AIAA JOURNAL Vol. 40, No.

8, August 2002

Prediction of Residual Strength and Curvilinear Crack Growth in Aircraft Fuselages
Chuin-Shan Chen¤ National Taiwan University, Taipei 10617, Taiwan, Republic of China and Paul A. Wawrzynek† and Anthony R. Ingraffea‡ Cornell University, Ithaca, New York 14853
Two practical engineering approaches to assess the structural integrity of aircraft fuselages are presented. Both approaches combine fracture mechanics with thin-shell nite element analyses to predict structural responses quantitatively. The rst approach uses the crack-tip opening angle fracture criterion to predict residual strength of KC-135 fuselages. The second approach uses T-stress and fracture toughness orthotropy to predict crack-growth trajectory in narrow-body fuselages. For residual strength prediction 12 damage scenarios, which might occur in applications, are examined. It is found that the model with small multiple-site cracking and material thinning caused by corrosion damage has the worst load-carrying capacity. For curvilinear crack-growth simulation a directional criterion based on the maximum tangential stress theory accounting for the effect of T-stress and fracture toughness orthotropy is used to predict crack-growth trajectory. Both T-stress and fracture toughness orthotropy are found to be essential to predict the observed crack path, where the trajectory experiences crack turning and apping phenomena.

C

I.

Introduction

ONCERN about residual strength of aging aircraft prompted the development of an advanced analysis toolkit for fatigue and fracture problems.1 One component of the toolkit FRANC3D/STAGS has been developed to assess damage tolerance in critical aircraft structures.2;3 The program is capable of tracking crack growth using FRANC3D 4 and performing nonlinear thin-shell analysis using STAGS. 5 This paper presents two recent studies on simulating crack growth in airplane fuselages using the FRANC3D/STAGS program. The software program FRANC3D is available from the Cornell Fracture Group’s web site at www.cfg.cornell.edu.Animations of stable crack growth and plastic zone evolution in the KC-135 fuselage panel and curvilinear crack growth in the narrow-body fuselage panel are also available from the web site. The rst example is residual strength prediction for a typical KC-135 fuselage panel.6 The crack-tip opening angle (CTOA) fracture criterion7 and elastic-plastic thin-shell nite element analysis are used to characterize stable crack growth and predict residual strength. The objective of this study is to examine several damage scenarios that might occur in pressurized fuselages and their effects on residual strength prediction. In particular, lead crack growth, presence of multiple-site damage (MSD) from rivet holes, multiple crack interaction, and material loss as a result of corrosion damage are studied. The second example is curvilinear fatigue crack-growth simulation in a generic narrow-bodyfuselage.This is relatedto the growing interest by using crack turning phenomenon to improve the structural integrity of aircraft structures.8¡10 A reliable crack trajectory prediction for fatigue and fracture thus serves as a crucial step to use
Received 12 December 2000; revision received 24 February 2002; c 2002 by the American accepted for publication 1 March 2002. Copyright ° Institute of Aeronautics and Astronautics, Inc. All rights reserved. Copies of this paper may be made for personal or internal use, on condition that the copier pay the $10.00 per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923; include the code 0001-1452/02 $10.00 in correspondence with the CCC. ¤ Assistant Professor, Department of Civil Engineering, No. 1, Section 4, Roosevelt Road. † Senior Research Associate, Cornell Fracture Group, 641 Rhodes Hall. ‡ Dwight C. Baum Professor of Engineering, School of Civil and Environmental Engineering. 1644

this evolving methodology. In this study linear elastic fracture mechanics (LEFM) parameters including stress intensity factors and T-stress, together with orthotropy of fracture toughness, are used to predict the direction of fatigue crack growth. The objective is to study the effects of various fracture parameters on crack-growth trajectorypredictionand validate the analysis methodology by comparing with experimental observations.

II.

Residual Strength Prediction of KC-135 Fuselage

A crown lap-jointed KC-135 fuselage panel is chosen for analysis. This is a three-stringer-wide,three-frame-long, fuselage panel. The panel section has a radius of curvature of 72 in. (1829 mm). It contains a lap joint at the central stringer. The lap joint is a typical 3 three-row con guration with 16 -in. (4.76-mm)-diam countersunkhead rivets. The other two stringers are spot-welded to the skin. The upper and lower skins are 0.04-in. (1.02-mm)-thick, 2024-T3 aluminum alloy. The stringers and frames are 7075-T6 aluminum alloy. Frames are connectedto stringersby rivets.The panel con gurationsare shown in Figs. 1 and 2. The frame and stringerdimensions are shown in Fig. 3.
A.

All structural components including skins, stringers, and frames are modeled by shell elements. A piecewise linear representation is used for the uniaxial stress-strain curves for 2024-T3 and 7075-T6 aluminumalloys(see Figs. 4 and 5). Symmetric boundaryconditions are imposed on all of the boundary edges to simulate a cylinderlike fuselage structure. Pressure loading is applied on all of the external skins. Both geometric and material nonlinearities are included in the analysis. Rivets are modeled by one-dimensional elastic-plastic fastener elements that connect shell nite element nodes in the upper and lower skins. Each fastener element has six degrees of freedom, corresponding to extension, two shearing, two bending and twisting of the rivet. The stiffness of each degree of freedom is de ned by prescribing a force-de ection curve. The axial, exural, and torsional stiffnesses are computed by assuming that the rivet behaves like 3 a simple elastic rod with a diameter of 16 in. (4.76 mm). The elastic shear stiffness of the fastener element is computed by the following empirical relation developed by Swift11 : K rivet D ED [ A C C . D = t1 C D = t2 /] (1)

Numerical Model

and three-dimensional elastic-plastic fracture have been conducted to investigate the constraint effects on residual strength prediction. t1 and t2 are the thicknessesof the joined sheets. Fig.13. Although thin-sheetstructuresbehave essentiallyin plane stress.4 mm. 6. where E is the elastic modulus of the sheet material. WAWRZYNEK. Previous studies of residual strength prediction using two-dimensional elastic-plastic analyses had mixed success. This concluded that.18 The plane-strain core concept is used for all analyses presented here. 2 Detailed rivet spacing for the KC-135 fuselage panel.CHEN.15¡ 17 From these comes the concept of a plane-strain core. two-dimensional simulations with the plane-strain core concept seem to capture the effects for residual strength prediction. and A and C are empirical constants. 6).8 for aluminum rivets. 4 KC-135 fuselage panel: piecewise linear representation of the uniaxial stress-strain curve for 2024-T3 aluminum. 6 KC-135 fuselage panel: rivet shear stiffness and strength. the predicted residual strength would agree with test results up to a certain width of the panels. Although the height of the plane-strain core has a modest effect on residual strength prediction. the width of the plane-strain core has a relatively minor in uence. Fig. Fig. The criticalcrack-tipopeningangle (CTOA c ) is used to characterize elastic-plasticcrack growth and to predict residual strength. Fracture Criterion and Crack Con gurations . The discrepancy between twodimensionalpredictionsandtest resultsfor larger panelswas thought to be related to the three-dimensional constraint effects. 5 KC-135 fuselage panel: piecewise linear representation of the uniaxial stress-strain curve for 7075-T6 aluminum. 3 Dimensions of stringer and frame for the KC-135 fuselage panel (dimensions in inches. which equal to 5 and 0. 7). Once the fastener reaches its ultimate strength. The initial shear yielding and ultimate shear strength of the rivets are assumed to occur at load levels of 510 lb (2268 N) and 725 lb (3225 N). local to the crack tip. = 25. which assumes that for thin-sheet structures a small near-tip region reaches plane-strain conditions whereas the rest of the structural behaves essentially in plane stress.A 50% reductionof the shear stiffness is assumed after the initial yielding. although three-dimensional simulations are best to quantify the constraint effects. The shearing force-de ection curve is shown in Fig. For two-dimensional and thin-shell problems the width and the height de ne the plane-strain core region (Fig. D is the rivet diameter.7 Typically.12¡ 14 Extensive numerical studies of two. to approach plane-strain conditions.the constraint from the nite thickness of the specimens can cause the regions. Fig. 1 Dimensions of the KC-135 fuselage panel. modi ed after Ref. Fig. it breaks and loses its load-carryingcapacity. The CTOA fracture criterion asserts that the angle maintains a constant value during stable crack growth for a given thickness of a metallic material. AND INGRAFFEA 1645 Fig. B. 1 in. respectively.

(1. For the case without MSD. (254 mm) 10.edu. (0. similar to the conclusionsdrawn in Ref. 12 indicate the following: 1) The MSD cracks signi cantly reduce the residual strength of the fuselage panel. the bulging at the lower crack edge is much smaller than the opposing side. 20. (254-mm) initial lead crack and MSD (external view).046 in. Out-of-plane bulging is observed in the skin crack edges. (254-mm) initial lead crack but without MSD and corrosion damage is shown in Fig. dominant plastic zones accompanying the lead crack tips are observed.0 in.0 in. (1. (254-mm) initial lead crack is well below the required strength.cfg. 13. Predicted residual strengths summarized in Fig. Because of the stiffnessof the stringer. The plasticity distributions are highly in uenced by the multiple crack interactions. The plane-strain core height is assumed to be twice the sheet thickness. 7 Schematic of the plane-strain core concept.04 in. 2) A 10% uniform thickness degradation caused by corrosion damage reduces the residual strength by 3. The coupling of MSD and corrosion damage leads to the most severe damage scenario.cornell. for the cases with a 10-in.8 to 28. 9 and 10.14 in.64 mm) 0.03 mm).14 in. (181. .025 in. Detailed mesh around crack path for the KC-135 fuselage The effect of material thinning is modeled by a uniform reduction in thickness of the upper skin at the lap joint in the two center bays.64 mm) 0. The MSD pattern is symmetric about the lead crack at the three rivets in front of the lead crack. A mesh pattern with 0. C. (254-mm) initial lead crack as the panel reaches its residual strength. For the case with MSD.36 mm) 7.4 to 9. 8 Crack con gurations with a 10-in.a small crack with a lengthequal to the rivet diameter plus the MSD length is used to model the MSD crack. (2. Fig.The plastic zone is composed from those elements with computed effective stresses exceeding the initial yield stress. the evolving plastic zones are well-con ned by the elastic regions within the frames.0 in. The predicted residual strength for the MSD and/or corrosion damage cases with a 10-in. Six different crack con gurations with various lengths of lead and MSD cracks are studied. 14.18 kPa). measured 0. In addition to the effects of MSD. (254 mm) 10. (254 mm) MSD crack lengtha None 0. Fig.0%. As shown in Fig. The initial con gurations prior to crack growth are shown in Table 1. (181. (254-mm) initial lead crack are shown in Fig. the animation of the plastic zone evolution is available from the Cornell Fracture Group’s web site at www. 3) In general. Fig. Because rivet holes are not modeled explicitly in the nite element model.29-mm)-thick.17 mm) from both sides of a fastener hole.7 deg. The crack con gurations with a 10-in.02-mm) crack-tip elements is used. However.25 deg used in 0. 16 and 19. AND INGRAFFEA Initial crack con guration for KC-135 model Lead crack length 7. The CTOAc that is used is 5. increasing the lead and MSD crack lengths reduces the residual strength. Figure 11 shows the predicted results of the operating pressure loading vs the total crack extension for all of the cases conducted in this study.0% loss of residual strength as a result of the presence of small MSD is observed. (254-mm) initial lead crack residual strength seems to be relatively insensitive to the MSD crack sizes.09-in. The deformed structure at residual strength for the case with a 10-in.5 psi (86. 4) The required residual strength for the aircraft fuselages is 12. This pattern is similar to the one used in the at-panel simulation reported in Refs. (2. (0. A nite element mesh for the model is shown in Figs. The lead and MSD cracks are located along the upper rivet row in the upper skin of the joint.025 in.36 mm) 7. The unsymmetric out-of-plane bulging leads to an antisymmetric bending deformation eld at the crack tips. WAWRZYNEK.36 mm) 10. 8.046 in.1646 Table 1 Model 1 2 3 4 5 6 a Emanating CHEN. No experimental crack-growth data are available to date for this material and thickness. plastic zones are developed at the multiple crack tips. with a plane-straincore height equal to 0. Numerical Results and Discussion Fig. 17 and 19. (1.21 Figures 14 and 15 depict the predicted plastic zones for the cases with a 10-in.04-in.02 mm) behind the crack tip. 9 Finite element mesh for the KC-135 fuselage panel. material thinning as a result of corrosion damage is also studied.17 mm) None 0. this particular CTOA c value is estimated based on the 5. (1. (181. 10 panel. A 21. 2024-T3 bare material used in Refs.08 in. The lead crack is located symmetrically about the central frame line.14 in.

WAWRZYNEK.17-mm) MSD.0]. Fig.64-mm) MSD.025-in. (1. (0. magni cation factor = 5. N. M. (1. 15 Predicted plastic zones for 10-in. 0.025-in.64-mm) MSD.14-in. and s . corrosion.64-mm) MSD [pressure = 11. corrosion. (0. no MSD. Fig. (181. Skins and tear straps were 0. AND INGRAFFEA 1647 7. 13 Deformed shape of the KC-135 fuselage panel [pressure = 15. magni cation factor = 5. d . (181. frames.025-in. (0. 11 Predicted operating pressure vs total crack extension for the KC-135 fuselage panel: ¤.91-mm) thick. and ——.025-in.036-in. 12 Predicted residual strength vs initial lead crack length: ¤. (254-mm) initial lead crack without MSD [pressure = 15. stringer clips.3 psi (77. 0. and stringer clips were 7075-T6 clad aluminum alloy.0].4-mm) initial lead crack 10-in. no MSD. Curvilinear Crack Growth in a Narrow-Body Fuselage A.3 psi (105.5 kPa). curve t. 0.3 psi (105. 0. (0.9 kPa). and frames was tested by the Boeing Commercial Airplane Group.CHEN. (0. 0.0]. The tear straps were hot-bonded Description of Experiment Fig.4-mm) initial lead crack with corrosion damage 10-in. no MSD. Stringers. (254-mm) initial lead crack with corrosion damage Fig.17-mm) MSD.046-in. stringers. (254-mm) initial lead crack with 0. corrosion.5 kPa).046-in.14-in. (254-mm) initial lead crack 7. ¥.046-in. M. A generic narrow-body fuselage panel with tear straps. (1. magni cation factor = 5. .17-mm) MSD. 0. 2024-T3 clad aluminum alloy. s . 14 Predicted plastic zones for 10-in. Fig.64-mm) MSD. III.

16 Structural features of a generic narrow-body fuselage panel (modi ed after Ref. (508 mm) of crack extension.cornell.edu.cfg. (1880 mm) to match narrow-bodyairplanes. More information about panel dimensions can be found in Ref.27 Figure 18 shows the predicted propagation angle with the effect of T-stress.PR=2/ ¢ L is assigned.2 It is assumed that the crack will start to grow from the tip in the direction perpendicularto which the tangentialstress ¾µ µ is maximum. using a globallocal hierarchical modeling approach3 could require continual updating of the boundary conditions from the preceding model in the hierarchy as a result of the crack growth. This criterion was developed by Erdogan and Sih. Considering only the singular stresses (or the rst-orderterms) from the asymptoticexpansions. The panel was inserted into a test xture with a radius of curvature of 74 in. 17 panel. µ / Maximum K c .µc =2/] 8 (3) £ ¤ 1 ¯p C 2 T sin µc cos µc D 0 £ ¤ Fig. (127-mm) initial saw cut in the transverse-longitudinal orientationcenteredon the midbay tear strap and just above the stringer tear strap. K I . 25.8 kPa) was applied to propagate the crack.3 cos µc ¡ 1/ [ K I cos. AND INGRAFFEA C. where r c is a “critical” distance away from the crack tip and T is the constant stress or the T-stress.µc =2/ ¡ K II sin. 3. 10 A dimensionless parameter T 3 is introduced to normalize the T-stress effect.®/ D q ¯£ 1 The entire curvilinear crack-growth simulation consists of more than 20 in. and frames were modeled by quadrilateralshell elements.8 psi (53. Various directional criteria have been proposed. 9). K II . The saw cut went completely through both the skin and midbay tear strap. The physical meaning of rc has been studied in Ref. During the test. On this boundary edge. Finite element model for the generic narrow-body fuselage to the skins at midbay and at each frame station. the crack propagation angle µc is determined by solving 1 ¯p 2¼ r c ¡ 3 cos. 3 and online at http://www. In this study a four-stringer-bay wide and two-frame-bay long panel was analyzed.1= K ¤ (5) Fig. Crack-Growth Directional Criteria (2) where K I and K II are mode I and mode II stress intensity factors.A cyclic pressure of 7. B. stringers. 17. For this speci c problem only internal pressure was applied to the structure.3 cos µc ¡ 1/ D 0 Fig.2¼ r c / used in Ref. All structural components including skins. Zaal22 gives an excellent review of this subject. 16. 18 Predicted propagation angle from the maximum tangential stress theory with T-stress [Eq. thus. the positions of the crack tips were recorded.3 cos µc ¡ 1/] 4 2¼ r c ¡ 3 . K c .0-in. a simple numerical model using symmetric boundary conditions might suf ce to simulate the panel test. This would increase efforts substantially in performing the numerical analyses. The asymptotic stress expansion from the crack tip including the T-stress and other higher-order terms can be found in Ref. Each node of the shell element has six degrees of freedom. A simple elliptical functionis used to characterizethe anisotropic fracture toughness K c . where P is the applied pressure.®/ C 2 T cos 2µc < 0 µ ¶ D ¾µ µ Kc ´ critical at µ D µc (4) where ® is the angle characterizingthe material grain orientationand K c . it is assumed that the crack propagatesin the directionof maximum normalizedstress. Pressure loading is applied on the skin of the shell model. the equations used in this study to compute the fatigue crack-growth direction are outlined. and L is the arc length of the edge.23 If one includes the constant stresses (or the second-order terms) from the asymptotic expansions. Numerical Model N m /2 sin2 ® cos2 ® C .the directionof crack growth or the crack propagation angle µc is determined by solving K I sin µc C K II . This directional criterion was proposed by Williams and Ewing26 and later corrected by Finnie and Saith. that is. 28 and 29). Symmetric boundary conditions are imposed on all of the boundary edges of the model to simulate a cylinder-like fuselage structure. Geometrically nonlinear analyses are performed. As a result. A typical nite element mesh used in the simulation is shown in Fig.®/ is the strength parameter characterizingthe material fracture resistance. an axial force equal to .®/ (Refs. Uniform axial expansion is allowed at one longitudinal end.µc =2/ [ K I sin µc C K II . N ´ 8 . WAWRZYNEK. rc . R is the radius of the panel. 24. The detailed test data can be found in Ref. The structural features of the test panel are shown in Fig. T . T =K I /p. .1648 CHEN. In the following. The tested panel had a 5. (3)]. For crack growth in orthotropicmedia. Detailed derivations can be found elsewhere. such that ¾µ µ .

N m and the material orientaBoth the fracture orthotropy ratio K tion angle ® can affect the predicted angle of impending fracture propagation µc . (3) and (4) for the isotropic and orthotropicmedia. (3)]. From the coupon test results the fracture toughness p p for this material and thickness is about 100 mPa m) in the p ksi in: (109p p L direction and 105– 120 ksi in: (114 mPa m – 131 mPa m) in the T direction. Predicted results suggest the following: 1) The T-stress has a very mild in uence on the early crack trajectory prediction because of its relatively small magnitude. when the crack grows further into the tear strap region the inclusion of fracture orthotropy adversely alters the crack path predictionand does not predict apping as observedin the panel test.30 A simple displacement correlation method is used to evaluate the T-stress10. 2.CHEN. As a . (2. k 1 . Fig. Eqs. but with some discrepancy. measured. Fracture Parameter Evaluation puted deformed shapes during curvilinear crack growth. The crackgrowth directional criteria.09 in. This observation gives a preliminary indication of the experimental scatter that might occur in the panel test. during early stages of crack growth the predicted trajectories for the orthotropic case agree better with the experimental measurements than the isotropic case. K c . k2 /:3. T / is the fracture toughness along the transverse (T ) direction. M. Fig. T /= where K K c . Figure 21 plots the predictedcrack trajectories with r c D 0 (T-stress effectsignored) and r c D 0:09 in.09 in. 20 Effects of material orientation on predicted propagation angle. Effect of T-Stress and rc The effect of T-stress and rc on crack trajectory prediction is studied rst. L /. 24. as demonstrated in Figs. T-stress increases and plays an important role in the crack turning prediction. (4). (2. Predicted crack paths from both numericalsimulationsat the right and left crack tips are almost symmetric about the midbay. (2.29 mm) as well as the experimentalmeasurements.3 Thus. 19 and 20. 19 Effects of fracture orthotropy ratio on predicted propagation angle.32 its counterpart for curved shells is not availableto date. severe apping is predicted as the crack turns. Crack-growth simulation with fracture orthotropy also predicts crack turning as the crack approachesthe tear strap. Figure 23 shows the computed stress intensity factors and T-stress vs the half-crack extension at the right crack tip.29 mm) are comparable at the early stage of curvilinear crack growth. This drastically changes the computed values of fracture parameters. 21 are comparable to the experimental measurements. WAWRZYNEK. For the case with rc D 0:09 in. In subsequent analyses the orthotropic directional criterion. As shown in Fig.31 . a 5– 10% error in the computed T-stress is expected. N m is the fracture orthotropy ratio de ned as K N m ´ K c . the fracture toughness is assumed to be 10% higher in the T than in the L direction. The modi ed crack closure integral method is used to compute the membrane and bending stress intensity factors . 2) The computed fracture parameters for rc D 0 and 0:09 in.29 mm) are compared with those from the isotropic prediction and experimental measurements. The predicted crack trajectories with r c D 0. Although a better numerical method for T-stress computation using a path-independent integral is available for twodimensional models. However. E. But as the crack approaches the tear strap.Figure 22 shows the com- The predicted crack-growth trajectories depicted in Fig. Numerical Results 1.26. 3. that is. Several possible reasons might explain why the current methodology including the fracture toughness orthotropy does not predict the desired apping and should be examined in future research: 1) A characteristic feature of fracture orthotropy in the tear strap region is that the material orientation in the tear strap differs from that in the skin (that is. and s . K II . L / is the fracture toughness along the material longitudinal ( L ) direction and K c . Eq. Crack-growth direction is predicted by the isotropic directional criterion[Eq. sharp turning as the crack approaches the tear strap alters the deformation and stress elds. 21 Comparisons between predicted and measured crack trajectories (isotropic directional criterion with various magnitudes of rc ): ¤. but the measured crack paths are not. (2. is used to predict the propagationangle. K I .29 mm)]. D. are used to predict the propagationangle in thinshell structures. Yet. a sharp turning caused by T-stress is predicted as the crack approaches the tear strap. The disagreement during early stages of crack growth might be related to the fracture toughness orthotropy of the fuselage skins. Effect of Fracture Toughness Orthotropy Calculated deformation and stress elds near the crack tip are used to compute fracture parameters for crack-growth simulations. predicted [rc = 0. the transverse direction in the tear strap is along the longitudinal direction of the skin and vice versa). Using the simple displacementcorrelationmethod. AND INGRAFFEA 1649 Fig. predicted (rc = 0). Bulging caused by the applied pressure is observed. Moreover. (2.29 mm). The sign conventions of stress intensity factors follow those in Ref.

23 Computed stress intensity factors and T-stress vs half-crack extension.0 in. magni cation factor = 2.09 in. However. the material characteristics in this overlapped region might behave like a quasi-isotropic material with less fracture toughness orthotropy. 2) The skin and tear strap are assumed to be perfectly bonded in the current model. particularly in the crack-tip region. The hollow and solid markers denote the computed fracture parameters for the isotropic case with rc = 0 and 0. for example. 22 a = 23.6 in.0 in. Kc ( T )/Kc (L ) = 1. Computed deformed shapes during curvilinear crack growth [isotropic case with rc = 0. respectively. 5) The crack-growth directional criterion and its subsequent curvilinear crack-growth simulations explicitly assume that the crack is grown under small-scale yielding conditions. Fig.09 in. a = 21. as the length of the fatigue crack extends to tear strap. Yet. Fig. a = 12. 4) Accurate stress intensity factor and T-stress evaluations in this region are crucial to predict crack turning. as the crack grows into this region the adhesive bond between the skin and tear strap is likely to fail. predicted (isotropic). result.1). measured.1650 CHEN. predicted (orthotropic. Additional three- dimensional crack-growth simulations are needed to quantify the three-dimensionaleffect on crack turning prediction. AND INGRAFFEA a = 5.8 in. 24 Comparisons between predicted and measured crack trajectories [isotropic and orthotropic cases with rc = 0. WAWRZYNEK. Fig. 3) The thin-shell approximation does not capture all of the threedimensional complexities of the problem in the vicinity of the tear strap. (2. path-independent integrals for geometrically nonlinear shells. s .0]. and M.09 in.29 mm).29 mm). Current crack-growth simulations use a low-order polynomial degree of shape functions for thin-shell nite element analysesand use a displacementcorrelation method to extract the T-stress term from the nite element solutions. This inevitably alters the local crack-tip stress elds and consequently affects the crack-growth behavior. stable tearing and .Other numerical methods. which can be inaccurate. should improve the accuracy of fracture parameter evaluations. (2. (2.29 mm)]: ¤.

33 have reported numerical simulations for this problem previously. The kinematic boundary conditions on the 2 £ 2 bay model were not updated during crack growth. The subsequent crack growth after the initial crack de ection follows a trajectory where the local stress states are of a mode I dominant type. Thus. (2).34 ). AND INGRAFFEA 1651 extensive plasticity are likely to occur. Thus. s . more shallow crack trajectories are predicted in the present study. In the previous studies the kinematic boundary conditions used in the lowest level in the hierarchy were obtained from a global model with an initial 5. 26 at zero crack extension. This can be observed from the computed values shown in Fig. The hollow and solid markers denote the computed stress intensity factors from the current isotropic prediction with rc = 0 and those from Chen et al. In Potyondy et al.. Nevertheless. like crack growth in double cantileverbeam specimens. the computed fracture parameters are comparable with previous results.43 mm) away from the intersection of the skin and stringer as a result of limitations in the previous version Fig. predicted (current rc = 0). of interest to compare these results with the current prediction. Three hierarchical modeling levels were employed.33 a global-local hierarchical modeling approach was used to model the panel test. The driving force for this case would be less than the one with updated boundary conditions as the crack grows. a similar fatigue life is anticipated. predicted (Potyondy et al. Fig.The analysis methodologytogether with the FRANC3D/STAGS program is shown to be an effective tool to simulate 1) lead crack growth. From these results one can conclude the following: 1) The applied axial force used to model the longitudinal stress caused by a closed cylinder has little in uence on the computed stress intensity factors..23 that is. The directionalcriterionused in Potyondyet al. IV.21 and Chen et al. Figure 25 shows the predictedcrack-growth trajectories from previous and current studies as well as experimental measurements. Eq. Comparisons with experimental measurements suggest that the fracture toughness orthotropy plays an important role in predicting the early crack-growth trajectories.33 respectively. 25 Comparisons between predicted and measured crack trajectories (isotropic directional criterion with various magnitudes of rc ): ¤. F.21 respectively. Both studies analyzed early curvilinear crack growth but did not address the issue of sharp turning as the crack approaches the tear strap.33 respectively. 26 and 27. 3) multiple crack interaction. 33 at this stage basically represent the same boundary conditions and crack con guration except an axial force was applied in the current model. because they remain more or less the same for all cases. Figures 26 and 27 show the computed stress intensity factors at the right crack tip in comparison with Potyondy et al.. the current model and the model used in Ref. 2) The fact that the kinematic boundary conditions were not altered during crack growth in the previous studies has a mild affect on the crack trajectory prediction and stress intensity factor computation.21 and Chen et al. 2) MSD crack growth. (127-mm) crack. a 6 £ 6 bay-stiffened panel model. curvilinear crack growth in a generic narrow-body fuselage is studied. the longitudinal stress in this numerical model is expected to be less than that in the test xture because the test xture is a closed cylinder.45 in. and 4) material thinning as a result of corrosion damage in pressurized fuselages.21 and Chen et al. the boundary conditions applied to the global shell model corresponded with an open cylinder. measured. Acknowledgments This work was performed with support from the NASA Langley Aircraft Structural Integrity Program under Contracts NAG1-1184 and NAG-1-2069 and U. and §. As a result.21 and Chen et al. It is. of the FRANC3D program. Crack growth was only performed in the 2 £ 2 bay model.0-in. (11. thus. predicted (Chen et al. This is properly re ected on the computed K I values shown both in Figs. The active plastic zone and accumulated plastic wake caused by stable tearing would likely affect the crack-growth prediction. composed of a global shell model. the lowest level in the hierarchy. The initial crack location in Potyondy’s simulations was modeled at 0. and a 2 £ 2 bay-stiffened panel model.2 crack turning and apping as the crack approachesthe tear strap is highly related to the T-stress. comparisons are made with the isotropic prediction with rc D 0. Air Force Corrosion Damage .CHEN. In addition. This leads to a lower ratio of K II =K I in the current model with updated boundary conditions.. thus. Comparisons with Previous Studies Potyondy et al. This seems to have little effect on the computed values of K II . nevertheless. Also.S. WAWRZYNEK. Fig.21 ).33 corresponds to the Erdogan and Sih directional criterion. 26 Computed stress intensity factors vs half-crack extension. M. Conclusions The CTOA fracture criterion and elastic-plastic nite element analysis are used to predict fracture behavior and residual strength of KC-135 fuselage panels. 27 Computed stress intensity factors vs half-crack extension. The hollow and solid markers denote the computed stress intensity factors from the current isotropic prediction with rc = 0 and those from Potyondy et al.

P.. 1191– 1205. 4. Vol. 1997. Jr. pp. Bigelow. 30. “On the Crack Extension of Plates Under Plane Loading and Transverse Shear. O. and Instability. 1992. J. Dissertation. 6 Cope. edited by J. “Discrete Crack Growth Analysis Methodology for Through Cracks in Pressurized Fuselage Structures..” International Committee on Aeronautical Fatigue. Wawrzynek. K. edited by A. 19. “Numerical Assessment of T-Stress Computation Using a P-Version Finite Element Method. 39– 52. J. A. S. 1611– 1633. Dissertation. 21 Potyondy. “Elastic-Plastic Crack Growth Simulation and Residual Strength Prediction of Thin Plates with Single and Multiple Cracks. and Su. Vol. M. J. 1923– 1934. 85. TR D500-13008-1.4. 815– 829. ICF9. 1991.. 1999.. Banks-Sills.. M. American Society for Testing and Materials. 26 Williams. R. L. 16 Dawicke. Philadelphia. Springeld. 1987.D. and Ingraffea. Vol. O. 6.” Fracture Mechanics.. Vol. 33 Chen. Newman.. 5 Rankin.” Damage Tolerance of Metallic Structures: Analysis Methods and Application. J. “Prediction of Crack Growth Direction for Mode I/II Loading Using Small-Scale Yielding and Void Initiation/Growth Concepts.” International Journal of Fracture. Harris..” Boeing Defense and Space Group. P. A.. Wawrzynek. A. pp. 527– 532. Philadelphia.. Boone. VA. Philadelphia. T. C. Sutton.. H.” Ph. School of Civil and Environmental Engineering. 1998. KS. “Corrosion Damage Assessment Framework: Corrosion/ Fatigue Effects on Structural Integrity. “A Normal Stress Criterion for Crack Extension Direction in Orthotropic Composite Materials. 2001... C.” International Journal of Fracture. A. edited by R. 68. NASA CP 3160. 19. F..1652 CHEN. R.. L. “An Elastic-Plastic Finite Element Analysis of Crack Initiation. M.” Ph. B. 23 Erdogan.. and Worden. S. 8 Maclin. David S.. pp. J. C. M. 31 Sutton. A. 83.. “Computation of Membrane and Bending Stress Intensity Factors for Thin. L.. “Fracture Under Complex Stress— The Angled Crack Problem. 21– 38. “Finite-Element Analysis of Pressure Vessel Panels. 4. and Newman. C. F.K. WAWRZYNEK.D.-S.” Lockheed Martin Missiles and Space Co. 1997. 1995.. N.” NASA CR-2000-209342. Newman Jr.. 29. 1999. R. W. 14 Dawicke. G.-S. C.. D. 1984. Delft Univ.. 72. “A Note on the Angled Crack Problem and the Directional Stability of Cracks. 223– 242. Spring eld. Wang.. Jan. pp.. Palo Alto. No..” InternationalJournal of Fracture. B. pp. D. D. edited by T. D. CA. NY. 23.. 1995. 1990. Wichita. Chen.” Proceedings of the Federal Aviation Administration– NASA Symposium on the Continued Airworthiness of Aircraft Structures. 30 Viz. No. 69 – 107.” Journal of Composite Materials. 17 Dawicke. Vol.. pp. Chang and J. 4 Carter. J. U.. R. edited by F. Jr. 67 – 74. 2001.. American Society for Testing and Materials. C.. Philadelphia. S. X. pp.-S.” Fracture Mechanics: Fifteenth Symposium..” Proceedings of the Ninth International Congress on Fracture. S. Australia.Helpful discussionswith the members of the Cornell Fracture Group are also appreciated.” Fatigue and Fracture Mechanics. England. and Sih. NASA Symposium on the Continued Airworthiness of Aircraft Structures. 1995. TR LR-681. Karihaloo et al. Vol. J. A. M. American Society for Testing and Materials.. A. R. Sept. 22 Zaal. C. The Netherlands. “Performance of Fuselage Pressure Structure.. 544– 553. and Riks. pp.. M..” 1991 International Conference on Aging Aircraft and Structural Airworthiness.. pp. Elsevier Science.” International Journal of Fracture.. D. 177– 199. pp. 4. Sutton.. A. N. Cornell Univ. pp. of Technology. R. T. G...” Computational Mechanics. 9 Miller. edited by B.. edited by S. A. The authors would like to thank James C.-S. 1995. “Methodology for Fatigue Crack Growth and Residual Strength Prediction with Applications to Aircraft Fuselages. 1. Vol. 441– 446. D. 9. pp. Jr. Jr. J. Sanford. 3. Zehnder. VA. edited by C. 13 Newman. A. 45. 358– 379. and Ingraffea. edited by T. D.. R. “Validated Feasibility Study of Integrally Stiffened Metallic Fuselage Panels for Reducing Manufacturing Costs. S. A. Rudd. Ithaca. “In uence of MSD Crack Pattern on the Residual Strength of Flat Stiffened Sheets. C. Reynolds. and Saith. and Ingraffea.” Fracture Mechanics. R.... P.” Proceedings of the Federal Aviation Administration– 1 Harris. A. “The Effect of Crack Interaction on Ductile Fracture.. West Midlands. C. 2 Chen. pp.. and Hui.. “Finite Element Modeling of Fracture Propagation in Orthotropic Materials. R. Wawrzynek. C... No.. No. Engineering Materials Advisory Series. Erdogan. 484– 486. 20 Nilsson.. A... Vol. 28 Buczek.. N. T.-S. Bigelow. “Measurement and Analysis of Critical CTOA for an Aluminum Alloy Sheet. E. J.. Potyondy.. and Bigelow. D.. Jr.” Engineering Fracture Mechanics.” International Journal of Fracture. Brogan. 18 Sun. References C.. Philadelphia. “A Survey of Crack Path Stability Criteria and Their Application to Crack Flapping Phenomena in Stiffened Structures. pp. 1972. 6. L. Springeld... Vol. 93– 117.. C. A. R. T.” International Journal for Numerical Methods in Engineering. Newman. Vol.. 8. 1997. Cornell Univ. American Society for Testing and Materials. 2. and Bigelow. and McMeeking.. No. and Ingraffea. Atlanta.. Dawicke.” ComputationalMechanics. “ThreeDimensional CTOA and Constraint Effects During Stable Tearing in a ThinSheet Material. “A Methodology for Simulation of Curvilinear Crack Growth in Pressurized Shells. 1996. Vol. M. 26. J. E. Chen. 7 Newman. Kaelber. Rankin.. No. 25. 10. Philadelphia. and Newman. pp. NY. pp. No.. and Cabiness. Vol. School of Civil and Environmental Engineering. 1985. No. 185– 201.. T. “Process Zone Size Effects on Naturally Curving Cracks. L. C. A.. G. 519– 527. pp.. L. W. edited by C...-F.” Fatigue and Fracture Mechanics. G. No. C. P. C. “Residual StrengthPredictionsUsing a Crack Tip Opening Angle Criterion. 555– 566.. Reuter et al. 38. 1993. J. pp.-Y. 2. Vol. 29. P. F. D. “Fracture Analysis of Stiffened Structure. pp. 10 Pettit. 2.. D. C. and Dale Cope for their supportand encouragement. Loden. 27 Finnie.. R.. D. 1986. 161– 170. Ingraffea. A.. 10. K. Panontin and S. Atluri..” International Journal of Fracture. A... pp. Jr.. 63 – 88.. 1996. Vol. Starnes. M. pp. and Ingraffea... 17th Symposium. Vol. M. K. “A Topology-Based System for Modeling 3D Crack Growth in Solid and Shell Structures. C. G. and Ewing. Jr. A. 1999. Ithaca. 1973. No. No. P.” Proceedings of the Federal Aviation Administration– NASA Symposium on the Continued Airworthiness of Aircraft Structures. 32 Kfouri. A. C. Palazotto Associate Editor . Stable Crack Growth. Vol. O.” Proceedings of the International Workshop on Structural Integrity of Aging Airplanes. and Herakovich. Sydney. Wawrzynek. B. P. 1996. National Technical Information Service. VA. J. Inc.-S.” Journal of Basic Engineering. C. A. A. “Some Evaluations of the Elastic T-term Using Eshelby’s Method. Cracked Plate. A. A.. 275– 290. pp. Wawrzynek. P. 29 Boone. Zhao. “STAGES User Manual Version 2. AND INGRAFFEA Assessment Framework under Subcontract NCI-USAF-9128-001. Bigelow.. Sheppard. “A Fracture Criterion for Widespread Cracking in Thin-Sheet Aluminum Alloys. 25 Chen. J. Advanced Technology Center.. and Toh. A. edited by C. A. 443– 468. “Analytical Methodology for Predicting the Onset of Predicting Widespread Fatigue Damage in Fuselage Structure. L.. Blom. Aug. 107. pp. Panontin and S. National Technical Information Service.. Vol..” Engineering Fracture Mechanics. pp. I. A.” Faculty of Aerospace Engineering. J.. and Dawicke.. pp. G. 97– 113. Sheppard. Atlanta Technology Publications. 3 Potyondy.. Pettit.. 1993. 24 Pettit. pp. 15 Dawicke. 1997.American Society for Testing and Materials. 337– 348. No. 26. C. Dawicke. C. R. and Bigelow. D. C. 1963. 1984. E. D. 19 Chen. 103– 122. May 2000. edited by W. D. pp. National Technical Information Service. R.. edited by C. 1992. Krause.” International Journal of Fracture. and Ingraffea. “Crack Growth Simulationand Residual StrengthPrediction in Thin Shell Structures. 1999. “Large Crack Tip Opening in Thin Elastic-Plastic Sheets. J. 301– 315. E. 11 Swift. R. Delft.. 12 Hom. A. “Residual Strength Predictions for Multiple Site Damage Cracking Using a Three-Dimensional Analysis and a CTOA Criterion. Vol. C. S. American Society for Testing and Materials. C.. A.

Sign up to vote on this title
UsefulNot useful