5.06 Corrosion and Environmentally-Assisted Cracking of
Carbon and Low-Alloy Steels
H.-P. Seifert
Paul Scherrer Institut, Villigen PSI, Switzerland

J. Hickling
Independent Technical Consultant, Prastio-Avdimou, Cyprus

D. Lister
University of New Brunswick, Fredericton, NB, Canada

ß 2012 Elsevier Ltd. All rights reserved.

5.06.1

Introduction

107

5.06.2
5.06.2.1
5.06.2.2
5.06.2.2.1
5.06.2.2.2
5.06.2.2.3
5.06.3
5.06.3.1
5.06.3.2
5.06.3.2.1
5.06.3.2.2
5.06.3.2.3

5.06.3.2.4
5.06.3.2.5
5.06.4
5.06.4.1
5.06.4.2
References

Uniform and Flow-Accelerated Corrosion
Uniform Corrosion
Flow-Accelerated Corrosion
Controlling factors
Mechanisms and models
Service experience and mitigating actions
Localized Corrosion and Environmentally Assisted Cracking
Pitting
Environmentally Assisted Cracking
Basic types of EAC and major factors of influence
Corrosion fatigue and strain-induced corrosion cracking
Stress corrosion cracking
EAC mechanisms and models
Service experience and mitigation actions
Conclusions
Uniform and Flow-Accelerated Corrosion
Localized Corrosion and Environmentally Assisted Cracking

109
109
111
111
114

118
120
120
122
122
123
128
132
136
139
139
139
140

Abbreviations
AC
AGR
ANL
ASME
ASME BPV
ASME III
ASME XI
ASTM
BWR
BWRVIP
BWRVIP-60

Content of Cr, Mo and Cu in alloy in
EPRI ‘CHECWORKS’ FAC-Code
Advanced gas-cooled reactor

Argonne National Laboratory, USA
American Society of Mechanical
Engineers
ASME Boiler and Pressure Vessel
Code
Section III of ASME BPV Code
Section XI of ASME BPV Code
American Society of Testing and
Materials Standards
Boiling water reactor
Boiling Water Reactor Vessel and
Internals Program
Basis document for SCC
disposition lines for low-alloy
steels

CANDUW

CF
CRDM
CS
DCPD

DH
DL
DO
DSA
EAC
EC
ECP

ECPcrit

CANada Deuterium Uranium,
PHWR developed by Atomic Energy
of Canada Ltd.
Corrosion fatigue
Control rod drive mechanism
(housing)
Carbon steel
(Reversed) direct current potential
drop crack length measurement
method
Dissolved hydrogen (concentration)
Disposition line
Dissolved oxygen (concentration)
Dynamic strain ageing
Environmentally assisted cracking
Erosion corrosion
Electrochemical corrosion potential
Critical cracking potential (e.g., for
SICC)

105


106

Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

EPRI


Electric Power Research Institute,
USA
F & A model EAC model for CS & LAS developed
by P. Ford and P. Andresen (GE GR)
FAC
Flow-accelerated corrosion
FRAD
Film rupture anodic dissolution EAC
mechanism
HAEAC
Hydrogen-assisted EAC
mechanism
HAZ
Heat-affected zone of weldment
HCF
High-cycle fatigue
HT
High temperature
HWC
Hydrogen water chemistry
JSME
Japanese Society of Mechanical
Engineers
LAS
Low-alloy steel
LCF
Low-cycle fatigue
LEFM
Linear elastic fracture mechanics

LWR
Light water reactor
MT
Mass transfer in EPRI
‘CHECWORKS’ FAC-Code
NDT
Nondestructive testing
NMCA
Noble metal chemical addition
NRC
Nuclear Regulatory Commission,
USA
NWC
Normal water chemistry
PHWR
Pressurized heavy water reactor
PWHT
Postweld heat treatment
PWR
Pressurized water reactor
PWSCC
Primary water stress corrosion
cracking (in PWRs)
QþT
Quench and temper heat treatment
RPV
Reactor pressure vessel
SCC
Stress corrosion cracking
SEM

Scanning electron microscope
SHE
Standard hydrogen electrode
SICC
Strain-induced corrosion cracking
SS
Stainless steel
SSR(T)
Slow strain rate (test)
SSY
Small-scale yielding
UTS
Ultimate tensile strength
VGB
German Association of Large Power
Plant Operators
YS
Yield strength

Ceq

Symbols

KI
KI,i

C
Cb

Concentration of Fe(II) species at

the oxide–coolant interface
Concentration of Fe(II) species in
the bulk coolant

CODLL

d
D
da/dN
da/dNAir
da/dNCF

da/dtAir ¼
da/dNAir/
DtR
da/dtCF ¼
da/dNCF/
DtR
da/dtSCC
da/dtSICC
dCODLL/dt

de/dt
de/dtcrit
dKI/dt
EA
Fen

G
h


kc
kd

Thermodynamic equilibrium
concentration of Fe(II) species
Crack-opening displacement at
load line in precracked fracture
mechanics specimen
Pipe diameter
Diffusivity
Crack advance per fatigue cycle
Crack advance per fatigue cycle in
air
Corrosion fatigue crack advance
per fatigue cycle in hightemperature water
Time-based fatigue crack growth
rate in air
Time-based corrosion fatigue
crack growth rate in hightemperature water
SCC crack growth rate
SICC crack growth rate
Crack-opening displacement rate
in slow rising load or displacement
test
Strain rate (sometimes locally at
crack-tip)
Critical strain rate (e.g., for SICC)
Stress intensity factor rate in slow
rising load or displacement test

Arrhenius activation energy of
thermally activated process
Environmental correction factors,
ratio of fatigue life in air at room
temperature to that in water at
service temperature
Geometry factor in EPRI
‘CHECWORKS’ FAC-Code
Mass transfer coefficient for Fe(II)
species from the oxide–coolant
interface to the bulk environment
by convection
Geometry factor in Siemens-KWU
‘WATHEC’ FAC-Code
Dissolution reaction rate constant
of magnetite at the oxide–coolant
interface
Stress intensity factor (LEFM)
Stress intensity factor at the onset
of SICC crack growth in slow rising
load tests with precracked fracture
mechanics specimens


Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

KIJ

Stress intensity factor at the onset
of ductile crack growth

m
Paris law exponent for fatigue and
corrosion fatigue
n
Frequency exponent for corrosion
fatigue
pH % Àlog[Hþ] pH value
R
Load ratio of minimum to
maximum load
Flow-accelerated corrosion rate
RFAC
Dissolution rate of magnetite at the
Rd
oxide–coolant interface
Formation rate of magnetite at the
Rg
metal/oxide interface
Mass transport rate of Fe(II)
Rm
species from the oxide–coolant
interface to the bulk environment
by convection
Re ¼ dur/m
Reynolds number
Sc ¼ m/( rD)
Schmidt number
Sh ¼ hd/D
Sherwood number
t

Time
T
Temperature
u
Coolant velocity
Z
Reduction of area in tensile test
[X]
Concentration of element/species
X in water or in alloy
a
Steam void fraction in EPRI
‘CHECWORKS’ FAC-Code
DC ¼ Ceq À Cb Undersaturation in dissolved Fe(II)
species
Dd
Wall thinning
DK ¼ KI,max À Stress intensity factor range of
KI,min
fatigue cycle
Upper DK threshold for corrosion
DKCF,H
fatigue
Lower DK threshold for corrosion
DKCF,L
fatigue
DK threshold for fatigue in air
DKth, Air
Decline time (down-ramp) of
DtD

fatigue cycle
Hold time at maximum load of
DtH
fatigue cycle
Rise time (up-ramp) of fatigue cycle
DtR
e
Mechanical strain
Critical strain (e.g., for SICC)
ecrit
k
Specific electrical conductivity
m
Viscosity of coolant
n
Loading frequency
Upper critical frequency for
ncrit,H
corrosion fatigue

ncrit,L
r
s
scrit
t

107

Lower critical frequency for
corrosion fatigue

Density
Mechanical stress
Critical stress (e.g., for SCC)
Fluid shear stress at pipe wall

5.06.1 Introduction
Carbon and low-alloy steels (CS & LAS, Table 1)
and their associated weld filler metals are widely
used for pressure vessels and piping in both the
primary and secondary coolant circuits of watercooled reactors (light water reactors (LWRs) and
CANDUs – pressurized heavy water reactors
(PHWRs)), as well as in service water systems.1
The main reasons for the use of CS & LAS are
their combination of relatively low cost, good
mechanical strength and toughness properties in
thick sections (hardenability), and good weldability,
as well as their good stress corrosion cracking (SCC)
resistance in primary coolant environments. Compared with austenitic stainless steels and nickel-base
alloys, ferritic CS & LAS exhibit only moderate
corrosion and irradiation resistance. They also show
a ductile-to-brittle transition in toughness properties
at lower temperatures.
CS & LAS components in the primary circuit of
pressurized water reactors (PWRs) are clad (usually
with austenitic stainless steel) and thus do not generally come into direct contact with the reactor coolant.
This is also the case for the reactor pressure vessel
(RPV) in boiling water reactors (BWRs), although
the RPV head is sometimes left unclad and the cladding has been removed from the blend radius of many
RPV feedwater nozzles. In BWRs of German and of
newer General Electric designs, extensive use is also

made of unclad LAS and CS in both the feedwater and
steam lines, as well as in the condensate system. The
primary coolant piping in conventional CANDUs is
made exclusively of unclad CS. In secondary coolant
systems, the steam generator pressure vessel shell is
unclad, as are the feedwater, drain, and steam lines.
CS & LAS pressure-boundary components, in particular in the primary circuit such as the RPV, are
very critical systems with regard to plant safety and
lifetime (extension). Minimizing corrosion improves
plant availability and economics and is also fundamental for safe operation over extended periods of
50–60 years.


108

Typical CS & LAS piping and pressure vessel materials in Western LWRs (US designation, according to Section II of ASME BPV Code)


Designation

Type

Product
form

Cmax
(%)

Mn
(%)


Pmax
(%)

Smax
(%)

Simin
(%)

Cumax
(%)

Nimax
(%)

Crmax
(%)

Momax
(%)

Vmax
(%)

YS25 C
(MPa)

Heat
treatment


Microstructure

SA 106 Gr. B

CS
C–Mn

Pipe drawn

0.30

0.29
1.06

0.035a

0.035a

0.10

0.40a

0.40b

0.40b

0.15b

0.08b


! 240
300–400c

Normal.

Ferriticpearlitic

SA 333 Gr. 6

CS
C–Mn

Pipe drawn

0.30

0.29
1.06

0.025a

0.025a

0.10

–

–


–

–

–

! 240
300–400c

Normal.

Ferriticpearlitic

SA 516 Gr. 70

CS
C–Mn

Vessel plate

0.27d

0.79
1.30

0.03a

0.035a

0.13

0.45

–

–

–

–

–

! 260
300–400c

Normal.

Ferriticpearlitic

SA 533 B Cl.1

LAS
Mn–Mo–Ni

(R)PV plates

0.25

1.07
1.62


0.12e
(0.35)

0.15e
(0.35)

0.13
0.45

0.10e

0.37
0.73

–

0.41
0.64

0.05

! 345
450–550c

Q&T

Bainitic

SA 508 Gr. 3 Cl. 1


LAS
Mn–Mo–Ni

(R)PV
forging

0.25

1.20
1.50

0.12e
(0.25)

0.15e
(0.25)

0.15
0.40

0.10e

0.40
1.00

0.25

0.45
0.60


0.05

! 345
450–550c

Q&T

Bainitic

SA 508 Gr.2 Cl. 1

LAS
Ni–Mo–Cr

(R)PV
forging

0.27

0.50
1.00

0.12e
(0.25)

0.15e
(0.25)

0.15

0.40

0.10e

0.50
1.00

0.25
0.45

0.55
0.70

0.05

! 345
450–550c

Q&T

Bainitic

a

In modern steels, these values are less than 0.015%.
Combination shall not exceed 1.0%.
c
Typical range.
d
Carbon varies with thickness up to 0.31%.

e
Requirement for core belt region.
YS ¼ yield strength; Normal. ¼ normalized; Q & T ¼ quenched and tempered.
b

Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

Table 1


Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

Consideration of both uniform and flowaccelerated corrosion (FAC) behavior for all unclad
surfaces is important for corrosion product transport
and deposition (e.g., crud formation on fuel elements)
but – together with the assessment of resistance to
localized corrosion phenomena such as pitting and
environmentally assisted cracking (EAC) – is obviously also required for integrity reasons. In the case of
EAC, however, safety considerations furthermore
require that possible defects extending through the
cladding be taken into account, so that the susceptibility of the RPV must be assessed as if no cladding were
present. Sometimes, thick pads of Alloy 182 have been
welded directly onto the RPV to act as attachment
points for internal structures; the higher yield strength
of Alloy 182, the thicker section and its known SCC
susceptibility raise special concerns for these areas.
In such cases, it is possible that SCC or thermal fatigue
of the austenitic alloy will occur such that the crack
tip propagates to the interface between the austenitic
and ferritic alloys. Furthermore, leakage of coolant

from the primary circuit in PWRs poses a special
hazard for CS & LAS components, since the boric
acid it contains can concentrate and lead to uniform
corrosion, or ‘wastage,’ of external surfaces.
This chapter covers both the uniform and localized corrosion behavior of CS & LAS pressureboundary components in the primary (BWR, PWR,
and CANDU) and secondary (PWR and CANDU)
coolant systems of Western reactors, whereby the
discussion is focused on relevant US nuclear codes
and rules together with material standards in this
area. Special emphasis in Sections 5.06.2 and 5.06.3
is placed on FAC and on EAC, both of which have
resulted in serious pipe ruptures (FAC) or leaks (EAC)
during both nuclear and fossil service in the past.
In Section 5.06.2, the uniform and boric acid
corrosion behavior of CS & LAS, as well as the nature
of the protective oxide film on these materials, are
summarized first, followed by a condensed review of
the FAC behavior of these steels. The major factors
controlling FAC, the underlying mechanism and predictive models, as well as the relevant service experience and possible mitigation actions are discussed.
After a brief overview of pitting in CS & LAS in the
first part of Section 5.06.3, crack initiation susceptibility conditions and crack growth behavior are discussed in detail for the different types of EAC and
compared with the relevant design codes and crack
growth disposition curves for CS & LAS. This is
followed by a review of the mechanistic understanding of EAC and of existing EAC models. LWR service

109

experience and mitigation actions with regard to
EAC are then summarized and compared with this
experimental and theoretical background knowledge.

Finally, Section 5.06.4 summarizes the major conclusions of this review.

5.06.2 Uniform and FlowAccelerated Corrosion
5.06.2.1

Uniform Corrosion

Uniform or general corrosion does not normally
cause a problem for the structural integrity of CS
or LAS components in nuclear coolant systems.
Corrosion rates in typical circuits are generally of
the order of a micrometer per year (1 mm yearÀ1) or
less – higher than those of stainless steel or nickelbased alloys, for example, but quite acceptable.
Around 300  C, uniform corrosion rates of CS &
LAS are minimal at a slightly alkaline pH300  C of
$6–6.5 (neutral high-purity water has a pH300  C of
5.7) and intermediate dissolved oxygen levels. Under
some shutdown conditions, however, LWR primary
coolant can be aggressive to these materials, in particular in conjunction with increased oxygen levels
(e.g., through oxygen ingress from air); below
$100  C, corrosion rates may be high. Compact,
defect-free oxide films grown at higher temperatures
during service are kinetically quite stable at lower
temperatures and usually provide sufficient protection
against uniform corrosion during short shutdown periods. Nevertheless, reactor vessels and LAS piping in
PWRs are clad with stainless steel, which helps reduce
the build-up of crud on fuel and of radiation fields by
ensuring a high degree of water purity with a low level
of dissolved iron.
A particular concern in PWRs arises from the

leakage of borated coolant from joints such as gasketed flanges and its impingement on components
such as flange studs. Up to 2001, some 140 leaks had
been reported publicly.2 Solid boric acid at room
temperature and dilute, deaerated boric acid solutions regardless of temperature have little effect
on CS & LAS, but as the boric acid concentrates, corrosion rates up to about 1 mm yearÀ1 may be reached.
Aerated solutions can be much more aggressive,
with the attack increasing with acid concentration.
Note that as hot coolant escapes to the environment,
its boric acid content (which may be nominally
2000 ppm (1 ppm ¼ 1 mg kgÀ1; 1 ppb ¼ 1 mg kgÀ1) or
more as elemental boron) concentrates by evaporation.
At temperatures in the neighborhood of 100  C, which


110

Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

are attained by surfaces impacted by coolant flashing
to steam, corrosion rates can reach $250 mm yearÀ1.2
In some situations, flow effects can exacerbate the
attack, as described in Section 5.06.2.2.
The resistance of CS and LAS to corrosion is
dependent upon the protective properties of the
oxide film. Environments such as boric acid that
dissolve or erode the oxide then promote corrosion.
The predominant oxide on CS and LAS in coolant
circuits operating above about 130  C is magnetite –
Fe3O4. In deoxygenated alkaline water, the magnetite
forms a double layer that has been well characterized

in terms of materials performance in boiler systems at
temperatures of about 300  C.3 This morphology is
found on CS in CANDU primary circuits, and would
be found on pressure-vessel steel exposed to PWR
primary coolant in the absence of high-alloy cladding.
The layers are formed by the simple oxidation of
the steel by water:
Fe þ 2H2 O ¼ FeðOHÞ2 þ H2

½IŠ

The nascent hydrogen is absorbed by the metal and
diffuses to the exterior. Roughly half of the ferrous
species (often as the dissolved hydroxide – depending
on the pH) are precipitated oxidatively at the metaloxide interface as small crystallites of magnetite, each
a few tens of nanometers across, also releasing hydrogen to the coolant:
3FeðOHÞ2 ¼ Fe3 O4 þ 2H2 O þ H2

½IIŠ

The precise fraction precipitated is determined by
the density of the oxide relative to that of the metal,
since the inner layer occupies the volume of metal
corroded.3 The remainder of the dissolved iron diffuses through the oxide to the bulk coolant and precipitates according to eqn [II] as an outer layer of
magnetite crystals, each several micrometers across,
again releasing hydrogen to the coolant. If metal
species other than those of iron originate from alloy
components elsewhere in a circulating system, they
may coprecipitate and modify the locally formed
magnetite. An example of double-layer formation is

shown in Figure 1.
The concentration of dissolved iron in the coolant
governs the oxide formation. If the coolant is significantly undersaturated in iron, the outer layer cannot
precipitate and the inner layer may even dissolve at
the oxide–coolant interface. In nonisothermal systems,
temperature gradients create solubility differences and
transport iron around the circuit, modifying the oxide
films accordingly (the same phenomenon transports
different oxides around circuits containing other

Coolant flow
Precipitation
Dissolution
Outer oxide

o/s interface

Inner oxide

m/o interface
Corroding metal

Figure 1 Schematic of double layer oxide formation on
carbon steel in high-temperature water.

materials, such as the nickel-base alloys in PWRs).
Thick films may also spall and release oxide particles
to be distributed by the coolant. In circuits connected
to the reactor core, oxide transport may create deposits
on the fuel, impeding heat transfer and leading to

increased radiation fields around out-of-core components (note that the nickel-base alloys and stainless
steel in PWRs can produce deposits derived from
nickel ferrite, NiFe2O4; on high-burnup fuel undergoing subcooled boiling, these can harbor boron from
the coolant and provoke shifts in the neutron flux, as
well as affect radiation fields).
Evolved hydrogen also affects magnetite solubility
(by the one-third power of the concentration – as
indicated by eqn [II]). Such increased solubility at the
metal–oxide interface has been invoked as the reason
for the lack of precipitation within pores as iron diffuses to the oxide–coolant interface.4 Magnetite films
formed on steel surfaces that are pressure boundaries,
where the hydrogen evolved by eqn [I] continuously
effuses through the metal, tend to have a more adherent inner layer of larger crystallites than those formed
on totally immersed surfaces such as test coupons,
where the evolved hydrogen can only diffuse through
the oxide to the bulk coolant once the metal becomes
saturated.5
Under neutral oxidizing conditions, magnetite is
still the predominant base oxide formed on steels.6
However, since dissolved oxygen becomes the oxidant rather than water, hydrogen generation is suppressed and the basic oxidation reactions become:
2Fe þ 2H2 O þ O2 ¼ 2FeðOHÞ2
6FeðOHÞ2 þ O2 ¼ 2Fe3 O4 þ 6H2 O

½IIIŠ
½IVŠ

The oxide layers – especially the outer one – then
tend to contain the more-oxidized forms maghemite
and/or hematite (both of formula Fe2O3), particularly in BWR circuits.7 Under reactor coolant conditions, corrosion rates and oxide solubilities under



Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

oxidizing conditions are generally substantially lower
than those under reducing conditions. At high oxygen
levels, however, the risk for pitting and EAC increase
significantly (see Section 5.06.3).
The forms of oxide that are thermodynamically
stable under various conditions in coolant circuits are
indicated by Pourbaix diagrams, which plot the equilibrium potentials of the oxidizing–reducing reactions
against pH; the higher the potential, the more oxidizing the environment. For dissolved species, the equilibria and therefore the lines in the diagram are
dependent upon the concentration; when illustrating
corrosion situations, a concentration of 10À6 M or less
is often assumed. It should be borne in mind, therefore,
that such diagrams are mainly indicative in nature and
illustrate the possibilities of species formation without
taking account of reaction kinetics. Figure 2, adapted
from Beverskog and Puigdomenech,8 is an example for
species pertinent to steel at 310  C, where a species
concentration of 10À8 M is representative. The hydrogen line in the figure represents the equilibrium:
2H2 O þ 2eÀ ¼ 2OHÀ þ H2

5.06.2.2

½VŠ

Flow-Accelerated Corrosion

5.06.2.2.1 Controlling factors


Flow-accelerated (or -assisted) corrosion (FAC), sometimes called erosion–corrosion (EC) in older literature,
0
−0.2
Fe2O3

−0.4

Fe(OH)4−

−0.6

Fe3O4

E(v)

Fe(OH)+

Fe(OH)2

−0.8

Hydr

ogen

line

Fe(OH)3−

−1

Fe

−1.2
−1.4
6.5

7

7.5

8
pH310 ЊC

8.5

9

9.5

Figure 2 Pourbaix diagram for iron at 10À8 m at 310  C.
Reproduced from Beverskog, B.; Puigdomenech, I. Corros.
Sci. 1996; 38(12): 2121–2135.

111

is essentially the dissolution and erosion of the normally
protective oxide film on CS (or LAS with a Cr-content
< $0.2 wt%), exacerbated by fluid flow effects, resulting in excessive corrosion rates and substantial pipe wall
thinning. Nowadays, the term EC implies the involvement of a significant mechanical component as an abrasive (e.g., by dispersed solid particles in the liquid phase)
or cavitation-induced (mechanical) removal of surface

material; it should therefore be differentiated from FAC,
which is primarily caused by a flow-induced increase in
the mass transfer of dissolving and reacting (corrosive)
species at high-flow or highly turbulent locations,
although fluid shear stress on the oxide film at the
material surface may also make substantial contributions to the damage in some situations.
FAC is a pervasive problem in most types of
steam-raising system and has caused feedwater line
ruptures, occasionally with fatal consequences, in
both fossil and nuclear plants.9,10 In primary coolant
systems also, less serious (though costly) FAC occurs
chronically in the CS outlet feeders of conventional
CANDUs,11 and flow effects are implicated in the
corrosion of PWR pressure-vessel steel by borated
coolant leaking through cracked penetrations in the
RPV head.12 FAC thus occurs in the regions of high
turbulence in both single and two-phase flows, but
never in systems with dry steam.
FAC depends on hydrodynamics (mainly steam
quality, flow rate, fluid shear stress at the wall,
turbulence intensity, and mass transfer coefficient),
environmental factors (mainly temperature, pH,
dissolved oxygen, hydrogen, and iron concentrations)
and material parameters (metal composition – Mo,
Cu and, in particular, Cr content).9 The critical
parameter combinations for the occurrence of FAC
in feedwater systems and the main parameter effects
are schematically summarized in Figure 3.
The conditions leading to increased FAC rates
are usually related to regions with turbulent flow, to

low electrochemical corrosion potentials ECP (i.e.,
to chemically reducing conditions), and to low iron
concentrations in the water (Figures 3 and 4).
Depending on the pH, the maximum FAC rates
occur at about 130  C in single-phase flow, and at
about 180  C in two-phase flow (in the latter, it is the
condition in the liquid layer at the steel surface that
controls the FAC rate, but this is difficult to measure
or predict). Note that FAC can still be a problem
at other temperatures, even though rates are lower.
For example, feeder FAC in CANDU primary coolants occurs at 300–310  C at the core outlet, and FAC
is also significant in feedwater systems at the low


Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

Log FAC rate

112

pH 7

Hydrodynamics
Hydrodynamics
Hydrodynamics
pH 9

• Shear stress at surface
• Flow rate
• Turbulent intensity

• Mass transfer coefficient

Flow

Critical conditions
for high FAC risk
in feedwater

∼150 °C

Material

Environment

• [Cr] in metal < 0.2%

Log FAC rate

Log FAC rate

pH 7
• Low [Fe]
• pH < 9.2
• 120 °C < T < 180 °C
• [O2] < 2–40 ppb

pH 7

∼0.2% Cr


pH 9

T

∼40 ppb

Log FAC rate

pH 7

∼1−2 ppb
pH 9

Log FAC rate

pH 9
[Cr] in metal

∼130 °C

pH 9

pH
[O2]

Figure 3 Critical parameter combinations for flow-accelerated corrosion (derived from Uchida, S. et al. In: Proceedings
of the 13th International Conference on Environmental Degradation of Materials in Nuclear Power Systems, CD-ROM.
Whistler, British Columbia, Canada, 19–23 August, King, P., Allen, T., Busby, J., Eds.; Toronto, ON: The Canadian
Nuclear Society, 2007) and major parameter effects on flow-accelerated corrosion under feedwater conditions.


temperature of condensate extraction. Specific geometries like elbows, bends, protruding weld roots,
orifices, and valves cause local turbulence, which
significantly increases FAC rates at, or immediately
downstream of, the location concerned. Systems such
as the moisture-separator/reheater drain lines, where
steam has condensed and relatively iron-free water is
flowing, are particularly susceptible. In primary coolant systems, there is the desire to keep iron concentrations low to prevent crud build-up and radiation
transport problems, hence the frequent use of highalloy materials that are resistant to FAC as cladding.
However, it must be recognized that a recirculating
system will always tend toward equilibrium; in other
words, dissolved iron concentrations on average will
vary around solubility values, depending upon oxide
dissolution and precipitation kinetics, temperature
gradients around the circuit, and the capacity of
sinks such as the purification circuit.
Most studies of FAC have been performed under
feedwater conditions, which generate high rates of
attack that can reach several millimeters per year in
some situations. Neutral chemistry, low-oxygen conditions at about 140  C, as may be found in BWR

feedtrains, can give high FAC rates, so dual-cycle
PWRs or PHWRs routinely add an amine such as
ammonia to raise the pH in the secondary coolant
circuit. The actual pH employed depends upon the
materials of construction; for all-ferrous feedtrains,
a pH25  C from 9.3 to 9.6 is usually specified, but the
value is kept below 9.2 to avoid excessive corrosion of
copper-base alloys, if these are present. Also, to
achieve a more even distribution of additive around
the circuit, an amine (such as morpholine) with a

coefficient of distribution between the steam and
liquid phases closer to unity than that of ammonia
may be used.
Oxygen dissolved in the coolant is also a powerful
inhibitor of FAC; it has been added routinely to feedwater systems in BWRs and certain fossil boilers for
some time. Depending upon the rate of attack, levels
of oxygen from a few ppb to several tens of ppb are
sufficient to stifle FAC completely. Maintaining a
dissolved oxygen content >$30 ppb, which raises
the corrosion potential ECP in the feedwater system above the Fe3O4/Fe2O3 phase boundary in the
Pourbaix diagram in Figure 2, is particularly crucial
in BWRs operating on hydrogen water chemistry


Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

113

St37.2 (A414 Gr B-carbon steel)

Flow (kg h−1)
983

491

15Mo3 (A161 Gr T1–0.5% Mo)

907
756
605


378
302
227

15NiCuMoNb5
13CrMo44 (A213 Gr T12–1% Cr, 0.5% Mo)
10CrMo910 (A213 Gr T22–2.2% Cr, 1% Mo)
3000

4.0
pH = 9.04

pH = 7.0

2.0

Increasing flow

Loss rate (mm year−1)

3.0

1.0

Specific material wear rate (μg cm−2 h−1)

1000
300
100

30
10
3.0
1.0
0.3
0
90
(a)

100 110 120 130 140 150 160 170
Temperature (ЊC)

0.1
50
(b)

100

150

200

250

300

Temperature (ЊC)

Figure 4 Effect of temperature, flow rate (a) (data from Bignold, G. J. et al. In: Proceedings of the International Specialist’s
Meeting on Erosion-Corrosion of Steels in High-Temperature Water and Wet Steam, Les Renardie´res, France, 11–12 May;

EDF: France, 1982) and material (b) (data from Heitmann, H. G.; Schub, P. In: Proceedings of the Third Meeting on Water
Chemistry of Nuclear Reactors, pp. 243–252, Bournemouth, UK, October; British Nuclear Engineering Society (BNES):
London, UK, 1983) on single-phase flow-accelerated corrosion under different flow and chemistry conditions. Reproduced
from Dooley, R. B. Power Plant Chem 2008, 10(2), 68–89.

(HWC) with high rates of hydrogen injection into the
feedwater. If HWC is combined with noble metal
chemical addition (NMCA), the FAC risk is reduced,
since much lower hydrogen injection rates are then
adequate to mitigate SCC in stainless steel recirculation piping and reactor internals. (Recombination of
hydrogen and oxygen to lower the ECP requires the
radiation fields present in the RPV.) Oxygen levels
significantly above $50 ppb may increase the risk
of strain-induced corrosion cracking and corrosion
fatigue in CS & LAS feedwater piping (see Section
5.06.3). Furthermore, the deliberate addition of oxygen
to feedwater systems in dual-cycle reactors may
pose problems, since residual oxygen entering the
steam generators can provoke SCC of the high-alloy
steam-generator tubes. Nevertheless, severe FAC of

components in the feed train of advanced gas-cooled
reactors (AGRs) has been successfully mitigated since
the early 1980s by oxygen additions.13
Material properties have a significant impact
on FAC rates, but typically the plant operator has
no control over this (unless a replacement of piping
is an option). Certain elements in the steel can act
to retard FAC, as mentioned earlier; for example,
chromium is particularly effective and a concentration of 0.1% in the metal reduces FAC in 180  C

ammoniated water and water–steam mixtures at
pH25  C 9 by about 70%.14 Moreover, under CANDU
primary coolant conditions of 310  C and pH25  C 10.5
(adjusted with lithium), increasing the chromium content of SA-106 Grade B CS from 0.019% to 0.33%
reduces FAC by about 50%.11


114

Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

5.06.2.2.2 Mechanisms and models

As with uniform corrosion (discussed in Section
5.06.2.1), FAC is governed by the ability of the
oxide film to protect the metal. Magnetite forms on
the steel at the metal–oxide interface and is degraded
at the oxide–coolant interface by fluid flow effects
and by dissolution according to the general equation
[VI] (which indicates the dependence of the dissolved
species on pH under reducing conditions and which is
equivalent to eqn [II] for b ¼ 2). The turbulence
in the coolant and the solubility of the magnetite
are then paramount in determining the severity of
the attack.
Fe3 O4 þ 3ð2 À bÞHþ þ H2 ¼ 3½FeðOHÞb Šð2ÀbÞþ
þ ð4 À 3bÞH2 O
½VIŠ
with b ¼ 0, 1, 2, or 3.
Mass transfer is often assumed to control the

mechanism.15 This derives from the postulate that
the magnetite film attains a steady-state thickness as
it dissolves at the rate Rd at its outer surface in
coolant undersaturated in dissolved iron and forms
continuously at the metal–oxide interface at the
same rate Rg. Since the magnetite formation at
the metal–oxide interface accounts for only about
half of the corroded metal, the other half diffuses
through the magnetite to the oxide–coolant interface, and with the iron from the magnetite dissolution is transported to the bulk coolant at the rate Rm.
The FAC rate RFAC is thus twice the dissolution rate
Rd of the magnetite at the oxide–coolant interface.
This concept of two processes in series – dissolution
Rd and mass transfer Rm– leads to the equation for
the steady-state FAC rate RFAC¼ dm/dt ¼ Rm¼ 2Rd
with all the variables in equivalent units of iron per
unit surface and time.
1. Steady-state assumption for the serial process:
Rg ¼ Rd ¼ 0:5Rm

½1Š

2. Dissolution rate of magnetite at the oxide–coolant
interface according to eqn [VI] (assuming firstorder kinetics):
Rd ¼ 0:5 Á dm=dt ¼ kd ðCeq À CÞ

½2Š

where kd is the dissolution reaction rate constant, C is
the concentration of Fe(II) species at the oxide–coolant
interface, and Ceq is their equilibrium concentration

according to eqn [VI], which corresponds to their
maximum solubility in the coolant.

3. Transport of Fe(II) species from the oxide–coolant
interface to the bulk environment by turbulent
mass transfer:
Rm ¼ h Á ðC À Cb Þ

½3Š

where Cb is the concentration of Fe(II) species in the
bulk coolant and h is the mass transfer coefficient,
which is dependent on flow conditions and geometry.
From eqns [1]–[3] it follows that:
RFAC ¼

h Á kd Á DC
ð0:5 Á h þ kd Þ

½4Š

where DC ¼ ðCeq À Cb Þ is the undersaturation in
iron. Models of FAC are based on the principles
behind eqn [4]. We expect that kd strongly increases
with temperature according to an Arrhenius law
for a thermally activated process (although there are
no data to confirm this over the temperature ranges
of interest), whereas h only shows a moderate
increase through the temperature dependence of the
properties in eqn [6]. If mass transfer controls, h is

small compared with kd (h ( kd) and the equation
reverts to:
RFAC ¼ hDC

½5Š

For a coolant of constant conditions containing little or
no dissolved iron (i.e., Cb % 0), the driving force DC
approaches a constant value – the solubility of the
oxide, Ceq – and RFAC varies as the mass transfer
coefficient (which increases with increasing flow rate
and turbulence). The mass-transfer model then
implies that the effects of materials composition and
coolant chemistry on FAC rate are brought about by
their effects on oxide solubility (Figure 5). According
to eqn [VI], the saturation concentration or solubility
Ceq depends on temperature, pH and H2 concentration
by simple chemical equilibrium thermodynamics.
Accordingly, the effect of chromium in the steel
can be attributed to the relative stability of mixed
oxides containing chromium (iron chromite, FeCr2O4,
for example, is virtually insoluble in reducing coolant
and accounts for the protection afforded by stainless
steel and similar alloys). As corrosion proceeds and
the magnetite dissolves, chromium is not leached out
in concert but continually concentrates in the film. It
is interesting to note that the inhibition occurs immediately at the start of exposure and continues at about
the same level, suggesting that the mechanism is the
rapid formation at the metal–oxide interface of a more
protective layer of oxide that is maintained throughout exposure.16 It appears that the higher the chromium content of the steel, the more protective that



Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

100
NH3
(mg kg−1) pH
0.1
80

Fe (μg kg−1)

60

8.75

0.2

8.90

0.3

9.00

0.5

9.20

1.0


9.40

2.0

9.60

300

350

40

20

0
0

50

100

150

200

250

Temperature ( ЊC)

Figure 5 Solubility of magnetite/iron as a function of

temperature at various ammonia concentrations.
Reproduced from Dooley, R. B. Power Plant Chem 2008,
10(2), 68–89.

oxide layer; hence, the immunity of stainless steel and
certain LAS like 13CrMo44 or 10CrMo910. These
observations cannot be fully explained with the DC
term in the simple mass-transfer model of eqn [5],
which would indicate an increasing inhibition with
time as the chromium concentrates in the film. One
mode of inhibition of FAC by additives in the coolant
may be via incorporation in the magnetite to make it
less soluble. From the experience with chromium,
whatever species is added should be available to affect
the metal–oxide interface consistently, presumably by
being kept permanently in solution in the coolant.
So far, titanium has shown promise as a coolantborne inhibitor of FAC under CANDU primary coolant conditions. Its effectiveness has been attributed
to its ability to form a mixed oxide with iron –
ulvo¨spinel – along with the magnetite that forms
on corroding CS.17 An in-plant demonstration of
titanium addition to a CANDU primary system is
described in Section 5.06.2.3.
The simple mass-transfer model also indicates
that temperature should affect FAC partly through
its influence on magnetite solubility. In ammoniated

115

water at pH25  C 9.0, there is a strong temperature
dependence and the maximum FAC rate occurs

between 130 and 140  C, depending on flow rate.18
The solubility of magnetite under the same conditions increases from the range 5–15 ppb at 25  C to
a maximum of about 30 ppb that persists over the
range 110–150  C,9,19 while mass transfer coefficients
at the same mass flow should approximately double
between 25 and 140  C. Similarly, in neutral water, the
maximum attack for several materials occurs at about
150  C,20 while the magnetite solubility increases from
about 70 ppb at 25  C to a maximum of about 140 ppb
at 120–130  C. The rough correspondence between
the temperatures of maximum FAC rate and of maximum magnetite solubility, as well as the effect of
temperature itself on solubility, indicate the strong
influence of oxide film dissolution on the FAC mechanism. It is likely that at low temperatures dissolution
rates of magnetite are low enough for kd to have an
effect through eqn [4] and lower the flow dependence
accordingly.18
The inhibiting effect of amines and high pH at
feedwater temperatures should also be realized
mainly through the solubility of magnetite. Thus, in
neutral water at 140  C, the solubility of magnetite is
about 119 ppb, but if the pH25  C is raised with ammonia to 9.2, the value falls to the range 14–26 ppb.9,19
This would suggest that a reduction in FAC rate by a
factor of 8.5–4.5 might be expected from ammoniating the coolant to pH25  C 9.2; however, experiments
indicate a reduction by a factor of only about 2.16
It is also instructive to consider the mass transfer
implications of the model according to eqn [5]. Mass
transfer in pipe flow in aqueous systems can be
described via a correlation of the mass transfer coefficient h with dimensionless numbers:
h ¼ Sh


D D
¼ ARe p Sc q
d
d

½6Š

in which the Sherwood number, Sh (given by hd/D,
where d is the pipe diameter and D is the diffusivity),
characterizes mass transfer in terms of the flow (via the
Reynolds number, Re, given by dur/m, where u is the
coolant velocity, r is the density, and m is the viscosity)
and physical properties (via the Schmidt number, Sc,
given by m/(rD)); A, p, and q are constants.
Typically, experiments on mass transfer of dissolved species yield values between about 0.6 and
0.9 for the exponent p.21,22 Recent experiments in a
water loop on FAC under neutral conditions at
140  C, however, indicated that the FAC rate RFAC
correlated rather weakly with Re.1.2,23 An alternative


116

Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

mass transfer analysis gave an excellent correlation
with fluid shear stress at the pipe wall, t:
Ru ¼ Pt

½7Š


FAC rate ϫ Flow rate (mm year–1ϫm s–1)

where P is a constant (see Figure 6). Thus, a steel
containing 0.019% chromium gave the correlation
RFACu ¼ 0.07t, while a steel containing 0.001% chromium in parallel experiments gave RFACu ¼ 0.18t,
where RFAC is in units of millimeters per year, u is
in meters per second, and t is in pascals.16
The predominance of mass transfer in developing
such correlations depends upon the dissolution rate
constant, kd in eqn [4], being large enough to make
the mass transfer coefficient, h, controlling. This
would seem to be valid under neutral chemistry conditions, where the solubility of magnetite is high, but
under high-pH conditions, where the solubility is
reduced, kd may be reduced also and its influence
may become significant. However, although recent
80

indications24 are that FAC in 140  C ammoniated
water at pH 9.2 is not correlated well by the simple
mass-transfer model leading to eqn [6], those experiments also indicated a greater dependence on flow
rate or shear stress, viz., t raised to the power 1.5–2.0.
This cannot be attributed to an increasing influence
of kd in eqn [4]; apparently, a different mechanism is
involved.
Surface texturing usually accompanies FAC. In
steam–water mixtures, ‘tiger-striping’ is caused by
the streaming pattern of the liquid film on the surface, while in single-phase water, ‘scalloping’ sculpts
the attacked surface with grooves, flutes, or shallow
depressions (Figure 7(a) and 7(b)). However, in

experiments in neutral water at 140  C, in which corrosion rates of several millimeters per year were seen
in tubular test sections, a low-chromium steel developed no scallops, even though it corroded at more than
twice the rate of a higher chromium steel that developed distinct scallop patterns.23 The scalloping that
was seen was approximately related to the pipe flow
via a characteristic ‘scallop Reynolds number’:

70

Resc ¼ 1:55 Â 104 Æ 2:6 Â 103

60
50
40
30
20
10
0
0

200

400

600

800

1000

1200


Shear stress τ (N m−2)

Figure 6 Correlation for flow-accelerated corrosion at
140  C in neutral water: carbon steel with 0.019% Cr.

in which the characteristic dimension is the average
scallop spacing. While the scallops were formed by the
corrosion of the metal, it was significant that distinct
oxide forms developed and were related more to scallop crests than to valleys. Those forms, shown in
Figure 7(c), occurred over pearlite grains in the
metal and may be described as ‘coral-like.’ They provide further confirmation of the importance of oxide
dissolution in the mechanism, since they are no doubt
formed by the different solubilities of the different
compositions of oxide overlaying the lamellae of
cementite and ferrite in the pearlite. As the magnetite

200 μm

(a)

½8Š

3 μm

(b)

Diameter of piping

(c)


Figure 7 Surface texturing in flow-accelerated corrosion. (a) ‘Scalloping’ or ‘horseshoe pits’ in single-phase flow. (b) ‘Tiger
striping’ in two-phase flow. (c) ‘Coral-like’ oxide on carbon steel undergoing flow-accelerated corrosion in neutral water at
140  C. (a) and (b) courtesy of COMSY™ with permission of AREVA NP.


Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

generally dissolves, that over the cementite lamellae is
less soluble and is left standing proud. It was noted in
the experiments that the ‘coral’ oxide concentrated
about 50% more chromium than the surrounding
oxide on the ferrite grains, possibly because the underlying pearlite contained more chromium as carbide
associated with the cementite.
Loop studies using tubular test sections of the
material of interest16 under reactor feedwater conditions establish the basis for adding oxygen with minimal residual concentrations left at the end of piping
systems. At 140  C in neutral water, about 40 ppb of
dissolved oxygen are required to stifle FAC, whereas at
pH25  C 9.2 with ammonia, only 1–2 ppb are required.
The concentrations required for stifling are related to
the measured FAC rates and it is clear that – as oxygen
is progressively added to the corroding system – the
cathodic reaction of water being reduced to hydrogen
is replaced by oxygen reduction; at the stifling concentration, the oxygen sink disappears and with continuing
addition its concentration in the loop jumps sharply.
However, although there is an obvious relationship
between the FAC rate at stifling and the stoichiometric
flux of oxygen by mass transfer to the surface, a
straightforward linear correspondence may not apply.13
While several mechanistic models of FAC in feedwater systems based mostly on the principles behind

eqn [4] have been developed, empirical models have
been applied extensively for some time. In the 1980s, for
example, parametric studies at the laboratories of the
then Siemens-KWU led to the formulation of a correlation between pipe wall thinning Dd and the system
variables u (flow velocity), T (temperature), pH, O2
(oxygen concentration), M (materials composition –
Cr, Mo, and Cu), and t (exposure time):
Dd ¼ kc f ½u; T ; pH; O2 ; M; t Š

½9Š

where kc is a geometry factor. The correlation
was developed initially from data for single-phase
water flow, but was adapted to two-phase steam-water
flows, with the bulk velocity u substituted by the mean
velocity of the annular film of water covering the pipe
wall. The resulting computer code, ‘WATHEC,’ was
restricted to steels with the content of Cr plus Mo less
than 5% and exposure times greater than 200 h. The
predictions of wall thinning for a large number of
situations were equal to or greater than the measured
values in 85% of the cases – in other words, the code
was considered to be suitably conservative.25 Later, the
data management tool ‘DASY’ was added to the code.
The EPRI-sponsored computer code ‘CHECWORKS™’ combined an empirical equation, which

117

had some basis in mechanisms such as that leading to
eqn [4], with a comprehensive data management

scheme.26 The data management includes analysis
of ultrasonic test data, calculation of critical wall
thickness for components at risk, and organization
of pertinent databases. The FAC rate RFAC is written
as a function of the system variables:
RFAC ¼ f ½T ; AC; MT; O2 ; pH; G; aŠ

½10Š

where T is temperature; AC is alloy content of Cr, Mo,
and Cu; MT is mass transfer; O2 is concentration of
dissolved oxygen; G is a geometry factor; and a is the
steam void fraction. The factors in eqn [10] are interrelated and the equation is nonlinear. While the absolute
predictions of RFAC in CHECWORKS™ are not generally of high precision, iterations incorporating plant
measurements can identify the locations of risk and
can rank components in the order of vulnerability.27
The FAC of CS is most pronounced under feedwater conditions, but it also occurs at higher temperatures in the primary coolant systems of PHWRs.
The phenomenon was identified in the late 1990s at
the Point Lepreau CANDU in New Brunswick,
Canada, where surfaces of affected outlet feeders of
CS were scalloped and the wall thinning rates plotted
against coolant velocity indicated a dependence on
the velocity raised to the power $1.5.28 Regions of
high turbulence, such as the tight-radius bends close
to the reactor face, were more severely affected.
It was also noted that the coolant at the core outlet
was unsaturated in dissolved iron, since it entered the
core at 265  C saturated after its passage through the
steam generators of nickel alloy and the inlet feeders
of CS; as its temperature rose in the fuel channels

the solubility at the high pH rose in concert (the
CANDU core contains no iron-bearing alloys, so it
cannot act as a source of dissolved iron).
Although the high-turbulence (and therefore highmass-transfer) regions are again the most affected in
primary coolant FAC, it is unlikely that the mechanism
in primary coolant is straightforward mass-transfer
control based on eqn [11]. First of all, the velocity
dependence is too high (the power of 1.5 rather than
0.6–0.9 as expected from correlations such as eqn [6]).
Second, measurements of the dissolution rate of magnetite under chemistry conditions close to those of
CANDU coolant29 have given a value of kd in eqn [4]
very much lower than the mass transfer coefficient h,
which would put the mechanism squarely under dissolution control with no velocity effect at all. The
alternative theory proposed is that dissolution of magnetite works in synergy with fluid shear stress at the


118

Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

surface to degrade the oxide. Thus, the loosening of
the magnetite crystallites in the film makes them susceptible to removal by the fluid forces and as they are
eroded away the film becomes less protective. A mathematical model developed on this principle was able to
predict quite well the thinning of the walls of outlet
feeders at an operating plant in terms of the development of the oxide film, the pattern of attack around
representative bends, and the corrosion potential ECP
and velocity-dependence of FAC rate in individual
feeders.30 The model was adapted for predicting corrosion under conditions when the coolant is saturated in
dissolved iron and gave reasonable predictions of oxide
film growth and general corrosion in CANDU inlet

feeders, where corrosion rates are quite low.
It is probably more than a coincidence that FAC
under these primary coolant conditions, when magnetite solubility is low, seems not to be controlled directly
by mass transfer, while similar indications apply under
feedwater conditions at high pH, when solubilities
are also low. The parallel between the two situations
could be clarified if measurements of kd under feedwater conditions were available and the measurements
under primary coolant conditions were verified.
The high rate of general corrosion of CS caused
by aerated concentrated solutions of boric acid originating from leaking PWR coolant was described in
Section 5.06.2.1. Some of the studies that quantified
the attack were done with dynamic systems, such as
evaporating sprays, and it became clear that flow has
an effect.2 Of immediate concern is the corrosion of
RPV steel caused by borated coolant leaking through
cracked penetrations housing control rod drive
mechanisms. At the Davis Besse PWR in 2002, such
corrosion had threatened the integrity of the vessel.
The sequence of events that can lead to cavity formation next to a nozzle was postulated12 to be in three
phases: initially, slow seepage of coolant into the external annulus (crevice) in the head would be accompanied by low corrosion rates; next, when the crevice had
opened enough and the crack had lengthened to give
substantial leak rates, an evaporating coolant jet would
accelerate the attack through flow effects; finally, leakage into a cavity would create a turbulent evaporating
pool, extending the attack sideways.
An extensive testing program sponsored by the
Electric Power Research Institute (EPRI), Palo Alto,
California, investigated the phases of boric acid
attack at Davis Besse. The second phase, which experienced substantial flow effects, was simulated with
laboratory experiments in which a flashing jet of
borated coolant was directed onto a heated sample


of pressure-vessel steel and the damage assessed
in terms of system parameters – notably, coolant
chemistry and flow rate.31 Volumetric (or massive)
metal loss was correlated with volumetric coolant
flow and seemed to behave differently from metal
penetration, which was correlated with jet velocity.
FAC was in evidence through miniature scallops in
the damage craters that formed around (but some
distance away from) the points of jet impact. Metal
loss rates attained about 3 cm3 yearÀ1 at a flow rate of
200 ml minÀ1 with a boric acid concentration equivalent to 1500 ppm [B] and pH300  C of 6.9 adjusted with
lithium; the rate depended on the volumetric flow
in the jet raised to the power 0.84. Under the
same chemistry conditions, the penetration rate
reached 200 mm yearÀ1 at a jet velocity of 140 m sÀ1
and the two were correlated via the velocity raised to
the power 4.3. It was notable that neither pH300  C nor
the boron concentration was the controlling chemistry parameter; rather, it was the ratio [B]/[Li].
5.06.2.2.3 Service experience and mitigating
actions

Many incidences of feedwater pipe thinning by FAC
from two-phase coolant were reported in the 1980s.
In 1985 March, a line downstream of a level control
valve for a feedwater heater at the Haddam Neck
PWR actually ruptured because of FAC induced by
flow-impingement. However, the first major incident
in a nuclear plant was the catastrophic pipe break at
the Surry Unit 2 PWR in December 1986, which led

to five deaths and several injuries. The 0.46 m diameter line thinned and ruptured at an elbow, 0.3 m from
a 0.6 m header, as a section of the pipe wall 0.6 Â 1.2 m
was blown out. Until then, FAC by steam–water mixtures had been considered to be more serious than
FAC by single-phase coolant. Six months later, excessive thinning of a feedwater line was reported at the
Trojan plant and, in September 1988, Surry Unit 2
reported 20% wall loss in the suction line to a feedwater pump over a 1.2-year period.
Reports of serious thinning of feedwater piping
continued after the Surry incident, even though
plant inspections had generally become more rigorous
and chemistry control had tightened. In May 1990,
the Loviisa Unit 1 WWER (Eastern type PWR) in
Finland suffered a break in a 0.3 m diameter line in
the turbine hall, releasing about 50 m3 of steam and
water into the building, and in February 1993, a
similar incident occurred in Unit 2.
The latest major FAC incident in a nuclear plant
was the rupture of a feedwater line at the Mihama


Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

Unit 3 PWR in 2004, which led to four deaths and
seven serious injuries.32 The thinning of the pipe wall
from 10 to 0.6 mm by FAC caused a large section to
peel back after rupture, allowing the coolant at

Figure 8 The ruptured feedwater line at Mihama-3.

140  C to flash to steam on release. The damaged
pipe is shown in Figure 8. Maximum thinning rates

were observed just downstream of an orifice plate,
where turbulence intensity was high. It should be
noted that chemistry had been maintained at high
pH with ethanolamine and that hydrazine was used to
scavenge oxygen. However, the location had not been
inspected since the plant start-up in 1976. The possibility of adding oxygen to the feedwater is being
considered and inspection procedures have been
revised extensively.
In 1997, at the Point Lepreau CANDU PHWR
in New Brunswick, Canada, the outlet feeder pipes
of CS that carry the heavy water coolant from the
core to the steam generators were found to be corroding excessively. The same problem has since then
been seen at all CANDUs in operation before 2000.
Figure 9 shows the arrangement of pipes at a reactor
face. The feeders are about 76 mm diameter and carry

3

4

1

119

2
5

9

7


6

8
10

1. Reactor outlet header
6. Calandria end shield face
2. Reactor inlet header
7. Tube spacers
3. Reactor outlet header
8. Support brackets
4. Reactor inlet header
9. Walkway
5. Feeder tube upper supports 10. End fittings

Figure 9 CANDU reactor face; cutaway view of feeder pipe arrangement, with permission from Atomic Energy of
Canada Limited.


120

Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

coolant at velocities between about 12 and 22 m sÀ1.
At the coolant temperature of 310  C and pH $ 10.6
(adjusted with lithium), the solubility of magnetite is
relatively low – about 1.7 ppb – and the FAC rates
accordingly attain only about 120 mm yearÀ1. The
attack is only a fraction of that observed under

feedwater conditions and has caused no safety issues,
but it means that feeder integrity may be lifelimiting and has necessitated replacements at some
reactors. Mitigating actions have been taken by
reducing the primary coolant alkalinity to the bottom of the recommended pH range, where the magnetite solubility is close to the minimum, while
plants in service since 2000 employ feeders with
a relatively high chromium content ($0.3% in
contrast with the $0.02% in earlier reactors).
One trial of a coolant additive was made at the
Darlington Unit 3 900 MW CANDU in 2002,
when a titanium dioxide slurry was added to one
channel to give a concentration of ten or so microgram per kilogram at the outlet feeder33; significant
reductions in FAC rate were recorded, but it
was decided not to undertake the further development needed to proceed to the next stage of
full-plant addition. Meanwhile, feeder replacement
has become a manageable – if costly – operation at
severely affected plants.
The boric-acid corrosion at the Davis Besse PWR
in Ohio, which in 2002 was found to have a large
wastage cavity in the RPV head adjacent to a penetration housing a control rod drive mechanism, was
postulated to be partly due to FAC (see Figure 10).
About 2040 cm3 of the pressure-vessel steel had
corroded away and about 106 cm2 of stainless steel
cladding were exposed at the bottom. Substantial
quantities of solid boric acid had deposited close by.
Subsequent investigations12 determined that coolant
had leaked from a crack in the adjacent Alloy 600
nozzle into the surrounding annulus and in time
had widened it. The crack was an example of primary
water stress corrosion cracking (PWSCC), of which
numerous incidences have been recorded in PWRs.

No simple means of mitigation have been proposed
for existing plants, since the coolant boron level is
fixed by reactivity considerations. Long-term prevention entails avoiding operating with coolant
leaks (as is, in fact, the regulation in some jurisdictions), for example, through minimizing the possibility of PWSCC by using less-susceptible Alloy 690
material for the CRDM penetrations. In the meantime, more rigorous inspection regimes are being
implemented.

Figure 10 Cavity in the reactor pressure vessel head at the
Davis Besse pressurized water reactor.

5.06.3 Localized Corrosion and
Environmentally Assisted Cracking
5.06.3.1

Pitting

In CS & LAS, a shallow form of pitting can occur in
the complete absence of anionic water impurities as
the electrochemical corrosion potential (ECP) at the
steel surface is raised, for example, through oxygen
and other oxidizing species. Such corrosion pits often,
but not exclusively, initiate at MnS inclusions which
intersect the steel surface. Figure 11, originally compiled by Hickling, shows the critical boundaries
between uniform surface attack and shallow pitting
in high-purity water at low flow rates as a function of
temperature and dissolved oxygen level. The critical
oxygen concentration for pitting drops with decreasing temperature and is further reduced by a simultaneous mechanical straining of the surface, or by
increased sulfate and chloride impurity levels. Furthermore, pitting in CS & LAS is favored by high
steel sulfur contents and quasi-stagnant flow conditions. In the absence of impurities, increasing the flow
rate of water across the metal surface mitigates the

aforementioned form of pitting corrosion. Alkalization


Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

shifts this boundary to much higher values, as does the
introduction of buffering and passivating species (e.g.,
on the secondary side of steam generators).34–37
Some degree of pitting corrosion is inevitable
after long-term exposure of unclad CS & LAS surfaces to water in LWR systems and is not usually a
threat to either coolant purity or to structural integrity. Shallow pitting has been observed primarily in
specific piping systems with residual water because of

121

incomplete draining during nonoperational periods
(shutdown corrosion). This can be avoided by adequate wet or dry preservation techniques. If pitting
happens during normal plant operation at high temperatures, however, it indicates conditions under
which EAC may also occur (since this is controlled
by similar parameters, see Section 5.06.3.2) and can
even be directly implicated in the initiation of EAC
(Figure 12).

300

ck

ce

ta

at

g

ce

tin

fa

pit

ng

ur

tti

ls
ra
ne
Ge

Pi

ne

ra

ls


ur

fa

200
Ge

Temperature (ЊC)

at

ta

ck

Pitting with
straining

Pitting
without
straining

100

SSRT

0
10


lzumiya, Tanno
Videm
Mizuno et al.
Ford

Coupon
specimens

1000

100

10 000

Dissolved oxygen content DO (ppb)
Figure 11 Boundaries between uniform corrosion and pitting attack in carbon and low-alloy steel in quasi-stagnant
high-temperature water. Compiled from The general and localized corrosion of carbon and low-alloy steels in
oxygenated high-temperature water. EPRI-NP-2853; EPRI: Palo Alto, CA, 1983; Electric Power
Research Institute. BWR environmental cracking margins for carbon steel piping. EPRI Report NP-2406; EPRI: Palo Alto, CA,
1982; Indig, M. et al. Rev Coat Corros 1982, 5, 173–225; Videm, K. In Proceedings of the 7th
Scandinavian Corrosion Congress, Trondheim, Norway, May 26–28, 1975; pp 444–456.

(a)
20 mm

(b)

Acc.V Spot Magn
20.0 kV 5.9 1696x


20 mm
Det WD
SE 11.0 5/1;250C;HigN

Figure 12 Strain-induced corrosion cracks initiating from a corrosion pit or a (dissolved) MnS inclusion at the surface of a
low-alloy and carbon steel in high-temperature water in slow strain rate experiments. Adapted from Congleton, J. et al.
Corros. Sci. 1985, 25, 633–650; Atkinson, J. D. et al. In Proceedings of the 13th International Conference on Environmental
Degradation of Materials in Nuclear Power Systems, Whistler, British Columbia, Canada, Aug 19–23; King, P., Allen, T.,
Busby, J., Eds.; The Canadian Nuclear Society: Toronto, Canada, 2007; CD-ROM.


122

Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

5.06.3.2 Environmentally Assisted
Cracking
5.06.3.2.1 Basic types of EAC and major
factors of influence

EAC is used here as a general term to cover the
full spectrum of corrosion cracking ranging from
stress corrosion cracking (SCC) under static load to
corrosion fatigue (CF) under cyclic loading conditions (Table 2).38,39 Strain-induced corrosion cracking (SICC) involving slow, dynamic straining with
localized plastic deformation of material, but where
obvious cyclic loading is either absent, or is restricted
to a limited number of infrequent events such plant
startup and shutdown, is increasingly used as an
appropriate term to describe the area of overlap
between SCC and CF.38,39

Under critical parameter combinations, EAC is
observed in all wrought and welded CS & LAS in
high-temperature water. The EAC crack path is
usually perpendicular to the direction of maximum
Table 2

tensile stress and transgranular in nature, with a
quasicleavage appearance showing a feathery morphology at high magnifications. The general fracture
appearance is similar for SCC, SICC, and even CF
(at least for strong environmental acceleration of
fatigue crack growth), thus confirming that EAC is
governed by the same basic process for all three
loading modes. In the case of cyclic loading at frequencies !$10À3 Hz, the fracture surface also usually contains both ductile and brittle fatigue striations,
which are perpendicular to the local crack-growth
direction.38
EAC initiation and growth in CS & LAS are
governed by a complex interaction of environmental,
material, and loading parameters, and most influencing factors are both interrelated and synergistic. The
major parameters of influence, which have been
identified so far, are summarized in Table 3.38,39
The effect of these parameters on EAC initiation
and crack growth (including key thresholds) is

Basic types of EAC in CS & LAS and relevant nuclear codes

Mechanism

Environmentally assisted cracking (EAC)
CF
Corrosion fatigue


SICC
Strain-induced corrosion cracking

SCC
Stress corrosion cracking

Type of loading

Cyclic: low-cycle, high-cycle

Static

LWR operation
condition
Characterization of
crack growth
Characterization of
crack initiation

Thermal fatigue, thermal
stratification, . . .
ASME XI
Code Case N-643 (PWR)
ASME III
Fen-approach

Slow monotonically rising or very
low-cycle
Start-up/shut-down, thermal

stratification
High-sulfur line of F & A model as
upper bound
Susceptibility conditions:
ECPcrit, de/dtcrit, ecrit

Transient-free, steady-state
power operation
BWRVIP-60 disposition lines
scrit

Source: Seifert, H. P.; Ritter, S. Research and service experience with environmentally-assisted cracking of carbon & low-alloy steels in
high-temperature water. SKI-Report 2005:60; SKI: Stockholm, Sweden, 2005; ISSN 1104-1374. alsakerhetsmyndigheten.
se/.
Hickling, J. et al. PowerPlant Chem. 2005, 7, 31–42.

Table 3

Major influencing factors for EAC in C & LAS

Environmental parameters

Material parameters

Loading parameters

Corrosion potential, dissolved oxygen
content
Temperature


S-content, morphology, size, spatial distribution
and chemical composition of MnS

Frequency, loading or strain
rate
Level of load, KI, stress,
strain, DK
Type of loading

2À
À
ClÀ, SO2À
4 , S , HS

Flow rate

Susceptibility to dynamic strain ageing,
Concentration of interstitial C and N
Hardness/yield stress if >350HV5/800 MPa

Residual stress

Source: Seifert, H. P.; Ritter, S. Research and service experience with environmentally-assisted cracking of carbon & low-alloy steels in hightemperature water. SKI-Report 2005:60; SKI: Stockholm, Sweden, 2005; ISSN 1104-1374. />Hickling, J. et al. PowerPlant Chem. 2005, 7, 31–42.


Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

discussed in detail in Seifert and Ritter38 and an
interpretation of their synergism is given both there
and in Section 5.06.3.2.4.

5.06.3.2.2 Corrosion fatigue and straininduced corrosion cracking
5.06.3.2.2.1 Initiation and susceptibility
conditions

Slow strain rate (SSR)38,40 and low-cycle fatigue
(LCF) tests40–42 with smooth specimens have clearly
shown that CF and SICC can occur in CS & LAS in
oxygenated, high-purity, high-temperature water if
the following conjoint threshold conditions are
simultaneously satisfied:
 Water temperature >$150  C. In LCF experiments, susceptibility then increases with temperature up to $320  C. SSR tests, on the other hand,
usually indicate a maximum of susceptibility
between 200 and 270  C, depending on strain rate.
 Corrosion potential > ECPcrit ¼ $À200 mVSHE or
dissolved oxygen content >$30 ppb. Above this
threshold, EAC susceptibility then generally
increases with increasing ECP/oxygen content,
but saturates at very high levels.
 Loading which leads to (local) macroscopic strains
at the water-wetted surface above the elastic limit.
The susceptibility then increases strongly with the
degree of plastic strain. SSR experiments with
tapered specimens, and LCF tests with different
waveforms, indicate a minimum critical strain of
0.15–0.2%, which is in a similar range to typical
oxide film rupture strains on CS & LAS ($0.05–
0.2%) in high-temperature water.
 Positive strain rates below $10À3 sÀ1. The EAC
susceptibility then increases with decreasing strain
rate de/dt. In most LCF investigations, saturation

of the decrease in fatigue life is observed below
a strain rate of 10À5 sÀ1, but SSR tests indicate a
maximum of susceptibility between 10À5 and
10À7 sÀ1, depending on ECP and temperature.
 EAC susceptibility increases with increasing steel
sulfur content and a lower threshold is often quoted
at $0.003 wt%, but experimental evidence for the
latter is weak. Distinct material susceptibility
to dynamic strain aging (DSA) in the critical
temperature/strain-rate range, or a low yield
stress, may also favor crack initiation by SICC.
If one or other of these conjoint threshold conditions
is not satisfied, SICC initiation is extremely unlikely
and no, or only a minor, environmental reduction of

123

fatigue life is observed in high-temperature water.
Furthermore, in high-purity water, a high flow rate
may completely suppress SICC susceptibility and significantly retard CF crack initiation (in particular,
for small strain amplitudes or slow strain rates) compared to quasi-stagnant conditions, since the risk for
the formation of an aggressive, occluded water chemistry within small surface defects is significantly
reduced by convection. Note, however, that high levels
of chloride or sulfate may extend the range of susceptibility to less severe conditions (e.g., to lower ECP and
strain levels).
The range of system conditions where EAC crack
growth from incipient cracks may occur is significantly extended compared to the initiation susceptibility conditions specified earlier. For example,
CF crack growth has been observed in high-purity
PWR water at ECPs below À500 mVSHE under certain cyclic loading conditions (10À2 to 10 Hz).38
Apart from local stress raisers such as welding

defects, which may help overcome the strain threshold
in the field, the effect of initial surface condition (surface roughness, cold work, residual stress, oxide film,
and preoxidation) on SICC and CF initiation is much
less pronounced than for (high-cycle) fatigue in air, or
with SCC of stainless steels or Ni-base alloys. SICC
cracks usually, but not exclusively, initiate at MnS
inclusions or corrosion pits.38,40,43,44 Pitting, particularly if occurring actively, therefore facilitates SICC
initiation (Figure 12). CF cracks, on the other hand,
initiate mainly along slip bands, carbide particles, or at
the ferrite–pearlite phase boundary, and less frequently
at micropits or MnS inclusions.40–42 The effect of
pitting and MnS inclusions on CF initiation is thus
moderate, but may become more pronounced in the
case of deep, high-aspect-ratio pits, mild environmental conditions, or at small strain amplitudes.40
5.06.3.2.2.2 SICC initiation and crack growth from
incipient cracks

In high-purity water in the absence of any significant
fatigue contribution, CS & LAS show distinct SICC
susceptibility only in highly oxidizing environments.
For example, it is almost impossible to initiate relevant
SICC crack growth in precracked specimens in slow
rising-load tests with constant load rate at KI values
<$70 MPa m1/2 in high-purity water at an ECP of
<$À100 mVSHE. Even under highly oxidizing conditions (ECP ! þ50 mVSHE), KI values of $25 MPa m1/2
have to be exceeded to initiate SICC in slow, risingload experiments in high-purity water. A maximum
in SICC initiation susceptibility (i.e., a minimum in


124


Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

(a)

101

102

60

103

SA 533 B Cl. 1, 0.018% S
288 ЊC, 8 ppm O2’ 65 ppbSO24−

10−6

50

10−7

45
10−8

40
35
30
10−7


10−9
10−6

10−5

10−4

10−3

−1

Crack opening displacement rate dCODLL/dt (mm s )

(b)

10−8

50

40

10−9

30
Weld filler material, 0.007% S
dCODLL/dt = 2–4 ϫ 10−6 mm s−1
20
100

10−2


8 ppm O2, 65 ppp SO24−

150

200

250

300

SICC crack growth rate (ms−1)

55

100

Stress intensity factor at onset of
SICC crack growth Kl,i (MPam1/2)

60

SICC crack growth rate (m s−1)

Stress intensity factor at onset of
SICC crack growth Kl,i (MPa m1/2)

Stress intensity factor rate dKl/dt (MPa m½ h−1)
10−1


350

Temperature (ЊC)

Figure 13 Effect of loading rate (a) and temperature (b) on Strain-induced corrosion cracking initiation and crack growth
from incipient cracks in oxygenated high-temperature water in a low-alloy steel reactor pressure vessel steel. Reproduced
from Seifert, H. P.; Ritter, S. J. Nucl. Mater. 2008, 378, 312–326.

the KI,i value needed to initiate SICC) occurs at intermediate temperatures around 200–250  C and at slow
loading rates (Figure 13). Furthermore, the SICC
initiation susceptibility is affected by environmental
and material parameters in a very similar way as in
SSR tests with smooth specimens.38,45
Under highly oxidizing conditions (ECP ! þ
50 mVSHE), SICC cracks can grow without any significant fatigue crack growth contribution and, in extreme
cases, SICC growth rates during plant transients with
severe thermal stratification can reach high values (up
to several hundreds of micrometers per event, or several millimeters per day). Under such circumstances,
SICC crack growth rates in both high-purity water and
water with increased sulfate or chloride levels are
comparable for all CS & LAS, independent of their
sulfur content. Once growth is initiated, crack velocities rise with increasing loading rates dKI/dt (and thus
crack-tip strain rates) and with increasing temperature
(Figure 13), at least up to $250  C.38,45
5.06.3.2.2.3 CF initiation and crack growth from
incipient cracks

Under cyclic loading, CS & LAS show a distinct
susceptibility to initiation of corrosion fatigue (CF)
from incipient cracks in all LWR environments.

However, environmental acceleration of fatigue
crack growth in these materials only occurs for
certain combinations of loading and environmental parameters. The cycle-based crack growth rate
da/dN then usually depends strongly on loading frequency and temperature, in contrast to fatigue in air,
and can achieve 10–100 times higher values under
certain conditions (although the actual effects on

component integrity in the field are much less severe
than might appear at first glance from isothermal test
data). The combination of material, loading, and
environmental conditions, where strong environmental acceleration of fatigue crack growth (factor of
!10) usually occurs, extends over a broad range in
the case of highly oxidizing BWR/NWC conditions,
but is restricted to a much narrower range at low
ECPs (Figures 14 and 15(a)), that is, in the case of
PWRs and of BWRs operating on HWC (with or
without NMCA (noble metal chemical addition)).
In contrast to SICC initiation, there is no threshold ECPcrit for CF in high-purity water, and material
parameters usually play a less pronounced role
(Figure 15(b)).46,47 The major parameters affecting
CF crack growth are shown in Table 4, and the
observed cracking behavior can be briefly summarized
as follows.38,47
Above an upper critical frequency ncrit,H of
$1–100 Hz and/or an upper DKCF,H threshold, environmental acceleration of fatigue crack growth disappears, because air fatigue crack growth rates are
higher than the maximum EAC rates for CS & LAS
in high-temperature water. These upper ncrit,H and
DKCF,H thresholds are shifted to lower values with
increasing DK and frequency, respectively, and do not
depend noticeably on environmental and material

parameters.
Strong environmental acceleration of fatigue
crack growth (factor of !10) in LWR environments
occurs with all CS & LAS below these upper thresholds for the combination of temperatures !$100  C,
loading frequencies ! ncrit,L ¼ f (ECP, DK, . . .), and
DK values ! DKCF,L ¼ f (ECP, n, . . .). If one or more


Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

102
0.4 or 8 ppm O2
+50 to + 200 mVSHE
0.2 ppm O2
0 mVSHE

ASME Xl ‘air’

10−4

10−3 10−2 10−1
Frequency n (Hz)

(a)

100

101

dK/dt , n


t’

10−2

T = 288 ЊC, R = 0.7 − 0.8, Δ K = 14 − 22 MPa m1/2
SA 533 B Cl. 1 (0.018% S), 20 MnMoNi 5 5 (0.015% S)

10−5

10−1

we

< 0.005 ppm O2
−500 mVSHE

100

Xl ‘

ASME Xl ‘Wet’

10−1

10−2

101

ME


100

SA 533 B Cl. 1,0.0018% S
250 ЊC, kout £ 0.07 μS cm–1
0.4 ppm O2

10−3

102
(b)

AS

101

da/dNCF (μm cycle–1)

da/dNCF (μm cycle–1)

102

125

r’

ai

E


‘
Xl

M

AS

Kl.max = 58 MPa m1/2
0.5 MPa m1/2h−1
5 MPa m1/2h−1
35 MPa m1/2h−1
370 MPa m1/2h−1

1
10
100
Stress intensity factor amplitude Δ K (MPa m1/2)

(a)

10−6

SA 533 B Cl.1 (0.018 wt % S)
T = 274 − 288 ЊC

10−8

10−10
/d


t Air

10−12

da

NWC, + 150 mVSHE
NWC, − 100 mVSHE
HWC, − 500 mVSHE

10−12
10−10
10−8
10−6
Fatigue crack growth rate in air da/dtAir (m s−1)

Corrosion fatigue crack growth rate in
high-temperature water da/dtCF (m s−1)

Corrosion fatigue crack growth rate
in high-temperature water da/dtCF (m s−1)

Figure 14 Effect of electrochemical corrosion potential, loading frequency (a) and DK and loading rate (b) on corrosion
fatigue crack growth in low-alloy steel in high-temperature water and comparison with ASME XI ‘Air’ and ‘Wet’, crack growth
rates. Reproduced from Seifert, H. P.; Ritter, S. Corros. Sci. 2008, 50, 1884–1899.

(b)

10−3
Superposition-model da/dtCF = da/dtENV + da/dtAir


10−5
10−7
10−9
10−11
10−13
10−13

T = 240 − 288 ЊC
0.4 − 8 ppm O2
<1 or 65 ppb SO42−
n = 3E − 6 to 8E − 3 Hz
240 ЊC
R = 0.2–0.8
288 ЊC 250 ЊC
ΔK = 11 − 62 MPa m1/2
r
SA
533
B
Cl.1
(0.018% S)
i
t
20 MnMoNi 5 5 (0.004% S)
/d A
20 MnMoNi 5 5 (0.015% S)
da

10−11


10−9

10−7

10−5

10−3

Fatigue crack growth rate in air da/dtAir (m s−1)

Figure 15 Corrosion fatigue crack growth in the time-domain as a function of electrochemical corrosion potential (ECP) for
a high-sulfur low-alloy steel (a) and as a function of material and loading parameters at high ECPs > $ þ 50 mVSHE (b).
Reproduced from Seifert, H. P.; Ritter, S. Corros. Sci. 2008, 50, 1884–1899.

of these threshold conditions are not satisfied, the
environmental effect is moderate (factor of 5), or
even absent altogether. When these conjoint threshold
conditions are simultaneously satisfied, the environmental acceleration of fatigue crack growth increases
with increasing temperature (sometimes with a maximum around 250  C) and decreasing loading frequency and DK values, often showing maximum
enhancement close to the lower thresholds.
The lower thresholds ncrit,L and DKCF,L depend on
material, environmental, and loading parameters (in
particular, ECP and DK or frequency – see Figure 14
and Table 4). They can reach very low values of
10À6 Hz and of $2 MPa m1/2 (for load ratio R!1)

or $5 MPa m1/2 (R! 0) under highly oxidizing conditions (ECP ! þ100 mVSHE). Below these lower
thresholds, the environmental acceleration of fatigue
crack growths drops to moderate values (factor of 5)

and the cycle-based crack growth rate da/dN becomes
independent of frequency. The DK-dependence is the
same as in air and the effect of temperature is usually
moderate. Such behavior is sometimes designated
as true corrosion fatigue and corresponds to the
low-sulfur curve of the Ford and Andresen model
(see Section 5.06.3.2.4; Ford and Andresen48 and
Ford49). Once conditions exceed these lower thresholds, the material behavior corresponds to the highsulfur curve of the model.


126
Table 4

Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

Major parameter effects on CF crack growth with typical parameter values
High-sulfur, !1–5 ppm S2À

0.1 ppm S2À

Crack-tip environment

Low-sulfur,

Range
da/dtCF/dat/dtAir

n < ncrit,L
2–5
6¼ f(n)

m $ 3, that is, parallel to air curve
6¼f(n), that is, n ¼ 0
$DKth, Air ¼ f(R)

da/dNCF ¼ BÁDKm
da/dNCF ¼ AnÀn
DKCF,L ¼ f(n, R, ECP, . . .)
ncrit,L ¼ f (DK, R, ECP, . . .)

No evidence for lower threshold for
low-sulfur behavior

Temperature

Moderate effect

ncrit,L n < ncrit,H $ 1 Hz
!10–100 (10 000)
n# ! da/dtCF/dat/dtAir"
m $ 1–2
n ¼ 0.5–0.6
n", R", ECP", SO2À
4 ", S", DSA"! DKCF,L#
!þ100 mVSHE: DKCF;L $ 2–5 MPa m1/2
À500 mVSHE: DKCF,L $ 10–30 MPa m1/2
ECP", SO2À
4 ", S" DK" ! ncrit,L#
!þ100 mVSHE: 1 EÀ5 Hz to <1 EÀ6 Hz
À500 mVSHE: 1 EÀ3 Hz to 1 EÀ1 Hz
T > $100  C: T" ! da/dtCF"

EA $ 40–50 kJ molÀ1
T# ! DKCF#
With minimum of DKCF at intermediate T?

Above 1–10 Hz, the environmental effects on crack growth disappear.
Source: Seifert, H. P.; Ritter, S. Research and service experience with environmentally-assisted cracking of carbon & low-alloy steels in
high-temperature water. SKI-Report 2005:60; SKI: Stockholm, Sweden, 2005; ISSN 1104-1374. alsakerhetsmyndigheten.
se/; Seifert, H. P.; Ritter, S. Corros. Sci. 2008, 50, 1884–1899.

The water flow rate has little effect on CF crack
growth of long/deep cracks, in contrast to CF initiation and short-crack growth. In the latter cases, CF
effects can disappear entirely if flushing out of the
local crack-tip electrolyte occurs, but this is unrealistic for deep, semi-elliptical cracks.
5.06.3.2.2.4 Adequacy and conservatism of
fatigue design according to Section III of ASME
BPV Code in the context of environmental effects

Design against fatigue of CS & LAS primary pressureboundary components is often based on Section III of
the ASME BPV Code.50 It relies on the use of fatigue
curves and endurance limits, derived mainly from
strain-controlled LCF tests with small, smooth specimens in air at room temperature, which do not explicitly consider the possible effects of LWR environments.
The accumulated good service experience with
CS & LAS primary pressure-boundary components does not indicate any generic deficiencies in
the current fatigue design procedures arising from
lack of consideration of environmental effects (see
Section 5.06.3.2.5), even though laboratory investigations clearly reveal that the fatigue lives of CS &
LAS can be substantially reduced in LWR environments (Figure 16)38,41,42,51,52: Although the microstructures and cyclic-hardening behavior of CS &
LAS differ significantly, the effects of the environment on the fatigue life of all these steels are very
similar. The magnitude of the reduction in this


depends on temperature, strain rate, oxygen level in
the water (i.e., ECP), and sulfur content of the steel.
The decrease is significant only when four conditions
are satisfied simultaneously, that is, when the strain
amplitude, temperature, and dissolved oxygen content in the water are above certain threshold values
($0.15%, 150  C, 0.04 ppm) and the strain rate is
below a key threshold value (10À3 sÀ1). Although
only a moderate decrease in life (by a factor of <2)
is observed if any one of the aforementioned threshold conditions is not satisfied, fatigue lives of CS &
LAS can be more than a factor of 20 lower in the
coolant environment than in air under certain environmental and loading conditions (Figure 16). Such
observations have thus raised some concern with
respect to the possibility of insufficient safety margins for the fatigue design of certain CS & LAS
pressure-boundary components.
Conservatism in the ASME Code fatigue evaluations may arise from (a) the fatigue evaluation procedures themselves and/or (b) the fatigue design
curves they use. Sources of conservatism in the procedures include the use of design transients that are
significantly more severe than those experienced in
service, conservative grouping of transients, and use
of simplified elastic–plastic analyses that result in
higher stresses/strains. The design margins of 2 and
20 on stress and cycles, respectively, in the ASME III
design curve were intended to cover the effects of
some variables (e.g., surface finish, material


Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

127

10.0


1.0

0.1
101
(a)

A533-Gr B low-alloy steel
288 ЊC, ea ≈ 0.4%

150−250 >250
0.05−0.2 >0.2
0.01−0.4 <0.01
³0.006 ³0.006

Temp( ЊC) :<150
DO (ppm) :£0.05
Rate (%/s) :³0.4
S (wt.%) :>0.006

Fatigue life (cycles)

Strain amplitude ea (%)

Carbon steel

Mean curve
RT air

ASME design curve

102

103

104

104

103

Air
Simulated PWR
≈0.7 ppm DO

102
105

10−5

106

Fatigue life (cycles)

(b)

10−4

10−3

10−2


10−1

100

Strain rate (%/s)

Figure 16 Comparison of fatigue initiation life of carbon steel in high-temperature water with ASME III mean and design
curves in air (a) and effect of strain rate and oxygen content on fatigue initiation life in low-alloy steel. Reproduced from US
NRC. Effect of LWR coolant environments on the fatigue life of reactor materials. NUREG/CR-6909; US NRC: Washington,
DC, 2007, Work performed by Argonne National Laboratory, managed and operated by UChicago
Argonne, LLC, for the US Department of Energy under Contract No. DE-AC02-06CH11357.

variability, load sequence, and size effects) that can
influence the fatigue life of components, but were not
actually investigated in the tests which provided the
original data for the curves. They were not intended to
cover the effects of LWR environments,38 but a recent,
detailed analysis41,42 revealed that the factors of 2 on
stress/strain and 20 on cycle number not only provide
appropriate margins for the intended factors but may
also contain excess conservatism that would partially
counteract reductions in fatigue life due to EAC.
Based on large research programs, methods have
been developed in the United States42 and in
Japan51,53 for incorporating the effects of LWR coolant environments into fatigue evaluations. Experimentally derived fatigue life correction factors Fen
(defined as the ratio of life in air at room temperature
to that in water at the service temperature) are used
to adjust component fatigue usage values for environmental effects. Such approaches have recently been
implemented in the Japanese JSME Code54 and the

new US NRC Regulatory Guide 1.207,55 which is to
apply to new plants.
The apparent discrepancy between isothermal
laboratory results (strong environmental effects) and
field experience (only a few CF incidents under very
specific circumstances, predominantly related to thermal transients) mainly arises from the large degree
of conservatism in the fatigue evaluation procedures
mentioned earlier. The significantly higher strains of
design transients may fully outweigh possible environmental effects for real plant transients. Furthermore,
one or more threshold conditions are often also not

satisfied for most transients. Even plant transients
with strong environmental effects (e.g., at slow strain
rates) are usually not very damaging with respect to
fatigue damage accumulation because of their small
strain amplitudes, together with the rather limited
number of cycles during the whole component lifetime. Finally, as discussed earlier, the turbulent flow
rate at most component surfaces significantly retards
corrosion fatigue crack initiation (in particular for
small strains and slow strain rates) compared to
the quasi-stagnant conditions used in most lab
investigations.38,41,42
5.06.3.2.2.5 Adequacy and conservatism of
fatigue flaw tolerance evaluations according to
Section XI of ASME BPV Code in the context of
environmental effects

Fatigue flaw tolerance evaluations in combination
with periodic in-service inspections are based on
Section XI of the ASME BPV Code. Article A-4300

in Appendix A of Section XI contains a set of reference fatigue crack growth rate da/dN curves for CS
& LAS in air (‘air curves’) or in LWR coolant environments (‘wet curves’).56 The current ASME XI wet
curves (Figure 17) are based on lab data obtained
prior to 1980. They depend explicitly on DK and load
ratio R, but not on other variables that are known to
be important in CF, such as loading frequency, ECP,
or steel sulfur content. The same curves are used for
different types of CS & LAS and for BWR/NWC,
BWR/HWC/NMCA, and PWR primary or secondary side conditions. System conditions (or thresholds)


Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

100
10−1
10−2

101

Δtrise = 14 s
R = 0.95
0.018% S

100
10−1
10−2

43 s
−6 0
N 20

e
=
s
ca t ise
e Δ r
od .65
C 0
=
R

HWC, 274/288 ЊC

3
64 s
N- 14
e =
s
e
ca Δt ris
de 95
Co = 0.
R

e

rv

10−3

ME

AS X I ‘W
et,
ME
’R
XI
Ն0
‘W
.65
et,
’R
Յ
0.2
5

101

N-643, EAC-curve for S > 0.013 % S, R = 0.65, ΔtR = 100 s
N-643, EAC-curve for S > 0.013 % S, R = 0.65, ΔtR > 1 s
N-643, non-EAC curve, R = 0.65
ASME XI ‘Wet,’ R = 0.65
ASME XI ‘Air,’ R = 0.65

AS

da/dN (μm per cycle)

102

da/dNCF (μm per cycle)


128

ir’

E

XI

‘A

M

AS

Δtrise = 200–2000 s
R = 0.3–0.7
0.004–0.018% S

cu

AC

-E

on

N
10−3
10−4
1

10
100
1
10
100
(a)
Stress intensity factor amplitude ΔK (MPa m1/2)
(b)
Stress intensity factor amplitude ΔK (MPa m1/2)

Figure 17 ASME XI ‘Air’ and ‘Wet’ curves and Code Case N-643 for high and low-sulfur steels (a) and comparison of cyclic
corrosion fatigue crack growth rates under hydrogen water chemistry conditions for different loading conditions with the
corresponding curves (b). Reproduced from Seifert, H. P.; Ritter, S. Corros. Sci. 2008, 50, 1884–1899.

where environmental effects on fatigue crack growth
can be neglected or excluded are not defined in the
present ASME Section XI Code.38,47
A more specific Code Case N-64357 for fatigue
crack growth in CS & LAS exposed to PWR primary
environments has been developed since 2000 and
may be used as an alternative to the ASME XI wet
reference fatigue crack growth rate curves for this
specific environment (Figure 17). Depending on system conditions, the Code Case N-643 procedure
predicts either lower or higher crack growth rates
than the general Section XI approach. The main
advantage of this newer Code Case is that it contains
criteria for the onset/cessation of EAC and that it
better reflects the experimentally observed cracking
behavior in PWR environments, since it considers
steel sulfur and frequency–loading-rate effects to a

certain extent. However, this approach has not (yet)
found general acceptance, primarily because of the
difficulty of defining appropriate rise times for actual
plant transients and the complications involved in
including these in component analyses.
In a similar way to that discussed earlier for
fatigue life design, conservatism in fatigue flaw tolerance evaluations may arise from (a) the fatigue flaw
tolerance evaluation procedures themselves (e.g., by
the use of design transients) and/or (b) the fatigue
crack growth rate curves. As discussed in Seifert and
Ritter,38,47 the current ASME XI wet curves conservatively cover the CF crack growth rate laboratory data
under most combinations of loading, environmental,
and material parameters. Even under highly oxidizing
BWR/NWC conditions, they are only significantly
exceeded under some very specific (but plantrelevant) circumstances (Figures 14 and 17), which

have caused some isolated CF cracking incidents in
the past. The Section XI curves might therefore be
regarded as an adequate, general bounding approach
under most system conditions, but they do not
realistically describe and reflect the experimentally
observed CF crack growth behavior of CS & LAS in
oxygenated high-temperature water. The curves predict crack growth rates which are either significantly
too high (e.g., n 10À2 Hz and ECP < À200 mVSHE)
or too low (e.g., n 10À2 Hz and ECP > 0 mVSHE).
Furthermore, system conditions or thresholds (e.g.,
n > 1 Hz), where environmental effects on fatigue
crack growth can be neglected, or even excluded, are
not defined in ASME XI.
5.06.3.2.3 Stress corrosion cracking

5.06.3.2.3.1
conditions

Initiation and susceptibility

Table 5 shows an assessment scheme according to
Hickling,58 based on both laboratory and field experience, for SCC initiation susceptibility and crack
growth in CS & LAS at normal strength levels
under BWR/NWC conditions.
Initiation of propagating SCC cracks from
smooth, defect-free surfaces under static load in
high-purity water is only observed for the following
conjoint conditions: stresses at the water-wetted surface above the high-temperature yield strength,
quasi-stagnant flow conditions, and dissolved oxygen
contents !$0.2 ppm. Furthermore, if complete exhaustion of low-temperature creep is allowed to occur
before the specimens are exposed to high-purity,
high-temperature water, no SCC is observed, thus
indicating ‘nonclassical’ SCC behavior and confirming the importance of slow dynamic surface straining


Assessment scheme for SCC susceptibility of CS & LAS

O2 (ppm)

Operating medium: HT water or steam condensate with T > 170  C
Flow conditions

k (mS cmÀ1)

Crack initiation by SCC?


Derivation

Crack growth by SCC?

Derivation

<0.2
<0.2
<0.2

Typical for reactor
Quasi-stagnant
Quasi-stagnant

Typical, that is, 0.1
$0.2
Raised (e.g., by impurities)

1
1
2

Typical for reactor
Quasi-stagnant

Typical, that is, $0.2
$0.2

0.2–0.4


Quasi-stagnant

Raised (e.g., by impurities)

)0.4

Typical for reactor

Typical, that is, $0.2

)0.4

Quasi-stagnant

<1 often (0.2)

)0.4

Quasi-stagnant

Raised (e.g., by impurities)

No susceptibility
No susceptibility
Possibility cannot be excluded,
perhaps after incubation time
No susceptibility
Possibility cannot be excluded,
perhaps after incubation time

Possibility cannot be excluded,
perhaps after incubation time
Susceptibility is suppressed
through flow
Possibility cannot be excluded,
perhaps after incubation time
Possibility cannot be excluded,
perhaps after incubation time

1
2
3

0.2–0.4
0.2–0.4

No susceptibility
No susceptibility
For stress levels at the water-wetted surface in
the region of the HT yield pointa
No susceptibility
For stress levels at the water-wetted
surface > HT yield pointa
For stress levels at the water-wetted surface in
the region of the HT yield pointa
In general, no susceptibility (? at stress
levels)HT yield point)
For stress levels at the water-wetted surface !
HT yield pointa
For stress levels at the water-wetted in the

region of the HT yield pointa

1
2
2
1, 3
2
2

a
Possibility cannot be excluded, perhaps after long incubation time.
1 – from experiments in more aggressive environments; 2 – from appropriate autoclave experiments; 3 – no direct experimental evidence.
Source: Hickling, J.; Reitzner, U. VGB Kraftwerkstech. 1992, 72, 359–367.

1
2
2
2
2
2

Corrosion and Environmentally-Assisted Cracking of Carbon and Low-Alloy Steels

Table 5

129