Form Approved REPORT DOCUMENTATION PAGE OMB No. 0704-0188 Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing this collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden to Department of Defense, Washington Headquarters Services, Directorate for Information Operations and Reports (0704-0188), 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202- 4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to any penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number. PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS. 1. REPORT DATE (DD-MM-YYYY) 2. REPORT TYPE 3. DATES COVERED (From - To) 15 Apr 2010 to 14 Apr 2013 4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER Application of LCR Waves to Inspect Aircraft Structures 5b. GRANT NUMBER FA9550-10-1-0151 5c. PROGRAM ELEMENT NUMBER 6. AUTHOR(S) 5d. PROJECT NUMBER Dr. Auteliano A. Santos Jr 5e. TASK NUMBER 5f. WORK UNIT NUMBER 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION REPORT NUMBER A ND ADDRESS(ES) School of Mechanical Engineering University of Campinas – UNICAMP 13083-970 Campinas, SP – Brazil 9. SPONSORING / MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR’S ACRONYM(S) AFOSR/IOS Air Force Office of Scientific Research 11. SPONSOR/MONITOR’S REPORT 875 N Randolph St NUMBER(S) Ste 325 Rm 3112 Arlington, VA 22203 12. DISTRIBUTION / AVAILABILITY STATEMENT Distribution A. Public Releasable 13. SUPPLEMENTARY NOTES 14. ABSTRACT The stresses and strains imposed on mechanical components can result in mechanical failures. Such stresses can be applied, pre-existed, or both. The pre-existed stresses, also known as residual stresses, result from internal residual strain caused by elastic energy within the mechanical element. In metals, they are generated during the thermomechanical processing of the material [1-2] or loading in service. In composites, the concept is not the same and residual stresses are not expected to perform a role as significant as for the metals in the failure process. Notwithstanding, for both type of materials, the applied stresses in service plays an important role in the expected life. This study aimed to evaluate stress in aircraft structural components based on the acoustoelectric effect. The study concentrated efforts in the structural materials, emphasizing Aluminum and composites. The method used Lcr waves (p-waves) as the main tool. The final objective was to evaluate the suitability of such method to be used on-board in the future, after further development. 15. SUBJECT 16. SECURITY CLASSIFICATION OF: 17. LIMITATION 18. NUMBER 19a. NAME OF RESPONSIBLE PERSON Unclassified OF ABSTRACT OF PAGES Dr. Auteliano A. Santos a. REPORT b. ABSTRACT c. THIS PAGE UU 16 J19rb . TELEPHONE NUMBER (include area UU UU UU code) +55 (19) 3521-3179 Standard Form 298 (Rev. 8-98) Prescribed by ANSI Std. Z39.18 UNIVERSITY OF CAMPINAS FINAL REPORT Grant/Collaborative Research with Air Force Office For Scientific Research FA9550-10-1-0151 Main Researcher: Dr. Auteliano A. Santos Jr Associate Professor School of Mechanical Engineering University of Campinas – UNICAMP 13083-970 Campinas, SP – Brazil Phone: +55 (19) 3521-3179 - Fax: +55 (19) 3289-3722 Email: [email protected] INTRODUCTION The stresses and strains imposed on mechanical components can result in mechanical failures. Such stresses can be applied, pre-existed, or both. The pre-existed stresses, also known as residual stresses, result from internal residual strain caused by elastic energy within the mechanical element. In metals, they are generated during the thermomechanical processing of the material [1-2] or loading in service. In composites, the concept is not the same and residual stresses are not expected to perform a role as significant as for the metals in the failure process. Notwithstanding, for both type of materials, the applied stresses in service plays an important role in the expected life. On-boar health monitoring of aircrafts requires reliable inspection systems, protected against environmental changes and damages caused by loading. Knowing the stress in critical spots give the flight commander the information required to take decisions about what action would be performed. Also, knowing the plastic damage in selected spots allows the maintenance personnel to predict when the component should be serviced or replaced. Because most of the methods of measuring stresses are destructive, they cannot be used with aircrafts. Strain gages can be used, but they are not adequate for most critical locations. Besides, they are fixed transducers, which cannot be used in another place after the first measurement, so a large amount of them is required to inspect structures like in aircrafts. Thin-solid films have the same limitations. Nondestructive methods, like X-ray and neutron diffraction usually do not meet the safe requirements. 1 Among those non-destructive techniques for measuring stresses, one notable for its low cost and portability (for field applications) utilizes ultrasonic waves. It also allows easy and fast measurements, has good resolution, and is safe. Measuring stresses with ultrasound is possible from what is known as the acoustoelastic effect. The acoustoelastic effect relates the strain of a body subjected to a static state of stress with the propagation speed of elastic waves in it [3]. To measure stress, we can apply acoustoelasticity employing three main techniques: Rayleigh waves, acoustic birefringence, and critically refracted longitudinal waves, or L waves. The Rayleigh cr wave technique relies on the main characteristic of this waves—the propagation in the surface of the medium—and is thus suitable for surface stress measurements. Researchers have applied Rayleigh waves with adequate results [4-7] to investigate both external and residual stress measurements. The acoustic birefringence technique relates the stress to the difference in speed of two ultrasonic shear waves propagating in the same direction, but polarized in orthogonal directions [8-11]. L waves are cr longitudinal subsurface waves (body waves) and are less sensitive to surface imperfections than Rayleigh waves. Since they are longitudinal, L waves are more sensitive to stress variations [7, 12-14]. cr According to acoustoelastic theory, a change in wave velocity is related to stress by acoustoelastic coefficients, which depend on the second- and third-order elastic constants of the material. We can use these constants to calculate the acoustoelastic coefficients or obtain them experimentally by applying a known stress and then measuring the variation in the wave speed [15-19]. Because the elastic properties of a material directly influence the acoustoelastic effect and the former may differ according to direction, materials with dissimilar properties in different directions answer differently to the wave propagation. In metals, the non-uniformity results, in part, from the manufacturing process. An example of a process that generates these differences is rolling. Rolling is used in the manufacturing of aircraft Aluminum, where alloys for coatings and structural reinforcements are laminated. This process produces a symmetry known as orthorhombic, allowing us to classify the material as orthotropic [18]. In composites, the actual distribution of the fibers is the factor that influences the non-uniformity of the material. For many of the structural parts nowadays used, the fibers are oriented in just one direction, leading to a complete different set of speed results depending on the wave type and its propagating direction. Most researchers working with applied stress measurements using ultrasound ignore the anisotropy generated by rolling in metals [19-20]. Instead, they relate stress and wave velocity using an isotropic model. This is done mainly because such a model is simple and because, even for second-order elastic constants, the anisotropy generated by rolling is generally weak [15]. The problem with this method is that third-order elastic constants can also influence the results. However, when using longitudinal waves propagating in the direction of the unidirectional applied stress, it is expected that the anisotropy has small effect on the results, since the elastic properties in the main directions are known. A study is required to analyze the texture effect. For composites, these assumptions require even deeper studies. As stated before, the L method seems to be the correct selection to measure stresses in many cr instances. It can be used for metallic parts, as shown by Pereira Jr. [21] in his MSc thesis, like Aluminum or other lighter metallic components. Because the composites are usually applied in thin plates or slender beams, the L is also an adequate alternative, since the longitudinal wave travels about one cr wavelength from the surface. This report describes the findings of the Acoustoelastic Laboratory of University of Campinas, in the tentative of answer two main questions: Can L method give the information required when used to cr inspect stresses in aircraft structural metallic components? Can we inspect composites structural parts using a similar L based measurement technique? Those questions motivated the development of a cr scientific research sponsored by the Air Force Office for Scientific Research with a grant of $ 69,420.00 for a three year period, started in April 2010. The report shows the objectives and methodology, the development of the study, the answers found, and the proposal for new studies. Objectives The study aimed to evaluate stress in aircraft structural components based on the acoustoelastic effect. The study concentrated efforts in the structural materials, emphasizing Aluminum and composites. The method used L waves (p-waves) as the main tool. The final objective was to evaluate cr the suitability of such method to be used on-board in the future, after further development. 2 METHODOLOGY The methodology presented in the proposal is described as follows. It has seven steps and, as already mentioned in the previous report. As can be seen, the targets for each step were fully accomplished. a. Reviewing of the state of art. In this first step we intended to know the systems used in the present, their characteristics and limitations. We would also search for researches in the area to learn how we can contribute to the safety of the aircrafts, through their papers and reports or direct contact. b. Selection of the materials to be studied. The Aluminum was the main focus of study. The Aluminum was our main focus of study. We would ask AFOSR to select a composite to be evaluated too, but, instead, we consulted Brazilian companies in the aeronautic area to find which material would be more adequate for our study. Regarding the format of the parts to be studied, our main focus would be on plane surfaces, like sheets to make aircraft hulls. c. Development of the inspection system. This step required the selection of the commercial products to perform the experimental evaluation, and the development of those that cannot be bought; the assembly of the system; initial tests, and calibration for each material under evaluation. d. Development of the routines for data treatment. Because of the dispersive nature of the measurement in non-uniform materials, especially in composites, we would work to implement the best routines to get the useful results. It would require fundamental research about wave propagation and signal processing. There was no guarantee that we could inspect composites with such technique, so the study would also evaluate different strategies, aiming to get adequate results. To change the type and characteristics of the sensors was one of the possibilities. e. Tests for influence factors. The main factors at that time identified for measurements with L waves were temperature, surface uniformity, texture (metals), and system cr characteristics, like transducer natural frequency, size, probe size, and so on. We would look for new factors and evaluate the influence of them with the already known. f. Application of the system to evaluate the stress and strain in the structural components. After controlling the influence factors, we would evaluate the behavior of L wave in the cr selected materials and study the viability of applying the method in the inspection. g. Final report and technical communications. The papers to communicate our results would be submitted at this point of the research. The information to be included would depend on the AFOSR authorization. The final report would also be issued at this time, including suggestions to future research works and developments. The initial schedule proposed would cover 24 months total. DEVELOPMENT The development of the study followed the methodology presented above. As mentioned before, the initial schedule would cover 24 months total. Because of some delay in the processing of the funds sent by the US government inside Brazilian Finance System, the period for this research ranges from April 2010 to April 2013. To describe how the research was performed and the results, each step of the methodology is discussed in this section, including the references for the sources, when applicable. They are: a. Reviewing of the state of art. This step included the search for advances about acoustoelasticity, including equipment, papers about innovations, composites, and methods for stress measurement. The results of this research allow the developing of three MSc. Theses, two Undergraduate Scientific Researches and two undergraduate final works. The complete review can be consulted in the Theses. Unfortunately, the theses are in Portuguese, but most of the references are in English and can be consulted if necessary. The documents are in the following links: 3 • Author: Tainá Gomes Rodovalho – Sponsor: AFOSR http://www.bibliotecadigital.unicamp.br/document/?code=000871644 • Author: Rodrigo Junqueira Leão – Sponsor: AFOSR http://www.bibliotecadigital.unicamp.br/document/?code=000871675 • Author: Paulo Pereira Júnior – Sponsor: CAPES - Brazil http://www.bibliotecadigital.unicamp.br/document/?code=000812437 We also included the theoretical development in those theses and in other documents cited in this report. The main development in this area was to understand and list in a clear format how we can measure stresses using the wave speed. The Thesis of Eng. Pereira Jr [21], the third cited above, describes deeply the theory about it. b. Selection of the materials. After consulting the people who work directly with manufacturing of aeronautic parts, we decided to choose Aluminum alloy 7050, used as structural parts. We also decided to use carbon fiber composites with epoxy matrix (HexTow® AS4 / Hexply® 8552) in the form of sheets. The fibers are arranged one directionally, as in most structural parts. The research had two main branches: analytical/numerical simulations and experimental evaluation, both of for Aluminum and composite. The first looked to understand the effect of stresses in wave propagation trough simulations. Experimental work aims to measure the effect of applied stresses on the selected structural materials taking in account all effects. Provided our numerical models fit the experiments adequately, we expected to refine it progressively until we can simulate the behavior of the structural materials in service conditions, in a future research. Before conducting experimental work, experimental plans were developed. We needed to know if the wave speed was sensitive to the stresses, so we had to build samples of both materials that could be measured in two situations: with and without stresses. Besides, we had to develop samples to get other parameters associated with the measurement processes, like the elastic constants and the effect of the rolling direction (Aluminum) and fiber direction (composite). We also have to develop the models to the simulations. Because of the lack of experience and knowledge about this subject, we decide to employ the models presented in the literature and improve them according to our needs, if necessary. Samples for experiments In this section, to facilitate the understanding of the report, we present the main samples employed in the tests performed. The Aluminum samples were mostly in the format of bars, as shown in Figure 1. They were extracted from a rolled sheet bought in the market, but the study of the elastic constants involves several different formats of samples, as shown in Figure 2. Of course, care was taken to do not introduce stresses in the bars during cutting processes. In this case, we did not perform a thermal process for stress relief, as we use to do with steel. This could lead to slight different initial stresses in each bar, but not to different response to applied stresses. The bars were used to find the acoustoelastic coefficient L1 , which is employed in the 11 calculation of the stresses from Equation (1) using longitudinal waves. In that equation, υ is 1 the wave of a longitudinal speed propagating in the main principal direction; ε is the deformation; σ is the stress in the principal direction; E is the Young’s modulus. The same 1 bars where used to verify the effect of the influence variables, like temperature, as will be presented later in this report. ( ) ( ) dv(1) v(1) E dv(1) v(1) E dv(1) v(1) (1) 1 1 = 1 1 = L1 ⇔ dσ = 1 1 dε dσ 11 1 L1 1 11 4 The development of equation (1) can be seen in one of the papers published by our research group about the determination of the effect of the anisotropy in the wave speed [22]. It’s worth to mention that the acoustoelastic coefficient for isotropic metallic materials can be calculated from the elastic constants or from the elastic properties, as show in Equation (2). In that equation, l, m and n (not showed) are third-order elastic constants; λ e µ are the Lamé’s constants. ( ) dv(1) v(1) µ+ 2m+υµ 1+2l (2) L1 = 1 1 = 2+ λ 11 dε λ+2µ Figure 1 – Format of the Aluminum bars used in the AFOSR research [21] Figure 2 – Format of the Aluminum samples used to find the elastic constants [22] The method we use for measuring stresses – the L wave method, requires the wave hits the cr surface in the first critical angle. This angle is calculated using the Snell’s Law and depend on the wave speed in the propagation direction. For isotropic metals that speed is known and can be found in published compendiums. For orthotropic metals the speeds can be measured. We did that with the Aluminum. However, for composites the wave speed strongly depends on the fiber direction. In braided composites, the effect is probably minimized by the fiber distribution, but for unidirectional structural composites we need to evaluate the wave speed in other directions away from the fiber’s. This information would also be valuable to understand the response to stress in each direction. So, the composite samples to evaluate the effect of the fiber direction was built in the format of a 24 side polygonal specimen. Figure 3 shows the sample. For measuring stresses, the samples were similar to the bars for Aluminum tests. Figure 4 shows the bars (5) and the measurement probe. For the tensile test equipment to leave the 5 fibers of the sample undamaged, it was necessary to bond aluminum plates (1) on it to minimize the effects of load grips. The contact of the fixing fixture could generate higher stress concentration points, favoring rupture in the region, and could also break the superficial fibers of the sample. The aluminum plates were fixed on the sample surface using Araldite® Professional, epoxy glue. Figure 3 –Polygonal samples used to find the elastic constants [23] Figure 4 – Samples of unidirectional composites in format of bars for tensile stress tests Because we used the L wave technique, it was necessary to employ angled wedges (2). cr Those angles were the critical angles calculated using Snell’s Law, as mentioned, and the polygonal specimen. The wedges (rexolite@ shoes) were fixed to a connection bar (3) that blocked them from moving away. In the right side of the figure we show a positioner made of foam (a), used to keep the probe in the same longitudinal position, and the wood clamp, to keep the probe in contact during the measurements. Both parts were necessary because we used a vertical tensile stress machine and the probe could fall from the composite bar. The composite samples used in our tests were manufactured by GME Aerospace, a Brazilian company that produces parts for the aeronautic industry. As mentioned, they were made with material composed of carbon fibers (HexTow® AS4), unidirectional, pre-impregnated (prepreg) with epoxy matrix (HexPly® 8552 from Hexcel®). Table 1 shows the materials’ properties. The process of manufacturing the composites follows very strict rules for aerospace materials. The process includes numerically controlled cutting and autoclave in vacuum as showed in Figure 5. Details about the process are in Ms. Rodovalho Master Thesis [23]. 6 Figure 5 – Manufacturing of composite parts. Left: cutting machine. Right: Autoclave Table 1. Physical and mechanical properties of prepreg Hexply® AS4/8552 Lamina density (g/m³) 1.59 Nominal fiber volume (%) 58.52 Lamina thickness (mm) 0.187 Longitudinal elastic modulus (GPa) 142 Transversal elastic modulus (GPa) 9.5 Shear modulus (GPa) 5.0 Longitudinal tensile strength (MPa) 2336 Transversal tensile strength (MPa) 81 Shear strength (MPa) 114 c. Development of the inspection system. The system is basically the same for both, metal and composite applications. It was derived from the one already used for railroad wheels and oil pipes. The differences were on the transducers: Aluminum required 5 MHz transducers [16] and, for composites we used 1 MHz. Before deciding which frequency we should use, several lower and higher frequencies were tested for both materials. Figure 6 shows one of the arrangements used, in this case, for composites. Similar probes were used for Aluminum, but the angles of incidence in the shoes were different. Details about the system for composites can be seeing in the paper “Stress analysis in carbon/epoxy composites using Lcr waves”, recommended for publication by the reviewers of Journal of Composite Materials and attached to this report in the recommended version. Figure 7 shows one of the many arrangements used with the complete system along the measurements. In that case, it was used to measure the wave speed in the polygonal specimens. 7 Figure 6 – Scheme of the system to measure the effect of stresses in the wave speed. Figure 7 – Complete system used to measure the wave speed for composite polygonal samples. d. Development of the routines for data treatment. When we started to study the possibility of using L waves to measure stresses in aerospace cr materials, we already have confidence (based on previous experiments and literature) that we could get results in Aluminum, but we would need to improve the method. However, the measurement in composites was totally new for our group. So, we had to study the ways to obtain useful data from the measurements. The result of this study generated a paper submitted to ASME IMECE 2013 (Comparison of signal filtering techniques for ultrasonic waves used in inspection of composite materials), attached to this report. There we compared the techniques based on digital filters IIR (Infinite Impulse Response), FIR (Finite Impulse Response), and Discrete Wavelet Transform (DWT). The results showed that for the bulk longitudinal and L wave propagating in most directions cr in the composite polygonal samples, the filters IIR and FIR presented better results than the DWT, aiming significant improvement in the Signal to Noise ratio (SNR). For bulk waves, the mean improvement in SNR where the FIR filters had the best results was 61%; for the IIR filters the improvement was 27% and 38% for the DWT. Only in one direction of 8 propagation, 15° in relation to the fiber direction, the DWT had better results. However, analyzing the waveforms after the signal processing, it can be seen undesirable distortions in the waveform. This kind of effect does not happen when we use digital filters. These findings were used to improve the programs developed for data acquisition employing the L waves, as the one who the main screen is shown in Figure 8. cr Figure 8 – Software developed to measure stress with L waves. cr e. Tests for influence factors. The main factors identified for measurements with L waves at the beginning of the cr research were temperature, texture (metals), and system characteristics, like transducer natural frequency, size, probe size, and so on. We would look for new factors and evaluate the influence of them with the already known. After several studies, we reach the following conclusions: i) The system characteristics influences for both Aluminum and Composites. So, any new equipment has to be calibrated to be used in wave speed measurements, for the resolution we require to get useful results. Notwithstanding, provided that we chose the equipment that responds adequately, we can get consistent results. ii) Probe size influences on the how deep the wave travels. That information came from literature and our previous findings with API steel. In our evaluations for this research, all samples were tested under uniform traction. So, this factor would not influence. More studies are required to evaluate parts under bending. iii) Texture for Aluminum was evaluated analyzing the differences in elastic properties in three main directions: the rolling one and two perpendicular ones. Table 2 shows the results determined using ultrasonic measurements. We employed two models to calculate them: Orthotropic and Isotropic, according with the literature. As can be seen, there is only small differences in the results. However, because L values 11 are different, since they depend on third-order elastic constants, it is necessary to identify the rolling direction and correct the stress for any direction from it. iv) Fiber direction is a major influence factor for composites, as expected. We built the polygonal specimens to evaluate it. The experimental plan of measurements allows us to know the influence of direction and temperature effects, simultaneously. Figure 9 shows the results. As can be seen, there is a huge effect of the direction, in comparison with the temperature. Because we are interested in the wave that 9