Corrosion Science 78 (2014) 233–243

Contents lists available at ScienceDirect

Corrosion Science
journal homepage: www.elsevier.com/locate/corsci

An arbitrary Lagrangian–Eulerian model for modelling the

time-dependent evolution of crevice corrosion
Wen Sun, Lida Wang ⇑, Tingting Wu, Guichang Liu ⇑
Department of Materials Science and Chemical Engineering, School of Chemical Engineering, Dalian University of Technology, 2 Linggong Road, Dalian 116024, China

a r t i c l e i n f o a b s t r a c t

Article history: An arbitrary Lagrangian–Eulerian (ALE) model is proposed to model the time-dependent evolution of a
Received 20 April 2013 crevice corrosion that can achieve the simulation of the transient potential distribution, the dynamic con-
Accepted 2 October 2013 centration profiles and the time-dependent geometry deformation for a corroding crevice. The time-
Available online 14 October 2013
dependent deformation of the crevice geometry, which is caused by metal dissolution and corrosion
product precipitation, is implemented via the ALE method. The model reasonably predicted the dynamic
Keywords: pH distribution inside the crevice. It is also predicted that the deposition of corrosion product accelerates
A. Stainless steel
crevice corrosion. The current model would be helpful in elucidating the mechanism of crevice corrosion
B. Modelling studies
C. Crevice corrosion
and other forms of related localised corrosion.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction
Crevice corrosion is an especially dangerous type of localised
corrosion for its unfeasible detection, which can cause sudden
and often catastrophic failure [1–5]. It may occur on the surface
of all kinds of metallic alloys as a result of small occluded cells un-
der crevices [2,6]. During the crevice corrosion process, as the cre-
vice mouth narrows gradually, the mass transfer between crevice
solution and bulk solution becomes more difficult, resulting in
the formation of more aggressive crevice solution [7–9]. A good
understanding of the crevice corrosion mechanism cannot only im-
prove our ability to predict and prevent crevice corrosion, but also
provide further insights into other forms of localised corrosions
[10], such as pitting corrosion [11,12] and stress corrosion cracking
[13,14].
Crevice corrosion is an extremely complex phenomenon [15–
18]. As far as crevice corrosion of iron in chloride solution is con-
cern, the iron inside the crevice dissolves releasing Fe2+ ions. Then
Fe2+ ions react with water to generate H+. With the development of
crevice corrosion, cations concentrate inside the crevice. In order to
maintain charge balance of the crevice solution, the anions outside
the crevice (such as Cl, OH) immigrate into the crevice and con-
centrate inside the crevice. Subsequently, the crevice solution be-
comes saturated with ferrous chloride [19]. Consequently, after a
series of complex physical and chemical processes, corrosion prod-
uct precipitates at the crevice mouth. The deposition of corrosion
product changes the geometry and the crevice corrosion behav-
iour. It is known that an active local corrosion cell can be sealed
⇑ Corresponding authors. Tel./fax: +86 411 84986047.
E-mail addresses: ldwang@dlut.edu.cn (L. Wang), gchliu@dlut.edu.cn (G. Liu).
off by a special restrictive geometry via limiting the mass transfer
between crevice solution and bulk electrolyte. Thus, the common
features of restrictive geometry and local acidity lead to the view-
point that crevice corrosion is a general phenomenon of ‘‘occluded
cell corrosion’’ [20].
Up to now, crevice corrosion has been studied experimentally
using methods such as freezing methods using a micro-electrode
[21], simulation of corrosion in the occluded space [22], cyclic
potentiodynamic polarisation [23], the Potentiodynamic–Galvano-
static–Potentiodynamic technique [24] and dynamic electrochem-
ical impedance spectroscopy [25]. However, the prediction of the
crevice damage via those methods still remains difficulties. With
the development of computational mathematics and computer
technology, mathematical modelling, which is important for pre-
dicting corrosion rate, has also been extensively used to analyse
the crevice corrosion. Sharland and Tasker [26] established a model
to predict the steady-state solution chemistry and electrode kinet-
ics within an active corrosion cavity. The model includes a solid
hydroxide precipitation reaction. They found the corrosion rate de-
pends on many parameters, such as external electrode potential
and crevice dimensions. Sharland et al. [27] developed a mechanis-
tic model of the propagation stage of an established crevice. The
model considers the time evolution of the solution chemistry and
electrochemistry within a corroding cavity. Vuillemin et al. [28]
published a work to simulate steady state crevice corrosion prop-
agation by considering the formation of solids and gases. Kennell
et al. [4,10] developed a critical crevice solution and IR drop crevice
corrosion model. They reported that cathodic reactions are likely to
occur towards the tip of the crevice. In order to quantitatively as-
sess the risk of the corrosion of pipelines under the disbonded
coatings, Song [29–33] developed several crevice corrosion models

0010-938X/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.corsci.2013.10.003
234 W. Sun et al. / Corrosion Science 78 (2014) 233–243
Nomenclature
a anode
c cathode
bi Tafel slope of species i involved electrochemical reac-
tion (V/decade)
ci concentration of species i (mol/m3)
ci,0 concentration of species i in bulk electrolyte (mol/m3)
D distance from crevice mouth (cm)
Di diffusion coefficient of species i (m2/s)
E electric field strength (V/m)
Ei electrode potential (V vs. SCE)
E0i equilibrium electrode potential (V vs. SCE)
F Faraday’s constant (96,485 C/mol)
ij current density of electrochemical reaction j (A/m2)
iM anode current density of metal M (A/m2)
i0i exchange current density (A/m2)
ip current density at artificial boundary (A/m2)
J current density vector (A/m2)
Ji mass diffusion flux (mol/(m2 s))
k rate coefficient of corrosion product deposition (m10/
(mol3 s))
0
k reaction constant of corrosion product deposition (mol/
m3)
~
ki forward reaction constant for hydrolysis reaction of spe-
cies i (the unit is related to hydrolysis reaction)
ki backward reaction constant for hydrolysis reaction of
species i (the unit is related to hydrolysis reaction)
Keq,i equilibrium constant for the hydrolysis reaction of spe-
cies i
MCP molecule weight of corrosion product (kg/mol)
Mi molar mass of species i (kg/mol)
n normal vector
nj number of electron transferred of reaction j
of metals with different influenced factors, such as the oxygen dif-
fusion, the solution resistance and the variation of the disbond-
ment gap.
However, since the appearance of corrosion product inside or
outside a crevice will automatically have an impact on the crevice
geometry, steady state simulations together with transient simula-
tions without considering time-dependent geometry deformation
of crevice cannot precisely describe the whole crevice corrosion
process. Vuillemin et al. [28] suggested that one of the challenges
in the simulation of crevice corrosion propagation is the rigorous
treatment of non-aqueous species (gas or solid), since the appear-
ance of a new phase in a crevice will automatically have an impact
on its geometry. To overcome this challenge, moving mesh method
would be a promising approach. At present, there are several inter-
esting works about modelling corrosion processes on the basis of
moving mesh method [34,35]. Xiao and Chaudhuri [36] presented
a FEM-based corrosion model. Their work aim to provide a ‘‘deter-
ministic’’ modelling approach that can systematically identify, cou-
ple and characterise multiple physical and (electro)chemical
phenomena under a modular framework. Deshpande [37] pre-
sented a moving mesh model to predict the corrosion rate of gal-
vanic couples. The numerical corrosion rates were in good
agreement with those obtained from the experimental techniques.
Subsequently, this model was used to capture the evolution of
cross-sectional microstructure to investigate its effect on corrosion
behaviour of magnesium alloys [38]. Onishi et al. [39] presented a
novel numerical method to analyse time-dependent localised cor-
rosion by combining the finite volume method and the voxel meth-
od. The method successfully reproduced the experimental results
N grid number near crevice mouth
Ni molar flux of species i (mol m2/s)
R universal gas constant (8.3145 J/(Kmol))
Ri reaction rate of species i (mol/(m3 s))
Rei Electrochemical reaction rate of species i (mol/(m2 s))
s area of corrosion product deposition (m2)
T temperature (K)
ui mobility of species i (mol m2/(J s))
v moving velocity vector (m/s)
V moving velocity caused by corrosion product deposition
rate (m/s)
x x co-ordinate in spatial frame (m)
X X co-ordinate in reference frame (m)
y y co-ordinate in spatial frame (m)
Y Y co-ordinate in reference frame (m)
zi charge number of species i
Greek
e relative permittivity
e coverage rate of corrosion product deposition at the cre-
vice mouth
e0 initial coverage rate of corrosion product deposition at
the crevice mouth
h porosity of corrosion product deposition
h0 initial porosity of corrosion product deposition
/ potential (V vs. SCE)
mij stoichiometric coefficient of species i in reaction j
q charge density (C/m3)
qCP density of corrosion product (kg/m3)
qi density of species i (kg/m3)
r electric conductivity (S/m)
C boundary
of a capillary electrophoresis problem and several examples of cre-
vice corrosion. Sarkar et al. [40] presented a numerical framework
to simulate corrosive dissolution over long periods of time by using
an adaptive meshing strategy, with which the evolution of the io-
nic fields, current density, and movement of the anodic front can
be calculated. They demonstrated that such a mathematical/
numerical framework can capture the complex dynamics of elec-
trochemical behaviour in the corroding systems. However, most
of those models are developed to simulate the movement of anodic
front. As far as we know, there have been no transition models that
can be used to study the effect of the deposition from the corrosion
product on the crevice corrosion behaviour.
During crevice corrosion, the precipitation of corrosion product
and the dissolution of crevice walls naturally occur, and that lead-
ing to the timely deformation of crevice geometry. Since the cre-
vice geometry is one of the main factors to influence the
development of crevice corrosion, analysing the deformation of
crevice geometry would help to understand crevice corrosion
[28,41]. On the basis of our published work [42,43], a mathemati-
cal model is established in this paper to illustrate the time-depen-
dent evolution of geometry corrosion by considering the crevice
deformation caused by metal dissolution and corrosion product
deposition.
2. Simulation
The numerical results predicted by the current model are vali-
dated against the experimental results of Alavi and Cottis [44]. In
their experiment, a plate of AISI 304 stainless steel, which was
 W. Sun et al. / Corrosion Science 78 (2014) 233–243 235

coupled to a larger piece of the same steel, and a Perspex electrode

holder were used to fabricate the experimental crevice measuring
8 cm deep and 2.5 cm wide with a 90 lm gap. The crevice was ex-
posed to free-corrosion conditions in aerated NaCl solution (0.6 Artificial Boundary
moI/L, pH = 6.5) at room temperature. Electrodes were placed
along the crevice wall to measure the dynamic concentration pro- y
Bulk Solution
files of Cl and H+ ions.
The presented model is developed based on electrochemical (0, 0) x
reactions, hydrolysis reactions, mass transfer and overall charge CMP
balance. The distribution of potential is based on transient electric- Cathode Wall Crevice Mouth
ity. The electrochemical reactions that involved in the crevice cor-
rosion process of AISI 304 stainless steel are described by Butler–

Crevice Wall
Volmer equation and Tafel equation. Multistage hydrolysis reac-
tions of the metallic ions and their complex reactions with anions,
Crevice Solution
as well as the hydrolysis equilibrium of water, are taken into con-
sideration in this paper. The crevice deformation velocity caused
by the metal dissolution is presented by Faraday’s Law, while that
caused by corrosion product deposition is obtained on the basis of
Crevice Tip
the kinetics of corrosion product deposition. However, up to now,
reports on the kinetics of corrosion product deposition are still rare
and no experimental data for the kinetic rate of corrosion product CTP(0, -8)
deposition is available. Experimental techniques applied to charac-
terise corrosion products in chloride-containing environments
show that the main component of the corrosion products is Fe(III) Fig. 1. A schematic diagram of the crevice geometry and the concepts used in this
work. Coordinate system used in the text is also shown in this picture. CMP, whose
oxyhydroxides (b-FeOOH) or Fe(II) hydroxychloride (b-Fe2(OH)3Cl) coordinate is (0.0045, 0), denotes crevice mouth position, and it is a reference
[45–50]. Herein, to simplify the simulation, the deposition kinetics position. CTP is short for crevice tip position (0.0045, 8).
of corrosion products is supposed to be simply related to concen-
trations of Fe2+, Cl and O2. On the basis of this hypothesis, the 
@/ X
kinetics of corrosion product deposition is described mathemati- r ¼ ic;j ð5Þ
@nC j
cally to qualitatively present the deformation of crevice geometry,
which contributes to a better understanding of the time-dependent
evolution of crevice corrosion. In this work, the computations While for crevice walls, the boundary condition can be ex-
involving the partial differential equations of current model are pressed as follows:
performed by using the COMSOL MultiphysicsÒ software program, 
@/
which allows flexible meshing of model geometry and setting of r ¼ itotal ð6Þ
@n C
boundary conditions [51].

2.1. Mathematics description for electricity 2.2. Mathematics description for mass transfer

For the physical problem of transient electricity, several funda- 2.2.1. Governing equation for mass transfer
mental equations should be taken into account [42,51]. These fun- The crevice reactive system includes the following twenty spe-
2
damental equations involved in the current model are shown as cies: Na+, Cl, OH, H+, Fe2+, FeCl+, FeCl2, FeCl4 , Fe(OH)+, Fe(OH)2,
þ þ
follows: Cr , CrCl , CrCl2 , Cr(OH) , CrðOHÞ2 , Cr(OH)3, Ni2+, NiCl+, Ni(OH)+,
3+ 2+ 2+

Ohm’s law Ni(OH)2. The governing equations are deduced firstly from the mass
J ¼ rE ð1Þ balance principle, neglecting the advection term and considering
mass transfer by diffusion and electromigration in the case of dilute
aqueous solutions. Then the Nernst–Planck equation is given as
The equation of continuity @ci
þ Ni ¼ Ri ð7Þ
@q @t
þ rJ ¼ 0 ð2Þ
@t
The total flux of species i, Ni, is defined accordingly as
And Gauss’ law Ni ¼ Di r  ci  zi ui Fci r  / ð8Þ
r  ðeEÞ ¼ q ð3Þ
The potential in Eq. (8) is provided by coupling the electricity
part with current model. Furthermore, the mobility for species i,
The boundary condition for artificial boundary (as shown in
ui, is given by the Nernst–Einstein equation.
Fig. 1) is given by
 Di
@/ ui ¼ ð9Þ
r ¼ ip ð4Þ RT
@n C

Herein, ip is the current density which makes the crevice corrosion
system at the boundary satisfy Kirchhoff’s first law. Following
Ohm’s law, the potential flux at the stainless steel surface equals
the total current density multiplied by the solution resistivity. For
cathode walls, the boundary conditions satisfy
Herein, the diffusion coefficient for species i, Di, is taken from
Ref. [1,6]. In addition to considering mass transfer due to diffusion
and electro-migration, the electroneutrality condition is taken into
account to ensure that there is always a zero net charge at any local
point in a dilute solution
236 W. Sun et al. / Corrosion Science 78 (2014) 233–243

@ci zi ui F 2 ci X Table 2
¼ zj cj ð10Þ Hydrolysis reactions used to model the crevice corrosion of AISI 304 stainless steel
@t e j and their equilibrium constants.

Reaction log10(Keq) References No.

2.2.2. Electrochemistry Cr3þ þ H2 O $ CrðOHÞ2þ þ Hþ 3.8 [6,11] a
The electrode potential evolved in electrochemical reaction j is CrðOHÞ2þ þ H2 O $ CrðOHÞþ
2 þH
þ 6.2 [6] b
calculated by Nernst equation [3] CrðOHÞþ
2 þ H2 O $ CrðOHÞ3 þ H
þ 6.2 [6] c
!  2þ 0.149 [5,6] d
RT Yn Cr3þ þ Cl $ CrCl
v
Ei ¼ E0i þ ln cj ij ð11Þ Cr3þ þ 2Cl $ CrCl2
 þ 0.158 [5,6] e
nij F j¼1 Fe2þ þ H2 O $ FeðOHÞþ þ Hþ 8.3 [6,52] f
FeðOHÞþ þ H2 O $ FeðOHÞ2 þ Hþ 11.1 [6,52] g
 þ [5,6] h
Fe2þ þ Cl $ FeCl 0.161
Then the electrochemical reactions, which are assumed to occur 
Fe2þ þ 2Cl $ FeCl2 2.45 [5,6] i
at the metal/solution interface, are listed in Table 1. All anodic  2 1.90 [5,6] j
Fe2þ þ 4Cl $ FeCl4
reactions occur simultaneously in the crevice. It is assumed that [6] k
Ni2þ þ H2 O $ NiðOHÞþ þ Hþ 9.5
metallic elements in AISI 304 stainless steel contribute to the over- þ
NiðOHÞ þ H2 O $ NiðOHÞ2 þ Hþ 9.1 [6] l
all corrosion process in proportion to their mole fraction. And AISI 2þ  þ 0.996 [5,6] m
Ni þ Cl $ NiCl
304 stainless steel is composed of 71% iron, 19% chromium, 10% H2 O $ OH þ Hþ 14 [53] n
nickel, while other impurities are not considered. Thus, the overall
anodic reaction can be expressed as
ia ¼ 0:71iFe þ 0:19iCr þ 0:1iNi ð12Þ Then, the consumption rate of Fe2+ involved in the first step of
2+
Fe hydrolysis is given as
The physical parameters fixed in this part are taken from Ref. [1].
R0Fe2þ ¼ ~
kFe;1 cFe2þ þ k Fe;1 cFeOHþ cHþ ð15Þ
2.2.3. Homogeneous hydrolysis
The hydrolysis reactions involved in this work and their corre-
In the same way, the source terms for different species are listed
sponding equilibrium constants at room temperature are listed in
in Table 3.
Table 2. The equilibrium constants greatly limit the concentrations
of free ions (such as H+, Fe2+, Cl) inside the crevice. And the stabili- 2.2.4. Boundary conditions
sation of hydrolysis products can be deduced according to their cor- At the boundary (the cathode wall and the crevice wall as
responding equilibrium constants. For example, it can be concluded shown in Fig. 1), mass diffusion flux follows Fick’s diffusion law
from Table 2 that FeCl+ is more stable than FeOH+ because the equi- 
@ci 
librium constant of reaction h is greater than that of reaction f. Con- Ji ¼ Di ð16Þ
@nC
sequently, the pH profile inside the crevice would be changed in the
presence of Cl because Fe2+ is more likely to react with Cl instead of
hydrolysing to generate H+ [5,28]. Since hydrolysis reactions inside For species i which takes part in electrochemical reaction, its
the crevice are very complex, only fourteen reactions and nineteen production or consumption rate depends on electrochemical reac-
aqueous species are taken into consideration in the present case. tion rate and can be calculated by Faraday’s law
For species i, the reaction rate, which is the source term in the mij ij
Nernst–Plank equation, is expressed by related hydrolysis kinetics Rei ¼ ð17Þ
nj F
constants. For instance, the first step of Fe2+ hydrolysis is

ð13Þ Since the diffusion flux of species i equals its production or con-
sumption rate, the net flux of species i is given as

! @ci  X g v ij itotal
i
where k Fe;1 and k Fe;1 are the forward and reverse kinetic constants, Di  ¼ ð18Þ
@n C j
nij F
respectively. And they are related to the equilibrium constant
~
kFe;1
K eq;Fe1 ¼ ð14Þ Parameter gi is the mole fraction of metallic component for an-
k Fe;1 ode reaction, while it equals 1 for cathode reaction. If species i does

Table 1
Electrochemical reactions occurring within the corroding crevice system.

Description of reaction Reaction Current density (A/m2) References


Reduction of water 2H2 O þ 2e ! 2OH þ H2 ð/EH O Þ
2 [32]
0 bH O
iH2 O ¼ iH2 O 10 2

þ ð/EH Þ
[1,3,28]
2H þ 2e ! H2 2

iH2 ¼ i0H2 cHþ 10 bH

2


Reduction of oxygen O2 þ H2 O þ 4e ! 4OH c O2
ð/EO Þ
2 [1]
iO2 ¼ i0O2 cO2;0 10 bO
2

O2 þ 4Hþ þ 4e ! 2H2 O 0 c O2
ð/EO Þ
bO
2 [4]
iO2 ¼ iO2 cO2;0 10 2

!
Oxidation of Fe Fe ! Fe2þ þ 2e /EFe ð/EFe Þ [32]
iFe ¼ i0Fe 10 bFe;a  ccFe;0
Fe
10 bFe;c

Oxidation of Cr Cr ! Cr3þ þ 3e [1]

/ECr

iCr ¼ i0Cr 10 bCr

Oxidation of Ni Ni ! Ni 2þ
þ 2e
/ENi
[1]
iNi ¼ i0Ni 10 bNi
 W. Sun et al. / Corrosion Science 78 (2014) 233–243 237

Table 3
The source terms for different species in Nernst–Plank equation. The above equations dictate smooth deformation of the mesh
Species Source terms by considering the imposed boundary conditions. For the inner
Fe(OH)+ ! walls of the crevice, if the deformation velocity of crevice walls is
RFeðOHÞþ ¼ k Fe;1 cFe2þ  k Fe;1 cFeðOHÞþ cHþ  RFeðOHÞ2
!
denoted as v, its normal component can be calculated by the fol-
Fe(OH)2 RFeðOHÞ2 ¼ k Fe;2 cFeðOHÞþ  k Fe;2 cFeðOHÞ2 cHþ lowing equation
FeCl+ !
RFeClþ ¼ k Fe;3 cFe2þ cCl  k Fe;1 cFeClþ
FeCl2 ! XX g iij M i
RFeCl2 ¼ k Fe;4 cFe2þ c2Cl  k Fe;4 cFeCl2 n  v jC ¼ i
V ð23Þ
2
FeCl4
!
RFeCl2 ¼ k Fe;5 cFe2þ c4Cl  k Fe;5 cFeCl2 j i
nij F qi
4 4

Cr(OH)3 !
RCrðOHÞ3 ¼ k Cr;3 cCrðOHÞþ  k Cr;3 cCrðOHÞ3 cHþ
2

CrðOHÞþ !
2 RCrðOHÞþ ¼ k Cr;2 cCrðOHÞ2þ  k Cr;2 cCrðOHÞþ cHþ  RCrðOHÞ3
2 2
In the above-mentioned equation, the first term of the right-
Cr(OH)2+ ! hand side of Eq. (23) represents the mobile rate of crevice walls
RCrðOHÞ2þ ¼ k Cr;1 cCr3þ  k Cr;1 cCrðOHÞ2þ cHþ  RCrðOHÞþ
2

CrCl2+ ! caused by the metal dissolution, while the second term is walls’
RCrClþ ¼ k Cr;5 cCr3þ c2Cl  k Cr;5 cCrClþ
2
!
2
movement rate caused by the deposition of corrosion product.
CrCl2+ RCrCl2þ ¼ k Cr;4 cCr3þ cCl  k Cr;4 cCrCl2þ Unfortunately, up to now, no kinetic data for the deposition of cor-
Ni(OH)+ !
RNiðOHÞþ ¼ k Ni;1 cNi2þ  k Ni;1 cNiðOHÞþ cHþ  RNiðOHÞ2 rosion product are currently available. In this work, in order to
Ni(OH)2 !
RNiðOHÞ2 ¼ k Ni;2 cNiðOHÞþ  k Ni;2 cNiðOHÞ2 cHþ qualitatively model the deposition of corrosion product, the
NiCl+ ! parameter k (kinetic constant of corrosion product deposition) is
RNiClþ ¼ k Ni;3 cNi2þ cCl  k Ni;1 cNiClþ
OH ! assumed to be 108 m10 s1 mol3. Furthermore, as the generation
ROH ¼ k w  k w cHþ cOH
Fe2+ RFe2þ ¼ RFeðOHÞþ  RFeClþ  RFeCl2  RFeCl2 mechanism of corrosion product is very complex, in order to sim-
Cr3+ RCr3þ ¼ RCrðOHÞ2þ  RCrClþ  RCrCl2þ
4
plify the simulation, we assume the mobile rate is
2

Ni2+ RNi2þ ¼ RNiðOHÞþ  RNiClþ

Cl RCl ¼ RFeClþ  2RFeCl2  4RFeCl2  RCrCl2þ  2RCrClþ  RNiClþ
V ¼ kcFe2þ cOH cCl ð24Þ
4 2
+
H RHþ ¼ RFeðOHÞþ þ RFeðOHÞ2 þ RCrðOHÞ2þ þ RCrðOHÞþ þ RCrðOHÞ3
2

þRNiðOHÞþ þ RNiðOHÞ2 þ ROH The cathode or the outer walls of crevice are assumed to be
non-corroding. Then the normal moving velocity of the outer walls
not participate in electrochemical reactions, its boundary condition is considered to be
is presented
 by
@ci  n  v jC ¼ kcFe2þ cOH cCl ð25Þ
Di ¼0 ð19Þ
@nC
And the artificial boundaries are considered to have zero dis-
However, at the artificial boundary, all species satisfy the fol- placement. Their boundary conditions is
lowing boundary condition
dxjC ¼ 0; dyjC ¼ 0 ð26Þ
ci jC ¼ ci;0 ð20Þ

Finally, it should be noted that the corrosion product deposition

2.3. Moving mesh technique
covering the crevice greatly affects the electricity and mass trans-
fer near the crevice mouth. Hence, corresponding equations must
In order to track the movement of crevice walls, the computa-
be modified because the corrosion product deposition is supposed
tional geometry has to be realigned timely with new position of
to be porous [57]. Furthermore, a very thin deposition is assumed
the walls via a proper moving mesh technique. Among the existing
to cover the crevice opening at time t = 0. The initial thickness is
moving mesh techniques, the ALE method is a finite element for-
proposed to be 109 cm so that it would not affect the behaviour
mulation in which the computational domain is not preferentially
of crevice corrosion. The potential distribution and mass transfer
fixed in space or attached to material. The ALE-based finite element
in the porous corrosion product deposition is different from that
simulations possess both advantages of the traditional Lagrangian-
in solution. Handling problems of electricity and mass transfer in
based and Eulerian-based finite element simulations. Furthermore,
porous media are detailed reported in our published paper [43].
the ALE method allows moving boundaries without the need of
During crevice corrosion, the coverage rate and the porosity of
mesh‘s movements and can capture greater deformation with
corrosion product deposition can be calculated by the following
higher resolution [54,55].
equation
For ALE method with mesh refinement, if the mesh displacement
becomes large and the mesh quality degrades during the calcula-  Z Z 0
tion, the computational domain would be re-meshed and all solu- 1 @e 1 MCP k cFe2þ cCl cO2
 ¼ dl ð27Þ
tion variables would be mapped from the old mesh to the new 1e @t ð1  h0 Þs qCP
one. In the present case, Laplacian smoothing is used as mesh-up-
date strategy for ALE method. As in mesh generation or smoothing 0
techniques, the rezoning of the mesh nodes consists in solving a La- 1 @h M CP k cFe2þ cCl cO2
 ¼ ð28Þ
place or Poisson equation for each component of the node velocity h @t qCP
or position so that the mesh forms lines of equal potential on a log-
ically regular region [54,56]. The mesh displacement is governed by qCP
0
@ 2 @x @ 2 @y k ¼ k ð29Þ
þ ¼0 ð21Þ M CP
@X 2 @t @Y 2 @t
where CP is short for corrosion product, the initial porosity h0 is as-
@ 2 @y @ 2 @x sumed to be 1, the initial coverage rate e0 is ‘‘crevice opening
2 @t
þ 2 ¼0 ð22Þ
@X @Y @t length/outer crevice length = 0.0045’’.
238 W. Sun et al. / Corrosion Science 78 (2014) 233–243

3. Results and discussion reasonable agreement with the experiment. Thus, it can be con-
cluded that the current model can also reasonably predict the
3.1. Model validation time-dependent pH evolution of the crevice solution.

The pH profile along the crevice is obtained by considering the
interaction of cation hydrolysis, mass transfer and hydrogen evolu-
tion. Vuillemin et al. [28] pointed out that the hydrogen evolution
occurs only for cathodic polarisation when crevice width is greater
than 5 lm, while for smaller gaps, hydrogen evolution can also oc-
cur under anodic polarisation. Kennell and Evitts [10] argued that
the hydrogen evolution would naturally occur at the crevice tip
after the dissolution of metal and the establishment of consider-
able gradients of concentration and potential within the crevice.
Furthermore, Alavi and Cottis’ [44] also suggested that the cathodic
3.2. The influences of crevice deformation on crevice corrosion
When the mesh deformation has become so large, the quality of
the mesh is too bad to continue calculation. It is necessary to gen-
erate a new mesh for the deformed configuration. On the other
hand, because the calculation accuracy is closely related to the
4.5
a Experiment [44]

4.0 Heppner's model [6]

hydrogen evolution or water reduction may occur in the depth of Current model
the crevice. Therefore, it is necessary to consider the hydrogen evo-
lution at the crevice tip in the analysis of the mass transfer and the 3.5
electrochemical phenomena. Fig. 2 shows the experimental data of
pH profile along the crevice and the corresponding numerical re- 3.0
sults predicted by several published models. The current model

pH
predicts that pH decreases firstly and then increases with the de- 2.5
crease of the distance from the crevice mouth position (CMP,
whose coordinate is (0.0045, 0) as shown in Fig. 1). It also can 2.0
be seen that the pH displays a minimum value near the crevice
mouth (0.0045, 1.0). It is obvious that the numerical result pre- 1.5
dicted by current model is in much better agreement with the pub-
lished experimental data than that predicted by other models. The 1.0
0 20 40 60 80 100
good agreement between the numerical results and the experi-
Time t / h
mental data indicates that the current model is of high accuracy
and is able to reasonably predict the pH profile along the crevice. 4.5
The presented model is further validated by the comparison be- b Experiment [44]
Heppner's model [6]
tween the numerical and the experimental pH-time curves. The 4.0 Current model
experimental result shows that the pH inside the crevice decreases
rapidly at first as shown in Fig. 3. At the position near crevice
3.5
mouth (0.0045, 1.0), the pH decreases to a minimum value of
about 1.66 when the experiment runs for about 7 h, and then it in-
creases gradually. At the position in the middle of the crevice 3.0
pH

(0.0045, 4.0), the pH keeps decreasing until the experiment

runs for 52 h and then increases gradually for the rest time of 2.5
the experiment. At the position near the crevice tip (0.0045,
7.5), pH deceases slowly until the end of the experiment. The pre-
2.0
dicted pH profiles at position (0.0045, 1.0), (0.0045, 4.0) and
(0.0045, 7.5) are also shown in Fig. 3, respectively. It can be seen
that the numerical results predicted by current model are in 1.5
0 20 40 60 80 100

Time t / h
5.5
Evitts et al. [15]
Walton et al. [5]
4.5 c Experiment [44]
5.0 Heppner's model [6]
Sharland et al. [26] Current model
White et al. [58] 4.0
4.5
Heppner et al. [6]
Kennell et al. [10]
4.0 The current model 3.5
Experimental data [44]
pH

3.5
pH

3.0
3.0
2.5
2.5

2.0
2.0

1.5 1.5
-8 -7 -6 -5 -4 -3 -2 -1 0 0 20 40 60 80 100
y / cm Time t / h

Fig. 2. Comparison of different numerical models and experimental data for a Fig. 3. Comparison of predicted pH profile to experimentally determined profile in
corroding AISI 304 stainless steel crevice. (See above-mentioned references for AISI 304 stainless steel crevice at position (0.0045, 1), (0.0045, 4), (0.0045,
further information.) 7.5), respectively.
 W. Sun et al. / Corrosion Science 78 (2014) 233–243 239

mesh quality, selecting an appropriate re-meshing method to
ensure the mesh quality of the computational domain is very
important. In this work, the automatic re-meshing technique is ap-
plied to re-mesh the geometry. The mesh quality distribution of
computational domain and the deformed crevice geometry are
shown in Fig. 4(a) and (b), respectively. It can be seen that the com-
putational domain is re-meshed during the calculation process.
The mesh quality of the re-meshed computational domain is larger
than 0.378, indicating that the automatic re-meshing can ensure
high mesh quality. To re-mesh the computation domain, follow
these steps: 1 add a stop condition; 2 view the deformed mesh;
3 copy the solution; 4 create a deformed configuration; 5 remesh
the deformed configuration; 6 continue solving with the new mesh
[51]. In COMSOL MultiPhysicsÒ, the stop condition of computation
in mesh I can be invoked when the mesh quality decreases to a
given value (this value is set to 0.2 in this work). When the mesh
quality is smaller than the given value, the computation is termi-
nated and the computational domain is re-drawn and re-meshed
(such a new mesh is denoted as mesh II). Subsequently, the com-
putation is re-started for further time steps. The above procedure
is repeated iteratively throughout the computation scheme in or-
der to obtain a smooth meshing without entanglement along the
moving interface. From Fig. 4(a) or (b), it can be seen that the ano-
dic dissolution and the corrosion product precipitation lead to
large deformation near the crevice mouth. Then the elements
may elongate considerably there, which could potentially result
in mesh entanglement for further deformations. However, the
automatic re-meshing method applied in solving the ALE model

Fig. 4. The geometry deformation of crevice predicted by current model. (a) Initial mesh quality distribution; (b) Mesh quality distribution at 90 h.
240 W. Sun et al. / Corrosion Science 78 (2014) 233–243

eliminates those mesh elongation and smearing. Furthermore, it 3.5

-9 10
k=10×10 m s mol
-1 -3 Corrosion Product
provides a smoother mesh with high quality as shown in Deposition at the
3.0 -9 10 -1
k=9×10 m s mol
-3
Fig. 4(b). As crevice corrosion proceeds, the corrosion depth in- -9 10 -1 -3
Crevice Mouth
k=7×10 m s mol
creases with time. Meanwhile, more and more corrosion product 2.5 -9 10 -1 -3
k=5×10 m s mol
may precipitate outside the crevice. Fig. 5 shows the simulated -9 10 -1
k=3×10 m s mol
-3
2.0
morphologies of corrosion product deposition and crevice wall at

-log (-y / cm)

-9 10 -1 -3
k=1×10 m s mol
different time. It can be observed that the geometry deformation 1.5
caused by corrosion product deposition near the crevice mouth is
rather obvious, while the mesh quality remains as high as 0.65 at 1.0
Corrosion Depth
90 h. Therefore, it can be concluded that the ALE model established 0.5 Original Crevice
in this work is enable to track the movement of corroded crevice Wall
wall and corrosion product deposition and to provide high mesh 0.0
quality timely to ensure high precision calculation results at the -0.5
same time.
In order to detailedly investigate the influences of corrosion -1.0
-0.025 -0.020 -0.015 -0.010 -0.005 0.000
product deposition on crevice corrosion, models with different val-
x / cm
ues of k are calculated via a parameter sweep method. Fig. 6 re-
veals the calculated deformation of crevice wall under different Fig. 6. Crevice deformation under different reaction constant of corrosion product
values of k at 90 h. In order to present the crevice deformation deposition at 90 h.
more clearly, the vertical axis is selected as log(y). For different
precipitation reaction constant, it can be seen that there is always a than the original profile of crevice wall. And the presence of such
certain region along the crevice wall where the coordinate value in a region is attributed to the electrochemical dissolution of crevice
x-coordinate (the coordinate system is shown in Fig. 1) is smaller wall. Meanwhile, as shown in Fig. 6, the corrosion product precip-

Fig. 5. Numerical morphologies of corrosion product deposition and crevice wall at different time. (a) 0 h; (b) 30 h; (c) 60 h; (d) 90 h.
 W. Sun et al. / Corrosion Science 78 (2014) 233–243 241

3h crevice corrosion. Fig. 6 shows that the crevice closure and the cor-
3.0
20 h rosion depth inside the crevice increase with the increase of the
2.5 40 h precipitation reaction constant (marked by cyan arrow in Fig. 6).
60 h Hence, it can be inferred that crevice corrosion is more severe
2.0 80 h
when crevice product precipitates at the crevice mouth. Fig. 7 dis-
90 h
plays the time-dependent crevice deformation. It is obvious that
-log (-y / cm)

1.5
Original Creivce the corrosion depth and the thickness of corrosion product deposi-
1.0 Wall tion increase with time. Moreover, the corroded area inside the
crevice also increases with time (marked by cyan arrow). The cor-
0.5
roded crevice shape and the peak position of the corrosion depth
0.0 predicted by the ALE model qualitatively agree with the published
experimental studies [2,59–64].
-0.5 The corroded crevice shape is influenced by many factors, such
-1.0 as concentrations of dissolved oxygen and aggressive ions, applied
-0.025 -0.020 -0.015 -0.010 -0.005 0.000 potential or current, initial crevice geometry and corrosion prod-
x / cm uct. Onishi et al. [39] reported that the corroded crevice shape is
influenced by the width of initial crevice and the pH distribution.
Fig. 7. Crevice deformation at different time when reaction constant of corrosion
They found that a smaller initial crevice has a sharper and larger
product deposition is assumed to be 108 m10 s1 mol3.
peak for the corrosion depth; the high pH solution slows down
the corrosion rate. Jakupi [63] suggested that the propagation pre-
3.5
dominantly leads to penetration into the alloy at low applied cur-
ip=0.020 A/m
2
Obvious Corrosion Damage
rents, whereas corrosion damage spreads across the alloy at higher
3.0 2 at the Crevice Mouth applied currents. It should be noted that the corroded crevice
ip=0.015 A/m
shape shown in Fig. 6 and 7 may be just a typical shape observed
2.5 ip=0.010 A/m
2
in reported experiments. Due to the diversity of actual conditions
2.0 in the crevice, corroded crevice shapes are diverse. Matsuhashi
et al. [65] investigated the crevice corrosion of stainless steel in
-log (-y / cm)

1.5
natural seawater. The test pieces were subjected to a potentiostatic
1.0 test for about 921- 51960 s arbitrarily at 440 mV vs. SSC. It is noted
Original Crevice that the applied potential is always higher than its natural poten-
0.5 Wall
tial during the whole immersion process. That is to say that the
0.0
crevice is under anodic polarisation. Then they found the corrosion
damage occurred firstly at the crevice mouth and then gradually
-0.5 developed towards the crevice tip, which is different from the typ-

ical one shown in Fig. 6 and 7. In this paper, ip = 0.016 A/m2 (an
-1.0
-0.025 -0.020 -0.015 -0.010 -0.005 0.000 average result of Eq. (4)) represents the situation that there is no

x / cm applied potential or current. When ip < ip , the crevice is under ano-
dic polarisation. In contrast, the crevice is under cathodic protec-
Fig. 8. Crevice deformation under different current density applied at artificial tion. Fig. 8 shows the crevice profiles under different ip. It can be
boundary at 90 h.
seen that both the corrosion depth and the corroded area increase
itates inside the crevice in the area where the x-coordinate value is with the increase of ip. And the corrosion product does not deposit
smaller than 45 lm. These numerical results also indicate that faster inside the crevice than the electrochemical dissolution of
the crevice geometry greatly deforms at the propagation stage of crevice mouth when ip increases to 0.02 A/m2. Such a corroded

O2 Na+, Cl-

Bulk solution Diffusion Electro-migration & diffusion

Deposition of corrosion products

reduces the crevice width
OH-
Couple plate Fe 2+ + (2 - x)Cl- + x OH - → Fe(OH)x Cl 2-x or
4Fe(OH)2 + Cl - + 2H 2 O → Fe3II Fe III (OH)8 Cl ⋅ 2H 2 O + e

Metallic ions release

Metal dissolution enlarged

the crevice width

M → M n + + ne Homogeneous hydrolysis or
reaction occur in crevice
M n + + xH 2 O → M(OH) +x (n - x) + xH +
M n + + x Cl → MCl +x (n - x)

Crevice

Fig. 9. Schematic of the corrosion behaviour of AISI 304 stainless crevice in NaCl solution.
242 W. Sun et al. / Corrosion Science 78 (2014) 233–243
crevice shape qualitatively agrees with the experimental observa-
tion of Matsuhashi et al. [65] and partly agrees with the experi-
ment of Jakupi et al. [63]. Consequently, it is suggested that the
current model can be potentially applied to predict the corroded
crevice shape.
The long-term crevice corrosion behaviour is described in Fig. 9.
The reduction of oxygen occurs mainly on the cathode wall gener-
ating OH ions, while the electrochemical dissolution of crevice
wall occurs inside the crevice releasing Fe2+ ions. When the OH
ions, the dissolved oxygen and the Fe2+ ions are transported to
the crevice mouth, reaction takes place to produce corrosion prod-
ucts at the crevice mouth. The corrosion product subsequently pre-
cipitates at the crevice mouth closing the crevice mouth. The
deposition of corrosion product promotes the ‘‘occlusion effect’’
of the crevice. Therefore, the mass transfer between the crevice
solution and the bulk solution becomes more difficult and the IR
drop inside the crevice becomes larger which leads to further sta-
bilisation of crevice corrosion. The concentration gradients, crevice
solution composition and crevice geometry consequently enlarge
the differences of physical and chemical environment along the
crevice, which may meet the initial condition for hydrogen evolu-
tion towards the crevice tip [10].
It has been reported that the hydrogen evolution can also occur
at the crevice tip even without considering the geometric deforma-
tion caused by the deposition of corrosion products [4,10]. The fact
indicates that the hydrogen evolution in current study may be
mainly introduced by the elongated geometric structure of crevice.
However, previous papers argued that the hydrogen evolution,
which occurs during the subsurface intergranular attack of
AA2024-T3 aluminium alloy, can be caused by the precipitation
of corrosion product [66,67]. Therefore, it can be reasonably con-
cluded that hydrogen evolution during crevice corrosion is induced
by the deposition of the corrosion product or simply by the elon-
gated geometric structure in a corroding crevice. Finally, it is pro-
posed that the current model not only describes the time-
dependent evolution of crevice corrosion process, but also provides
a potential way to understand the mechanism of other forms of
localised corrosion, such as pitting corrosion, stress corrosion crack
and intergranular attack.
4. Conclusions
An arbitrary Lagrangian–Eulerian model for analysing the time-
dependent crevice corrosion has been developed in this work. The
current model comprises a transient potential distribution calcula-
tion, a mass balance calculation and an update of crevice geometry.
In the transient potential calculation, Ohm’s law, the continuity
equation and Gauss’ law are adopted to analyse the electrodynam-
ics of the crevice corrosion system. In the mass balance calculation,
the conventional Nernst–Plank equation is applied to satisfy the
mass conservation by considering diffusion, electromigration, elec-
trolyte electroneutrality, hydrolysis reactions and electrochemical
reactions. In the update of crevice geometry, the ALE method is
used to conduct the analysis of domain deformation, which is
caused by metal dissolution or corrosion product deposition, with
automatic remeshing schemes.
The accuracy of the established model is validated by Alavi and
Cottis’ experiment and other corresponding experiments. The
numerical results are in good agreement with the experimental re-
sults. And the current model predicts that the corrosion product
precipitating at the crevice mouth closes the crevice mouth and
can further exacerbate crevice corrosion. The current model is also
capable of predicting the corroded crevice shape. Furthermore, the
ALE model proposed herein is able to predict detailed time-depen-
dent evolution of crevice corrosion. And the current model also
helps researchers to better understand the mechanism of crevice
corrosion and other relevant localised corrosion, such as stress
corrosion crack and pitting corrosion. However, it should be noted
that the numerical results are qualitative predictions because the
parameters for the corrosion product deposition kinetics are hypo-
thetical. Improvement in the ALE model of the crevice corrosion
and further work for validation of the proposed model is in
progress.
Acknowledgments
The authors thank the Special Support of Marine Public Service
Project (No. 201005028-3) and the support of the Fundamental Re-
search Funds for the Central Universities (No. 852012). The authors
would like to especially thank Dr. Qing Mao for his help on paper
writing.
References
[1] W. Wang, H. Sun, L. Sun, Z. Song, B. Zang, NumericaI simulation for crevice
corrosion of 304 stainless steeling sodium chloride solution, Chemical
Research in Chinese Universities 26 (2010) 822–828.
[2] M.I. Abdulsalam, Behaviour of crevice corrosion in iron, Corrosion Science 47
(2005) 1336–1351.
[3] K.L. Heppner, R.W. Evitts, J. Postlethwaite, Prediction of the crevice corrosion
incubation period of passive metals at elevated temperatures: Part I –
Mathematical model, Canadian Journal of Chemical Engineering 80 (2002)
849–856.
[4] G.F. Kennell, R.W. Evitts, K.L. Heppner, A critical crevice solution and IR drop
crevice corrosion model, Corrosion Science 50 (2008) 1716–1725.
[5] J.C. Walton, G. Cragnolino, S.K. Kalandros, A numerical model of crevice
corrosion for passive and active metals, Corrosion Science 38 (1996) 1–18.
[6] K.L. Heppner, R.W. Evitts, J. Postlethwaite, Prediction of the crevice corrosion
incubation period of passive metals at elevated temperatures: Part II – Model
verification and simulation, Canadian Journal of Chemical Engineering 80
(2002) 857–864.
[7] M.I. Abdulsalam, H.W. Pickering, The effect of crevice-opening dimension on
the stability of crevice corrosion for nickel in sulfuric acid, Journal of the
Electrochemical Society 145 (1998) 2276–2284.
[8] A. Naganuma, K. Fushimi, K. Azumi, H. Habazaki, H. Konno, Application of the
multichannel electrode method to monitoring of corrosion of steel in an
artificial crevice, Corrosion Science 52 (2010) 1179–1186.
[9] J.W. Oldfield, W.H. Sutton, Crevice corrosion of stainless-steels. 2.
Experimental studies, British Corrosion Journal 13 (1978) 104–111.
[10] G.F. Kennell, R.W. Evitts, Crevice corrosion cathodic reactions and crevice
scaling laws, Electrochimica Acta 54 (2009) 4696–4703.
[11] N.J. Laycock, J. Stewart, R.C. Newman, The initiation of crevice corrosion in
stainless steels, Corrosion Science 39 (1997) 1791–1809.
[12] S.M. Sharland, A review of the theoretical modeling of crevice and pitting
corrosion, Corrosion Science 27 (1987) 289–323.
[13] G. Bombara, M. Cavallini, Stress-corrosion cracking of bone implants,
Corrosion Science 17 (1977) 77–85.
[14] G.R. Engelhardt, D.D. Macdonald, M. Urquidi-Macdonald, Development of fast
algorithms for estimating stress corrosion crack growth rate, Corrosion Science
41 (1999) 2267–2302.
[15] R.W. Evitts, Modelling of Crevice Corrosion, Thesis (Ph.D.), University of
Saskatchewan, Saskatoon, SK, 1997.
[16] M.I. Abdulsalam, H.W. Pickering, Effect of the applied potential on the
potential and current distributions within crevices in pure nickel, Corrosion
Science 41 (1999) 351–372.
[17] K.H. Cho, M.I. Abdulsalam, H.W. Pickering, The effect of electrolyte properties
on the mechanism of crevice corrosion in pure iron, Journal of the
Electrochemical Society 145 (1998) 1862–1869.
[18] H.W. Pickering, The significance of the local electrode potential within pits,
Crevices and Cracks, Corrosion Science 29 (1989) 325–341.
[19] M. Pourbaix, Theoretical and experimental considerations in corrosion testing,
Corrosion Science 12 (1972) 161–190.
[20] B.F. Brown, Concept of occluded corrosion cell, Corrosion 26 (1970) 349–350.
[21] A. Naganuma, K. Azumi, Observation of corrosion behavior of stainless steels in
a salt manufacturing plant environment using the multichannel electrode
method, Corrosion Science 53 (2011) 1165–1173.
[22] M. Pourbaix, in: J.C. Scully (Ed.), The Theory of Stress Corrosion Cracking in
Alloys, Proceedings of the NATO Ericeira Conference, Brussels, Belgium, 1971,
17.
[23] S.T. Kim, S.Y. Kim, I.S. Lee, Y.S. Park, M.C. Shin, Y.S. Kim, Effects of shielding
gases on the microstructure and localized corrosion of tube-to-tube sheet
welds of super austenitic stainless steel for seawater cooled condenser,
Corrosion Science 53 (2011) 2611–2618.
 W. Sun et al. / Corrosion Science 78 (2014) 233–243
[24] N.S. Zadorozne, C.M. Giordano, M.A. Rodriguez, R.M. Carranza, R.B. Rebak,
Crevice corrosion kinetics of nickel alloys bearing chromium and
molybdenum, Electrochimica Acta 76 (2012) 94–101.
[25] S. Nagarajan, N. Rajendran, Crevice corrosion behaviour of superaustenitic
stainless steels: dynamic electrochemical impedance spectroscopy and
atomic force microscopy studies, Corrosion Science 51 (2009)
217–224.
[26] S.M. Sharland, P.W. Tasker, A mathematical-model of crevice and pitting
corrosion. 1. The physical model, Corrosion Science 28 (1988) 603–620.
[27] S.M. Sharland, C.P. Jackson, A.J. Diver, A finite-element model of the
propagation of corrosion crevices and pits, Corrosion Science 29 (1989)
1149–1166.
[28] B. Vuillemin, R. Oltra, R. Cottis, D. Crusset, Consideration of the formation of
solids and gases in steady state modelling of crevice corrosion propagation,
Electrochimica Acta 52 (2007) 7570–7576.
[29] F.M. Song, Simple algorithms for 1D oxygen concentration profile in an
occluded region, Corrosion Science 50 (2008) 3287–3295.
[30] F.M. Song, Theoretical investigation into time and dimension scaling for
crevice corrosion, Corrosion Science 57 (2012) 279–287.
[31] F.M. Song, Predicting the chemistry, corrosion potential and corrosion rate in a
crevice formed between substrate steel and a disbonded permeable coating
with a mouth, Corrosion Science 55 (2012) 107–115.
[32] F.M. Song, N. Sridhar, Modeling pipeline crevice corrosion under a disbonded
coating with or without cathodic protection under transient and steady state
conditions, Corrosion Science 50 (2008) 70–83.
[33] F.M. Song, A mathematical model developed to predict the chemistry and
corrosion rate in a crevice of variable gap, Electrochimica Acta 56 (2011)
6789–6803.
[34] M. Bohm, J. Devinny, F. Jahani, G. Rosen, On a moving-boundary system
modeling corrosion in sewer pipes, Applied Mathematics and Computation 92
(1998) 247–269.
[35] S. Scheiner, C. Hellmich, Stable pitting corrosion of stainless steel as diffusion-
controlled dissolution process with a sharp moving electrode boundary,
Corrosion Science 49 (2007) 319–346.
[36] J. Xiao, S. Chaudhuri, Predictive modeling of localized corrosion: an application
to aluminum alloys, Electrochimica Acta 56 (2011) 5630–5641.
[37] K.B. Deshpande, Validated numerical modelling of galvanic corrosion for
couples: magnesium alloy (AE44)-mild steel and AE44-aluminium alloy
(AA6063) in brine solution, Corrosion Science 52 (2010) 3514–3522.
[38] K.B. Deshpande, Numerical modeling of micro-galvanic corrosion,
Electrochimica Acta 56 (2011) 1737–1745.
[39] Y. Onishi, J. Takiyasu, K. Amaya, H. Yakuwa, K. Hayabusa, Numerical method
for time-dependent localized corrosion analysis with moving boundaries by
combining the finite volume method and voxel method, Corrosion Science 63
(2012) 210–224.
[40] S. Sarkar, J.E. Warner, W. Aquino, A numerical framework for the modeling of
corrosive dissolution, Corrosion Science 65 (2012) 502–511.
[41] Y. Xu, H.W. Pickering, The initial potential and current distributions of the
crevice corrosion process, Journal of the Electrochemical Society 140 (1993)
658–668.
[42] W. Sun, G.C. Liu, L.D. Wang, Y. Li, A mathematical model for modeling the
formation of calcareous deposits on cathodically protected steel in seawater,
Electrochimica Acta 78 (2012) 597–608.
[43] W. Sun, G.C. Liu, L.D. Wang, T.T. Wu, Y. Liu, An arbitrary Lagrangian–Eulerian
model for studying the influences of corrosion product deposition on
bimetallic corrosion, Journal of Solid State Electrochemistry 17 (2013) 829–
840.
[44] A. Alavi, R.A. Cottis, The determination of Ph, potential and chloride
concentration in corroding crevices on 304-stainless steel and 7475-
aluminum alloy, Corrosion Science 27 (1987) 443–451.
[45] M.L.A. Gil, A. Santos, M. Bethencourt, T. Garcia, S. Fernandez-Bastero, A. Velo, L.
Gago-Duport, Use of X-ray and other techniques to analyse the phase
transformation induced in archaeological cast iron after its stabilisation by
the electrolytic method, Analytica Chimica Acta 494 (2003) 245–254.
243
[46] T. Misawa, T. Kyuno, W. Suetaka, Shimodai.S, Mechanism of atmospheric
rusting and effect of Cu and P on rust formation of low alloy steels, Corrosion
Science 11 (1971) 35.
[47] P. Refait, J.M.R. Genin, The mechanisms of oxidation of ferrous
hydroxychloride beta-Fe2(OH)3Cl in aqueous solution: the formation of
akaganeite vs goethite, Corrosion Science 39 (1997) 539–553.
[48] P.H. Refait, M. Abdelmoula, J.M.R. Genin, Mechanisms of formation and
structure of green rust one in aqueous corrosion of iron in the presence of
chloride ions, Corrosion Science 40 (1998) 1547–1560.
[49] C. Remazeilles, P. Refait, On the formation of beta-FeOOH (akaganeite) in
chloride-containing environments, Corrosion Science 49 (2007) 844–857.
[50] Z.S. Wang, C.C. Xu, X.Q. Dong, Localized corrosion and phase transformation of
simulated archaeological iron, Chinese Journal of Chemical Engineering 16
(2008) 299–305.
[51] COMSOL MultiphysicsÒ, version 4.3, <www.comsol.com>.
[52] H.Y. Chang, Y.S. Park, W.S. Hwang, Initiation modeling of crevice corrosion in
316L stainless steels, Journal of Materials Processing Technology 103 (2000)
206–217.
[53] B. Malki, T. Souier, B. Baroux, Influence of the alloying elements on pitting
corrosion of stainless steels: a modeling approach, Journal of the
Electrochemical Society 155 (2008) C583–C587.
[54] J. Donea, A. Huerta, J.-P. Ponthot, A. Rodriguez-Ferran, Encyclopedia of
Computational Mechanics, E. Wiley, R. Stein, R. de Borst, T.J.R. Hughes (Eds.),
John Wiley and Sons, Chichester, 2004.
[55] J. Wu, H. Wang, X. Chen, P. Cheng, G.F. Ding, X.L. Zhao, Y. Huang, Study of a
novel cathode tool structure for improving heat removal in electrochemical
micro-machining, Electrochimica Acta 75 (2012) 94–100.
[56] A.M. Winslow, Equipotential Zoning of Two-dimensional Meshes, University of
California, Lawrence Livermore Laboratory, Livermore, California, Report No.
UCRL-7312, 1963.
[57] F.M. Song, D.W. Kirk, J.W. Graydon, D.E. Cormack, Steel corrosion under a
disbonded coating with a holiday – Part 1: the model and validation, Corrosion
58 (2002) 1015–1024.
[58] S.P. White, G.J. Weir, N.J. Laycock, Calculating chemical concentrations during
the initiation of crevice corrosion, Corrosion Science 42 (2000) 605–629.
[59] M.H. Wang, H.W. Pickering, Y. Xu, Potential distribution, shape evolution, and
modeling of pit growth for Ni in sulfuric-acid, Journal of the Electrochemical
Society 142 (1995) 2986–2995.
[60] J.S. Lee, M.L. Reed, R.G. Kelly, Combining rigorously controlled crevice
geometry and computational modeling for study of crevice corrosion scaling
factors, Journal of the Electrochemical Society 151 (2004) B423–B433.
[61] B.P. Cai, Y.H. Liu, X.J. Tian, F. Wang, H. Li, R.J. Ji, An experimental study of
crevice corrosion behaviour of 316L stainless steel in artificial seawater,
Corrosion Science 52 (2010) 3235–3242.
[62] P. Jakupi, F. Wang, J.J. Noel, D.W. Shoesmith, Corrosion product analysis on
crevice corroded Alloy-22 specimens, Corrosion Science 53 (2011)
1670–1679.
[63] P. Jakupi, J.J. Noel, D.W. Shoesmith, The evolution of crevice corrosion damage
on the Ni–Cr–Mo–W alloy-22 determined by confocal laser scanning
microscopy, Corrosion Science 54 (2012) 260–269.
[64] H. Shu, F.M. Al-Faqeer, H.W. Pickering, Pitting on the crevice wall prior to
crevice corrosion: Iron in sulfate/chromate solution, Electrochimica Acta 56
(2011) 1719–1728.
[65] R. Matsuhashi, S. Tsuge, Y. Tadokoro, T. Suzuki, Estimation of crevice corrosion
life time for stainless steels in seawater environments, Nippon Steel Technical
Report 99 (2010) 62–72.
[66] A.M. Glenn, T.H. Muster, C. Luo, X. Zhou, G.E. Thompson, A. Boag, A.E. Hughes,
Corrosion of AA2024-T3 Part III: propagation, Corrosion Science 53 (2011) 40–
50.
[67] A.E. Hughes, A. Boag, A.M. Glenn, D. McCulloch, T.H. Muster, C. Ryan, C. Luo, X.
Zhou, G.E. Thompson, Corrosion of AA2024-T3 Part II co-operative corrosion,
Corrosion Science 53 (2011) 27–39.