Engineering Failure Analysis 14 (2007) 1605–1617

www.elsevier.com/locate/engfailanal

Aircraft engine bleed system tubes: Material and failure

mode analysis
A.M.L. Adib *, C.A.R.P. Baptista, M.J.R. Barboza, C. Haga, C.C.F. Marques
Escola de Engenharia de Lorena, EEL/USP, Cx. Postal 116, CEP 12602-810, Lorena, SP, Brazil

Received 14 October 2006; accepted 30 November 2006

Available online 30 January 2007

Abstract

The failures of T-ducts made of Ti A40 with 50 mm in diameter and 0.52 mm in thickness, used in the pneumatic system
of a commercial aircraft, are examined as part of a research program aimed to improve the development of such systems.
The investigation included the fractographic analysis of the burst parts, microstructure and internal duct surface observa-
tions and mechanical tests performed in samples taken from the failed ducts and from unused material. The failed duct
fracture surfaces showed typical features of transgranular fatigue cracking. Delamination and oxide particle deposits were
found in their internal surfaces. Tensile and fatigue tests revealed some effects of aging, but with no evidence of embrit-
tlement. The fracture mechanisms were the same for the failed ducts and for the unused material. The internal surface dam-
age, associated to the geometry of the duct and to the solid-solution hardening increases the local stress concentration.
These facts suggest that cracks had initiated at multiple sites around the inside surface, so that further increments of crack-
ing occurred on subsequent cyclic pressurization during service life until the occurrence of sudden final fast fracture.
 2006 Elsevier Ltd. All rights reserved.

Keywords: Aircraft failures; Fractography; Titanium; Fatigue

1. Introduction

The investigation of failures in aircraft parts and structures provides an important source of information
with regard to design improvement, material selection and prevention of further incidents. An assessment
of the frequency of failure modes in airplanes, made from case histories data, showed that fatigue is present
in 55% of in-service failures followed by corrosion and overload, responding for 16% and 14% respectively [1].
In the present paper the failures of thin-walled titanium ducts used in the pneumatic system of a commercial
aircraft are examined as part of a research program aimed to improve the development of such systems. The
investigation includes the fractographic analysis of the burst parts, microstructure and internal duct surface
observations and mechanical tests performed in samples taken from the failed ducts and from unused
material.

*
Corresponding author. Tel.: +55 12 3159 9914; fax: +55 12 3153 3006.
E-mail address: [email protected] (A.M.L. Adib).

1350-6307/$ - see front matter  2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfailanal.2006.11.053
1606 A.M.L. Adib et al. / Engineering Failure Analysis 14 (2007) 1605–1617

2. The engine bleed system

An important energy source in commercial aircrafts is the high pressure, high temperature air that is bled
from the engines and routed throughout the airframe to secondary systems responsible for the air conditioning
in the cabin and ice buildup prevention on the fuselage shapes [2]. This hot air is transported by the engine
bleed system, also called pneumatic system, a set of metallic ducts as shown in Fig. 1, containing curved
and straight sections, joints, welded parts, valves and sensors whose reliability and durability are important
factors for flight safety, maintenance cost reduction and for complying with certification requirements [3,4].
Several factors can affect the performance of the pneumatic system parts. From the development stage,
these include the quality of the raw material, the cold draw and weld processes, as well as the presence of stress
raisers in the duct parts. The various in-service factors that may lead to failure include fatigue and creep pro-
cesses, oxidation, stress corrosion and internal surface damage. Therefore, the effects of aging must be taken in
account when predicting the useful life of the ducts.
Among the various metallic materials suited for use in the pneumatic system, titanium stands out because
of its favorable characteristics, such as high strength-to-weight ratio, corrosion resistance, good formability
and appropriate mechanical properties [5]. Commercially pure titanium is indicated for application in the
air distribution lines, primarily for parts requiring strength up to 205 C and oxidation resistance up to
315 C [6]. A number of in-flight failures of titanium pneumatic ducts, all of them associated with cracking
adjacent to welds, have been described in technical literature [7,8]. The investigations lead to conflictant con-
clusions upon the failure causes, pointing out the embrittlement due to migration of hydrogen to the weld zone
in the presence of residual stresses [7] or the crack growth occurring mainly by fatigue from pre-existing
cracks, with hydrogen-assisted cracking playing a minor role in the process [8].

2.1. Properties of commercially pure titanium

Pure titanium has a hexagonal close-packed structure with a c/a ratio of 1.587, which is lower than the ideal
c/a ratio (1.633), and is strongly plastically anisotropic at room temperature due to texture and the wide vari-
ety of deformation mechanisms [9,10]. In hcp materials, the most common slip modes are f1 0 1 0gh1 1 2 0i,
f1 0 
1 1gh1 1 
2 0i and f0 0 0 1gh1 1 
2 0i. They constitute a total of four independent slip systems. In pure titanium,
the slip occurs most easily by the activation of dislocations with Æaæ type Burgers vector on prismatic planes,

Fig. 1. Typical engine bleed air system installation [2].

 A.M.L. Adib et al. / Engineering Failure Analysis 14 (2007) 1605–1617 1607

and to some extent on basal and pyramidal planes. Since the Æaæ slip alone cannot provide five independent slip
systems, deformation by Æc + aæ slip or by twin systems must be activated in addition to Æaæ slip in order to
maintain the compatibility of polycrystalline deformation [10].
At the macroscopic level, the deformation mechanisms associated with plastic deformations of commer-
cially pure titanium are strongly dependent on the temperature and strain rate. Deformation by slip occurs
predominantly on a f1 0  1 0g plane in a h1 1 
2 0i direction over a wide range of temperatures with secondary
slip more prevalent at higher temperatures. Besides, the cross slip of Æc + aæ dislocations from the ð1 1 0 0Þ
plane to the ð1 1 
2 2Þ or ð1 0 
1 1Þ plane becomes energetically favorable above 300 C. Deformation by twinning
has been observed at low temperatures and the density of twins increases with increasing strain rate, strain and
decreasing temperature [11,12].
These complex deformation mechanisms and the high reactivity of titanium with interstitial elements (oxy-
gen, carbon, nitrogen, and hydrogen) can be related to the aircraft structures degradation and to the mechan-
ical behavior in-service changes under different stress-temperature regimes. The affinity of titanium for oxygen
is one of the main factors that limit its application as structural material at high temperatures. The absorbed
oxygen combines chemically with titanium to form a hard and brittle oxide layer on the material’s surface.
This oxidized scale of very complex structure can be composed of a set of parallel layers, containing several
distinct oxides [13,14]. The mass increase of titanium exposed at high temperatures can be assumed to be due
to the diffusion of oxygen. Additionally, diffusion within the grain increases faster than grain boundary diffu-
sion as the temperature increases. At lower temperatures grain boundary diffusion becomes more influential
[13,15].
The effects of the interstitial elements (C, O, N) on mechanical properties of titanium can be summarized as
follows. Like most impurities in metals, they have certain virtues and certain drawbacks. Small amounts of
these interstitials offer a substantial strengthening together with little loss in tensile ductility at room temper-
ature. They increase the strain rate sensitivity, produce yield-point phenomena and the dynamic strain-aging
caused by the interaction between moving dislocation and mobile points defects. They lose their strengthening
effect at elevated temperatures, but at subzero temperature promote embrittlement and have a deleterious
effect on toughness, weld ductility, machinability [9,11,13,16]. However, the improvement in static strength
rarely results in comparable improvements in fatigue properties and is usually accompanied by loss in fracture
toughness and in greater susceptibility to environment-induced cracking [17]. In combination with hydrogen,
their detrimental effects can be magnified. In this case, hydrogen embrittlement of metals used in aircraft com-
ponents can cause serious deterioration of their mechanical properties. Especially, vulnerable are the titanium
tubular ducts which are exposed to high temperatures and pressures. This is a type of deterioration that can be
linked to corrosion as it involves the ingress of hydrogen into a component, and can seriously reduce the duc-
tility and load-bearing capacity, which causes cracking and catastrophic brittle failures [8,18,19]. Eventually,
the oxide film formed by thermal oxidation of titanium can prevent corrosion and hydrogen absorption [15].
Besides, deformation twinning plays an important role in maintaining a generalized plastic flow and also in
mechanical behavior of titanium at all temperatures. Experimentally, it is observed that deformation twinning
can effectively strengthen a material under some circumstances and weaken it in others [9,11,16]. Previous
studies suggested that the twins could play a role in the initiation of shear bands and fatigue cracks [17,20,21].

3. The failed titanium ducts

The failures occurred mainly in distribution line T-ducts made of Ti A40 (ASTM grade two equivalent)
tubes with 50 mm in diameter and 0.52 mm in thickness. According to the typical flight profile, the operating
pressure peak is 1.73 MPa and the working temperature can reach 300 C. The general appearance of the
ducts, failed after 2500–5000 flight hours, is shown in Fig. 2. Before the manufacturing of these parts, the
raw material tubes are stress-relieved at 560 C for 2.5 h and cooled in air. A straight section is cut and
cold-formed before the circumferential welding that results in the T-shaped part. An arc weld process with
non-consumable tungsten electrode and argon atmosphere protection is adopted in the production of the
ducts. The in-flight bursts resulted in the formation of one or more through-thickness cracks in the duct walls.
The cracks, with up to 30 mm in length, are positioned in two distinct sites, as seen in Fig. 2 (i) in the straight
section opposite to the T leg or (ii) following the 90 cold drawn derivation where the T leg is welded. The
1608 A.M.L. Adib et al. / Engineering Failure Analysis 14 (2007) 1605–1617

Fig. 2. General appearance of the burst ducts.

cracks did not emanate from the welds. The investigation conducted in this work is aimed to determine the
fracture mechanisms and to compare the characteristics of the failed duct material to those of the unused tubes
by means of microstructure and surface analyses, as well as tensile and fatigue tests. Hence, samples taken
from the raw material (RM), the stress-relieved material (SR) and from the failed ducts (FD) are evaluated.

3.1. Microstructure analysis

The micrographs of the titanium samples are shown in Fig. 3. In the RM condition (Fig. 3a), the analysis
revealed the presence of a large amount of deformation twins. The activation of twinning was dependent on
the local crystallographic orientation of each matrix grain. Some grains have accommodated deformation
mainly by slip. Twinning planes of the types f1 0 
1 2g, f1 1 2 1g and f1 1 2 2g are active during plastic deforma-
tion. The observed twin types have been associated with the magnitude of the twinning shear and to twinning
planes. Formation of numerous mechanical twins and intersection among these twins divide the grain interior,
resulting in microstructure refinement. Additional grain refinement occurred by the almost simultaneous for-
mation of secondary and the tertiary twins in addition to subdivision of twins due to crossing twins. The
microstructure of the SR material is shown in Fig. 3b, together with the solidification structure formed in
the melted weld zone. In this case, deformation twins are practically absent. The same microstructure is pres-
ent in the FD material (Fig. 3c), comprising single-phase, equiaxed a grains with an average grain size of
10 lm.

3.2. Internal surface examinations

All of the samples were subjected to ultrasonic cleaning with acetone and carefully dried before the SEM
(scanning electron microscopy) surface examinations. Very distinct features are present in the internal surfaces
of the ducts, as shown in Fig. 4. The RM samples have smooth surface related to the finishing obtained by the
rolling process (Fig. 4a). The stress-relieving treatment produces a blue heat tint on the surface of the tubes.
The SR material shows also an increased roughness revealing the grain structure of the material (Fig. 4b). The
failed ducts were entirely brown-colored as a consequence of the service use. Their internal surfaces are also
generally smooth, although they are full of white particle deposits, see Fig. 4c. In certain regions, intense sur-
face damage was found, including clusters of oxide particulates and delamination areas. Furthermore, surface
micro-cracks were observed in the regions near the fractures (Fig. 4d). EDS (electron diffraction spectrometry)
analyses were performed in order to identify the elements deposited in the FD internal surfaces. A polished
 A.M.L. Adib et al. / Engineering Failure Analysis 14 (2007) 1605–1617 1609

Fig. 3. Micrographs of titanium samples: (a) RM condition; (b) SR; (c) FD.

surface, from which the deposits were removed, was also examined. Fig. 5 shows the spectra obtained from the
polished surface (Fig. 5a) and from two distinct particles (Fig. 5b and c). Obviously the polished surface is
composed basically by titanium, with small amounts of Si, P, S and Fe totalizing not more than 0.15% in
weight. The interstitial C and O are also present, but it is well known that the assessment of their amounts
1610 A.M.L. Adib et al. / Engineering Failure Analysis 14 (2007) 1605–1617

Fig. 4. Internal surface profiles: (a) RM condition; (b) SR; (c) and (d) FD.

by this technique is not reliable. The deposits are composed by considerable amounts of S, Cl, K, and Ca, as
well as Si, Al, and Na, which clearly represent external contaminants that were absorbed during the service
operation.

3.3. Room temperature mechanical properties

The tensile and fatigue tests were conducted in a MTS servo-hydraulic machine at room temperature in
laboratory air. Ring-shaped specimens were taken from the tubes. In the case of the failed ducts, the samples
were cut from intact areas of the parts. A test rig designed and built especially for these tests was mounted in
fracture mechanics grips, as shown in Fig. 6a. In such a test configuration, the material is loaded in the same
direction of the highest principal stress due to the duct pressurization. The ring’s width was 10 mm and the
gripping setup provided also 10 mm as the initial gage length on each side of the specimen. Fig. 6b shows
an untested tensile specimen besides a tested one. The test-pieces fractured at one of the two gage length
regions. Only the fractured region was considered in the assessment of the elongation. At least five specimens
of each material condition (RM, SR, FD) were tested. The average results are given in Table 1. These results
 A.M.L. Adib et al. / Engineering Failure Analysis 14 (2007) 1605–1617 1611

Fig. 5. EDS spectra: (a) base material; (b) and (c) surface particles.

show that the FD material presented some increase in the tensile properties compared to the SR material, and
this was accompanied by a ductility loss. This behavior is probably due to solid solution strengthening
occurred during service conditions. On the other hand, the slightly higher tensile strength presented by the
RM condition is attributed to the presence of deformation twins.
Ring-shaped specimens were also adopted for the fatigue tests, but in this case they had a continuous radius
which gave them a hourglass profile, see Fig. 6c. The outer surfaces of the specimens, together with their edges,
were carefully ground with emery paper. The internal surface was left in its original condition. The fatigue
1612 A.M.L. Adib et al. / Engineering Failure Analysis 14 (2007) 1605–1617

Fig. 6. Mechanical tests: (a) test rig; (b) tensile specimens; (c) fatigue specimen.

tests were conducted under load control, with a stationary sinusoidal waveform and a frequency of 10 Hz. The
average results are given in Fig. 7 as S/N curves. From these preliminary results it can be inferred that the SR
material shows improved fatigue behavior compared to the RM condition. This can be attributed to the pres-
ence of twin-matrix interfaces acting as crack nucleation sites in the latter condition. It can also be observed
that the FD material shows a drop in its fatigue resistance, probably due to the accumulated surface damage.
 A.M.L. Adib et al. / Engineering Failure Analysis 14 (2007) 1605–1617 1613

Table 1
Tensile properties at room temperature
Material condition Yield stress (MPa) Tensile strength (MPa) Elongation (%) in 10 mm
Raw material 289 440 66
Stress relieved 284 430 72
Flown tube 339 461 63

320

RM
300
SR
FD
280

260
S (MPa)

240

220

200

180

160
10000 20000 50000
N (Cycles)

Fig. 7. S/N curves for the three material conditions.

4. Fractographic analyses

The fracture surfaces of the failed ducts, as well as of the mechanical test specimens, were examined in a
scanning electron microscope (SEM), operating in the secondary electrons mode.

4.1. Failed ducts

As already shown (Fig. 2), the failed ducts presented large cracks, with some amount of plastic deformation
in their vicinities. Fig. 8 is a general view of such a crack, showing the complex macroscopic fracture path with
deviations and distinct crack planes. Fig. 9a is a representative SEM fractograph showing a typical feature of
transgranular fatigue failure and subsidiary cracks associated with striations. The fatigue striations can be
clearly seen in the higher magnification Fig. 9b. Due to the working temperature, these striations are probably
related to the occurrence of cross slip. The striation spacing of about 1 lm denotes a high fatigue crack growth

Fig. 8. General appearance of the fracture surfaces.

1614 A.M.L. Adib et al. / Engineering Failure Analysis 14 (2007) 1605–1617

Fig. 9. Fracture mechanisms in the failed ducts: (a) transgranular brittle fracture; (b) fatigue striations.

rate. The striation marks did not ran parallel to striations on adjacent grains (see arrows in Fig. 9a). No evi-
dence of titanium hydrides was found in this investigation. On the other hand, the internal surface damage (see
Fig. 4d), associated to the geometry of the duct and to the solid-solution hardening increases the local stress
concentration. This fact suggest that cracks had initiated at multiple sites around the inside surface, so that
further increments of cracking occurred on subsequent cyclic pressurization during service life until the crack
became large enough to cause the sudden final fast fracture.

Fig. 10. Typical ductile fracture observed in the tensile specimens.

 A.M.L. Adib et al. / Engineering Failure Analysis 14 (2007) 1605–1617 1615

4.2. Tensile and fatigue specimens

Examination of the tensile specimens tested at room temperature revealed, in all of the material conditions
(RM, SR, and FD), high concentration of voids caused by the low work-hardening behavior of titanium,
allowing more plastic deformation to occur before a stress sufficient to cause failure was reached (Fig. 10).
The slip systems and the presence of more that one operative twinning system are the main reasons for tita-
nium to exhibit this extensive ductility [16]. The fractured surfaces of the fatigue specimens also presented the
same features for the three tested conditions. In Fig. 11, referred to a SR specimen, these features are given.

Fig. 11. Typical fatigue fracture: (a) general view; (b) crack propagation; (c) final fracture.
1616 A.M.L. Adib et al. / Engineering Failure Analysis 14 (2007) 1605–1617

The fracture surfaces usually presented two regions (Fig. 11a), corresponding to stable crack growth and final
fracture. Multiple crack initiation was found in some specimens, leading to non-propagating cracks. The
transgranular crack propagation region presents plain fatigue striations (Fig. 11b). There are not parallel-fis-
sured striation marks. The final fracture area demonstrates visible plastic deformation (Fig. 11c). The residual
forced fracture is a transgranular dimpled fracture. In spite of the smaller fatigue resistance, similar fracture
appearance was observed in the FD specimens. The fatigue mechanisms of titanium crystals have been studied
and, depending on the grain orientation, many deformation modes have been observed. The complex slip and
twinning behavior produced by the competition between these systems make it difficult to understand the fati-
gue mechanisms of this material. However, the fractographic features observed in this work are typical of fati-
gue in titanium [22,23].

5. Conclusion

The results presented in this paper show that, due to service aging, the Ti A40 used in pneumatic ducts suf-
fers solid solution hardening, associated with an increase in the tensile strength, and surface damage consisting
of delamination and oxide particle deposits, causing a decrease of the fatigue resistance. The internal surface
damage, associated to the geometry of the duct and to the solid-solution hardening increases the local stress
concentration. This fact suggest that cracks had initiated at multiple sites around the inside surface, so that
further increments of cracking occurred on subsequent cyclic pressurization during service life until the cracks
became large enough to cause the sudden final fast fracture. The striation features, related to high crack prop-
agation rates, and the internal surface damage characteristics are probably consequences of the reached in-ser-
vice temperature levels.

Acknowledgements

C. C. F. Marques and C. Haga acknowledge FAPESP for the grants. This research project supported by
FAPESP (proc. 04/02352-2) will also allow the performing of laboratory simulations of service conditions in
aircraft pneumatic systems, aiming to assess the aging process involved with various potential materials for use
in such systems.

References

[1] Findlay SJ, Harrison ND. Why aircraft fail. Mater Today 2002;11:18–25.
[2] SAE ARP 1796. Aerospace Recommended Practice: Engine Bleed Air System for Aircraft, 1987. p. 23.
[3] SAE ARP 699. Aerospace Recommended Practice: High Temperature Pneumatic Duct Systems for Aircraft, 1997. p. 103.
[4] FAR part 25. Code of Federal Regulations, Aeronautics and Space. Issued by Office of the Federal Register, National Archives and
Records Administration, Washington, DC, 2000.
[5] Kao YL, Tu GC, Huang CA, Liu TT. A study on the hardness variation of a- and b-pure titanium with different grain sizes. Mater Sci
Eng A 2005;398:93–8.
[6] SAE AMS 4941C. Aerospace Material Specification: Titanium tubing, 8P, 1984.
[7] Lynch SP, Hole B, Pasang T. Failures of welded titanium aircraft ducts. Eng Failure Anal 1995;2:257–73.
[8] Barta ER, Byer RR, Narayanan GH. Delayed hydrogen embrittlement in commercially pure titanium. In: Proceedings of symposium
on testing and failure analysis (ISTFA 88): ASM; 1988. p. 387–95.
[9] Salem AA, Kalidindi SR, Doherty RD. Strain hardening of titanium: role of deformation twinning. Acta Mater 2003;51:4225–37.
[10] Balasubramanian S, Anand L. Plasticity of initially textured hexagonal polycrystals at high homologous temperatures: application to
titanium. Acta Mater 2002;50:133–48.
[11] Chichili DR, Ramesh KT, Hemker KJ. The high-strain-rate response of alpha-titanium: experiments, deformation mechanisms and
modeling. Acta Materialia 1998;46:1025–43.
[12] Yoo MH, Agnew SR, Morris JR, Ho KM. Non-basal slip systems in HCP metals and alloys: source mechanisms. Mater Sci Eng A
2001;319–321:87–92.
[13] Bertini YA, Gacougnolie JL, Parisot J, Fouquet J, Beshers D. Mechanism of stress induced oxygen diffusion in titanium. In: Kimura
H, Izumi O, editors. Titanium science and technology. Metall soc, vol. 1. Kyoto: AIME; 1980. p. 529–34.
[14] Pitt F, Ramulu M. Influence of grain size and microstructure on oxidation rates in titanium alloy Ti–6Al–4V under superplastic
forming conditions. J Mater Eng Perform 2004;13:727–34.
[15] Fukuzuka T, Shimogori K, Satoh H, Kamikubo F. On the beneficial effect of the titanium oxide film formed by thermal oxidation. In:
Titanium science and technology. Metall Soc, vol. 3. Kyoto: AIME; 1980. p. 2783–92.
 A.M.L. Adib et al. / Engineering Failure Analysis 14 (2007) 1605–1617 1617

[16] Nemat-Nasser S, Guo WG, Cheng JY. Mechanical properties and deformation mechanisms of a commercially pure titanium. Acta
Mater 1999;47:3705–20.
[17] Wanhill RJH. Ambient temperature crack growth in titanium alloys and its significance for aircraft structures. Aeronaut J
1977:68–82.
[18] Hack JE, Leverant GR. The influence of microstructure on the susceptibility of titanium alloys to internal hydrogen embrittlement.
Metal Trans A 1982;13:1729–38.
[19] Hardie D, Ouyang S. Effect of hydrogen and strain rate upon the ductility of mill-annealed Ti6Al4V. Corros Sci 1999;41:155–77.
[20] Chen CQ, Li SX. Tensile and low-cycle fatigue behaviors of commercially pure titanium containing c hydrides. Mater Sci Eng A
2004;387–389:470–5.
[21] Xiaoli T, Haicheng G. Fatigue crack initiation in high-purity titanium crystals. Int J Fatigue 1996;18:329–33.
[22] Brooks CR, Choudhury A. Metallurgical failure analysis. Boston: McGraw Hill; 1993.
[23] Azevedo CRF. Failure analysis of a commercially pure titanium plate for osteosynthesis. Eng Failure Anal 2003;10:153–64.