Table of Contents ECNDT '98 Session: Aerospace

Ultrasonic Inspection of Adaptive CFRP-Structures J. Pohl* - Uni. of Magdeburg, Germany. *Corresponding Author Contact: Otto-von-Guericke Univ. IWW, D-39016 Magdeburg, Germany ; Juergen.Pohl@Masch-Bau.Uni-Magdeburg.DE

Introduction

Adaptive structures receive and process environmental information, thus, enabling them to react selfregulatively to changes of outer influences. Such a structural response allows compensation of undesired mechanical deformation for shape control, artificially increased stiffness or effective damping of vibrations and noise as examples of application. The resulting structural systems are termed smart, adaptive or intelligent structures [1].

The adaptive function requires both inherent sensoric as well as actuatoric capabilities of the material. One possible variant of such an adaptive structure are fibre reinforced composites, for example, carbon fibre reinforced polymers (CFRP) with embedded piezoceramic plates as sensors and actuators.

Development, manufacturing and optimisation of this material system require non-destructive examination for structural characterisation and detection of manufacture flaws. Furthermore, load-induced flaws which affect serviceability have to be proved non-destructively. For a complete characterisation various non-destructive methods have to be applied [2] but among them ultrasonic techniques are of special interest for their extended possibilities of flaw detection and structural characterisation.

The adaptive material system

Carbon fibre reinforced polymers are a very attractive material for advanced constructions because of their high stiffness and strength as well as their low weight. Another advantage is the possibility to design anisotropic properties in manufacturing according to the demanded load direction and the desired purpose of application.

Fig 1: Typical design of an adaptive CFRP-structure (left) and micrographs of real RTM- and Prepreg-structures (right)

Fig 2: Typical flaws in an adaptive CFRP-structure

Special sets of specimens have been designed to investigate typical material structures and imperfections and to optimize test parameters. Figure 1 shows a typical example of the principle layout of the employed CFRP-structures which are developed and manufactured by the German Aerospace Research Center (DLR). The plate-like composite consists of various layers of CFRP with desired orientations of carbonfibres. Both unidirectional and woven raw materials are used. Typical thickness of one layer ranges from 0,12 to 0,25mm. Piezoceramic plates of lead zirconate titanate (PZT) coated with a very thin metallic layer of copper nickel alloy are embedded within the CFRP layers. Typical dimensions of these plates are 30x50x0,2mm³. Insulating layers of aramid-epoxy material provide the necessary electric insulation of the piezoceramic. The overall thickness of the specimen ranges from 1 to 3mm. Carbon fibre rovings (fibre bundles) or a complete CFRP-layer serve as electric connections. Other implemented variations regard manufacturing techniques. So beside conventional prepreg-lamination technology (PP) a differential-pressure-resin-transfer-moulding (DP-RTM) was employed.

Several structural features as lateral and depth position of the piezoceramic plate, ply stacking sequence, thickness of layers (CFRP and insulation layers), orientation and waviness of fibres are of special interest for these subjects. Beside this manufacture induced flaws may occur and can affect integrity and serviceability of the structure. Examples of typical flaws are given in figure 2 and include local fibre and resin concentration, porosity, fracture of the piezoceramic, fibre undulation, delamination and debonding. Both artificial produced and natural developed flaws of these types are included in the specimen programme. To verify and compare non-destructive examination results cross-sections of interesting areas of the specimens are prepared and examined by microscopy in various cases.

Ultrasonic techniques

The variety of flaws and structures demands application of different ultrasonic techniques with various image presentations. This requires to take into consideration the distinctiveness of ultrasonic propagation in adaptive CFRP-structures with their anisotropy, heterogeneity and complex internal geometry.

As known from other anisotropic media in addition to well-known longitudinal and shear waves modes appear neither being pure longitudinal nor shear waves. Their specific behaviour sensitively depends on direction of propagation, polarisation and anisotropy of material properties [3]. In addition, plate modes (Lamb modes) may occur at oblique incidence if a sufficient small thickness to wavelength ratio exists [4].

Considering the influence of integrated sensors and actuators with their very different acoustic properties (acoustical impedance of piezoceramic differs from that of CFRP by scale of 10), the conditions for the propagation of ultrasonic waves increasingly get complex. Fortunately the normal incidence of longitudinal waves to the plate corresponds to the existence of the pure longitudinal wave mode [3] and admits the detection of most structural flaws with sufficient sensitivity.

Other problems arise from the layered structure [5]. The material is a combination of alternating layers (fibres in the matrix, epoxy layers, insulation layers, piezoceramic plates) of different thickness, density and elastic properties. At each interface partial transmission and partial reflection with different amounts depending on the acoustic properties take place. So reflectivity and transmissivity of layer's interface are important for the ultrasonic response of the whole structure. The numerous interfaces produce a large number of echos because multiple reflections and interference effects occur. Local variations of ultrasonic velocity in the plate's plane make it more difficult to predict and eliminate the influence of these multiple reflections.

Furthermore the material shows high attenuation which causes a frequency downshift of the peak frequency of the ultrasonic pulse with increasing travel path.

For the examination of the adaptive CFRP-composites high frequency (up to 25 MHz), highly focused transducers with suitable ultrasonic equipment (bandwidth up to 100 MHz) were employed. This was aimed to achieve high axial and lateral resolution. A stable coupling was ensured by immersion technique. The received ultrasonic signals were stored in a computer with a digitisation rate up to 200 MHz with full waveform capture. While scanning averaging over a desired number of A-scans at every point for noise reduction was applied. The stored A-scans allow subsequent off-line processing and analysis of the data with different image or signal processing and various presentation formats. So, for example, it is possible to generate projection C-scans with varying gate width and B-scans of different sections in real time and evaluate phase information or measure geometric parameters at the A-scan at the same screen. The so given completing information of different presentations is a valuable help for interpretation of results. If necessary, further signal or image processing was applied. The employed testing techniques can be divided into three categories:

In most cases pulse echo technique with longitudinal waves at normal incidence was used. In some cases this was completed with a double transmission technique using an additional reflector for integral characterisation of the structure. Because of the high reflectivity and therefore caused low transmissivity of the piezoceramic this technique proved not to be very favourable in all cases. Oblique incidence of waves with evaluation of backscattered signals was applied, too. Beside this above listed common ultrasonic techniques a pitch-catch technique was employed for testing overall performance of only the piezoceramic. In this case the piezoceramic plate of the composite was driven by an electric pulse and acted as an ultrasonic transmitter in a broad frequency range. The generated signal was received and scanned by an ultrasonic transducer. This was performed at different frequencies as well as in a broad-band low frequency range (mid frequency about 0,5 MHz) and at high frequencies (between 10 and 20 MHz).

Results

Fig 3: Bent actuator - schematic left, C-scan and B-scan at marked line of C-scan

Fig 4: Short-circuited actuator (left: C-scan, centre: B-scan at marked line, right: micrograph of marked area in B-scan)

Structural features
Ultrasonic is the preferred non-destructive technique for images of inner structures in different planes. It is possible to resolve the layer structure by means of short ultrasonic pulses and so measure structural dimensions in depth direction. This requires knowledge of ultrasonic velocity in the different layers of the material. An evaluation of possible velocity variance dependent on manufacturing technique, structure of carbon-fibre material and fibre-resin concentration allows to determine the precision of measurement of structural dimensions. Figure 3 shows a typical example for this. The material was carbon fibre fleece with a relative low fibre-resin concentration. As the principal view shows, a rectangular window was cut into the insulation layer above the piezoceramic for electrical contacting at this side. The B-Scan reveals a resulting deformation of the actuator. The maximal bending was determined to 0,23 mm which within 20 m agreed with the actual value, measured at the plastographically prepared cross-section. The additionally shown C-scan admits the determination of extension of the deformation in the actuator plane and detection of cracks with different orientations.

Another example of special structural features is given in figure 4. Here multiple short-circuit occurred between connections and actuator, identifiable as nearly round bright areas at the actuator edges in the ultrasonic picture. Additionally, some bright lines become visible. The B-scan shows them laying short beneath the surface and the phase information from the A-scan identifies them as fibre rich regions. These estimations were affirmed by the cross-section which shows a small fibre bundle at this place in the host material of low fibre-resin ratio. With this an explanation for short-circuits is given because these fibre bundles with varying depth may contact different electric poles.

Flaws
Detection and representation of flaws are strongly influenced by anisotropy and heterogeneity of CFRP-structure. So differences of acoustic impedance between carbon fibres and resin matrix (especially in the case of woven material) supply laterally local changes of ultrasonic reflectivity and consequently transmission. Thus structures of preceding layers may be impressed to representations of deeper layers and mask their structural features. Here a feature based image processing may be advantageous. With use of a-priori information of depth position of the desired feature a weighted superposition (addition or subtraction) of C-scans or projection C-scans diminishes the display of the overlaid CFRP-structures and emphasizes the display of desired features. Furthermore conventional image processing may be added, but it has to be kept in mind that sometimes these routines suffer from removing the desired information unintentionally, too.

Fig 5: Cracks in woven RTM-material ( left: unprocessed C-scan at straight incidence, centre: processed C-scan, right: C-scan at oblique incidence)

Fig 6: Broken PZT-actuator in prepreg-material (left: C-scan, right: B-scan at marked line)

Fig 7: Delamination caused by impact loading: a) micrograph, b) straight incidence - scanning from downside, c) oblique incidence - scanning from downside, d) straight incidence - scanning from upper side, e) double transmission

Fig 8: Damages in high speed impact tested specimens

Figure 5 gives an example of disturbing layer structures in specimen with intentionally broken actuators. The display of woven RTM-material is disturbed by the fibre-structure so that in the unprocessed C-scan the cracks are barely visible. Better detectability is provided by the described image processing. Here the crack network partially becomes visible in the case of straight incidence. Oblique incidence technique is also suited, but only cracks perpendicular to scanning direction are detected. The other example regards a prepreg laminate with more unique material properties. Here beside the cracks adhesive tape stripes for fixing the actuator are to be seen. The B-scan in figure 6 displays the multi-layered structure of the specimen and indicates areas where the actuator echo is attenuated due to the flow out of epoxy resin at actuator cracking.

Other examined damages regard action of mechanical loading. The following example denotes a conventional impact test. Here a classical delamination was caused at the rear side. At the front side regarded to loading a rounded delamination between insulation and CFRP developed, because this area was weakened by porosity (nearly rectangular area in figure 7d)). A typical elliptic shaped delamination arised under the actuator at the interface of different orientated CFRP-layers. Additionally to the C-scans taken from both sides of the plate a
C-scan from a double transmission technique using an additional reflector is depicted where delaminations due to shadowing are detected.

In a series of other experiments a high speed loading in a split Hopkinson bar system was employed. This causes a short duration pressure/tension pulse in the material. In difference to classical impact testing of CFRP the load acts not at a point but at a distinct circular area of 20 mm diameter. Further details of the testing system are given in [2].

The first picture in figure 8 shows the damage in a prepreg material. The shaded region in the centre corresponds to the impact area. Here partly delamination at the CFRP-insulation interface occurred but mostly a degradation of the piezoceramic with small cracks parallel to the plate's plane took place. In the other parts of the actuator normal cracks developed which are barely visible as horizontal and oblique lines. In RTM-material with woven CFRP different damages developed. The action of the load leads to a ray-like pattern of cracks in the piezoceramic plate, radial outgoing from the loaded circular area in the centre. Due to better material quality no delamination or parallel cracking happened in this case.

Because here only some examples of flaw detection may be given, table 1 lists a description of detectability of different flaws with ultrasonic techniques in a summarising manner. This includes experiences with various types of flaws, manufacturing and raw material which affect detectability in different order. In cases where CFRP-structures are more homogeneous, the detectability becomes better than the given one in the table. Not included is required effort for testing. So techniques with oblique incidence have to be assessed as expensive because several scan directions must be applied for detection of flaws with arbitrary orientation.

Flaw type --------- ultrasonic technique	piezo-ceramic cracking	Debonding	delamination	fibre and resin concentration	matrix cracking	undulation	fibre orientation	pores
normal incidence	3	1	1	1	3	2	2	3
oblique incidence	2	4	3	2	2	2	1	1
Explanation: 1-very good 2-good 3-possible 4-hardly possible 5-impossible

Table 1: Detectability of flaws

Performance of piezoceramic


Fig 9: Testing of bent actuator (compare figure 3) with pitch-catch technique, a) C-scan at low frequency, b) and c) C-scans at high frequency

The above described pitch-catch technique allows an estimation of piezoceramic performance. The surrounding material stays untested in this case. As figure 9 shows, the behaviour changes with frequency remarkably. In the low frequency range the piezoceramic plate is stimulated to oscillate in a higher mode of natural frequencies. This is only less affected by cracks but debonding and delamination alter the behaviour considerably. With the use of high

frequencies beside performance of the piezoelectric plate it is possible to resolve structural features. Here cracks and deformations in the piezoceramic become clearly detectable so that this technique admits a very effective estimation and evaluation.

Summary