Вы находитесь на странице: 1из 26

28.

Hydrogen-Induced Blister Cracking of Linepipe Steel

Mako lino
Central R&D Bureau
Nippon Steel Corporation
Tokyo, Japan

TABLE OF CONTENTS
1. Introduction
2. Industrial Sour Gas Environments and Estimated Hydrogen
Fugacities
3. Mechanism of HIBC
3.1 Trapping of Hydrogen By Sulfur-Associated Defects in
Linepipe Steel
3.2 Influence of External Stress Upon HIBC Extension
4. Current Status of Development of HIBC Resistant Steel
4.1 Control of Corrosion and Hydrogen Entry by Alloying
4.2 Estimation of Hydrogen Concentration Immediately Below
Surface in Early Stage of Exposure - Influence of
Alloying Element
4.3 Influence of Alloying Upon Kinetics of Corrosion of
Hydrogen Evolution Type
4.4 Control of Non-Metallic Inclusions
4.5 Improvement of Through-Thickness-Directional Fracture
Toughness by Heat Treatment
5. New Sour Gas Linepipes
6. Conclusion
7. References
APPENDIX A. The Meaning of C$5
Absorption of Hydrogen into a Plate Containing Traps
Liberation of Hydrogen from a Plate Containing Traps
APPENDIX B. log/t J(t) - 1/t Relation in Permeation Transient
for Changing Boundary Condition
1. INTRODUCTION
Phenomenologically it has been established that hard
steels, typically steels with Rockwell hardness Hj^ above 22 in
industrial application (1), are susceptible to hydrogen-induced
cracking. Qualitatively this phenomenon is to be explained as
follows: Hard steel, in itself, means steel hardened by intro-
duction of defects by alloy additions, heat treatment, cold
working and so forth; the steel then becomes susceptible to
aggression by hydrogen which has entered steel by any means.
Contrariwise, soft steels are resistant to hydrogen-induced
embrittlement, in part because of sufficiently low defect
density. This implies that a defect-rich region that could act
as a crack-embryo might be stabilized by being surrounded by a
sufficiently soft and crack-resistant matrix. However, soft
steels suffer from hydrogen-induced blistering, external or
internal, when exposed to hydrogenating environments of suffi-
ciently high hydrogen fugacity (see Table 1, which classifies
hydrogen-induced degradation phenomena (2)). The internal
blisters interact with each other to cause cracking in between
or at the periphery of the blisters (3). This phenomenon is
well known today and is called hydrogen-induced blister cracking,
HIBC. "Blistering" proceeds parallel to the plate surface
independently of the process of relaxing externally applied
stress and is naturally strongly influenced by the hot-rolling
structure, namely parallel-to-surface defects caused by the hot
rolling in the plate manufacturing process (Table 2) (3).
Hydrogen-induced cracking at the weld can result from
possible or accidental formation of hardened regions at or
around the weld. It has been reported, however, that submerged-
arc welds and their heat-affected zones are not weak points with
respect to HIBC because of usual absence of regions with HRQ
above 22 for linepipe steel of grade up to X-65 (4). Therefore,
only the problem of HIBC of linepipe steels is treated in the
present chapter.

2. INDUSTRIAL SOUR GAS ENVIRONMENTS AND ESTIMATED HYDROGEN


FUGACITIES
The HIBC of industrial linepipe steels referred to in
Section 1 is caused by hydrogen which results from corrosion,
the entrance of which into the steel is enhanced by H2S.
Hydrogen sulfide is one of the most effective reagents in pro-
moting hydrogen entry at a steel surface, although the exact
mechanism of promotion of hydrogen entry remains to be establish-
ed today (5,6).
Because the best steel matrix must resist the hydrogen
pressure built up in HIBC embryo which corresponds to the hydro-
gen fugacity of the corrosive environment, we will have to know
as exactly as possible the aggressiveness or hydrogen fugacity
of the environment for success in developing resistant steels in
a rational manner.
Figure 28-1 shows how CH' is influenced by the partial
pressures of hydrogen sulfide and carbon dioxide which are
present in industrial sour gas linepipe environments and which
promote hydrogen entry when dissolved in water. Here, C&5 is
the volume of hydrogen absorbed in a steel specimen during 96 h
exposure to brine equilibrated with ^S and C02 at the pressures
Table 1
Typical Usual Source Typical Failure
Materials of Hydrogen Conditions Initiation Mechanisms
1:L 3
Hydrogen Steels, Ni-base Gaseous \\2 p=10~ ~10 atm Unknown I n c u - Unknown
Environment alloys, titanium T = - 1 0 0 ~ 70O 0 C bation Deriod
Embrittlement alloys not observed
(HEE)
-Hydrogen Carbon and low Thermal pro- XH=O . 1 ~ lOppm I n t e r n a l Crack Internal
Stress Crack- alloy s t e e l s cessing e l e c - T = - 1 0 0 ~ 10O 0 C initiation diffusion
ing (HSC) trolysis , Incubation to s t r e s s
corrosion periods observ- concentra-
ed tion
-Loss in Ten- S t e e l s , N i - b a s e Gaseous H ? , X H = O . ! ~ lOppm Surface and/or Unknown
sile ductili- alloys, Be-Cu Internal R T=-100 ~ 70O0C internal effect
ty bronze from e l e c t r o -
chemical charg-
ing
-Hydrogen Carbon and Gaseous H2 UD to p=10 3 atm Surface ( d e c a r - C diffu-
Attack low alloy T = 2 0 0 ~ 60O 0 C burization) ; sion de-
steels i n t e r n a l carbide carburiza-
interfaces tion) H
(methane bubble diffusion;
formation) nucleation
and growth
(bubble
formation)
-Blistering Steels H2$ corrosion OH eauivalent I n t e r n a l defect H diffu-
Electrolytic to 2 . 1 0 2 ~ 103 sion ;
charging atm nucleation
T = O - 15O 0 C and growth
of bubble
-Shatter Steels ( f o r g i n g s W a t e r vapor Precipitation Internal defect H diffu-
Cracks , for t u r b i n e reacting with of dissolved sion to
Flakes, Fish- rotors) molten steel H during ingot voids
eyes cooling
Table 1 C o n t ' d .

Typical Usual Source Typical Failure


Materials of Hydrogen Conditions Initiation Mechanisms
3 3
Microper- Steels Gaseous F^ p=2.10 ~8.10 Unknown Unknown
f oration (compressors) atm
T=20~100°C
Degradation Iron, steels, Gaseous or XTT=l~10pr)m AdsorDtion
H
in Flow nickel base internal (Fe,20°Cl to disloca-
Properties alloys hydrogen Up to p=10 3 tions ,
atm solid solu-
(T>^M.P.) tion
Metal V, N b , Ta, T i , Zr Internal H au>solubility Internal Hydride
Hydride from m e l t ; near 20°C defect preciDita-
Formation corrosion ; p = 1 - 103 tion
electrolytic atm
charging

*HD = Hydrogen D e g r a d a t i o n , HE = Hydrogen E m b r i t t l e m e n t


p = gaseous hydrogen p r e s s u r e , a^ = hydrogen a c t i v i t y , x^ = total hydrogen c o n t e n t , T =
temperature
TABLE II: Morphological Features of Hydrogen-Induced Cracking
and Blistering

External appled stress O (J=O <7~0.7<7y


(an ^ <b)
Low strength steel 0 -«=». blistering
— <^ —
e.g. line pipe steel ^ — ^ .
mild steel used S > 0.01% O __ «» ^"
in chemical plant (32) 0^ ' *»
^ internal —- ^
<==> ~*
«0° blistering
^. (Ci) IGHIC
S < 0.01%
High strength steel |ow suscept,bil.ty — —
e.g. OCTG to blistering O— (02) TGHIC — O
weldment of .
high. ten. steels _^

Features of morphology
(at) Extensive blistering takes place in soft steels with sulfur content S>0.01%.
(32) Small scale but stepwise formed internal blistering takes place in soft steels
with sulfur content S ^0.01%.
(b) Internal blister-array forms perpendicular to external stress axis (3)
Final failure takes place by cracking region between internal blisters.
(Ci & C2) Cracking takes place at carbide-ferrite interface etc or at crystal grain boundary
in very hard steels.
IGHIC =intergranular hydrogen-induced cracking
TGHIC =transgranular hydrogen-induced cracking
(a & b) Blistering takes place to relax internal gas pressure PH, or shear stress field
caused by PH,.

(C) Cracking takes place to relax stress field caused by external applied stress (J
or to lower mechanical free energy.
Crack will form perpendicular to £7-axis, because this is the most effective way
of the stress relaxation.
shown in the figure.* The hydrogen fugacity of a typical high
H2S-pressurized wet environment may be estimated (7,8) to be
approximately 3000 atm assuming that lattice-dissolved hydrogen
concentration, C L , is related to C^5 as Cj|5 = (D/D*)Cj * 70 C L ,
where a typical apparent diffusivity, D*, of hydrogen in line-
pipe steel of 10-6cm2/s, and a value of 7 x 10~5cmz/s for true
diffusivity, D, of hydrogen in iron (9) are used. An HIBC
embryo of size 2a in a steel with KJP of ~40 kg mm" 3 / 2 would not
grow beyond ^l mm, where KJQ is the local fracture toughness
around the HIBC embryo.

*CH is measured as the volume of hydrogen liberated from the


specimen during 72 h at 45 C, after exposure to the H 2 S-CO 2
saturated brine; the C^5 can be regarded as a good measure of
the volume of hydrogen absorbed in steel during tests, as a
calculation given in APPENDIX A shows. See APPENDIX A also for
5
the relation between CrV
H and CLT.
Figure 28-1: Influence upon C^5 of partial pressures of H2S
and CO2.

It should be added that the hydrogen fugacity of actual


sour gas pipeline environments, which may be influenced by other
factors such as the transmission gas velocity to develop values
different from those produced in laboratory tests, remains a
great concern to researchers in this field.

3. MECHANISM OF HIBC
Hydrogen which has entered steel causes various detrimental
effects, the phenomenological classification of which is given
in Table 1. Hydrogen-induced degradation in linepipe steels
generally takes a form of HIBC, which is induced by'hydrogen
trapped at interfaces between matrix and, typically, hot roll-
flattened manganese sulfide inclusions. The trapped hydrogen
atoms combine to build-up molecular gas pressure at easily
separable interfaces.
3.1 Trapping of Hydrogen by Sulfur-Associated Defects in Line-
Pipe Steel
Manganese sulfide inclusions elongated in the hot-rolling
process in sample preparation are found, by metallographic
examination of the cross section of specimens, at the number
density corresponding to the sulfur content in the steel. In
order to examine the behavior of hydrogen trapping by sulfur-
associated defects in linepipe steel permeation experiments were
carried out by the present author by an electrochemical method on
steel sheets, 0.1 cm thick, with various sulfur contents (10,15)
including very high contents. The experimental results are
given in Figure 28-2 (10). Hydrogenation was performed electro-
lytically in an alkaline electrolyte (O.IN NaOH with 2.7 x 10"5N
NaAsO2) to avoid electrochemical interaction between the electro-
lyte and the sulfide inclusions appearing on the surface which
would take place if an acid solution is used (11) . As seen
from Figure 28-2, the time of rise in permeation rate increased
and the maximum attainable permeation rate decreased with
increase in sulfur content in steel.

a=0.1cm
Stee! S
A 0.0008 wt pet
B 0.008
Permeation Current * (mA)

C 0.08
D 0.08*
* treated with
rare earth metals
(0.08 wt pet)
t permeation 2
area 3.8cm

Time (min)

Figure 28-2: Hydrogen permeation transients for steels with


various sulfur contents.

Hydrogen can be captured at or around lattice-dissolved


sulfur and at interfaces between sulfides and the ferrite
matrix. The above-mentioned result suggests that the steel with
a higher sulfur content contains more abundant reversible as
well as irreversible trap sites. A diffusion analysis made for
a system with both reversible and irreversible traps (12)
predicts the observed behavior.

The hydrogen captured in irreversible traps and immobilized


at ambient temperature would be released upon heating the speci-
men holding the captured hydrogen. The behavior during the
release would provide information on the trap depth. After
removal of hydrogen at 25 C the volume of hydrogen subsequently
released at various temperatures up to 1000 C is presented in
Figure 28-3. The cylindrical specimen 0.5 cm in diameter and
5 cm long is hydrogenated electrolytically for 24 h under the
same conditions as for the case of Figure 28-2. The volume of
hydrogen released every 5 min was measured by gas chromatography,
using argon as carrier gas flowing through an infrared beam
furnace for heating the specimen at a rate of l°C/min up to
hydrogenation time

Steel C (S-0.073wt pet)

Temperature (0C)
Figure 28-5: Liberation of Hydrogen at High Temperatures

1000 0C. The accuracy of measurement of hydrogen volume was


10~4 cm^. An amount equal to about ten percent of the hydrogen
released at 25°C, Cfl , is released at temperatures around 80 to
110 0C and 370 to 420°C as illustrated in Figure 28-3; the
hydrogen must have been captured at traps so deep that the
trapping was effectively irreversible at 25°C. These traps
should have affected the hydrogen permeation, as depicted in
Figure 28-2.
3.2 Influence of External Stress upon HIBC Extension
An externally applied stress influences considerably the
manner of HIBC extension; in the absence of external stress HIBC
develops by stepwise linking of internal blisters along with
considerable plastic deformation as shown in Figure 28-4. Under
externally applied stress the HIBC develops by linking blisters
formed in stacked arrays perpendicular to the applied stress as
shown in Figure 28-5. Linking of blisters occurs with the aid
of hydrogen-induced cracks formed in deformation bands connect-
ing blister tips (3).

The shear stress distribution induced around the blister


is sensitively influenced by external stress applied parallel
to the blister plane. Stress analysis shows that a pronounced
shear stress peak produced by internal hydrogen pressure at a
crack tip is reduced by external stress applied parallel to the
crack plane (13). This shear stress is considered to play an
important role in nucleating secondary blister cracks by acting
on manganese sulfide-matrix interfaces to trigger opening and to
accelerate hydrogen precipitation there. Positions characteriz-
ed by high enough shear stress and high tensile stress,
(together with sufficiently low compressive stress in the plate
thickness direction) and which are at planes of weak bonds,
such as at flattened manganese sulfide-matrix interfaces, are
expected to become the nucleation sites. Such positions around
a first-formed blister crack lie at the trajectory of a stepwise
crack extension in the absence of external stress, explaining
ssau^omi
9ieid Plate Width

Figure 28-4: HIBC extension in specimen free from external


stress.
Stress

Plate Thickness

Figure 28-5: HIBC extension under stress.

the blister array in Figure 28-4. Under external stress


however, such are disposed alongside and at a fair distance
apart from both tips and center of the first-formed blister
crack. This is consistent with blister array shown in Figure
28-5.

This difference in the manners of HIBC extension with and


without externally applied stress could lead to different
evaluations of the material, as discussed elsewhere (3).
4. CURRENT STATUS OF DEVELOPMENT OF HIBC-RESISTANT STEEL
Various studies have been performed concerning factors
influencing the HIBC resistance of steel in wet environments
containing hydrogen sulfide. These factors include; alloying
effects in steel corrosion and therefore hydrogen evolution,
non-metallic inclusions as blister nuclei, segregation of phos-
phorus and other elements accelerating blister crack extension,
and metallographic structure and second phases. The influence
of each factor is discussed individually first, and then to-
gether, showing examples of industrially developed HIBC-resistant
linepipe steels for use in aggressive sour gas environments.
4.1 Control of Corrosion and Hydrogen Entry by Alloying
Figures 28-6 to 8 show corrosion and hydrogen absortpion
behavior in H2S-saturated brine of pH 5.2, as influenced by
alloying by small amounts. The CJj^ factor appearing in these
figures is the hydrogen volume defined in Section 2. The C^ is
interpreted to be composed of hydrogen upon normal sites, CL,
and hydrogen reversibly trapped in abnormal sites,
(N/NL)CLexp(Et/kBT),

C S
H = V1 + Sr e*p(id^
L D
CD
which is shown to be approximately valid by the diffusion
analysis given in APPENDIX A. Thus C^ successfully compares
CT among specimens with very similar trap characteristics
(N , Et).
The results given in Figures 28-6 to 8 are summarized as,
(a) Alloying with small amounts of Ni or other Group VIII ele-
ments in the periodic table (0.08 wt pet Pd or Pt, 0.02 wt pet
Co or Rh) remarkably reduces hydrogen absorption into steel.
Corrosion rate (mdd)

C4H5 (cm3/10Og)

Alloy content (wt pet) Alloy content (wt pet)


Figure 28-6: Corrosion rate and CH as influenced by alloying
pH 5.2 (1).

0.08 wt pet Bi also decreases hydrogen absorption conspicuously.


(b) A similar effect is caused by alloying with Group Ib ele-
ments like Au and Cu.
(c) In Group Via Cr causes a considerable effect.

The effects of alloying elements are mainly attributable


(PPLU) 9JBJ UO(SOJJOO

C4H5 (cm3/10Og)
Cr (wt pet) Cr (wt pet)

Figure 28-7: Corrosion rate and C^ as influenced by alloying


pH 5.2 (2).
Corrosion rate (mdd)

CS5 (cm 3 /10Og)

Alloy content (wt pet) Alloy content (wt pet)

Figure 28-8: Corrosion rate and Cu


H
as influenced by alloying
pH 5.2 (3).

to protective films of corrosion products formed on steel


surface which control further progress of corrosion and of
hydrogen evolution. When the pH of the corrosive environment is
lowered to around 4, Cu and Cr tend to lose their effects; only
Ni and some other Group VIII elements (0.2 wt pet Pd and O.2 wt
pet Pt, etc.) remain effective in reducing hydrogen absorption
(14).
4.2 Estimation of Hydrogen Concentration Immediately Below
Surface in Early Stage of Exposure - Influence of Alloying
Element
The surface condition of steel exposed to wet sour gas
environments is expected to shift as Fe++ ions dissolved in the
media deposit upon it as iron sulfide. The rate of dissolution
of metal grows to a maximum approximately three hours after the
commencement of exposure, decaying thereafter, as mentioned in
the following section. In order to know to what degree the
alloying exerts an effect upon reduction of hydrogen absorption
when the steel surface is not influenced by protective films,
the alloy effect at an early stage, i.e., within ^l h after
commencement of exposure, is examined.
The hydrogen concentration, Cj3 immediately below a steel
surface after ^l h of exposure to a wet sour gas environment can
be estimated by the analysis of hydrogen permeation through a
sheet %0.1 cm thick in the case of linepipe steel with a hydro-
gen diffusivity of ^10~^ cm /s. The initial stage of permeation
is defined as the time in which the permeation rate J(t) grows
from ^O up to 0.95J(°°), where J(°°) is the asymptotic value of
J(t) at long times. For J(t) < 0.95J(«>) it can be shown (15)
that the log /t J(t) - 1/t plot is a straight line when Cb =
constant, and shifts from line B^A^ to B O A O in Figure 28-9 when
Cb changes from (1 + e)Cb to Cb; the gradient of the straight
line gives a^/4D, from which the hydrogen diffusion coefficient,
D, may be obtained (a is the sheet thickness). See Appendix B).

Figure 28-9: /T J-l/t Relations as influenced by change in


boundary condition.

Figure 28-10 compares the log [/t J(t)] - 1/t plot for a
steel containing 0.4 wt pet Ni or 0.27 wt pet Cu as a reference;
the arrows indicate the time corresponding to the anodic current
ia(t) = 0.95ia(°°). The slopes of the lines extending to the
right of this point give good estimates of D; the intercept on
the ordinate with the thus-obtained D gives the final value of
Cb in the initial stage of permeation, C0. The C0 values thus
obtained are listed in Table 3. The observed influence upon C0
of Cu or Ni is small, i.e., insufficient to explain the effect
shown in Figure 28-6 or 8. Thus the main portion of the effects
caused by these alloying elements must be explained by their
effects in later stages of exposure, which is treated in the
following section.
4.3 Influence of Alloying upon Kinetics of Corrosion of
Hydrogen Evolution Type
In this section the corrosion rate, p, is examined as a
function of time, and possible influences of small amounts of
alloying element upon the corrosion kinetics parameters are
discussed.
Figure 28-11 shows the change with time of the corrosion
rate during exposure to IH^S-saturated brine of pH 5.2, for
steels with Cu, Cr or Ni as alloying additions. The p(t) is
defined by
HaS-saturated brine
pH 5.2
25 C
1mm

Figure 28-10: Influence of alloying upon initial permeation


characteristics.

HzS-saturated brine
pH 5.2
25±2C
SH
AQ (0.27 Cu)
T2 (0.56Cr)
T1 (0.37Ni)

t (day)
Figure 28-11: Corrosion kinetics as influenced by alloying.
Table 5 Table 4
.. Steel D*(cm2/s) C 0 (wt ppm) Steel t max kj* e*
"SId.he ~
0 . 4 % Ni 1.19xl(T 6 3.6 SH 3" 350 0.3
0.27ICu 0.91x10-6 6.0 A0 , (a) 258 (a) 0.29
ref. 0.55x10'° 9.1 (0.27Cu) (b
> 297 (b
> °'64
T2 , (a) 812 (a) 0.64
(0.56Cr) 3 (b) 523 (b) 0.64
T1 1140 1
(0.37Ni) <1.5
"IaJ and (b) indicate values for
the upper and lower p(t) value
respectively
PCt) = ™ (2)

where W(t) is corrosion weight loss as a function of time, t.


p(t) is observed to assume its maximum value at 3 h or less,
thereafter decreasing following the relation
P(t) = p(3)(|)e (e = 0.3 to 1.0) (3)
It is characteristic that the Ni-bearing steel, Tl, is more
prone to corrosion at an earlier stage, but p(t) diminishes at a
greater rate and becomes more stable thereafter than occurs with
steels containing Cu or Cr. The corrosion rate for the Cu-
bearing steel, AQ, is unstable when compared with that for Tl;
the experimental plots for p(t) range widely and separate into
the upper and lower lines (see Fig. 28-12), while a single
stable line, shown in Figure 28-11, is obtained for Tl. p(t)
data for Cr-bearing steel, T2, range moderately and are separat-
ed into the parallel upper and lower lines, as illustrated in
Figure 28-12, implying that the stability of the corrosion rate
varies in the sequence AQ < T2 < Tl (greatest decay rate). p(t)
for SH also is unstable (the lower lines alone are plotted in
Figure 28-11 for clarity). Thus p(t) for the Ni-bearing steel
assumes values as low as 0.75 mdd after a 64 day exposure to the
solution from Figures 28-11 and 12 and are summarized in Table 4.

H2S-saturated brine
pH 5.2
25±2C
AO (0.27 Cu)
T2 (0.36Cr)

Figure 28-12: Unstable corrosion kinetics observed for AQ


(0727 Cu) and T2 (0.56 Cr).
The change with time of Cfl as influenced by alloying is
shown in Fig. 28-13.* It is noticed that Cft of steel alloyed
with small amounts of Ni decreases to a considerable degree
after 24 h exposure to the environment, corresponding to the
greatest decay rate of p(t) observed in this steel.

*CH is defined by the NTP volume of hydrogen liberated during


96 h from a specimen kept at 25°C in mercury, after exposure to
the test solution for the prescribed time indicated on the
abscissa.
HzS-saturated brine pH5.2 25C

Figure 28-15: Change of C^ with time of exposure as influ-


enced by alloying.

The corrosion kinetics observed and mentioned above may be


explained by the following model. The sulfide corrosion rate at
around ambient temperature could hardly be explained by diffu-
sive movements of anions and cations through the lattice of a
corrosion film, as would be possible in a high-temperature oxi-
dation process. So it might well be expected that penetration
of sulfide solution through leakage paths or discontinuities in
the film, blocking of leakage paths, generation of compressive
stress, upheaval and cracking of the film and repetition of
these processes should proceed during the corrosion under con-
sideration.
If the unblocked points are N per unit area, and the mass
increment of sulfide is dW-^, then
- dN = k 1 N dW1 (4)

since it is expected that the rate of blocking of leakage spots


caused by increasing the mass of the sulfide, should be propor-
tional not only to dW-^ but also to N (16) . Here dW-^ is related
to corrosion loss dW in dt as
dto^ = n(t)dW (5)
n(t) being the precipitation ratio, which is expected generally
to change with time. The rate of dissolution or corrosion is
expected to increase proportionally to N. Thus

a£ = kzN W
Combining equation (4) with equation (6), we obtain, for
large enough t,

dW _ 1 m
dt
Ic1 ['dt n(t)
^O

Comparison of equation (7) with the experimentally obtained


relation
P(t) = p(3)(|)£ (= 0.3 to 1.0)
suggests that
n(t) = constant (8)
for Ni-bearing steel, implying that an ideally stable sulfide
continues to precipitate at a constant ratio to the dissolution
rate of the steel, which would control further progress of
corrosion and consequently hydrogen entry at the steel surface.

It could also be considered that the precipitation ratio,


n(t), should decrease with time for all the other steels during
exposure to the corrosive environment according to a decay curve,
n(t) = n(3)(3/t)6(x) (9)
where 6(x) = 1-e. Thus,

TTO-: (k33 =ki3<HxJ


dt -tl-6(x)
3? I'M*n(x) ) ^
and we know that p(3) is related to n(3) as p(3) = e/k}3ri(3),
B = I - 6(x) (O < 6(x) < 1). 6(0.37 Ni) = O as mentioned-above.
e for the other steels is given in Table 4. The equation (10)
with <5 (x) = O has been known to hold in the case of low-tempera-
ture oxidation where a protective film is formed on iron surface
(17).
4.4 Control of Non-Metallic Inclusions
Hydrogen absorbed in steel is reversibly and irreversibly
trapped at interfaces between the matrix and sulfides of various
dimensions (3.1). Of these, the interfaces at crack-like, flat
sulfides of width £ 10 ym form HIBC nuclei when hydrogen is
trapped to precipitate as pressurized gas therein. A flat
inclusion-matrix interface, when separated and pressurized with
hydrogen gas to form a crack, will accelerate stress-induced
hydrogen diffusion into the crack-tip region, the number of
hydrogen atoms to be deposited at crack tip, n(t), being pro-
portional to (KvDtAsT)V5:

n(t) = eC(™t)4/5 (ID


B

where K is the stress intensity factor of the crack concerned


and is proportional to the pressure of hydrogen gas precipitated
within the crack, V the partial molal volume of hydrogen, D the
diffusion coefficient of hydrogen, t the time, kg Boltzman's
constant, T the absolute temperature, C the average concentra-
tion of lattice-dissolved hydrogen and 3 = (5/2) (3ir/4) 1'5
[/2"/ir(l+v)/3]4/ , v being Poisson's ratio (18). The hydrogen
deposited at the crack tip will facilitate crack growth. Around
a globular inclusion, on the other hand, no stress - induced
diffusion takes place because of lack of triaxial stress concen-
tration there. This is the very reason why inclusion shape
control is essential for improvement of HIBC resistance of line-
pipe steels.
Actually shape control of sulfides by rare earth metals or
a Ca treatment in combination with Ti addition has been shown
to be very effective in improving HIBC resistance in immersion
tests as well as in tests under externally applied stress (14).
Figure 28-14 shows the effect of sulfide shape control by
Ti-REM treatment upon HIBC resistance of steel under tensile
stress (14). Hydrogenation, in this example, is electrolyti-
cally performed in IN sulfuric acid with 7.7 x 10~% sodium
metaarsenite (NaAsO2) added as a hydrogenation promoter,
cathodic current density being 80 A/m . The considerable effect
Normalized stress 0/0b

Time to failure t F (h)


Figure 28-14: Effect of sulfide shape control by Ti-REM
treatment upon HIBC resistance.

of Ti-REM treatment upon prolongation of time to failure is


clear from this figure. As an alloying element controlling
hydrogen entry at steel surface, Cu loses its effect in this
test medium. Through metallurgical observations one learns that
Ti causes adequate dispersion of REM sulfide or oxy-sulfide
inclusions through the uniformly dispersed nature of TiN, which
serves to provide nucleation sites for sulfide formation (14).
Thus, excessive sulfur segregation, leading finally to flattened
MnS after hot rolling, will be suppressed.
Recent steel-making technology has extended interest to Ca
treatment as a means of controlling sulfides in steel. It has
been found that Ti addition in combination with Ca also retards
time to failure, tp, considerably in comparison with that for
simple Ca addition. Figure 28-15 is an example illustrating the
effect: the combined effect of Ti and Ca treatment to consider-
ably prolong tp is clear from this figure, while the treatment
by Ti or Ca alone produces an insufficient or a small effect.
Hydrogen could precipitate also at oxide-matrix interfaces,
which would act as HIBC nuclei in linepipe steels. It has been
shown that time to failure under the same hydrogenation condi-
tions as in Figure 28-14 is considerably increased by lowering
the oxygen content of steel with low sulfur content and with
pearlite-banded structure, although this tendency is reduced
for steel with acicular ferrite structure and free from well-
defined pearlite bands (19).

4.5 Improvement of Through-Thickness-Directional Fracture


Toughness by Heat Treatment

For sour gas linepipe steels with a high strength require-


ment, higher than X-46, API grade, control of metallurgical
Normalized stress 0/0^

Time to failure tr(h)

Figure 28-15: Combined effect of Ti and Ca upon HIBC resis-


tance .

structure is more important. For such steels the choice of


strengthening mechanism without intervention of hard or brittle
phases becomes difficult because P, Mn, V, Nb, etc. will segre-
gate in phases of lower solidifying temperatures, when molten
steel solidifies, and the segregation of these elements will
facilitate formation of hard or brittle phases. In this sense
it can be said that the development of sour gas linepipe steel
involves the problem of selecting a strengthening mechanism with
as uniform as possible a metallurgical structure, provided that
inclusion control is nearly complete.

If a hydrogen embrittlement-sens itive phase like high-


carbon martensite is present locally in front of microcracks or
HIBC nuclei, these microcracks will easily grow by cracking it
and merging to form extended parallel-to-plate cracks.
Thermally transforming the HE-sensitive phase to an HE-insensi-
tive one may raise the HIBC resistance of the steel. Actually,
tempering linepipe steels with martensitic and bainitic phases
in their metallurgical structure at 650°C for 30 min. or more,
especially when preceded by quenching from 930°C, considerably
improves HIBC resistance, as evaluated in immersion tests in
I-^S-saturated brine. This is true, provided that adequate
sulfide shape control is applied (14).

Influence of QT: Experiments have shown that quenching


from 93O0C (kept for 30 min.) and tempering can reduce by as
much as several tens of pet the HIBC resulting from exposure to
H 2 S-saturated brine of pH as low as 3.0 to 3.8. Figure 28-16
illustrates this situation; this figure suggests that quench-and-
tempering could completely eliminate HIBC upon exposure to the
environment of pH as low as 3.0 to 3.8 (MK in Figure 28-16). It
is to be noted that sulfide shape control is nearly perfect in
MK, i.e., an experimentally obtained measure of success of sul-
fide shape control, ESSP = (Ca/1.25S) (1-124 O), is around unity
(1.06), Ca, S and O being expressed in wt pet. Figure 28-16
also shows that quenching and tempering does not improve HIBC
resistance for steel AQ. (After quenching followed by tempering,
slight banding is observed to be inherited from the as-rolled
structure both for MK and AQ). This suggests that success of
QT in improving HIBC resistance requires sulfide shape control
AQ (ESSP=O) MK (ESSP = 1.06)

as 15min eOmir 15min|60min as 15mm 60mm 1mm 15min|60min


rolled Temper OT rolled Temper O OT

Figure 28-16: Results of immersion test in HoS-saturated


brine of pH 3.0 to 3.8, 25°C.

to a certain degree; note that ESSP is zero for steel AQ.


Figure 28-16 suggests again that a single tempering would not be
expected to improve HIBC resistance appreciably for pearlite-
banded steels.

We have learnt from the heat treatment study mentioned


above that light banding is inherited from the as-rolled struc-
ture after the Q-T process. If steels are placed in an exces-
sively hostile environment, we would have to expect that the
light bands could be an easy path of HIBC propagation. It might
well be expected, in this connection, that no substantial diffu-
sive motion leading to an alleviation of segregation could occur
in this Q-T treatment, except by the three fastest elements, H,
if present, C and N. It has been observed that P, and carbide-
forming elements like Mn, V, Nb, Ti, Cr, and Si, etc. segregate
to a greater or lesser extent in the bands. Of these, the
segregation of P and Mn (20) is the most influential in that the
segregation factor, defined by the ratio of the EPMA (electron
probe microanalysis) peak height to the mean concentration,
itself is the greatest (P) or the absolute concentration in that
zone is the highest (Mn).

Very naturally the problem is "Does phosphorus segregation


itself reduce HIBC propagation?" The Q-T treatment which dis-
played a favorable effect could not be expected to alleviate
segregation of P as well as of Mn; the diffusion distance 2/EKyt,
for the faster element, P, caused by the Q-T treatment is as
small as 2 x 10 cm, compared with a typical band spacing
5 x 10~3 cm< (Jn this case Q-T apparently contributed to an
improvement of the metallurgical structure). We have assumed
that the diffusion coefficient of phosphorus, Dp, in cm^/s, is
given by (12):
D = 2.9 exp(-55000/RT) (12)

It should be noted that the success of the Q-T treatment in


the complete elimination of HIBC, as mentioned above, requires
a sulfide shape control to a certain extent.
5. NEW SOUR GAS LINEPTPES
By combining the above-mentioned techniques, development of
sour gas linepipes intended for use in aggressive environments
is possible. Some examples are given below.
(1) The first examples of sour gas linepipes developed are those
which withstood an exposure test in pressurized, wet H2S-CC>2 gas
(22), the composition of which simulates that of the Lacq sour
gas field in France: H2S 21 to 8 mol pet, CC>2 20 to 11 mol pet,
the remainder of the gas being N2 under a total pressure of
13.7 MPa.
Substitute ocean water was introduced into the pipe to
simulate the worst condition for the tested pipes, the inner
surface of the pipes being exposed to wet gas and aaueous phases
in part. The pressurized medium was stationary, and the
temperature of test was 20 to 30°C. The chemical compositions
and mechanical properties of the tested pipes A, B, C and D are
given in Table 5. All these are UOE pipes of API grade X65 with
30 inches in diameter, 0.625 inches in wall thickness.
The test results showed that all the pipes did not burst
after a test duration of 1000 h; careful ultrasonic testing
after exposure revealed three spots of a very small reflected
area, indentified as HIBC, for pipe C in the aqueous zone, and
freedom from hydrogen-induced defects for pipes A and B. In the
reference pipe D, extended and densely populated HIBC were
observed to have formed during the test. After several compara-
tive tests with diffusion analysis for hydrogen in pipes and
small test pieces, it turned out that the aggressiveness of the
pipe test was approximately equivalent to that of an immersion
test in H2S-saturated brine of pH 5.2 (small piece test).
(2) Other examples of steels developed for sour gas pipe and
intended for use in more aggressive environments, with chemical
compositions and mechanical properties, are given in Table 5.
For steel E (21) both inclusion-shape and microstructural
control were applied for the prevention of formation and propa-
gation of HIBC. The inclusion shape control is perfect in this
steel; ESSP = 1.77 > 1. The microstructure was controlled to be
band-free. For steel F both Ni addition for reduction of hydro-
gen entry, and nearly perfect inclusion shape control, for pre-
venting HIBC nucleation (ESSP = 0.99 ~ 1) are employed. It has
been confirmed that both these steels, E and F, resisted HIBC
formation perfectly upon exposure to 10 atm of H2S-saturated
brine for 96 h at 25 C. Further experiments in an autoclave
confirmed the perfect resistance of steel F to HIBC formation
(HIBC% by UST is zero) upon exposure to saturated brine under
20 atm H2S + 50 atm C02 phase, and to the wet gas phase for 96 h
at 31 0C. The temperature of the last test was chosen to avoid
the formation of a solid ^S • 6^0 phase above 10 atm H2S,
below 29.5 OG.
Table 5: Examples of HIBC-Resistant Linepipes and Reference
Steel (D)

Steef* C Si Mn P S Cr Ni Mo Nb V Ti Ca REM BLC Test pav 10atm H1S 20atm HaS-50atm COt
C(56 HIBC % pa v CS5 HIBC %
A 0.089 0.31 1.49 0.012 0.001 0.19 0.20 0.033 0.069 0.015 0.0010 free from mdd <*/100g mdd <*/100g
HIBC
B 0.048 0.24 1.41 0.009 0.001 0.60 0.083 0.0060 a
C 0.023 0.29 1.59 0.013 0.002 0.21 0.22 0.036 0.075 0.011 0.0030 almost free
from HIBC
D 0.089 0.29 1.52 0.018 0.004 0.26 0.082 HIBC
E 0.090 0.15 0.83 0.009 0.0007 0.025 0.0036 212.5 1.28 O
F 0.054 0.27 1.19 0.017 0.0009 0.37 0.031 0.016 0.0017 74.5 0.08 O 75 0.22 O
* Metallurgical structure : A1C acicular fernte
B, D, F pear lite-banded
E. band-free

6. CONCLUSION
When compared with general HIC mechanisms, the mechanism of
HIBC of linepipe steel is clear: the trapping of hydrogen at
interfaces between matrix and, typically, hot-roll-flattened
manganese sulfide inclusions, the separation of interfaces to
form hydrogen gas-pressurized crack-like cavities (blister
cracks), and the growth of blister cracks and linkage between
them produce the HIBC. The manner of linking of blister cracks
differs depending upon whether externally applied stress is
present or not. The hydrogen enters steel as a result of corro-
sion in the presence of water and ^S, the efficiency of hydro-
gen entry being maximum around ambient temperature.
As a means of preventing the HIBC in linepipe steel, a
materials approach is presented which consists of (a) control
of hydrogen entry at the steel surface, (b) control of the
nuclei for deposition of a small amount of hydrogen, and
(c) improvement of the toughness of the hydrogen-impregnated
matrix around the tips of the HIBC nuclei. Examples of HIBC-
resistant linepipes are given.

REFERENCES

1. Warren, D. and Beckman, G. W., Corrosion 13: 631t (1957).


2. Hirth, J. P. and Johnson, H. H., Corrosion 32: 3 (1976).
3. Hill, R. T. and lino, M., Proceedings of the 1st Interna-
tional Conference, Current Solutions to Hydrogen Problems
in Steels, Washington, DC, November 1-5, 1982, American
Society for Metals, p. 196.
4. Nakasugi, H., Presented at National Association of Corro-
sion Engineers Middle East Conference in Bahrain.
5. Smialowski, M, in "Stress Corrosion Cracking and Hydrogen
Embrittlement of Iron-Base Alloys," R. W. Staehle, J.
Hochmann, R. D. McCright and J. E. Slater, eds., NACE,
Houston, Texas, p. 405 (1977).
6. McCright, R. D., ibid, p. 306.
7. Phragmen, G., Jernkontret Annaler A.R.G. 128: 37 (1944).
8. de Kazinczy, F., Acta Metall. 7: 725 (1959).
9. Volkl, J. and Alefeld, G., Topics in Applied Physics Volume
28, Hydrogen in Metals 1, Berlin Heidelbery New York:
Springer Verlay p. 321 (1978).
10. lino, M., Acta Metall. 30: 377 (1982).
11. Evans, U. R., The Corrosion and Oxidation of Metals, London:
Edward § Arnold (1960), p. 393.
12. lino, M., Acta Metall. 30 367 (1982).
13. lino, M. Metall. Trans. 9A: 1581 (1978).
14. lino, M., Nomura, N., Takezawa, H. and Gondoh, H., Revue de
Metallurgie No. 8-9: 591 (1979).
15. lino, M., Metall. Trans. A, to be published.
16. Davies, D. E., Evans, F. R. S. and Agar, J. N., Proc. Roy.
Soc. 225A: 443 (1954).
17. Uhlig, H. H., Corrosion and Corrosion Control, New York:
John Wiley § Sons p. 169 (1963)."
18. lino, M., Eng. Frac. Mech., 10: 1 (1978).
19. Reference (3) p. 159.
20. Miyoshi, E., Tanaka, T., Terasaki, F. and Ikeda, A., Trans.
AIME 98: 1221 (1976) .
21. Seibel, M. G., Academie de Science: 4661 (1963).
22. Nakashima, A., lino, M., Matsuda, H. and Yamada, K.,
Presented at 20th Annual Conference of Canadian Institute
of Metallurgists, Hamilton, Ontario, Canada, 23 to 27
(August, 1981).

APPENDIX A. THE MEANING OF C^5


C^ played an important role in interpreting experimental
data in 4.1. Here the physical meaning of cA~> is examined by a
diffusion analysis. This consists of the analysis of hydrogen
absorption into a plate containing traps and of hydrogen libera-
tion from the plate.
Absorption of Hydrogen into a Plate Containing Traps
The absorption of hydrogen into a place containing reversi-
ble and irreversible traps will be considered first. It will
be assumed that lattice-dissolved hydrogen concentration
immediately below the surfaces of the plate attenuates as C o e~ at
(t > O, boundary condition), and that the plate is free from
hydrogen before exposure (initial condition). In a steel-
hydrogen system the low lattice hydrogen concentration approxi-
mation may be applied to the diffusion analysis. Thus, the
trapping rate, in non-dimensional form is given by*

~ = Xu - yv (A.I)
where u = C L /C O , v = C r /C o , T = Dt/a 2 , X = Nk(a2/D) and y =
p(a2/D).
The ratio y/X is related to the binding energy between
hydrogen and a trap as y/X = (N^/N)e~E/^BT, where kg is the
Boltzmann constant and T the absolute temperature. For large
enough E, say E >0.8eV (12), y can be neglected when compared

*CL and C^ are the concentration of hydrogen upon normal lattice


sites and trap sites respectively. D is the diffusion constant,
t is the time and 2a is the plate thickness. k and p are pro-
bability of capture and release from the trap respectively. N
and NL are the densities of the trap site and normal lattice
site respectively. x is the spatial coordinate.
with A; the trap is better treated as irreversible. If Cj[ is
the concentration of hydrogen in irreversible trap sites, then
3C^/8T = KC or writing w = Ci/Co

^=KP (A.2)

under the low trap coverage approximation (12). K is the para-


meter of capture.
Solving the diffusion equation with respect to u,
2
a| (u + v + w) = i-£ (A. 3)
BT ^2
where £ = x/a, under the initial and boundary conditions
mentioned above, we obtain

-2 n|Q (-l)n(n + ^)TTCOs(n + ^ ) T T £ c o n j u g a t e ]


(A.4)
which, using an effective diffusion constant D* = D/(l + —),
approximates to ^

(A.5)

when A + y»(y) »K, where y = /a(l + —-—)-K and

2s1 = K + A + y+(n + %)2Tr2 + /[K + A - y + (n + %)2ir2]2 + 4Ay


If we assume here D* = 10"6 cm /s, D = 7.10"5 cm /s, 2a =
1 cm, t = 3.6 1O 5 S, a = 0.01, a(l + ^) - K ^ 0.7 - K * 0.5
(K - 0.2), then we calculate from (A.5J that the hydrogen con-
centration at midthickness of the plate after 100 h exposure,
u(0, 3.6 x 105s) - 0.4. This is approximately the same as that
at the steel surface, u(±l, 3.6 x 105s) = e~ aT - 0.36, meaning
that the hydrogen concentration, u, can be assumed to be nearly
uniform through the specimen after 100 h exposure. Hydrogen
captured in reversible traps is understood to be in thermo-
dynamic equilibrium with hydrogen upon normal lattice sites. So,
v also is uniform through specimen after 100 h exposure.
Liberation of Hydrogen from a Plate Containing Traps
Since we now know that hydrogen concentrations u and v are
nearly uniform through specimen after 100 h exposure to the
hydrogenating environment, the problem of hydrogen liberation
from the plate can successfully be treated under uniform initial
conditions u(£,t) = 1, v(£,v) = X/y for t = O and boundary con-
dition u(±l,t) = O for t>0.
Solving equations (A.3) with (A.I) and (A.2) under these
initial and boundary conditions, and differentiating u with
respect to £, and letting £ = ±1, we obtain the hydrogen flux j
at boundaries in units of DCo/a,

(A.6)
T
The total flux during t after exposure, g(r) = f dTJ(l,T) in
units of C0, is given by ^o

(A.7)
The quantity attained after sufficiently long time, g(°°), may be
shown to be

(A.8)

which can be proved to reduce to

(A.9)

S0 in equation (A.7) is estimated to be

(A.10)

when X + y»(-£) »K. Since we know X/y = 70 as a typical value,


and generally y»l, (12) then for small enough K, K«(y) , SQ *
0.035. Therefore the largest exponential term e~s6T in equation
(A.7) is as small as e"so - 0.03 after 100 h liberation time,
meaning g(100 h) - g O) . Thus, we know that the C^5 represents
substantially the total hydrogen volume released from normal
lattice sites and reversible trap sites of specimen, when we
take into consideration that C^, the hydrogen volume released
in 72 h from a specimen kept at 45 °C, is equivalent to
C0g(143 h), t i.e., the volume to be released for 143 h from
specimen kept at 25 °C, and g(100 h) < g(143 h) * gO). The
adequacy of the assumption K«(Z)2 for usual linepipe steels is
discussed elsewhere (15). Thus for steels under consideration

tTo be more exact Co?zshould be replaced by Cp* =


CQ exp(-3.6xl05Da/a ) which is ^ C0 for small enough a.
The quantity C0(I + A/y) can also be shown (12) to equal COD/D*.

APPENDIX B. log/t J(t) - 1/t RELATION IN PERMEATION TRANSIENT


FOR CHANGING BOUNDARY CONDITION
Let the initial and boundary conditions be
u U, T) = o (T < o) (B.I)
u (o, T) = o (T > o)
u (1, T) = £(T) (B.2)
for the diffusion equation under consideration,

3u _ 3 2 u rR -,
8T ~
Tr a?27" ^ * ^)

Expressing the Laplace transform of a function h(£,T) by h(£,s),


and solving equation (B.3) with respect to j = Ja/DCo =
9u/H S=0 J we obtain
j = sf.g or If . g (B.4)

where g is a function which equals j when f = unit step function,


i.e. ,

g = (B.5)
/s sinh/s"
when f ( t ) changes as

r( , fl + e (o < T < T)
UTJ
Ml (T < T) (B.6)
where T = Dt^/a. j in equation (B.4) is

j = [1 + e(l - e"Ts)] — i (B.7)


/s" sinh/s
By expanding l//s~ sinh/s~ as

(B.8)

and applying the inverse transformation to equation (B.7), we


obtain

(B.9)

with

(B.10)

showing the transitional response.


Equation (B.9) converges very rapidly, so that for
o <. j CT) £ 0.95
2 -J-
J(T) = e 4T [1 + En0(T)] (B.11)
/TTU

is a very good approximation to J(T) in equation (B.9) (15).


From equation (B.11) we know that log/t J - 1/t plot produces a
transition from line B-^Aj to B Q A O in Figure 28-12, the slope of
line B0A0 giving a^/4D and the intercept on the ordinate, with
the thus-obtained D, giving the final value of C0 in the initial
stage of permeation.

Вам также может понравиться