Engineering Failure Analysis 13 (2006) 516–525

www.elsevier.com/locate/engfailanal

Corrosion fatigue phenomena learned from failure analysis

R. Ebara *

Department of Advanced Materials Science, Kagawa University, 2217-20 Hayashi-cho, Takamatsu, Kagawa 761-0396, Japan

Received 12 July 2004; accepted 11 December 2004

Available online 1 April 2005

Abstract

The purpose of this paper is to present corrosion fatigue phenomena learned from failure analysis. The emphasis is
focused upon the importance of the role of corrosion pit in the corrosion fatigue crack initiation process and intergran-
ular fracture surface in the corrosion fatigue crack propagation process of steam turbine blade. The hydrogen effect is
also stressed in the crack propagation process in the corrosion fatigue crack propagation of the cargo oil tank ship
structural members for very large crude oil carrier. Finally the future problems should be solved in corrosion fatigue
failure are touched in brief.

2005 Elsevier Ltd. All rights reserved.

Keywords: Failure analysis; Corrosion fatigue; Steam turbine blade; Ship hull structure

1. Introduction

Corrosion fatigue has been studied for more than eighty years since McAdam [1] and Gough and Sop-
with [2]. The valuable conference proceedings and book [3–6] related to corrosion fatigue engineering have
been published so far. However we have still corrosion fatigue failure of components in machines and struc-
tures. This is due to an increase use of the higher strength materials, severer environments and an applica-
tion of limit design. Among corrosion fatigue failure analyses experienced by author and his colleagues
corrosion fatigue of steam turbine blade for compressor drive turbine and corrosion fatigue of cargo oil
tank structural members for very large crude oil carrier are demonstrated.
Characteristics of corrosion fatigue failure were exemplified for steam turbine blade and cargo oil tank
members. Then corrosion fatigue mechanism, influencing variables and countermeasures are briefly sum-
marized on 12 Cr stainless steel for steam turbine blade and high strength steel for ship hull structure
mainly from the experimental results obtained by author and his colleagues.

*
Tel.: +81 87 864 2391; fax: +81 87 864 2438.
E-mail address: [email protected].

1350-6307/$ - see front matter
2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfailanal.2004.12.024
 R. Ebara / Engineering Failure Analysis 13 (2006) 516–525 517

2. Corrosion fatigue failure of steam turbine blade

Corrosion fatigue failure of steam turbine blades has been a big concern in the world for these more than
30 years. Fatigue failure of steam turbine moving blade due to resonance can be easily identified on fracture
surface from beach mark observed by naked eye. Intergranular fracture identified by SEM observation and
fatigue failure in lower stress than design stress raised the question for corrosion fatigue failure of the steam
turbine blade.
Corrosion fatigue failure of steam turbine blade is characterized as follows.

(1) Corrosion pits are observed at initiation area.

(2) Plural microcracks in association with corrosion pits are observed at corrosion fatigue crack initiation
area as shown in Fig. 1 [7].
(3) Intergranular fracture surfaces are observed at corrosion fatigue crack propagation area as shown in
Fig. 2 [7].

Steam environment is not consisted of pure steam alone and contains harmful impurities. Impurities are
oxidized scale from superheater, dissolution from the vapor contained in the boiler feedwater, or dissolu-
tion in the mist accompanying saturated steam [7]. More than eighty kinds of chemical compounds such as
oxide, silicate and sulfate are identified on deposits found on blades of turbine by X-ray diffraction method
[8]. Most compounds are washed away by water in the wet stage of the turbine. In the wet-dry alternating
stage impurities concentrate at irregular parts of the turbine. Nowadays, it is well recognized that corrosion
fatigue failure of the steam turbine blade occurs in a repeated stress area if impurities should concentrate in
a wet-dry alternating stage of the low pressure steam turbine. However the mechanism for impurities con-
centration has not yet fully understood. The most deleterious impurities, which reduce fatigue strength of
turbine blades, are NaCl and NaOH. The number of cycles is reached to 1010–1012 for design life of 20 years
[9]. Therefore many corrosion fatigue tests of 12 Cr stainless steel for steam turbine blade have been con-
ducted in various content of NaCl and NaOH aqueous solution.
In the results, it was clarified that corrosion fatigue crack initiated from corrosion pit at corrosion fati-
gue crack initiation area and propagated with intergranular fracture. The stage of crack initiation from cor-
rosion pit and the appearance of crack propagation from corrosion pit can be clearly identified in Fig. 3.
Crack branching and crack blunting at the corrosion fatigue crack tip, which show characteristics in

Fig. 1. Surface crack with corrosion pit on failed turbine blade [7].
518 R. Ebara / Engineering Failure Analysis 13 (2006) 516–525

Fig. 2. Intergranular fracture surface observed on the failed turbine [7]. Arrow shows crack propagation direction.

corrosion fatigue process, can be clearly identified from this figure. The maximum depth, diameter and frac-
tion of area of corrosion pit increases with increasing number of cycles. The maximum depth of corrosion
pit at crack initiation stage is 23.5 lm in plane bending stress of 228 MPa (N/Nf = 0.968) and 17.8 lm in
plane bending fatigue stress of 280 MPa (N/Nf = 0.645) (Fig. 4) [10]. The above observations revealed that
crack initiated at the bottom of corrosion pit where stress concentration is large and is presumably electro-
chemically active.
Intergranular fracture was predominantly observed at crack propagation area. The fraction of intergran-
ular fracture was small near crack initiation area. It increased and then decreased with Kmax. The maximum
fraction of intergranular fracture can be observed at Kmax = 16MPa m1/2 as shown in Fig. 5 [10].
Fig. 6 shows the conventional S–N diagrams of plane bar specimens for 12 Cr stainless steel with testing
speed of 60 Hz in 3–3 · 10
4% NaCl aqueous solution [11,12]. It can be clearly observed from this diagram
that the fatigue strength of 12 Cr stainless steel remarkably decrease in 3% NaCl aqueous solution. The
fatigue strength decrease due to 3% NaCl aqueous solution at 107 cycles is approximately 75%. The NaCl
content dependency can be also observed on fatigue strength. The number of cycles in 3 · 10
2% NaCl
aqueous solution with same reduction rate of fatigue strength move to more longer life of 109 cycles. From
these findings it can be mentioned that corrosion fatigue strength of the turbine blade decreases even in a
mild steam environment if very small quantity of Cl
concentrate corrosion fatigue occurs at the irregular
area of the turbine blade.
It was also clarified that in 40% NaOH aqueous solution the same reduction of fatigue strength with
3%NaCl aqueous solution was observed for 12 Cr stainless steel [7].
Fig. 7 shows the ultrasonic corrosion fatigue testing results of 12 Cr stainless steel in various content of
NaOH aqueous solution [13]. The corrosion fatigue strength of 12 Cr stainless steel decreases with increase
of NaOH content and depends on NaOH content In these S–N curves the inflection point can be observed
at the higher number of cycles on S–N curves in 5% and 20% NaOH aqueous solution. Moreover it was
clearly observed that the fine corrosion fatigue cracks propagated from the bottom of the blunted corrosion
fatigue cracks on the unbroken specimens in 5% NaOH aqueous solution. From these findings it can be
mentioned that the corrosion fatigue strength decreases even in a dilute NaOH aqueous solution if an
aggressive environment like 40% NaOH aqueous solution is formed in a bottom of fatigue crack. The
mechanism of the change of the impurities contents from bulk solution to crack solution is necessarily
to clarify.
 R. Ebara / Engineering Failure Analysis 13 (2006) 516–525 519

Fig. 3. Corrosion fatigue crack initiation and propagation process in association with corrosion pit, 228 MPa, 3% NaCl aqueous
solution (318 K) [10]: (a) 4.95 · 106, (b) 5.45 · 106, (c) 5.48 · 106, (d) 5.485 · 106, (e) 5.495 · 106, (f) 5.515 · 106 and (g) 5.55 · 106 cycles.
520 R. Ebara / Engineering Failure Analysis 13 (2006) 516–525

Fig. 4. Maximum depth of corrosion pit [10].

It has been reported that an inflection point appeared on the long life range of S–N curves of carbon steel
[14,15] and high strength steel [16] in corrosive environments.
The dissolution of the specimen due to corrosion and impurities concentration at the crack bottom are
the most possible reasons of this phenomena.

Fig. 5. Fraction of intergranular fracture surface vs. Kmax [10]. Aerated 3% NaCl aqueous solution, 353 K.
 R. Ebara / Engineering Failure Analysis 13 (2006) 516–525 521

Fig. 6. S–N curves of 12 Cr stainless steel in various content of NaCl aqueous solution [11,12], 60 Hz, R(rmin./rmax.) =
1.

Fig. 7. S–N curves of 12 Cr stainless steel in various content of NaOH aqueous solution [13], 14.9 kHz, R =
1.

Environmental variables influencing corrosion fatigue strength of 12 Cr stainless steel are dissolved oxy-
gen, pH and temperature. Metallurgical variables are microstructure induced by tempering temperature
Mo content and surface roughness. Mechanical variables are stress concentration factor, mean stress. A
couple of methods to improve corrosion fatigue strength of 12 Cr stainless steel have been developed
[17]. The most effective method to prevent corrosion fatigue failure for steam turbine blade is the reduction
of the design stress.

3. Corrosion fatigue failure of cargo oil tank ship structural member

Fig. 8 shows the fracture surface of the side longitudinal member in VLCC cargo oil tank [18]. It can be
judged from the macroscopic features that the cause of this crack can be fatigue. The fatigue crack prop-
agation area generally consisted of relatively smooth fracture surface. However, the extremely up and down
fracture surface can be observed on fracture surface in web plate. Unfortunately, the microfractographical
characteristics cannot be identified on this fracture surface because of the heavy scaled on fracture surfaces.
However the suspicion of the effect of H2S in crude oil on this type of fracture surface was felt from the
failure analysis. Therefore corrosion fatigue tests for ship hull structural steels were conducted in sour crude
oil environment.
522 R. Ebara / Engineering Failure Analysis 13 (2006) 516–525

Fig. 8. Macroscopic fracture surface of the face plate [18].

Fig. 9. The da/dN
DK and s
DK curves for HT50(TMCP) base metal in sour crude oil [19].
 R. Ebara / Engineering Failure Analysis 13 (2006) 516–525 523

Fig. 9 shows the da/dN
DK curves for the CT specimens of the TMCP manufactured high strength
steel, HT50(TMCP) obtained in sour pffiffiffiffi crude oil containing 400 ppm H2S and in air [19]. In the region in
which DK approximately 16 MPa m and above, the da/dN in the sour crude oil is accelerated as compared
with that in air. The acceleration becomes conspicuouspwith ffiffiffiffi the increase in DK. The acceleration rate is
about seven
pffiffiffiffi times when DK is approximately 30 MPa m and it reaches about fifteen times when DK is
50 MPa m. The acceleration of the da/dN in sour crude oil was also observed for the weldments [20]
and mild steel.
The acceleration of the da/dN in the sour crude oil environment was first reported by Vosikovsky [21].
He found that the da/dN for X65 line pipe steel was significantly accelerated in both the intermediate and
high DK region in the sour crude oil containing 350 ppm H2S as compared with that in air. He also found
that the da/dN was strongly dependent on the H2S content in the sour crude oil [22].
Fracture surfaces in the sour crude oil were cleavage fracture where the fatigue crack propagation rate
was accelerated. In the region where the da/dN was not accelerated, the ductile striation was predominantly
observed. It was also revealed that the cleavage fracture surfaces were dominated by brittle striation. Judg-
ing from the shape of the etch pit found on cleavage fracture surface where the da/dN was accelerated in the
sour crude oil, the corrosion fatigue fracture surface can be clearly identified as (1 0 0) cleavage (Fig. 10)
[23].
The relation between the measured brittle striation spacing per cycle, S and the stress intensity factor
range, DK is also shown for the HT50(TMCP) in Fig. 9. The da/dN
DK is in well coincident with S
DK.
This means that the acceleration of fatigue crack propagation in the sour crude oil environment is predom-
inantly characterized by the brittle striation. The embrittlement mechanism of steels by external gaseous
environments has already been proposed by Wei [24]. It is assumed that in relatively high DK region molec-
ular hydrogen produced through the reaction of steels with H2S and H2O in the sour crude oil dissociates to
atomic hydrogen which enter the plastic zone of the fatigue crack tip and accumulates there in large quan-
tities, causing the plastic zone to turn into the hydrogen embrittlement zone and thus resulting in the accel-
eration of fatigue crack propagation rate (Fig. 11). The reason that the higher DK value, the more the crack
propagation rate accelerated is probably because the crack tip plastic zone is large permitting the corre-
spondingly large quantities of atomic hydrogen to accumulate there and hence causing the crack tip zone
of hydrogen embrittlement to enlarge.
Environmental variables influencing corrosion fatigue crack propagation of ship hull structural steels in
sour crude oil environment are hydrogen sulfide content and water.

Fig. 10. Etch pits observed on cleavage fracture surface in acceleration with brittle striation [23] HT50(TMCP), DK (MPa m1/2): 34.7.
Arrows shows crack propagation direction.
524 R. Ebara / Engineering Failure Analysis 13 (2006) 516–525

Fig. 11. Schematic illustration of corrosion fatigue crack propagation mechanism in sour crude oil environment [19]. Note: *1)SCO,
Sour Crude Oil; *2)HEZ, Hydrogen Embrittlement Zone; *3)PZ, Plastic Zone; *4)A, da/dN in SCO is accelerated; *5)B, da/dN in SCO is
not accelerated.

Development of corrosion fatigue resistant steels is desired. To prevent corrosion fatigue failure for the
cargo oil tank ship structural members reduction of the design stress of the component is the most effective.

4. Concluding remarks

This paper has briefly summarized corrosion fatigue phenomena learned from failure analysis for steam
turbine blade and cargo oil tank ship structural member. The important role of corrosion pit and intergran-
ular fracture were stressed in corrosion fatigue failure for steam turbine blade. While the emphasis is fo-
cused upon the hydrogen related corrosion fatigue behavior for cargo oil ship structural members. It
can be concluded that the valuable findings of corrosion fatigue phenomena have been obtained from fail-
ure analysis.
The following works are recommended for corrosion fatigue of engineering materials.

(1) Clarification of corrosion fatigue crack initiation mechanism and quantitative analysis of corrosion
fatigue crack initiation process.
(2) Accumulation of long term corrosion fatigue strength.
(3) Development of accelerated corrosion fatigue testing method.
(4) Development of the effective improvement methods to prevent corrosion fatigue failure.

References

[1] McAdam Jr DJ. Proc ASTM 1926;26:224–54.

[2] Gough HJ, Sopwith DG. J Inst Metals 1932;49:93–112.
[3] Devereux OF, McEvily AJ, Staehle R, editors. Corrosion fatigue, chemistry, mechanics and engineering. NACE; 1972.
[4] Craig Jr HL, Crooker TW, Hoeppner DW, editors. Corrosion fatigue technology. ASTM STP, 642. ASTM; 1978.
[5] Crooker TW, Leis BN. Corrosion fatigue, mechanics, metallurgy, electrochemistry and engineering, ASTP801; 1983.
 R. Ebara / Engineering Failure Analysis 13 (2006) 516–525 525

[6] Endo K, Komai K. Met Corr Fatig Strength Des 1982 [in Japanese].
[7] Ebara R, Kai T, Mihara M, Kino H, Katayama K, Shiota K. Corrosion fatigue behavior of 13Cr stainless steel for turbine
moving blades. In: Jaffee RI editor. Corrosion fatigue of steam turbine blade materials. New York: Pergamon Press; 1983. p. 4-
150–4-167.
[8] Jonas O, Bias LG. Analyzing steam-turbine deposits. Power 1976;120:26–9.
[9] EPRI. Corrosion fatigue of steam turbine-blading alloys in operational environments. CS-2932; 1984.
[10] Ebara R, Yamada T, Kawano H. ISIJ Int 1990;30:535–9.
[11] Ebara R, Kai T, Inoue K. Corrosion-fatigue behavior of 13Cr stainless steel in sodium-chloride aqueous solution and steam
environment.. In: Craig Jr HL, Crooker TW, Hoepnner DW, editors. ASTM STP, 642. ASTM; 1978. p. 155–66.
[12] Ebara R, Kai T, Inoue K, Masumoto I. Fractographic analysis of corrosion fatigue cracking of 13Cr stainless steel in NaCl
aqueous solution 1978;27:64–8.
[13] Ebara R, Yamada Y. Availability of ultrasonic fatigue testing for evaluation of corrosion fatigue behavior of structural materials.
In: Wu XR, Wang ZG, editors. FatigueÕ99, 4/4. p. 2323–8.
[14] Iwamoto K. On the S–N curves of carbon steel under rotary bending condition in city water. Trans JSME 1964;30:500–3.
[15] Endo K, Komai K, Kinoshita S. Long life corrosion fatigue strength under continuous and alternate immersion in 1% NaCl. J
Mater Sci Jpn 1976;25:894–8.
[16] Hirakawa K, Kitaura I. Corrosion fatigue of steel in sea water. Sumitomo Search 1981;26:136–51.
[17] Yamanouchi H, Kino H, Ebara R. 82-JPGC-Pwr-18, ASME, p. 1–6.
[18] Watanabe E. PhD Dissertation. Hiroshima University; 1998.
[19] Ebara R, Yamada Y, Fushimi A, Sakai D, Watanabe E, Yajima H. J Soc Naval Achitect Jpn 1993;173:337–47.
[20] Nakano Y, Matsumoto S, Sugie E, Yajima H, Ebara R, Watanabe E. J Soc Naval Architect Jpn 1993;174:571–8.
[21] Vosikovsky O. Corrosion 1976;32:472–5.
[22] Vosikovsky O, Rivard A. Corrosion 1982;38:19–22.
[23] Ebara R, Yamada Y, Sakai D, Fushimi A, Yajima H. J Soc Mater Sci Jpn, No. 488, 1994;43:580–586.
[24] Wei RP. In: Fong JT, editor. Fatigue mechanisms. ASTM STP675; 1979. p. 816–31.