Detection of fibre waviness using ultrasonic array scattering data

Size: px

Start display at page:

Download "Detection of fibre waviness using ultrasonic array scattering data"

Barbra Sullivan
6 years ago
Views:

1 18th World Conference on Nondestructive Testing, April 2012, Durban, South Africa Detection of fibre waviness using ultrasonic array scattering data Damien PAIN, Bruce DRINKWATER Abstract Ultrasonics and NDT Research Group, Department of Mechanical Engineering, University of Bristol. Progress regarding the composites manufacturing process has been progressively improved, small imperfections such as fibre waviness are inevitable. Any localised departure of a ply from the desired lay-up direction and is known to adversely affect strength. Therefore, end users are now particularly interested in detecting subtle defects such as fibre waviness at various stage during prototyping or in completed composite components. In this paper, an ultrasonic array is used to extract information that characterises the scattering of the interior of the composite structure. A method for extracting the scattering matrix from experimental array data over a spatially localised region is presented. Ultimately this leads to the ability to map the distribution of scattering behaviour within the composite. It is shown experimentally that the scattering matrices obtained from areas of correctly placed ply layers and areas that are wrinkle free are statistically separable. Keywords: ultrasonic array, fibre waviness, carbon fibre composite 1. Introduction Composite materials owe their success to the ability to produce favourable mechanical properties in specific directions while reducing the weight of components relative to traditional materials. In this paper we consider a subtle defect type, termed fibre waviness, which is known to adversely affect the mechanical properties of a structure. For example, a fibre misalignment of only 0.25 is sufficient to reduce the compressive strength from 2720 MPa to 1850 MPa and down to 700 MPa for a 3 deviation [1]. Therefore manufacturers and end users are now interested detecting fibre waviness at various stages during manufacturing or after manufacturing. Fibre waviness can be described as ply deformation that exhibits a wave like pattern occurring within the composite material component. Two types of fibre waviness can exist and are termed out-of-plane waviness in-plane waviness with respect to the plane of the plies. Here only the out-of-plane or through thickness fibre waviness is treated. -ray Computed Tomography (CT) is considered to be the reference in terms of detectability of fibre waviness. CT images offer a level of detail down to the fibre distribution that is so far not achievable with any other technique; however CT is expensive and often not practical for on-site inspection[2-6]. Ultrasonic techniques are commonly used for composite material inspection and have been particularly effective for true characterisation of delaminations, porosity or voids using both single element probes and multi-element array [7], [8].However, the fibre arrangement within the matrix generates both scattering and anisotropy that makes ultrasonic methods difficult to apply. In this paper we show that a scattering matrix approach can be used to detect and quantify out-of-plane fibre waviness.

2 2. Data Capture and Imaging One of the most powerful post-processing imaging algorithms is the Total Focussing Method (TFM) [9]. Its efficiency has been demonstrated in previous publications [10], and is considered by some authors as the gold standard [11]. In essence this uses all the elements in the array to focus on every imaging pixel and hence produce a high resolution image over a potentially wide area. Once the full matrix of data has been captured it is usually filtered to remove any noise that is not within the known bandwidth of the array elements. The filtration step is realised in frequency domain by applying a Fast Fourier Transform (FFT) to each time signal contained in the data matrix to give the complex spectra.the subsequent spectra are multiplied with Gaussian function to give. An inverse Fourier transform can be applied to but in practice it is useful to apply the Hilbert transform to obtain. In order to generate an image, the TFM focuses on every pixel of the image as shown by equation (1), (1) (2) where is the image intensity, is the Hilbert transform of the time-trace for the transmit element located at position and the receive element located at position, and where is the ultrasonic longitudinal wave velocity. Alternatively the phase angle, extracted from the Hilbert Transform can be used in the TFM algorithm to produce a phase image. 3. AVERAGE SCATTERING MATRI It has been shown that sub-wavelength discontinuities such as cracks or voids can be characterised using the ultrasonic scattered field [12]. When an ultrasonic wave is incident on a discontinuity within a component scattering occurs and the scattered field carries information about the shape, orientation and size of the discontinuity. The scattering matrix (or S-Matrix) is defined as a matrix containing all the far field scattered amplitudes from all incident and scattered directions [13], (3) where is the distance from the reflector to the receive element, is the amplitude of the scattered wave, the amplitude of the incident wave and the wavenumber. A discontinuity can be fully characterised if its far field scattering amplitudes (or S-matrix) is known over all angles. Experimentally this means that a transmitter-receiver pair needs to be moved completely around the discontinuity. In reality it is rarely possible to have access to all

$positions surrounding a defect since structures adopt various shapes that make this impossible and therefore only a fraction of the scattering matrix is measureable.$

3 positions surrounding a defect since structures adopt various shapes that make this impossible and therefore only a fraction of the scattering matrix is measureable. The S-matrix can be extracted from experimental data by a number of techniques. Here we use a sub-aperture technique [14]. The challenge is to obtain scattering information from spatially localised region over as wide a range of angles as possible. In the sub-array approach a series of images are formed from different transmitter(tx) and receiver(rx) subarrays as shown in Fig. 1. For a pair of sub-arrays and the amplitude of the intensity of the TFM image at the location of the scatterer can be expressed as follows, (4) Sub-array transmitter (m) = SA (m) Sub-array receiver (n) = SA (n) Q Figure 1. Sub-array principle schematic. It is now possible to use the above scattering matrix approach to interrogate composite material in order to detect fibre waviness. However fibre waviness generally has a length (wavelength) and amplitude greater than the ultrasonic wavelength used to carry out the ultrasonic inspection. Fig. 2 shows a schematic of a simplified out-of plane fibre waviness with the region of interest (ROI) where fibres are misaligned. The approach used in this paper is to average the extracted scattering matrices over a number of locations within the ROI. The computation is realised such that intensities from scattering matrices located at a same angular positions are averaged together. SA (m) SA (n) Figure 2. Schematic of the region of interest containing out-of-plane fibre waviness.

4 50mm 20mm 4. Experimental apparatus A 5-MHz linear array with 64 elements (manufactured by Imasonic, Besancon, France) was used in the experiments, and its parameters are shown in Table 1. A commercial array controller (Micropulse MP5PA, Peak NDT, Ltd., Derby, UK) was used to capture the complete set of time-domain signals from every transmitter-receiver element pair of the ultrasonic array. The captured data were then exported and processed using MATLAB (The Math- Works, Inc., Natick, MA). TABLE 1. Array parameters Array parameter Value Number of elements 64 Element width (mm) 0.53 Element pitch (mm) 0.63 Element length (mm) 15 Centre frequency (MHz) Bandwidth at -6dB (MHz) 5 4 Two test samples were used during the experiments for the validation and test the average scattering matrix algorithm. The geometry and dimensions of each sample is shown in Fig. 3 and Fig. 4. Sample#1 is an aluminium block 50 mm deep with a 45 o notch located at 20 mm from the top surface and 1mm wide. Sample#2 was designed by Rolls-Royce to develop a methodology to distinguish the different defects and establish the resolution capabilities of various NDE techniques in flat simple laminates. Sample #2 is a 30 mm thick carbon fibre sample made from 124 plies of carbon fibre unidirectional following a ply staking (-45). Three different fibre folding regions were created comprising of one, two and three ply folding with the folded packs being 0.25 mm (1 ply- small waviness), 0.5 mm (2 plies-medium waviness) and 0.75 mm (3 plies- large waviness). The back face and top face of the sample are finished with pre-peg woven plies. The three regions of waviness have an estimated amplitude of 2mm and a wavelength of to 10 mm. 1 mm ROI Figure 3. Sample#1, Aluminium block with slot at 45 o

5 10mm 30mm 150mm... 3 folds waviness 2 folds waviness 1 fold waviness 10 mm Non wavy region 300 mm Y Position 20 Position 2 Position 1 50 mm 150 mm 250 mm Figure 4. Sample#2, drawing of the carbon sample including the three different types of fibre waviness. 5. Results and discussions 5.1 TFM: Amplitude and phase imaging The amplitude TFM was shown to be a reliable imaging technique [9], also it is possible to image the internal structure of a component using the phase information which can sometimes produce more details regarding the characteristics of a defect. Fig 5. shows imaging of the three different levels of waviness as shown if Fig 4. The images are plotted up to 20 mm deep so the back wall located at 30 mm in not visible in the images. Amplitude TFM images are shown at the top (Fig 5a-5c) and the phase TFM is presented at the bottom (Fig 5d-5e). Images of the large waviness (3 plies) show a clear disturbance at 10 mm deep indicating the presence of the fibre waviness. The medium waviness (2 plies fold) is discernable around 10 mm but the signal is lost when looking at the small waviness (1 ply fold). The phase imaging exhibits a repetitive pattern throughout the image. The pattern is linked to the plies layout, however does not represent every single plies of the sample since the physical resolution achievable is limited by the wavelength which in this case is 0.59mm.

a) b) c) d) e) f) Figure 5. Comparison; Amplitude TFM a), large waviness, b), medium waviness and c) small waviness; phase TFM d), large waviness, e), medium waviness and f) small waviness.

6 a) b) c) d) e) f) Figure 5. Comparison; Amplitude TFM a), large waviness, b), medium waviness and c) small waviness; phase TFM d), large waviness, e), medium waviness and f) small waviness. Array and composite sample used are described in table I and figure 3 respectively. 5.2 Validation of scattering matrix code: Aluminium sample and Rexolite sample The aluminium sample (sample#1) was used in order to validate the scattering matrix extraction algorithm against a feature with a known orientation. The slot is 1mm wide and 5 mm long at an orientation of -45 o from the vertical axis. The region of interest (ROI) was chosen to cover the slot and had an aperture of 6mm along axis (-3 mm to 3 mm) and 6 mm along the axis (-18mm to -23 mm). The resolution chosen when extracting the scattering matrices was 0.5 mm which results in 144 pixels in the ROI. The average scattering matrix generated is visible in Fig. 6a) and an inspection of the amplitudes shows that the highest amplitude is obtained from the pulse-echo elements at a reflected angle of o which is 1.5 o off the expected reflected angle. The darker area in the top right corner of the matrix is due to the flat end of the slot reflecting at +45 o. The middle pattern (lighter colour) is an artefact inherent to all single scattering matrices contained in the ROI and exacerbated when averaging scattering matrices together; the pattern shows that due to the orientation of the slot, only a small quantity of the ultrasound reflected is detected by the array along the specular diagonal.

Figure 6. Averaged scattering matrix over a 6 6mm ROI for sample#1 with a 45 o slot. 5.

7 shows results obtain for wavy and non wavy zones. Regarding the waviness, Fig.7a) is representative of a typical pattern obtain when waviness is present.

7a), conversely those areas are not present when no waviness is detected as seen in Fig. 7b.

7 Figure 6. Averaged scattering matrix over a 6 6mm ROI for sample#1 with a 45 o slot. 5.3 Carbon fibre sample The experiments using sample#2 have shown that the average scattering matrix present a typical pattern when there is waviness. The average scattering data presented in Fig. 7 shows results obtain for wavy and non wavy zones. Regarding the waviness, Fig.7a) is representative of a typical pattern obtain when waviness is present. The average scattering matrix for a wavy location exhibits darker areas symmetrical in respect to the specular diagonal (Fig. 7a), conversely those areas are not present when no waviness is detected as seen in Fig. 7b. In order to statically quantify our results, data were collected at different locations over the wavy regions as well as over a region that was thought to contain no waviness; for each wavy area several batches of 20 measurements were made along the fibre waviness as shown in figure 4c with a 5 mm pitch between each location. The non wavy region was chosen between two wavy regions again as seen in Fig. 4. The waviness was made to be localised and so it was reasonable to assume that the zone between two wavy regions was not affected by the surrounding waviness. The same sampling process was then followed for the non wavy region with a 5mm pitch between each position. To avoid any tilting of the probe about the axis, the array was moved along a guide to maintain constant vertical orientation. Analysis of the scattering matrices indicated that the largest differences in the scattering matrices occurred in dashed box regions of Fig. 7. Note that since two similar regions are visible within the average scattering matrix, two windows were generated. For each window the mean amplitude of the pixels included in the window is calculated and the two windows of each average scattering matrix averaged together. In order to compare the results with the non wavy results identical windows were selected for non wavy average scattering matrices. a) b) Figure 7. Comparison S-matrix, a) sample with waviness and b) zone without waviness

8 Density in % If the information provided by the average scattering matrix allows us to statistically separate wavy and non wavy groups then the potential exists to define a meaningful waviness threshold above which detection is possible, and below which it is not. Fig. 8. shows an example of the differentiation of the two groups in the case of the large waviness. The representation of the probability density function of the two data sets, i.e. waviness and no waviness shows that it is possible to separate the wavy sample from the non wavy sample. Note that the cut-off point (CoP) is arbitrary and is defined here as CoP=( ),where is the mean value of the wavy group and is the standard deviation of the wavy group as shown in Table 2. Table 3 summarises the distribution test between the two groups. TABLE 2. TABLE 3. Waviness No waviness waviness Mean (µ) test present % absent % Standard Deviation (σ) positive True Positive (TN) 87.5 False Positive (FP) 10 negative False Negative (FN) 12.5 True Negative (TN) CoP TN TP FN FP no waviness waviness mean no waviness mean waviness waviness 2 std deviations no waviness 2 std deviations Amplitude Figure 8. Statistical differentiation between wavy and non wavy data. 6. Conclusion It has been shown that it is possible to use ultrasonic array scattering data for the detection of fibre waviness. Also the paper is a proof of principle further work using different samples is necessary to accurately determine the sensitivity of the technique and its resolution. Importantly, microscopic examination will be used to allow precise measurement of the actual extend of the waviness. Acknowledgement The work was funded through the Engineering and Physical Sciences Research Council (EPSRC) and Rolls-Royce.

9 References [1]M. Wisnom, The effect of fibre misalignment on the compressive strength of unidirectional carbon fibre/epoxy. Composites, Volume 21, Number 5. September [2] R. Oster, Computed Tomography as a Non-destructive Test Method for Fiber Main RotorBlades. Computerized Tomography for Industrial Applications and Image Processing in Radiology, March 15-17, 1999, Berlin, Germany. [3] P. Joyce, D. Kugler, and T. Moon, A Technique for Characterizing Process-Induced Fiber Waviness in Composite Laminates - Using Optical Microscopy. Journal of Composite Materials 1997;31(17): [4] G. Requena, G. Fiedler, B. Seiser, P. Degischer, M. Di Michiel, T. Buslaps 3D-Quantification of the distribution of continuous fibres in unidirectionally reinforced composites. Composites: Part A 40 (2009) [5] J.S.U. Schell, M. Renggli, G.H. van Lenthe, R. Muller, P. Ermanni, Micro-computed tomography determination of glass fibre reinforced polymer meso-structure. Composites Science and Technology 66 (2006) [6] M. Kosek, P. Sejak, Visualization of voids in actual C/C woven composite structure. Composites Science and Technology 69 (2009) [7]R.D. Adams, and P. Cawley, A review of defect types and non destructive testing techniques for composites and bonded joints. NDT International Volume 21, Number 4 August [8] R J Freemantle, N Hankinson, and C J Brotherhood, Rapid phased array ultrasonic imaging of large area composite aerospace structures. Insight - Vol. 47 No. 3, March [9] C. Holmes, B. W. Drinkwater, P. D. Wilcox, Post-processing of the full matrix of ultrasonic transmit receive array data for nondestructive evaluation. NDT&E International 38 (2005) [10] J. hang, B. W. Drinkwater, P. D. Wilcox, A. J. Hunter, Defect detection using ultrasonic arrays: The multi-mode total focusing method. NDT&E International 43 (2010) [11]O. Oralkan, S. Ergun, J.A. Johnson, M. Karaman, U. Demirci, K. Kaviani, T.H. Lee, B.T. Khuri-Yakub, Capacitive micromachined ultrasonic transducers: nextgeneration arrays for acoustic imaging?, IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control 49 (11) (2002) [12] J. hang, B. W. Drinkwater, and P. D. Wilcox, Defect Characterization Using an Ultrasonic Array to Measure the Scattering Coefficient Matrix. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, vol. 55, no. 10, October [13] L. W. Schmerr, Fundamentals of Ultrasonic Nondestructive Evaluation A Modeling Approach. New York: Plenum Press, [14] P. D. Wilcox, C. Holmes, and B. W. Drinkwater, Advanced Reflector Characterization with Ultrasonic Phased Arrays in NDE Applications. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, vol. 54, no. 8, august 2007.

Extending Acoustic Microscopy for Comprehensive Failure Analysis Applications

Extending Acoustic Microscopy for Comprehensive Failure Analysis Applications Sebastian Brand, Matthias Petzold Fraunhofer Institute for Mechanics of Materials Halle, Germany Peter Czurratis, Peter Hoffrogge