Inspection of Adhesively Bonded Aircraft Repair ...

Technical Paper

Inspection of Adhesively Bonded Aircraft Repair Patches using Ultrasonic Guided Waves Padmakumar Puthillath, Cliff J. Lissenden and Joseph L. Rose Engineering Science and Mechanics Department 212, EES Bldg., Pennsylvania State University, PA 16802, U.S.A Email: [email protected]

ABSTRACT Aircraft structures subject to service loads and chemical environments develop structural weaknesses like cracks, corrosion etc. In order to mitigate the resulting reduction in the useful service life of aircrafts, repairs are done in the form of adhesive bonding of plates such as a titanium plate to aluminum fuselage. Typical of adhesive joints, these bonded repair patches also have interfacial (adhesive) or bulk (cohesive) weaknesses in the joint. Nondestructive inspection of these structures is required to ensure the quality of the repair patches. Conventional inspection approaches using normal or oblique incidence of ultrasonic waves has limited capability in detecting both the adhesive and the cohesive weaknesses present in the adhesive joints. A systematic approach for the inspection of adhesive repair patches using ultrasonic guided waves is demonstrated in this work. Among the multiple modes and frequency combinations possible in a structure, defect sensitive guided wave modes were selected from theoretical studies. The selected mode6frequency combination is ensured to work over a range of bondline thicknesses that can occur in real joints. An angle beam wedge arrangement to generate defect sensitive modes was used to successfully inspect epoxy bonded titanium 6 aluminum samples prepared with simulated adhesive and cohesive weaknesses. Keywords: Adhesively Bonded Repair Patch, Ultrasonic Guided Waves, Dispersion, Adhesive and Cohesive Weakness

1. INTRODUCTION Aircraft structures are subject to in6service loading like fatigue, thermal and chemical environments that can initiate points of weakness within the structure, such as fatigue cracks, corrosion, and delamination, thus leading to a reduction in their service life. In the military, aging induced structural weaknesses in aircrafts are mitigated using appropriate repairs because replacement is prohibitive in terms of time and cost [Pyles 2003]. Repairs can be performed using mechanically fastened or adhesively bonded patches. In comparison to mechanical fastened repairs, a bonded repair produces minimal alteration to the aerodynamic contours, results in weight savings in addition to avoiding the stress raisers associated with bolt/rivet holes. Adhesively bonding metal or composite patches to the damaged surface of aircraft, after appropriate surface treatment, can improve the stiffness of the weakened part [Pyles 2003]. The adhesive bonding used is susceptible to interfacial (or adhesive) and bulk (or cohesive) defects, making nondestructive inspection essential in order to ascertain the quality of repairs. Practical cases of adhesive repairs can be found in the literature [Baker 2009, t’Hart and Boogers 2002]. Recently, researchers have demonstrated a method for health monitoring of repairs using strain gages bonded to the aircraft skin and repair patch [Baker 2009]. The ratio of strains was used to detect debonding between the skin and the repair patch. This is a good approach for local inspection, but requires either an extensive coverage of the repair patch using strain gages or optical fibers to obtain a global inspection. The use of commercial bond Journal of Non destructive Testing & Evaluation

testing equipment like Bondascope [Baker 2009] and the Fokker Bondtester [t’Hart and Boogers 2002] in successfully detecting cohesive defects is also shown in the literature. This is again a local inspection approach, and similar to an ultrasonic C6scan, is not successful in detecting adhesive weakness in bonding. Ultrasonic wave propagation through structures is dependent on the material elastic properties. Ultrasound provides a nondestructive means of adhesive bond quality assessment. Pilarski and Rose [1988] have shown the importance of generating shear at the interface between the adhesive and adherend. This was an improvement of the bulk wave approach that needed very high frequencies (> 10 MHz). Ultrasonic guided waves are special kinds of waves propagating primarily under the influence of the geometry and boundary conditions of a waveguide. They are characterized by dispersion which is captured in the form of phase and group velocity variation with frequency [Rose and Pilarski 1988]. Rose and co6workers [1996] have successfully demonstrated mode selection principles by employing modes from the overlap between dispersion curves of the individual plate that form the adhesive bond. This study comprehends the progress made in mode selection for inspection of defects in an adhesive joint 6 titanium patch bonded to aluminum aircraft skin using epoxy. In this paper, a theoretical study is carried out where the guided wave phase velocity dispersion curves are used in conjunction with wave structures to determine optimal conditions for inspection of adhesive and cohesive Vol. 9, Issue 3 December 2010

Technical Paper

weakness in continuous waveguides. Epoxy bonded aluminum 6 titanium repair patches were prepared with interfacial weakness conditions simulated by using teflon inserts and other surface variation techniques. The inspection technique presented here is applicable to the inspection of bonded repair patches under the condition that both the transmitter and the receiving transducers rest on the bonded joint. The optimal guided wave mode was generated in the bonded sample using an ultrasonic transducer mounted on an acrylic angle beam wedge. The difference in transmission in terms of the signal content was successfully analyzed and used to discriminate between the defective and non6defective regions in the structure. This work has been recently reported by Puthillath and Rose [2010].

2.

GUIDED WAVE MODE SELECTION

Ultrasonic guided wave dispersion curves provide the theoretically possible phase velocity and frequency combinations that can exist in a structure having free

boundaries. Each point on a dispersion curve has a unique wavestructure and it holds the potential to solve different inspection problems. In the literature there have been instances [Rose and Pilarski 1992] where the guided wave mode selection has been carried out to address different defect detection scenarios. The Lamb wave phase velocity dispersion curves for a typical adhesive repair patch 6 epoxy (0.66 mm thick) bonded titanium (1.6 mm), aluminum (3.175 mm) joint – is shown in Fig. 1. Since the repair patch geometry under study is not mid6plane symmetric, the modes are referred by numbers rather than the conventional Antisymmetric (A) or Symmetric (S) notation. In this work, both the adhesive and the cohesive weakness in the epoxy layer are studied. The adhesive weakness is assumed to be located between the aluminum and the epoxy layer. This assumption comes from the fact that repairs are performed on field and hence the possibility that the surface of aluminum does not meet the cleanliness condition required for bonding is high.

Fig. 1 : Lamb wave dispersion curves for aluminum6epoxy6titanium adhesive repair patch and two wave structures or cross6sectional displacement profiles (at locations 1 and 2 on the dispersion curves). The dotted lines demarcate the aluminum, epoxy and the titanium regions, with aluminum being at the bottom. A larger in6plane displacement (ux) at the aluminum6epoxy interface can be noticed at location 2. Vol. 9, Issue 3 December 2010

Journal of Non destructive Testing & Evaluation

Technical Paper

In order to inspect the interfacial weaknesses at the aluminum6epoxy interface in the bonded repair patch, the in6plane displacement at that interface was used as the criterion. The normalized displacement wavestructures for two neighboring modes (modes 17 and 18) on the dispersion curves at the same phase velocity value (14.37 km/s) and frequency 2.5 MHz is also shown in Fig. 1. The interfacial in6plane displacement profile across the frequency6phase velocity space of choice is shown superimposed on the guided wave modes on Fig. 2 as an easy reference tool. In addition to the interfacial in6plane displacement feature, in order to simplify the interpretation of the experimentally measurements waveform, it is mostly preferred to have minimum number of modes excited within the structure. The guided wave mode selected should also have a smaller wavelength to improve sensitivity to smaller defects. The mode 18 (from Fig. 1), identified with an arrow in Fig. 2, at a higher phase velocity range (14616 km/s) was thus selected for inspection of the bonded joint. A big challenge in practical adhesive bond inspection problems is the change in the bond6line thickness. Despite attempts made to control the thickness of the bond line namely the use of cured epoxy strips, glass beads, shim6 stocks etc., local variation in adhesive thickness is possible. In the guided wave based inspection scenario, a change in bond line thickness implies a change in the thickness of just one layer of the layered waveguide – implying a new problem to be solved. The effect of 100 % variation in the thickness of the bond6line on the Lamb wave phase velocity dispersion curves was studied by varying the thickness from 0.4318 mm to 0.8636 mm. Though not shown here, it was noted that the selection of ~15 km/s and 2.5 MHz was found to result in a high interfacial in6plane displacement throughout the range of the thickness considered.

3. PREPARATION OF SAMPLES WITH SIMULATED DEFECTS In order to verify the theoretical work, small repair patch samples – i.e. epoxy bonded titanium6aluminum joint were created. Aerospace grade sheet epoxy – EA9696 was used as the adhesive. Defects were introduced at controlled depths in order to study the wave propagation across cohesive and adhesive weaknesses. aluminum (3.175 mm) and titanium (1.6 mm) plate samples were degreased using a solvent, polished using abrasive disc pads, cleaned with acetone followed by coating with sol6gel and water based primer. The use of the sol6gel is to improve the adhesion for bonding metals. The primer provides better mechanical properties and corrosion resistance. For each repair patch, two epoxy layers were stacked between the prepared faces of the aluminum and titanium plates and cured under vacuum conditions with the application of appropriate pressure and temperature inside an autoclave. Square defects with 0.5" sides were introduced at the aluminum6epoxy interface for creating adhesive weakness and between the layers of epoxy to create cohesive weakness. A folded strip of teflon was also suitably placed for creating both adhesive and cohesive weaknesses in the repair patch. A bubble wrap was used to create another instance of adhesive weakness. The repair patches prepared were cut to form samples for testing shear strength using ASTM 3165. From the test results, it was observed that the good bond region in the repair patch sample was stronger than any simulated weaknesses. The results from the ASTM 3165 tests thus point to the reliability of the fabrication procedure implemented and also provide support to the methods of simulating weaknesses in the bonding.

4. EXPERIMENTAL WORK AND RESULTS There are various techniques for exciting guided waves in structures for experimental work viz. acrylic wedge, oblique incidence in a water immersion mode, or a comb transducer with or without time delays. A variable angle acrylic wedge arrangement was adopted due to its ease of implementation and flexibility to generate guided wave modes at different phase velocities.

Fig. 2 : Normalized in6plane displacement value is shown superimposed over the dispersion curves for titanium6 epoxy6aluminum joint. The arrow on top of the figure indicates the mode 18 – that has one of the highest in6 plane displacements at the aluminum6epoxy interface. Journal of Non destructive Testing & Evaluation

The mode identified in Fig. 2 (mode 18) was generated using a variable angle beam acrylic wedge set to an incidence angle of 10°. Experiments were performed by arranging the wedge mounted transducers (2.25 MHz, and 12.7 mm in diameter) in pitch6catch configuration and varying the excitation frequency from 1 MHz to 3 MHz in steps of 50 kHz. The RF signals collected were squared and summed to obtain an energy quantity for each excitation signal in the range of frequencies 1 MHz and 3 MHz at 50 kHz intervals. A comparison of the variation in the transmitted energy quantity obtained from the frequency sweep experiments is shown in Fig. 3. Vol. 9, Issue 3 December 2010

Technical Paper

The signals collected in pitch6catch mode across the simulated defects using a pair of 2.25 MHz, 6.7 mm diameter commercial transducers mounted on fixed angle wedges is shown in Fig. 4. The pulse input to the transmitter was a tone burst cosine pulse at 2.5 MHz. The RF signals from Fig. 4 clearly show that the guided wave mode selected from the theoretical study was able to distinguish between the different cases of interface conditions simulated and a good joint among the repair patch samples fabricated. It is interesting to note that the mode selected specifically for sensitivity to adhesive weakness is also sensitive to the cohesive weakness because the displacement at the middle of the epoxy layer is still significantly high. Fig. 3 : Energy transmission from frequency sweep experiments using variable angle acrylic wedges adjusted to 10° incidence and reception angle in pitch6catch mode. Commercially available 2.25 MHz transducers were used as transmitter and receiver and tone burst input was supplied to the transmitter.

It can be observed from the Fig. 3 that the range of excitation frequencies 6 2 to 2.5 MHz, with incidence and reception angles of 10°, is sensitive to the adhesive and cohesive defects simulated in the bonded repair patch samples. The transmission is maximum in the case of a good bond and minimum in the case of a cohesively weak bond. The energy transmission in the adhesive weakness cases lies between the two extremes. The variable angle wedges were replaced by small fixed angle wedges having the same incidence angle (10°), thus reducing the number of contacting interfaces between each of the transducers and the bonded plate by one from the initial count of three.

CONCLUSIONS In this work, a theoretically driven systematic approach to selection of guided wave modes for the inspection of adhesive and cohesive weaknesses in an adhesively bonded repair patch, comprised of epoxy bonded aluminum and titanium repair patch, is presented. One of the guided wave modes with large in6plane displacement at the aluminum6epoxy interface in a titanium6 epoxy6aluminum bonded joint was selected for the inspection. Several repair patch samples 6 epoxy bonded aluminum6titanium plates 6 were fabricated in the lab with simulated interfacial weakness conditions. Acrylic angle beam wedges set to an angle of 10°, with 2.25 MHz transducer mounted on top was found to be able to generate interface sensitive mode in the bonded repair patch. Using a matching receiver, it was possible to distinguish between the good and bad repairs. The applicability of the selected experimental parameters over a very large range of adhesive thickness was ensured in order to keep the selection valid for a real repair patch sample where the bondline thickness can vary. Though the solution presented in this paper is specifically tailored to detection of adhesive weakness at the aluminum6 epoxy interface in a repair patch sample, the general approach laid here can be applied to the problem of detecting defects in any layered anisotropic media. In such a case, modes can be selected for sensitivity to the different interfaces present there.

REFERENCES 1. Raymond Pyles. Aging Aircraft: USAF workload and material consumption life cycle patterns, RAND Pittsburgh, U.S.A. 2003. 2. Alan Baker, Nik Rajic and Claire Davis. Towards a practical structural health monitoring technology for patched cracks in aircraft structure, Composites: Part A, 40 (2009) 134061352. Fig. 4 : The RF signals collected by placing fixed angle wedges (10°) with 2.25 MHz transducers mounted on top, across the defective and non6defective repair patches in a pitch6 catch mode. A tone burst excitation source was used to pulse the transmitter. Vol. 9, Issue 3 December 2010

3. W. G. J. t’Hart and J. A. M. Boogers, Patch repair of cracks in the upper longeron of an F616 aircraft of the Royal Netherlands Air Force. Structures and Materials Division, National Aerospace Laboratory NLR6TP620026294. (2002).


Technical Paper 4. J.L. Rose, Ultrasonic wave in solid media, Cambridge University Press, Cambridge, (1999). 5. A. Pilarski and J.L. Rose. A transverse wave ultrasonic oblique6 incidence technique for interfacial weakness detection in adhesive joints, Journal of Applied Physics, 63 (1988) 3006307. 6. J.L. Rose, K.M. Rajana and J.N. Barshinger, “Guided waves for composite patch repair of aging aircraft,” in Review of Progress in QNDE, 15B, eds. D.O. Thompson and D.E. Chimenti, Plenum Press, New York (1996) 129161298.


7. A. Pilarski and J.L. Rose. Lamb wave mode selection concepts for interfacial weakness analysis, Journal of Nondestructive Evaluation, 11(34) (1992) 2376249. 8. P. Puthillath and J.L. Rose. Ultrasonic guided wave inspection of a titanium repair patch bonded to an aluminum aircraft skin. International Journal of Adhesion and Adhesives, 30 (2010) 5666 573.

Vol. 9, Issue 3 December 2010