~ Pergamon INTEGRITY WITH Eng&eering Fracture Mechanics Vol. 51, No. 3, pp. 361-380, 1995 Elsevier Science Ltd. Printed in Great Britain. 0013-7944(94)00252-5 OF AIRCRAFT MULTI-SITE STRUCTURAL FATIGUE ELEMENTS DAMAGE? JAI H. PARK,,+ RIPUDAMAN SINGH,§ CHANG R. PYO¶ and SATYA N. ATLURIH FAA Center of Excellence for Computational Modeling of Aircraft Structures, Georgia Institute of Technology, Atlanta, GA 30332-0356, U.S.A. Abstract--Integrity evaluation of aging structures is extremely important to ensure economic and safe operation of the flight vehicle. A two step analytical approach has been developed to estimate the residual strength of pressurized fuselage stiffened shell panels with multi-bay fatigue cracking. A Global Finite Element Analysis is first carried out to obtain the load flow pattern through the damaged panel. This is followed by the Schwartz-Neumann Alternating Method for local analysis to obtain crack tip stresses and the relevant crack tip parameters that govern the onset of fracture. Static residual strength is evaluated using fracture mechanics based, as well as net section yield based, criteria. The presence of holes, with or without Multi-Site Damage (MSD), ahead of a dominant crack is found to significantly degrade the capacity of the fuselage structure to sustain static internal pressure. An Elastic-Plastic Alternating Method has been developed and applied, to evaluate the residual strength of fiat panels with multiple cracks. The computational methodologies presented herein are marked improvements to the present state-of-the-art, and are extremely efficient, both from engineering man-power, as well as computational costs, points of view. 1. INTRODUCTION A LARGEpercentage of in-service aircraft has exceeded its initial design life and is prone to wide spread fatigue damage. Residual strength estimations of critical aircraft components with detectable cracks are imperative for the safe operation of these aging airplanes. This paper presents a part of the major effort being made to develop reliable computational tools for structural integrity evaluation of aging, as well as new, airplanes. These computational tools are envisioned to involve the least amounts of engineering man-power, as well as computer, resources. Micro-cracks emanate at material/structure imperfections and discontinuities. These cracks grow under static and fatigue loading, and coalesce to form detectable cracks. These detectable cracks can linkup and become catastrophic, at times, within two inspection schedules. In order to ensure safety, it is important to understand the severity of cracks and also, have an estimate of collapse strength/residual strength of the damaged structure. This paper primarily deals with the investigation of longitudinal cracks in stiffened fuselage shell panels, even though the proposed methodologies apply equally well to a variety of wide spread fatigue damage situations at a variety of locations in the pressurized fuselages of commercial planes. In an airliner fuselage, pressurization causes stresses in the shell structure. The stiffeners (stringers, frames and tear straps) take a part of the load, but the major fraction is taken by the skin. Large cracks in the skin cause significant redistribution of load flow, which becomes fairly complex due to various stiffening elements and the presence of holes/MSD ahead of a dominant crack. Thus, the problem essentially calls for a determination of the load flow pattern through the panel, before the crack tip stresses and other relevant crack tip parameters e.g. the stress intensity factor (SIF), T*, etc. that govern the onset of fracture, can be determined. The finite element alternating method has been found to provide good quantitative estimates of SIFs at various crack tips and a qualitative insight into the mutual interactions of small, long and even multi-bay cracks in the shell panel or at a row of fastener holes. This analysis adds to the understanding of the residual strength of the damaged structure. Global finite element analysis tPaper presented at the 1CAF-11, Stockholm, June 1993. SVisiting Scholar from Chungbuk National University, Cheongju, Korea. §Postdoctoral Fellow, member AIAA. ¶Postdoctoral Fellow. IlInstitute Professor and Reagents" Professor of Engineering, Fellow AIAA. 361 362 J.H. PARK et al. of the multi-bay shell panel with cracks, generates information about the realistic load flow pattern through the panel. The cracked portion of the shell is then isolated with the corresponding sheet stresses and the fastener loads (if any). At this stage, for local analysis purposes, the uncracked skin segment subjected to tractions on its boundaries (as determined from global analysis) and any fastener loads, is modeled and a finite element analysis is performed to obtain the stress field at the location of the cracks, which are then erased through the alternating method, using analytical solutions for an infinite body with pressurized cracks, to determine very accurately the crack tip stresses and the crack tip fracture parameters [1]. This sequential combination of two step FE Methods forms an efficient computational tool for the estimation of crack tip parameters and the corresponding static residual strength. For materials with significant ductility, an Elastic-Plastic Alternating Method has been developed by Nikishkov and Atluri [2], which generates a very accurate elastic-plastic stress state solution near the crack tips by using only the finite element model of the uncracked sheet. This technique has been applied to study the Multi-Site Damage (MSD) in A1 2024-T3 skin in the present paper. Both the techniques [1, 2] use a single finite element mesh to study a variety of cracks, of varying lengths, in the skin. This makes the present analysis extremely efficient. 2. PROBLEM DEFINITION The fuselage shell panels under consideration are typical of commercial airliners. They are stiffened along the longitudinal direction by stringers and circumferentially by frames and tear straps. Tear straps are generally at frame locations, but can also be present at mid-frame stations depending upon the design philosophies. A typical configuration is shown in Fig. 1. Two shell geometries, one of each kind, have been considered. The geometrical details are presented below. Shell radius R Shell skin thickness t Distance between frames Distance between stringers Distance between tear straps Width of T-straps Thickness of T-straps t t Frame area Frame moment of inertia Frame neutral axis offset Stringer area Stringer moment of inertia Stringer neutral axis offset Fuselage internal pressure Rivet diameter D Pitch of rivets Material: skin, frame, stringers Material: tear straps Shell 1: Shell 2: 118.5 in. 0.071 in. 20.0 in. 8.0 in. 20.0 in. 3.0 in. 0.025 in. 0.7171 infl 1.4320 in. 4 3.25 in. 0.6721 in] 0.1020 in. 4 0.68 in. 8.6 psi 0.1875 in. 1.25 in. AI 2024-T3 Ti 8-1-1 74.0 in. 0.036 in 20.0 in. 9.25 in. 10.0 in. 2.0 in. 0.036 in. 0.160 in. 2 0.120 in. 4 3.15 in. 0.186 infl 0.040 in. 4 0.78 in. 9.0 psi 0.15625 in. 1.0 in. A1 2024-T3 A1 2024-T3. The lap joint configuration considered is as follows: Length of overlap No. of rivet rows Pitch of rivets No. of rivets in each bay Rivet diameter D Adhesive layer thickness ta Material 3.0 in. 3 1.0 in. 20x3 0. ! 5625 in. 0.00025 in. AI 2024-T3. Integrity of aircraft structural elements (a) 363 Frames Skin Lap.lomt ~ ~'"" ' Shell panel Stringers Straps (b) Cracked Skin S e g m e n t Fig. 1. Fuselageshell panel configuration.(a) Shell panel. (b) Cracked skin segment. The material properties of A1 2024-T3 and Ti 8-1-1 are taken as follows: Young's modulus E Shear modulus G Poisson's ratio v Yield strength ay Ultimate tensile strength au 1 Crack tip linkup stress 5(ay + au) % Elongation Fracture toughness Klc Adhesive shear modulus is Ga = A1 2024-T3 Ti 8-1-1 10.5 x 10 3 k s i 4.2 × 10 3 k s i 0.32 47.0 ksi 64.0 ksi 55.5 ksi 18% 93.0 ksi ,,/~. 17.5 × 10 3 ksi 6.7 × 10 3 ksi 0.32 135.0 ksi 145.0 ksi 140.0 ksi 8% 1.09 x 105 psi. Four situations of cracks are considered in this article: (1) (2) (3) (4) a single dominant multi-bay crack in shell 1, a single dominant multi-bay crack at a row of fastener holes in shell 1, long cracks and MSD at lap splice in shell 2, and a single dominant multi-bay crack at lap splice in shell 1. In addition, a full elastic-plastic analysis of a flat panel with multiple cracks is performed, the configurational details of which are given separately in the corresponding section. 3. ANALYTICAL APPROACH The method of analysis adopted, is to first evaluate the load flow through the damaged panel and then perform a refined analysis to obtain stress fields at the various crack tips. A global-local finite element analysis procedure has been developed for this purpose. 3.1. Global analysis (FEM) Conventional linear elastic finite element analysis of the multi-bay stiffened shell panel with cracks, is performed as a part of the global analysis• The FEM model is briefly described below: The fuselage skin is modeled by four-noded shell elements, with five degrees of freedom per node. The element used is strain-based and was developed by Ashwell and Sabir [3]. Tear straps are also modeled using the same element• The frames and stringers are modeled as two-noded, three degrees of freedom per node, curved/straight beam elements with their shape functions degenerated from those of shell elements. This is done to ensure compatibility within the stiffeners and the sheet. The cracks are incorporated into the problem as unconnected nodes belonging to respective elements. For the purpose of present global analysis, the crack tip singularity is not modeled as 364 J.H. PARK et al. the crack will be modeled analytically in the second step, i.e. the local analysis. The fasteners are modeled as two degrees of freedom connectors between the corresponding nodes and the fastener stiffness is represented by the empirical relation developed by Swift [4]: ED with A = 5.0 and C --- 0.8 for A1 rivets. Wherever there is a crack, the stiffness along the crack length is relieved as the fastener will not be able to bear the load in that direction. Adhesive is also modeled as a two-noded, two degrees of freedom per node connector between the sheets. The adhesive stiffness is modeled as area ga = fld ta 1- 3 / 5 t2 ) , (2) aa'8ka+ where Pd is the degradation factor. The values of 0 and 1 represent total degradation and perfect conditions, respectively. A value of 0.1 means 90% degradation. "area" Represents the bond area being lumped at the nodal connections. Appropriate multi-point constraints have been imposed to prevent criss-crossing of sheet nodes in the lap joint zone. The fuselage internal pressure is applied as a uniformly distributed normal outward load on the shell panel. The four edges of the panel are permitted to undergo only radial displacement in a cylindrical system. A typical problem size for the configurations considered is of the order of 15 000 degrees of freedom and the computer time is of the order of min on an HP 9000/700 series workstation. 3.2. Elastic local analysis [ E - F E A M ] From the global analysis, the skin segment containing the cracks, holes and fasteners of interest, is isolated with corresponding sheet stresses. The fastener holes are now modeled as circular and the bearing loads (if any, from global analysis) are distributed as sinusoidal variations over the periphery. The stresses due to the misfit of the rivet can also be accounted for at this stage. This problem is solved using the Schwartz-Neumann Finite Element Alternating Method which involves two solutions. (1) An analytical solution to the problem of a row of cracks of arbitrary lengths with crack faces being subjected to arbitrary tractions, (2) Finite element solution for a strip, with/without a row of holes, but without cracks; the strip being subjected to sheet stresses and pin bearing loads. Since the finite element solution is only for the uncracked body and the cracks of arbitrary lengths are accounted for analytically in step 1 above, the computational finite element mesh remains the same as the cracks grow. This technique and the analytical solution to the problem of multiple cracks in an infinite sheet are presented in detail in an earlier work by Park et al. [5]. Eight-noded isoparametric elements with two degrees of freedom per node are employed in this finite element analysis. The crack tip stress intensity factors and the stress field are obtained directly from the FEAM analysis. The net section stress for any ligament is obtained by taking an average over the ligament length. To compute the plastic zone size, Irwin's formula does not seem to give a reasonable approximation, probably due to the complexity of geometry and vicinity of other cracks/loadedholes. So the plastic zone size is estimated from the computed stress field by doubling the distance from crack tip to the point where the stress falls to yield stress. Critical pressure for the fuselage is that value of applied pressure differential for which either the crack tip SIF becomes equal to K~c of 93.0 ksi x / ~ . , or the net ligament stress equals the linkup stress of 55.5 ksi. For linear elastic analysis, this can be computed directly from the obtained values of K~ and average ligament stress tray. KIc Critical pressure differential = applied pressure x - Ki Integrity of aircraft structural elements Critical pressure differential = applied pressure x 365 linkup stress O'av 3.3. Elastic-plastic local analysis [EP-FEAM] Elastic-plastic analyses for fiat panels with multiple cracks are conducted using the ElasticPlastic Finite-Element Alternating Method in conjunction with the initial stress method as recently developed by Nikishkov and Atluri [2]. This is briefly described below: The initial stress method [6] for the solution of elastic-plastic problems is an iterative scheme, in which the elastic-plastic displacement and stress fields are sought as the summation of the respective elastic solutions due to artificial volume loads. Since the initial stress method requires only elastic solutions in the calculation procedure for an elastic-plastic problem, it is possible to develop an elastic-plastic alternating method using the usual alternating method, which provides an elastic solution at each iteration. Then, the algorithm for the elastic-plastic alternating method based on stress is as follows: Initial conditions: { q~)} = = [b °q = Iterative procedure: {Aq~~} = { q~)} = {a~ )} = {t~i) } = Approximation {t~i) } = Analytical solution {a~i)} = {0} {e } [0] [Ko]-'{~'"-~} {q~-')} + {Aq(oi) } [Di[B]{ q~)} - [n¢]{o'~)} - [b(i)]{ U} {a~([b")])} {if(i)} .~. [DCP][D] l({o.~) } + y,~¢i)~t_c ,J {~b¢')} = {P} - S,,[B]V{a 'i,} dV Convergence criterion: IIAq~) II < ~ IIq HHere, indices " o " and "c" are used to denote quantities relating to an uncracked finite element model and to the analytical solution for the infinite plate with the crack, respectively; { q } and {Aq } are the displacement vector and its increment; {~, } is the finite element residual vector; {P } is the current level of applied load; [K] is the global stiffness matrix; {a} is the current stress level; [B] is the displacement derivative matrix; [D] is the elasticity matrix; [D e°] is the elastic-plastic constitutive matrix; {to} is the traction vector on the crack surface; [b] is the complex coefficient of the approximation; { U} are Chebyshev polynomials and V is a volume of the finite element discrete model. This single step algorithm can be divided into one with several load steps. In this case, the displacement vector { q} should be treated as an increment for the current load step. In the present method, stresses {ac} in an infinite cracked body are computed directly from an analytical solution at finite element integration points and are used to calculate the incremental stresses. For computing the elastic-plastic stress increment with sufficient accuracy, it is necessary to divide the strain increment into small sub-increments and apply a stress correction at the end of each sub-increment. The details of the algorithm of the elastic-plastic alternating technique are presented in Nikishkov and Atluri [2]. The elastic-plastic algorithm for multiple cracks is similar, except for the part of calculating Chebyshev polynomials. In the present case, Chebyshev polynomials for a crack must be computed by also accounting for the stress effects of other cracks. The complete analytical procedure for multiple cracks is schematically shown in Fig. 2. In the analytical model, the material is presumed to obey the plastic flow theory with a Mises yield surface. The deformation curve is taken as a = Cry+ ke P, where tr is the uniaxial stress, k is the hardening coefficient given as (au - O'y)/(0.18 - try~E) and eP is the uniaxial plastic strain. 366 J.H. PARK et al. lastic-Plastic Alternating Technique lot Multiple Cracks [ Ptotal 4444 Rcvcrsc Stresses at Crack Locations. / /" / .f / / eresidual i / j~ / J J [ l Ate Resldual Load Vectors Zero ? [ ~ EvaluateElastic Stress Distributions by Using Chebyshev Polynomials. Yes p;a Elastic-Plastic Stresses by Using Initial Stress Method. Fig. 2. Schematicof the elastic-plastic alternating method for multiplecracks. 4. SINGLE DOMINANT MULTI-BAY CRACK IN SKIN Consider a panel of shell 1, consisting of seven frames (six bays) and six stringers (five bays). There are seven tear straps, one at each frame location. Consider a single dominant crack aligned longitudinally and located centrally, i.e. halfway between two stringers and symmetrical about a frame/strap location. Swift [7] has clearly defined the design requirements for the fuselage from the damage tolerance point of view. The fuselage stiffened shell must be able to sustain a two-bay-long crack with broken central strap. For this requirement, consider the central tear strap also to be cracked along with the skin. All other stiffening elements are intact. Analyses have been carried out for various crack lengths up to 50 in., i.e. the crack spreading in four bays. Since there is only one crack and there are no holes in the vicinity of the crack tip, it is difficult to reasonably define an intact ligament length, thus the critical pressure is computed only from the fracture mechanics point of view (Fig. 3). Integrity of aircraft structural elements Frame --] T strap 367 I I T strap Frame I I 35.0 \ "5- a / "E 0.0 0.0 I I I l 10.0 I I i I 8 6 psi 1½ Half Crack Length / " b 0jI • 2410 - I Local Analyses Region Fig. 3. Critical pressure diagram for a single dominant multi-bay crack in the shell panel. A point on this curve gives information about the cabin pressure required to make the crack tip SIF critical and does not imply that the panel will fail catastrophically. The curve falls below the applied pressure of 8.6 psi for crack lengths of 10.8 and 18.2 in. (points " a " and "b"). The implication of this is as follows. Considering an initial half crack length of, say, 2 in. and the pressure never exceeding 8.6 psi, the crack cannot grow under static loading, it will grow under cyclic loading till it reaches point "a". At this stage, fast fracture will occur and the crack will instantaneously grow up to point "b", where it will be arrested as the SIF falls below the critical value. Further growth of the crack will be again under cyclic loading. The effect of plasticity is not included in this calculation, and stable tearing that follows crack growth initiation is not modeled. 5. TWO-BAY CRACK W I T H H O L E S NEAR T H E CRACK TIP Consider a panel of shell l, consisting of seven frames (six bays) and seven stringers (six bays). There are seven tear straps, one at each frame location. Consider a single dominant crack aligned longitudinally along the fastener holes. The crack is located centrally and is symmetrical about a frame/strap location. The central tear strap is broken. All other stiffening elements are intact. The presence of holes alters the stress flow. The distance between the crack tip and the hole periphery is found to have substantial effect on the crack tip SIF and also, for very small ligaments, the section will yield before the crack becomes unstable. A situation of holes ahead of a two-bay crack is analysed to study the effects of holes on the critical pressure. Figure 4 shows the effect of holes ahead of a two-bay crack. For the sake of convenience we will use a term "overhang", defined as the extent of crack length from the closest hole center involved in the same crack. Zero overhang implies a hole at the crack tip and so the SIF has no definition. For extremely small overhang, the SIF value is low and for longer overhangs, the SIF rises sharply. The ligament stress is found to increase steadily with overhang. For any crack length and overhang, the critical pressure is the lower of the two values corresponding to K~ = K~c and a~v = linkup stress. Figure 4 shows the critical pressure from both considerations and interestingly, the structure is critical from the net section failure point of view over most of the region. Again, any point on this curve represents the pressure differential required to either make the crack unstable, or cause tensile failure of the ligament. It is seen from this analysis that the presence of holes (even without cracks emanating from them) ahead of a lead crack, significantly lowers the residual strength of a stiffened panel; as surmized by Swift [8] and earlier in ref. [7]. EFM 51/3~ 368 J.H. PARK et al. Frame -w--w- Critical y " t ' - - ~ strap --"1 ~ [ Net Section Yield With Holes 35.0 I o.o-r ~ I i I I 2~0 | | I 24.0 Half Crack Length [] - Local Analyses Region Fig. 4. Critical pressure diagram for a single dominant two-bay crack with holes ahead of the crack tip. 6. L O N G CRACK AND MSD IN LAP S P L I C E We now consider the problem of a riveted/bonded longitudinal lap joint in the shell panel. Skin stresses have to flow through this joint. So long as the adhesive is in good condition, it transfers most of the load across the lap, but the adhesive is found to degrade fairly fast with usage and time, and make fasteners the major load carriers. The presence of cracks at rivet holes causes further load redistribution. An important aspect of the situation is how one crack effects other cracks in its close vicinity. Consider the fuselage panel 2 with a longitudinal riveted/bonded lap joint. Let there be cracks at the outer (critical) row of fastener holes. The adhesive is treated to be degraded, so that fasteners share a major part of the hoop load. The fuselage panel considered for analysis, consists of three bays in the longitudinal direction and six bays along the circumferential direction. This panel contains four frames, seven tear straps and seven stringers. Tear straps exist at all frame and mid-frame stations. The cracks are only considered in the central bay. Figure 5 shows 12 c a s e s [B-M] of crack configurations that have been analysed in this section. Case A is the uncracked situation and forms a reference. These configurations have no relationship with each other and the long crack configurations are not necessarily arrived at through linkup of small crack situations. In all these cases, the stiffening elements (frames, T-straps and stringers) are all intact. 6.1. Global analysis The untracked panel is analysed as the first case, to obtain a reference and understand the original load flow path. After the adhesive has virtually degraded (99% for the present problem) the rivets transfer a large part of the hoop load. In this situation 83% goes through the fasteners and only 17% of the load is taken by the frames/straps. Among the three rows of fasteners, the outer two carry ca 36% each and the rest of the 28% is taken by the central row. In a flat panel lap joint, normally, the center row carries a much lesser portion. The curvature seems to be a cause of more even row-wise load distribution. The maximum sheet hoop stress is found to be 13.4 ksi. The presence of cracks at rivet holes causes load redistribution at the joint. For various cases considered [A-M] the load changes in straps/frames in the damaged bay are given in Table i. It can be observed from this table that MSD, or long cracks, [B-I] do not significantly overload the frames and straps. Even for case I, where one may say that the T-strap load has gone up from Integrity of aircraft structural elements i i A 369 0110 0 0 0 0 0 0 O 0 O O O O O O O O OIio I I I OIO t I I w B I I I o ,o % ~ %~=,~%o o O O O O O O O O O 0 0 0 0 0 0 0 0 0 OlO I I I 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 010 I I I 010 I I I I I I C 0,o % ~ % ~ % ~ o D o Io o i ~ %~o-~s o E ! I I 010 I I I 0 0 I0 F 01o o-o-~ ~>-o-~ o I ( i 0 0 -0---0"--~. ! I I G 0,0 0-%0--o-%° o H o 1o - o - - o - o - ~ 010 O O O O O O O O O OIio 1 ! I 0 010 I I I ol, o 0 0 0 0 0 0 0 0 0 0 0 0 010 I I I 0',0 I I ! o o 0 I I I I o o -.o~o--o--o--o-o J o o oa~ %~a~O o o 0 ~*~:~0~'=0~¢ 0 Ii 0 t I ! K L M o o ~ o ° o [ ! I I ~ ^ ~ I oto ~ Ic I I 1 I I .o--o--.o -o--o b b o I I I o I o I i I l l I I I I I I ~ O i I i O I I ° ! ! I IO T strap 4 T strap 3 j t t o o,o e o I oto T stra 5 Frame 3 Frame 2 = 0.25" = 0.3" = C.s" Fig. 5. Crack configurations analysed to study the long crack and M S D interactions at lap splice. 24.5 to 84.4 lb, numerically both these values are insignificant as compared to the yield strength of 3384 lb (2 x 0.036 × 47000 = 3384) for the strap. Some amount of overloading for the central T-strap can be seen in cases J and K, where cracks exist across it. From cases K - M , it is evident that if two cracks approach a tear strap from both sides and linkup at a location where there is no frame, the strap will yield. Table 1. Load redistribution at frames/T-straps for various cases (lb) Case A B C D E F G H I J K L M Frame 2 T-strap 3 T-strap 4 T-strap 5 Frame 3 1127 1142 1142 1146 1142 1168 1153 1232 1231 1104 1284 1246 1535 24.5 34.6 34.5 34.9 32.5 47.5 40.2 86.6 84.4 -9.4 100 19.9 230 228 211 218 248 248 250 221 240 252 1510 489 4163 4558 24.5 24.1 23.9 21.8 22.0 20.4 22.6 16.6 32.3 -4.5 90.0 19.9 230 127 127 127 126 126 126 127 125 151 121 1271 1246 1535 In cases L and M, tear strap has yielded, load > 3384 lb. 370 J.H. PARK et al. Table 2. Row-wise rivet load distribution (lb) Row 1 Row 2 Row 3 Case Load % Load % Load % Total load A B C D E F G H I J K L M 3394 3305 3278 3280 3302 3267 3361 3223 3201 2778 2937 1669 1101 36.1 35.2 34.9 35.1 35.3 35.2 35.9 35.2 35.1 34.1 34.7 34.6 34.5 2600 2682 2696 2673 2661 2654 2627 2615 2633 2421 2485 1442 952 27.7 28.6 28.7 28.6 28.5 28.6 28.1 28.6 28.9 29.7 29.4 29.9 29.8 3405 3395 3402 3385 3382 3366 3358 3315 3272 2956 3035 1718 1141 36.2 36.2 36.3 36.3 36.2 36.2 35.9 36.2 35.9 36.2 35.9 35.6 35.7 9398 9383 9377 9339 9347 9288 9347 9154 9107 8156 8458 4831 3195 F o r all the cases, the row-wise rivet l o a d d i s t r i b u t i o n within the d a m a g e d panel is given in T ab l e 2. It can be o b s e r v e d that as the crack length increases an d the load is shed f r o m the rivets, the relative share o f the three rows changes marginally. Thus, as the cr ack ed r o w sheds the load, the o t h e r two rows also do so a n d effectively, the load goes into the skin an d the frames/straps. Physically, this a p p e a r s correct as the rivet rows are 1 in. a p a r t and with a crack o f a few inches in the first row, the second and third rows will hardly get any load to share. T h e rivet-wise load redistribution p a t t e r n for the critical r o w for all the cases [ A - M ] is given in Tab l e 3. T o have an overall q u a li ta ti v e idea o f the load redistribution, the loads are also p l o t t ed to scale in Fig. 6. R iv e t n u m b e r s 1-10 run f r o m left to right in the d a m a g e d bay. A crack diverts the l o a d t o w a r d s its tips. So, intuitively a crack on one side o f the hole should o v e r l o a d the fastener, an d cracks on b o t h sides will u n l o a d it. Rivets 5 an d 6 o f cases A an d B clarify this point. O n a similar logic, the stresses in the l i g a m e n t between t w o crack tips will be high an d further increase with crack growth. A t s o m e stage o f loading, the ligament will yield, causing a crack linkup to f o r m a single long crack. C o m p a r i n g the rivet load patterns for cases H an d I, we can see that even long cracks across the tear strap do n o t affect each other. T h e rivets i n v o l v e d in long cracks take very little load, thus significantly o v e r l o a d i n g the n e i g h b o r i n g rivets (case H). In these cases, the n e i g h b o r i n g rivets are also f o u n d to bear some side l o a d as a c o n s e q u e n c e o f inclined load flow p a t h t h r o u g h the skin, at the r o w o f rivets. T h e side load on rivet 8 o f case I is o f the o r d e r o f 55 lb t o w a r d s rivet 7, an d Table 3. Rivet load distribution in the critical row (lb) Case Rivet A B C D E F G H I J K L M I 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 135 148 164 174 180 182 181 181 179 168 168 179 181 181 182 180 174 164 148 135 141 146 166 123 198 135 132 187 171 191 186 175 180 181 182 180 174 164 148 135 141 146 167 123 200 138 140 131 182 195 187 175 180 181 182 180 174 164 148 135 143 149 172 133 238 126 58 116 217 208 191 176 180 181 182 180 174 164 148 135 141 145 163 186 229 122 57 115 216 208 191 176 180 181 182 180 174 164 148 135 151 171 236 62 150 154 63 121 224 211 192 177 180 181 182 180 174 164 148 135 145 156 195 107 273 60 106 222 183 196 188 176 180 181 182 180 174 164 148 135 176 246 141 39 26 35 101 305 214 210 193 178 181 181 182 180 174 164 148 135 176 246 142 40 26 35 101 305 215 213 204 186 223 106 60 273 108 196 157 147 148 151 170 125 202 140 140 210 269 75 18 8 11 21 46 175 322 216 171 150 203 311 107 33 18 17 29 74 371 294 279 354 111 35 19 19 33 104 319 196 246 388 133 41 20 12 8 3 - 5 - 14 -14 - 5 3 8 12 20 41 133 388 246 365 120 35 17 11 9 8 3 -6 -15 -15 -6 3 8 9 11 17 35 120 365 Integrity of aircraft structural elements 371 A I B I o C D o Z, 6 E F G H J ~)~ K Scale: to'Z z' I I I 0 I ~ 200 Ibs Fig. 6. Rivet load redistribution pattern due to cracks at fastener holes in lap splice (Table 3). rivet 9 in case J is 48 lb towards rivet 10. Very long cracks (over 10 rivets) are found to even reverse the load direction on some of the centrally located rivets (cases L and M). This happens due to inplane bending, again as a consequence of load flow diversion due to the crack. This analysis is, of course, not correct as the tear strap has yielded. Thus, the cases L and M are not further analysed. Looking at Table 3 one can see that case I is essentially the case when the configuration H exists on the left side of the tear strap and the configuration G exists on the right side of the tear strap. In this case, although the strap is overloaded, there is no mutual interaction between the cracks and so it need not be analysed as an independent case. 6.2. Local analysis The finite element alternating technique is employed to cases of interest, viz. B-H, J and K, by isolating the cracked portion (row and bay) from the panel. Crack tip parameters and stress fields are obtained from these analyses. The net section stresses and plastic zones are then computed. Figure 7 shows the maximum net section stresses and the residual strength in terms of cabin pressure based on the linkup stress of 55.5 ksi. Case J is intended to study the effect of a large crack on MSD and so the local analysis is performed only on the MSD region and the indicated maximum stress is not for the complete section, which is likely to be between rivet 9 and 10. The following observations are made from this figure: • Case B and C - - T h e maximum net section stress is not significantly increased by MSD and adequate residual strength still exists in the panel. • Case C and D - - L i n k u p shoots up the maximum net section stress. • Case B and J - - A long crack on or across the tear strap marginally increases the net section stress in the MSD zone. • Case D and E - - M S D near a long crack increases the ligament stress. • Case F - H and K - - O n e or two long cracks involving five or more rivets push the maximum stress beyond yield stress, thus substantially reducing the residual strength. The sections at the crack tips either yield (F, H, K) or tend to yield (G). 372 J.H. et al. PARK 0 B O 0 O- -0- O- -O-~ -O - 0 0 C O 0 O- -0- O- - O - "q - O - -0- 0 D O 0 O- --0- O - ~--O O O- 0 E O O 0 O 0 "' "O O O- 0 F O 0 0 O .....0 ~'~ - C~ 0 G O 0 0 O- ~ 0 - ~ 0 ~ - - 0 0 0 H O 0 -0 d O 0 K 0 0'°---0 ~ - 9 25 psi ~ 0 0 0 ,0 0-~0 -0- O- - O -'~ "O'- O 0 0 O~ 0 eZ 0-- 0 .0 0 m 0 9 25 psi Critical Pressure Link up Stress = 55.5 ksi Fig. 7, Maximum net section stresses and the correspondingcritical cabin pressure (psi) based net ligament stress criterion (Table 4). The stress intensity factors and the corresponding residual strength obtained with K,c = 9 3 . 0 k s i v / ~ . are shown in Fig. 8. The following observations are made from this figure: • Case B and C - - S h o r t and medium M S D cracks at neighboring rivets have marginal effects, e.g. SIF at rivet 7 goes up from 14.2 to 14.5 due to a crack at rivet 8. Short/medium cracks, one rivet hole apart, do not interact. Adequate residual strength exists in these situations. • Case C and D - - L i n k u p causes significant increase in the crack tip SIF, but has little effect on the SIF at M S D in the neighborhood. • Case D and E - - L i n k e d up cracks in close proximity affect each other marginally and do reduce the residual strength. • Case H and K - - I n moving from H to K, one more fastener hole is involved. Interestingly, the SIF on one end falls and goes up on the other end. This is due to the load redistribution. The residual strength for these cases has just fallen below the applied load o f 9.0 psi. The maximum SIF, net section stresses and the corresponding residual strengths are listed in Table 4. Looking at Figs 7 and 8, and Table 4 it is quite clear that the net ligament yields before the crack becomes unstable, under static loading conditions. Plastic zone sizes for all the cases have been computed (Fig. 9). For M S D cases B, C and J, the plastic zones are insignificantly small. They are found to grow as cracks link up. Table 5 gives the various K, and rp values for all the crack cases analysed. Plastic zone size becomes significant once K, exceeds c a 50.0, which happens when three or more rivets are involved in a single crack. Integrity of aircraft structural elements 373 0 B o o oJ toJ to J o o C o o oJ toJ toJ to J o D o o o~ E 0 o 0 o ~DJ o F o o 0-----0 O o 0 o 0 0 H o o o o J o o O O K 0 0 0 o 25 psi 9 o .4 o-1 0 o o 0 o---o C~j o o m I n i I ~Frame 0 25 psi 9 Critical Pressure K i c = 93.0 ksi i ~ n Fig. 8. Stress intensity factors and the corresponding critical cabin pressure (psi) based on fracture mechanics criterion (Table 4). 7. TWO-BAY CRACKING AT LAP SPLICE We now consider the aspect of a multi-bay crack with, and without, an MSD ahead of it. In this section, we consider a panel of shell 1, consisting of six bays in the longitudinal direction and six bays along the circumferential direction, and a lap joint. This panel contains five frames, five tear straps and seven stringers. Tear straps exist at all frame stations. Let there be a long crack extending equally on two sides of the central broken tear strap. All the other stiffening elements (frames, T-straps and stringers) are considered intact. We are interested in the residual strength for various lead crack lengths and the effect of MSD ahead of the crack tip. 7.1. Single dominant crack at fastener holes First consider the situation of a single dominant crack only. Global analysis is carried out to obtain the rivet bearing loads and then local FEAM is applied to obtain the crack tip SIF and net section stresses. Table 4. Residual strengths for various cases considered (cabin pressure, psi) Fracture mechanics Case B C D E F G H J K Max K~ 14.6 14.8 53.4 52.7 53.4 55.6 98.2 15.9 108.9 Net section stress Critical pressure Max a.~ Critical pressure 57.33 56.55 15.67 15.88 15.67 15.05 8.52 52.64 7.68 22.3 22.6 38.4 30.8 55.5 44.8 59.9 23.7 61.9 22.40 22.10 13.00 16.21 9.0 11.15 8.34 21.08 8.07 374 J . H . P A R K et al. B 0 (D- -<)- O- -0- -<)- 0 0 0 0 0 -0- -0- -0- 0 0 0 0 C 0 0 O- -0- O- D 0 0 O- -0- O- e-O 0 O-e 0 0 0 0 E 0 0 0 0 0 e<) 0 O-e 0 0 0 0 F 0 0 0 0 o - e q~O 0 O~ 0 0 0 0 G 0 0 0 0-t 0 0 ~ 0 0 0 0 0 H 0 oI-o o o o o4o o 0 0 0 J 0 0 -0- 0-- -0- O- oI-o K o o o -0- o 0 o-1---o- 0 o-0o I I 1 I I 0 0 I 0 I I i = Plastic zones smaller than 2" = i 0.035 " are not shown Fig. 9. Plastic zones for various crack configurations (Table 5). In the last section, we saw that the magnitude of crack length from the hole center (overhang) is an important factor in determining the critical pressure. For a single dominant crack, at the outer critical row of fasteners, spreading equally on both sides of a broken tear strap, the critical pressure diagram is shown in Fig. 10. Linkup stresses for the lead crack (ahead of which there are fastener holes, but without MSD cracks) correspond to the net section yield between the crack tip and the fastener hole. Up to c a 40% overhang, the shell panel is K,c critical and for the later half, it has Table 5. SIFs and the corresponding plastic zones for all crack tips Case B C Kl rp 10.5 13.4 13.5 12.5 14.5 14.6 0.0160 0.0269 0.0271 0.0205 0.0303 0.0300 14.4 14.2 0.0299 0.0296 10.5 13.5 0.0161 0.0272 13.6 0.0274 12.5 14.7 0.0208 0.0313 14.8 14.6 0.0311 0.0304 14.5 13.8 13.4 0.0301 0.0285 0.0279 K, rp 11.0 14.4 14.9 13.2 53.4 50.8 0.0176 0.0318 0.0331 0.0181 0.2142 0.2010 52.7 50.3 0.2101 0.1986 53.2 53.4 0.2882 0.2089 53.4 0.2121 G 29.2 55.6 0.1090 0.2237 H 92.5 98.2 0.2923 0.3210 Case D E F Case J K Kl rp 10.6 13.8 13.7 12.6 15.2 15.0 0.0171 0.0274 0.0274 0.0214 0.0307 0.0303 15.9 15.8 0.0314 0.0316 108.9 82.5 0.3395 0.2873 Integrity of aircraft structural elements Frame Frame --] T strap 35.0 [, ~/[ ~ Critical Section ,~- .~ ~ _ - 375 8.6 psi T strap Yieldin~1g I i i/ SIF 1 I 0.(I-0.0-- _! 10.0 Half Crack Length Frame 20.0 ~ [ Tear Strap Fig. 10. Critical pressure diagram for a single dominant multi-bay crack at outer critical row of fastener holes in a lap splice. too little section strength. Interestingly, a two-bay crack is fully section strength critical. Frames/tear straps appear to provide adequate residual strength to the panel. 7.2. Single dominant crack and MSD at fastener holes We now explore the effect of MSD near fastener holes ahead of the dominant crack in a lap splice. The important parameters are the lead crack length, the lead crack overhang from the nearest fastener hole, and the number and lengths of MSD cracks near fastener holes ahead of the lead crack. The MSD considered for the present purposes is over the three rivets immediately ahead of the lead crack tip, as the far away MSD cracks have insignificant effect on the lead crack tip stress field and also have lower intact ligament stress. In the MSD zone, cracks of equal length are presumed to be present on both sides of the rivet holes. To understand the MSD effects, for various lead crack lengths, starting from a situation where the lead crack spans over only two holes, to a situation of a multi-bay crack involving more than 40 rivets, the following two sets of crack configurations (overhang and MSD length) are analysed. (1) MSD length of 0.12in., which corresponds to the maximum that can hide under the countersunk head and stays undetected during regular economical nondestructive inspections. Two values of lead crack overhang, viz. 0.25 in. and 0.50 in., are considered. (2) Lead crack overhang of 0.25 in., and the MSD lengths of 0.12, 0.25 and 0.35 in. The analyses for the configurations of the two sets are performed and the residual strengths are computed, based on linear elastic fracture mechanics, as well as net section stress criteria (Tables 6, 7). From Table 6, it can be seen that the small MSD cracks (0.12 in.) have marginal effect on the residual strength of the shell panel, whether it is computed based on fracture mechanics or section yielding criteria. Of course, for relatively large overhang values, i.e. 0.75 in., the section stress shoots up with the presence of MSD. J. H. PARK et al. 376 Table 6. Effect of MSD on K[, a~v and residual strength. Set 1, constant MSD crack length of 0.12 in. LEFM Net section stress Half crack length Kj Critical pressure No MSD MSD No MSD 1.75 3.75 5.75 7.75 9.75 11.75 13.75 15.75 17.75 18.75 19.75 20.75 21.75 22.75 27.5 41.2 53.8 64.4 72.7 79.4 84.8 88.0 81.9 68.8 47.6 28.6 30.7 36.4 28.4 42.1 54.2 64.0 71.5 77.5 82.2 83.1 79.2 65.5 43.2 25.2 28.6 35.1 28.5 19.0 14.5 12.2 10.8 9.9 9.2 8.9 9.6 11.4 16.4 27.4 25.5 21.5 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 19.00 20.00 21.00 22.00 23.00 27.7 41.2 53.9 64.6 73.1 80.0 85.6 88.6 81.5 67.4 47.5 31.5 32.6 37.5 28.3 41.8 54.1 64.2 72.0 78.3 83.2 84.3 78.9 64.2 43.6 28.4 30.8 36.4 28.3 19.0 14.5 12.1 10.7 9.8 9.1 8.8 9.6 I 1.6 16.5 24.8 24.0 20.9 MSD aav No MSD Critical pressure MSD No MSD MSD 12.4 17.4 22.4 26.9 31.0 34.4 37.1 38.5 36.0 30.5 23. I 19.7 18.3 19.1 39.4 28.6 22.2 18.5 16.1 14.5 13.5 13.0 13.8 16.1 21.I 24.6 26.7 25.7 38.5 27.4 21.3 17.7 15.4 13.9 12.9 12.4 13.3 15.6 20.7 24.2 26.1 25.0 20.6 29.1 37.5 44.9 51.2 56.5 60.6 62.2 57.2 47.1 34.3 28.5 27.6 29.9 24.6 17.6 13,7 11.4 10.0 9.0 8.4 8.1 8.8 10.7 14.4 16.9 17.8 16.7 23.2 16.4 12.7 10.6 9.3 8.4 7.9 7.7 8.3 10.1 13.9 16.7 17.3 16.0 Lead crack overhang = 0.25 in. 27.6 18.6 14.4 12.2 10.9 10.1 9.5 9.4 9.9 11.9 18.1 31.1 27.4 22.3 12.1 16.7 21.5 25.8 29.6 32.9 35.4 36.7 34.7 29.6 22.6 19.4 17.9 18.6 Lead crack overhang = 0.50 in. T h e effect o f M S D 27.7 18.7 14.5 12.2 10.9 10.0 9.4 9.3 9.9 12.2 17.9 27.6 25.4 21.5 19.4 27.1 34.9 41.8 47.8 52.8 56.7 58.6 54.2 44.8 33.2 28.2 26.8 28.6 crack length on the critical pressure, c o m p u t e d b a s e d on linear elastic f r a c t u r e m e c h a n i c s c r i t e r i a , is m a r g i n a l ( T a b l e 7). T h e i n c r e a s e in M S D c r a c k l e n g t h h a s t w o effects: (1) It r e d u c e s t h e i n t a c t l i g a m e n t l e n g t h , r a i s i n g t h e n e t s e c t i o n stresses; as is v e r y c l e a r f r o m t h e r i g h t h a l f o f T a b l e 7. T h i s i n c r e a s e s t h e l e a d c r a c k tip S I F . (2) It relieves t h e f a s t e n e r l o a d , d i v e r t i n g t h e s h e e t s t r e s s e s a w a y f r o m t h e l e a d c r a c k tip. T h i s l o a d f l o w r e d i s t r i b u t i o n r e d u c e s t h e l e a d c r a c k tip S I F . Table 7. Effect of MSD crack length on residual strength. Set 2 Residual strength based on Half lead crack length 1.75 3.75 5.75 7.75 9.75 11.75 13.75 15.75 17.75 18.75 19.75 20.75 21.75 22.75 LEFM Net section stress MSD crack length MSD crack length No MSD 0.12 in. 0.25 in. 0.35in. No MSD 0.12 in. 0.25 in. 0.35 in. 29.0 19.4 14.9 12.4 II.0 10.1 9.4 9.1 9.8 11.6 16.8 28.0 26.1 22.1 28.2 19.0 14.8 12.5 11.2 10.3 9.7 9.6 10.1 12.2 18.5 31.7 27.9 22.8 28.1 18.8 14.6 12.4 ll.l 10.2 9.7 9.6 10.1 12.1 18.5 31.9 27.9 22.6 27.9 18.5 14.3 12.1 10.9 10. I 9.5 9.4 10.0 12.1 18.4 32.1 27.9 22.4 39.4 28.6 22.3 18.5 16.1 14.5 13.5 13.0 13.7 16.1 21.1 24.5 26.7 25.7 38.5 27.5 21.3 17.7 15.4 13.9 12.9 12.4 13.3 15.6 20.7 24.3 26.1 25.0 30.2 21.4 16.6 13.8 12.0 10.8 10.0 9.7 10.4 12.3 16.5 19.6 20.8 19.8 23.0 16.2 12.5 10.4 9.1 8.2 7,7 7.4 8.0 9.6 13.0 15.6 16.3 15.3 Integrity of aircraft structural elements 4o.o-I [] ... v,so 377 . I 20.0 10.0 0.0 I I I I I I I I I I I I 1 I 1.75 3.75 5.75 7.75 9.75 11.75 13.75 15.75 17.75 18.75 19.75 20.75 21.75 22.75 Half Lead-Crack Length (in) Fig, 11. Effect o f M S D crack length o n residual strength, based o n net section yield criterion (Table 7). The net effect is the sum total of these two. For smaller MSD crack lengths, the load flow redistribution dominates, causing a fall in the SIF. For longer MSD crack lengths, the net section overloading dominates and raises the SIF. But overall, the effect of MSD on the lead crack length SIF is marginal. However, the maximum net ligament stress increases steadily with increase in MSD lengths as the two crack tips come closer; in this case, the plasticity dominates and crack linkup based on the net section stress criterion is highly likely. We see that the MSD does not significantly alter the lead crack tip SIF, but does raise the net ligament stress when the two crack tips come closer. The use of linear elastic fracture mechanics to study the effect of MSD infested holes ahead of a lead crack is thus questionable. It may be concluded that the effect of plasticity and net section yield dominate the situation when holes with MSD cracks are present ahead of a lead crack. The residual strength estimate, based on net section yield, is significantly reduced in the presence of MSD near holes ahead of a lead crack. Figure 11 represents the fall of residual strength with MSD crack length in the form of a bar chart. Also, judging from the results in Section 5, it is the effect of holes ahead of the lead crack that is most dominant, whether or not these holes have MSD cracks emanating from them. 8. E L A S T I C - P L A S T I C ANALYSIS OF FLAT PANELS W I T H MSD 8.1. Problem definition Consider an A1 2024-T3 tensile sheet of width 20 in., height 40 in. and thickness 0.04 in. with a center crack of length " 2 a " and two MSD cracks of length " 2 b " each. Let the spacing between the main crack and the closest MSD crack be denoted by " c " and the MSD cracks be spaced apart by a distance " d " . The configuraton is shown in Fig. 12. Two configurations listed in Table 8 have been analysed. Because of the double symmetry, only one quarter of the specimen needs to be considered. The finite element mesh of the uncracked body (which is only needed in the present Elastic-Plastic Finite-Element Alternating Method) consists of 160 quadratic, eight-noded, isoparametric elements with 40 subdivisions in the crack line direction and eight subdivisions in the Table 8. Crack configurations (Fig. l l) Configuration a b c d No. 1 3.80 0.25 4.50 1.50 No. 2 2.50 0.50 3.50 2.00 378 J.H. PARK et al. ' " ' Ill l l lil l l l lIl l )l l l l l l l l " tllllllllllllltllltllllllllllllllllll eg'°n ' 40 l l tl l l li]l l l l l l l l l l l l l !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! IIIIIIIIIIIIIIIIIIIIIIII"""'IIIIIIII tJ= t t t t 20 " t M e s h subdivision -I~ ~ Fig. 12. Tensile sheet with two MSD cracks (Table 8). load direction. Though the crack configuration changes (Table 8), a single finite element mesh of the uncracked body is adequate in the present alternating procedure. 8.2. Results and discussion For the two configurations considered (Table 8), both elastic [E-FEAM] and elastic-plastic [EP-FEAM] analyses have been performed. To have a reference, an elastic analysis for a single crack was performed, K~ obtained and the critical strength of the panel computed. This is shown by a dashed line in Figs 13 and 14 (based on K~c = 84 ksi x/~.). Critical strengths of the two panels, 50.0 ..... a~ 40.0 ~ ~ £ Single crack Based on Kic( 84 ksi- i41K) Based on ligament yield ( 47ksi ) Based on T* ( 413 psi-in ) 30.0 O1 ¢- 20.0 O ~3 10.0 0.0 0.0 ,~ 2.0 4.c I t~.q 8.0 Half Crack Length (inches) I [~}::!i!~:MS r J i 10.0 i D!g!i!iis?i~::;ii!il!iiii?iii!i!!!ii!iiiii!::i::i!iiiiiii!::i!iii?ii:i!i:i] I i Fig. 13. Critical strength diagram for configuration no. 1. Integrity of aircraft structural elements 50.0 i ~ \\ 40.0 e- 379 . . . . . Singlecrack -------0----Basedon Ktc(84 ksi- J[-ff) .----o--- Basedon ligament yield ( 47ksi ) I }~\ ~ Based on T* ( 413 psi-in ) I 30.0 c- ~. 20.0 No. 2 O .=_ ~3 10.0 0.0 0.0 2.0 I 410 I 610 8.0 Hall Crack Length (inches) i i I 10.0 I :::::::::::::::::::::::::::::::::::~i::~:::::~::~i~:.`!i~::!`::i::~i~::::~i~!~iii~i!~:~i::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: :i Main cra,e.k ~i~{%':~!:i!~!!!!i MSD i! i~:!:i::::r:!i~:~i~]i!~:.ii!iiiii iiiiiiiii~ii] ::!::::;:::::::::i:::i:!:i:h;.~:;:::::::::~:i:!~!z~i:i::::;:::::::;:~:i:;;~i};i~`~:;..:~!:;:~;~`~`!~!;;.~i~t::;:;:i:!:i:i:i:;:;:;:::::::::5:::::::::~:~ Fig. 14. Critical strength diagram for configurationno. 2. computed using three different criteria, are shown by solid lines in these figures. A point on these curves represents the value of applied tensile stress to the panel that would cause (1) Main crack tip SIF = 84 ksi ~/i-n. (from elastic analysis), (2) Complete yielding of the uncracked ligament (from elastic-plastic analysis), (3) Main crack tip T * = 413 psi-in. (from elastic-plastic analysis). The values of K~c = 84ksi ~ / ~ . and T * = 413 psi-in., used in the present section, have been obtained from the experimental results of Jeong [9]. The following observations can be made from these figures: (1) Critical strength of the panel computed using elastic-plastic analysis is lower than that computed using elastic analysis. (2) For small main crack lengths, the panel is critical from fracture mechanics consideration, i.e. as the applied load is increased, the crack would become unstable before any of the sections yield under stress. When the main crack tip approaches the MSD crack tip (distance between the tips 1.25-1.4 times the MSD length), the net section stresses go up and the intact ligament yields before the crack becomes unstable. (3) After first linkup, both the panels are found to be section stress critical up to the second linkup. The following should be noted concerning the residual strength predicted by T*: (1) Each point on the curves based on T* represents the value of T* needed to initiate the growth of the crack from its current length, but does not consider the actual growth of the crack from one length to another. (2) The effect of the plastic-wake during the actual growth is not modeled. However, an actual simulation of the growth from one length to another is necessary to fully understand the linkup process. The linkup criteria can be based on the crack tip values of T* [energy flux to the crack tip region, of radius El, the CTOA (Crack Tip Opening Angle) or CTOD (Crack Tip Opening Displacement). However, it can easily be shown that T*, CTOA and CTOD 380 J.H. PARK et al. are all linearly related. It has been shown earlier by Atluri [10] that T * remains constant during elastic-plastic stable crack growth. It is also well-known that C T O A remains constant during stable growth. Thus, either T* or C T O A can be used as criteria for stable crack growth and linkup. However, C T O A being a geometric parameter, its c o m p u t a t i o n requires a very, very refined finite element mesh. On the other hand, T * can be calculated t h r o u g h the Equivalent D o m a i n Integral M e t h o d developed by Nikishkov and Atluri [11], as an integral over the d o m a i n between the far-field c o n t o u r and a c o n t o u r o f radius E near the crack tip. Thus, finite element meshes (Fig. 12) are sufficient to c o m p u t e T * . Thus, the T * based linkup criteria would be far more economical both from engineering man-power, as well as c o m p u t e r time, points o f view. These results will be presented shortly in a separate report. 9. C O N C L U S I O N S Analyses o f fuselage shell panels with longitudinal cracks at various locations have been carried out using G l o b a l - L o c a l F E M . The conclusion that can be drawn from all these analyses is that the presence o f rivet holes (with or without M S D emanating from these holes) ahead o f a lead crack reduces the residual strength o f the d a m a g e d panel significantly. The most important parameter turns out to be the location o f the lead crack tip with respect to the holes and M S D crack tips. If the net ligament is small, the section stresses shoot up, bringing the residual strength to a very low value. Acknowledgements--The authors are grateful for the financial support from Fedral Aviation Administration to the Center of Excellence for Computational Modeling of Aircraft Structures, Georgia Institute of Technology. They also acknowledge with pleasure the discussions and support from various colleagues, in particular Mr Tom Swift and P. W. Tan of FAA, and Dr Zhao, Mr Wang, Mr Shenoy and Dr Nikishkov of Georgia Tech. REFERENCES [1] J. H. Park and S. N. Atluri, Fatigue growth of multiple cracks near a row of fastener holes in a fuselage lap joint. Proc. Int. Workshop Structural Integrity o f Aging Airplanes (Edited by S. N. Atluri, C. E. Harris, A. Hoggard, N. Miller and S. G. Sampath), pp. 91-116. Atlanta (1992). Also in Computational Mech. 13, 189-203 (1993). [2] G. P. Nikishkov and S. N. Atluri, An analytical-numerical alternating method of elastic-plastic analysis of cracks. Computational Mech. 13, 427-442 (1994). [3] D. G. Ashwell and A. B. Sabir, A new cylindrical shell finite element based on simple independent shape functions. Int. J. Mech. Sci. 14, 171-183 (1972). [4] T. Swift, Fracture analysis of stiffened structures. A S T M STP 842, 69-107 (1984). [5] J. H. Park, T. Ogiso and S. N. Atluri, Analysis of cracks in aging aircraft structures, with and without composite patch repairs. Computational Mech. 10, 169-202 (1992). [6] G. C. Nayak and O. C. Zienkiewicz, Elasto-plastic stress analysis. A generalization for various constitutive relations including strain softening. Int. J. numer. Meth. Engng 5, 113-135 (1972). [7] T. Swift, Damage tolerance in pressurized fuselages. 11th Plantema memorial lecture, ICAF, Canada (1987). [8] T. Swift, Residual strength of stiffened structures. Lecture at Georgia Tech. (I0 February 1993). [9] D. Y. Jeong, Preliminary analysis of FMI flat panel experiments. Foster-Miller and Teledyne Engineering Services, Draft report (December 1992). [10] S. N. Atluri, Energetic approaches and path-independent integrals in fracture mechanics, in Computational Methods in the Mechanics o f Fracture (Edited by S. N. Atluri), Elsevier Science (1986). [11] G. P. Nikishkov and S. N. Atluri, An equivalent domain integral method for computing crack tip integral parameters in non-elastic, thermo-mechanical fracture. Engng Fracture Mech. 26, 851-867 (1987). (Recewed 9 March 1994)