ARTICLE IN PRESS

International Journal of Fatigue xxx (2008) xxxxxx

Contents lists available at ScienceDirect

International Journal of Fatigue

journal homepage: www.elsevier.com/locate/ijfatigue

Fatigue crack growth history in damage tolerance design of aircraft structures

Kunz Jan Siegl *, Ivan Nedbal, Jir
Czech Technical University in Prague, Faculty of Nuclear Sciences and Physical Engineering, Department of Materials, Trojanova 13, 120 00 Prague 2, Czech Republic

a r t i c l e

i n f o

a b s t r a c t
Knowing the real history of the fatigue process is a valuable asset to the for design, development, and reliable operation of structures exposed to time-variable loading service conditions. Information of this type is encoded in the fracture surface morphology. This information can be decoded by quantitative fractography and transformed it into a form interesting from the engineering point of view. Application of these fractographic methods is conditioned by the existence and detectability of fractographic features the characteristics of which are correlated with fatigue crack growth rate. In the paper presented, three various fractographic features are used for fractographic reconstitution striations (in the case of constant amplitude loading), beach marks (for simple program loading), and special inserted fracture marks (for complex program loading). The importance and irreplaceable role of the fractographic analysis is illustrated by some case studies dealing with full-scale fatigue tests of aircraft structure parts. 2008 Elsevier Ltd. All rights reserved.

Article history: Received 26 February 2008 Received in revised form 15 April 2008 Accepted 15 May 2008 Available online xxxx Keywords: Crack growth rate Quantitative fractography Fractographic reconstitution of fatigue crack growth Aircraft components

1. Introduction Prevention of failures the many of which can be catastrophic is conditioned by the development of effective failure analysis methods. The cost of failure analysis may exceed the value of the fractured part, but the cost of service failures can usually be far above cost of failure analysis [1]. As a traditional ex-post experimental method, fractography has played an important and often irreplaceable role in analysing service failures, in full-scale testing, etc. Fracture surface topography created by a growing crack is the result of interactions between material microstructure and service conditions, especially character of loading, temperature, environment, etc. [2]. Information on a failure process is encoded in the fracture morphology, and the main tasks of fractography is to decode this information, i.e., to nd a correlation between the identied fracture morphological features and the corresponding physical processes, and to transform it into a form interesting from the engineering point of view. Each fractographic feature can be characterised by a set of qualitative and/or quantitative parameters. Qualitative parameters give fundamental information about the fracture character (static or fatigue, brittle or ductile, transgranular or intergranular, etc.). Quantitative parameters are measurable characteristics of fracture surface, e.g., the area percentage of the particular fractographic feature, its size (e.g., ductile dimple diameter, striation spacing, cleavage facet size, etc.), local crack growth direction and others. A typical feature of current fractography is an effort to increase the volume and quality
* Corresponding author. Tel.: +420 224 358 508. E-mail address: jan.siegl@fj.cvut.cz (J. Siegl). 0142-1123/$ - see front matter 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijfatigue.2008.05.006

of quantitative data supplying better and deeper understanding of the fracture processes, and to integrate the of the experimental results with the theoretical knowledge of fracture mechanics. One of the main contributions of quantitative fractography is the reconstitution of fatigue process history, i.e., setting of dependence of the crack length (or some other variable expressing the failure extent) on the number of cycles (or on some other time related variable, e.g., service time, number of ights, etc.). The reconstitution is conditioned by the existence and detectability of a fractographic feature, some quantitative characteristic of which is correlated with the fatigue crack growth rate, e.g., striations, beach marks, marking lines, etc. According to the standards [3], a special consideration must be given to wide spread fatigue damage and included into the damage tolerance evaluation. Probable locations and modes of fatigue damage should be demonstrated with sufcient full-scale test evidence. The efciency of these exacting and expensive experiments can be considerably increased by quantitative fractography. Whereas possibilities of visual monitoring methods are limited (especially in the case of hidden and inaccessible cracks), fractography can provide precious supplementary data. The information encoded in fracture micromorphology is objective and often unobtainable in other ways. The methods of fractographic reconstitution can give detailed information on the fatigue process in the tested structure parts e.g., the length of crack initiation period; crack growth rates in various locations of the failed area; timedependent, two-dimensional description of fatigue crack growth; time sequence of individual failures and their interaction; consequences of applied repairs, etc. (e.g. [2,46] etc.). In addition to the assets mentioned, quantitative fractography is indispensable

Please cite this article in press as: Siegl J et al., Fatigue crack growth history in damage tolerance design of aircraft structures, Int J Fatigue (2008), doi:10.1016/j.ijfatigue.2008.05.006

ARTICLE IN PRESS
2 J. Siegl et al. / International Journal of Fatigue xxx (2008) xxxxxx

to studying of fundamental principles of fatigue mechanisms [7], to evaluating prediction models of fatigue crack growth under variable amplitude loading, [8] etc. 2. Fractographic reconstitution of fatigue crack growth 2.1. Method based on striation spacing measurements The basic problem of this method is to nd the relation between striation spacing and macroscopic crack growth rate. Application of this method, which can result in the most detailed information on fatigue crack growth, is illustrated in the following case study. The wing spar of the aircraft LM 200 was subject to the experimental study of fatigue cracks propagation [9,10]. The web plate 1.5 mm thick was riveted to the L-prole ange plates with dimensions 5 40 40 mm. On both ange plates, sheet strips 2 mm thick modelling the wing skin were furthermore riveted. All the main structural components of the wing spar were made of Al-alloy type 2124. The model of wing spar was loaded with four-point bending with constant sinus cycle at stress ratio R = 0 and loading frequency f = 2 Hz. In the critical area, the loading force induced a stress with rmax = 100 MPa. During the test two fatigue failures propagated in the spar. One of them initiated in the spar web on vertical sharp notches in the hole for the rivet connecting the spar web and the lower ange plate (cracks C1S1 and C1S2 in Fig. 1). The second one failed the lower ange plate (cracks C1P1 and C1P2 in Fig. 2). The growth of the main fatigue crack C1S2 in the spar web could be visually monitored only in the second part of the fatigue test (i.e., in the period 68 226 < N < 117 500). There is no macroscopic information about the growth of the other three mentioned fatigue cracks (i.e., C1S1, C1P1, and C1P2). The main results of qualitative and detailed quantitative fractographic analyses of both failures were very similar and can be summarised as follows:

 The both fatigue cracks in the web (C1S1 and C1S2 see Fig. 1) initiated on articial sharp notches in the rivet hole. Fatigue cracks in ange plate (C1P1 and C1P2 Fig. 2) initiated on the drilling traces in the rivet hole.  All fatigue cracks propagated mostly by striation mechanisms.  Striation spacing s was measured and presented as a function of crack length a, where a = distance of measured micro-area under study from rivet hole axis. Each point in the graphs in Figs. 3 and 4 represents a weighted mean value of about ten striation spacing data at the same crack length (over 700 individual measurements of striation spacing were carried out on the investigated fracture surfaces).  The results of striation spacing measurements for spar web (Fig. 3) imply that both fatigue cracks C1S1 and C1S2 were propagating at very similar rates.  The results for cracks C1P1 and C1P2 (Fig. 4) in the lower ange plate lead to the analogical conclusion, i.e., both cracks were propagating at very similar rates. The ex-post reconstitution of the fatigue process history was based on the results of quantitative microfractographic analysis, i.e., on the dependence of striation spacing on crack length (si, ai), i = 1,2, . . . ,n. Details of the method used were described previously by Nedbal et al. (e.g. [4,5]). The data processing consists of two steps: (a) recalculation of the striation spacing into the macroscopic crack growth rate, (b) integration of the results obtained into the relation of the crack length as a function of the number of applied cycles. Generally, striation spacing s and macroscopic crack growth rate v cannot be taken as identical. The ratio of the variables

Fig. 1. Failure of the spar web two fatigue cracks C1S1 and C1S2 propagated from notched rivet hole.

Fig. 2. Failure of the lower ange plate two fatigue cracks C1P1 and C1P2 were propagated from rivet hole.

Please cite this article in press as: Siegl J et al., Fatigue crack growth history in damage tolerance design of aircraft structures, Int J Fatigue (2008), doi:10.1016/j.ijfatigue.2008.05.006

ARTICLE IN PRESS
J. Siegl et al. / International Journal of Fatigue xxx (2008) xxxxxx 3

Fig. 3. Striation spacing vs. crack length measured for the cracks in the spar web.

Fig. 4. Striation spacing vs. crack length measured for the cracks in the lower ange plate.

v s

can vary within several orders (103 < D < 101), e.g., Nedbal [5,11]. This fact is a consequence of the three following phenomena (see, e.g. [2,4]):  Discontinuous propagation of fatigue crack front both in time and space, i.e., the existence of the so called idle cycles.  Deviation of local direction of fatigue crack growth from the macroscopic one.  Synergy of striation micromechanism with other fracture micromechanisms, e.g., ductile dimple fracture, inter crystalline decohesion, quasi-cleavage, etc. An a priori quantitative information on ratio D is very desirable for practical use. For these purposes, the most suitable form is the relation D = D(s), which can be determined from laboratory tests on simple bodies made of the same material as the fractured struc-

tural part under study. Also, close correspondence of testing conditions with the service ones is very advisable. In addition to the above information, knowledge of some couple of corresponding data (ai, Ni) contributes noticeably to the quality and exactness of nal results of fractographic reconstitution. This pair may be given by the number of cycles and corresponding crack length at the moment of service interruption, termination, etc. If dependence D = D(s) and one couple of corresponding data (ai, Ni) are available, the following equation can be used to reconstitute the fatigue crack growth history

Nx

ax ai

da Ni ; Ds sa

where Nx is the number of cycles corresponding to the given crack length ax, and (Ni, ai) is a priori known or determined couple of corresponding data. For all the observed fatigue cracks in the wing spar of LM 200, dependences s = s(a) were determined fractographically by means

Please cite this article in press as: Siegl J et al., Fatigue crack growth history in damage tolerance design of aircraft structures, Int J Fatigue (2008), doi:10.1016/j.ijfatigue.2008.05.006

ARTICLE IN PRESS
4 J. Siegl et al. / International Journal of Fatigue xxx (2008) xxxxxx

Fig. 5. Result of fractographic reconstitution of fatigue crack growth curves. Lengths of crack C1S2 measured optically during fatigue test are plotted for comparison. (a) Macroscopic character of fracture; (b) micromorphology of fracture surface.

of SEM see Figs. 3 and 4. For D = D(s) in Eq. (2), we used the relation from Fig. 5. This relation is based on our previous laboratory experiments on a similar Al-alloy [5]. As an initial condition (ai, Ni) necessary for reconstitution purposes by Eq. (2), the couple of data Nf = 117 500 cycles and af = 98.5 mm corresponding to the fatigue test termination was applied. The results of fractographic reconstitution of fatigue cracks growth in the tested model of wing spar are plotted in Fig. 5. For comparison, this graph also presents the results of macroscopic crack length monitoring. Fig. 5 clearly shows the benet of fractographic reconstitution. Based on quantitative fractographic analysis, we were able to describe the detailed history of fatigue processes in the critical area of the investigated wing spar. The most important information obtained fractographically refers to the fatigue cracks growth in the visually inapproachable hidden parts. 2.2. Method based on beach mark spacing measurements In the case of program loading spectrum, e.g., ight simulation loading, the reconstitution of fatigue crack growth can be based on a fractographic identication of morphological features created by specic inherent parts of the spectrum (e.g., gust cycles, the most severe ights, etc.). If the beach marks are detectable and their spacing is measurable, neither marking nor special modication of the original loading spectrum is required. In this case, no difculties arise in ascertaining the relation between microscopic and macroscopic crack growth rate. In addition to the crack growth rate, observation of beach marks gives information on the crack front shape and its changes in time and space. On this basis, two-dimensional description of fatigue degradation of body cross

section is possible. Application of this method for fractographic reconstitution of fatigue crack growth in aircraft structure parts was presented, e.g., by Kunz et al. [6] and Siegl et al. [12]. As an example, in the full-scale fatigue test, a small civil aircraft wing was loaded by means of ight-by-ight spectrum simulating real service condition. One loading sequence of this spectrum consists of nine randomly arranged ight types (in total 3156 ights) and corresponds to 2000 simulated ying hours (SFH). Individual types of ights denoted F1F9 have different incidence, maximum loading level and number of individual loading cycles (see Table 1). The wing hinge (made out of steel M300) was fractured after 119082 SFH, i.e., after 59.54 loading sequences [13]. Failure of the hinge corresponded to the propagation and coalescence of three fatigue cracks initiated in the hole for bolt (Fig. 6a). The cracks propagated mostly by the striating mechanism, and due to variable amplitude loading fracture micromorphology was characterised by the occurrence of more or less distinct beach marks. Individual beach marks determined the shape and position of fatigue crack fronts corresponding to the different ights. A detailed study of fracture micromorphology in combination with loading spectrum analysis enabled identication of beach marks corresponding to the application of the ve most severe ights (i.e., 1 ight F1 and 4 ights F7 in each individual sequence). Based on the beach marks identication, the reconstitution of fatigue failure of the wing hinge was carried out (see Fig. 6). 2.3. Method based on fracture marking If the testing loading unit does not include a suitable cycle with a detectable fractographic repercussion, it is possible to

Table 1 Characterisation of nine ight types F1F9 in the loading sequence of 3156 ights Flight Incidence Rel. max. level Number of cycles F1 1 3.17 95 F2 938 1.58 21 F3 600 1.58 13 F4 50 2.13 75 F5 468 1.58 19 F6 800 1.73 23 F7 4 2.99 133 F8 11 2.38 119 F9 20 2.38 127

There are the ve most severe ights in each sequence (in table marked with bold letters).

Please cite this article in press as: Siegl J et al., Fatigue crack growth history in damage tolerance design of aircraft structures, Int J Fatigue (2008), doi:10.1016/j.ijfatigue.2008.05.006

ARTICLE IN PRESS
J. Siegl et al. / International Journal of Fatigue xxx (2008) xxxxxx 5

Fig. 6. Fracture surface of failed wing hinge: fatigue crack fronts corresponding to the application of the most severe ights F1 and F7 are labelled.

Table 2 List of fractographic marks in Morse code (see [14]) Mark number Symbol 1 2 3 4 5 6 7 8 9 10 11 12

use some specic marking cycles or blocks inserted intentionally into the testing loading spectrum. The fracture marking is relevant especially in the case of complex random loading. Unlike other marking techniques as a dye-penetrant method, heat tinting, etc. the methods based on the application of special marking loading cycles or blocks make it possible to mark all fractures in the tested structure, including hidden or unknown cracks. Using this marking method is unfortunately complicated by two quite opposing requirements [14]: (a) good and unequivocal detectability of marks on the fracture surface,

(b) minimal inuence on fatigue crack initiation and growth, i.e., minimal change of damaging effect of the basic loading spectrum. A considerable experimental effort was devoted to developing different marking methods especially in aircraft industry. In principle, there are two main types of fracture marking based on special loading cycles: (1) marking by means of special loading cycles inserted into the original loading spectrum (see, e.g. [14,15] etc.), (2) marking by means of rearrangement of the original loading spectrum (see, e.g. [16]).

Fig. 7. Micromorphological characteristics of marks Nos. 1 and 12 (see Table. 2).

Please cite this article in press as: Siegl J et al., Fatigue crack growth history in damage tolerance design of aircraft structures, Int J Fatigue (2008), doi:10.1016/j.ijfatigue.2008.05.006

ARTICLE IN PRESS
6 J. Siegl et al. / International Journal of Fatigue xxx (2008) xxxxxx

An example of fracture marking, full-scale fatigue test of the main beam of ghter body is given in Nedbal et al. [14]: service conditions were simulated by means of the ight-by-ight loading spectrum. This very complicated spectrum consists of many cycles with very different amplitudes divided into three different loading sequences. The fractographic analysis of simple test bodies has conrmed that for the fractographic reconstruction of fatigue crack process neither the striation spacing measurements nor beach marks identication can be used. Therefore for the full-scale test of airframe special coded marking loading blocks had to be developed. One marking set consists of 12 different marks the symbols of which are given in Table 2. The symbols used are similar to Morse code: the dot consists of 100 constant amplitude cycles, the dash consists of the same 800 cycles. The micromorphological features of fracture surface corresponding to marks Nos. 1 and 12 are documented in Fig. 7. Individual dots and dashes were separated by a special cycle with higher stress amplitude. In one marking set, it was not difcult to identify individual marks each mark differs from each other and thus is unequivocal. Should it be difcult to directly identify some mark, it is possible to use a second level of information, i.e., dene it more precisely by identifying the neighbouring (the last before and the next after) marks, the sequence of the individual marks being constant and known. The same marking set is repeatedly applied during the full-scale fatigue test. Identication of individual marks on the fracture surface enables a complete fractographic reconstitution of the fatigue crack growth in the main aircraft beam fracture. The results were summarised as a two-dimensional description of fatigue process depending on the number of simulated ying hours (similarly as for wing hinge, see Fig. 6).

Acknowledgments Financial support for this research by the Ministry of Education, Youth and Sports, through project MSM6840770021, Diagnostic of Materials is gratefully acknowledged. References
[1] Parrington RJ. Fractography of metals and plastics. Pract Fail Anal 2002;2(5):1646. [2] Nedbal I. One lecture on fractography of fatigue failures. cole dt dveloppements rcent en fatigue des matriaux et des structures. Paris: cole Centrale Paris ICTM; 1997. [3] Federal Aviation Regulation. Part 25 Airworthiness standards: transport category airplanes. Sec. 25.571 Damage-tolerance and fatigue evaluation of structure; 1978. [4] Nedbal I, Kunz J, Siegl J. Quantitative fractography possibilities and applications in aircraft research Basic mechanisms in fatigue of materials. In: Luk P, Polk J, editors. Proceedings of the international colloquium. Materials science monographs, vol. 46. Amsterdam: Elsevier; 1988. p. 393403. [5] Nedbal I, Siegl J, Kunz J. Relation between striation spacing and fatigue crack growth rate in Al-alloy sheets. Advances in fracture research. In: Salama K, Ravi-Chandar K, Taplin DMR, Rama Rao P, editors. Proceedings of the seventh international conference on fracture, vol. 5. Oxford: Pergamon Press; 1989. p. 348391. [6] Kunz J, Nedbal I, Siegl J. Application of fractography in full-scale tests of aircraft structure parts Fracture behaviour and design of materials and structures. In: Firrao D, editor. Proceedings of the eighth European conference on fracture, vol. I. Cradley Heath: EMAS; 1990. p. 16629. [7] Kunz J, Nedbal I, Siegl J, Prtl O. Fractographic remarks to fatigue crack growth rate Fatigue 93. In: Bailon J-P, Dickson JI, editors. Proceedings of the fth international conference on fatigue and fatigue thresholds, vol. II. Cradley Heath: EMAS; 1993. p. 64954. [8] Schijve J. The signicance of fractography for investigations of fatigue crack growth under variable-amplitude loading. Fatigue Fract Engng Mater Struct 1999;22(2):8799. k A. Experimental data determination for the damage [9] Augustin P, Pte tolerance analysis of the wing spar with cracked web. In: Proceeedings of the 38th international conference on experimental stress analysis. Brno: Technical University of Brno, 2000. p. 914. k A. Application of quantitative [10] Kunz J, Siegl J, Nedbal I, Augustin P, Pte microfractography in damage-tolerance and fatigue evaluation of wing spar. In: Proceedings of the 24th international congress of the aeronautical sciences. Edinburgh: Optimage, 2004 [CD-Rom]. [11] Nedbal I, Kunz J, Siegl J. Microfractographic aspects of fatigue crack growth in , editor. Proceedings of the 7010 aluminium alloy. Fractography97. In: Parilk L international conference, Koice: IMR SAS; 1997. p. 26470. [12] Siegl J, Schijve J, Padmadinata UH. Fractographic observations and predictions on fatigue crack growth in an aluminium alloy under miniTWIST ightsimulation loading. Int J Fatigue 1991;13(2):13947. k O, Kunz J. Fractographic study of fatigue processes. In: [13] Siegl J, Nedbal I, Kovr ek Z, Strnadel B, editors. Fracture damage of structural parts. Ostrava: VBz Bu TU Ostrava; 2004. p. 16572. [14] Nedbal I, Kunz J, Siegl J. Some remarks on fatigue fracture marking. Fatigue 99. In: Wu XR, Wang ZG, editors. Proceedings of the seventh international fatigue congress, vol. 4/4. Beijing: HEP, Cradley Heath: EMAS; 1999. p. 23738. [15] Dainty RV. Use of marker blocks as an aid in quantitative fractography in full-scale aircraft fatigue testing: a case study. In: Mecholsky Jr JJ, Powell Jr SR, editors. Fractography of ceramics and metal failures ASTM STP 827. Philadelphia: ASTM; 1984. p. 285308. [16] Prtl O, Schijve J. Reconstitution of crack growth from fractographic observations after ight simulation loading. Int J Fatigue 1990;12(3):17583.

3. Conclusion The application of fractography to failure analysis of fatigued aircraft structures contributes to more precise identication of causes and failure processes. In order to satisfy the current damage tolerant design criteria that require the experimental determination of crack growth rates for various aircraft components, considerable effort has been devoted to the development of quantitative techniques specically directed to the evaluation of fatigue crack growth. Quantitative fractography is indispensable not only for the study of fundamental principles of fatigue mechanisms, but it can also offer detailed information on the fatigue process in the tested structures e.g., the duration of crack initiation period, crack growth rates, two-dimensional time-dependent description of fatigue crack growth, time sequence of individual failures and their interaction, consequences of applied repairs, etc.

Please cite this article in press as: Siegl J et al., Fatigue crack growth history in damage tolerance design of aircraft structures, Int J Fatigue (2008), doi:10.1016/j.ijfatigue.2008.05.006