$ /90 ${}^{\circ}$ -ply. Once an interlaminar defect occurs, the fiber-contact network between adjacent laminae is destroyed, and hence the induced currents cannot flow across the interface and interact with each other, implying that a net-zero current flows through the boundary forcing the normal component of the current to zero. As a result, the two components $J_{x}$ and $J_{y}$ induced in samples M2 and M3 experience a steep drop (see Figs 5f and 6f) and a square hole appears in the contour plot of the $J_{z}$ distribution (see Figs 4f, 5f, and 6f).

Figure 7.

Current density $J_{x}$ (extracted at $x=$ 0 mm, $y=$ 1.3 mm) and $J_{y}$ (extracted at $x=$ 1.3 mm, $y=$ 0 mm) in the through-thickness direction of M3: (a) without delamination; (b) with delamination at the 2nd interface; (c) with delamination at the 4th interface. (The abscissa is the $z$ coordinate from the top surface of the sample to the bottom surface.)

From the foregoing numerical analyses, delamination does influence the ECs near the interface between two different plies, causing a repartition of the current path and a steep attenuation of the current density. In-plane current density profiles in the through-thickness direction of sample M3 without and with delamination at various depths (Fig. 7) provide a quantitative comparison. Note that the induced current is found to be highly concentrated around all the interfaces of M3 regardless of delamination. However, a significant decline in current density takes place at interior delamination sites (Fig. 7b and c, marked by open ellipses). This is consistent with the fact that delamination defects are capable of impeding current flowing across the interface, resulting in a sharp decrease in current density around the delamination site.

Moreover, the current density of the 1st lamina drops from its maximum value at the top surface (Fig. 7a–c), whereas a maximum value appears at the top surfaces in all other seven laminae. Here the fiber contact through the laminae enables current to flow across the interface and combine with other currents, leading to current concentrations around the interfaces. In contrast, delamination impedes currents flowing across the interface, and therefore provides no contribution to that of subsequent laminae. Hence the amplitude of the current density around the intact interface is larger than that around the delamination site.

To describe the extent of the influence of delamination, the voltage variation of the EC probe is numerically calculated. The $z$ component of the magnetic flux density $B_{z}$ is used to obtain the voltage. According to Faraday’s law, the induced voltage $V$ is expressed as

$\displaystyle V=j\omega N\int\!\!\!\int B_{z}dxdy,$ (5)

where $\omega$ is the angular frequency, and $N$ is the number of turns in the coil. To obtain the change in voltage, a FE model without delamination was established to obtain the voltage without delamination $V_{1}$ . Therefore, the change in voltage caused by delamination is then evaluated by subtracting $V_{1}$ from that with delamination $V_{2}$ ,

$\displaystyle\Delta V=V_{2}-V_{1}.$ (6)

Figure 8.

Voltage changes obtained from samples M1–M3 of various plies with delamination at different depths.

The change in output voltage of the EC probe is plotted in Fig. 8. For the unidirectional plate M1, the output voltages obviously remain almost unchanged, whereas a significant voltage change is found when delamination occurs in multidirectional laminate M2 and cross-ply one M3. From the analysis above, current concentrations occur around the intact interfaces of the two laminates but attenuate steeply at delamination sites, resulting in a significant reduction in output voltage amplitudes. However, the influence of delamination on the ECs in the unidirectional sample is negligible, and hence a slight output voltage is obtained. The change in output voltage therefore reveals the delamination influences affecting the ECs when fiber directions on the opposing sides of the interface are different. Furthermore, the larger the angular difference is between adjacent laminae, the greater are the current density and the voltage change near the interface. Also, a shallower delamination depth produces stronger current density attenuations and larger voltage changes.

4. Experimental studies and discussion

Experiments were conducted to detect and evaluate artificially induced delamination damage in CFRP laminated materials of different plies. Meanwhile, size and location of the delamination were compared with those imaged by an acoustic emission device (C-scan, Physical Acoustics Corp. (PAC)) using a scanning frequency of 5 MHz.

Table 4
Parameters of fabricated CFRP samples

Samples	Stacking sequences	Delamination area	Delamination depth
S1	0 ${}^{\circ}$ ${}_{24}$	20 mm $\times$ 20 mm	0.25 mm
S2	( $-$ 45 ${}^{\circ}$ /0 ${}^{\circ}$ /45 ${}^{\circ}$ /90 ${}^{\circ}$ ) ${}_{6}$
S3	(0 ${}^{\circ}$ /90 ${}^{\circ}$ ) ${}_{12}$
S4	(0 ${}^{\circ}$ /90 ${}^{\circ}$ ) ${}_{12}$		0.5 mm
S5	(0 ${}^{\circ}$ /90 ${}^{\circ}$ ) ${}_{12}$		0.75 mm

Figure 9.

Schematic showing the fabrication of a laminate with artificial delamination.

Figure 10.

(a) Photo of the EC experimental setup; (b) Schematic of the experimental setup.

Figure 11.

Voltage signals obtained from the different CFRPs with and without inside delamination.

Figure 12.

(a) Eddy current scanning image of sample S1; (b) Ultrasonic C-scanning image of sample S1.

Figure 13.

As for Fig. 12 but with sample S2.

Figure 14.

As for Fig. 12 but with sample S3.

Figure 15.

Eddy current scanning images of (a) sample S4, (b) sample S5 when spacing $S=$ 0, (c) sample S5 when spacing $S=$ 10 mm; (d) T-R probe selects the deep penetration field trajectories.

4.1 Materials

Five hand-made samples, marked as S1–S5 (Table 4) were fabricated by laminating 24-layer tailored unidirectional prepregs in different directions 0 ${}^{\circ}$ , $\pm$ 45 ${}^{\circ}$ and 90 ${}^{\circ}$ (Fig. 9). Insulation sheets (20 mm $\times$ 20 mm) were inserted between adjacent sheets to simulate the delamination. The sheets was inserted at designated interfaces (0.25 mm depth in samples S1–S3, 0.5 mm depth in sample S4, and 0.75 mm depth in sample S5) before curing. The laminates were then placed between two heating platens of a hot press, and the temperature of the platens was increased from room temperature to 80 ${}^{\circ}$ C for 30 minutes under a pressure of 2 MPa. Afterwards, they were cured at 130 ${}^{\circ}$ C for two hours under a pressure of 12 MPa and subsequently cooled to room temperature automatically. The cured samples have a dimension of $L\times W\times H=$ 300 mm $\times$ 300 mm $\times$ 3 mm.

4.2 ECT system

The EC-based testing system for scanning and imaging of CFRP materials was developed in the authors’ laboratory (Fig. 10). In the system, a TR-type probe (see Section 2, Fig. 1a) is fixed to a flexible supporting bar for inducing and sensing ECs; the driving frequency is the obtained frequency $f_{\textit{NERSE}}$ of 2.48 MHz. The input of the probe is connected to an arbitrary waveform generator (33220A, Agilent) via a coaxial cable, whereas the output of the probe is connected to a high-frequency lock-in amplifier (SR844, Stanford Research Systems). The CFRP samples were placed between the EC probe and bottom plate. A stepping motor in the $x-y$ plane and its controller provide the adjustments and control of the probe movement. For data acquisition, a digitizer (PCI-6251, National Instruments) samples the signal and PC-based LabVIEW software was developed to produce the analysis. The scanning origin is located on the intact zone of the samples, and the zero-calibration of the probe signal is made by automatically adjusting the output offset before each measurement.

Even if the EC probe operates at frequencies close to resonance, the obtained voltage signal is still small (Table 1). To extract the useful weak signal, the lock-in amplifier is used as the processing unit, enabling signal measurements to be taken with high precision and inspections of the small defects.

4.3 Results and discussion

Figure 11 shows the profiles of the output voltage signal of EC probe obtained from different samples with and without inserted delamination. The voltage signals are quite distinct. A significant voltage variation can be observed from both samples S2 and S3 in the simulated delamination zone, whereas a very small variation is observed between sample S1 with a delamination defect and S2 without delamination. For the unidirectional sample S1, delamination produces little influence on ECs around the interface because fibers are aligned (Fig. 4a–f), and thus the obtained change in voltage by the probe is almost zero. The experimentally obtained nonzero voltage is due to some inevitable experimental uncertainties, for example, from the periodic fiber structures, the fabrication of the test sample, liftoff, and tilt of the probe. According to the heterogeneous properties of the CFRP materials, the periodic fluctuation signal is a response from the periodic fibers structure, which is caused by fluctuations in conductivity of the carbon fibers and resin matrix.

Figures 12–15 show the images of the artificial delamination defect in the different laminated CFRP samples using the EC approach and ultrasonic C-scan experiments; the plots give the voltage signal recorded at each scanning point. The scanning distances are in the plane of the laminate (Fig. 3, the $x$ - and $y$ -directions). The delamination defect at depth 0.25 mm cannot be detected in the unidirectional laminate S1 from the EC experimental image (Fig. 12a). However, both shape and location of this delamination in S1 is clearly evident from the C-scan image (Fig. 12b). The experimental result indicates that the EC method lacks insensitivity to interface delamination in unidirectional CFRP laminates because the current concentration around the interfaces is too small (Fig. 4a–f).

Experimental images obtained from the multidirectional and cross-ply laminates (Figs 13 and 14) at depth 0.25 mm clearly show the defects (marked by open squares). The actual size and location of all defects are accurately reflected in the images and are in good agreement with those in the C-scan images. The conventional EC approach, therefore, looks promising in the detection of delamination defects in multidirectional and cross-ply laminates, in which fiber directions of adjacent laminae are different. This is because delamination results in a distortion of the EC distribution near the interfaces between the two laminae (see Figs 5a–f and 6a–f).

From the experimental images of samples S4 and S5 obtained using the EC method (Fig. 15a–c), changes in the output signals become smaller with increasing delamination depth. The delamination defect at 0.5 mm depth in sample S4 is distinguishable (Fig. 15a) but not clearly identifiable in sample S5 (Fig. 15b) when the spacing ( $S$ ) between the transmitter and receiver coils of T-R probe is 0. As shown in Fig. 15d [16], the magnetic field of the transmitter coil penetrates and spreads out into the material. The receiver coil only picks up the flux penetrated deeply in the material when increasing coil spacing. Thus, the larger the coil spacing is, the deeper is the detectable flux lines (delamination at depth 0.75 mm of S5 is clearly evident), but the weaker the signal becomes. Therefore, in practical applications, one trades off coil spacing against weak signal amplitudes.

Hence, we concluded that the practicability of the conventional EC technique in detecting subsurface delamination defects is promising, and the inspection sensitivity is found to depend on fiber directions on the opposing sides of the interface.

5. Conclusion

The detection of delamination in laminated CFRP composites were investigated from numerical modeling and experiments using conventional EC testing. Characterization of the EC probe was done through a frequency sweep method for improving the detection sensitivity. Eddy current distributions in CFRP laminates with delamination defects was investigated using FE analyses. Fiber orientation on both sides of the delamination interface was found to have significant influence on the ECs and thus affected inspection sensitivity. For adjacent laminae with different fibers orientations, the current density distribution is altered in the delamination site undergoing a steep drop that causes changes in the output voltage of the pickup coil. For adjacent laminae with the same fiber orientation, however, the current density distribution is almost unaltered in the delamination site, resulting in little or no change in output voltage signal.

Finally, delamination detectability at different depths was investigated in experiments using a 24-layer laminate with different stacking sequences. The amplitude variations of the probe output signal obtained were qualitatively in agreement with numerical analyses. Size and location of the delamination defect were compared with those obtained from ultrasonic scanning images. The experiments showed that the subsurface delamination with a maximum depth of 0.75 mm in cross-ply CFRP could be detected by increasing the spacing between the two coils but subsurface defects in unidirectional plates were undetectable.

Footnotes

Acknowledgments

This work was supported by Key Project of National Natural Science Foundation of Jiangsu Province (No. BK20150061), the Fundamental Research Funds of the Central Universities (No. NE2015001), the Research Fund of State Key Laboratory of Mechanics and Control of Mechanical Structures (Nos. MCMS-0517K04 & MCMS-0517G01), the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD). In addition, this work was also partly supported by the JSPS Core-to-Core Program, A. Advanced Research Networks, and “International research core on smart layered materials and structures for energy saving”.

References

Heuer

Schulze

Pooch

and Petrenz

, Review on quality assurance along the cfrp value chain-non-destructive testing of fabrics, preforms and cfrp by hf radio wave techniques, Composites Part B: Engineering 77 (2015), 494–501.

Gerhard

Rolf

and Ole

, Non-destructive characterisation of carbon-fibre-reinforced plastics by means of eddy-currents, Composites Science and Technology 61 (2001), 865–873.

, Eddy current techniques for nor-destructive testing of carbon fiber reinforced plastic, Ph. D. Dissertation, University of Manchester, 2012.

Bardl

Nocke

Cherif

and Klein

, Automated detection of yarn orientation in 3d-draped carbon fiber fabrics and preforms from eddy current data, Composites Part B: Engineering 96 (2016), 312–324.

Schulze

M.H.

Heuer

Küttner

and Meyendorf

, High-resolution eddy current sensor system for quality assessment of carbon fiber materials, Microsystem Technologies 16 (2010), 791–797.

Mizukami

Mizutani

Kimura

Sato

and Todoroki

, Detection of in-plane fiber waviness in cross-ply cfrp laminates using layer selectable eddy current method, Composites Part A: Applied Science and Manufacturing 82 (2016), 108–118.

Mizukami

Mizutani

Kimura

Sato

Todoroki

Suzuki

and Nakamura

, Visualization and size estimation of fiber waviness in multidirectional cfrp laminates using eddy current imaging, Composites Part A: Applied Science and Manufacturing 90 (2016), 261–270.

Mizukami

Mizutani

Todoroki

and Suzuki

, Detection of in-plane and out-of-plane fiber waviness in unidirectional carbon fiber reinforced composites using eddy current testing, Composites Part B: Engineering 86 (2016), 84–94.

Y.Z.

Tian

G.Y.

Pan

M.C.

and Chen

D.X.

, Non-destructive testing of low-energy impact in cfrp laminates and interior defects in honeycomb sandwich using scanning pulsed eddy current, Composites Part B: Engineering 59 (2014), 196–203.

10.

Koyama

Hoshikawa

and Hirano

, Investigation of impact damage of carbon fiber reinforced plastic by eddy current nondestructive testing, in: Smart Materials Structures & NDT in Aerospace Conference, 2011.

11.

Gros

X.E.

and Takahashi

, Monitoring delamination growth in cfrp materials using eddy currents, Nondestructive Testing and Evaluation 15 (1998), 65–82.

12.

Cheng

and Tian

G.Y.

, Comparison of nondestructive testing methods on detection of delaminations in composites, Journal of Sensors (2012), 1–7.

13.

Reimche

Duhm

Zwoch

Bernard

and Bach

, Development and qualification of a process-oriented nondestructive test method for weld joints to operate with remote field eddy current technique, Proceedings of 9th ECNDT (2006), 1–13.

14.

Mook

Michel

and Simonin

, Electromagnetic imaging using probe arrays, Journal of Mechanical Engineering 57 (2011), 227–236.

15.

Owston

C.N.

, A high frequency eddy-current, non-destructive testing apparatus with automatic probe positioning suitable for scanning applications, Journal of Physics E: Scientific Instruments 3 (1970), 814–818.

16.

Hughes

Fan

and Dixon

, Near electrical resonance signal enhancement (NERSE) in eddy-current crack detection, NDT & E International 66 (2014), 82–89.

17.

Blodgett

Sathish

and Boehnlein

, Application of resonant frequency eddy current technique on a shot-peened nickel-based engine-grade material, American Institute of Physics: Melville (2007), 1608–1615.

18.

Pratap

and Weldon

, Eddy currents in anisotropic composites applied to pulsed machinery, IEEE Transactions on Magnetics 32 (1996), 437–444.

19.

COMSOL Multiphysics Reference Manual.

20.

Qiu

Cheng

and Takagi

, Interlaminar contact resistivity and its influence on eddy currents in carbon fiber reinforced polymer laminates, NDT & E International 94 (2018), 79–91.

Detection of delamination in laminated CFRP composites using eddy current testing: Simulation and experimental study

Abstract

Keywords

1. Introduction

2. EC probe characterization

3.1 Numerical model

Table 2 Stacking sequence of samples

Table 4 Parameters of fabricated CFRP samples

4.2 ECT system

4.3 Results and discussion

5. Conclusion

Footnotes

Acknowledgments

References

Table 2
Stacking sequence of samples

Table 4
Parameters of fabricated CFRP samples