Table of Contents ECNDT '98
CS-Pulsed Eddy Current Inspection for Cracks in Multi-Layered Joint Al-Alloy Aircraft StructuresW. Bischoff, H.-A. Crostack, M. Maass*
Dortmunder Initiative zur rechnerintegrierten Fertigung e.V, Dortmund, Germany
Th. Meier, G. Tober
Daimler-Benz Aerospace Airbus GmbH, Bremen, Germany
|TABLE OF CONTENTS|
Aluminium structures of aircraft in operation are bearing alternating loads of continuous vibrations and shock waves, which inevitably cause fatigue cracks in the ageing aircraft material. Eddy current testing offers good opportunities to detect this cracks and is widely used in this field /1, 2/. Eddy current testing is especially sensitive on high conductive materials like aluminium and furthermore it is not restricted by any safety regulation, it is easy to use and mechanically robust.
However the sensitivity of eddy current testing systems depends on depth and volume of the damage and the reliability is reduced by disturbing effects. In multi-layered rivet-joint aluminium structures, strong influences on the NDT are caused by the rivets. Since cracks in ageing rivet joint aluminium structures are starting from the drilling holes of the rivets, the sensitive detection of cracks in hidden layers, separated from the strong rivet influence, is still demanding further improvements from the NDT-methods. In particular in deeper layers of multi-layered aluminium structures, bending cracks, which don't go through the whole thickness of the layer, are difficult to detect. Since the eddy currents are just displaced but not blocked by these cracks, the sensitivity of the measurement is strongly reduced.
This paper presents CS-pulsed eddy current measurements (CS = Controlled Signals, /3-6/) on rivet-joint aluminium alloy structures of aircrafts and demonstrates the opportunities of this technique for a sensitive crack-detection despite the disturbing influence of the rivets. Beside using a conventional type of eddy-current sensor, also the performance of the new rotating field eddy current sensor will be demonstrated /7, 8/.
Fig 1: Drawing of the investigated specimen
The CS-pulsed eddy current testing system can be divided in three parts: PC with software, measurement device, sensor. The software is controlling the whole measurement procedure, which simplifies the creation of an automatic testing system. By this the CS-pulse is generated, optimised and together with all parameters settings send digitally to the measurement device, which transmits the analogue signal to the sensor. After a measurement the PC reads the received CS-pulse from the measurement device and signal-evaluation and presentation can be performed. The CS-technique allows a fast evaluation by reducing the whole signal evaluation to the storage of the maximal peak of the received pulse and the time where the maximal peak appears within the duration of a pulse.
The maximal peak can be seen as a summation of the changes of the all amplitudes within the frequency range of the pulse, and the time-shift is a summarised phase shift over the whole frequency range. Due to a specific optimisation of the transmitting CS-pulse during a calibration procedure before the measurement, which affect the shape of the amplitude and the phase of the CS-pulse over the whole frequency range, the CS-pulse is already sensitised to the measurement effect (in this case: cracks) and automatically reduces other influences. The measurement device mainly consists of A/D and D/A converter, for generating the analogue pulse, and transmitting and receiving amplifier, for adapting the measurement device to the different impedance characteristics of sensors.
|Fig 2: Eddy current sensor for crack detection in multi-layered aluminium structures|
|Fig 3: Rotating field eddy current sensor|
Figure 2 shows the used conventional sensor with view to the touch down surface. The measure in centimetre beneath gives the size of the sensor. Under the dark area on the left side of the sensor the transmitting coil is build in, on the right side the receiving coil. This type of sensor shows a good behaviour for detecting cracks in rivet-joint multi-layer aluminium structures and is therefore part of testing manuals in the aircraft industries /1/. However, due to the concept of having the transmitting and the receiving coil separated beside each other and not wound around the same axis, makes the sensitivity of this sensor dependent on the crack orientation: it is very sensitive to cracks parallel to the longitudinal direction of the sensor, but not sensitive to cracks vertical to it.
The detection of cracks running in several directions requires area scans with different sensor orientations or rotating sensors. A new multi-transmitter sensor allows the excitation of a rotating eddy-current density without a mechanical movement. For the generation of rotating magnetic fields, the measurement device contains several transmitter. Each transmitter is connected to a different transmitting coil of the rotating field sensor. The results presented in this paper were received by using 3 transmitter and a rotating field sensor with 3 transmitting coils. In order to generate a rotating field, the 3 transmitter are parallel triggered, sending a series of CS-pulses. Within a series of pulses of each transmitter, the amplitude of the pulses is changing in form of an amplitude modulation (in this case: sinus-modulation). By giving different phase shifts to the sinus-modulation of the 3 transmitter, the eddy current density beneath the 3 transmitting coils of the rotating field sensor will move. If 120° phase shifts are applied between the 3 transmitter (containing sinus modulation), the density of the eddy current field beneath the sensor can perform 360° rotations without a mechanical rotation of the sensor itself.
Figure 3 shows the rotating field sensor with the 3 transmitting coils in form of a symmetric triangle and the receiving coil in the middle. The measure in centimetre beneath the sensor describes the size and the distances between the coils.
This photo is showing the touch-down surface of the sensor, already fixed in the holding device of the automatic handling system. The arrangement of each single transmitting coil to the central receiving coil is in principle comparable to the sensor shown in Figure 2. Through enhancing the number of transmitting coils from 1 (figure 2) to 3 and placing them symmetrically around the receiving coil (figure 3), the sensitivity of this sensors is no longer dependent on the orientation of the cracks, but can detect the orientation and position of a defect.
|Linear grey scaling||Threshold scaling|
|Fig 4: Area scan along rivet row 1, (two different grey scaling)|
Measurement by using CS-eddy current technique with a conventional sensor
During the measurement an automatic handling system for the sensor placement was used, which is necessary to perform the measurements in form of area and line scans. Best results were received with a CS-pulse covering a frequency range from 1 kHz to 3 kHz. Since the evaluation of the maximal peak of the CS-pulse was already sufficient for crack detection, the time shift was not taken into account here. Beside the presentation of the evaluated pulse amplitudes (normalised) in a linear grey scaling, another scaling was chosen with selected amplitude levels for emphasising the defect recognition (threshold grey scaling). In both scaling the lighter colours stand for higher amplitudes of the peak of the received pulse. Figure 4 shows the results of an area scan over rivet row 1 in both grey scaling. Between the two scan presentations, an information concerning the crack length is given in tabular form.
The edges effects of the rivets lead to higher amplitudes compared with the areas between the rivets. However the orientation dependent sensitivity of the sensor only display the upper and lower edges, not the right and left ones. The cracks are indicated by a further increase of the amplitude, visible by the lighter colours. In practice these testing system could easily indicate cracks just by choosing an adequate threshold for the amplitude.
The threshold scaled presentation to the right in figure 7 additionally indicates the length of the crack with semi-dark grey colours between the rivets. With an increasing crack length, the black gap between to rivets is more and more closed by the semi-dark colours. This is already visible between the rivets 12 and 13 (12 mm crack length) and the gap becomes closer between the rivets 15 and 16 (15 mm crack length). In the end the continuous crack between the rivets 18 and 19 is indicated by a closed area of semi-dark colours.
Measurement by using CS-eddy current technique with rotating field sensor
For the presented measurements all three transmitter send a series of CS-pulses, each pulse in the same frequency range (1 kHz - 3 kHz) like the single pulse of the above presented results. The whole time for sending and receiving the series of pulses is adjusted by the settings of the sampling frequency, the time-resolution of the single pulse and of course the number of pulses: the whole time is in this case 26 ms for a 360° rotation.
|Fig 5: Pol-diagram of a defect-free measurement between the rivets|
A good demonstration of the performance of the rotating field sensor is given by the receiving signals from scans over a rivet, figure 6. The schematic drawing top-left in figure 6 visualise 5 sensor positions on and around a rivet.
Fig 6: Edge effect influence on the pol-diagrams of the rotating field sensor in defect free areas
Fig 7: Filter for suppressing the rivet edge effect
The centre pol-diagram (C in figure 6) presents the received series of pulses with the sensor directly over the rivet. Because of the rotation symmetry of the circular rotating field around the cylindrical rivet the receiving signal shows similar low amplitudes like Figure 5.
The other pol-diagrams were taken from sensor positions at the rivet edge, 2 mm away from the centre of the rivet. The received pol-figures posses higher amplitudes and the '8' is pointing to the rivet. The higher amplitude of the bottom pol-figure (B in figure 6) is caused by the position of the sensor close to the overlapping edge of the top skin.
Similar to the measurements with the conventional sensor, the presence of a crack is indicated by a further increase of the amplitude of the '8', as shown in the bottom-left pol-diagram of figure 7. However, the strong edge effect prevents the orientation of the '8' from following the crack influence ('8' vertical to the crack). This can be solved by using an adequate inverse filter to suppress the edge-effect on the orientation of the '8'. The top-right poldiagram of figure 7 shows a filtered pol-figure of a defect free measurement. This is equivalent to the pol-diagram in figure 5, showing the result from a defect free area without the disturbing influence of a rivet edge. The second part of figure 7 presents the result of a crack measurement after filtering the edge effect. In figure 7 bottom-right, the orientation of the '8' is determined by the crack and therefore showing an orientation of the '8' vertical to the crack.
The presented application of the CS-pulsed eddy current technique on multi-layered rivet-joint aluminium specimen from the aircraft industry demonstrates function and possibilities of these technique with very promising results. Compared with continuous waves, pulses allow a higher current excitation, which lead to a higher signal to noise ratio in the receiving signal and consequently increases the sensitivity. The CS-technique supports a suppression of interference effects. of the sensor instead of a excitation Beside a conventional eddy-current sensor, which is commonly used in aircraft industry for this task, a new rotating field sensor was introduced. This new sensor, which posses a sensibility independent from the orientation of the sensor to the defect orientation, proofed the ability of indicating the orientation of cracks.
The results of this investigation show, that the CS-eddy current technique especially in combination with the rotating field sensor is a good approach to tackle the problem of bending crack detection in multi-layered aluminium structures.