Cracks occur most often near the bottom of the tank where there is a change in wall thickness. Simulations using CIVA were performed to determine the optimal probe configuration and inspection strategy. One proposed solution is to use a shear-wave inspection at 45 degrees using a linear array with a wedge (see below), combined with electronic scanning. As a first step, the beam profile is optimized using CIVA to gain maximum resolution in the area of interest. The number of elements to be fired at the same time and the focusing delay laws are deduced from analysis of the beam calculations.

Once the beam is optimized, wave-defect interactions are calculated (see animation above). A reference case is simulated (usually a calibrated side-drilled hole) and a parametric study is conducted to determine the effect of crack orientation. Here the crack is rotated from –20 to +20 degrees, with a 10º step. Results are shown in the plot below, which shows echo amplitude versus crack orientation.

By correlating these simulation results to experimental measurements on the calibrated defect, the sensitivity of the procedure can be inferred; i.e., the inclination range that can be detected. Note that similar parametric studies can be performed to determine the smallest resolvable defect, and the coverage zone.

Practical example:

For equipment with a 80-dB dynamic range, let’s suppose that the calibrated defect is observed experimentally at a 50dB gain with an echo at 100% of the screen height and a noise below 5% of the screen height.

Let’s now assume that any full-height echo measurement is acceptable as long as the noise level is below 30% of the screen height.

As a first approach, the remaining dynamic range to keep the measurement acceptable is:

20xlog (30/5) = 15.5 dB

Any echo within a 15.5dB range of the side-drilled-hole echo will therefore be accepted. Going back to the simulation results that show that any echo from a rotated crack (within ± 20º) remains within 10 dB of the reference echo (see plot), we can conclude that all defects with rotations within this range will be detected using the proposed inspection strategy.

The conventional inspection technique requires the fastener to be removed. The challenge in developing an easier and less expensive inspection strategy is to design a technique that can be used from the skin side, that does not require removal of the fastener, and that provides the same or better resolution than the conventional method (see Figure 1).

Figure 1: Generic aircraft fastener joining the skin and spar. Large stresses in the fixation area make it a potential site for crack initiation.

The phased-array concept is to use a large linear array in a pseudo-tandem configuration, in which different elements of the same probe are used for transmission and reception [2]. The transmitting elements are phased to achieve a focused incident wave along a sectorial scan path after reflection off the bottom surface (see Figure 2).

Figure 2: Transmission delay laws are computed to focus shear waves after reflection off the bottom surface.

The reception elements are phased to perform a focused sectorial scan (see Figure 3).

Figure 3: Reception delay laws are computed to focus shear waves directly beneath the probe wedge.

The convolution between the transmitted and received signals defines the active focal spot for the measurements (see Figure 4).

Figure 4: Resolved focal points using separate transmission and reception delay laws. The entire thickness is covered by the inspection procedure. The linear array replaces a large number of tandem probes (one for each thickness).

The objective of combining the pitch-catch mode and sectorial scanning in this way is to optimize both the zone coverage and the consistency of the resolution. The optimized inspection strategy uses a 96-element linear array at 5MHz (see animation in Figure 5).

Figure 5: Ray tracing of the focal points using separate transmission and reception delay laws. The entire thickness of the part is covered by the inspection procedure.

Advantages:

• Requires only one probe for the complete inspection.
• No dead zones.
• Tunable focusing for transmission and reception.
• Adaptable to any geometry.
• Consistent resolution through the whole thickness.

The first inspection strategy considered used the same number of elements (40) for both transmission and reception. Simulation results calculated using CIVA (see Figure 6) show good consistency of the focal-spot size throughout the thickness, with a 7dB-amplitude variation between the top and bottom focal spots and a 4mm-diameter focal spot at -6dB. The results also show a lack of resolution in transmission underneath the front surface and in reception at the bottom surface.

Figure 6: 40-element configuration. Beam visualization shows sufficient resolution for crack detection.

Using these results, a new strategy was designed that uses 54 elements for transmission and 42 elements for reception. The elements are phased in both transmission and reception to improve resolution. Results (see Figures 7, 8 and 9) show better resolution throughout the entire thickness (2mm-diameter focal spot at -6dB), as well as better consistency (2.5 dB variation from top to bottom). The linear array replaces a large number of tandem probes (one for each thickness).

Figure 7: Visualization of the focused beam after reflection off the bottom surface using 54 elements for transmission.

Figure 8: Visualization of the focused beam using 54 elements for reception.

[1] Neau G., Hopkins D., Tretout H, and Boyer L., “Phased-array applications for aircraft maintenance, manufacturing and development”, Aerospace Testing Expo 2006, UKIP Media & Events 2006.

[2] Mahaut S., Chatillon S., Raillon-Picot R. and Calmon P., “Simulation and application of dynamic inspection modes using ultrasonic phased arrays”, Review of Quantitative Nondestructive Evaluation Vol. 23, ed. by D. O. Thompson and D. E. Chimenti, American Institute of Physics, 2004.

Figure 1: T-section stiffener. Expected defects are located inside the stiffener. The corner piece prevents inspection using only normally incident waves.

Without the corner part, the inspection could be performed with good accuracy using normally incident pressure waves. However, with the corner piece, pressure waves at normal incidence cannot sample every zone of the stiffener, and the shadow zone (or silent zone) is too large to be acceptable. The inspection strategy must therefore to be adapted to be able to detect defects that lie underneath the corner piece.

One solution is to use a linear phased-array probe. By performing a sectorial scan and using delay laws to steer the acoustic beam, the full width of the part can be inspected and the formerly silent zone is eliminated. As illustrated in Figure 2, cracks lying underneath the corner part can be detected using this procedure. In this figure, the inspector can observe the multiple reflections of the thin part (occurring at normal incidence, on the left side of the sectorial scan), as well as other structural reflections (at high steering angles, right side of the sectorial scan).

Figure 2: Sectorial scan superimposed on a cross section showing the stiffener geometry. The sectorial scan allows cracks lying beneath the corner piece to be detected.

Image depicts crack detection in an aluminum sample using an M2M system, an Imasonic probe and a displacement encoder. Courtesy of Dassault Aviation.

Our largest system is a fully parallel 128-channel system. With some development, we can provide a fully parallel 256-channel system.

Our product brochures can provide you with the technical specifications of each system.

• Multi-X and Multi-2000
• Multi-2000 Pocket

Image shows inspection of an advanced composite at Dassault Aviation, using an M2M system to drive the linear array probe. Used by permission.

Image shows a radiated acoustic beam using a sectorial probe. 64 elements are used to compute the delay laws and the acoustic field. The beam can be focused at any location in the specimen.

Modeling results obtained using CIVA (developed by the Commissariat à l’Energie Atomique in France) are used here to illustrate some of the unique features of phased array systems, and to demonstrate how modeling can be used to determine the optimal inspection strategy, which, in turn, can be used to specify the appropriate probe and to determine hardware needs.

Understanding and visualizing the beam radiated in the test specimen

The example discussed in detail here is the use of the same linear-array probe with two different sectorial scanning strategies. In this case, simulation is used to understand and visualize the beam shape, to help the engineer find the optimal inspection procedure. The simulation image shown in Figure 5a is the acoustic beam resulting from firing 7 elements (of a 64-element linear array) with focusing at a distance of 35 mm. The images correspond to the case where the probe is used with a wedge angled at 45° on a steel specimen. The acoustic beam shown in the upper-left image of Figure 5 is from the first shot in a sectorial scan. For subsequent shots, the beam is steered in increments of one degree up to 70 degrees, while maintaining the focal point at a distance of 35 mm (the middle and final shots of the sequence are displayed in the center and lower-left images of Figure 5). What the simulation shows is that the beam is not well focused, meaning that resolution and the ability to size defects will not be optimal with this configuration. In addition, a side lobe is evident (shear wave at 45 degrees), that becomes more and more significant for angles greater than 62 degrees. The creation of side lobes results in signals that are more complicated and generally more difficult to interpret.

To improve the inspection, simulations were run using different numbers of elements to optimize the beam in the sample. Recall that the right-hand column of Figure 5 shows the ultrasonic beam obtained using 16 elements focused at a fixed distance of 35 mm for each angle in the sectorial scan. By comparing the left- and right-hand columns, it is easy to see that the beam in the second case (left-hand column) is much better focused, which allows detection of smaller defects and improved sizing. Using the -6dB sizing technique, the focal spots can be determined and compared for both cases.

Figure 5: Radiated beams (shear waves) at 45, 58 and 70 degrees (first to third rows, respectively) for 7- and 16-element probe configurations (left- and right-hand columns, respectively). The dots on the images indicate the targeted focal points.

Although this relatively simple case might not warrant a modeling study, the complex geometries encountered in practice, along with physical constraints that limit access, make modeling an extremely valuable tool for determining optimal inspection strategies. For example, in those cases were access to the part is limited, it is very useful to be able to determine the minimum size and number of elements necessary to perform the required measurements. For the case presented here (Figure 5), it is possible to compare the 7- and 32-element configurations to determine the optimal tradeoff between size and detection capability.

Wave-defect interaction: evaluating the sensitivity of an NDT procedure

Using CIVA simulation software, it is not only possible to characterize the acoustic field for any phased-array configuration, but it is also possible to determine the sensitivity of the proposed inspection procedure. Even with sophisticated modeling tools there is still a need for calibration experiments, but they can usually be reduced to validation experiments performed on reference specimens (for example, a block with side-drilled holes). The reference test specimens are modeled and the results are compared to experimental measurements.

Figure 6: Sectorial scans (top images) and dynamic echo curves (graphs below scans). Laboratory measurements are displayed on the left, and the results of the corresponding simulations are shown on the right. Experimental and simulation results are within 1 dB agreement.

The CEA is continually validating CIVA with experimental data [1], and the results displayed in Figure 6 are an example of how modeling results are validated. In this case, experimental and simulation results are shown for an aluminum block containing side drilled holes obtained from a focused, sectorial scan using 40 elements of a 64-element Imasonic probe. The sensitivity of proposed inspection protocols is determined by quantifying the defect response in terms of gain compared to the reference case (calibrated defects); i.e., if the gain required to identify the defect is within the dynamic range of the phased-array controller, then it will be possible to detect the defects in question. A series of parametric studies is often carried out, for example, to study the dependence between detectability and the size of the defect, its orientation, and/or its geometry.

1. Mahaut S., Chatillon S., Kerbrat E., Porre J., Calmon P. and Roy O., “New features for phased array techniques inspections: simulation and experiments”, Proceedings of the WCNDT, 2004.

Less movement, faster inspection, better reliability

The advantages of phased-array systems include the ability to perform electronic scanning of the ultrasonic beam, which can reduce inspection times by eliminating or reducing the need move the probe. As illustrated in Figure 3, electronic scanning is accomplished by firing successive groups of elements in the array. A complete C-scan image can be obtained with a matrix phased array with the probe in a fixed position. The reliability of inspections can also be improved by reducing the need to move the probe. As is well known, good coupling between ultrasonic probes and the part undergoing inspection is crucial for good acoustic measurements. Each time the probe is moved, there is a risk of losing or degrading coupling. Thus, minimizing the number of times the probe is moved helps to maintain uniform conditions for multiple measurements.

Real-time imaging, easier interpretation

Phased arrays allow a broad spectrum of inspection strategies that improve performance, for example, sectorial scanning and focalization after reflection off the back surface of the test specimen. The most advanced phased-array systems include tools such as dynamic-depth focusing. With real-time imaging, inspections are easier to perform and the reliability of the measurements is also greatly improved. Because thousands of signals are captured and displayed at once, the struggle that operators often have in locating and visualizing defects on the screen is greatly reduced. In addition, the number of false alarms is diminished because of reduced operator dependence, and data recording and traceability are improved. Experimental results obtained using a sectorial scan are shown in Figure 4. Measurements were performed using 32 elements of a 64-element linear array with a frequency of 5 MHz. The test specimen was an aluminum reference block containing planar defects.

Applying delay laws to improve performance and simplify procedures

Figure 4: Example of sectorial scanning used for crack sizing. The solid rectangular line indicates the geometry of the test specimen under examination, including two parallel saw cuts. The corner echoes (labeled “a”) resulting from the cuts were easily detected using 32 elements of a linear array with a central frequency of 5 MHz. Diffraction from the crack tips (labeled “b”) is only observed when the beam is appropriately focused and directed.

Phased arrays can also replace an entire tool kit of conventional transducers. A single phased array used in conjunction with appropriate delay laws can reproduce the same acoustic beams achieved with numerous conventional probes, while also providing greater functionality. Using a phased-array controller that allows several types of delay laws per inspection, the results of several different sets of measurements that comprise a complete NDT procedure can be visualized simultaneously in real time.