Advances in the Bonded Composite Repair o f Metallic Aircraft Structure phần 9 pptx

440 Advances in the bonded composite repair of metallic aircraft structure in good general agreement with theoretical predictions.. ea's., Advances in the Bonded Composite Repairs of

Chapter 15 Graphite/epo.xy patching eficiency studies 439

Fig 15.18 Different damage configurations of "equivalent" width

that the combined contribution of Kl, and KIIr to the effective stress intensity factor was less than 8% for the configurations studied

Fatigue testing confirmed that various forms of damage could be repaired effectively with single patches For example, the fatigue lives of panels containing 40mm diameter holes and either single cracks or four 45 kinked cracks were improved by factors of 5.1 and > 15, respectively, by single-sided repairs with a

1 mm thick patch [80 mm x 80mml The measured fatigue crack growth rates were

Table 15.5

Comparison of stress intensity factor ranges for four 45" kinked cracks

and two diametrically opposite cracks, with tips at x, =4Omm and hole

radius = 20 mm ( R = 0.1, omrx = 65 MPa)

Crack configuration Patch thickness, mm AKP AKu AKp/AKL'

2 x diam opposite 2.0

cracks 1 .o 11.3 13.7 27.7 27.7 0.41 0.49

440 Advances in the bonded composite repair of metallic aircraft structure

in good general agreement with theoretical predictions For example, the mean

crack growth rate of 9 x 10~9m/cycle measured for four kinked cracks at a half

crack length of 35mm during R = 0 1 , am,,=41.25MPa loading, was in good agreement with predicted values of A K p = 5.7 MPam’I2 and Kmin/Kmax = 0.51, and da/dN-AK data for 2024-T35 1 aluminium alloy sheet Furthermore, the observed crack paths indicated little effect due to Mode I1 loading In the case of a 40mm

hole and two cracks, double sided patching resulted in crack arrest, in agreement with theoretical predictions Work is in progress to determine the effectiveness of patch repairs for other damage configurations

15.10 Future work

Although adhesively bonded gr/ep patch repair of cracked metallic structures has been studied extensively and service experience with repairs has been good, it appears that further work is required to address some remaining problems and to assess the full potential of the repair technique Specific research objectives include the following:

(a) To investigate the effect of variable amplitude loading spectra on patch debonding and hence on patch efficiency There is a clear requirement for a model to predict debonding, and for incorporation of this in a general model, which will enable the effect of patching on fatigue crack growth to be predicted for a wide range of loading spectra

(b) To assess the influence of hot-wet fatigue test environments on patch efficiency, and the effect of long-term pre-exposure to hot-wet environments on such behaviour

(c) To establish the advantages and limitations of repairs carried out by co-cure of prepreg and adhesive

(d) To develop and assess bonded patch repair schemes for applications involving elevated service temperatures

(e) To investigate the effectiveness of bonded patches for the repair of various forms of corrosion damage and battle damage in aluminium alloy structures

( f ) To develop and assess patch repairs for applications involving bonding over fasteners

(g) To assess the potential of bonded patches for the repair of SPF/DB titanium alloy structures, and to develop optimum repair schemes

(h) To develop “Smart” patches for monitoring repair performance in service, and

improved NDE techniques for (i) inspecting pretreated surfaces prior to

bonding, and (ii) assessing the strength and durability bonded patch repairs

0 British Crown Copyright 2001 Published with the permission of the Defence Evaluation and Research Agency on behalf of the Controller of HMSO

Chapter 15 Graphitelepoxy patching efficiency studies 44 1

References

I Kemp, R.M.J., Murphy, D.J., Butt, R.I., et al (1983) RAE Technical Report TR 83005

2 Sutton, G.R., Stone, M.H., Poole, P et a / (1984) In: Repair and Reclamation, The Metals Society;

3 Poole, P., Stone, M.H Sutton, G.R., et al (1986) In: The Repair of Aircraft Structures Involving

4 Sutton, G.R and Stone, M.H (1989) RAE Technical Report TR 89034

5 Dowrick, G., Cartwright, D.J and Rooke, D.P (1980) RAE Technical Report TR 80098

6 Young, A,, Cartwright, D.J and Rooke, D.P (1988) Aeronautical J., pp 41&421

7 Young, A,, Rooke, D.P and Cartwright, D.J (1989) Aeronautical J., pp 327-332

8 Ball, A S (1993) MOD Contract SLS 41B/2093, Final Report BAe-KDD-FCP-0104

9 Poole, P., Brown, K and Young, A (1990) RAE Technical Report TR 90055

pp 17.1-17.6

Composite Materials, AGARD-CP-402, pp 9.1-9.21

10 Poole, P., Lock, D.S and Young, A (1991) In: Aircraft Damage Assessment and Repair The

11 Poole, P and Young, A (1992) In: Theoretical concepts and Numerical Analysis of Fatigue [A.F

12 Baker, A.A (1988) In: Bonded Repair of Aircraft Structures (A.A Baker and R Jones, eds.),

13 Poole, P., Young, A and Ball, A S (1994) In: Composite Repair of Military Aircraft, AGARD-

14 Poole, P., Lock D.S and Young, A (1997) In: Proc of 1997 U S A F Aircrufi Structural Infqrit.)’

15 Poole, P., Brown, K., Lock, D.S et a/ (1999) In: Proc of I W 9 USAF Aircruft Structural Infegrit?,

16 Poole, P., Stone, M.H., Sutton, G.R., el al (2000) In: Proc of 2000 USAF Aircraft Structural

Institution of Engineers, Australia, pp 85-91

Blom and C.J Beevers, eds.], EMAS, pp 421438

Chapter 16

REPAIR OF MULTI-SITE DAMAGE

R JONES and L MOLENT"

Defence Science and Technology Organisation, Air Vehicles Division, Monash

16.1 Introduction

The phenomenon of multi-site damage (MSD) in aircraft has been under

examination in recent years by many in the aviation industry This section investigates the feasibility of applying advanced bonding technology to commercial aviation structures containing MSD The validity of this technology has already

been proven in its application to fatigue and stress corrosion in military aircraft, as described in other chapters of this book

The consequence of the undetected presence of MSD was dramatically illustrated

through the in-flight failure of a fuselage lap joint on an Aloha Airlines B-737 aircraft on April 28, 1988 Essentially this failure occurred due to numerous small cracks along a fastener line linking together, causing the residual strength of the

fuselage to be exceeded under pressurization A test programme was conducted to reproduce this type of failure, and an adhesively bonded boron/epoxy doubler for use as a repair or preventative measure has been developed

This chapter presents the results of a fatigue test programme, which also considers environmental and damage tolerance aspects, conducted using specimens representative of wide-bodied commercial aircraft fuselage lap joints This work was reported in detail in [l-lo]

Two separate generic specimens were considered, one representative of Boeing Commercial Aircraft Company (Boeing) and the other of Deutsche Airbus GmbH (Airbus) aircraft fuselage lap joints

* Air Frames and Engines Divbion, Aeronautical and Maritime Research Laboratory, Fishermum Bend, Virtoriu 3207 Australia

Baker, A A , Rose, L.R.F and Jones, R (ea's.), Advances in the Bonded Composite Repairs of Metallic Aircraft Structure

Crown Copyright 0 2002 Published by Elsevier Science Ltd All rights reserved

443

444 Advances in the bonded composite repair of metallic uircraft structure

Following the development of a bonded-composite repair for MSD in the fuselage lap-joint of wide bodied transport aircraft a number of full scale demonstration repair/reinforcements were undertaken

16.2 Specimen and loading

Figure 16.1 details a typical configuration of a modern Boeing wide bodied aircraft pressurized fuselage construction For the purpose of this work attention is focused on the lap joint area Local details of this location vary depending on the age of the aircraft and specific manufacturers’ details

It should he noted that, in many cases, in addition to fasteners, the lap joints are bonded together, either using hot or cold setting adhesives This is done for the purpose of increasing the fatigue life of the joint In service, environmental degradation may cause this bond to become ineffective, and corrosion of the mating skins could accelerate the onset of MSD (as was the case in the Aloha

incident) For these reasons bonded lap joints are not considered in this work

In the present investigation a worst-case scenario was assumed, viz: a non- bonded, full depth, upper plate, counter-sunk configuration as shown in Figure 16.2(a) Here the counter-sunk rivet hole, if accompanied by improper rivet head seating, leads to a phenomenon known as “knife-edging’’ (i.e the tip of the counter-sunk in the top plate “cutting” into the lower plate) This in turn leads

to a reduction in the fatigue life of the joint, relative to the case where the counter- sunk does not fully penetrate the plate, due to the sharp corners accelerating the initiation of cracks

The basic specimen used in this investigation (referred to as the “Boeing” type) consisted of two 2024-T3 clad aluminium alloy sheets 1.016mm (0.04 inch) thick, fastened with three rows of BACR15CE-5, 100” shear head counter-sunk rivets,

3.968 mm (5/32 inch) diameter, as shown in Figure 16.3 The width of the specimen

was chosen to coincide with the typical distance between tear straps of a B-737 aircraft

The upper row of rivet holes contained crack initiation sites, induced by means of

an electrical spark erosion technique, on either side, nominally 1.2 mm long This length was chosen so that the defect was obscured by the fastener head, which is representative of possible undetectable flaws These flaws were achieved by drilling the rivet holes undersize (3.85mm diameter), spark eroding the initiation sites to 1.225 mm, and then machining the counter-sunk (4.039 mm) to achieve the required hole diameter The accuracy to which the latter was performed determined the final configuration of the defects In some cases the defects only remained on one side of the hole, the other being removed by the tool Following this the fasteners were inserted

The specimens were manufactured by the then Australian Airlines (now QANTAS), from material supplied by them, to aircraft standards End tabs were

Chapter 16 Repair of multi-site damage

Frame

station

445

Fig 16 I , Typical wide body fuselage construction (from Boeing)

bonded to the base specimen to ensure failure would not initiate from the specimen ends (see Figure 16.3)

Since the amount of out-of-plane bending due to fuselage curvature in a typical fuselage joint was unknown, the local bending was prevented by testing the specimens bonded back-to-back and separated by a 12.5 mm thick honeycomb

446 Advances in the bonded composite repair of metallic aircraft structure

Joint description configuration

Chapter 16 Repair of multi-site damage 447

2024-T3 plate thickness = 0.040in

End tabs to be bonded

Fig 16.3 Uniaxial "Boeing" type lap joint specimen Note rivet numbering used in this chapter

core Details of the procedure used to construct the test specimens can be found in [2] In this configuration strain gauge results indicated no global bending or parasitic stiffening due to the honeycomb

One drawback of this method of testing is that the failure of one face (i.e the base specimen) terminates further testing of the other The over-riding advantage of

this technique is that due to symmetry, a heat cured repair can be applied to the base specimens without inducing extensive bending due to the thermal expansion mismatch of the parent material and that of the repair A view of the assembled test

specimen is given in Figure 16.4

The specimens were tested in various capacity servo-hydraulic test machines The specimens were loaded in tension to give a remote plate stress of 92 MPa (13.4 ksi) This figure was determined from operational data obtained for the US DOT MSD

Committee Review Board, see Table 16.1 (from [ll]), for the B-737 aircraft

448 Advances in the bonded composite repair of nietallic aircraft structure

Fig 16.4 Back to back bonded lap joint specimens

Table 16.1

MSD committee review results [l 11

Typical maximum normal operating stresses for Boeing 727 and 737 fuselage splices

Primary skin stress is pressure hoop stress

Actual

PR/T At frames highest Comment

10400 Midway between frames Maximum applied shear stresses are less than 25% of the 13000 psi hoop stresses

Maximum principal stresses are:

Tension ~ less than 110% of hoop stress Shear less than 60% of hoop stress

Chapter 16 Repair multi-site damage 449

The applied constant amplitude loading used was:

For the unreinforced specimen the tests were terminated when the upper row of fasteners had completely failed (i.e all cracks linked) For the reinforced specimens, eddy current techniques (see [2]) were used to check for crack growth beneath the doubler

16.2.2 Airbus lap joints

Initial work had concentrated on the development of a bonded composite repair for MSD in joints similar to those encountered in Boeing aircraft After discussions

with Airbus, four “generic” specimens typical of current (7050-T73) fuselage construction were provided by Airbus for repair and testing This work was reported in detail in reference [SI The specimens had previously been fatigue tested

to failure and each specimen had failed in the first row of rivets in the upper skin

On average the life to failure of the unrepaired specimens ranged from between

77200 to 2175600 cycles, depending on the test load which ranged from 80 MPa for the former to 43.7MPa for the latter The specimens were 115mm wide and (nominally) 420mm long Two of the specimens had also failed at the ends of the specimen near the edge of the change of thickness at the joint The skin thickness varied from 2.3mm near the joint, to 1.6mm elsewhere (see Figure 16.5)

In an attempt to reproduce the level of local constraint, as seen in service aircraft the specimens were “paired” and mounted back to back on a 25.4mm thick aluminium honeycomb core (this has negligible stiffness in the direction of the load,

as demonstrated for the “Boeing” specimens)

Whilst these specimens had failed at the first row of rivets in the upper skin, the third row of rivets in the lower skin is also a potential failure location To simulate cracking at this location a saw cut was inserted between the middle rivets in the lower skin of each of the four specimens (Figure 16.5) The specimens, along with the aluminium honeycomb and end tabs, were then assembled and bonded together

450 Advances in the bonded composite repair metallic aircruft structure

Fig 16.5 "Airbus" lap joint specimen [8]

using an epoxy-nitrile structural thin-film adhesive which was cured at 120 "C in an

autoclave

The repaired specimens were fatigue tested under constant amplitude loading,

with a maximum load of 47.49 kN and a minimum load of 4.83 kN These loads

correspond to a remote stress in the skin, suggested by Airbus, of 128.9MPa and

13.1 MPa respectively

16.3 Repairs

Conventional repair methods for detected cracking in fuselage lap joints involve

removing the damaged material and the use of a riveted scab patch, see Figure 16.6

Chapter 16 Repair of multi-site damage 45 1

Fig 16.6 Typical lap joint fastened repair

The concern here is that this introduces further stress concentrations, due to the increase in fastener holes, and that the close proximity of these repairs may lead to

a compromise in the damage tolerance of the structure The objective of this investigation was to evaluate a possible bonded repair or life enhancement for mechanically fastened fuselage lap joints

I6.3.1 Repair philosophy

In 1990, with the support of the then Australian Civil Aviation Authority (CAA, now Civil Aviation Safety Authority), a world wide study into the commercial application of bonded repair technology was performed [12] Thirty four organisations in eight countries, including ten manufacturers and seven Regulatory Authorities were consulted The following proposed design rules and procedures were subsequently adopted by the CAA; viz:

1 Designs shall be substantiated against the Damage Tolerance provisions of the United States Federal Aviation Regulations (FAR) Part 25.57 1 at Amendment

45

452 Advances in the bonded conyosite repair of metallic aircraft structure

2 The repair of any structural component which contains damage sufficient to reduce the aircraft structure to below design limit load residual strength shall not normally be attempted

3 Service time degradation, environmental and impact damage substantiation evidence shall be provided for the composite material and the structural bond,

as appropriate to the design This should include sufficient work to enable the composite repair to meet the intent of the damage tolerance requirements

4 Quality control consideration should include, for all critical areas, wedge testing

of bond strips produced during the repair process

In the CAA Airworthiness Advisory Circular it was stated that:

" civil requirements do not mandate an initial flaw approach However, it is often convenient to do so and this may reduce the threshold fatigue testing requirement This may be in recognition of leaving the initial crack in the metal unchanged but also may cover the presence of an un-bonded region in the joint." The bonded repairs/reinforcements described below were designed and tested to fulfil the above requirements

16.3.2 Repair details

A boron fibrelepoxy composite was chosen for the repair applications because of its high stiffness, relatively high coefficient of thermal expansion, low void content and the environmental resistance of the cured epoxy resin The specimens were surface treated prior to doubler application using the standard AMRL pre-bonding surface treatment, which includes a thorough degrease, mechanical abrade and application of an aqueous silane solution The boron/epoxy laminate was cured in

an autoclave to form the composite doubler and the repair was bonded to the specimens with the same epoxy-nitrile structural thin-film adhesive (FM73, Wayne, New Jersey) as described in other chapters, in an oven using mechanical pressure

16.3.2.1 Boeing specimens

The bonded doubler used in this investigation was a multi-segmented unidirectional boron/epoxy laminate, containing ten plies at the greatest thickness The laminate (203.2 mm long by 203.2 mm wide), was bonded over the joint, with one segment butting up against the skin step, to provide an alternative load path for each row of fasteners, see Figure 16.7, similar to the concept used in the F-111

Chapter 16 Repair of multi-site damage 453

wing pivot fitting upper plate repair [13] The doubler was applied after the specimens had been pre-cracked In order to evaluate the fail-safe nature of the repair, two specimens were also repaired following complete failure of the upper row of fasteners

In addition to the hot-curing adhesive (FM-73), a cold-setting adhesive (Flexon

241, Victoria, Australia) was also investigated

The doublers were applied both as a reinforcement (specimens with short cracks emanating from rivet heads) and as a repair (specimens completely cracked through)

In one specimen pair (A9/A10) the stiffness of the doubler (upper section) was reduced by a third, in order to investigate the optimum thickness or stiffness required

The three regions scanned were as follows, viz:

1 Region 1 A scan encompassing all three rows of rivets and regions of both the upper and lower skins The area scanned was 113mm by 98mm The scan generated an array of 114 by 105 data points

2 Region 2 A scan encompassing three rivets in the first (Le critical) row of rivets The area of the specimen scanned was 88 mm by 47 mm generating an array of

116 by 67 data points

454 Advances in the bonded composite repair of metallic aircraft structure

Fig 16.8 Location of SPATE regions

3 Region 3 A scan of a rivet in the first row The area of the specimen scanned

was 15 mm by 15 mm generating an array of 82 by 80 data points, each data point corresponding to a region approximately 0.2 mm square

In each case the scan region was sprayed matt black to achieve a uniformly high infrared emissivity The location of these regions are shown in Figure 16.8 and the resultant scans, in uncalibrated stress units, of regions 1-3 were shown in [6] The SPATE scan for region 1 is shown in Figure 16.9

16.4.1 I Discussion of results

Figure 16.9 clearly shows the load path taken in the specimen There are three

distinct zones, in each of which the bulk stress is relatively uniform The first zone is the region up to and including the first row of rivets At the line of the rivets there is

a rapid decrease in the bulk stress to a second relatively constant value This decrease in the bulk stress reflects the load transferred at the first line of rivets An analysis of this data reveals that approximately 45% of the load is transferred by this row of rivets

The bulk stress undergoes another decrease at the second line of rivets At this scale the stress concentration, in the bulk stress, at the first row of rivets, was not apparent

A more detailed view of the bulk stress field around the first row of rivets was taken for region 2 [6] This confirmed the rapid decay in the load at the first row of rivets, due to load transfer to the lower skin Again a t this scale, the stress concentration in the bulk stress was not apparent This also illustrated the rapid

decay of the stress concentration on the upper surface of the upper skin, However,

we know that, in the fatigue test program outlined in Section 16.5.1, cracks always

Chapter 16 Repair of multi-site damage 455

P

Fig 16.9 SPATE scan of region 1

initiated at the first row of fasteners This highlights the role that the full depth counter-sunk plays in creating a fracture critical location at the lower surface Unfortunately this surface is not observable via standard thermal emission techniques

An even more detailed picture of the stress field around a fastener in the first row

of rivets was taken for region 3 [6] The concentration in the bulk stress at the upper

surface of the upper skin was now visible However, it was seen that this stress concentration, which is quite small, was very localised

From these scans it was apparent that the lack of a bond between the upper and lower skins results in the major load transfer occurring at the first row of rivets This will exacerbate the stress concentration effect of the full depth counter-sunk holes at the lower surface of the upper plate even though the stress concentration at the upper surface is particularly localised

Had the upper and lower skins been bonded, a substantial proportion of the load would have been transferred prior to the first row of fasteners The boron/epoxy

doubler, described in Section 16.3.2.1, uses this concept to increase the damage

tolerance of the joint In this approach a boron/epoxy laminate is bonded to the

Advances in the honiied composite repair of metallic aircraft structure

external surface of the joint so as to provide an alternate load path, from the upper

to the lower skin, thus by-passing the critical row of fasteners

16.4.2 Finite element analyses

A three dimensional finite element analysis of the specimen was undertaken to confirm the above results The model contained 4296 nodes and had 10680 degrees

of freedom with the counter-sunk rivets modelled separately as three dimensional isoparametric elements, see Figure 16.10 This analysis confirmed the rapid decay

of the stress concentration in the bulk stress The load transfer at the first row of rivets was also consistent with experimental measurements

To confirm the low stress concentration at the rivet hole, the “Boeing” specimen was extensively strain gauged A strip of ten micro-gauges was located on the surface of the upper skin adjacent to a rivet in the first row The resultant strains for

a remote stress of 92MPa (13.4ksi) are given in Table 16.2 These results confirm the results of the previous investigation, and yields a localised strain concentration,

at the hole, of approximately 1.45

16.4.2.1 Elastic-plastic analysis

As will be described in Section 16.5.1.1 it was noted in the fatigue tests performed

on these “Boeing” specimens, that the linking of adjacent cracks appeared to occur when the remaining distance between the crack tips was approximately 2 mm To examine this phenomenon a 3D elastic-plastic finite element analysis was performed A main objective of this analysis was to determine if this “failure”

was due to net section failure of the remaining 2mm ligament, between the crack tips, or due to ductile fracture The basic finite element model used is as described

Chapter 16 Repair of multi-sire damage 451

Table 16.2 Surface strains

Distance from edge

of rivet head (mm) Hoop strain (w)

0.5 1.5 2.5 3.5 4.5 5.5 6.5 7.5 8.5

For mechanically fastened structures with multiple load paths, the commonly used J integral is path dependent, even for monotonic loading To overcome this

shortcoming the path independent r* integral, see [15] for more details, was calculated along a number of separate paths for a remote stress of 92 MPa, which was the maximum stress applied to the specimens in the fatigue tests At the peak load the equivalent stress intensity factor K , which was defined, as in linear elastic

fracture, in terms of r“ and E ( Young’s modulus) as:

K = ET* ,

was found to be approximately 23.9 MPam’” which is well below the fracture

toughness of the material However, the von Mises equivalent stress was found to exceed its ultimate permissible value for more than 0.5 mm of the 2.0 mm between the crack tips This implies that failure of the ligament, which was a precursor to the total failure of the specimens, was due to net section failure This result is consistent with experimental work (see [1,2] and Section 16.5.1) where it was found

that, regardless of the length of the cracks emanating from two adjacent fastener holes, linking of these cracks was always observed when the remaining ligament was approximately 2 mm

Whilst repairs to MSD in the first row of rivets has received considerable attention, and the fundamental mechanisms underpinning this technique deter- mined, the ability of an external bonded doubler to repair cracks in the third row of rivets in the lower (hidden) skin has not previously been investigated analytically

To address this problem the lap joint described in Section 16.2.2 was also modelled,

using two-dimensional eight-noded iso-parametric elements, with a 75 mm edge

458 Advances in the bonded composite repair of metallic aircraft structure

v v v v v v v v v v

loaded to 129 MPa

Fig 16.11 Schematic of FEM lap joint with edge crack [8]

crack in the lower skin at the third row of rivets (Figure 16.1 1) [8] In this analysis a

remote stress of 129 MPa was applied to the aluminium skins and the mechanical properties used for the composite, adhesive and aluminium alloy are given in Table

16.3

Table 16.3

Material properties used for the numerical analysis

E,,=208.116 x 10'MPa G.ry = 750 MPa E.yx=70.1 x 103MPa

Chapter 16 Repair of multi-site damage 459 Table 16.4

Critical composite and adhesive stresses (see Fig 16.7) [8]

The stress intensity factor K was calculated, for both the unrepaired and repaired

“Airbus” cases, using Eq 16.1, see [16]:

Here v is the crack opening displacement at a distance 1 behind the crack tip and

a is the half-crack length For the unrepaired case K was calculated to be

47.1 MPam”*, and for the repaired case K reduced to 6.89MPam‘’2 which is

approximately the fatigue threshold value for this material

The stresses at each of the critical points, see Figure 16.7, are given in Table 16.4 and were found to be beneath the design allowables The adhesive shear stress was also found to be beneath the endurance limit of 25MPa [17] This implies that, during fatigue testing, crack growth should be very slow, the adhesive should exhibit adequate damage tolerance with minimal degradation in its performance and that delamination should not occur

16.5 Specimen fatigue test results

16.5.1 Unreinforced baseline fuselage lap joint specimens

Only the “Boeing” type fuselage lap joint specimens were tested in this program Ten unreinforced specimens were tested (see Table 16.5) It should be noted that

in the initial process used to bond specimens Al-A4 to the honeycomb, adhesive flowed between the two aluminium sheets and across the upper row of fasteners This was detected subsequent to the failure of specimens Al/A2 This, in effect, enhanced the fatigue life of the upper row of fasteners, as can be seen from Table 16.5 This specimen failed within 140000 cycles of the last inspection, at which no damage had been detected The failure of A2, although probably influenced by the final failure of AI, occurred through the inner plate at the lower fastener row, in contrast to all other specimens This may indicate a possible danger in lap joint modifications which consider only the upper row This is of concern, since damage occurring at the lower fastener is only detectable from the interior of the aircraft

460 Advances in the bonded composite repair of metallic aircraji structure

Table 16.5

"Boeing" fuselage lap joint fatigue program

cycles at failure at Ligaments joined crack

A6/2" Reinforced n/a see below prior ~

A5/2'"! Reinforced n/a > 1300000d prior -

A6/2"" Reinforced n/a > 1316400 prior -

A9/2'" Reinforced' n/a 1000000 prior -

A 10/2'"' Reinforced' n/a I 1 11000 prior -

* Adhesive seepage occurred across top fastener row (enhanced fatigue life)

a No crack detected at 837000 cycles

Failed lower fastener row

Patch stiffness decreased by a third

Failed in grip area of A5

e Environmentally conditioned

+ Continuation of specimen test

('? Repaired after complete failure, using room temperature curing adhesive (Flexon 241)

@ Testing not completed

To rectify the seepage of adhesive into the lap joint, a teflon strip was placed over the skin step

During the cycling of the unreinforced specimens periodic crack measurements were made A mean life to failure of approximately 59000 cycles was obtained It was observed that, in each case, the initial linking of two adjacent cracks occurred when the remaining ligament was approximately 2 mm long

Figure 16.12 presents typical crack growth data for specimen A6 which suffered total failure

Chapter 16 Repair of multi-site damage

sp.~tmen M cirm a moats

Fig 16.12 Specimen A6 crack growth data

Figure 16.13 presents a view of a failure surface By referring to Figure 16.14 (from [18]) it is seen that these failures reflect the nature of in-service crack growth

In Figure 16.15, a comparison of crack growth data for the two most prominent cracks for a number of specimen was plotted, from the same common initial crack length as used in data given by Boeing (see Figure 16.16) In these figures only data for the most significant cracks occurring on a specimen were plotted In each figure the two curves corresponding to the least number of cycles, were the cracks which first linked

Comparison of the experimentally obtained crack growth rates to the crack growth rates obtained from fleet data, shows good agreement The experimental crack growth data are bounded, above and below, by the crack growth data obtained from service aircraft This agreement implied that this specimen geometry could be used to study the repair of fuselage lap joints

It was observed that the variation in the fatigue lives of the unreinforced specimens was partly due to the nature of the cracking In general, the shortest life occurred when the largest cracks in a specimen grew towards each other from adjacent fastener holes A longer life occurred when cracks initiated and grew from

widely separated holes

462 Advances in the bonded composite repair of nietallic aircraft structure

Fig 16.13 Failure of specimen A12

Even though, due to manufacturing tolerance, some notches were visible to one side of a fastener (i.e notch larger than had been intended), this did not necessarily facilitate the initiation of the most prominent crack in a specimen The “knife- edging” phenomenon was felt to be the dominant factor determining crack initiation

-Yeer d mgwfschre 1981 -HgM hours rot a d a b l e -Fibhi C&le8 39,523