Ultrasonic techniques for hidden corrosion detection in aircraft wing skin

INTRODUCTION

Ultrasonic inspection is a non-destructive method in which beams of high frequency sound waves that are introduced in to the material are used to detect surface and sub surface flaws. The sound waves travel through the material with some attend loss of energy and are reflected at interfaces .the reflected beams is detected and analyzed to define the presence and location of flaws. Different ultrasonic techniques are there for finding the flaws or defect in which guided waves demonstrate an attractive solution where conventional ultrasonic techniques are less sensitive to defects such as corrosion/disbands. In thin multilayered wing skin structures and hidden exfoliation under wing skin fasteners. Guided lamb waves have been used for long range or large area corrosion detection and the evaluation of adhesively bonded structure. Ultrasonic guide waves are promising but require procedure development to ensure high sensitivity and reliable transducer coupling.

ULTRASONIC TESTING
Basic principle and general application
Most commercial ultrasonic testing is done at frequencies between 1 and 25 MHz. The elastic wave travel in to the material with some loss of energy due to attention and are reflected at the interfaces. The reflected beam is analyzed to detect and locals the defect and for their quantitative evaluation. These waves are completely reflected at solid-gas interfaces, partial reflection at solid-liquid or solid-solid interfaces. The reflected energy depends mainly on acoustic impedance of the material at the interfaces and the acoustic impedance of a given material is the product of density and velocity. This widely used NDT method has a lot of applications like defining bond characteristics, measurement of thickness of the components, estimation of corrosion and determination of physical properties, structure, grain size and elastic constant. In the case of austenitic stainless steel welds because of the problems associated with high scattering and attenuation ultrasonic waves at low frequencies are employed. For successful application of ultrasonic inspection, the testing must be suitable for type of inspection being done and the operator must be successfully trained and experienced.

TYPES OF ULTRA SONIC WAVES
On the basics of the mode of vibration or particle displacement ultrasonic waves are classified as longitudinal waves, transverse waves, surface waves and lamb waves
Longitudinal waves
Some times called compression waves are the type of ultrasonic waves most widely used in the inspection of metals. They travel through metal as a series of alternate compression and rarefaction in which the particle transmitting the wave vibrate back and forth in the direction of the travel of the waves. Because of the easy generation and detection, this type of ultra sonic waves is most widely used in ultra sonic testing. These types of waves can propagate in solids, liquids and gas.
Transverse waves.
This type of ultrasonic waves is called transverse wave or shear wave because the direction of particle displacement is at right angles to the direction of propagation. Unlike longitudinal waves transverse wave cannot be supplied by the elastic collision of adjacent molecular or atomic particles. For the propagation of transverse wave it is necessary that each particle exhibits a strong force of attraction to its neighbors so that as the part moves back and forth its pull its neighbors with in thus causing sound to move through the material with the velocity associated with the transverse waves which is about 50% of the longitudinal wave for the same material
Surface waves
They are another type of ultrasonic waves used in the inspection of metals. These waves travel along the flat or curved surface of relatively thick solid parts. In the surface waves, particle vibration generally follows an elliptical orbit. Surface waves are subjected to less attenuation in a given than are longitudinal or transverse waves. They have a velocity approximately 90% of the transverse wave velocity in the same material.
Lamb waves
They are also known as plate waves, are another type of ultrasonic waves used in the non destructive inspection of metals. The propagation characteristics of lamb waves depend on the density, elastic properties and structure of metals and also are influenced by the thickness of the metal test face and the cyclic frequency. Their behavior in general resembles that observed in the transmission of electromagnetic wave through wave guides. There are two basic forms of lamb waves
a) Symmetrical: Here compressional particle displacement along the neutral axis of the plate and elliptical particle displacement on each surface
b) Asymmetrical: Here there is a shear displacement along the neutral axis of the plate and an elliptical displacement, particle displacement on each surface.
Ultrasonic inspection is conventionally used for corrosion detection in aircraft wings. But the conventional inspection method carries with it certain defects like:
(i) It scans perpendicular to the surface and hence rate of scanning (from point to point) is less and hence highly time consuming.
(ii) Conventional method is not capable of detecting disbonds between layers and cracks at fastener holes.
These defects are over come by a newly developed inspection method using guided ultrasonic waves.

GUIDEDWAVES

Propagating wave packets which are superpositions of various modes are often called guided waves. There are various types of guided waves available in practice. Wave packets, resulting from appropriate stress and strain boundary conditions, which travel on the surface of a solid body, are know as surface waves. Surface waves usually exhibit large amplitudes and travel slower than other types of guided waves.
Rayleigh waves are the best known surface waves. They are non-dispersive for uniform material properties. However, their mechanism of propagation is very complex; waves are polarized and surface particles are moved around an ellipse. The components of Rayleigh waves can couple with a medium surrounding the surface of the body. This coupling affects the amplitude and velocity of the wave. The amplitude of the wave decreases rapidly with depth. The rate of decrease depends on the wavelength. Therefore inspection methods based on Rayleigh waves are used mostly to detect surface defects.




Guided waves demonstrate an attractive solution where conventional ultrasonic inspection techniques are less sensitive to defects such as corrosion/disbonds in thin multilayered wing skin structures and hidden exfoliation under wing skin fasteners. Moreover, with their multimode character, selection of guided wave modes can be optimized for detection of particular types of defects. Mode optimization can be done by selecting modes with maximum group velocities (minimum dispersion), or analysis of their wave mode structures (particle displacements, stresses and power distributions). Guided Lamb modes have been used for long range/large area corrosion detection and the evaluation of adhesively bonded structures. Ultrasonic guided waves are promising but require procedure development to ensure high sensitivity and reliable transducer coupling and to provide a mechanism to transport the probe(s) over the area to be scanned. This paper describes some practical inspection setups and procedures based on guided wave modes for corrosion damage detection in single and multilayered wing skin structures and exfoliation detection immediately adjacent to fasteners in aircraft wing skin. It describes the results of their application to detection of corrosion in simulated and real components of aircraft wing skin. Using a tone burst system, the wave modes are selected, excited and tested in pulse echo and pitch catch setups. Launch angles were obtained from the calculated dispersion curves. Theoretical group velocities were compared to tested group velocities to confirm the excited modes at frequency thickness product and launch angle. The simulated corrosion in single and multilayered wing skin structures and exfoliation located under several rivets was successfully detected. Some guided Lamb modes proved to be more sensitive to corrosion type defects and produced better results.

Monitoring Strategy

Real engineering structures under inspection are usually quite complex when compared with simple plates studied in laboratory conditions and reported in the literature. Various types of joints, stiffeners, rivets, complicated shapes or varying thickness determine the complexity. This causes the entire analysis to be much more complicated and requires an appropriate monitoring strategy.
The methodology or strategy of monitoring is extremely important for successful damage detection. The basic factors, which determine the Lamb wave based damage detection analysis are related to properties of the structure under inspection, transducer schemes, choice of excitation input signal, and appropriate signal processing. Other elements include various aspects related to transducer coupling methods, optimal sensor locations and sensor validation procedures. The last but not least is the hardware used for monitoring, graphical interface and data storage organization.
The dispersive nature of lamb waves and also the finite number of modes at a given frequency makes long-range inspection very difficult. To overcome these problems, low frequency-thickness products are often utilised for damage detection. In this case only two fundamental modes A0 and S0, are used. It is important to limit the bandwidth of the excitation to a range over which there is little dispersion (i.e. the phase velocity does not change significantly with frequency).
The first applications of Lamb waves for damage detection used bulky wedge transducers. It appears that piezoceramic elements are now the most widely used transducers due to the fact that they can be used as sensors and actuators at the same time. Often piezoceramics can become an integral part of the monitored structure. Recent advancements in this area include optical fibre sensors and MEMS (Micro-Electro-Mechanical Systems) sensors. Both types of sensors can be used not only for Lamb waves detection but also for strain measurements. MEMS devices can additionally generate Lamb waves. A number of different practical aspects need to be considered once choosing transducers for Lamb waves detection. This includes: coupling, connectors and environmental protection. The cold bonding is usually preferred to hot bonding. The process of bonding must be as easy (if not easier) as the procedure for bonding strain gauges. Also, bonded sensors are better than embedded due to possible sensor failures and replacements. Reliable connectors and environmental protections are required to prevent sensor failures. Wireless applications are possible with piezoceramic and MEMS sensors. Coupling, connectors and environmental protection are particularly important in the case of optical fibre and piezoceramic sensors.
Different types of signals are used as input excitation. It is considered that the simpler the input signal the simpler the output signal for damage detection. The choice of input excitation is often a compromise between the amplitude and the mode generation. Low-voltage signals are possible when the input frequencies are within transducer resonance frequencies. This is often associated with intelligent signal processing to remove undesired noise and extract features related to damage. In practice transducers resonance frequencies do not coincide with single Lamb wave mode frequencies. Previous studies show that even a simple input signal can lead to complex output signals due to various attenuation and dispersion effects which are not related to damage. This clearly shows that intelligent signal processing is one of the most important elements of the Lamb waves based damage monitoring strategy.
Once the transducers excitation signals are chosen the question is where to put sensors for optimal damage detection. Recent years have shown considerable progress on the problem of determining the number and location of sensors in engineering structures.

CORROSION DETECTION WITH GUIDED WAVES
Guided Lamb modes are dispersive waves and their velocities are a function of the frequency thickness product. Therefore, any material changes such as corrosion/exfoliation or lack of adhesion between two layers will affect the propagating mode amplitude, velocity, frequency spectrum and its time of flight.
RF waveforms from guided modes going through a corroded area have a relatively low transmitted signal amplitude and time of flight shift, while no corroded areas are associated with stable time of flight & high signal amplitude.

.

Inspection of lap splice joint with guided waves in a pitch catch setup permits a selected guided wave mode to travel from the sender toward the receiver probe, producing relatively low amplitude RF signal when corrosion exists between the two bonded parts. Otherwise, if there is no corrosion, the excited mode will leak into the second joint producing relatively high amplitude RF signal (Figure 1). In a pulse echo setup, a low RF signal is obtained in the presence of corrosion and high RF signal is obtained for absence of corrosion.


Figure 1 Transmission results from a) non-corroded area b) corroded area

Fatigue cracks and exfoliation under the shadow of fastener heads in aircraft skin structures can be detected using ultrasonic guided waves. Guided modes are selected and launched from outside the exfoliated and hidden area to interrogate the interested rivets. In pulse echo setup, the received modes associated with RF signals include indications and reflections from exfoliation.

EQUIPMENT AND INSTRUMENTATION
The system used in our experimentation is Tektrend's PANDA® Guided Wave System (Figure 2). The new PANDA® Guided Wave System unit is an advanced modular and portable automated scanning system. It can be configured for conventional UT and ET transducer positioning, providing C scan images. The PANDA® can be configured for guided wave inspection, providing cost effective, practical nondestructive evaluation.




The PANDA® Automated Scanning System is self contained in a single unit in which all the electronic boards are mounted in the system computer workstation. It offers advanced analysis and interpretation capabilities, where intelligent scans can be performed with a pre designed intelligent classifier. The system contains tools to tag signals for export to an integrated pattern recognition package. The positioning control, ultrasonic control, data acquisition, displays and analysis software are all integrated into a single software package, ARIUS IV®.
The Guided Wave System is hosted on flexible rail to allow scanning of curved surfaces and to enable complete automation of the ultrasonic field inspection. An adaptable spring loaded piston design for holding transducers is mounted on the Y axis scanning arm, which moves on the X¬- axis. The system is fitted to the inspected surface with a vacuum control system. The PANDA® Arm can operate in vertical and horizontal orientations and scan contoured and edged surfaces. Measurement can be made in pulse echo as well as pitch catch modes with piezoelectric transducer probes (optional with EMAT probes) with 0.005 and 0.002 inch maximum scanning accuracy and resolution with a maximum scanning rate of 6 inches/second at maximum resolution. The transducer probes are driven by a tone burst pulser to excite narrow band guided wave modes and to provide high power to launch the wave over long distances. With tone burst excitation, the operating frequency and the pulse characteristics of the transmitter can be controlled in a repeatable manner.

INSPECTION RESULTS
Detectability of corrosion in aircraft wing skins was investigated for three cases. One layer corrosion using controlled thinning areas, two layers corrosion detection in lap splice joints and corrosion detection under fasteners of wing skin structures. Tests were performed using three aluminum specimens with different types of simulated corrosion.
(i) Layer corrosion using controlled thinning areas:
The first specimen represented 460x405x I mm aluminum plate with controlled thinning in designated areas. To demonstrate the sensibility of the excited wave modes, corrosions were induced in three places with different levels of thinning (10%, 15% and 25%). Measurements were made using the pitch catch setup which consisted of two variable angle broadband transducers with central frequencies at 3.5 MHz, one of the transducers acts as transmitter used to generate the guided wave mode and the other one was used to receive the generated mode and its interaction with the corroded structure.
The first set of tests demonstrates detectability of the open corrosion on the aluminum plate using the pitch catch setup with piezo composite transducers to generate the A1 mode at 2.2 MHz with an incident angle of 200. Figures 3 b, 3 c and 3 d show the RF waveforms obtained with transducers positioned perpendicular to the corroded areas (three locations), while Figure 3a shows RF waveform obtained with transducers perpendicular to the noncorroded area.
A1 guided mode signals passing through the corroded area have a transmitted signal low amplitude and higher time of flight which is consistent with theoretically calculated group velocity dispersion curves, while signals from the noncorroded area are associated with stable time of flight and high received signal amplitude. Therefore, wave propagation behavior in corroded areas allows estimation of the percentage of the corrosion material loss. Mode selection and optimization can improve the resolution of material loss estimation



(ii) Corrosion detection in lap splice joints
Tests were also carried out on 406x322x1.0 mm lap joint (Figure 4). The width of the bonded area was 68.5 mm. The lap Joint was assembled and subjected to accelerated corrosion in a salt fog chamber.
Guided wave inspection was performed on the lap joint specimen and inspection results were evaluated in terms of the sensitivity and repeatability. Scanning was carried out over the sample illustrated in Figure 4 along the X direction using two transducers in the pitch catch setup to excite the S0 mode at 1.5 MHz.

The corroded area between the second and the first aluminum layers created a disbond and permitted bad transmission of the generated mode from the sender toward the receiver without any energy leakage in the additional bonded aluminum layer. In the noncorroded area, there was a good bond between the second and the first layer; therefore, the transmitted signal amplitude was attenuated due to leakage of the transmitted energy into the second layer.
Figure 5 a shows single line modified C scan results of this inspection and presents a series of signals in three dimensional format. Transducer displacement (X-¬direction), time of flight (Y direction) and signal amplitudes (Z direction). The well bonded (non corroded) areas are characterized by high amplitude signals (signals indicated by red colour). Poorly bonded areas (caused by corrosion) resulted in a reduction of amplitude of the received signals as shown by the low amplitude echoes at both ends of the specimen. The high and low amplitude signals are represented by the lighter and heavier colors, respectively. The interruptions between signals in Figure 5a are due to the presence of rivets. To verify the guided wave results, these specimens were also inspected using, an eddy current technique as well as an enhanced optical technique (D Sight). Corrosion was detected in the two ends of the specimen by both techniques as shown in Figure 5b and 5c. The red and orange colors in the eddy current image show areas of severe corrosion while the green and blue represent areas having very light corrosion. In the D Sight image, the existence of corrosion is inferred by the presence of waviness (pillowing) between the rivets, which is caused by the formation of corrosion products (aluminum oxide and hydroxide) at the interface between the two plates.
(iii) Corrosion detection under fasteners of wing skin structures

The third series of tests were performed on fasteners of wing skin structures to detect corrosion damage immediately adjacent to the fastener holes in airframe structures as shown in Figure 6. Fatigue cracks commonly initiate at fasteners since high stresses around it are created. Water and humidity then are infiltrated to create exfoliation and corrosion around and under the rivets. As the guided waves penetrate within and beyond the region of the fastener head, ultrasonic energy is reflected from discontinuities (corrosion, mechanical damage) present in the region of interrogation.
In this test, once again, a linear manual guided wave scan was performed by moving a single transducer in a pulse-echo mode at 3.5 MHz with in incident angle of 370 along the specimen in the Y direction parallel to the fastener row at a distance varying from 0.1” to 0.5” from the line of holes (Figure 7). The displacement of the wedge/transducer assembly was performed using the PANDA® automated scanner shown in Figure 3 which encoded position in both the x and y directions. One full RF waveform was acquired at every 0.12mm along the scan path. The RF waveform was digitized at 100 MHz and contained 2048 points. The acquired signals were averaged and filtered and all the data for each scan were saved in a file for later retrieval and analysis.



Figure 8 shows single line of scan results of this inspection. The image is color coded according to the reflected amplitude (ultrasonic energy); i.e., blue corresponds to minimum reflected energy and white to maximum (Figure 8). The time scale increases vertically from top to bottom and the horizontal scale corresponds to the scan displacement at an increment.
The reflected energy front the left hand Cluster shows a trail of small reflections on both the left and right of the fastener. These regions are indicated in boxes in Figure 8. These reflection trails are clearly distinguishable from the indication of a defect free cluster shown on the right in Figure 8. Interpretation of the fastener hole integrity is based on the presence of a trailing shadow below the fasteners on either side of the main reflection. Although the exfoliation reflectors are more diffuse than the discrete reflectors provided by crack-like defects, the indications are clear.
Performance and repeatability tests were performed on similar specimens. The initial inspection and immediate interpretation provided 46% identification rate of all defects in the 15"x12"x0.2” specimens (68% If we include the possible defects) with two false calls. Based on the experience obtained during Inspection Sessions, a subsequent interpretation session gave a detectability score of 90% with 5 false calls. The false calls were subsequently attributed to coupling inconsistencies and possibly stray signals produced by the presence of the stringer attached to the specimen.











CONCLUSION
A practical inspection procedure was demonstrated using guided waves for fast and effective inspection to detect and locate defects in layered aircraft structures. Lamb wave inspection can be carried out either by using two probes in pitch catch or one probe in pulse echo configurations. It can detect corrosion in lap-splice joints in a single scan and the procedure setup is suitable for presentation of the results as an image relating the amplitude and time of flight to facilitate interpretation. It has also the capability to detect the extend of corrosion.
Results form exfoliation under the shadow of fastener heads was detected using ultrasonic guided waves launched from outside the area with imaging to assist in interpretation. However, results form thicker tapered wing skin specimens were not conclusive, the guided wave technique did not seem to apply appropriately to these samples. It appears that some bulk shear components dominated the scan results and provided extra reflection from the countersink and the exfoliation. It also suffers from the drawback of the need of highly sensitive and reliable transducers.





















REFERENCE

1) M.Brassard, A.Chahbaz and A. Pelletier”Combined NDT Inspection Techniques for Corrosion Detection of Aircraft Structures”, 15th world conference on NDT, ROMA (ITALY), 15-21 October 2000.
2) Metals hand book, 8th edition vol- 2
3) Non-destructive evaluation-A tool in design, manufacturing and service by Don E.Bray & Roderick K.Stanley.
4) NDTnet 1998 June, Vol.3 No.6