Fatigue crack growth and life prediction methods
| Suitability of Different Crack Growth Models for Analysing Multi-Site Damage in Aircraft Structural Configurations Alvin Ng, Cees Bil Abstract: Fuselage lap joints, wing chord-wise splices, rib to skin attachments and wing skins at run outs are typically susceptible to Multi-site Damage (MSD). However, due to the differences in configuration, different crack growth models are better suited to different configurations. This investigation compared several different MSD crack growth models and assessed their suitability to different structural configurations. |
Fatigue crack growth in Fibre Metal Laminates under variable amplitude loading Sharif Ullah Khan, Rene Alderliesten, Rinze Benedictus Abstract: Fibre metal laminates (FMLs) are hybrid materials consisting of alternating layers of uni-directional impregnated fibre lamina and thin metallic sheets adhesively bonded together. These materials have been developed primarily for aircraft structures as a substitute to high strength aluminium alloys due to their better mechanical and damage tolerance properties. For standard GLARE, aluminium 2024-T3 sheets and S2-glass fibres are bonded together with FM94 epoxy adhesive to form a laminate. FMLs are famous for their exceptional fatigue properties. These properties are due to the a complex fatigue mechanism. During fatigue loading, crack nucleation and growth occurs in the metal layers. The fibres being insensitive to the occurring of fatigue stresses remain intact and bridge the cracks in the metal layer. This crack bridging restrains the crack opening and reduces the crack-tip stress intensity factor resulting in slow crack growth rate. Secondly, due to this mechanism delamination starts to grow at the metal-fibre interface. The excellent fatigue properties of FMLs are due to a so called coupling process between the delamination and crack growth. FMLs have been extensively investigated in the past, especially the fatigue mechanisms. After developing the understanding and knowledge about FMLs at Delft, a generic and physically sound analytical fatigue crack growth prediction model has been developed. The model works quite well for the Constant Amplitude (CA) loading. Further research is done in order to enhance the capabilities of this model to predict under Variable Amplitude (VA) and Flight Spectrum Loading. To make the step towards a VA fatigue crack growth prediction model for FMLs from the CA prediction model, certain limitations are identified. These limitations are the lack of quantitative information about plastic zone formation and its size, shear lips and stress state. The size of plastic zone is vital information required to investigate the effect of overloads. Smaller or constant plastic zone sizes will reveal the presence of limited retardation/interaction effects. Shear lips and its shape reveal the effect of load variation on material properties as well as the influence of crack length. Presence of plane stress or strain needs to be studied for FMLs due to the presence of thin metallic sheets on the other hand the fibres and adhesive can influence the stress state under load variation. A research program has been executed in which experiments have been performed to quantify the above mentioned parameters. Three types of materials have been considered for the test program, namely Monolithic Aluminium, Metal Laminates (MLs) and FMLs. The correlation between the monolithic aluminium and MLs provides information on stress state and shear lips. While the effects of intact fibres and their influence combined with the effect of VA loading has been investigated by correlating the MLs and FMLs. Plastic zone sizes, crack tip plasticity and delamination shapes has been measured using Digital Image Correlation (DIC) technique. The effects of VA loading on these parameters has also been investigated. The sizes of plastic zone measured during these tests have been compared and evaluated using the Irwin's plastic zone size relation and a better correlation has been observed for FMLs than monolithic metal and MLs. This paper presents the experimental work and its analysis. Fatigue crack growth results will be presented in the form of plastic zone sizes visualized using DIC compared with the Irwin's relation. Influence of overload ratio on plastic zone sizes and delamination shapes will be discussed using quantitative data. Finally these quantitative results have been implemented in already existing CA prediction model to enhance its prediction capabilities to VA loading. |
Effect of shot peening residual stresses on fatigue strength and crack propagation threshold: experience on helicopter components made in Al 7475 alloy Ugo Mariani, Giuseppe Ratti, Marco Giglio, Mario Guagliano Abstract: CS 29, under paragraph 29.571, requires that flaw tolerance capabilities are to be demonstrated for fatigue assessment. For metallic parts of helicopters, subjected to high loading frequencies, this means to design at stress levels which do not allow crack propagation since the use of the traditional LEFM crack growth approach leads to very short inspection intervals. The presence of a small surface flaw (scratch, dent, crack) significantly reduces the fatigue allowable of a mechanical part since the crack nucleation phase is either very short or even totally missing. Within this context the fatigue limit in the presence of flaws assumes a different meaning to that used in traditional safe-life concepts. It can be considered to be the threshold stress for non propagation of cracks emanating from the original flaw. Shot peening is a mechanical process which cold works the surface of a structural part by means of a propelled stream of spherical shots. It is used to improve the fatigue properties of the part by introducing on the surface and in a small layer underneath beneficial compressive stresses which retards or sometimes prevents fatigue cracking. After shot peening application a surface flaw is partially or totally embedded in the compressive stress field. When considering the traditional LEFM methodology, this compressive stress field is the main source of the closure of the crack, which therefore grows at a highly reduced rate. The same closure effect is also present when considering the threshold stresses for non propagation of cracks emanating from the original flaw, so that the detrimental effect due to the presence of the flaw is reduced. The shot peening intensity according the Almen scale is used to indicate the way to perform shot peening on a metal part. However, a given value of Almen intensity can be achieved by means of different combination of treatment parameters (shot size, material, velocity and angle) inducing different in-depth residual stress profiles. Instead flaw depth is the most important parameter for surface flaws. Hence the set of combinations shot peening parameters - flaw depth is very wide and the effect on the threshold to propagation can be very different with varying the combination chosen. In addition other parameters influence the compressive stress field so that theoretical methods available in the literature can lead to results not consistent with test experience. Within this context the present paper will show the effect of shot peening on the threshold to propagation of a semicircular surface flaw 0.36 mm deep for a typical aluminum alloy used for helicopter dynamic components (Al7475-T7351). A numerical procedure based on finite element simulation has been developed being the aim the prediction of the residual stress field induced by shot peening by varying the peening parameters. On the base of the numerically obtained results, an approach was developed to predict the fatigue strength of peened parts including notches or surface flaws. The results were compared to those experimentally obtained leading to the definition of a criterion aimed to the optimization of the shot peening parameters versus fatigue strength and threshold to crack propagation of small defects. Keywords: shot peening, flaws, flaw tolerance, compressive stresses, threshold to crack propagation |
Experimental validation of stress intensity factor solutions for the pin loaded lug Alten Grandt, David Child, Nicholas Moyle Abstract: Predicting crack growth behaviour is an important element of managing the structural integrity of aircraft fleets. It is common for fleet managers to employ software packages, such as AFGROW, to automate the prediction of crack growth behaviour. AFGROW has traditionally used empirically derived stress intensity factor solutions. However, stress intensity factor solutions generated using Finite Element Models (FEMs) are now being employed within AFGROW to provide a greater level of accuracy and flexibility. Irrespective of the move towards using FEMs, stress intensity factor solutions must still be appropriately verified in order for the results to be utilised in structural integrity management. Two independent research programs were conducted at Purdue University to review stress intensity factor solutions for cracked, pin-loaded lugs – a commonly used configuration in aircraft primary structure. The purpose of the research was to experimentally validate stress intensity factor solutions derived from FEMs. Component fatigue tests were conducted on corner, oblique and through cracked, pin loaded aluminium lugs for a range of lug geometries typical in aerospace applications. Crack propagation was measured using direct optical and marker banding techniques. Non dimensional stress intensity factor (geometry factor) solutions were calculated utilising a back-tracking method and compared with results from representative StressCheck® Finite Element Models. In the case of the through crack configuration, the comparison of experimental and StressCheck® derived geometry factors showed a close correlation and were an improvement to solutions provided in AFGROW at the time. Based on the results of this research, StressCheck® pin-loaded lug geometry factors have since been incorporated into the AFGROW software. In the case of the corner crack configuration, there was a correlation between experimental and StressCheck® results. However, this was highly dependant on the choice of pin-loading boundary conditions in the FEM. The results of this research have been utilised to define the default AFGROW software pin-loaded lug boundary condition assumptions. |
Property for fatigue crack propagation of friction stir welded 2024-T3 aluminum alloy Takao Okada, Kazuya Kuwayama, Shinya Fujita, Motoo Asakawa, Toshiya Nakamura, Shigeru Machida Abstract: Friction Stir Welding (FSW) is one of new weld process with the capability of welding high strength aluminum alloys of the 2xxx and 7xxx types. From a point of reduction of production cost and structural weight, FSW is expected to apply aircraft primary structure as alternative to rivet joint. However, the recent regulation for damage tolerance and fatigue evaluations of aircraft structure requires understanding the location of probable fracture origin and fatigue crack growth property. Especially, it is important to investigate relationship between fatigue crack growth property of FSW panel and residual stress on the panel, because fatigue crack growth property is affected by residual stress around the weld line. In this reason, many researchers have investigated the effect of residual stress on crack growth property. But their studies do not evaluate the effect of inclined angle of FSW weld line to direction of crack growth. This presentation describes experimental results of fatigue crack growth test in order to clarify property for fatigue crack growth of FSW panel. The effects of residual stress on crack growth rate and redistribution of residual stress during crack growth were investigated. Additionally, the effect of inclined angel of weld line on crack growth rate and direction of crack growth were discussed. Sheets of 2024-T3 with 2mm thickness were joined by FSW butt joint. Specimens have FSW of inclined angles of 0 or 30 degree to the direction of applied stress. Width and length of the specimen are 400 and 1,000 mm, respectively and the starter notch is introduced at the center. The specimens are subjected to cyclic loading with R = 0.1 and stress amplitude is 50 MPa. Test frequency is 5 Hz. Crack length was measured by CCD and the scale. Residual stress field was determined by the hole drilling method. Furthermore, residual stress redistribution caused by the crack propagation was observed by strain gages located in front of notch. The peak value of residual stress in longitudinal direction, about 300 MPa, was found on weld line. Residual stress outside of the weld line was apparently smaller than that in weld line. And the residual stress in transverse direction was smaller than that in longitudinal direction by one order of magnitude. The da/dN-dK curve of the FSW specimen showed that crack propagation rate was accelerated by the tensile residual stress when the crack tip located around the weld line. After the crack tip ran through the weld line, crack propagation rate gradually close to that for base metal. The relationship between redistribution of longitudinal stress and crack length shows that tensile stress redistributes in front of the crack tip after the crack tip ran through the weld line. This tensile stress accelerated crack propagation rate of the crack crossing the weld line. Comparison of crack path between the FSW specimen for 0 degree weld line and that for 30 degree shows that direction of crack growth was not affected by its inclined angle. Regardless of weld line angle, the crack grew perpendicular to the loading direction. Next, the relationship between crack length and angle of maximum principle stress in front of the notch is investigated. The direction of maximum principle stress is inclined to the weld line in case the crack tip is located near the line. If the maximum principal stress is a dominant factor for the direction of crack growth, the crack path has inclined angle．But actual crack path is mostly perpendicular to the weld line. This result indicated that the maximum principle stress is not the primary factor for the direction of crack growth. The direction of the maximum stress amplitude seems to be a primary factor. Crack propagation rate measured by inclined angle of 0 degree is almost same as that measured by the angle of 30 degree. It is supposed that this result is caused by cancel of crack opening stress for FSW specimen, because the maximum tensile residual stress, equal to 300 MPa, is apparently higher than the applied maximum stress, 55.6 MPa. Therefore da/dN-dKeff curve of base metal was evaluated using equation of the stress intensity range ratio, U, suggested by Schijve. However, da/dN-dK of FSW and da/dN-dKeff of base metal did not coincide, even if crack tip was located on weld line. It implied that crack opening stress for each weld line angle is not completely canceled by the tensile residual stress. Now we are conducting the test with specimens which have weld line of inclined angle of 45 degree. In addition, measurement of crack opening stress is conducted to directly evaluate dKeff of the specimen. These results are presented at the symposium. |
Load Sequence Effect and Enhanced Fatigue Life Prediction under Spectrum Loading Lei Wang, Yongkang Chen, William Tiu, Yigeng Xu Abstract: Damage tolerant design of airframe structures relies on the life prediction of fatigue cracks propagating from a detectable size to the critical size. Enormous research efforts have been made to understand the crack growth behavior of airframe structures under the spectrum loading. It is however recognized that accurate life prediction for airframe structures is challenging due to the complex load sequence effect. This paper aims at gaining a further scientific understanding of the complex influence of the loading history on fatigue crack growth, which leads to the development of an enhanced fatigue life prediction methodology for airframe structures. The spectrum loading is systematically broken down into a number of simple yet representative loading scenarios with overload/underload superimposed onto the baseline constant amplitude fatigue loading. Detailed finite element (FE) simulation of the plasticity-induced crack closure (PICC) has been carried out to catch the transient behavior of PICC. The variation of the PICC is quantified to rationalize the transient crack growth behavior under the spectrum loading. Research efforts have been focused on the relationship between the local crack closure and the global load transfer, fatigue damage for the load below the conventionally defined crack closure load, and the load sequence effect on crack propagation. Results so obtained are implemented to the proposed fatigue life prediction model to achieve more reliable life predictions under spectrum loading. It has been concluded that while care should be taken in defining the crack closure the concept provides a powerful tool in rationalizing the load sequence effect on fatigue crack growth behaviour. Good agreement is observed between the preliminary results from the proposed life prediction model and the results reported in the literature under typical spectrum loadings. |
Fatigue Crack Growth in Thick Plate 7050 Aluminum Joel Schubbe Abstract: Abstract A study has been accomplished to characterize the fatigue crack growth rates and mechanisms in thin and thick plate commercial 7050-T7451 aluminum plate in the L-S orientation. Crack growth and crack shielding with branching behavior of long, through thickness cracks is examined and compared to L-T and T-L oriented growth data. Compact tension specimens and the compliance method were used to determine crack growth rates. Constant ΔK data showed significant retardation of growth rate curves for the L-S orientation in the range of 10 to 13 MPa√m where branching and splitting parallel to the load axis are dominant growth mechanisms. Aerospace alloys are continually being developed, modified, tested and qualified for use in flight vehicles around the world. Extended service life is a prime consideration in every current design and both structural design and material selection is necessary to increase life and ensure safe flight. Every designer is striving to find that optimal mix of properties to include strength, corrosion resistance, fatigue resistance, stiffness, or other desired characteristics. The limitations of use of many available systems are the testing and qualification of these materials in the environment in which they are intended to survive. Testing, machining, cost of supply, and space requirements drive engineers to use well-known materials. Only small configurational or orientation changes are necessary to make a known material “less known” in any or all of the material selection property areas. 7050-T7451 is purported to be a superior aluminum alloy system, exceeding strength properties of 7085-series alloys and improving upon corrosion characteristics. Historically, the “banding” of properties for selection and design is used to simplify the down select process, narrowing choices to specific manufacturing materials. Testing for the extremes is common practice and helps to reduce the number of tests required the characterization of any specific alloy. Unfortunately, the most modern, tailored alloys tend to be highly anisotropic in nature and must be fully tested with loads representative of those expected in service for the expected orientation to understand the required range of properties for selection. The 7050-T7451 alloy is one of these highly anisotropic materials that need full characterization for use. Most strength properties are fully defined but if experimental crack growth rate data and fatigue properties for the L-S orientation are to be used for life estimates, they have not been tested at sufficient levels for these predictions. This study was initiated to generate an initial data set for prediction of growth rates in the L-S orientation and to characterize the morphology of the crack. In this case, it has been determined that the manner in which the crack progresses may be important to the failure mode for machined parts with multi-axial stress distributions. Specimens tested in this study were cut from 7050-T7451 aluminum plates of parent thicknesses 1.27 cm and 10.16 cm. L-T and T-L oriented specimens were tested to validate the test method against legacy data and L-S oriented specimens were tested to generate new da/dN vs. ΔK crack growth data plots and examine the influence of crack length on the mechanisms associated with the material crack morphology. Testing in an approximate range of 5 to 35 MPa√m was accomplished and at load ratios of R=0.05, 0.1, 0.3, and 0.5. It was found that at R=0.7, the specimen configuration defined for these tests, combined with the required ASTM loading levels, was incompatible for performing this testing. Crack initiation at the R=0.7 load ratio resulted in immediate splitting parallel to the load and little or no forward crack progression. Repeat tests for the other load ratios were accomplished to examine repeatability and to compare to L-T and T-L data. A MTS 810 electric servo-hydraulic 22kip test stand was used to pin load the specimens and control was accomplished using an external MTS Flex SE Test controller with MTS PC interface software. Load considerations included the load ratio, thickness to specimen depth, K-gradients, and maintaining a small scale yielding criteria. Tests were conducted in a tension-tension mode due to specimen type and Cyctest software was used to regulate the growth loads and gradients of continuous load shedding during test. Cycle by cycle feedback was provided to enable real-time load corrections. In addition, crack tip opening displacement (CTOD) data from the MTS clip gage was recorded and used to determine crack growth rates using the crack closure compliance method. Optical checks were performed to guarantee the crack length accuracy being used for compliance calculations during test. Crack lengths were measured using a Gaertner traveling optical telemicroscope with digital readout. |
Prediction of fatigue life in carbon fiber reinforced plastics using damage mechanics Min Sung Kang, JungHun Choi, HongSun Park, Soo Park, JaeMean Koo, ChangSung Seok Abstract: Recently, the amount of CFRP used has been increasing in applications that require structural weight reduction, such as aircraft structures. Due to their intrinsic anisotropy, composite materials show quite complicated damage mechanism with their fiber orientation and stacking sequence and especially, their fatigue damage process is sequential occurrence of matrix cracking, delamination and fiber breakage. In the study, to propose new model capable of describing damage mechanism under fatigue loading, fatigue analysis of composite laminates based on damage mechanics, are performed. Fatigue damage curves are obtained from hysteresis loop and assessed by the fatigue damage analysis. Then, static and fatigue damage analysis are combined. Expected results such as stress-cycle relation are verified by the experimental results of fatigue tests. |
Industrial applications of the extended finite element method for crack propagation simulations in aeronautical structures Eric Wyart, Marc Duflot, Hans Minnebo, Frederic Lani Abstract: ABSTRACT In this contribution, the authors present several examples of computation of stress intensity factors and simulation of the propagation of cracks in aeronautical structures. The featured applications involve the use of an innovative numerical method called the eXtended Finite Element Method (XFEM) [1]. The XFEM is particularly suited for the solution of complex fracture mechanics problems as it allows for introducing an a priori knowledge of the solution in the finite element approximation space, e.g. the displacement discontinuity (crack opening) and the functions of the development expansion of the crack tip displacement field in a linear elastic solid, which avoids the need for a conforming mesh. The XFEM has been the subject of intense research efforts for an entire decade now. It has recently reached a maturity level that allows its transfer into industrial damage tolerant approaches, supplementing standard procedures based on the use of simplified solutions gathered in software such as Esacrack, Nasgro, Afgrow. It also permits to generalize the use of complete three-dimensional finite element simulations of cracked structures thanks to a significantly reduced effort devoted to meshing and CAD/CAE operations. This contribution summarizes the basic principles of the method, presents the main challenges faced during its implementation in an industrial setting and shows some of the applications performed at CENAERO on aeronautical structures in the frame of European projects and consultancy studies [2, 3]. KEYWORDS Fracture mechanics, crack propagation, extended finite element method REFERENCES [1] N. Moës, J. Dolbow, and T. Belytschko. A finite element method for crack growth without remeshing. International Journal for Numerical Methods in Engineering, 46:131–150, 1999. [2] É. Wyart, D. Coulon, M. Duflot, T. Pardoen, J.-F. Remacle, and F. Lani. A substructured FE-shell/XFE-3D method for crack analysis in thin walled structures. International Journal for Numerical Methods in Engineering, 72: 757—779, 2007. [3] Marc Duflot, Éric Wyart, Frédéric Lani, Philippe Martiny and Sébastien Sagnier. Application of XFEM to multi-site crack propagation in an aeroengine component. Engineering Fracture Mechanics, in press, 2008. |
Fatigue life evaluation on tension type fitting Nicolas Barea, Marie Masse Abstract: The fatigue life evaluation on tension type fitting is a difficulty in aeronautical structure designing. A multitude of fatigue calculation methods and test results exist for shear joints but very few recognized ones for tension fittings. The goal of this study is to try to define a fatigue life evaluation method on tension type fitting. Finite element analysis shows the difficulty to have consistent results, because of the high influence of parameters like friction and contact, pre-load value and its precision, way to model fastener and fitting, elements type choice. This numerical approach is heavy and can be replaced by an easier and faster approach as it is often made for shear joints. More than hundred specimens in several configurations (type of fastener, geometry, 1,2 or 3 wall fittings, loading, pre-load) have been tested and also real aircraft components failures (from Airbus A380, Embraer ERJ170 and Dassault F7X partial tests or full scale tests) have been used, thus a simple method has been developed and correlated to these fatigue tests. The method steps are described succinctly here. First of all, the geometry and loading must be well known (thickness, distance to wall, radius, hole diameter, fastener geometry, fitting tension load). Then loads are determined with tension fitting simple equilibrium according to geometry, prying effect (contact stress), and assembly stiffness. To do this, the research works of J. Guillot (ref.[1]) on thread screws assembly have been used. Knowing the loads equilibrium of the fitting, maximum stresses are calculated for the different critical sites, in radius between wall and end pad, near fastener head and in fastener. Test results have permitted to define failure criterion based on the comparison of fastener and fitting stiffness. Finally tension fitting fatigue life is estimated using the maximum stress and linear SN fatigue law depending of the fitting configuration (formed or machined fitting, flexible or stiff fitting), material type and its protection, scale and stress ratio effects. This fatigue law is correlated by test. Also fatigue life of fastener can be estimated with a method based on ESDU (ref.[2]) and our test specimens. Thus the method obtained gives very satisfactory fatigue life prediction for tension fitting, with: • 100% of test results in prediction band width [N/5;5N] • 91.2% of test results in prediction band width [N/3;3N] For the fastener predictions, results are also satisfactory, fastener test results are near the fastener law based on ESDU. This test program and obtained life prediction method give a good answer for our specific tension type fitting problem and help efficiently and economically assembly design, avoiding heavy complex finite elements models. Prediction method based on finite elements models and the final method based on fitting equilibrium and tests have been compared. The finite elements model predictions are less close to test results than the others, which is easy to understand because the approach is only numerical. Trust blindly numerical approach is a bad behaviour, and we have to pay attention not to forget to come back to fatigue test to take into account real aircraft part and fatigue dispersions. The work summarized here has permitted a better understanding of tension type fitting and its several parameters, and so to find explanations to test results or aircraft’s failures. As it has been shown tension fitting life predictions can be obtained very fast with accuracy with a simple method and generic tests. REFERENCE LIST [1] Guillot Jean (1987), Assemblages par éléments Filetés. Techniques de l’ingénieur 8 – 1987, B 5 560 –B 5 561-B 5 562. [2] ESDU. (1984), Fatigue strength of external and internal steel screw threads under axial loading. ESDU Fatigue - Endurance Data, Vol 4 Stress and Strength, Vol 5 item n°84037. [3] FDT stress department (2008), Note fatigue dépliage. FDT technical report N°E0010MM-A, LATECOERE internal note. |
A new Approach of Estimating Crack Growth Rate and Its Life of Aeronautical Mechanical Component Based on S-N Curve Moucun Yang, Hong Nie Abstract: The prediction of stochastic crack growth rate and its life is important for the reliability analysis of structures as well as the scheduling of inspection and repair/replacement maintenance. The celebrated Paris law prevails for crack propagation estimation. But much testing data are required in the method, such as stress intensity factor and crack grow rate which costs much. It is significative if other fatigue data, such as Wöhler S-N curve, can be used to Paris law. The relation between Wöhler S-N/P-S-N law and Paris law under constant amplitude loading is established firstly in the paper. And then, the method of forecasting crack growth rate and life is presented under constant amplitude, block-spectrum and stochastic loading. Paris curve parameters are also deduced through the parameters S-N and P-S-N curves without power function form. Finally, the idea of improving the precision of the presented method also is discussed in the paper. The dependence of the new method on test data is low and, at the same time, S-N and P-S-N fatigue data are used adequately, which decreases the product development cycle and saves much. |
Fatigue crack propagation characterization of woven CFRP composite plate with hole subject to pin load Sangyoung Kim, Junghun Choi, Hongsun Park, Minsung Kang, Jaemean Koo, Changsung Seok Abstract: Recently, CFRP (Carbon Fiber Reinforced Plastic) is more and more using in aircraft structural component. In the manufacture of CFRP aircraft structural components, various independent components are joined with those by bolts and pins. Holes for bolts and pins have an effect on the fatigue characteristic of such structures, because those act as notch in structures. Also, because load of structures with the joints transfers through pins and bolts, the fatigue crack propagation characteristic of such structures are different from those with holes subject to remote load. Consequently, a study of fatigue properties about pin load is very important. But, methods of test and analysis for CFRP material are not established at present. Recently, studies of fatigue crack propagation characteristic of CFRP material are progressing using test method of metal. But, the most of studies are progressing about remote load. In this paper, fatigue crack propagation of woven CFRP composite plate with a hole subject to pin load was characterized. For this study, plain plate specimens were manufactured according to ASTM D 3039 and CT (Compact Tension) specimens were manufactured according to ASTM E 1820. Hole of diameter 10mm was made in the center of plain plate specimens. From experiments, crack propagations per cycle and slopes of the stress-strain curves at regular time intervals were obtained. And the relations among crack length and a slope of the stress-strain curve and damage of CFRP composite plates were evaluated. Also, fatigue crack propagation characteristics are compared and analyzed in remote load and pin load. |
Accelerated fatigue testing on hydraulic shaker Mathieu Fressinet Abstract: In term of fatigue initiation, fatigue tests are commonly stopped at 107 cycles assuming that a fatigue limit exists. Indeed, under this limit the material is not damaged by fatigue and so no fatigue failure occurs. For Aluminium, it is known that such a limit does not exist and the Wöhler’s curve is extended thanks to models like the Haibach ones or thanks to statistical approaches. Unfortunately, despite these precautions, experience feedback on in service aircrafts have shown that structural elements could fail after 109 cycles. As a consequence many studies have been carried out since the beginning of the nineties to explore the domain of gigacycle fatigue (108-1010 cycles) and to characterize precisely materials behaviour. Besides, as reaching 109 cycles at 100 Hz lasts more than 16 weeks, which is not economically acceptable, many efforts have been undertaken to increase test frequencies and to set up new test processes. Fully concerned by such a phenomenon, CEAT launched its own study which is part of a project aiming at understanding the initiation and the growth of some cracks in vibratory environment, phenomena that standard tools are not able to predict satisfactorily. To accelerate the test frequency, the idea is to use hydraulic shakers usually employed to test the robustness of electronic devices in vibratory environment. The philosophy, inspired from [1] and [2], is to excite the test specimen clamped on the shaker near one of its resonance mode. The paper describes in a first part the different stages to elaborate and validate the test procedure and in a second part the first results obtained thanks to this fatigue testing method. Thus after having chosen specimen’s general design, the first task was to model it by FEM what enables to specify the different sizings necessary to get the right mode shape at the expected frequency with the right stress level. For this study, the specimen is made in 2091T3 and designed with a stress concentration area to be representative of a real structure; it also enables to reach the expected stress level at low energy. During the tests run at more than 800 Hz, special care was also taken to ensure that no consequent damping heating appears. Actually it is a well known fact that the influence of the test frequency is related to the temperature and it is assumed that its influence is minimized if the heating is not significant. To control the stress level, gauges were glued to the specimens. Unfortunately their life duration does not exceed 106 cycles and correlations had to be done between the strain given by the gauge, displacements measured by a laser and the acceleration in entry. In fact these correlations enable to control the test trough the displacement of a chosen point on the sample and so to carry out the test with the greatest accuracy even with a broken gauge. Once the test procedure was set up, tests on different samples were carried out to verify the reproducibility of such a test. Besides tests were performed near 105-106 cycles to be compared with 4 point bending tests performed with a standard fatigue test device at 5 Hz on specimens specially designed to have the same stress concentration in the studied area. In such a way the influence of test frequency was quantified. Finally after having performed tests in the gigacycle domain, this fatigue testing methodology was used to study crack initiation on pitting corrosion and to demonstrate that very short cracks could propagate even subjected to a very low stress level. [1] Methods of testing for endurance of structural elements using simulated acoustic loading, ESDU 93027 [2] Goodman Diagram Via Vibration-Based Fatigue Testing, Tommy J. George, M.H. Herman Shen, Onome Scott-Emuakpor, Theodore Nicholas, Charles J. Cross, Jeffrey Calcaterra, Journal of Engineering Materials and Technology, January 2005, Vol.129 pages 58 to 64 |
Methods for FEM Analysis of Riveted Joints of Thin-Walled Aircraft Structures within IMPERJA Project. Jerzy Jachimowicz, Jerzy Kaniowski, Wojciech Wronicz Abstract: The paper deals with the modelling of riveted joints in aircraft structures with Finite Element Method. Presented works were carried out within Eureka project No. E!3496 called IMPERJA. The goal of the IMPERJA project is to increase the fatigue life of riveted joints, which will lead to an increase of the aircraft service life, a smaller number of inspections and lower operation costs of an aircraft. The project assumed FEM modelling of the operating aircraft’s structure at three different complexity levels, namely considering the complete structure, a structural detail and a single riveted joint. The paper presents analyses of various rivet models and calculations of a structure and a riveted joint. In the first part examples of various models, at global and local level, were presented and usefulness of them was discussed , influence of the following simplification was analysed; • neglection of rivets in a model (elements are jointed continuously) • rivet as a rigid element (MPC) • neglection of contact phenomenon • neglection of secondary bending • neglection of residual stresses after riveting process The basis of the analysis was the asymmetric butt joint model with 14 rivets. The model which took into account secondary bending and contact phenomenon was analysed as well. The method of modelling residual stresses with temperature and thermal coefficient was used. In the second part, the example of analysis of riveted joint was demonstrated for a wing of PZL M28 Skytruck aircraft. It’s is a twin-engine, high-wing, cantilever monoplane of all-metal structure with maximum take-off and landing weight 7500 kg. A submodeling technique was used there. At first, part of the wing model, based on a CAD model, was built. It includes 7 ribs and 6 bulkheads between them. Dimensions of the model eliminate stress perturbation, connected with boundary conditions, in the area near the middle rib. It was a shell model. The boundary conditions were taken on a basis of operation data. Presence of rivets wasn't taken into account. Instead of this, parts were connected continuously (nodes were merged). The Linear model of material was used. The purpose of the part of the wing model was to gain accurate boundary conditions for next model of riveted joint on the middle rib. The behaviour of whole model is correct but stress distribution around rivets is not correct. A shell model of riveted joint was build. A boundary conditions were set on a basis of result from previous analysis. Forces, instead of displacements, were used, as boundary conditions, on account of a large stiffness difference between part models (part of wing and riveted joint model). The nonlinear model of material was used. A contact effect and secondary bending were taken into account. Thanks to that, phenomena around rivets were represented considerably better. Results from this analysis could be used as boundary conditions in a detailed calculation of one or few rivets with solid elements. Such a model was consider as well. The presented method allows to analyse phenomena that appear around a rivet in a real structure, during operation. Analyses were performed with MSC PATRAN and NASTRAN software. |
On the crack growth behavior under simple compresive loads of 2024-T3 aluminium alloy Milan Krkoska, Rene Alderliesten, Rinze Benedictus Abstract: In real life, many engineering components and structures are subjected to complex variable load conditions. Sudden changes in loading are known to significantly influence the fatigue behaviour of materials, commonly known as loading “interaction” effect. In this sense, the extension of a crack during particular load cycles can be influenced by loads applied during preceding cycles, and in fact, it can influence the crack extension occurring during subsequent single or multiple load cycles. Load interaction effects may introduce serious challenges in engineering prediction calculations. Load interaction can be studied in terms of crack growth rates changes on the cycle-by-cycle level from striation spacing measurements or they can be detected as the change in crack plane. Therefore, the fractographic observation of fatigued samples is indispensable technique for fatigue failure investigations. Several models were proposed in the past, regarding striation formation and crack extension during fatigue. These models assume simplified cracking conditions, particularly the constant amplitude loading, with crack plane more or less flat and perpendicularly oriented with respect to the direction of loading. Simplified, Mode I opening conditions are considered. Resulting crack growth models are mostly symmetrical in the shape. These models are not fully able to explain the physical concept of the fatigue process under variable loading conditions, since the fractured surfaces are rather rough; especially on the micro level. Additionally, the fractured surface of the specimen that failed under variable amplitude loading contains surface features that are not present at the fractured surface of specimen that failed under constant amplitude loading. Obviously, there is a great need for an increase in understanding of fatigue processes as detected in materials and structures during variable amplitude loading; reevaluation of existing models and development of useful concept for engineering practice that can be successively employed in design and analysis. Most literature concerns interaction between overloads, and constant amplitude cycles, both usually also tension. Several authors attributed this interaction to different mechanisms. To name a few: the state of residual stresses ahead of the crack tip, different crack closure mechanisms, material changes ahead of the crack tip (strain hardening, softening), crack front irregularities and sample geometry (plain stress/strain conditions). However, less literature is available on the interaction between underloads, particularly in compression mode and constant amplitude cycles. Understanding of such interaction is indispensable for understanding of the fatigue failure. This paper will present the details about the crack growth behavior of investigated 2024-T3 aluminium alloy in terms of crack growth rates and fracture surface appearance as a function of specific loading sequences; particularly the effect of underload (compressive) cycles and constant amplitude (tensile) cycles. Test data showing the effect of individual single underloads, repeated underloads and groups of underloads on fixed and varying number of constant amplitude loads will be presented. The main focus will be given to the crack growth rates and analysis of fractured surfaces by means of electron and light microscopy. |
An analytical model for load transfer in a mechanically fastened, double-lap joint Ligieja Paletti, Calvin Rans, Rinze Benedictus Abstract: In mechanically fastened joints, fatigue cracks have been observed, both in test and in practice, to nucleate in one of two primary locations. The first location is close to the hole edge, along the joint net-section. The second location can be found at some distance from the hole edge, causing gross-section failure. The preference for crack to nucleate at either location is determined by the prevalent load transfer mechanism. For cracks nucleating at the hole edge, bearing has been proved to be the main load transfer contribution. When the crack nucleate remote from the hole edge, friction is believed to be the primary cause. Existing models for the load transfer in mechanically fastened joints usually ignore the contribution of friction. Neglecting friction implies neglecting the related failure scenarios. Therefore, fatigue models based on these load transfer models give predictions whose validity is limited to bearing-dominant cases. Existing models for frictional load transfer describes the behavior of the friction force acting between the sheets of the joint. Those models focus only on the contacting bodies. The drawback is that the applied load is assumed to be transferred fully by friction; this assumption is wrong in a joint, where also the effect of bearing must be considered. The knowledge of the load split between the two mechanisms must then be determined. A load transfer model which includes the contribution of friction is needed in order to improve the accuracy of fatigue models. The predominant load transfer mechanism indicates the most reliable failure model for a particular joint configuration. In order to distinguish between the two main mechanisms, a parameter is introduced which identifies the prevalent load transfer mechanism for a chosen joint configuration. In this paper a model for the load transfer in a double-shear, push-fit, bolted joint is presented. By using this configuration, it can be assumed that only bearing and friction act in the joint and all the other contributions to load transfer are neglected. A key input for the model is the pressure distribution around the hole after the tightening process. The clamping pressure acting between the plates is modeled as the Hertzian pressure distribution. The frictional load is therefore calculated by using the contact mechanics equations and the Coulomb’s model of friction. An additional input is a model for the bearing load; in this study, bearing is modeled assuming a cylindrical, push-fit, rigid pin inserted in a hole without friction. A parameter a represents the splitting factor between the two load mechanisms considered. This parameter is defined as the percentage of the external applied load transferred by bearing and depends on both the clamping force and the external applied load. An iterative procedure is established for the determination of a. The output of the analytical model allows the determination of the prevalent load transfer contribution, and, therefore, the appropriate failure model. Moreover, a transition point is identified for a certain value of a. At this point the load transfer mechanism changes from friction only to a combination of friction and bearing. This transition is reflected in a difference in crack nucleation location. Experimental evidences are presented. The model presented has to be considered as a starting point for the development of a more general model. In order to describe the behavior of more generic lap joint, the complexity of the model must be increased, including multiple fastener rows and time-varying clamping pressure distribution due to bending. |
Simulation of Fatigue Crack Growth in the High Speed Machined Panel under the Constant Amplitude and Spectrum Loading Petr Augustin Abstract: Application of new production methods such as high speed machining, laser beam welding or friction stir welding in the aerospace industry leads to the reduction of manufacturing costs. Structures designed for these production techniques are typical by the integral character. Comparison of their behavior with differentially stiffened designs shows lack of damage tolerance characteristics and that is why this topic is investigated. The paper describes methodology of numerical simulation of fatigue crack growth under constant amplitude and spectrum loading and its application on integrally stiffened panels manufactured using high speed machining. Relatively large experimental program comprising both the constant amplitude and spectrum tests on integral panels and CCT specimens was undertaken at the Institute of Aerospace Engineering laboratory in order to acquire crack growth rate data and enable verification of analyses. The work was performed within the scope of the 6th Framework Programme project DaToN - Innovative Fatigue and Damage Tolerance Methods for the Application of New Structural Concepts. Presented methodology of crack growth simulation starts by the calculation of stress intensity factor function from finite element results obtained using MSC.Patran/Nastran. Prediction of crack propagation in stiffened panel requires determination of stress intensity factors for large number of crack configurations and that is why simple FE models comprising shell elements were built. The crack closure technique was used for calculation of stress intensity factor values. Subsequent crack growth analysis is done in NASGRO and uses description of crack growth rates by the Forman-Newman-de Koning relationship. Two crack growth models were applied for spectrum loading: non interaction and Willenborg model. A typical feature of integrally stiffened panels is branching of cracks growing through the stiffener. Simulation of this phenomenon requires consideration of parallel propagation of two cracks during the cycle-by-cycle computational procedure. This problem was treated using two-dimensional growth option in NASGRO. The methodology described above was applied for analysis of two-stringer panel tested within the experimental program of the DaToN project. The panel was made of 2024-T351 aluminium alloy using high speed cutting technique. Fatigue cracks in the skin were initiated from the central saw cut with the size of 2a = 20 mm. First analyses and verification tests of panels were performed under the constant amplitude loading at maximum nominal stress of 80 MPa with stress ratio R = 0,1 and at maximum nominal stress of 110 MPa with R = 0,5 respectively. For predictions of crack growth and experimental verifications using the spectrum loading, a load sequence representing service loading of the transport airplane wing was prepared. Applied load spectrum was measured on B737 airplane within the joint FAA/NASA collection program. The load sequence is composed of 10 flight types with different severity, analogous to the standardized load sequence TWIST. Generation of random sequence of loads and flights is realized by the in-house computer program. Before application on the stiffened panels, a calculation of crack growth under the spectrum loading was performed for simple CCT specimen geometry. Since analytical solution of stress intensity factor function is known in this case, it was possible to verify crack growth models used without the influence of inaccuracy of stress intensity factor determination based on FEM. The paper finally presents comparison of simulations of fatigue crack propagation in two-stringer stiffened panel under the spectrum loading with verification tests of two panels performed in the Institute of Aerospace Engineering lab. |
Bonded composite patch to repair metallic structures: Disbond propagation testing and modelling Pierre Madelpech Abstract: Our general objective is to progress in the certification of metallic structure repair process by bonding a composite patch on the damaged part. On military aircraft, some countries already use that repair and its mechanical efficiency is now well established. The composite patch acts as an efficient stress bypass avoiding any supplementary hole drilling and associated stress concentration. Nevertheless the process is not used in France because of a lack of confidence in bonding. This is still today a major certification concern, with a lot of remaining needs to assure the bond quality and durability. Debonding can initiate due to wrong bonding process, to environmental aggression or as a result of a static or fatigue loading. Although this is a real problem observed on in-service patched structures and during tests, until now, no hazardous consequences have been noticed. Moreover a disbond can be detected by classical NDI. Therefore, disbond propagation can be controlled if a structural inspection program is defined and followed. With such a procedure, damage tolerance philosophy requirements may be fulfilled. The study is part of a wider damage tolerance approach of bonding defect initiating and propagating between the metal and the composite. For this purpose, initiating and propagating must be predictable with sufficient confidence. In fact, this paper only addresses the problem of predicting the growth of an initiated disbond between the metallic structure and the repair. Added to the current problem of studying bonding, the composite repair of metallic structures is more complex, since the assembly needs a curing at 120°C. Indeed such a process raises a problem, as the difference of thermal dilatation coefficients between the two materials creates residual stresses after cooling. The modelling of the disbond growth has been realised, based on total energy release rate. The numerical calculation of an aluminium plate covered by a composite patch has been performed using Samcef. The adhesive has been modelled by interface elements. It gives precisely access to the stress in the adhesive avoiding all at once too refined mesh and degenerated elements. It is essential for the study of a phenomenon occurring in the adhesive. This approach based on energy release rate does not differentiate the different kind of solicitation (related to stress intensity factors KI and KII). The initial bonding defect has been supposed to be circular and the propagation path has been imposed according to common sense but also to test results. Step by step, the disbond is modelled and the energy release rate is computed. From the propagation law, the number of load cycles necessary to cause that size increment is determined. Thus, the propagation is predicted and the life time of the bonding is obtained. The study also includes test to determine the unidirectional propagation law of a disbond. The total energy rate is more efficient when the ratio of solicitation mode (KII/KI) remains homogeneous between computing and testing. Fatigue testing was performed in order to characterize the linear elastic fracture parameters of the adhesive. The design of the fatigue coupons was therefore chosen as close as possible to repair conditions on aircraft. Moreover, it was not relevant to perform traditional tests on the adhesive as debonding can be cohesive or adhesive. The two different interfaces (Carbon – adhesive and Aluminium – adhesive) must consequently be present in the specimen. During the test, the growth has been regularly measured by common NDI and the coefficients of the Paris law have been deduced. The successive forms of the disbond have been used to define and control the modelling of the disbond evolution. Finally, it gives a tool to estimate the propagation law of a disbond occurring under a composite patch. |
Three Dimensional Crack Growth Prediction Sarah Galyon, Saravanan Arunachalam, Scott Fawaz, James Greer Abstract: In complex aircraft structure, crack growth rarely propagates in the idealized fashion assumed in durability and damage tolerance analyses (DADTA). Usually the applied loading is not perpendicular to the crack nucleating feature and subsequent crack propagation. This situation is known as mixed mode crack growth or in more general terms, three dimensional crack growth. Most DADTA’s conducted assume mode I only; thus, engineering judgment is used to estimate the amount of error present in the idealized models. The Center for Aircraft Structural Life Extension (CAStLE) at the United States Air Force (USAF) Academy completed a project to generate three dimensional (3D) crack growth data and predict the measured crack growth rate, crack trajectory and residual strength using state-of-the-art stress analysis and life prediction tools. Specifically, we generated the 3D fatigue crack growth test data using 1.6 mm (0.063 inch) thick aluminum alloy (AA) 2024-T351 ARCAN specimens in an ARCAN test fixture. The ARCAN test fixture allows the ARCAN sample to be rotated to produce different mixtures of mode I and mode II loading (with 0° being pure tension/compression (mode I) and 90° pure shear (mode II)). ARCAN specimens were tested at angles of 0°, 30°, 60° and 75° under constant amplitude loading and a stress ratio (R) of 0.1. The stress amplitude was adjusted to control the crack growth rate and plastic zone size. A grid on the surface of the sample was used to optically track crack trajectory and crack growth rate. While mechanical testing was being completed, we also developed a crack prediction model of the ARCAN specimen using FRANC3D/NG, a solid modeling, mesh generation and fracture mechanics code from Cornell University Fracture Group. FRANC3D/NG should be able to predict the cracking behavior observed in the ARCAN tests. A parallel effort was also undertaken to develop an engineering model of mixed mode crack growth where contributions to mode I and mode II growth were accounted for using KI and KII and the appropriate baseline crack growth data. For both the FRANC3D/NG and engineering model analysis, crack growth rate data is required and was produced per ATSM E647 using a 15.24 cm (6 inch) wide, 1.6 mm thick AA 2024-T351 M(T) specimen in both the LT and TL orientations under constant amplitude loading and an R of 0.1. The fatigue crack growth trajectory prediction for the ARCAN specimens was assessed with three different measures. The first was the point-wise comparison of the measured and predicted crack angle for each of the test conditions. The second was a block comparison of the measured and predicted crack angle. A block is defined as a discrete amount of crack growth, so the crack growth was compared at ¼, ½ and ¾ the total crack length. The third was examining the effect of cracking angle on the residual strength or critical crack size of the specimen. Comparisons were also made of the predicted and observed fatigue life, number of cycles at the end of the test and crack growth rate. The success of the prediction model is based on how the correlation of the model affects the ability to predict the measured fatigue and fracture performance based on the mechanical testing. Application of the UniGrow Model for Fatigue Crack Growth Pr |