NDT of welded joints in advanced composite materials

TWI Bulletin, August 1987

Graham Edwards, BSc (Hons), DMS, MlnstNDT, MITD, is a Senior Research Engineer in the NDT Department.

Increasing use of fibre reinforced polymeric composite materials for critical applications has created a demand for non-destructive testing (NDT) techniques capable of detecting significant defects. Currently available techniques are reviewed here, and the potential for future developments and improvements in performance discussed.

Fibre reinforced polymeric materials have come into widespread use for applications where high strength-to-weight ratios and considerable stiffness are needed. The ability to use a variety of fibre and matrix materials, and to alter the fibre lay-up, enables material properties to be tailored to specific applications. Soundness and fitness for service of glass fibre reinforced polyester composites for non-critical applications, such as the hulls of relatively small boats, have traditionally been maintained by application of straightforward quality control procedures during manufacture. Precise knowledge of the properties achieved is not generally required. However, the development of carbon fibre-epoxy composites for load-bearing structures in aircraft and, more recently, of the range of weldable carbon fibre/thermoplastic polymer composites, also aimed at the aircraft industry, has increased the need to gain an understanding of the significance of defects and to develop reliable means of non-destructive testing (NDT) of both parent material and joints.

This article examines the features of both thermosetting and thermoplastic fibre reinforced polymeric materials which influence their properties or fitness for purpose, and reviews the available NDT methods for detecting such features. Current NDT practice is almost exclusively confined to the testing of thermosetting composites used for aircraft structures and therefore the review concentrates on this application. The ability of NDT to detect faults in welded joints in thermoplastic composites is important and recommendations are given for further development in this area. However, optimum NDT procedures cannot be defined until a study has been made of the significance of defects in welded composite structures.

Advanced composites

Fibre reinforced composites are designed for high strength-to-weight ratio and high stiffness. These materials are therefore particularly well suited for aircraft structures, where the aim is for dimensional stability and high payloads. There is also an interest in fibre composites from the automotive industry, where the light weight and strength of the materials would make cars safer. However, the materials are presently too expensive to use in all but the most critical of applications.

Advanced composites are now regarded as essential materials in the design of military aircraft. In the AV8B vertical take-off aircraft, 25% of the airframe weight is of composite materials.^[1]

The nature of advanced composites

A useful introduction to composite materials exists in published work.^[2-5] These materials derive their mechanical properties from a combination of the tensile properties of the fibres and the quality of the bonds between the fibres and the polymeric matrix. To achieve maximum benefit from this arrangement the fibres must be strong. Typical fibre materials include carbon, glass and boron. Carbon fibres are the most widely used in the aerospace industry.^[6]

Continuous fibres may be used in composites as unidirectional prepreg (pre-impregnated) tape, filament wound into cylindrical shapes, or braided. Alternatively, the fibres may be chopped into short lengths and incorporated into short staple prepreg, random prepreg, sheet moulding compound,(SMC), dough moulding compound (DMC) or injection moulding compound. Chopped fibre composites may be formed easily, but for high performance applications the fibres must be continuous.

Most composites are made with polymer matrices, although there is a growing interest in glass and metal matrices. These latter materials are beyond the scope of this article. The matrix serves a number of purposes. The matrix:

1. protects the fibres from damage;
2. holds the fibres in place;
3. is the medium by which stress is transferred to the fibres;
4. acts as a barrier to crack propagation between the brittle fibres.

Epoxide resins are favoured as matrices for high performance carbon fibre reinforced plastic. The carbon fibres are pre-impregnated with the epoxide to form tapes which are laid up in a laminate and hardened or cured. The cured laminate cannot be softened by heating and so it cannot be formed or welded.

Thermoplastic polymers have become increasingly used as matrix materials.^[7] A thermoplastic can be softened or melted by heating, then shaped in a plasticised condition and cooled to re-solidify.

Thermoplastics suffer no chemical alterations during processing and may be processed repeatedly. In practice, however, there is some degradation of properties.

One of several types of thermoplastic used in advanced composites is the aromatic polymer poly-ether-ether-ketone (PEEK). This is available in consolidated sheets made up by laminating the prepreg under controlled conditions. The sheets can be shaped by a number of processes, including hydro-forming, press forming, roll forming and diaphragm shaping.^[8]

The feasibility of welding a similar aromatic polymer thermoplastic composite material, APC-2, is currently under investigation.^[9] The welds are made by locally heating the material above 400°C by one of a variety of methods and applying a constant pressure across the joint surfaces while it cools down.

Defects in advanced composite materials

The study of defects in advanced composites has been confined almost exclusively to aircraft structures. The significance of defects in composite materials is not well understood, and there is no generally accepted classification.^[11] Failure mechanisms are different from those in metals. For example, single delaminations may not cause failure by virtue of the fact that they tend to lie parallel to applied stresses as a consequence of the design of the structure. However, delaminations which are open to the surface, for example at the edges of a composite structure or around fastener holes, will propagate rapidly because of water ingress.

The complexity of the materials has led to empirical defect criteria, but more recently attempts have been made to rationalise the criteria based on a better understanding of thermal properties, fatigue processes, toughness, impact tolerance and chemical resistance of composites.^[12]

Defects can arise as faults in the prepreg, delaminations and voids in the laminate, poor bonds in the structure or because of service conditions. This is true for composites with thermoset and thermoplastic matrices.

Thermoset composites

Defects in the prepreg

The fibre to matrix bond has an important bearing on the transverse, flexural and shear behaviour of the composite. If the bond is too weak, the fibre and matrix tend to behave independently and if too strong, may lead to brittleness. The strength of the bond is heavily dependent on the surface treatment given to fibres prior to impregnation with the polymer.

The fibre to matrix volume percentage is normally kept to about 60%. If the proportion of matrix increases, the strength and stiffness of the prepreg will decrease. If the matrix volume decreases, the fibres may come into contact with each other and become damaged. It is essential that volume fraction ratio is controlled during manufacture.

Epoxy prepregs have to be kept carefully in rolls interleaved with polythene and refrigerated to avoid premature cross linking of the resin. They have limited shelf life at room temperatures.

Defects in the laminate

The prepregs are draped over moulds to form the finished part before curing in an autoclave. Laying-up is a skilled operation done in dust free conditions. Contaminants on the prepreg surface will cause delaminations.

A laminate only 2mm in thickness can contain 32 plies, and so every opportunity is taken to reduce the risk of misaligned fibres. The use of prepreg laying machines on large flat surfaces reduces the possibility of human error in the lay-up of the laminate.

During curing, moisture is expelled from the resin, and may give rise to porosity. The laminate is pressed between a porous release film inside the autoclave to absorb the moisture. Excess moisture can be left in the surface of the prepreg if it has not been dried properly after removal from refrigeration. The presence of porosity is unlikely to constitute a particularly significant defect, but it does make the detection of gross defects, such as delaminations in the laminate, difficult. Porosity levels up to 4% of the surface area of a component are normal, but in some structures the level may reach 15% and be acceptable.

Delaminations are the predominant factor in determining the service life of the composite. They may occur because of incomplete curing, which has failed to create a bond between the plies, or surface contaminants on the prepreg which have prevented adhesion. They may even be caused by the presence of the interleaves between which the prepreg is stored and which should be removed prior to laying-up. Acceptance levels for delaminations are based on the density and size of indications in the C-scan image of the ultrasonic test described later. As indicated above, porosity interferes with such assessment, therefore allowances must be made for the presence of porosity.

Defects in the bond

The laminates are incorporated into complex aircraft structures by a number of processes. As much as possible of the structure is cured in one stage, but inevitably parts have to be joined together. This may be accomplished mechanically with bolts, usually of titanium, by hot curing together parts of the structure which have been partially pre-cured, or by using adhesives.

Figure 2 shows a bonded wing section of an aircraft made in composite materials.^[13] The spars are designed in the form of a J and in the uncured state, placed between pre-cured skins, which are located in a jig to control final dimensions. The internal spars are pressurised against the skins by inflated rubber bags, and the assembly cured and bonded in one stage. If the pressures on the bonded surfaces are not uniform, disbonds will occur. The curing of such complex geometries in one operation presents problems of access for NDT, because the tests cannot be carried out before assembly.

Weak bonds may occur because contaminants on the surfaces prevent proper adhesion. This is of particular concern when carrying out repairs.

Thermoplastic composites

Little experience has been gained in the fabrication of aircraft structures from prepregs of carbon fibre composites with a thermoplastic matrix, but it is likely that the defects will be similar to those in thermoset composites. The fibres are more difficult to impregnate with some of the thermoplastics and the prepregs are stiffer and more difficult to work with than thermoset prepregs. It is therefore likely that consolidated sheets will be used that can be formed into the shapes required, rather than using structures which have been laid-up by hand.

Defects in consolidated sheets

Laminate in the form of consolidated sheet can be laid-up under controlled conditions to reduce porosity levels to, generally less than 1%. The consolidated sheet can then be formed after heating to soften the matrix. The forming may have detrimental effects on the composite material that will constitute defects. For example, the fibres may become distorted or damaged, and the molecular structure of the polymer altered.

Defects in welds

The consolidated sheet that has been formed into the required shape, for example a wing spar, may be bolted to the structure or bonded using adhesives. Alternatively it may be welded.

The welding of thermoplastic composites is in the development stage and the nature and significance of defects have not been established. In addition to the disbonds encountered in adhesive joints there are likely to be defects caused by disruptions to the fibres and changes in the molecular structure of the matrix. Moreover, each welding process, and there are several under investigation, may have unique types of defect associated with it.

Defects arising from service conditions

A major concern in the use of composites in aircraft structures is impact damage.[11] In metals, subcritical impact loads lead to visible indentation because of plastic deformation and damage is localised. In composites, high energy impacts can lead to catastrophic failure, while low energy impacts can lead to so called barely visible impact damage (BVID). This significantly reduces the strength of the composite in subsequent service. A fir tree effect occurs, where slight damage at the surface of the laminate spreads outwards between the plies at greater depths. Broken fibres, delaminations and water ingress may lead eventually to failure.

Current practice for the NDT of composite aircraft structures

The NDT methods discussed in this article are summarised in Table 1. The NDT of thermoset composites is well covered in published work.[14-18] During the manufacture of aircraft structures the primary NDT method is ultrasonics, with radiography used to investigate areas identified as possible defect locations. Vibration techniques are capable of detecting disbonds, particularly between composite skins and the internal honeycomb structure. Because it is simple to operate, vibration testing equipment is used extensively to detect BVID on aircraft in service.

Table l Possible defects that can be detected by NDT

x Not applicable    *Special techniques

Ultrasonic testing

Ultrasonic testing of composite airframe structures during fabrication has a number of distinctive features. To couple the ultrasound transducers to a structure, it is either immersed in a tank of water or, if too large for a tank, scanned with jet probes while free-standing.

The immersion test is more sensitive. The structure is laid flat, a few centimetres above a glass plate that provides a monitor echo for the ultrasonic flaw detector (Fig.3). The ultrasonic transducers scan in a raster fashion above the structure, using ultrasound in a pulse-echo mode of operation.

Jet probes maintain a column of water between the transducer and the test surface. The water pressures are high to maintain a constant level of coupling and the jet nozzles are designed to reduce turbulence in the water. Even so, the pulses reaching the receiver are 'noisy' and too strongly attenuated for pulse-echo mode of operation. Consequently, a through-transmission technique is used, with the transmitting and receiving transducers on opposite sides of the structure. For access to corners in the structure, the pulses are guided along brass tubes (Fig.4).

High levels of ultrasound attenuation are encountered because resins absorb the ultrasound energy and fibres scatter the high frequency components of the pulse. Therefore shear waves are attenuated strongly by the composite, and compression waves are restricted to frequencies below 5MHz. To provide pulses of high amplitude, the transducers are often as large as 25mm diameter and are excited by RF tone bursts instead of single spike voltages.

The results of the ultrasonic test are normally displayed in hard copy as a C-scan image of ultrasound attenuation in the structure. As the probe scans in a raster fashion across the structure, its movement is duplicated by the pen over the paper. A one-to-one gearing ratio is used between pen and probe, so that plotters for hard copy of large structures can be several metres long, although more recent systems using a microcomputer enable a video display to be used. The ultrasonic flaw detector monitors either the echo from the glass plate underneath the structure in an immersion test, or the through-transmitted pulse between the jet probes. Changes in the amplitude of the monitor echo or pulse are transmitted as signals to control the voltage on the electrostatic pen. The C-scan then becomes a map of the ultrasonic attenuation properties of the structure.

The ultrasonic test is particularly sensitive to delamination in a structure which, if larger than the cross sectional area of the sound beam, will obliterate the monitor echo or pulse. Voids and small clusters of delaminations scatter the ultrasound and reduce the amplitude of the monitor echo. These translate to the C-scan image as variations in the grey scale. Changes in the resin-fibre volume have a similar effect. A resin rich area in a bond line can increase the ultrasonic attenuation and decrease the amplitude of the monitor echo, giving rise to a false indication of poor bonding.

A C-scan gives a two-dimensional image of a three dimensional object so that defects can be superimposed upon each other. Acceptable levels of porosity may obscure significant delaminations and disbonds. Interpretation of the C-scan image therefore requires skill. Measurements are made of the number of indications per unit area and their location with respect to edges and other features in the structure. Typical accept/reject criteria for an aircraft structure are shown in Table 2. Sensitivity level A is for more critical areas in the bead seat or corners of curved structures. The C-scan is calibrated on laminates containing Teflon patches as reference reflectors for ultrasound.

Table 2 Reject criteria for C-scan indications

Radiographic tests

Generally, the size of composite structures makes 100% coverage with radiographs too expensive, but radiographs do provide a more complete image of the structure than ultrasonics and allow less significant flaws, such as voids, to be distinguished from delaminations.

Polymer matrices and carbon fibres have a low absorption coefficient for X-rays. Good image contrast is therefore difficult to achieve at X-ray energies above 20kV.

Radiographic image quality is also determined by definition. High definition radiographs require fine grain films and long focus-to-film distances. With a focus-to-film distance of typically 2m, large areas of the structure can be radiographed in one exposure and it is possible to magnify the image three or four times without undue loss in definition.

The effects of low kV and high levels of X-ray absorption in the air over long focus-to-film distances increase exposure times considerably, when compared with radiography of metals of comparable thickness. Instead of conventional X-ray tubes with a 10mA output, tube currents of 30-40mA are common for inspecting composite structures to increase the intensity of the X-rays and so reduce exposure times.

Radiography is not sensitive to delaminations or cracks if they lie in any plane which is not parallel with the radiation beam. At low kVs however, water becomes slightly opaque to the radiation.

Therefore, by taking radiographs after the ultrasonic C-scan, when the structure has been immersed in water, surface breaking delaminations and cracks are made visible on the radiograph because of water ingress.

Better contrast is achieved by using radio opaque penetrants, for example tetrabromoethylene. They are only practical for inspecting areas that are known to be defective, and the solvent residues may degrade the resin. They are frequently used to radiograph impact damage.

Radiography may be used to check fibre orientations, when they are brought into question by the forming techniques used in fabrication. Carbon fibres produce little contrast on a radiographic image, so filaments of lead silicate glass are introduced when making up the laminate.

The glass has a higher absorption coefficient for X-rays than carbon, therefore these fibres are highly visible on the radiograph and provide information about the lay-up of the carbon fibres themselves.

Vibration tests

Vibration tests are used to find disbonds in structures and BVID during in-service inspection.^[19] They are derived from simple coin tapping tests used in the early days of aircraft manufacture.

Modern instruments measure frequency and dampening variations in sonic and ultrasonic vibrations. They are sensitive to the matrix/fibre volume, fibre size, fibre direction and cracks in the matrix, as well as delaminations and disbonds. Because delaminations almost always occur with impact damage, the tests will detect BVID. There are three techniques.

Resonance technique

This technique, established by an instrument called the Fokker Bond Tester, detects changes in the resonant frequency of a composite structure as a broad band piezoelectric transducer is scanned manually across its surface. Resonance over a delamination is at a different frequency to resonance over the contiguous laminate (Fig.5). Although widely used, the technique requires skilled operation, because it is sensitive to surface roughness and the pressure applied to the probe. It also requires a liquid couplant between probe and surface.

Tapping technique

A more recent technique, which is not sensitive to surface conditions measures the dampening of vibrations. The surface is struck repeatedly by a hammer, designed to give reproducible impacts. The hammer is contained within a hand held probe, and strikes the surface at a repetition rate of 1-5Hz. A delamination or disbond decreases the structural stiffness of the composite resulting in a less intense force and longer impact duration from the hammer blow. The force-time history of each hammer blow is recorded by incorporating a force transducer in the hammer. Frequency spectra, derived by carrying out Fourier transforms on the force-time histories of each hammer blow, show that impacts on damaged areas have more energy at low frequencies, which fall off rapidly at higher frequencies (Fig.6).

Unlike resonance methods, the method does not need a liquid couplant between transducer and the surface. The instrument is set up initially on an area known to be free of damage. Gross damage can be detected audibly by the change in the pitch of the vibration, but minor damage can be resolved by observing a change in the frequency spectrum displayed by the instrument.

Impedance method

This technique forms the basis of the acoustic flaw detector.^[20] The instrument displays the difference in amplitude and phase between receiver and transmitter piezoelectric transducers inside a hand held probe (Fig.7). The test frequency, between 1-10kHz, is adjusted using a reference block containing a known disbond, to give maximum discrimination between good and defective areas. The stiffness of the laminate above the disbond affects the amplitude and phase of the vibration picked up by the receiver transducer.

Developments in the NDT of composite aircraft structures

The NDT of composite aircraft structures during manufacture puts emphasis on detecting the extent of a multitude of dispersed flaws rather than on detecting and measuring the size of individual flaws. Global NDT methods are therefore important and must be capable of gathering, storing and processing large amounts of data. For example in the ultrasonic C-scan, the filtering effects of the composite on the bandwidth of the echo may distinguish disbonds from resin rich areas, a condition which can be difficult to detect from changes in the sound attenuation alone. This has led to a rapid introduction of computers to control NDT in recent years.

Another development has been the arrival of thermographic methods as an alternative to ultrasonic tests.

Computerised ultrasonic C-scan systems

Many computerised ultrasonic C-scan systems have appeared recently for inspection of aircraft structures.^[21] There are many variations, but essentially the systems consist of three parts:

1. The mechanical scanning assembly and probe manipulator. This is capable of scanning complex structures with multiple probe arrays either in immersion tests, or with jet probes.^[22]
2. The multichannel flaw detector. This is capable of monitoring several ultrasound echoes simultaneously for changes in amplitude and flight times. Some instruments capture the whole of the ultrasonic A-scan waveform.^[23]
3. The host computer. This is a mini- or microcomputer, which processes the digitised data from the ultrasonic flaw detector and scanning assembly to provide C-scan images on a high resolution video monitor. It also runs the programs which control the whole system and may provide simple menu driven software for the operator to control and calibrate the test.

The principal advantage of a computerised system is that raw data from the test can be stored on disc and recalled at different sensitivity levels in the video image. For conventional C-scans in hardcopy, the structure would have to be rescanned if the test sensitivity had to be altered, a task which could take several hours.

The C-scan has already been described as an attenuation map of the structure. In a video image, each pixel colour represents an amplitude level in the monitor echo or through transmission pulse. The colour scale is carefully selected to prevent essential information being lost.

For large structures, each pixel may represent thirty or forty data points from the scan. The pixel colour may represent a mean amplitude value or, for attenuation mapping, the minimum amplitude value, or when an echo from a bond line is being monitored, the maximum amplitude value of responses from the bond.

Readily available software is used for 'zooming in' on areas of the video display, interrogating the image to give frequency distributions and line scans and for enhancing the image by increasing the contrast of indications and reducing noise.

Software especially tailored for C-scans has been produced to reduce noise in honeycomb structure. This is able to filter signals created by the regular and consistent changes in the ultrasound caused by the honeycomb pattern. Other software has been developed to interrogate the C-scan video image to measure the density of indications and thereby set the threshold levels between signal and noise to give optimum contrast to the defects.^[24]

The computer may also be used to construct B-scan images. For these, the flaw detector sets a monitor gate between the interface echo from the top surface of the structure and the backwall echo from the bottom surface. The time-of-flight to any echo within the gate, which exceeds a predetermined threshold level is recorded. The video image then becomes a 'slice' in the X-plane of the raster scan, with X-displacement along the top of the screen and time-of-flight down the screen. This B-scan format is used to locate flaws in the through-dimension of the structure. A new B-scan is presented on the screen at each increment in the Y-direction.

By combining C-scan with B-scan data, the computer can construct isometric projections of the composite structure on the video monitor. If the computer prints hard copies of two isometric projections with only a slight displacement in the grid axis, they can be viewed as stereographic pairs.^[25]

In summary, the digital processing of ultrasonic C-scan data is making a considerable contribution to the testing of structures in advanced composites. Several techniques may be incorporated into one system, for example ultrasonic spectroscopy, time-of-flight and attenuation measurements to allow measurements of ultrasonic stress wave factor^[26] or detect fibre orientations. ^[27]

Digital X-ray imaging

The advent of digital X-ray imaging (Fig.8) presents new possibilities for the inspection of advanced composites:

1. Large structures or batches of components can be scanned in real time.
2. The method is non-contact and does not require water coupling.
3. After the initial high capital cost, running costs are low, because there is no expenditure on films.
4. The low contrast of defects in composites can be enhanced in the digital image by using computer software for noise reduction, contrast stretching and spatial filtering.^[28]

The X-ray images are produced on a video screen, having been enhanced by the computer. They are weak and a considerable amount of intensification is necessary before they can be detected by a vidicon. The X-ray energies used to radiograph composites are low, typically 20kV. The efficiency of the X-ray image intensifiers used in digital X-ray imaging is poor at this level of excitation energy. For automated inspection, the X-ray intensifiers may be replaced by a linear array of solid state diodes, each coupled separately to the computer.^[29] An image is then built up in a similar manner to the C-scan, as the linear array, giving information for the X-direction, moves progressively in the Y-direction. Although spatial resolution is determined by the size of the diodes and is not as good as that attainable using fluoroscopic detectors, contrast is high because of their greater sensitivity to X-rays.

Microfocus radiography

Digital X-ray imaging and low kV X-rays are ideally suited for microfocus inspections in which resolution is improved by using a small point source of X-rays. The 10-50µm focal spot of the X-ray tube can be used to project an image of a test object over one or two metres to achieve high levels of magnification. Magnification helps to compensate for the poor spatial resolution of the digital image, when compared with the radiographic image on film.

Stereoradiography

Stereoradiography can offer quantifiable data on the spatial distribution of defects in three dimensions.^[30] A stereoscopic pair of radiographs may be viewed conventionally under a stereoscope, or alternatively, two stereoscopic video images, one on a red scale, the other on a green scale, may be superimposed. The stereoscopic effect is obtained by viewing the video image through red/green binoculars.

Thermography

Thermography is a technique whereby contours of equal temperature, or isotherms, are mapped over a surface. It has become a useful analytical tool in recent years since the advent of powerful infrared imaging cameras.^[31] For testing advanced composites, thermography is an attractive NDT method, because the low thermal diffusivity of the materials makes the capturing of perturbations in the isotherms because of defects relatively easy. Thermography is a rapid, non-contact method of inspection, which may provide an alternative to ultrasonic testing.

Surface and sub-surface defects can be detected, because they create local hot or cold spots in a thermal map of the surface. The defects may actively create the heat when the composite is subjected to vibrational loads, or they may passively interrupt the flow of heat imparted to the composite from an external source.

Thermochromic liquid crystals

Liquid crystals are materials which at certain temperatures possess both the properties of a liquid and of a crystal. They are viscous, because of a short range molecular structure which is characteristic of a crystalline solid. To apply a thin, even coating to the test surface, they are dissolved in a volatile organic solvent and are either sprayed or carefully brushed on.

When light falls on to a thermochromic crystal, part of it is polarised in a clockwise direction, and part of it in a counter clockwise direction.^[32] Above a certain temperature, one of these parts will be reflected as visible light, the other will be transmitted. As the temperature increases, so the inter-molecular distances increase, and the wavelength of the reflected light decreases from red through to blue and into the ultraviolet spectrum until a state of disorder exists and the liquid is isotropic and colourless. The effect is reversible. Various compounds can be mixed to actuate the colour changes at temperatures from -20 to 250°C and scale the temperature range from a few degrees to a fraction of a degree Celsius. Resolution as high as 0.005°C is achievable.

The temperature range over which the crystals are coloured must be chosen carefully. It produces a window within which useful thermal information can be recorded on a high speed photographic camera or video camera (Fig.9). The useful window may be only a few seconds in duration.

From changes in the isotherms at the test surface, the maximum temperature gradient can be ascertained. This will depend upon the thermal properties of the composite, the characteristics of the defect, the magnitude of the heat input and the sensitivity of the liquid crystals.

Liquid crystals are suitable for active and passive thermographic tests. Liquid crystals have been used to investigate defects in glass reinforced plastic,^[33] crack propagation in composites^[34] and carbon orientation in composites.^[35] Although a successful laboratory tool, the technique has yet to be used industrially. The recent development of encapsulated liquid crystals may help to produce practicable test procedures.

Infrared cameras

Infrared (IR) detectors held at a distance from a heated surface provide a non-contact method of thermography. Infrared radiation is detected using solid state photon effect devices, that produce electrical signals when irradiated by photons. They are sensitive to the wavelength of radiation. For maximum sensitivity, the semi-conductors are cooled in liquid nitrogen. The thermal image is created by the IR camera using familiar optical techniques. The development of an IR camera at the Radar Research Establishment in Malvern, which is capable of scanning the thermal field at fifty frames per second, has opened up new possibilities for thermography. At this scanning rate, the video images can be processed using readily available computer software for image enhancement and interrogation.

The main drawback in using IR cameras is their expense. Cameras which operate at 50 frames per second cost in excess of £25,000. It is probable that IR vidicons will become available in the near future, which use pyroelectric polyvinylidene fluoride (PVDF) films in place of the fluorescent screens used in optical vidicons. These will be cheaper and more robust.

Pulse-video thermography

Pulse-video thermography has been developed by the AERE at Harwell, and has been used to detect delaminations in composites.^[36] The technique is being used in an ESPRIT project to develop enhanced imaging techniques.^[37]

A pulse of heat is created by a xenon flash tube placed either behind, or in front of, the composite structure. As the thermal wave passes through the material, the area behind it cools to equilibrium, creating a transient event which is more sensitive to subtle discontinuities in the composite than a continuous heat flux. The time taken by the pulse to travel through the material depends upon the material's diffusivity.^[38] For a composite it may take one or two seconds, for a metal it will be a fraction of a second. A delamination in the path of the heat flux creates a lag in both the heating up and the cooling down of the surface temperature (Fig.10).

A model for the thermal signal has been derived,^[39] but a great deal of work is needed to characterise the inspection variables; heat input, exposure time and geometric set-up.

Vibrothermography

As an alternative to passive thermographic techniques, vibrothermography offers an active technique in which part of the composite structure is excited by mechanical vibrations.

High frequency mechanical vibrations lead to local hot spots in the composite and a heat flow that is disturbed by discontinuities. Friction on fracture surfaces of fibres and in delaminations also creates heat. The frequency of the vibration can be varied to set up local resonance in flawed areas creating high levels of viscoelastic heat.^[40]

Published work so far has described methods of scanning the thermal fields using conventional IR cameras, which have provided adequate resolution to reveal gross delaminations and impact damage in thin coupons of epoxy composite. The capabilities have yet to be explored of thermal imaging using computer enhancement techniques such as those employed in stress pattern analysis by thermal emission (SPATE).^[41]

Thermography offers considerable potential for NDT of composites. It may be non-contact, rapid and may be sensitive to defects which cannot be identified by other NDT methods.^[42,43]

However, research is needed to develop models to analyse the effects of defects on heat fluxes^[44] and to standardise test procedures.

Acoustic emission tests

As part of proof testing a composite structure, acoustic emission (AE) could have an important role in providing information about subcritical defects that may induce failure during service. Acoustic emission is detected by placing appropriate transducers at sensitive positions on the composite structure and then flexing the structure to produce the emissions.

Acoustic emission is particularly sensitive to delaminations. Under cyclic loading, fretting at the surfaces of the delamination gives characteristic AE events. Under heavier cyclic loads, the build-up of damage can be monitored as the brittle fibres fail and cracks develop in the resin matrix.

Acoustic emissions are elastic waves generated by the release of stored elastic energy as the material undergoes plastic deformation or fracture. The AE method involves one of two measuring techniques. The first called 'ring down counting' counts the number of times the AE signal exceeds a predetermined threshold value. The second called 'lock out counting' counts the number of AE events between a first predetermined threshold and a time after which no further AE has been recorded for a predetermined period.^[45]

The use of amplitude distribution analysis of AE is a potentially powerful tool in monitoring defects, but requires development. Peaks in the amplitude spectrum often relate to different failure modes.

There is some evidence that AE methods may be sensitive to weak bonds.^[17] Coherent adhesive bonds which adhere weakly to the surfaces of the adherend can be difficult to detect by other NDT methods.

Eddy current tests

Eddy current test methods find widespread use in the aircraft industry for the detection of cracks and corrosion in metal structures. For composite structures they are still in the early stages of development, although an eddy current test device is already available for detecting BVID.^[46]

Cracks perpendicular to the test surface can be detected when the composite fibres are conductive, as is the case with carbon and boron, because they interrupt the flow of eddy currents.^[17] The eddy current response is a combination of resistive effects parallel with the fibres and capacitive effects transverse to the fibres across the non-conductive resin matrix. Changes in the eddy current field can be related to cracks in the fibres, or changes in the fibre/matrix volume. The eddy currents are generated by a coil carrying a high frequency 1-10MHz) current, scanning the test surface. The coil impedance responds to changes in the eddy current field and is monitored by a voltmeter. However, this simple arrangement of equipment can make it difficult to distinguish relevant signals from non-relevant signals. For example in testing for BVID, the eddy current response has been found to be more sensitive to acceptable changes in the fibre matrix volume than to impact damage.

A more complex arrangement of equipment, which uses a cathode ray oscilloscope for phase analysis of the eddy current signal, multiple test frequencies and special coil configurations, may provide test data to distinguish between defect conditions.

NDT of joints in advanced composites

Adhesively bonded joints

These can present particular problems for NDT. The most common defect is the disbond caused, for example, by the surfaces of the adherend failing to make proper contact. The plane of the disbond will lie parallel with the surface of the structure in most cases and can therefore be detected by ultrasonics or vibration techniques. However, there may be poor adhesion across the bond surfaces in the absence of any gross disbonds. This is usually the result of poor surface preparation of the adherends that has given rise to minute bond line discontinuities which cannot be detected by conventional NDT methods.^[47]

The problem of detecting poor adhesion is of particular importance in composite repairs^[48] that play an important role in aircraft maintenance.^[49] Two test methods have been applied in the laboratory with some success. The use of ultrasonic spectrum analysis has revealed a correlation between the frequency spectrum of ultrasound passing through metal to metal adhesive joints and the degree of adhesion.^[50] In the other method, the velocity of interfacial ultrasonic Stoneley waves propagating along the adhesive between the surfaces of an adherend has shown a sensitivity to bond strength.^[51]

Welded joints

The problem of weak joints that are transparent to ultrasound and cannot be detected by vibration techniques may also present itself in welded joints in thermoplastic composites. Moreover, new NDT techniques may be needed to detect fibre disruption and changes in the properties of the matrix caused by high welding temperatures. For example, microfocus radiography has been used experimentally to image fibre misalignment and ultrasonic velocity measurements have been employed to detect changes in matrix crystallinity. However, it is not clear whether these structural and chemical changes will have any significant effect on the performance of the joint. Their significance must be established before developing new NDT techniques.

The welding process itself may affect the selection of NDT methods. For example, the use of wire implants for electro-fusion welding processes will severely hamper the detection of bond line defects with ultrasonics, because of the presence of wires that strongly reflect ultrasound.

Discussion

The NDT of composites in the aerospace industry is characterised by a high level of automation and diversity of test methods. Automation is advancing by an increase in the use of computerised test systems. These are able to conduct sophisticated signal and image processing on test data, scan complex composite structures and provide simple test control and calibration.

Until now, NDT has been confined almost exclusively to the inspection of advanced composites in thermoset matrices for aircraft structures. In the future there could be two important developments.

1. The introduction of advanced composites in thermoplastic matrices will follow a different manufacturing route. Non-destructive testing will be performed in three stages: on the pre-consolidated sheets, after forming, and on bonds and weld joints in the structure.
2. An increasing interest in the use of advanced composites on the part of other industries, for example the automotive and petrochemical industries.

The nature of defects in thermoplastic composites has yet to be established and although some account can be taken of defects that are found in thermoset composites, the different manufacturing and fabricating processes used with thermoplastics may introduce new defect conditions.

Good welds, for example, may rely on the absence of microstructural changes in the joint and disruptions in the fibres in addition to any gross lack of fusion. These conditions are beyond the resolution capabilities of present NDT methods.

One approach to overcoming this problem would be to use new NDT methods, such as ultrasonic spectrum analysis. Another approach would be to refine current NDT techniques. For example, ultrasonic C-scan systems could accommodate high resolution 10MHz ultrasound if computer techniques are used to reduce noise in the C-scan image from ultrasound scattering.

Non-destructive testing should be taken into account in design. For example, if fibre disruptions in welded joints are important, then glass marker fibres could be incorporated into the carbon fibre composite to make them visible in a radiographic image.

The discussion of suitable NDT methods for welded joints in thermoplastic composites is hypothetical until the significance of defects has been established. Therefore any programme of development work in NDT must be closely tied to one in mechanical testing.

The introduction of new materials and processes will have an effect on NDT, but so too will an interest shown in composites on the part of industries other than aerospace. In the automotive industry the emphasis may be more on rapid, reproducible tests and less on defect resolution. For this application thermography provides a rapid non-contact method of inspection. This technique requires a great deal of development work closely related to recent advances in thermal imaging.

For the petrochemical industry the NDT requirements will be for portable instrumentation to inspect pressure vessels and pipelines and for fixed installations to monitor plant performance. Eddy current methods may provide suitable portable instruments to inspect vessels made of carbon fibre composites. The behaviour of eddy currents in the fibres is more complex than it is in homogeneous metals and models are needed to evaluate the effects of test frequencies and coil configurations.

Acoustic emission is already established as a method for monitoring pressure vessels made of glass reinforced plastics. The use of advanced carbon fibre composites in clad pressure vessels for example, will require refinements to the technique.

In the future, automated test installations are likely to include 'expert systems' that are able to interpret test data.

Conclusions

The inspection of composites is an active area of NDT. For the testing of welded joints in thermoplastic composites there is a need to develop existing test procedures, for example with computerised ultrasonic immersion tests, and to introduce new methods such as thermography. However, development is hampered by a lack of knowledge of the significance of defects.

Acknowledgements

The author would like to thank Peter Mudge and other members of the NDT Department for their advice in the preparation of this article. The work was funded by Research Members of The Welding Institute.

References