The imaging technique-based NDE identifies the difference in the captured images before and after a defined time/deformation, which highlights the changes due to a flaw or defect. Some of the popular imaging NDE techniques are shearography testing, computed tomography, and digital image correlation, etc. The principles of these popular imaging technique-based NDE methods are discussed in this section.
1.2.1. Shearography-Based NDE
Shearography is a laser-based method, the basic layout of which is shown in
Figure 5. A laser source is used to illuminate the sample, which is imaged with the charge-coupled device (CCD) camera via the beam shearing element. The laterally shifted subsequent images of the sample surface that are continuously captured are coherently superimposed in the image plane by the optical beam shearing element. This captured lateral shift is termed as image shear and the created superposition is called a shearogram. The shearogram is an interferogram created over the reference object wave with the superimposition of a sheared object wave over it. With variable loading conditions, multiple images for shearograms are similarly recorded, wherein the induced deformation or variations are captured. The difference in the deformation state due to loading variations is then correlated with the interference fringe pattern resulting from the absolute difference recorded in subsequent shearograms. This resulting differential image is further termed as a “D-Image”. When processed, rather than providing deformation (as in holographic interferometry), the fringes provide the rate of change of the deformation. The surface and subsurface defects tend to modify due to the applied loading, resulting in minor alterations or major disturbances in the recorded loading fringe pattern, which is expected to appear more or less uniform for the no defect case. Hence, this principle is used for classifying and categorising various defects, depending on the extent of alterations or disturbances recorded in the shearographic fringe pattern. The simplified working principle of shearography testing is shown in
Figure 5. Although, it is essential to induce deformation in the sample as applied by vibration
[31], mechanical loading
[32], thermal expansion or contraction
[33][34], vacuum force
[35][36], and microwave heating
[37], and it could be applied in a static or dynamic way. The CCD camera captures the interferometric pattern, which leads to an edging image and consists of structural information
[38][39].
Figure 5. The working setup and principle of shearography testing-based NDE of composite components.
A loading system is used to stimulate deformation or to change the state of deformation of the sample surface, which is required in shearography testing. The loading systems which are normally used in shearography comprise of thermal pulse shearography, vibration shearography, and vacuum shearography. Thermal pulse shearography is efficient for inspecting impact damages or cracks that are barely visible to the naked eye. When the image shearing direction is not perpendicular to the orientations of cracks, the detected defect direction has sensitivity which is comparatively greater than the perpendicular image shearing
[33]. Vibration shearography is effectively utilised for the inspection of foam on the external tank of NASA’s space shuttle
[40], and also to disclose flat-bottom holes of variant sizes and locations at various depths in composite laminate
[31]. Vacuum shearography is efficient for imaging fibre debonding in composite laminate
[41], aluminium honeycomb panel
[42], and also in the composite element panel of the helicopter (honeycomb core with two outer layers of epoxy and graphite) and in the tail unit
[43] for core damages, core splice-joint separation, and delamination
[40]. There are other popular methods, such as thermoelastic stress analysis, that involves an infrared camera (replacing the laser source and CCD camera, as shown in
Figure 5) for picking any variations, which is very similar in principle with shearography
[44][45] and widely used for continuous monitoring of composites.
The advantage of shearography includes that by easily highlighting stress concentrations around the specific defect, it highlights the type and criticality of that defect, and since composite failure normally occurs by stress concentration, the degree of stress makes a lot of difference
[46]. The other advantage of shearography includes it being less prone to noise than other different types of NDE. This feature is useful because it does not require highly skilled operators for the inspection and determination of component usability without long-term training, since just comparing the deformed and undeformed shearograms becomes a lot easier. It has been found as very useful method for honeycomb and foam composite structures, with the ability to detect defects up to 2–3 mm in depth or sometimes even more
[22]. The major drawback associated with shearography is the difficulty of inspecting defects other than delamination. For that reason, it is sometimes combined with other NDE types which can help in identifying specific flaws
[3].
Another noteworthy shearography limitation is the requirement of applying appropriate external loading increments to the testing sample during examination. Therefore, appropriate loading systems are required. Furthermore, the alteration monitored in the displacement pattern derivative reduces with the defect depth or with an increase in its diameter. Therefore, the efficient digital shearography application for defects characterisation is difficult and is still dependent upon various factors, for example depth and defect type, material type, and laser illumination
[47]. Hence, this is another reason why shearography is sometimes coupled with other techniques of testing to detect flaws other than delamination
[3].
Overall, there are two major drawbacks associated with shearography. The first is the difficulty of detecting defects other than delamination. For this reason, it is sometimes combined with other NDE techniques to help in identify other types of flaws
[3]. Secondly, shearography requires applying external loading to the testing sample during examination, and consequently, an appropriate loading system is required. Furthermore, the alteration monitored in the derivative of the displacement pattern reduces with the subsurface depth of the defect or with the increase in its diameter. Consequently, application of shearography for defects’ characterisation is often challenging as it is dependent upon various factors, including the depth and type of defect (delamination, impact, crack, etc.), material type, and laser intensity
[47].
1.2.2. Computed Tomography-Based NDE
Computed tomography (CT) is an advanced form of conventional X-ray radiography, which is used for non-destructive 3D imaging of internal features of solids. The working setup of CT-based NDE is schematically presented in
Figure 6. This is an outstanding imaging technique to examine the details in terms of size and volume of structures with very high precision, and also in three dimensions, which is especially valuable for the inspection of structural integrity of complex geometries
[48]. The resolution of the technique depends inversely on the volume measured. Consequently, standard CT, microCT, and nanoCT techniques have been developed for increasing the resolution of feature sizes at the cost of the 3D volume that can be imaged.
Figure 6. Working setup and process of computed tomography-based NDE of composite components.
The extraction of information from a computational tomography dataset involves a series of steps. The data are acquired from multiple radiographs obtained as the sample is rotated relative to the X-ray source. A reconstruction algorithm combines all the angle-dependent radiographs into a 3D reconstructed image of the sample
[49][50]. Most computational tomography systems apply a filtered back projection (FBP) reconstruction algorithm, because of its predictable nature with regards to reconstruction times and computational cost
[50]. The accurate representation of an object using FBP can be attained by projecting the X-ray integrals for each X-ray path back through the object. This projection method has high accuracy with the additional feature of low noise of the projected images, although alternative iterative reconstruction techniques have considerable advantages in more problematic settings
[51][52]. Iterative algorithms utilise a linearised forward model of the X-ray acquisition method and use optimisation algorithms to reverse this model. Image viewing and processing techniques can be used to extract valuable information once a computational tomography volume is reconstructed, and this is called visualisation. The extremely high resolution achieved in nanoCT scans can detect details up to 0.2 µm for low absorbing materials
[53]. The obtained image quality is generally dictated by the variable control of spatial resolution
[54], contrast, noise, and artificial features, called artifacts
[54], such as scatter and beam hardening
[55][56][57].
The limitation associated with computational tomography is the sample size, which greatly affects the details obtained by the current generation of computational tomography systems
[58]. The resolution is restricted by the pixel size of the detector, which depends on the component geometry and is often 2 to 3 times the pixel size
[59]. The affected region which the detector covers is normally 2000–4000 pixels wide
[60], and therefore, the test object size is restricted by it. Other shortcomings are field of view limitation, in situ monitoring, and attenuation contrast. Comparing standard CT to microCT and nanoCT systems, the pixel size limitations indicated above can be translated into dimensions, i.e., millimetres, micrometres, or nanometres, which are more tangible and relevant than pixel size.
1.2.3. Digital Image Correlation-Based NDE
Digital image correlation (DIC) is a non-contact method to examine defects and has applications in structural composites. The sensing mechanism can perform inspections on active and passive structures. It is an optical technique which uses pattern matching and image registration methods for exact two- and three-dimensional calculations of change in the object shape which is being inspected
[61][62][63]. The DIC technique is useful to determine deformation, stress, strain, and displacement. This method has a number of applications in engineering and manufacturing techniques to determine the changes and provide measurements for finite element analysis, material and structural analysis, and quality control
[64][65][66].
The three-dimensional digital image correlation works on the principle of combined methods of image correlation with the photogrammetric location. Photogrammetry works on the triangulation principle, which is used for three-dimensional coordinate measurements
[64], as shown in
Figure 7.
Figure 7. Working setup and principle of digital image correlation method-based NDE of composites.
Objects being examined are targeted in photogrammetry and a series of photographs are taken from different angles for recreation of dimensional target locations of the object. The accurate location of every target can be acquired by triangulation with various different target views of the object being examined
[64]. Prior knowledge of the orientation and position of cameras for the images taken is important, and triangulation is dependent on these factors in photogrammetry. There are two cameras in 3D DIC, mounted at each end of a tripod camera (base) bar; therefore, the relative orientation and position of cameras is known with respect to each other. The cameras have the same working distance in this way, and therefore are easily removed from photogrammetry location measurements as a variable
[62].
When the load is applied, the pattern is deformed as the object being inspected is deformed. The structural deformation under specified loading conditions is captured and recorded by two DIC cameras. Unique correlation areas, which are called facets, are defined by initial image processing across the whole imaging area, and normally range from 5 to 20 square pixels in size
[64][65][66]. Every consecutive pair of images is tracked with sub-pixel precision from the measurement point located at the centre of each facet. The movements of these facets are tracked by an image correlation algorithm by using mathematical techniques to achieve maximum similarities determined from consecutive photographs
[64]. The software of image correlation is essentially designed with the purpose of pattern matching, which can be performed on both curved and flat surfaces
[62]. The locations of each facet in three-dimensions can be determined before and after every loading stage while examining in this way, and therefore resulting in three-dimensional displacements, the plain strain tensor, and the three-dimensional shape
[65][66]. Data of full-field displacement can be acquired from the measurement facet point tracking in the applied regular target patterns.