modelling of pitting corrosion in marine and offshore steel

lable at ScienceDirect

Journal of Loss Prevention in the Process Industries 37 (2015) 39e62

Contents lists avai

Journal of Loss Prevention in the Process Industries

journal homepage: www.elsevier .com/locate/ j lp

Modelling of pitting corrosion in marine and offshore steelstructures e A technical review

Jyoti Bhandari a, Faisal Khan b, *, Rouzbeh Abbassi a, Vikram Garaniya a, Roberto Ojeda a

a Australian Maritime College, University of Tasmania, Launceston, TAS 7250, Australiab Safety and Risk Engineering Group (SREG), Faculty of Engineering & Applied Science, Memorial University of Newfoundland, St. John's, Newfoundland A1B3X5, Canada

a r t i c l e i n f o

Article history:Received 27 May 2015Accepted 21 June 2015Available online 29 June 2015

Keywords:Pitting corrosionOffshore structuresPredictionCorrosion modellingSafety assessmentSteel

* Corresponding author.E-mail address: [email protected] (F. Khan).

http://dx.doi.org/10.1016/j.jlp.2015.06.0080950-4230/© 2015 Elsevier Ltd. All rights reserved.

a b s t r a c t

Corrosion is a major cause of structural deterioration in marine and offshore structures. It affects the lifeof process equipment and pipelines, and can result in structural failure, leakage, product loss, environ-mental pollution and the loss of life. Pitting corrosion is regarded as one of the most hazardous forms ofcorrosion for marine and offshore structures. The total loss of the structure might be very small, but localrate of attack can be very large and can lead to early catastrophic failure. Pitting corrosion is a localizedaccelerated dissolution of metal that occurs as a result of a breakdown in the protective passive film onthe metal surface. It has been studied for many years; however, the structural failure due to pit char-acteristics is still not fully understood. Accurate pit depth measurements, precise strength assessmenttechniques, risk analysis due to pitting, and the mathematical relationship of the environmental factorsthat causes pitting failure are also factors, which need further understanding. Hence this paper focuseson these issues. It reviews and analyses the current understanding of the pitting corrosion mechanismand investigates all possible factors that can cause pitting corrosion. Furthermore, different techniquesemployed by scientists and researchers to identify and model the pitting corrosion are also reviewed andanalysed. Future work should involve an in-depth scientific study of the corrosion mechanism and anengineering predictive model is recommended in order to assess failure, and thereby attempt to increasethe remaining life of offshore assets.

© 2015 Elsevier Ltd. All rights reserved.

1. Introduction

The scientific exploration and exploitation of the ocean depth isnow proceeding at a greatly accelerated rate (Schumacher, 1979).Consequently there has been particularly rapid progress in thedevelopment of equipment for deep-ocean investigation and un-derwater activities. The oil and gas industries have built a largenumber of offshore platforms, pipelines, ships, and underwaterstorage and shore facilities. However, the structural reliability ofthese structures is not fully understood (Stewart and Al-Harthy,2008). These marine and offshore structures are typically con-structed from steel and have the highest rate of critical incidentsdue to corrosion deterioration (Maureen et al., 2013). Corrosion iswell defined as the destructive attack on amaterial by reactionwithits environment (Popoola et al., 2013; Melchers, 2004a; Melchers,

2005a; 2010; Melchers and Jeffrey, 2008a). This material degrada-tion results in loss of mechanical properties of the structure such asstrength, ductility and impact strength. Material degradation leadsto loss of material and, at times, to ultimate failure (Popoola et al.,2013). Further, corrosion is also of major economic significance. TheWorld Corrosion Organization (WCO) estimates that the annualcost of corrosion worldwide is around $US2.4 trillion (3% of theworld's GDP) (Vel�azquez et al., 2014). The economic cost for allforms of corrosion in advanced economies such as the United Statesof America was recently estimated at around 4% of gross nationalproduct (GNP) (Melchers, 2008a). The environmental cost ofcorrosion can also be high; according to the Offshore HydrocarbonRelease Statistics and Analysis report 2002e2003 (HSR) (HSR,2003), a total of 2313 hydrocarbon releases were reported fromoffshore facilities in nine and half years. The most common causeswere ‘mechanical failure’ due to corrosion and other relateddegradation. A total of 1034 incidents (44.7% of the total incidents)were caused as a result of corrosion (HSR, 2003; Khan and Howard,2007).

J. Bhandari et al. / Journal of Loss Prevention in the Process Industries 37 (2015) 39e6240

Amongst all types of corrosion, pitting is the most common anddamaging form of corrosion in marine and offshore structures (Liet al., 2011). Pitting corrosion is very dangerous, widespread, anddifficult to detect; hence, it has been a matter of great concern tomarine and offshore industries for several years (Sidharth, 2009).Pitting corrosion is defined by Szklarska (Szklarska-Smialowska,1986) as a localized dissolution of metals that occurs due to thebreakdown of the protective passive film - notably the protectivecoating and paint on metal surfaces (Melchers, 2004a, b; Szklarska-Smialowska, 1986; Stewart, 2009). Pitting corrosion can also occurduring active dissolution if certain sections of the sample are moresusceptible and dissolve faster than the rest of the surface (Frankel,1998). Szklarska et al. (Szklarska-Smialowska, 1986) describespitting as a form of localized corrosion in which metal is removedpreferentially from vulnerable areas of the surface. More specif-ically, the dissolutions lead to the formation of cavities in passivatedmetals or alloys, which are exposed to solutions containingaggressive anions. Most pitting failures in the offshore sectors arecaused by chloride and chloride containing ions. It is very likely forpitting to occur when protective measures such as paint coating,galvanizing or cathodic protection are ineffective (Abdel-Ghanyet al., 2008). Pitting is one of the most vicious and insidiousforms of corrosion; the attack is extremely localized resulting inholes in the structure and thereby causing failures (Abdel-Ghanyet al., 2008). History shows that pitting corrosion is a dominantcause of structural failure in marine and offshore sectors. Thereason for this is due to the well-known fact that seawater is anaggressive corrosive environment and the structures are generallyfabricated with alloy steel which favours pitting corrosion. Thepitting in these structures is often very severe, not only undersustained immersed conditions, but also under general exposure toatmospheric conditions (Sidharth, 2009). The effects of pittingcorrosion can be catastrophic, as illustrated by several incidentsreported in the offshore oil and gas sectors. In March 1965 a naturalgas pipeline exploded in Louisiana, killing 17 people including 9children. The heat from the explosionwas very intense; 6 cars and 3trucks melted and 5 houses were scattered over a large area. In July1988 there was a massive leakage of gas condensate on Piper Alphadue to pitting corrosionwhich, when it ignited, caused an explosionleading to large oil fires. The scale of the disaster was enormous,killing 167 people on board, and it is regarded as the deadliestaccident in the history of the oil and gas industry (Maureen et al.,2013). In April 1992 a sewer explosion in Guadalajara, Mexico kil-led 215 people. In this incident a series of blasts damaged 1600buildings and injured 1500 people (Roberge, 2008). Roberge et al.(Roberge, 2008) described it as being one of the most catastrophicaccidents resulting from a single pit. A recent example of pittingcorrosion-related failure is the BP ULA accident in Norway. InSeptember 2012, an estimated 125 barrels of oil and 1600 kg of gasleaked at the North Sea platform due to pit in a valve; despite therebeing no injuries, production was stopped for 67 days (Maureenet al., 2013). Similarly, in December 2012 a buried, half a metrediameter, high-pressure natural gas pipeline operated by ColumbiaGas Transmission ruptured due to pitting corrosion. There were nofatalities; however, several houses were damaged and about 2million standard cubic metres of natural gas was released andburned. The total cost was expected to be about $9 million dollars(Board, 2014). Consequently, significant attention should be givento predict the occurrence of pitting corrosion in offshore structuresand adequate measures should be taken to prevent and control theconsequences.

Pitting corrosion in offshore steel structures has particularimportance where containment is critical, such as for pressurevessels, boilers, turbine blades and for metallic containers for toxicmaterials. It is also significant for localized structural strength in

piping, tanks, ships and marine structures (Abood, 2008; Melchers,1994). Melchers (1994) stressed the importance of pitting corrosionknowledge as being crucial for the marine and offshore sectors.They recommended a deep knowledge of pitting corrosion e itseffects, and the application of preventative methods e as this willassist in reducing material losses resulting from same (Melchers,2009). It will also improve the safety of operating equipment(Abood, 2008).

Pitting corrosion has been studied for several decades andconsiderable understanding of the pitting phenomenon has beengenerated. However in depth knowledge of pitting modelling andpitting measurement is still lacking (Frankel, 1998). There is also alimited understanding of why the pitting corrosion rate is notconstant over time and why data tends to be scattered. Hence, it isnecessary to review what actually happens when a steel surface isexposed to natural seawater and how this process progresses withtime (Melchers, 2008a).

Many researchers have studied factors affecting pitting corro-sion in marine and offshore environments (Melchers, 2004a, b;Caines et al., 2013; Soares et al., 2009; Valor et al., 2007, 2010;Zamaletdinov, 2007; Zaya, 1984); however, the factors investigatedare very limited and there is still a potential to study them morethoroughly. In this study, all possible factors that can cause pittingcorrosion will be discussed and the effects of a number of factors ininitiating the pitting corrosion will be summarized. In additiondifferent techniques employed by scientists and researchers, toidentify and model the pitting corrosion, are focused on in subse-quent sections.

2. General description of pitting corrosion

Pitting corrosion is defined as localized corrosion of a metalsurface, confined to a point or small area that takes the form ofcavities (Abood, 2008). Major consequences of pitting are thebreakdown of passivity; in general, pitting occurs when there is abreakdown of surface films exposed to the pitting environment.Pitting corrosion occurs when discrete areas of a material undergorapid attack while most of the adjutant surface remains virtuallyunaffected (Roberge, 2013). The attack leads to characteristic formsof cavity, such as pits or crevices in the metal surface (Melchers,1994). Research shows the common factors contributing to theinitiation and propagation of pitting corrosion are (Roberge, 2008):

� Localized chemical or mechanical damage to the protectiveoxide film

� Factors that can cause breakdown of a passive film, such asacidity, low dissolved oxygen concentrations and high chlorideconcentrations; these are likely to turn a protective oxide filmless stable, and thereby initiate pit

� Localized damage to, or poor application of, protective coating� Presence of non-uniformities in the metal structure of thecomponent such as non-metallic inclusions.

Pitting corrosion can produce pits with their mouth open (un-covered) or be covered with a semi-permeable membrane ofcorrosion products. Pits can be either hemispherical or cup-shaped.In some cases they are flat-walled, revealing the crystal structure ofthe metal but they may also have a completely irregular shape.Fig. 1 shows the common pit shapes divided in two groups namely:trough pits (upper) and sideway pits (lower):

Pitting cavities may fill with corrosion products and form capsover the pit cavities, sometimes creating nodules or tubercles.While the shapes of the pits vary widely, as evidenced in Fig. 1, theyare usually roughly saucer-shaped, conical, or hemispherical forsteel and many associated alloys (Roberge, 2008).

Fig. 1. Sketch of common pit shapes.

J. Bhandari et al. / Journal of Loss Prevention in the Process Industries 37 (2015) 39e62 41

Pitting corrosion is complicated in nature; according to Aboodet al. (Abood, 2008) the oxide films that form on different metalsvary one from another in electronic conduction, porosity, thicknessand state of hydration (Abood, 2008). Zaya (1984) studied pittingtheory and stages of pit development; the schematic of differentstages for the development of an individual pit can be seen in theFig. 2. Zaya (1984) distinguished various stages of the pittingcorrosion process and divided them into four stages (see Fig. 2).Stage 0 represents an un-attacked metallic surface which iscompletely covered with the passive films. Stage 1 involves therupture of the passive layer; the substrate is still protected exceptfor a small patch in contact with the electrolyte. The dimension ofthe small patch in stage 1 can be smaller or comparable to thethickness of the passive film. Subsequently, the dissolution of thesubstrate begins. Stage 2 is reached when the conditions for pitgrowth are met and repassivation cannot occur anymore i.e. the pitbegins to grow. Therefore, in stage 3 the dissolution of the substratebegins to grow and the pit becomes about 1e10 mm which can be

Fig. 2. Schematic stages of pit development (adopted from Zaya (1984)).

seen under optical microscope. Pits have a shape of hemisphere orof a polyhedron by stage 3. At the final stage 4 the pits can be seenwith the naked eye. The pit can have an irregular shape if partiallycovered around the mouth with solid corrosion products(Schumacher, 1979; Szklarska-Smialowska, 1986; Melchers, 1994;Zaya, 1984).

Many researchers, starting with Evens and Bannister in 1931(Evans et al., 1931; Evans, 1976), and later Richardson in 1973(Thompson et al., 1978), claim that local weak spots or defects arealways present in the passive films. Therefore stage 0 never existsand, immediately after immersion, the process starts at stage 1, i.e.metal and solution are in contact (Evans et al., 1931; Evans, 1976;Thompson et al., 1978). The induction period only corresponds tothe rate time necessary for the corrosion to be well developed anddetectable. Additionally, the corrosive attack is highly localised andthe precise location of pits appears to be unpredictable (Melchers,1994; Zaya, 1984).

Steel alloys, such as stainless steel and aluminium alloys, arecommonly used in marine and offshore sectors. These alloys arecomprised of passive films which are thin (nanometre-scale) oxidelayers that can form on the metal surface in marine environments(Roberge, 2008; Melchers, 1999). However, such passive films areoften susceptible to localized breakdown resulting in accelerateddissolution of the underlying metal (Schiroky et al., 2009). If theattack initiates on an open surface it is called pitting corrosion. Thisform of localized corrosion can lead to accelerated failure ofstructural components by perforation or by acting as an initiationsite of cracking (Melchers, 1994). Fig. 3 shows an example of thedeep pits on a metal surface.

Fig. 3. Deep pit in the metal (Roberge 2008, 2013).

Fig. 4. Schematic of anodic curve for a metal immersed in a solution containingaggressive ions (Adopted from Caines et al. (Szklarska-Smialowska, 1986; Frankel,1998; Caines et al., 2013).


Roberge et al. (Roberge, 2008) reported that the practicalimportance of pitting corrosion depends on the thickness of themetal and on the penetration rate. In general, it was found that therate of penetration decreases if the number of pits increases. This isattributed to closely spaced pits having to share the availableadjacent cathodic area which controls the corrosion current thatcan flow (Szklarska-Smialowska, 1986; Roberge, 2008, 2013).Abood et al. (Abood, 2008) studied different parameters that in-fluence pitting corrosion. The parameters investigated were limitedto environment, metal composition, pitting potential, temperatureand the surface conditions. Within these, environmental parame-ters were found to be the most critical factors. They includeaggressive ion concentration, pH, and inhibitor concentration(Szklarska-Smialowska, 1986; Roberge, 2008; Abood, 2008). Simi-larly, Frankel et al. (Frankel, 1998) provided an overview of thecritical factors that influence the pitting corrosion of metals. Theyfound that the effects of alloy composition, environment, pittingpotential and temperature are critical for pitting to occur. Thesecritical factors are further reviewed in the following section.

2.1. Stages of pitting corrosion

Pitting can consist in various stages: passive film breakdown, pitinitiation, metastable pitting, pit growth and pit stifling. Any ofthese stages may be considered to be critical (Szklarska-Smialowska, 1986; Abood, 2008; Alkire and Wong, 1988). Pittingcorrosion capitalizes on breakdown in the protective layer, eithernatural or applied, and provides a nucleation point for the forma-tion of pits in the presence of an electrolyte containing an aggres-sive anion (Caines et al., 2013). Once the passive film breaks downand a pit initiates, there is a possibility that a single pit will grow.The passive state is required for pitting to occur; however someresearchers believe that the minutiae of passive film compositionand structure play a minor role in the pitting process (Abood,2008). Regardless, pit growth is critical in practical applications offailure prediction in marine and offshore structures.

2.1.1. Passive film breakdownA passive film breaking theory was originally proposed by Agar

and Hoar (1947) in 1967 and was later extended on by many re-searchers (Zaya, 1984). Agar and Hoar (1947) suggested that theabsorption of the damaging ions on the surface of the passive filmproduces mutual repulsion which lowers the interfacial surfacetension. It was reported that when the repulsive force is sufficient,the passive film cracks (Zaya, 1984; Agar and Hoar, 1947). Thebreakdown of the passive film and the details of pit initiationcomprise the least understood aspect of the pitting phenomenon.Breakdown is a rare occurrence that happens very rapidly and on avery small scale, making direct detection of the breakdown difficult(Abood, 2008; Melchers, 1994; Zaya, 1984; Agar and Hoar, 1947).

Caines et al. (2013) described passive film breakdown in stain-less steel alloys commonly used in offshore installations. In thesealloys, the breakdown of the passive film provides sites for pitnucleation and, consequently, these breakdown sites are suscepti-ble to pitting corrosion. Passive films are present on the surface ofthe stainless steels in the presence of oxygen. At low temperatures,a true oxide layer is not formed; instead a thin passive film isformed and acts as a barrier and thereby provides corrosion resis-tance (Caines et al., 2013).

Fig. 4 illustrates the polarization curve (Melchers, 2008a;Szklarska-Smialowska, 1986; Frankel, 1998; Abood, 2008; Caineset al., 2013; Melchers and Ahammed, 1994) to estimate the sus-ceptibility of the metal alloy to pitting corrosion (Caines et al.,2013). The curve is used to find Epit and repassivation potential(ER). Higher Epit for a material in a given environment indicates

greater resistance to pitting (Szklarska-Smialowska, 1986; Caineset al., 2013; Strehblow and Marcus, 1995). Similarly, if the poten-tial is reduced below Epit, the surface may repassivate and pitgrowth can stop. However, if the potential is between Epit and ER,pitting can be expected (Szklarska-Smialowska, 1986; Caines et al.,2013; Strehblow and Marcus, 1995).

2.1.2. Pit initiationThe initiation stage of pitting is generally agreed to be of very

short duration. Melchers (1994) reported that the duration of pitinitiation can be as short as micro-seconds and that pitting initia-tion is strongly dependent on the surface conditions of thematerial.This is important in setting the overall areal pattern for subsequentlocalised corrosion (Melchers, 1994). Caines et al. (Caines et al.,2013) postulated that the initiation of pits is influenced by sur-face defects which may be due to manufacturing issues such asinstallation problems, improper maintenance procedures andenvironment changes (Caines et al., 2013). Caines et al. (Caineset al., 2013) also studied different factors that cause the site of pitinitiation. Some of these are:

� Damage to protective chemical or mechanical oxide layer� Environmental factors such as low pH, high chloride concen-tration causing protective layer breakdown

� Damage to applied protective coating� Discontinuities in the protective coating

Strehblow et al. (Strehblow and Marcus, 1995) and Frankel et al.(Frankel, 1998) represented pit initiation by categorizing it in threedifferent mechanisms: (a) penetration (b) adsorption and thinningand (c) film breaking mechanisms (Frankel, 1998). Penetrationmechanisms for pit initiation involves the transport of theaggressive anions through the passive film to the oxide interfacewhere aggressive dissolution is promoted (Frankel, 1998). Thismechanism is aided by the induction time for pitting and thepresence of chloride in an electrolyte. An adsorption theory ofpitting initiation is based on the material's adsorption of chlorideand oxygen. Strehblow et al. (Strehblow and Marcus, 1995)described an aspect of pit adsorption, based on X-ray PhotoelectronSpectroscopy (XPS) measurements, as being the exposure of Fe tochloride and other halides which cause the thickening of the pas-sive films, even under conditions where pit has not formed. This isas a result of the catalytically enhanced transfer of cations from theoxide to the electrolyte (Frankel, 1998; Strehblow and Marcus,


1995). The film breaking may occur due to mechanical stress atweak sites, or flaws, causing local breakdown events; howeverthese can also rapidly heal in nonaggressive environments (Frankel,1998; Strehblow and Marcus, 1995).

2.1.3. Metastable pittingMetastable pits are those that initiate and grow for a limited

period before repassivating, or before dying (Frankel, 1998; Wika,2012a). Pits which cease growing and repassivate in this mannerare described as metastable (Abood, 2008; Wika, 2012a; Pistoriusand Burstein, 1992a). The metastable pits may survive andbecome stable growing pits depending upon the composition of thematerial, the critical solution, mass-transport and the pitting po-tential at the bottom of the pit. Stable pitting corrosion occurs whenthe corrosion potential, Ecorr, exceeds the pitting potential, Epit(Frankel, 1998; Wika, 2012a). The measure of a material's ability toundergo stable pitting corrosion in a certain environment is definedas the pitting potential (Epit) (Pistorius and Burstein, 1992a). Galveleet al. (Galvele, 2005) described Epit as a limit where the growth ofpits happens above the surface, however a passive surface ismaintained below (Galvele, 2005).

Wika et al. (Wika, 2012a) described metastable pits as theincipient growth of initiated pits, and which must survive in orderto become stable growing pits. These are small pits, of only fewmicrons in size, which normally last only a few seconds before thesurface repassivates (Wika, 2012a).

Pistorius et al. (Pistorius and Burstein, 1992a) investigated thedevelopment of highly aggressive anolyte inside the metastable pitwhich has low pH as a result of hydrolysis of the dissolving metalcations. The anolyte also carries an enhanced concentration ofanions because these migrate into the pit to maintain analyticcharge neutrality. Hence, the pit growth becomes self-sustainingdue to the development of the aggressive anolyte in the pit.

2.1.4. Pit propagation (stable pitting)The pit propagation is the stage where the development of some

of the initiated pit occurs and, in particular, where its rate of growthincreases (Melchers, 1994; Caines et al., 2013; Wong and Alkire,1990). Caines et al. (Caines et al., 2013) defined this as the stagewhere pits grow and have the potential to increase beyond wallthickness, thereby leading to leaks (Caines et al., 2013). They alsoreported that certain conditions must be met for pits to propagate:

� Epit must be exceeded and remain above ER (repassivationpotential)

� An aggressive ion must be present� Localized breakdown of passive or applied film.

Melchers (1994) reported that the precise mechanics of pitpropagation is not fully understood; however it is known thathighly acidic conditions (i.e. low pH) is associated for pit to bepropagated (Melchers, 1994). Angal (2010) described pit growth inthe marine environment as an autocatalytic process (Angal, 2010).Fe2þ ions attract negative Cl� ions and, through the hydrolysis,creates a porous Fe(OH)2 cap over the pit. This process creates aself-propagating system where the increased acidity in the pitcavity increases corrosion of the steel walls of the pit (Caines et al.,2013; Angal, 2010). Some researchers (Frankel, 1998; Abood, 2008;Strehblow and Marcus, 1995; Wong and Alkire, 1990; Frankel andSridhar, 2008) reported that pit propagates at a rate that dependson material compositions, pit electrolyte concentration, and pit-bottom potential. It was suggested that, in order to understandpit growth and stability, it is essential to ascertain the rate-determining factors such as electrochemical reactions whichinclude charge-transfer process, ohmic effects, mass transport and

the combinations of these factors. Szklarska-Smialowska (1986)explained how the pit growth at low potential (i.e. below therange of limiting current densities) is controlled by a combinationof ohmic, charge transfer, and concentration over potential factors(Szklarska-Smialowska, 1986; Melchers, 1994). They suggested thatpit growth is nonlinear at higher chloride ion concentration.Additionally, nonlinear growth of the high chloride concentrationtn with n in the range of 0.33e0.5 has been suggested by Melchers(1994), Szklarska-Smialowska (1986) and Abood (2008).

3. Mechanism of pitting corrosion

Pits almost always initiate due to chemical or physical hetero-geneity at the surface such as inclusions, second phase particles,solute-segregated grain boundaries, flaws, mechanical damage, ordislocations (Szklarska-Smialowska, 1986). Most engineering alloyshave many or all such defects and a pit tends to form at the mostsusceptible sites first. For example, pits in stainless steels are oftenassociated with manganese sulphide (MnS) inclusions which arefound in most commercial steels. The role of MnS inclusions inpromoting the breakdown and localized corrosion of stainlesssteels has been recognized for some time (Sidharth, 2009;Szklarska-Smialowska, 1986; Frankel, 1998). Abood (2008)acknowledge that the pitting corrosion caused by passive filmbreakdown only occurs in the presence of aggressive anionic spe-cies, and that chloride ion is usually the cause. The severity of thepitting tends to vary with the amount of chloride concentration.The reason for the aggressiveness of chloride has been consideredfor some time and a number of investigations and examinationswere carried out in marine and offshore installations (Frankel,1998). Chloride is an anion of a strong acid and many metal cat-ions exhibit considerable solubility in chloride solutions. Hence, thepresence of the oxidizing agent (oxide) in a chloride-containingenvironment is extremely damaging and it can further enhancelocalized corrosion (Schiroky et al., 2009). Szklarska-Smialowska(1986) presented an experiment by means of ASTM G48, which isa common procedure when testing pitting corrosion, with thespecimens immersed in 6% of Fecl3. The results showed that theincrease in potential associated with oxidizing agents enhances thelikelihood of pitting corrosion (Szklarska-Smialowska, 1986;Frankel, 1998; Abood, 2008). Refer to Equation (1.1):

Me2 þ 2H2O/MeðOHÞþ þ Hþ (1.1)

Pitting corrosion normally starts with chloride rapidly pene-trating the protective oxide film which covers the metal surface.These points act as initiation sites for pitting corrosion (Melchers,1994; Schiroky et al., 2009). Selective dissolution is another waythat pitting corrosion can initiate and this occurs when one of thecomponents dissolves faster than other components (Szklarska-Smialowska, 1986).

Abood (2008) considered pitting to be autocatalytic in nature;once a pit starts to grow, the local conditions are altered such thatfurther pit growth is promoted. Fig. 5 shows that the anodic andcathodic electrochemical reactions that comprise corrosion sepa-rate spatially during pitting. The local pit environment becomesdepleted in cathodic reactant (e.g. oxygen) which shifts most of thecathodic reaction to the surface outside of the pit cavity. Refer toEquation (1.2):

O2 þ 2H2Oþ 4e�/4OH� (1.2)

In Fig. 5 themetal, M, is being pitted by an aerated NaCl solution.Rapid dissolution occurs in the pit while oxygen reduction takesplace on the adjacent metal surfaces.

The pitting phenomenon can be summarized as the local pit

Fig. 5. Autocatalytic process occurring in a corrosion pit (adapted from Abood (2008)).


environment depleted in the cathodic reactant, such as dissolvedoxygen, and enriched inmetal cation including anionic species. Thisacidic chloride environment is aggressive to most metals and tendsto prevent repassivation, thereby promoting continued propaga-tion of the pit. It was suggested that removal of oxidizing agents,e.g. removal of dissolved oxygen, is one powerful approach forreducing susceptibility to localized corrosion (Szklarska-Smialowska, 1986; Frankel, 1998; Abood, 2008).

A number of researchers (Frankel, 1998; Abood, 2008; Melchers,1994; Zaya, 1984; Galvele, 2005; Wong and Alkire, 1990; Melchers,2001a) have studied the local chemistries that form in pits using arange of techniques. Wong and Alkire (1990) described a way toisolate the pit solution by rapid freezing of the electrode in liquidnitrogen, removal of the surface excess, and subsequent thawing(Frankel, 1998; Wong and Alkire, 1990). This approach is mainlyused to study the pH in Aluminium pits and the chloride concen-tration in pits for stainless steel. Likewise, Frankel (1998) consid-ered a way to isolate the pit solution by using artificial pit electrodemethods. This is also known as one dimensional pit, or leadein-pencil electrode, which is a wire imbedded in an insulator such asepoxy (Frankel, 1998). This technique allows a considerable volumeof pit electrolyte to be analysed (Wong and Alkire, 1990). Anotherway to isolate the pit solution is by inserting microelectrodes intopits, cracks and cervices (Frankel, 1998). With this technique, oncethe local solution composition is fully characterized, it is possible toreassemble the local environment by reconstituting it in bulk formfrom reagent grade chemicals, and then determining the electro-chemical behaviour of a normal-sized sample extracted from a localenvironment.

Frankel (1998) also studied the electrochemistry of pittingcorrosion and have found that different characteristics potentialsexist within pitting corrosion (Frankel, 1998). It was found thatmetastable pit formation can occur below the Epit, however thiscannot be seen as initiation points of pitting (Szklarska-Smialowska, 1986; Frankel, 1998; Wika, 2012a). Metastable pitsare those pits that initiate and grow for a limited period beforerepassivating. Materials exhibiting higher value of pitting potential(Epit) and repassivating potential (Erep) are more resistant to pittingcorrosion therefore cyclic polarization experiments are generallyused to measure the pitting resistance of metals (Frankel, 1998).The value of Epit can be used to determine under what conditions

pitting corrosion occurs (Szklarska-Smialowska, 1986; Wika,2012a).

The environment (chemistry) and the material (metallurgy)factors determine whether an existing pit can be repassivated ornot (Roberge, 2013). Sufficient supply of oxygen to the reaction sitemay enhance the formation of oxide at the pitting site and thusrepassivate or heal the damage to the passive film. An existing pitcan also be repassivated if the material contains a sufficient amountof alloying elements such as Cr, Mo, Ti, W, N, etc. (Nu~nez, 2007).Amongst these elements, Mo can significantly enhance theenrichment of Cr in the oxide and thus repassivate the pit. Adetailed analysis of the influence of pit chemistry on pit growth andstability has been provided by Galvele (2005). The concentration ofvarious ionic species at the bottom of modelled one-dimensionalpit geometry is discussed by many researchers (Frankel, 1998;Abood, 2008; Wika, 2012a; Galvele, 2005). The concentration ofvarious ionic species is determined as a function of current densitybased on a material balance that considered generation of cationsby dissolution, outward diffusion, and thermodynamic equilibriumof various reactions such as cation hydrolysis (Frankel, 1998) (referto Equation (1)). Galvele (2005) found that a critical value of thefactor x$i, (where x is pit depth and i is current density), correspondto a critical pit acidification for sustained pit growth (Frankel, 1998).Current density in a pit is a measure of the corrosion ratewithin thepit and thus a measure of the pit penetration rate. The value “x$i”can be used to determine the current density required to initiatepitting at a defect of a given size (Abood, 2008). Increasing the pitdensity increases the ionic concentration in the pit solution, oftenreaching super saturation conditions. Frankel (1998) demonstratedan example of a solid salt film that may form on the pit surface.When the solid salt film is formed on the pit surface, the ionicconcentration drops to the saturation value and the pit growth rateis limited by mass transport out of the pit (Frankel, 1998; Abood,2008; Wong and Alkire, 1990).

Although many researchers claimed that salt films are requiredfor pit stability, Frankel (1998) disagreed. These authors suggestthat salt films will only enhance stability by providing a buffer ofionic species that can dissolve into the pit to reconcentrate theenvironment in the event of failures such as the sudden loss of aprotective pit cover (Frankel, 1998; Abood, 2008). The shape of thepit varies on the environmental conditions, and on whether the pitcontains a salt film or not. Abood (2008) and Frankel (1998) bothagreed that, under mass-transport-limited growth, pits can behemispherical with polished surfaces. In the absence of a salt film,pits can be crystallographically etched or irregularly shaped(Frankel, 1998; Abood, 2008).

4. Characteristics of pitting corrosion

In 1984, Zaya (1984) reported the following characteristics ofmetal that should be studied to characterise the pitting corrosion:

� Existence of a threshold value of the anodic potential belowwhich pitting does not occur in the given electrolyte system orthe potential of the metal must exceed a certain threshold (Zaya,1984).

� Some aggressive anion must be present. These ions correspondto strong acids and they tend to form complex and very solublesalts with the metal they attack (Dallek and Foley, 1976).

� Some other ions inhibit the effect of the aggressive ions aspitting agents. The relationship between the potential necessaryfor pitting and the concentration of aggressive and inhibitingions is given in Equation (1.3) (Szklarska-Smialowska, 1986;Zaya, 1984).


E ¼ E0 � a$log½Agg$� þ b$log½Inh$� (1.3)

� The delay (induction time), between the fulfilment of thenecessary conditions for pitting, and the detection of the firstpit, is due to the detection not taking place until the pit hasreached stage 3 (Fig. 2) (Zaya, 1984).

� The corrosive attack is highly localized on the surface butotherwise still passive. The precise location of a pit is unpre-dictable but the probability of occurrence is higher at grainboundaries, inclusions and other surface discontinuities (Zaya,1984; Ryan et al., 2002).

� Inside the pit the aggressive ions reach concentrations muchhigher than in the bulk of the solution. This is accompanied bychanges in the concentration of the other components of thesolutions e usually a drop in the pH (Zaya, 1984; Wong andAlkire, 1990).

These properties were selected after extensive observation ofpitting in combinations of several metal-solutions (Szklarska-Smialowska, 1986; Zaya, 1984).

4.1. Passive films

It is generally acknowledged that the susceptibility of metals,and the rate at which the corrosion processes take place, are closelyrelated to the quality of the passive film which normally occurs onthe metal surface (Szklarska-Smialowska, 1986). Many researchershave studied susceptibility of metals for many years. Amongstthese, Szklarska-Smialowska (1986) and Melchers (1994, 2007,2003a, 2004c) applied both theoretical and experimental knowl-edge to estimate the susceptibility of metals for pit to initiate inmarine and offshore structures. Szklarska-Smialowska (1986)described different ways to estimate the susceptibility of themetal alloy to pitting:

� Determination of characteristic pitting potential� Determinations of a critical temperature of pitting� Measurement of the number of pits per unit area, weight lossand, if possible, the size and depth of pits formed in a suitablestandard solution

� Determination of the lowest concentration of chloride ionscausing pitting (Szklarska-Smialowska, 1986).

Although the nature of the passive films is not fully understood,many researchers acknowledge that pitting susceptibility is relatedto local imperfections or discontinuities (Melchers, 1994; Wika,2012a; Moayed et al., 2003). Szklarska-Smialowska (1986) listedthe following understanding on the formation of weak spots inpassive film (Szklarska-Smialowska, 1986; Melchers, 1994):

� Boundaries between the metal matrix and non-metallic in-clusions where differences exist in the coefficient of thermalexpansion, causing either cracking or localised compressionzones.

� Boundaries between the metal matrix and second phase pre-cipitates, as these often have the ability to draw from the alloysome components responsible for the passive state (e.g. Crimpoverishment of CreNi stainless steels as a result of chro-mium carbide formation).

� Inclusions of greater chemical reactivity than that of the alloy ormetal itself.

In addition to non-metallic inclusions, the following featurescan create weak spots in the passive films:

� Grain boundaries at which some impurities segregate (i.e. notpurely physical defects)

� Flaws� Various kinds of mechanical damage� Three phases of metal (solid, liquid and gas)� Exit point of dislocations.

Melchers (1994) and Szklarska-Smialowska (1986) proposevarious ways in which the susceptibility of an existing structuralmaterials to pitting can be assessed under experimental conditions(Melchers, 1994). Among these, measuring susceptibility to pittingby using pitting potential and/or by temperature is commonly usedways and is discussed in the following sections.

4.1.1. Measuring susceptibility of pitting using pitting potentialElectrochemical studies of pitting corrosion reported that

pitting potential is related to the electrochemical process associ-ated to corrosion (Frankel, 1998; Melchers, 1994). When the elec-tronic potential is measured for a given metal electrolyte system, itwas observed that there exists a threshold value of the anodicpotential below which pitting does not occur (Melchers, 1994).Szklarska-Smialowska (1986) discussed various electrochemicalmethods used to determine characteristics pitting potentials and adetailed description is available (Szklarska-Smialowska, 1986).Different methodology used to determine characteristics pittingpotentials are:

� Measurements of the anodic polarization curve using a poten-tiostatic device (Greene, 1962).

� Measurements of the anodic polarization curve by galvanostaticmethods (Frangini and De Cristofaro, 2003).

� Measurements of current density vs. time at constant potential(Comotti et al., 2013).

� Measurements of potential vs. time at constant current (Maliket al., 1992).

� Repassivation time methods (Scratch Methods) (Szklarska-Smialowska, 1986).

� The pit propagation rate test (Wilde, 1972).� The critical pitting temperature test (B€ohni, 2000).� The scanning reference electrode technique (Power andShirokoff, 2012).

Malik et al. (1992) used these methods to conduct electro-chemical tests to investigate the pitting corrosion behaviour of AISI316L in Arabian Gulf Seawater and reported that the pitting po-tentials vary for the same materials under identical conditionsdepending upon the methods used (Malik et al., 1992).

4.1.2. Measuring susceptibility of pitting using temperatureAccording to DNV-RP-G101 (Veritas, 2002), temperature is the

main reason for pitting corrosion attack on offshore steel structures(Veritas, 2002). The reported experimental work on temperatureoften focused on finding the critical pitting temperature (CPT)which is defined as the lowest possible temperature where pittingoccurs (Wika, 2012a). Moayed et al. (2003) defines CPT as “thelowest temperature at which the growth of stable pits is possible”(Moayed et al., 2003). ASTM G 150 is a standard test method forelectrochemical CPT testing of stainless steel.

Several authors (Frankel, 1998; Melchers, 1994; Caines et al.,2013; Wika, 2012a) reported that, with an increase in tempera-ture, the Epit decreases and the damage caused by corrosion in-creases. Melchers (1994) reported that, for Carbon Steel, it appears


that the passive film becomes less protective with higher temper-ature and that this is related to the behaviour of chloride ions(Melchers, 1994). Wika (2012a) found that the increasing temper-ature causes higher current transients and promotes the conversionof metastable pits into stable growing pits (Wika, 2012a). B€ohni(2000) carried out electrochemical studies in micro and largescale e both in chloride free and in chloride solutions. In chloridefree solutions, the increase in temperature caused an increase indissolution of MnS inclusions, while in a chloride environment thegrowth of pits increases (B€ohni, 2000).

5. Factors affecting pitting corrosion in marine and offshoresteel structures

Steel alloys are used in numerous and diverse applications inmarine and offshore industries (Ryan et al., 2002). Stainless steelgrade 316 alloys are commonly used in offshore applicationsbecause of their corrosion resistant nature. It is regarded as safe fordesign life when choosing a material in an aggressive environment(Stewart and Al-Harthy, 2008; Ryan et al., 2002; Carpen et al., 2007;Isaacs and Kissel, 1972). Nevertheless, even though these alloysoffer a better resistance to general corrosion, they are still suscep-tible to pitting corrosion (Wika, 2012b). The most common causesof failure of stainless steel in marine environments is pittingcorrosion because the material can quickly be penetrated despitethat its general corrosion rate is very low (Malik et al., 1992).

Several findings on pitting corrosion of steel alloys, accom-plished by using numerical and experimental techniques, havebeen published by different investigators. Salh (1990) studied thepitting corrosion of carbon steel in sodium molybdite solution.They found that the sodiummolybdite is a good pitting inhibitor insolutions of moderate and low concentration of chloride, and thatthe corrosion potentials shifted to more positive value in thepresence of molybdite (Abood, 2008; Salh, 1990). Malik et al. (1992)investigated the pitting behaviour of 316L stainless steel in ArabianGulf seawater. The immersion test and electrochemical techniqueswere utilized to study different factors affecting the pitting corro-sion (Malik et al., 1992). Similarly, Melchers, (2008a) investigatedthe corrosion wastage in aged offshore structural steels. Theysummarized the progress in the development of the mathematicalmodels for corrosion loss and maximum pit depth under in-situconditions of steel as a function of time (Melchers, 2014a).Several published materials (Popoola et al., 2013; Darmawan, 2010;Darmawan and Stewart, 2007; Davydov, 2008; Guedes Soares et al.,2011; Nakai et al., 2006) are widely available.

The phenomenological factors which influence pitting corrosionin marine and offshore environments are generally similar to thoseof uniform corrosion. The influence of these factors differdepending on the types of marine environments, such as atmo-spheric, splash zone, tidal zone and shallow water immersion(Melchers, 1994; Melchers and Ahammed, 1994). The differenttypes of exposure can be sub-classified as illustrated in Fig. 6. Asummary of all the possible factors involved in pitting corrosion isgiven in Fig. 7. These factors are categorised into four differenttypes: (1) Physical factors (2) Chemical factors (3) Biological factorsand (4) Metallurgical factors.

A summary of the different types of exposure in marine envi-ronments is shown in Table 1. Factors which can affect pittingcorrosion given in Fig. 7 are composed from all types of exposurezones. Attention is not limited to just stainless steel; mild steel andlow alloy steels are also considered because literature indicates thatthe corrosion behaviour is similar to all (Melchers, 2004a, b, 1994;Melchers and Ahammed, 1994; Melchers, 2005b).

Environmental factors which influence marine immersioncorrosion are shown in Table 1 (Melchers, 2008a, 1994; Melchers

and Ahammed, 1994).Table 2 (Melchers, 1998). Not all of these factors are well un-

derstood although some have become clearer through recent in-vestigations (Melchers, 1998). The main factors governing pittingcorrosion in a marine and offshore environment are described insubsequent sections.

5.1. Physical factors

5.1.1. TemperatureTemperature is one of the critical factors in pitting corrosion,

because it greatly influences the corrosion behaviour of steels inseawater (Melchers, 2004a; Szklarska-Smialowska, 1986; Frankel,1998; Abood, 2008; Zamaletdinov, 2007; Strehblow and Marcus,1995; Wong and Alkire, 1990; Malik et al., 1999; Ghali et al., 2004).Many materials do not pit at a temperature below a certain valuewhich may be extremely high and reoccurring (Nu~nez, 2007).Pitting potential of stainless steel alloy is measured in the tem-perature range of 25e90 �C (Malik et al., 1999). The majority ofchemical and electrochemical reactions proceed more rapidly athigher temperature. Therefore it might be anticipated that the rateof pitting would increase with increasing (rising?) temperature(Szklarska-Smialowska, 1986). Almarshad et al. (Nu~nez, 2007)experimentally studied the effect of temperature on pitting byeither varying the temperature at a range of fixed applied poten-tials, or by varying the potential for a range of constant temperatureexperiments. They provided a plot of pitting and repassivationpotentials for three different stainless steels in 1 M NaCl as afunction of solution temperature (Abood, 2008; Acu~na-Gonz�alezet al., 2012). From these experiments it was observed thatextremely high breakdown potentials occur at low temperaturecorresponding to transpassive dissolution e but not with localizedcorrosion. However, just above the critical pitting temperature(CPT), pitting corrosion occurs at a potential that is far below thetranspassive breakdown potentials (Nu~nez, 2007).

Hadfield et al. (Hadfield, 1922) reported metal losses for ordi-nary steels over a 5 year exposure. Fig. 8 shows the pitting depth fortidal conditions as a function of annual mean temperature fordifferent experimental cities. It can be seen that pitting is incon-sistent as a function of temperature for tidal conditions, whereasthere is an increasing trend of pitting with temperature for generalcorrosion conditions (Melchers, 2008a, 1994; Melchers andAhammed, 1994).

The specimens were despatched to testing stations at Auckland,Colombo, Halifax (Canada), and Plymouth. Observations suggestthat, despite the differing sea and atmospheric conditions, im-mersion corrosion of carbon steel is very similar when comparingall five locations. The effect of temperature on atmospheric corro-sion was also thoroughly examined and showed that atmosphericcorrosion was generally consistent with temperature. Melchers(1994), 2002, Melchers and Ahammed (1994) stated that the re-action process for corrosion becomes faster with higher tempera-tures after the initial phase, suggesting that the corrosion rateincreases with increase (rise?) in temperature. However, in im-mersion conditions, oxygen concentration decreases with increasein temperature while biological activity increases with increase intemperature (Melchers, 1994; Melchers and Ahammed, 1994;Melchers, 2002).

5.1.2. pHThe pH of seawater may vary slightly depending on the photo-

synthetic activity. Plant matter consumes carbon dioxide and af-fects the pH during daylight hours (Schumacher, 1979; Younis et al.,2012). The carbon dioxide content in seawater (close to the surface)is influenced by the exchange with carbon dioxide in the

Fig. 6. Classification of environments encountered by pitting corrosion (Sørensen et al., 2009).


atmosphere (Schumacher, 1979). It is reported that the usual dailyfluctuations of pH (of between 8.0 and 8.2) has little direct effect oncorrosion rate. However it can be a factor in calcium carbonatedeposits which does affect the corrosivity (Schumacher, 1979;Melchers, 1994). Comparable conclusions were made by Melchers(1994). Chandler (1985) did not consider pH of seawater to beone of the main physical factors that influences pitting corrosionrate on steel (Chandler, 1985). Revie (2008) stated that, within therange of pH of about 4e10, the corrosion rate is independent of pHand depends only on how rapidly oxygen diffuses to the metalsurface (Guedes Soares et al., 2011; Revie, 2008).

Many experiments were carried out to understand the effect ofpH on pitting corrosion in marine environments. Malik et al. (1999)performed electrochemical tests on metal and found that corrosionrate increases with increasing acidity of the solution. They statedthat corrosion rate increases with rising pH between 4 and 9 (Maliket al., 1999). Similarly, Carpen et al. (2007) carried out potentio-dynamic experiments with distilled solutions containing chlorides.They found that a pH of 3 contributes a little lower pitting poten-tials than solutions of pH 5 with low chloride content. No effect ofpH was observed at higher chloride content. It was therefore sug-gested that pH has little effect on pitting potentials in chloridesolutions and does not much change the susceptibility to pittingcorrosion in the pH range 3e7 (Wika, 2012a; Carpen et al., 2007).The effect of pH on the breakdown potential was not understoodwell (Abood, 2008). Pistorius et al. (Pistorius and Burstein, 1992b)found that the breakdown potential (Eb) value is almost constantwithin a large range of pH values (Abood, 2008; Salh, 1990;Pistorius and Burstein, 1992b).

5.1.3. SalinitySeawater Handbook (Schumacher, 1979) defines salinity as the

total weight in grams of solid matter dissolved in 1000 g of water(Schumacher, 1979). The effect of seawater salinity is convention-ally considered to be a very important factor with regard to pittingcorrosion (Zakowski et al., 2014). The composition of seawater forsalinity is given in Table 3. It can be appreciated that the salinity ofseawater is composed of about 90% sodium chloride (NaCl). Thedissolved salt leads to a low resistivity so that the seawater acts as agood electrolyte, thereby enabling pitting corrosion (Melchers andAhammed, 1994).

Variations on salinity in open surface water appears to rangefrom 32,000 to 37,500 ppm (3.2e3.75%) with deep water meanaround 35,000 ppm (3.5%) (Melchers and Ahammed, 1994). A greatvariation in salinity is observed in some of the more isolated seas(see Table 3) and, because the salinity variations are accomplishedby other changes, the total effect on the pitting corrosion has to beestablished in each case (Schumacher, 1979).

Fig. 9 shows how the combination of chloride concentration and

dissolved oxygen concentration results in the maximum pittingcorrosion rate. The highest oxygen concentration can be achieved at3.5 weight precents sodium chloride (Yari, 2014). Melchers andAhammed (1994) reported that, in immersed conditions, thecorrosion rate is expected to increase due to higher dissolved ox-ygen concentration with reducing salinity (Fig. 9) (Melchers andAhammed, 1994). However, some researchers reported that, inthe longer term, the pitting corrosion rate effect of salinity may beless because the increased oxygen has a greater tendency to formprotective deposits (scale) or protective biofouling containing cal-cium carbonate (Melchers and Ahammed, 1994).

Some researchers (Guedes Soares et al., 2011; Melchers, 1998;Chandler, 1985; Mercer and Lumbard, 1995) reported that, forseawater conditions, salinity is of little practical importance tomarine corrosion. This is because corrosion of metals is notappreciably affected due to the salinity variations in open-oceansurface water range (ranging) from 32,000 to 37,500 ppm perthousand gram of water (Guedes Soares et al., 2011; Melchers,1998; Chandler, 1985; Mercer and Lumbard, 1995). According tothe experimental work ofMercer et al. (Mercer and Lumbard,1995),the effect of change in salinity appears to be marginal for stainlesssteel in calm conditions. Chandler et al. (Chandler, 1985) also didnot consider seawater salinity as one of the main environmentalfactors that influences corrosion rate of steel (Chandler, 1985).

5.1.4. High velocity of waterMany metals are sensitive to velocity effects in seawater

(Schumacher, 1979). For metals like iron or copper, there is a criticalvelocity beyond which corrosion becomes high. However, for steelstructures inmarine environments, the little effect of water velocitycan be ignored (Melchers, 2005c). Melchers (1998) reported that,with the possible exception of the effect on marine biologicalgrowth and the influence of the continuous supply of oxygen, theeffect of low and moderate water velocity on the rate of corrosioncan be ignored (Melchers, 1998, 2005c). HoweverMelchers (2004a)found that water velocity does increase the rate of pitting corrosionnonlinearly. They indicated that when corrosion products and/ormarine growth is disrupted or removed (as through erosion orabrasive action) the effect of water velocity on pitting corrosion canbe more severe (Melchers, 2003a; Guedes Soares et al., 2011).

The effect of higher water velocity on marine pitting corrosionshould be considered. Soares et al. (Guedes Soares et al., 2011) re-ported that the corrosion of steel by seawater increases as thewatervelocity increases. The effect of water velocity at higher levels isshown in Fig. 10 and illustrates that the rate of corrosion attack is adirect function of the velocity until critical velocity is reached(Abood, 2008; Guedes Soares et al., 2011). Special forms of corro-sion associated with higher seawater velocity are:

Fig. 7. Factors affecting pitting corrosion in marine and offshore environments.


Table 1Classification of marine environments (Melchers, 2008a, 1994; Melchers and Ahammed, 1994).

Marinezone

Description of environments Characteristics of steel corrosion behaviour

Interiorspaces

High humidity, higher than ambient temperature, periodic wetting possible. Particularly aggressive in areas where moisture can accumulate.

Atmosphere Minute salt particles present, corrosivity varies with height above water, winddirection and velocity, dew cycle, rainfall, temperature, pollution etc.

Sheltered surfacemay deteriorate more rapidly than those boldly exposed.Corrosion decreases rapidly with distance from sea.

Splash Wet, well aerated surface, no fouling. Most aggressive zone for steel. Protective coating is difficult to maintain.Tidal Well oxygenated, marine fouling likely to occur, oil coating from polluted

harbour water may be present.High attack for steel, however oil coating may reduce corrosion attack.Some protection on continuous steel pile.

Immersion Usually saturated with oxygen. Pollution, sediment, fouling, velocity etc. have keyroles in the degree of corrosion.

Biofouling restricts oxygen supply rate to surface, reducing corrosion.Protective coatings may be effective for limited corrosion control.

Table 2Environmental factors as shown in the flowchart Fig. 7 and its effect on the corrosion rate (Schumacher, 1979; Melchers, 1994; Melchers and Ahammed, 1994; Melchers, 1998).

Type Factor Effect on initial state corrosion Effect on steady state corrosion Influenced by

Biological Bacteria Little effect Controls rate Seawater temperatureBiomass NaCl concentrationPollutants Varies Varies Little known aboutMarine growth Little effect Controls rate Pollutant type/level

Chemical O2 Directly proportional NoneCO2 Little effect Little effectNaCl Inversely proportional Uncertain Little effect at seaCa Little effect Varies Little known aboutCarbonate solubility Little effect Little effect

Physical Temperature Proportional >10 �C Proportional Geographical locationspH Little effect Uncertain Little effect at seaPressure Uncertain Not none Oxygen effectWater velocity Little effect Uncertain Geographical locationsSuspended solids Uncertain Uncertain Geographical locationsSurface wetting and wave action Proportional for tidal/splash zone Proportional Location, weather patternsOil in the water Reduces for tidal zone Reduces for tidal zone Industrial development/shipping

Fig. 8. Pitting depth for tidal conditions as a function of annual mean water temper-ature (Melchers, 1994).

Table 3Composition of seawater and ionic constituents and total solids in Ocean waters (Schum

Constituent g/kg in 1000 g of water Cations, precent

Chloride 19.353 Naþþ 1.056Sodium 10.76 Mgþþ 0.127Sulphate 2.712 Caþþ 0.040Magnesium 1.294 Kþ 0.038Calcium 0.413 Srþþ 0.001Potassium 0.387 Total 1.262Bicarbonate 0.142Bromide 0.067Strontium 0.08Boron 0.004Fluoride 0.001


� Erosion-corrosion caused by high-velocity silt bearing seawater� Impingement attack e where air bubbles are present� Cavitation ewhere collapsing vapour bubbles cause mechanicaldamage and often corrosion damage as well (Schumacher,1979).

5.1.5. Physical sizePitting corrosion in seawater is influenced by the size of the

sample. The pitting susceptibility depends onmarine environmentssuch as tidal and/or splash zones; however, the size of the specimenmay alter the rate of pitting corrosion. The effect of physical size onmarine pitting is not yet fully understood.Melchers et al. (Melchers,1994) recommended that consideration be given to the physicalsize of specimen in order to model metal loss and pitting for real-istic structures (Melchers, 1994; Melchers and Ahammed, 1994).They also advised that a comparative study be performed with the

acher, 1979).

Anion, precent Body of water Salinity ppm

Cl� 1.898 Baltic Sea 8000So4� 0.265 Black Sea 22,000HCO3

� 0.014 Atlantic Sea 37,000Br� 0.0065 Mediterranean Sea 41,000F� 0.0001 Caspian Sea 13,000

2.184 Irish Sea 32,5000

Fig. 9. Combination of sodium chloride concentration versus the rate of corrosion inseawater (Yari, 2014).

Fig. 10. Effects of velocity of seawater on the corrosion rate of steel (Guedes Soareset al., 2011).


available observations on steel pylons in a marine environment andthe results from isolated test specimens (Melchers, 1994).

5.1.6. Water depth (hydrostatic pressure)Thewater depth (hydrostatic pressure) is also considered to be a

factor that increases the rate of pitting corrosion in marine envi-ronments. The effect of depth of exposure in seawater on theaverage corrosion rate of steels has been studied by some re-searchers (Carpen et al., 2007; Melchers, 2005b; A. International,2004; Qin and Cui, 2003; Salau et al., 2011; Zatkalíkov�a et al.,2010). Melchers, (2005b) studied the corrosion rate for mild steelcoupons exposed at variable immersion depths and at variousgeographical locations. Analysis of the data showed that there is notan obvious effect of water depth on short-term corrosion behav-iour. Nevertheless, oxygen concentration and water temperatureare important parameters influencing weight loss as a function oftime in different immersion depths. The effects of immersion depthfor long-term corrosion behaviour may not be limited to just theseparameters however, as it will also be governed by anaerobicconditions (Melchers and Ahammed, 1994; Melchers, 2005b, d).

Fozan et al. (Al-Fozan and Malik, 2008) conducted a laboratorytest to evaluate the effect of seawater level on the corrosionbehaviour of different alloys (Al-Fozan andMalik, 2008). Specimenswere fixed at three locations, namely: above seawater surface,semi-submerged in seawater and fully submerged in seawater. Theexperimental results show that, in a splash zone, the stainless steelusually has satisfactory performance but that the carbon and lowalloy steels do not. The stainless steels are susceptible to pittingcorrosion in the submerged zone because of factors such as oxygenconcentration, biological activities, pollution, temperature, salinityand velocity. Similarly the pitting corrosion rate for 316L stainless

steel is higher above the water level than when compared to sub-merged conditions. This is due to the higher oxygen concentrationabove the seawater. However it is just the opposite for 304 stainlesssteel; the pitting density and pit depth resulted in an increased rateof corrosion for this alloy when in a submerged location. Converselyfrom these experiments, it was found that the pitting corrosion rateof metals at semi-submerged conditions is higher than for sub-merged and atmospheric conditions (Al-Fozan and Malik, 2008;Zamanzadeh et al., 2004).

5.1.7. Atmospheric effectsIn atmospheric conditions, the intensity of the corrosive attack

is influenced greatly by the amount of salt particles or mist whichcollects on the metal surface (Little et al., 2008). Salt depositionvaries with wind and wave conditions, the height above sea level,and exposure etc. (Schumacher, 1979). Schumacher (1979) reportedthat several other factors which affect the atmospheric corrosionbehaviour in marine environments can include solar radiation, theamount of rainfall and fungi.

Melchers (1994), Melchers and Ahammed (1994) reported thatmacro biological activity is basically absent in atmospheric corro-sion, but that it may have some influence. Additionally, the airtemperature and exposure will have a direct influence on rate ofcorrosion (Melchers, 1994; Melchers and Ahammed, 1994). Newtonet al. (Friend, 1940) conducted an experiment to investigate thevariation of atmospheric corrosion as a function of length ofexposure in various locations. They reported difficulties in drawingconclusions because of the variation in several factors like expo-sure, humidity and distance from the sea. Additionally, there weresignificant differences in air temperature and annual rainfall(Melchers, 1994; Melchers and Ahammed, 1994; Friend, 1940).

5.1.8. Water current & tidal conditionsWater current may have an effect on marine biological growth

through the provision of an adequate rate of food supply. It also hasan influence on the continuous supply of oxygen (Schumacher,1979; Melchers and Ahammed, 1994). Lewis et al. (Lewis andMercer, 1984) reported that the effect of water current is difficultto quantify as it doesn't have a direct effect on pitting corrosion;however it can be a subsidiary factor for pit initiation. For example,it may increase the marine biological growth by dissolving theimpurities in the ocean which could then lead to pit growth(Melchers and Ahammed, 1994; Lewis and Mercer, 1984). Melcherset al. (Melchers, 1994) suggested that the effect of low to moderatewater currents on the rate of pitting corrosion can be ignored(Melchers, 1994).

The pitting susceptibility is higher in the tidal zone compared topitting susceptibility in a fully immersed zone (Melchers andAhammed, 1994; Southwell and Alexander, 1970). Melchers andAhammed (1994) gathered experimental observations and data,performed in cities such as Halifax, Colombo, Panama, Auckland,and Plymouth, which investigated the extent of corrosion loss as afunction of length of immersion exposure (Melchers andAhammed, 1994; Southwell and Alexander, 1970). The plot forpitting depth in tidal conditions as a function of annual meantemperature for different experimental cities is shown in Fig. 8.

5.1.9. Surface wettingThe degree and continuity of surface wetting is particularly

important for atmospheric, splash and tidal zone corrosion. Thesurface wetting depends on the location of the structural specimenrelative to the seawater (Melchers and Ahammed, 1994; LaQue,1959).

In the tidal and splash zone, wetting may be controlled by thelocal tidal range and climatic conditions such as water, air and


temperature. The effect of surface wetting is still not fully under-stood but the pitting due to surface wetting is believed to beminimal. Melchers et al. stated that no general rule exists todescribe functional relationships between surface wetting andclimatic conditions (Melchers, 1994; Melchers and Ahammed,1994; LaQue, 1959).

5.1.10. Initiation timeInitiation time is often called induction time and pertains to the

elapsed time before pitting corrosion begins (Vel�azquez et al.,2014). Schumacher (1979) and Malik (1999) described an experi-ment conducted in different stainless steels at 50 �C which wereimmersed in chloride solutions under open circuit potential mea-surements (Schumacher, 1979; Wika, 2012a; Malik et al., 1999). Itwas found that pitting potential could be presented as a linearfunction of initiation time, ti, where A and B are constants thatdepend on temperature. Refer Equation (1.4)):

Epit ¼ Aþ B$logðtiÞ (1.4)

Malik et al. (1992a) conducted electrochemical measurementsof the open circuit potential and found that the logarithm of theinduction time decreases linearly with increasing chloride con-tained and at increasing temperatures. These test temperatureswere conducted at 30, 50 and 80 �C (Malik et al., 1992a).

5.2. Chemical factors

5.2.1. High chloride ion concentrationThe presence of certain aggressive anions, such as chloride

concentration (Cl�), can migrate to the active corroding area andstabilize pitting corrosion (Pardo et al., 2000). The influence of Cl�

on the susceptibility to pitting corrosion has been studied innumerous metals and alloys and particularly in stainless steel.Pardo et al. (2000) stated that the presence of alloyed elements,such as Cr, Mo, and N, improve the resistance to localized corrosionof the stainless steel. However, the presence of chloride ions caneasily penetrate the film due to its high diffusivity (Wika, 2012a;Pardo et al., 2000). Frankel (1998) explained that pitting corro-sion will only occur in the presence of aggressive anionic species,which are usually chloride ions, and that the severity of pittingcorrosion tends to vary with the logarithm of the bulk chlorideconcentration (Frankel, 1998).

Fig. 4 shows the curves plotted for two different chloride con-centrations. It can be seen that Epit increases with decreasingchloride content. Many researchers (Frankel, 1998; Malik et al.,1992a, b; Newman, 2001) have published studies describing therelationship between Epit and the chloride content. Different po-larization measurements (Frankel, 1998; Malik et al., 1992a, b;Newman, 2001) demonstrated that the pitting potential was alinear function of the logarithm of the chloride concentration. Highalloyed stainless steels are less affected by increasing chloridecontent, and therefore Epit changes only slightly for thesealloys (Wika, 2012a).

5.2.2. Dissolved oxygenDissolved oxygen content is a major factor affecting the corro-

sivity of seawater. For many common metals, a higher oxygencontent is accomplished commensurate to the increase in the rateof pitting attack (Schumacher, 1979). The corrosion rate of localanodes is dependent on the cathode reaction and, therefore, de-polarization is more rapid with the increase of oxygen at thecathode. The depolarization is a function of the amount of dissolvedoxygen in the seawater and the velocity of flow at the surface(Abood, 2008; Zamaletdinov, 2007).

Various authors commented that there is a good correlationbetween dissolved oxygen and corrosion mass loss (Frankel, 1998;Malik et al., 1992; Guedes Soares et al., 2011; Melchers, 2006a).Melchers (2003a) stated that the corrosion rate usually has a linearrelationship with the concentration of dissolved oxygen (Melchers,2003a; Guedes Soares et al., 2011). They further stated that thedissolved oxygen levels in open ocean conditions reduce withdepth around 5e7 ml/l at surface, mainly as a result of increasedpressure (Melchers, 1994; Melchers and Ahammed, 1994). Hence,the oxygen concentration is critical in the immersion corrosionprocess during the early stage of seawater corrosion (Melchers,1994; Melchers and Ahammed, 1994). Schumacher (1979)mentioned that the oxygen level in seawater may range up to12 ppm. The factors that increase seawater oxygen could bephotosynthesis of green plants and wave action etc. Conversely, thebiological oxygen demand of decomposing dead organisms willreduce seawater oxygen level (Schumacher, 1979; Melchers, 1994).Regional variation exists in surface water temperature; hence,salinity and locally mixed conditions occur due to the regionalvariations of oxygen level (Melchers and Ahammed, 1994).

In waters containing high salt concentration, corrosion is pro-portional to the amount of oxygen dissolved in the water. If the saltconcentration in the water increases, then the solubility of oxygendecreases and, consequently, the corrosion rate is reduced(Zatkalíkov�a et al., 2010). Therefore, oxygen dissolved in water isprobably the most troublesome corrosion-producing substance. Anexception to this statement is metals which depend on a passivefilm for corrosion protection. Nevertheless, stainless steel oftencorrodes rapidly where the oxygen supply to the metal is restricted(Schumacher, 1979; Melchers and Ahammed, 1994; Ryan et al.,2002).

5.2.3. Carbon dioxideThe dissolved carbon dioxide in seawater produces carbonic

acid and, after ionisation, it forms bicarbonate and carbonate ions(Melchers and Ahammed, 1994). Few effects of carbon dioxide areknown for pit initiation in marine steel structures. However,Melchers et al. (Melchers and Ahammed, 1994) did report that thepresence of undissolved boric acid in oppositely charged ions suchas carbon dioxide may provide a constant and fairly high pH level(Schumacher, 1979; Melchers and Ahammed, 1994).

5.2.4. Effect of halogen ionsPitting corrosion can be caused by different halide anions

(Abood, 2008). The anodic process associated with metal passiv-ation is strongly affected by the presence of halide ions in theelectrolyte. With large concentration of halide ions the passive filmon a metal is susceptible to pitting and also suffers local damage;however, low concentration produces only an increase of anodiccurrent in the passivity range (Szklarska-Smialowska, 1986).Szklarska-Smialowska (1986) described chloride as the mostaggressive anions to produce pitting in several metals (Szklarska-Smialowska, 1986; Abood, 2008). The three main reasons for thespecific effect of chloride ions is explained by Abood et al.(Szklarska-Smialowska, 1986; Abood, 2008) as:

� Its ability to increase the activity of hydrogen ions in the pitelectrolyte

� Ability to form complexes with cations and hydroxide� Its ability to form a salt layer at low pH at the bottom of the pit.

5.3. Biological factors

5.3.1. BacterialThe corrosion of metals by sulphateereducing bacteria (SRB) is


well recognized as one of the factors that increase rate of pittingcorrosion in marine environments (Thomas et al., 1988). Recentstudies show that the hydrogen sulphide produced by these or-ganisms can have a serious effect on mechanical failure processesby increasing a metal's susceptibility to corrosion e especiallypitting corrosion (Thomas et al., 1988). Thomas et al. (Thomas et al.,1988) stated that the effect of bacteria is particularly important foroffshore structures; here it is envisaged that the combination ofbacteria activity undermarine fouling conditions, wave loading andan aggressive environment can result in the premature failure ofmetal components by pitting corrosion fatigue (Thomas et al.,1988).

Melchers and Ahammed (1994) explained that bacteria is mostcommonly associated with marine corrosion of steel. Desulfovibriois one of the most common; it is a genus of gram negative sulphatereducing bacteria. When this bacteria attached to the surface of ametal with other bacteria, they may not grow when there is apresence of oxygen (Melchers, 1994; Melchers and Ahammed,1994). However Melchers (1994), Melchers and Ahammed (1994)reported that, where anaerobic conditions are more conducive,they can grow well in the temperature range of 25 �Ce 44 �C andwith pH ranging from 5.5 to 9.0 (Melchers and Ahammed, 1994).

5.3.2. FoulingMarine fouling in offshore structures can have serious conse-

quences on their integrity, enhancing both the corrosion and stresscomponents of corrosion fatigue (Thomas et al., 1988). Fouling iscommonly considered to be growth beyond the bacterial stage.Algal growth may include seaweeds, coral, aurelias, barnacles,mussels etc. Melchers and Ahammed (1994) reported that commonfouling can tolerate a range of temperature and light intensities,with growth most productive in the range of 15 �Ce 20 �C(Melchers, 1998).

5.3.3. Marine growthsMarine growth can affect the rate of pitting corrosion in im-

mersion conditions (Lewis andMercer, 1984). Although the amountof marine growth varies considerably from location to location, itdoes not have an initial effect for short term immersion corrosion. Italso appears that an initial effect is not reflected in fouling growth(Melchers, 1994; Lewis and Mercer, 1984). Melchers et al.(Zamaletdinov, 2007; Melchers and Ahammed, 1994; Lewis andMercer, 1984) reported that, despite local and regional geograph-ical differences, the overall fouling pattern is somewhat similar.Moreover, the presence of local warm currents with adequate foodsupply appears to be the major factors for marine growth. It wasalso assumed that marine growth may be affected adversely byhigh currents and suspended solids (Melchers, 1994; Melchers andAhammed, 1994; Lewis and Mercer, 1984).

5.3.4. PollutantsWater pollution, particularly in harbours, may result in the in-

crease of corrosion rate. Under these conditions the water may bemore aggressive because of the greater concentration of ammoniaor sulphide which is also combined with lower oxygen levels(Schiffrin and De Sanchez,1985). Due to the pollutants, water is lessable to support marine growth of protective bacteria and biofouling(Melchers, 1994; Guedes Soares et al., 2011). Harbours and coastalregions may be susceptible to nutrient pollution from sewage oragriculture run-off. Offshore oilfields are also known to providesources of nutrient pollution (Melchers, 1994; Guedes Soares et al.,2011). Schiffrin et al. (Melchers, 1994; Schiffrin and De Sanchez,1985) reported the effect of nutrient-based pollution on thecorrosion of mild steel coupons. It was found that pollutants likeoils and grease lower the corrosion rate due to their inhibiting the

rate of oxygen transfer for the oxidation process. The effects ofpollutants for marine pitting corrosion is not fully understood;however, for coastal and harbour areas, the effect of pollutants forpitting corrosion is expected to be higher than that for uniformcorrosion (Melchers, 2008a, 1994; Melchers and Ahammed, 1994;Schiffrin and De Sanchez, 1985).

5.4. Metallurgical factors

5.4.1. Alloy compositionThe effect of alloy composition acts on the tendency for an alloy

to initiate pit and also affects rate of corrosion (Szklarska-Smialowska, 1986). This influence appears to be related to thequality of the magnetite layers found on the surface of steel andappears to be a function of the alloying elements (Melchers andAhammed, 1994). Schultze and Wekken (1976) provided a sum-mary of the effect of steel composition on corrosion loss. It wasreported that a particular alloy may have a favourable effect incorrosion resistance under some exposure conditions but have anunfavourable effect on others (Melchers and Ahammed, 1994;Schultze and Wekken, 1976).

Table 4 summaries the main effects of alloying elements. Theinfluence of steel composition on pitting corrosion susceptibilityand uniform corrosion susceptibility is similar in marine corrosion.However, the pitting resistance of low alloy steel may be less thanthat of carbon steel (Melchers, 1994).

Ting et al. (2011) reported that the small changes (say <0.5%) inthe alloying elements used in steel composition should have zero ornegligible effect on the degree of corrosion that occurs while oxy-gen diffusion controls the corrosion process (Ting et al., 2011). Morespecialized steel with larger alloy compositions will have a lowerinitial rate of corrosion. This is particularly so for alloying elementssuch as chromium, molybdenum and aluminium and, to a lesserextent, for nickel, silicon, titanium and vanadium (Melchers, 2004c;Ting et al., 2011). Szklarska-Smialowska (1986) studied the effect ofsmall increases in molybdenum in steel composition and it wasfound that only a small increase can greatly reduce pitting corro-sion susceptibility. In addition to this, a small increase in elementssuch as nitrogen and tungsten also have a strong influence on thepitting resistance of stainless steels (Szklarska-Smialowska, 1986;Abood, 2008).

5.4.2. Steel typesSteels differ in their relative corrosivity under different condi-

tions. ASTM (A. International, 2004) and Melchers and Ahammed(1994) categorised steels as: 1] ordinary steels, such as mild steel,2] low alloy steels, or high strength steels and 3] stainless orchromium steels (Melchers and Ahammed, 1994; A. International,2004).

Stainless steels are obtained by the addition of chromium(around 3%) to steel. This alloy is usually used for applications inmarine atmospheric conditions for maintaining the passivity pro-tective surface film. Melchers and Ahammed (1994) stated that,when local perforation of the stainless steel passive film occurs,active passive cells are formed and heavy local corrosion such aspitting occurs. For immersed conditions, the addition of chromiumto stainless steel is beneficial for short term exposure (a few years);however, it is not agreed if this is also the case for long termexposure. The resistance to corrosion of low alloy steels may rangefrom two to ten times that of ordinary carbon steel. This is because,due to the formation of basic salts on the surface, low alloy steelsform a tighter rust coating at a faster rate than that of coronarycarbon steels (Melchers and Ahammed, 1994; Melchers, 2002,2003b).

Table 4Effects of alloying elements on corrosion resistance of steels (Melchers and Ahammed, 1994; Schultze and Wekken, 1976).

Alloy Immersion zone Tidal zone

Aluminium Not conclusive, but perhaps beneficial long term effect Not conclusive, but perhaps beneficial long term effectChromium Favourable initially (<5 years) therefore undesirableCopper Probably detrimental long term e

Manganese Slight beneficial effect Probably detrimentalMolybdenum Unfavourable short-term, however favourable long term e

Nickel Little effect Beneficial effectPhosphorus Very detrimental long term, low concentration in steels e

Silicon Not importantSulphur May be detrimental even in small quantities


5.4.3. Surface conditions & surface roughnessThe state of the metal surface is known to affect pitting sus-

ceptibility (Szklarska-Smialowska, 1986; Zamaletdinov, 2007). Themore homogenous the surface is, both chemically and physically,the higher is the potential for pitting. Consequently the number ofpit is expected to be lower and metal resistance to pitting will in-crease (Szklarska-Smialowska, 1986; Abood, 2008; Malik et al.,1992). Surface roughness is caused by local ‘weak’ points in theprotective oxide film where a critical Cl� concentration can attend,or by homogeneities resulting from surface preparation; either canincrease the number of active sites for pit nucleation (Szklarska-Smialowska, 1986). Isaacs and Kissel (1972) studied the effect ofexposure time on active pit propagation on AISI 304 stainless steelin 0.4 M FeCl3 solution of pH 0.9, using a scanning reference elec-trode to measure the number of active pits. Both the number ofactive pits and the rate at which their growth was decreased(because of passivation) depended on the steel surface treatment(Szklarska-Smialowska, 1986; Isaacs and Kissel, 1972).

The roughness of a material's surface depends on the surfacepreparation. This preparation is done to ensure proper adhesion ofthe coating. The coating is used to protect the materials from theenvironment by making a smoother surface and, thereby, reducingthe possibility of localized corrosion occurring (Wika, 2012a).Moayed et al. (2003) reported, and it is well established, that attemperature above the CPT, the pitting potentials tend to decreaseas the sample surface roughness increases. The surface roughnessoffers advantages to change the characteristic of sites for pit initi-ation. Pits initiate at specific sites on the surface and a roughersurface generally provides the site with more passage geometry(Szklarska-Smialowska, 1986; Wika, 2012a; Moayed et al., 2003).Moayed et al. (2003) andMalik et al. (1995) studied the relationshipbetween CPT and surface roughness. They found that, with an in-crease in surface roughness, the CPT decreases and standard devi-ation of the test result increases. The higher chance of stable pittingin rough surfaces is attributed to the longer length of diffusion andlarger micro-crevices surrounding the inclusions (Moayed et al.,2003; Malik et al., 1995). Manning et al. (1980) established asimilar correlation on the effect of surface roughness on pittingpotential for single-phase and duplex stainless steel.

5.4.4. Protective coatingProtective coating systems, also known as anticorrosive coating,

is one of the methods developed to protect external surfacesagainst corrosion (Wika, 2012a). The protective coating systems ishighly recommended by classifications society rules such as DNV-RP-G101 (Veritas, 2009) and NORSOK M-501 (Norsok, 2004). Pro-tective coatings are critical for marine structures mainly becausethe environment is very aggressive and corrosive (Veritas, 2002;Sørensen et al., 2009; Norsok, 2004). Marine structures requireapplications consisting of primer, one or several intermediate coats,and a topcoat. The function of the primer is to protect the substrate

from corrosion and to ensure good adhesion to the substrate(Sørensen et al., 2009).

Fig. 11 shows the factors that affect the durability of an anti-corrosive coating system (Sørensen et al., 2009). Sørensen et al.(2009) stated that the coating provides the required colour andgloss for surfaces exposed to external marine environment. Theprotective coating should have a high resistance to ultraviolet ra-diation (Sørensen et al., 2009) as well as adequate resistance toaltering weather conditions and impact from objects. Bayer andZamanzadeh (2004) reported that environmental degradationcaused by moisture, temperature and ultraviolet radiation willreduce the lifetime of the coating. These authors reported six pri-mary causes of the majority of paint and coating-related failures: 1]Improper surface preparation, 2] Improper coating selection, 3]Improper application, 4] Improper drying, curing and over-coatingtimes, 5] Lack of protection against water and aqueous systems, and6] Mechanical damage (Bayer and Zamanzadeh, 2004).

5.4.5. Mill scaleThe presence of mill scale on structure's appears to lead to

greater pitting corrosion (Melchers and Ahammed, 1994). The ef-fect appears to be most significant for atmospheric corrosion andless for tidal and immersed corrosion conditions. Melchers andAhammed (1994) stated that the intensification of rate of pittingattack caused by mill scale is thought to occur due to electro-chemical action between the mill scale and the undecorated metal.Generally, mill scale tends to protect the metal underneath, exceptat cracks. At these points the scale becomes anodic and, because ofthe relatively large cathodic area, the rate of attack caused byelectrochemical action between mill scale and the bare metal atlocal anodes is instantaneous (Melchers, 1994; Melchers andAhammed, 1994).

5.4.6. Effect of PREN valuePitting Resistance Equivalent (PREN) is the capacity of an alloy to

resist pitting. A higher potential is needed to initiate pitting in al-loys with large PREN values thanwith alloys which have low values(Wika, 2012a; Malik et al., 1999). PREN is represented by theEquation (1.6):

PREN ¼ %Cr þ 3%Moþ 16%N (1.6)

PREN is mentioned in several studies when assessing whetheran alloy can resist pitting corrosion (Frankel, 1998; Abood, 2008;Caines et al., 2013; Melchers, 1999; Wika, 2012a; Malik et al.,1999; Xiong et al., 2012; Jargelius-Pettersson, 1998). Lorenz andMedawar (1969) initially established a good correlation betweenthe pitting potential of a wide range of stainless steels. It was foundthat the sum of %Cr þ 3.3 � (%Mo), indicating molybdenum addi-tions, were three times more effective than chromium additions inincreasing pitting resistance (Jargelius-Pettersson, 1998; Lorenzand Medawar, 1969). Malik et al. (1999) reported that, PREN >38

Fig. 11. Factors affecting the durability of an anticorrosive coating system (Sørensen et al., 2009).


conditions should be satisfied to provide pitting corrosion resis-tance to seawater-immersed conditions (Wika, 2012a; Malik et al.,1999). Johansen (2011) studied CPT as a function of PREN. Hisresearch demonstrated that the CPT increases almost linearly withincreasing PREN. The experiment involved testing of stainless steelalloy samples in seawater at 50 �C with the corrosion rate depen-dent on the PREN value. The results showed that the PREN valuedecreases with an increase in temperature (Wika, 2012a; Lorenzand Medawar, 1969; Johansen, 2011). Malik et al.( 1992b) sug-gested that Epit is also a linear function of the PREN value (Wika,2012a; Malik et al., 1992b).

6. Corrosion-related accidents in marine and offshore sectors

A summary of the offshore accidents due to pitting reported inthe literature (Maureen et al., 2013; HSR, 2003; A. International,2004; Zamanzadeh et al., 2004; A. International, 1992; Ne�si�c,2007; A. Stand , 2005) is summarised in Table 5.

7. Methods of identification of pitting corrosion

The first stage in understanding pitting corrosion of steel is tocorrectly identify the phenomenon. Pitting corrosion is character-ized by small flaws in the surface of a material as shown in Fig. 3(Caines et al., 2013). There are many techniques that can identifythe presence of pitting (Caines et al., 2013; A. International, 1992):

7.1. Visual inspection

To identify pitting corrosion, visual inspection can be done inambient light to determine location and severity of pitting. Caineset al. (2013) stated that photographic imaging is often used todocument the difference in appearance of pits before and afterremoval of corrosion products (Caines et al., 2013). Roberge (2008)describe this technique as easiest to employ as it does not requirespecialized equipment and is relatively economical (Roberge,2008). Recently, the use of remotely operated vehicles (ROV) andautonomous underwater vehicles (AUV) replaces dangerous hu-man effort for deep water inspection and underwater work. Theyincrease safety, reduce costs and increase efficiency. These tech-nologies utilize visual imaging and produce high resolution pho-tographs of the corrosion susceptibility in structures (Kros, 2011).

7.2. Metallographic examination

Metallographic examination is an investigative technique thatcan be used to determine the size, shape and density of corrosionpits (Caines et al., 2013). Jana (1995) stated that metallographicexamination is typically a destructive analysis technique as thespecimen must be cut from the component and examined with amicroscope (Jana, 1995). Power and Shirokoff (2012) studied thesimultaneous electrochemical analysis and in situ optical micro-scopy for 316L stainless steel samples submerged in sulphuric acidbased solutions. They reported that this technique provides both adetailed visual account of the corrosion process as well as a stan-dard electrochemical analysis of the pitting potentials and corro-sion rate (Power and Shirokoff, 2012). A brief discussion on themetallographic examination is provided by Caines et al. (2013).

7.3. Non-destructive testing (NDT)

Non-destructive testing (NDT) is a key technique used in in-dustry to evaluate the current state of component and equipmentin service and to aid in maintenance planning. It plays an importantrole in the continued safe operation of physical assets (Caines et al.,2013; A. Stand, 2005; Forsyth, 2011). American Society for Non-destructive Testing (ASNT) defines NDT as “the determination ofthe physical condition of an object without affecting the object'sability to fulfil its intended function. NDT techniques typically use aprobing energy form to determine material properties or to indi-cate the presence of material discontinuities” (Forsyth, 2011). ASTMG46-96 (A. Stand, 2005) stated that NDT technique is applicable toidentify pitting corrosion however, it is not effective at character-izing pitting as a destructive method. Additionally, specializedtraining is required to ensure realistic results (Caines et al., 2013; A.Stand, 2005). A brief description of each NDT method is presentedby Forsyth et al. (Forsyth, 2011) and Caines et al. (Caines et al.,2013).

An NDTmethod is classified according to its underlying physicalprinciple (Caines et al., 2013; Jana, 1995; Forsyth, 2011; Forsythet al., 2006) and common methods are:

� Visual and optical Testing (VT) (Forsyth, 2011).� Radiographic Testing (RT) (Souza et al., 2008).� Electrochemical and Electromagnetic Testing (EC, ET)(Szklarska-Smialowska, 1986).

Table 5Historical offshore accidents reported due to pitting corrosion.

Accident Year Fatalities Cost Descriptions

Umm Said Qatar (weld failure,gas processing plant)

April 1977 3 killed US$ 179million

A tank containing 236,000 barrels of refrigerated propane at 45 �F failed at weld. The possible causeof weld failure was corrosion due to the influence of sulphate reducing bacteria that remained insidethe tank after hydro test with seawater. The wave of liquid propane swept over the dikes beforeigniting a nearby tank containing 125,000 barrels of buthane. It took eight days to completelyextinguish the fire.

Ekofisk Norway (weld failure,offshore platform)

March1980

123 killed Alexandra L Kielland Platform, a semi-submersible oil-drilling platform located at Ekofisk field NorthSea, capsized during a storm. The platform was supported by five columns standing on five 22-mdiameter pontoons. The five 8.5 diameter columns on the pontoons were interconnected by anetwork of horizontal bracings. A cracked bracingmade five other bracings break off due to overload,and the vertical column connected with the cracked bracings became separated from the platform.The platform subsequently became unbalanced and capsized. The investigation showed that acorrosion fatigue crack had propagated from the double fillet near the hydrophone mounted to oneof the horizontal bracings.

Mexico Pipe Leaking (LPG) November1984

650 killed64,000 injured

US$ 29million

A 12-inch pipeline between cylinder and sphere storage ruptured. Initial blast caused a series ofBLEVEs (boiling liquid expanding vapour expansion). The outstanding cause of escalation was theineffective gas detection system and, due to this, of lack of emergency isolation. This explosion andfire is perhaps themost devastating incident ever. The high death toll was due to the proximity of theLPG terminal to a residential complex. The accident is not fully understood however; researchersclaim that this accident occurred due to the pitting corrosion in the pipe.

Piper Alpha North Sea (UK) July 1998 167 killed US$ 1.27billion

This was predominantly an operational error. Gas leaked from two blind flanges due to localizedcorrosion e then the gas ignited and exploded. A pump from two available pumps was tripped, andan operator accidentally changed the backup pump with a pressure relief valve, which had beenremoved, for maintenance. Severity damage of the explosion was due largely to the contribution ofoil and gas pipelines connected to Piper Alpha.

Sinking of the Erika Dec 1999 8 killed Erika broke into two near the coast of France whilst carrying approximately 30,000 tons of heavyfuel. 19,800 tons of fuel spilt along the coast of Brittany and France. This single oil spill was equal tothe total amount of oil spilt worldwide in 1998. A corrosion problem, which had apparently existedon the Erika since 1994, led to this devastating accident in 1999.

Roncador Brazil March2001

2 killed 8missing

US$ 515million

Investigation report of the fire, explosion, and sinking to P-36 (the largest offshore productionfacility) revealed a sequence of events started from the failure of the starboard emergency drain tank(EDT). Excessive pressure in Starboard EDT, due to a mixture of water, oil, and gas, caused a ruptureand leaking of EDT fluids into the fourth level of the column. The leaks were believed to haveoccurred due to corrosion.


� Ultrasonic Testing (UT) (Zhu et al., 1998).� Liquid Penetrant Testing (PT) (Caines et al., 2013).� Magnetic particle Testing (MT) (Caines et al., 2013; Forsyth,2011).

� Acoustic Emission Testing (AET) (Caines et al., 2013;Zamanzadeh et al., 2004) and

� Infrared and thermal Testing (IRT) (Caines et al., 2013).

7.4. Surface analysis technique

Methods of surface analysis are increasingly being used todetect and quantify elements present inside the passive layer (Zaya,1984; Zatkalíkov�a et al., 2010). Auger electrons spectroscopy (AES)is a common analytical technique used in the study of thecomposition of the outer 1e5 atomic layer of the surfaces of solids(Riviere, 1973). During AES, the sample is attacked with 1e10 KeVelectrons and the instrument analyses the emitted auger electrons.The sensitivity to individual elements is about 0.1%, however theaccuracy of the result is fairly poor (Zaya, 1984; Riviere, 1973).

X-ray photo-electron spectroscopy (XPS) consists of subjecting aspecimen to X-ray photons and analysing the ejected electrons. Themain advantage of this technique is that the energy of these elec-trons varies with the chemical state of the sample element. Thedepth sampled, and the sensitivity, is approximately the same as forAES. The main disadvantage of XPS is the poor lateral resolutionobtained due to the absence of focus by the incoming energy(Watts, 1994).

Secondary ion mass spectrometry (SIMS) is a technique forsurface and thin-film analysis. SIMS has been extensively reviewedfrom various instrumental aspects such as analytical applications,comparisonwith other surface analytical techniques, application of

surface studies and fundamental aspects of ion emission (Williams,1985). Usually, these techniques are associated with ion sputtering(ejecting the atom from a solid) to allow for in-depth analysis of thesample. However, sputtering has various disadvantages because itdestroys the chemical bonding which may have been present onthe surface, as well as at the original find topography. It may alsoform a cavity when sputtering is uneven, and some elements maysputter more slowly than others creating a new distribution in theremoved sample (Zaya, 1984; HSR, 2003; Riviere, 1973; Watts,1994).

7.5. Probabilistic approach for pit identification

Pitting corrosion has long been known to be a particularlyinconsistent and unpredictable process (Zaya,1984; Engelhardt andMacdonald,1998). More precisely, it is challenging to forecast whena pit will initiate and where this is going to take place, thereforeresearchers have tried to extract data from the distributions duringtimes of pit nucleation events (Valor et al., 2010; Zaya, 1984).Shibata and Takeyama (1977) were the first to postulate that thecritical potential necessary to induce pitting, and the inductiontime elapsed before pits become observable, are both statisticallydistributed quantities (Henshall, 1992). They asserted that thenucleation of a pit is a statistical process similar to the developmentof a crack in brittle material. They conjectured that the pit gener-ation process has the Markov property, i.e. that the future proba-bility of pit nucleation is uniquely determined once the state of thesystem at the present stage is known (Zaya, 1984; Shibata andTakeyama, 1977). Henshall (1992) found that the stochastic modelof pitting corrosion was useful in predicting corrosion damage ofhigh-level radioactive waste containers. They stated that the modelincludes simple phenomenological relationships describing


environmental dependence of stochastic parameters, and that itcan simulate pit initiation and growth under various environments,including those that change during exposure (Henshall, 1992).Similarly, Valor et al. (2007, 2013) used a new stochastic model forpit initiation and pit growth (Valor et al., 2007, 2010, 2013).

Spatial distribution models are commonly used in locationalanalysis, including spatial location of activities among the zones ofa region and measure of interaction between zones (Wilson, 1967).In the case of pitting corrosion models, spatial distributions is usedwhen the pits on the sample do not exactly follow a poisson dis-tribution (Zaya, 1984). Several researchers have modified thespatial distributions to model pitting corrosion (Vel�azquez et al.,2014; Khan and Howard, 2007; Valor et al., 2007, 2013; Wilson,1967; Aziz, 1956; Chaves and Melchers, 2014; Melchers, 2005e;Melchers et al., 2010; Melchers, 2008b). In addition, Aziz et al.(Aziz, 1956) introduced exponential distribution of the pit depth tocalculate the maximum depth by the statistic of extreme values(Aziz, 1956).

8. Corrosion modelling

The quantitative understanding (i.e. a model) of how thecorrosion process operates as a function of exposure time and un-der various environmental influences is necessary. It is alsorequired to predict the likely amount of corrosion in the future fordefined conditions (Davydov, 2008). Prediction and identificationof pitting corrosion in marine and offshore structures is a difficultproblem for a number of reasons (Frankel and Sridhar, 2008).Firstly, the events take place on a very small scale, with passive filmnanometres in thickness and with initiation sites of similar sizes.Immediately after initiation, the rate of pit growth can be extremelyhigh, even tens of A/cm2 (Frankel, 1998; Melchers, 1994; Frankeland Sridhar, 2008). Frankel and Sridhar (2008) considered thepitting location as an extremely dynamic one with rapidly movingboundaries and rapidly changing chemistries.

The modelling of pitting corrosion in marine and offshore con-ditions has been study of interest for some time. The effect of themain factors in pitting corrosion modelling, such as temperature,bacterial community, oxygen concentration, pH, and velocity, hasbeen considered in the past by several researchers (Melchers,2001a; Moayed et al., 2003; B€ohni, 2000; Melchers, 2002; Youniset al., 2012; Malik et al., 1992a; Pardo et al., 2000; Scheers, 1992).

Researchers (Melchers and Jeffrey, 2008a; Melchers, 2004b,2001a, 2014a, 1998, 2003b; Melchers et al., 2010; Melchers andJeffrey, 2008b, 2011; Melchers and Wells, 2006) proposed awidely acceptedmultiphase phenomenological model for corrosion

Fig. 12. General schematic of model for corrosion loss showing the changing behaviourof the corrosion process as a series of sequential phases (adapted from (Melchers(2003b)).

loss as a function of exposure period. As shown in Fig. 12, the modelwas developed based on corrosion science philosophy. The modelhas several phases, including kinetic, diffusion, transition, andanaerobic, and each of these phases is believed to control thecorrosion process (Melchers, 2005a, 2003b, 2003c). This model isbased on theoretical and empirical corrosion mechanics. In Fig. 12,phase 2 indicates the theoretical point at which anaerobic condi-tions is reached (Melchers, 2005a, 2010, 2004b, 2009, 2007). Thiscondition tends to encourage rapid marine growth and, by infer-ence, conditions favourable for sulphate-reducing bacteria (SBR)including activation and growth (Melchers and Wells, 2006). FromFig. 12, the number of sequential phases which correspond to thedifferent processes controlling the (instantaneous) rate of corrosionfor ‘at sea’ conditions is summarised as below (Melchers, 2005a,2003a, 2003b, 2003c, 2006b, 2008c):

Phase 0 e a very short period of time during which corrosion isunder ‘activation’ or kinetic control. This is governed by the rate atwhich local chemical reactions can occur unhindered by externaldiffusion or transportation limitations (Melchers, 2005b, d, 2003b).

Phase 1 e a period of ‘oxygen concentration’ control; thecorrosion rate is governed by the rate of arrival of oxygen throughthe water and the corrosion product layer to the corroding surface.It can be seen in the multiphase model that this phase can bemodelled, to an approximation, as a linear relationship betweendepths of corrosion vs. time (Melchers and Ahammed, 1994;Melchers, 2014a).

Phase 2 e in this phase the corrosion rate is controlled by therate of oxygen diffusion to the corroding surface through theincreasing thickness of the corrosion product. This phase is there-fore a non-linear function of time e the reason being is that, whenthe corrosion products build up, the oxygen flux will decline andeventually anaerobic conditions will be established in the zonebetween the corrosion product and the corroding metal (Melchersand Jeffrey, 2011; Melchers and Wells, 2006; Melchers, 2014b).

Phase 3 e in this phase the rate of corrosion is controlled by themetabolic rate of sulphate reducing bacteria (SRB) under anaerobicconditions. Often, this phase 3 is a period of rapid growth of SRBresulting from the conditions now being suitable for their increasedmetabolism and from the plentiful supply of nutrients (Melchers,2014b). The high rate of metabolite production initially producesa high rate of corrosion, but this reduces to a quasi-steady statesituation in equilibrium with the rate of supply of nutrients andenergy sources from the bulk water (Melchers, 2009; Melchers andJeffrey, 2011; Melchers, 2006b, 2014b, 2012).

Phase 4 e this represents the quasi-steady state situationreached at the end of phase 3 and is presumed to continue indef-initely. Melchers et al. (Melchers, 2014b) stated that presently thereis insufficient understanding of the precise mechanisms involved in

Fig. 13. Corrosion loss model and data fitting for long term corrosion data accessedfrom ASTM worldwide corrosion data.


this phase; however, the long term data for pitting corrosion sug-gests that it is closely linear with time (Refer Fig. 13). Observationsfor a variety of steels also suggest that the corrosion rate is largelyindependent of the actual activity levels of SRB and of the type ofsteel (Melchers and Jeffrey, 2011; Melchers, 2014b, 2012).

Modelling of the uniform/general corrosion in marine andoffshore structures is reviewed in the subsequent sections; how-ever, the focus is on modelling the pitting corrosion in marine andoffshore structures (Melchers, 2004a, 2005a; Vel�azquez et al., 2014;Melchers, 2008a; Khan and Howard, 2007; Melchers, 2004b, 2009;Valor et al., 2007; 2010; Melchers, 2003a; Guedes Soares et al.,2011; Melchers, 2005b, 2006a; Ting et al., 2011; Valor et al., 2013;Chaves and Melchers, 2014; Melchers et al., 2010; Melchers,2008b, 2003c, 2006b, 2003d, 2006c; Melchers and Jeffrey, 2008c;Melchers, 2001b; Rajani and Kleiner, 2001; Katano et al., 2003;Guedes Scares et al., 2005; Mao, 2007; Paul, 2012; Jain et al.,2013; Kolios et al., 2014; Melchers, 2003e).

8.1. Modelling the general corrosion in marine environments

Statistical and stochastic aspects of pitting corrosion were firstdeveloped in the late 1970s (Mao, 2007). In large-scale engineeringstructures, the measurement of pit depth and frequency isconsidered to be time-consuming and expensive. Therefore, onlythe deepest pits are studied since they are likely to be the cause offailure in these structures (Vel�azquez et al., 2014; Valor et al., 2010;Mao, 2007). The extreme value statistics developed by Gumbel arewidely used in the application of the maximum pit depth distri-butions. Extreme value distributions include three types ofasymptotic distributions for an infinite number of samples:

Type I : Gumbel distribution FðxÞ � expð � exp½ � x�Þ (1.7)

Type II : Cauchy distribution FðxÞ � exp�� x�K

�(1.8)

Type III : Weibull distribution FðxÞ � expð � exp½u� x�kÞ(1.9)

where x is a random variable representing the maximum pit depth,and k and u are constants.

Melchers et al. (Melchers, 2003e) studied the assessment of theremaining safe and serviceable life of deteriorating structures. Aprobabilistic model for immersion corrosion loss with time on mildand low alloy steels was developed based on the fundamentalphysiochemical model (Melchers, 2003a, 2003c). They stated that,for structural reliability assessment using probability theory, thestructural strength at any point in time t can be represented by R(t)with probability function fR (r,t), where R(t) is a random variableand r is a discrete, deterministic value. The generic form of theproposed model has material loss due to corrosion as a function oftime and is expressed by:

cðt; EÞ ¼ bðt; EÞ$f ðt; EÞ þ εðt; EÞ (2.0)

where c(t, E) is the average depth of penetration from one side ofsteel plate (mm); f (t, E) ¼ mean value function; b (t, E) ¼ biasfunction which should be unity when f (t, E) represents c (t, E)exactly under all conditions; ε(t,E) ¼ zero mean uncertainty func-tion; and E ¼ vector of factors which characterize the exposureenvironment, steel composition, and the surface finish factors.From this analysis Melchers (2003e) stated that the rate of pitgrowth is not a linear function of time (Melchers, 2003e). Addi-tionally, Melchers (2003c) proposed a similar probabilistic model-ling based on corrosion mechanics and environmental factors. The

environmental factors included were: temperature, dissolved ox-ygen, salinity, calcium carbonate, pH, water velocity and marinegrowth. The probabilistic model divided the corrosion into fourparticular stages: initial corrosion; oxygen diffusion controlled bycorrosion product and micro-organic growth; aerobic activity withlimited food supply; and anaerobic activity (Melchers, 2003b).

Neither of the previous studies attempted to include the effectsof the different operational or environmental factors on thecorrosion degradation expected through the lifetime of the struc-tures. Melchers (2005b) stated that in order to improve corrosionmodels it is necessary to not only account for time but also includecontributing variables (Melchers, 2005b). They also identified themain corrosion mechanisms that can be found in steel structures aswell as the main environmental factors that affect them (Melchers,2003a, c). Conversely, other studies tried to extend the previouslydeveloped, time-dependent corrosion models to include the effectof additional different environmental factors. These studies weremainly based on statistical analysis of field and experimental data(Melchers, 2003a, 2006a, 2003c).

Using literature field data as well as new field observations;Melchers (2004a, b, 2003b) calibrated the parameters of multi-phase time-to-corrosion models on mild and low alloy steel underfully aerated immersion conditions. They proposed a phenome-nological model for general corrosion of mild, low alloy steel undernear-surface conditions; this was adapted from a model of at-seaimmersion conditions which had already been proposed(Melchers, 2001b). They utilized literature data from the previouswork (Melchers, 2003b, 2001b) and applied a similar probabilisticframework (refer to Equation (1.7)) (Melchers, 2003d). The cali-brated parameters were then used to predict the corrosion degra-dation as a function of time and seawater temperature (Melchers,2004a, 2004b, 2003b).

Melchers (2006a) proposed examples of mathematical model-ling for long term general corrosion of structural steels in seawater.Considerationwas given to the early corrosion controlled by oxygenconcentrationwhich then later evolves to anaerobic conditions. Themodel was calibrated using extensive field data received from anoffshore oil platform site located in Haloong Bay, Vietnam, and fromcoastal sites along the China Coast (Melchers, 2006a). Furthermore,Melchers (2006b) proposed the probabilistic model for atmo-spheric corrosion of structural steels in ocean environments(Melchers, 2006b). However, they disagree with some of the as-sumptions previously made such as that corrosion is the linearfunction of time, i.e. that there is a constant corrosion rate and theassumption that the corrosion rate is a monotonically decreasingfunction of time (Melchers, 2006b).

8.2. Modelling of pitting corrosion in marine environments

Katano et al. (2003) proposed a predictive model for pit growthon underground pipes. They relied on the theory that the pittingcorrosion rate for metal depends on environmental factors. Theyestablished the relationship between pitting depth and environ-mental factors. This relationship was explored through regressionanalysis, with pitting depth as dependent variables and on envi-ronmental factors as independent variables. Pitting depth wasexpressed as a power of time t.

h ¼ gta (2.2)

where, g and a are constant. From the above relationship, theyderived a statistical model of pitting depth (y) as a dependent factorand environmental factors (x) as independent factors. A regressionmodel is expressed as:


y ¼ h expðεÞhε � N

�0;s2

�i; h ¼ ta exp

0@b0 þ

Xpj¼1

bjxj

1A (2.3)

where, a and b are regression coefficient.Melchers (2005b) extended the model previously developed

(Melchers, 2005e) to include the effect of dissolved oxygen in orderto obtain a multiphase, time-dependent pitting corrosion degra-dation model incorporating the effect of dissolved oxygen con-centrations (Melchers, 2005b). Furthermore, Melchers (2005b, c)gave consideration to how the previous observations could be usedto represent the effect of water velocity on the previously proposedmodels. They stated that it is essential to investigate the influenceof water velocity on early corrosion behaviour; only then could thecorrosion rate be correlated with the dissolved oxygen in the earlystages of corrosion. They thereby confirmed that the depth is not anindependent factor in marine immersion corrosion of mild steel;however, dissolved oxygen, water temperature, and local watervelocity were found to be the main influencing factors (Melchers,2005b, c).

Guedes Scares et al. (2005) studied the effects caused bydifferent environmental factors on the pitting corrosion behaviourof steel plates totally immersed in salt water conditions. Theyproposed a corrosion wastage model based on a non-linear time-dependent function. This model accounts for the effect of variousenvironmental factors including salinity, temperature, dissolvedoxygen, pH, and flow velocity (Guedes Scares et al., 2005).

Melchers (2005a) studied the statistical characteristics of pittingcorrosion, represented by the extreme value distribution ‘Gumbel’.Since then, considerable progress has been made in the rationaldescription and mechanistic modelling of individual pit initiationand early pit growth. Regardless, the modelling for growth in pitdepth tends to be largely empirical and laboratory observations areunlikely to be representative of field conditions. In particular, ex-periments cannot include the precise effect of bacterial activity inpitting corrosion (Melchers, 2005a, 2005b, 2005e). They proposedthe phenomenological model for pit growth and extreme value ofpit depth using 7 years of field data. From this analysis, they statedthat only those pits that initiate immediately upon exposure, andthen grow as stable pits, can become extreme depth pits. Distri-butions of all pit depths indicate the probability of occurrences ofmaximum pit depth for short and long marine immersion condi-tions. A Gumble distribution is commonly adopted as extreme valvedistributions because it provides for less conservative probabilityestimates (Melchers, 2005a, 2005b, 2005e).

Melchers (2004a) studied the variability of maximum pit depthof mild steel specimens subjected to marine immersion. Here theauthors considered the effect of anaerobic conditions for immer-sion pitting in mild steel. The datawas calibrated andmaximum pitdepth was shown to be a function of seawater temperature(Melchers, 2004a, 2005e). When considering the multiphasephenomenological model, it was found that pitting was most se-vere when widespread anaerobic conditions were established atthe corroding surface. It was acknowledged that the Gumbel dis-tributions plot that was used previously for expressing pit depthvariability was combined with a simple power function. However,they reported that the pit growth is a more complex functioninvolving several stages such as pit initiation, metastable pit andstable pit (Melchers, 2004a, 2005a, 2003b).

Melchers (2005e) then introduced super stable pit growthwhich initiates immediately after exposure and grows withoutsignificant metastable activity (Melchers, 2005e). They reportedthat, as a result of super stable pit there is likely to be a high degreeof dependency among the depths of extreme (deepest) pits. By this

time, when the super stable pits are introduced the use of extremevalue distributions (Gumbel distributions) started to becomedoubtful because it was now known that the statistical populationof super stable pits are likely to be different from the remainingpits. Melchers (2005e) recommended the use of normal distribu-tions (modelling the probability of occurrences) to represent theextreme pit depth of super stable pitting. Using the structuralreliability system theory, the distributions of the pit depth (for thedeepest depth) can be expressed as:

FXmaxðaÞ ¼ P½Xmax < a� ¼ P½all pits xi < a� (2.4)

where, event xi is the pit depth represented as random variable; P[ x i<a] the probability of occurrences; and X is the random vectorwhich hold all component xi. From this analysis, Melchers (2005e)identified that if the probability distributions for the deeper pits arenormal, the probability distributions for the extreme pits shouldalso be normal distributions. They further established that theseprobability distributions have more certainty than those estimatedby conventional approaches such as Gumbel distributions(Melchers, 2005e).

Khan and Howard (2007) proposed statistical methods toimprove the estimation of degradation rate. They indicated thatcombining the statistical methods with a reliability assessmentwould offer a potential for better use of inspection results, and forthe prediction of the probability of future leaks in offshore pipes aswell as remaining life assessment (Khan and Howard, 2007;Kowaka and Tsuge, 1994). In pitting corrosion, the average degra-dation rate does not represent the real status of material degra-dation due to the nature of pit initiation and to its usually difficultlocation. Hence, for pit corrosionmodelling purposes, focus is givento establishing the relationship between the maximum pit depthsin the given exposure. Khan et al. (Khan and Howard, 2007) rec-ommended a use of extremes value statistics in pitting corrosion toinvestigate extreme values and proposed linear, power law andlogarithmic extreme value models (Khan and Howard, 2007;Kowaka and Tsuge, 1994).

Linear model : x� x0 ¼ kðT � TiÞ (2.5)

Power law model : x� x0 ¼ kðT � TiÞn (2.6)

Logarithmic lawmodel: x� x0 ¼ k logðT � TiÞ (2.7)

where, x0 is the threshold depth of degradation (pit depth) atinitiation time Ti; x is the measured depth at time T; and k is thepitting corrosion rate. Depths exceeding x0 would grow, whereasdepths lower than the x0 may fail to grow (Khan and Howard, 2007;Kowaka and Tsuge, 1994). Khan and Howard (2007) and Kowakaand Tsuge (1994) provided a brief list of the applicable distribu-tion types for localized degradation.

Melchers (2008b) proposed an extreme value analysis for longterm marine pitting corrosion of steel affected by corrosive agentsulphate reducing bacteria (SRB). They argued that the conven-tional use of Gumbel distributions is no longer appropriate forderiving extreme value statistics for maximum depth of pits inpitting corrosion (Melchers, 2008b). They stated that conventionaluse of Gumble distribution does not consider either stable or metalstable pit growth.

SRB is the main corrosive agent for longer term pitting corro-sion. Modelling of long term pitting corrosion is rather difficultbecause of the lack of sufficient data in one exposure location hencethe use of Frechet extreme value distributions is appropriate as itcombines the data from different exposure times (Melchers,2008b). Melchers (2008b) proposed a statistical model for longer-


term, maximum pit depth which considers the effect of anaerobicconditions.

xðt; TÞ ¼ C½eðt; TÞ=rðtÞ�t (2.8)

where, e(t,T) denotes the rate of supply of nutrients to the corrodingsurface; t is the time elapsed since the commencement of overallanaerobic activity; and T is themeanwater temperature. Also, if r(t)is the proportion of the surface covered by pits, then the rate ofsupply of nutrient per pit at any time t is proportional to e(t,T)/r(t)(Melchers, 2008b).

Mohammad et al. (2012) proposed a prediction model of pittingcorrosion characteristics using Artificial Neural Networks (ANNs).They used a pre-corroded steel specimen and immersed it in cor-rosive ferric chloride solutions in different concentrations. It wasfound that the ANNs results agreed well with those obtained fromlaboratory tests. They further stated that, by increasing the corro-sive concentration with extended immersion duration, it resultedin an increase in pitting density and pitting depth (Mohammadet al., 2012).

Valor et al. (2013) proposed Markov chain models for the sto-chastic modelling of pitting corrosion (Valor et al., 2007, 2010,2013). Two different models of pitting corrosion were proposed:1) a continuous-time, nonhomogeneous linear growth Markovprocess was used to model external pitting corrosion in under-ground pipelines, and 2) the distribution of maximum pit depths inpitting experiments was modelled combining pit initiation and pitgrowth processes (Valor et al., 2013).

Vel�azquez et al. (2014) proposed amethodology for probabilisticmathematical modelling of the pit initiation process and its depth-of-growth process. Two stochastic models were developed: 1) thePoisson process which was used to modal pit initiation, and 2) theGamma process to model the pit depth-growth (Vel�azquez et al.,2014). After obtaining the pit depth, the maximum pit depth wasstudied using Block Maxima (BM) and the Peak-Over-Threshold(POT) methods.

Vel�azquez et al. (2014) described the pit initiation as Non-Homogenous Poisson Process (NHPP). According to NHPP, pitinitiation time was assumed with intensity function lm(t), where l

is defined as the mean pit density per area unit and m(t) can be anarbitrary function. Hence, the expected number of pits in giventime t can be expressed as:

EðNðtÞjlZt

s¼0

mðsÞ ¼ lðmtÞ (2.9)

The pit growthwasmodelled using the Gamma process which isa continuous-time stochastic process with independent gammaincrements. The Gamma probability density function can beexpressed by:

Gað$jk; qÞ ¼ yk�1q�kexpð�y=qÞGðkÞ (3.0)

where k is shape parameter which controls the rate of the jump; qthe inverse of the scale parameters which controls the jump sizes;Gað$jk; qÞ is the Gamma probability density function; and y > 0 and

GðkÞ ¼Z ∞

t¼0tk�1 expð�tÞdt is the Gamma function.

Finally, Vel�azquez et al. (2014) reported that the use of statisticalsimulations in the modelling pitting corrosion is significant as to befamiliar with different pit characteristics such as pit depth, pitdensity and rate of pitting. They ensure that the NHPP process candetermine the evolution of the total number of pits and, likewise,the Gamma process for the pit growth (Vel�azquez et al., 2014).

9. Challenges

There are many studies conducted on the mechanism, field-testing and laboratory testing in the field of pitting corrosion;however, there is no development of detailed studies for failures inoffshore structures due to pitting characteristics. One criticalrequirement is the investigation of how pitting characteristics, suchas rate, depth, density and distance, can cause a structural failure.Previously, empirical and statistical degradation models weredeveloped by either fitting field or laboratory data. However thesemodels, even though useful for specific site or operating conditions,still carry high uncertainty. Currently, several aspects of currentknowledge on pitting corrosion are deficient and require furtherinvestigation. Some of these challenges have been identified andlisted below:

� Structural failures due to pitting characteristics such as pitdepth, pit rate, pit density and interfacial distances between pitsis not fully understood; better understanding of these charac-teristics is crucial.

� The process of pit propagation and the rate of pit growth are notfully understood.

� Precise prediction mechanism of long term anaerobic corrosionrequires essential study in order to develop a failure model.

� The total depth of pit and its rate of growth are of concern inmarine and offshore structures; the precise mechanics of pitgrowth mechanism needs to be developed.

� There is still uncertainty in the pit depth measurement; a pre-cise pit depth measurement technique needs to be developed.

� There is no precise way to observe the depth of deepest pitwithout destroying the specimen; this needs to be resolved.

� Several factors that can affect pitting corrosion in marine envi-ronments have been established in this paper; however, theeffect of these factors on pit growth is still not fully understood.

� Mathematical relationship for the factors that affect pitting ratesuch as surface wetting, humidity, oil in the water, suspendedsolids etc. is not yet understood.

� The influences of pH on pitting corrosion under dynamic con-ditions may also be studied and the role of metal defect on pitinitiation needs further analysis.

� Another possible expansion is the study of the repassivationmechanisms, including super saturation of the solution, pre-cipitation of the passivating phase and subsequent increase inthe electrical resistance of the pore.

� In this paper, several conventional and recent methods formodelling of pitting corrosion are discussed to evaluate the rateof corrosion; however, there is a need to find an appropriatemethod of estimating the corrosion rate and how it developswith time.

Provided that all factors that can cause pitting corrosion areknown e as based on the mechanism of corrosion and by using riskassessment methodologies e a probabilistic risk assessment can beapplied to predict and evaluate future failure due to pittingcorrosion.

10. Conclusion

Pitting corrosion is a complex but important problem that is atthe root of many structural and system failures. It has been studiedfor many years however crucial phenomena remains unclear. Theaim of this paper was to identify and evaluate the parameters thataffect pitting corrosion in marine and offshore environments. Thispaper has reviewed and discussed the mechanisms and charac-teristics of pitting corrosion, several factors that affect its


development, as well as identification methods and modellingtechniques. Based on the literature reviewed it is clear that pittingcorrosion is a stochastic, probabilistic phenomenon that requiresinterdisciplinary concepts that incorporate surface science, metal-lurgy/material science, hydrodynamics and chemistry. From thisstudy, the following conclusions suggestive of recent knowledge onpitting corrosion can be summarized as:

� Pitting corrosion is considered as one of the most destructiveforms of corrosion; pitting mechanism and the characteristics ofpitting is summarized in this paper.

� It is generally acknowledged that there are three stages ofpitting.

� Pit can be initiated inmany different ways and the growth of pitscan be attributed to different phenomena.

� The several factors that affect pitting corrosion and the rate ofpitting have been investigated; the critical factors that havemost effect on the pitting rate are temperature, pH, bacterial andflow velocity, however, this varies on marine zones such as at-mospheric, splash, emersion etc.

� Various experimental investigations have shown that differenteffects may serve as stabilizing factors for localized corrosionand it depends on the stage of development of corrosion pit andenvironmental conditions.

� Pit depth has been acknowledged as a critical factor and is thekey parameter to describe pitting rate.

� Some standard laboratory methods of determining the pittingcorrosion rate, and the effect of different factors on rate of pitgrowth, are considered to be inaccurate.

� Although a standard exists for statistical analysis of laboratorycorrosion data, no such standard exists for the analysis of in-spection data relating to corrosion measurement.

� Short term pitting corrosion can be modelled from the experi-mental field data; however, this data cannot be relied on formodelling of long term pitting corrosion.

� In recent years, there have been different attempts made onprobability modelling for general corrosion as a function oftime; however, less is available for pitting corrosion undermarine immersion conditions.

� The use of statistical simulations in pitting corrosion is valuablefor determining pitting characteristics such as pit initiation andpit growth.

� The power law is commonly used to express pitting depth as afunction of time. Some of the statistical distributions are: Pois-son, Exponential, Normal, Log normal and Extreme valuedistributions.

� Extreme value analysis is considered most appropriate for thestudy of pitting corrosion; extrapolation from a small-inspectedarea to a large area is possible with this method.

� A Gumbel distribution is widely used for the application ofextreme value statistics in corrosion engineering.

� The conventional use of the Gumbel distributions is no longerappropriate to derive the extreme value statistics for maximumdepth of pits in pitting corrosion. It is suggested that, for longer-term pitting corrosion, the use of Frechet extreme value distri-butions is more appropriate.

� Field test is suggested for generating long-term data. This wouldallow for the collection of relevant environmental data andwould develop further understanding of degradation mecha-nisms and pitting corrosion rates.

Acknowledgements

Authors thankfully acknowledge support provided by NationalCentre for Maritime Engineering and Hydrodynamic (NCMEH) and

Australian Maritime College (AMC).

References

Abdel-Ghany, R., Saad-Eldeen, S., Leheta, H., 2008. The effect of pitting corrosion onthe strength capacity of steel offshore structures. In: ASME 2008 27th Inter-national Conference on Offshore Mechanics and Arctic Engineering, AmericanSociety of Mechanical Engineers, pp. 801e805.

Abood, T.H., 2008. The influence of various parameters on pitting corrosion of 316Land 202 stainless steel. In: Department of Chemical Engineering of the Uni-versity of Technology. University of Technology.

Acu~na-Gonz�alez, N., Gonz�alez-S�anchez, J.A., Dzib-P�erez, L.R., Rivas-Menchi, A., 2012.Early Corrosion Fatigue Damage on Stainless Steels Exposed to TropicalSeawater: a Contribution from Sensitive Electrochemical Techniques.

Agar, J., Hoar, T., 1947. The influence of change of size in electrochemical systems.Discuss. Faraday Soc. 1, 158e162.

Al-Fozan, S.A., Malik, A.U., 2008. Effect of seawater level on corrosion behavior ofdifferent alloys. Desalination 228, 61e67.

Alkire, R.C., Wong, K.P., 1988. The corrosion of single pits on stainless steel in acidicchloride solution. Corros. Sci. 28, 411e421.

Angal, R., 2010. Principles and Prevention of Corrosion. Alpha Science International.Aziz, P., 1956. Application of the statistical theory of extreme values to the analysis

of maximum pit depth data for aluminum. Corrosion 12.A. Stand, G46-94, 2005. Standard Guide for Examination and Evaluation of Pitting

Corrosion. ASTM International, West Conshohocken, PA.A. International, 2004. A.S.f. Testing, Materials, Annual Book of ASTM Standards.

American Society for Testing & Materials.A. International, 1992. ASM Handbook: Friction, Lubrication, and Wear Technology.

ASM International.Bayer, G.T., Zamanzadeh, M., 2004. Failure Analysis of Paints and Coatings, Pub-

lished Internally by Matco Associates SUR FAC EPRE PA R AT IONANDC OATINGAPPLI CAT IONPR AC TICES, 331.

Board, N.T.S., 2014. Columbia Gas Transmission Corporation Pipeline Rupture.B€ohni, H., 2000. Influence of temperature on the localized corrosion of stainless

steels. Russ. J. Electrochem. 36, 1122e1128.Caines, S., Khan, F., Shirokoff, J., 2013. Analysis of pitting corrosion on steel under

insulation in marine environments. J. Loss Prev. Process Ind. 26, 1466e1483.Carpen, L., Hakkarainen, T., Sarpola, A., Riihimaki, M., Riimo, J., Kinnunen, P.,

Pohjanne, P., 2007. Localized corrosion risk of stainless steels under evaporativeconditions. Corrosion paper#07197.

Chandler, K.A., 1985. Marine and Offshore Corrosion.Chaves, I.A., Melchers, R.E., 2014. Extreme value analysis for assessing structural

reliability of welded offshore steel structures. Struct. Saf. 50, 9e15.Comotti, I.M., Trueba, M., Trasatti, S.P., 2013. The pit transition potential in the

repassivation of aluminium alloys. Surf. Interface Anal. 45, 1575e1584.Dallek, S., Foley, R., 1976. Mechanism of pit initiation on aluminum alloy type 7075.

J. Electrochem. Soc. 123, 1775e1779.Darmawan, M., 2010. Pitting corrosion model for reinforced concrete structures in a

chloride environment. Mag. Concr. Res. 62, 91e101.Darmawan, M., Stewart, M., 2007. Effect of pitting corrosion on capacity of pre-

stressing wires. Mag. Concr. Res. 59, 131e139.Davydov, A., 2008. Analysis of pitting corrosion rate. Russ. J. Electrochem. 44,

835e839.Engelhardt, G., Macdonald, D., 1998. Deterministic prediction of pit depth distri-

bution. Corrosion 54, 469e479.Evans, V., 1976. The Corrosion and Oxidation of Metals (Second Supplementary

Volume).Evans, U., Bannister, L., Britton, S., 1931. The velocity of corrosion from the elec-

trochemical standpoint. Proc. R. Soc. Lond. Ser. A 131, 355e375.Forsyth, D.S., 2011. Non-destructive Testing for Corrosion, Corrosion Fatigue and

Environmentally Assisted Cracking in Aging Military Vehicles (RTO-AG-AVT-140).

Forsyth, D.S., Yolken, H.T., Matzkanin, G.A., 2006. A brief introduction to nonde-structive testing. AMMTIAC Q. 1 (Number 2), 7e10.

Frangini, S., De Cristofaro, N., 2003. Analysis of the galvanostatic polarizationmethod for determining reliable pitting potentials on stainless steels in crevice-free conditions. Corros. Sci. 45, 2769e2786.

Frankel, G., 1998. Pitting corrosion of metals a review of the critical factors.J. Electrochem. Soc. 145, 2186e2198.

Frankel, G., Sridhar, N., 2008. Understanding localized corrosion. Mater. Today 11,38e44.

Friend, J.N., 1940. Deterioration of Structures of Timber, Metal, and ConcreteExposed to the Action of Sea-water.

Galvele, J., 2005. Tafel's law in pitting corrosion and crevice corrosion susceptibility.Corros. Sci. 47, 3053e3067.

Ghali, E., Dietzel, W., Kainer, K.-U., 2004. General and localized corrosion of mag-nesium alloys: a critical review. J. Mater. Eng. Perform. 13, 7e23.

Greene, N., 1962. Predicting behavior of corrosion resistant alloys by potentiostaticpolarization methods. Corrosion 18, 136te142t.

Guedes Scares, C., Garbatov, Y., Zayed, A., Wang, G., Melchers, R., Paik, J., 2005. Non-linear corrosion model for immersed steel plates accounting for environmentalfactors. Discussion. Trans. Soc. Nav. Archit. Mar. Eng. 113, 306e329.

Guedes Soares, C., Garbatov, Y., Zayed, A., 2011. Effect of environmental factors onsteel plate corrosion under marine immersion conditions. Corros. Eng. Sci.


Technol. 46, 524e541.HSR, 2003. Offshore hydrocarbon release statistics and analysis. In: H. HID Statistics

Report 2002. Health and Safety Executive, UK. UK, 2002.Hadfield, R., 1922. The Corrosion of Iron and Steel, Proceedings of the Royal Society

of London. Series A, Containing Papers of a Mathematical and Physical Char-acter, pp. 472e486.

Henshall, G.A., 1992. Modeling pitting corrosion damage of high-level radioactive-waste containers using a stochastic approach. J. Nucl. Mater. 195, 109e125.

Isaacs, H., Kissel, G., 1972. Surface preparation and pit propagation in stainlesssteels. J. Electrochem. Soc. 119, 1628e1632.

Jain, S., Beavers, J.A., Ayello, F., Sridhar, N., 2013. Probabilistic model for stresscorrosion cracking of underground pipelines using Bayesian networks. Corro-sion 2013.

Jana, S., 1995. Non-destructive in-situ replication metallography. J. Mater. Process.Technol. 49, 85e114.

Jargelius-Pettersson, R., 1998. Application of the pitting resistance equivalentconcept to some highly alloyed austenitic stainless steels. Corrosion 54,162e168.

Johansen, N.A., 2011. Korrosjon på AISI 316-rør. Utfordringer og tiltak. STATOIL.Katano, Y., Miyata, K., Shimizu, H., Isogai, T., 2003. Predictive model for pit growth

on underground pipes. Corrosion 59, 155e161.Khan, F., Howard, R., 2007. Statistical approach to inspection planning and integrity

assessment. Insight Non-Destr. Test. Cond. Monit. 49, 26e36.Kolios, A., Srikanth, S., Salonitis, K., 2014. Numerical simulation of material strength

deterioration due to pitting corrosion. Procedia CIRP 13, 230e236.Kowaka, M., Tsuge, H., 1994. Introduction to Life Prediction of Industrial Plant

Materials: Application of the Extreme Value Statistical Method for CorrosionAnalysis. Allerton Pr.

Kros, H., 2011. Performing detailed level 1 pipeline inspection in deep water with aremotely operated vehicle (ROV). In: Offshore Technology Conference, OffshoreTechnology Conference.

LaQue, F., 1959. Corrosion testing of electrodeposited coalings. In: Proc. AES, p. 141.Lewis, J.R., Mercer, A.D., 1984. Corrosion and Marine Growth on Offshore Structures.Li, H., Brown, B., Ne�si�c, S., 2011. Predicting localized CO2 corrosion in carbon steel

pipelines. Paper Corrosion. paper#11253.Little, B.J., Lee, J.S., Ray, R.I., 2008. The influence of marine biofilms on corrosion: a

concise review. Electrochim. Acta 54, 2e7.Lorenz, K., Medawar, G., 1969. Über das Korrosionsverhalten austenitischer Chrom-

Nickel-(Molybd€an-) St€ahle mit und ohne Stickstoffzusatz unter besondererBerücksichtigung ihrer Beanspruchbarkeit in chloridhaltigen L€osungen. Thys-senforschung 1, 97e108.

Malik, A., Mayan Kutty, P., Siddiqi, N.A., Andijani, I.N., Ahmed, S., 1992. The influenceof pH and chloride concentration on the corrosion behaviour of AISI 316L steelin aqueous solutions. Corros. Sci. 33, 1809e1827.

Malik, A.U., Andijani, I.N., Siddiqi, N.A., 1992. Corrosion Behavior of Some Conven-tional and High Alloy Stainless Steels in Gulf Seawater. Technical report. SalineWater Conversion Corporation, KSA.

Malik, A., Siddiqi, N., Ahmad, S., Andijani, I., 1995. The effect of dominant alloyadditions on the corrosion behavior of some conventional and high alloystainless steels in seawater. Corros. Sci. 37, 1521e1535.

Malik, A.U., Ahmad, S., Andijani, I., Al-Fouzan, S., 1999. Corrosion behavior of steelsin Gulf seawater environment. Desalination 123, 205e213.

Malik, A.U., Al-Fozan, S.A., 22 October, 1992. Pitting Behavior of Type 316L SS inArabian Gulf Sea Water. Technical Report No. SWCC (RCD).

Manning, P., Duquette, D., Savage, W., 1980. Technical Note: the effect of retainedferrite on localized corrosion in duplex 304L stainless steel. Weld. J. 59,260e262.

Mao, D., 2007. Bayesian Modeling of Pitting Corrosion in Steam Generators.Maureen, W., Lisa, V.A.A., Lorenzo, V.W., 2013. Corrosion-related Accidents in Pe-

troleum Refineries: Lessons Learned from Accidents in EU and OECD Countries.Publications Office of the European Union.

Melchers, R.E., 1994. Pitting Corrosion in Marine Environments: a Review. Depart-ment of Civil Engineering and Surveying, University of Newcastle.

Melchers, R., 1998. Probabilistic modelling of immersion marine corrosion. Struct.Saf. Reliab. 3, 1143e1149.

Melchers, R.E., 1999. Corrosion uncertainty modelling for steel structures. J. Constr.Steel Res. 52, 3e19.

Melchers, R.J.R.E., 2001. Corrosion Tests of Mild Steel in Temperature Seawater.Research Report.217.12.2001. The University of Newcastle, Newcastle Australia.

Melchers, R.E., 2001. Probabilistic models of corrosion for reliability assessment andmaintenance planning. In: 20th International Conference on Offshore Me-chanics and Arctic Engineering, pp. 3e8.

Melchers, R., 2002. Effect of temperature on the marine immersion corrosion ofcarbon steels. Corrosion 58, 768e782.

Melchers, R.E., 2003. Probabilistic models for corrosion in structural reliabilityassessmentdpart 2: models based on mechanics. J. Offshore Mech. Arct. Eng.125, 272e280.

Melchers, R., 2003. Modeling of marine immersion corrosion for mild and low-alloysteels-part 1: phenomenological model. Corrosion 59, 319e334.

Melchers, R.E., 2003. Probabilistic models for corrosion in structural reliabilityassessment e part 1: empirical models. J. Offshore Mech. Arct. Eng. 125,264e271.

Melchers, R., 2003. Modeling of marine immersion corrosion for mild and low-alloysteels e part 2: uncertainty estimation. Corrosion 59, 335e344.

Melchers, R.E., 2003. Probabilistic model for marine corrosion of steel for structural

reliability assessment. J. Struct. Eng. 129, 1484e1493.Melchers, R., 2004. Pitting corrosion of mild steel in marine immersion environ-

ment e part 2: variability of maximum pit depth. Corrosion 60, 937e944.Melchers, R., 2004. Pitting corrosion of mild steel in marine immersion environ-

ment e part 1: maximum pit depth. Corrosion 60, 824e836.Melchers, R.E., 2004. Effect of small compositional changes on marine immersion

corrosion of low alloy steels. Corros. Sci. 46, 1669e1691.Melchers, R., 2005. Statistical characterization of pitting corrosion e part 1: data

analysis. Corrosion 61, 655e664.Melchers, R., 2005. Effect of immersion depth on marine corrosion of mild steel.

Corrosion 61, 895e906.Melchers, R.E., 2005. The effect of corrosion on the structural reliability of steel

offshore structures. Corros. Sci. 47, 2391e2410.Melchers, R., 2005. Effect of nutrient-based water pollution on the corrosion of mild

steel in marine immersion conditions. Corrosion 61, 237e245.Melchers, R., 2005. Statistical characterization of pitting corrosion e part 2: prob-

abilistic modeling for maximum pit depth. Corrosion 61, 766e777.Melchers, R., 2006. Examples of mathematical modelling of long term general

corrosion of structural steels in sea water. Corros. Eng. Sci. Technol. 41, 38e44.Melchers, R.E., 2006. Advances in mathematical probabilistic modelling of the at-

mospheric corrosion of structural steels in ocean environments. In: 3rd Inter-national ASRANet Colloquium.

Melchers, R., 2006. Pitting corrosion of mild steel under marine anaerobic condi-tions e part 1: experimental observations. Corrosion 62, 981e988.

Melchers, R., 2007. Influence of seawater nutrient content on the early immersioncorrosion of mild steel e part 1: empirical observations. Corrosion 63, 318e329.

Melchers, R., 2008. Corrosion Wastage in Aged Structures, Condition Assessment ofAged Structures.

Melchers, R.E., 2008. Extreme value statistics and long-term marine pitting corro-sion of steel. Probab. Eng. Mech. 23, 482e488.

Melchers, R.E., 2008. Development of new applied models for steel corrosion inmarine applications including shipping. Ships Offshore Struct. 3, 135e144.

Melchers, R.E., 2009. Experiments, Science and Intuition in the Development ofModels for the Corrosion of Steel InfrastructurejNOVA. The University ofNewcastle's Digital Repository.

Melchers, R., 2010. Transient early and longer term influence of bacteria on marinecorrosion of steel. Corros. Eng. Sci. Technol. 45, 257e261.

Melchers, R.E., 2012. Modeling and prediction of long-term corrosion of steel inmarine environments. Int. J. Offshore Polar Eng. 22, 257e263.

Melchers, R., 2014. Modelling long term corrosion of steel infrastructure in naturalmarine environments. Underst. Biocorros. Fundam. Appl. 66, 213.

Melchers, R.E., 2014. Microbiological and abiotic processes in modelling longer-term marine corrosion of steel. Bioelectrochemistry 97, 89e96.

Melchers, R., Ahammed, M., 1994. Nonlinear Modelling of Corrosion of Steel inMarine Environments. Department of Civil Engineering and Surveying, Uni-versity of Newcastle.

Melchers, R., Jeffrey, R., 2008. The critical involvement of anaerobic bacterial activityin modelling the corrosion behaviour of mild steel in marine environments.Electrochim. Acta 54, 80e85.

Melchers, R.E., Jeffrey, R., 2008. Probabilistic models for steel corrosion loss andpitting of marine infrastructure. Reliab. Eng. Syst. Saf. 93, 423e432.

Melchers, R., Jeffrey, R., 2008. Modeling of long-term corrosion loss and pitting forchromium-bearing and stainless steels in seawater. Corrosion 64, 143e154.

Melchers, R.E., Jeffrey, R., 2011. Bacteria Have Transient Influences on MarineCorrosion of Steel.

Melchers, R.E., Wells, T., 2006. Models for the anaerobic phases of marine immer-sion corrosion. Corros. Sci. 48, 1791e1811.

Melchers, R., Ahammed, M., Jeffrey, R., Simundic, G., 2010. Statistical characteriza-tion of surfaces of corroded steel plates. Mar. Struct. 23, 274e287.

Mercer, A., Lumbard, E., 1995. Corrosion of mild steel in water. Br. Corros. J. 30,43e55.

Moayed, M.H., Laycock, N., Newman, R., 2003. Dependence of the critical pittingtemperature on surface roughness. Corros. Sci. 45, 1203e1216.

Mohammad, H.M., Hammadi, N.J., Lafta, R.M., 2012. Prediction of pitting corrosioncharacteristics using artificial neural networks. Int. J. Comput. Appl. 60.

Nakai, T., Matsushita, H., Yamamoto, N., 2006. Effect of pitting corrosion on theultimate strength of steel plates subjected to in-plane compression andbending. J. Mar. Sci. Technol. 11, 52e64.

Ne�si�c, S., 2007. Key issues related to modelling of internal corrosion of oil and gaspipelines e a review. Corros. Sci. 49, 4308e4338.

Newman, R., 2001. WR Whitney award lecture: understanding the corrosion ofstainless steel. Corrosion 57, 1030e1041.

Norsok, M., 2004. 501: Surface Preparation and Protective Coating. Norsok StandardM-501, Rev, 5.

Nu~nez, M., 2007. Prevention of Metal Corrosion: New Research. Nova Publishers.Pardo, A., Otero, E., Merino, M., L�opez, M., Utrilla, M., Moreno, F., 2000. Influence of

pH and chloride concentration on the pitting and crevice corrosion behavior ofhigh-alloy stainless steels. Corrosion 56, 411e418.

Paul, S., 2012. Modeling to study the effect of environmental parameters oncorrosion of mild steel in seawater using neural network. Int. Sch. Res. Not.2012.

Pistorius, P., Burstein, G., 1992. Metastable pitting corrosion of stainless steel andthe transition to stability. Philos. Trans. R. Soc. Lond. Ser. A Phys. Eng. Sci. 341,531e559.

Pistorius, P., Burstein, G., 1992. Growth of corrosion pits on stainless steel in chloride


solution containing dilute sulphate. Corros. Sci. 33, 1885e1897.Popoola, L.T., Grema, A.S., Latinwo, G.K., Gutti, B., Balogun, A.S., 2013. Corrosion

problems during oil and gas production and its mitigation. Int. J. Ind. Chem. 4,35.

Power, R.J., Shirokoff, J., 2012. Techniques for in situ corrosion studies of 316Lstainless steel in sulfuric acid solutions. Recent Pat. Corros. Sci. 2, 2e21.

Qin, S., Cui, W., 2003. Effect of corrosion models on the time-dependent reliabilityof steel plated elements. Mar. Struct. 16, 15e34.

Rajani, B., Kleiner, Y., 2001. Comprehensive review of structural deterioration ofwater mains: physically based models. Urban Water 3, 151e164.

Revie, R.W., 2008. Corrosion and Corrosion Control. John Wiley & Sons.Riviere, J., 1973. Auger electron spectroscopy. Contemp. Phys. 14, 513e539.Roberge, P.R., 2008. Corrosion Engineering: Principles and Practice. McGraw-Hill,

New York.Roberge, R., 2013. Forms of corrosion in pictures. In: Welcome to Corrosion Doctors.

Kingston Technical Software, Kingston.Ryan, Mary P., Williams, D.E., Chater, R.J., Hutton, B.M., McPhail, D.S., 2002. Why

stainless steel corrodes. Nature 415 (6873), 770e774.Salh SH.M, 1990. Pitting Corrosion of Carbon Steel in Sodium Molybdate Solutions.

University of Toronto, U.T.O.Salau, M., Esezobor, D., Omotoso, M., 2011. Reliability assessment of offshore jacket

structures in Niger Delta. Pet. Coal 53, 291e301.Scheers, P., 1992. The effects of flow velocity and pH on the corrosion rate of mild

steel in a synthetic minewater. J. South Afr. Inst. Min. Metall. (South Afr.) 92,275e281.

Schiffrin, D., De Sanchez, S., 1985. The effect of pollutants and bacterial microfoulingon the corrosion of copper base alloys in seawater. Corrosion 41, 31e38.

Schiroky, G., Dam, A., Okeremi, A., Speed, C., 2009. Pitting and Crevice Corrosion ofOffshore Stainless Steel Tubing. Oilonline.

Schultze, W., Wekken, C.v.d., 1976. Influence of alloying elements on the marinecorrosion of low alloy steels determined by statistical analysis of publishedliterature data. Br. Corros. J. 11, 18e24.

Schumacher, M., 1979. Seawater Corrosion Handbook.Shibata, T., Takeyama, T., 1977. Stochastic theory of pitting corrosion. Corrosion 33,

243e251.Sidharth, A.A.P., 2009. Effect of pitting corrosion on ultimate strength and buckling

strength of plate-a review. Dig. J. Nanomater. Biostruct. 4, 783e788.Soares, C.G., Garbatov, Y., Zayed, A., Wang, G., 2009. Influence of environmental

factors on corrosion of ship structures in marine atmosphere. Corros. Sci. 51,2014e2026.

Sørensen, P.A., Kiil, S., Dam-Johansen, K., Weinell, C., 2009. Anticorrosive coatings: areview. J. Coat. Technol. Res. 6, 135e176.

Southwell, C., Alexander, A., 1970. Corrosion of metals in tropical waters, structuralferrous metals. Mater. Prot. 9.

Souza, E., Correa, S., Silva, A., Lopes, R., Oliveira, D., 2008. Methodology for digitalradiography simulation using the Monte Carlo code MCNPX for industrial ap-plications. Appl. Radiat. Isot. 66, 587e592.

Stewart, M.G., 2009. Mechanical behaviour of pitting corrosion of flexural and shearreinforcement and its effect on structural reliability of corroding RC beams.Struct. Saf. 31, 19e30.

Stewart, M.G., Al-Harthy, A., 2008. Pitting corrosion and structural reliability ofcorroding RC structures: experimental data and probabilistic analysis. Reliab.Eng. Syst. Saf. 93, 373e382.

Strehblow, H.-H., Marcus, P., 1995. Mechanisms of Pitting Corrosion, CorrosionMechanisms in Theory and Practice, pp. 201e238.

Szklarska-Smialowska, Z., 1986. Pitting Corrosion of Metals. National Association ofCorrosion Engineers.

Thomas, C., Edyvean, R., Brook, R., 1988. Biologically enhanced corrosion fatigue.Biofouling 1, 65e77.

Thompson, G., Furneaux, R., Wood, G., Richardson, J., Goode, J., 1978. Nucleation andGrowth of Porous Anodic Films on Aluminium.

Ting, O.S., Potty, N.S., Shahir Liew, M., 2011. Prediction of corrosion rates in marineand offshore structures. In: National Postgraduate Conference (NPC), 2011. IEEE,pp. 1e6.

Valor, A., Caleyo, F., Alfonso, L., Rivas, D., Hallen, J., 2007. Stochastic modeling ofpitting corrosion: a new model for initiation and growth of multiple corrosionpits. Corros. Sci. 49, 559e579.

Valor, A., Caleyo, F., Rivas, D., Hallen, J., 2010. Stochastic approach to pitting-corrosion-extreme modelling in low-carbon steel. Corros. Sci. 52, 910e915.

Valor, A., Caleyo, F., Alfonso, L., Vel�azquez, J., Hallen, J., 2013. Markov chain modelsfor the stochastic modeling of pitting corrosion. Math. Probl. Eng. 2013.

Vel�azquez, J., Van Der Weide, J., Hern�andez, E., Hern�andez, H.H., 2014. Statisticalmodelling of pitting corrosion: extrapolation of the maximum pit depth-growth. Int. J. Electrochem. Sci. 9, 4129e4143.

Veritas, D.N., 2002. Risk Based Inspection of Offshore Topsides Static MechanicalEquipment.

Veritas, D.N., 2009. Recommended Practice DNV-RP-G 101: Risk Based Inspection ofOffshore Topside Static Mechanical Equipment. Det Norske Veritas, Høvik.

Watts, J.F., 1994. X-ray photoelectron spectroscopy. Vacuum 45, 653e671.Wika, S.F., 2012. Pitting and Crevice Corrosion of Stainless Steel under Offshore

Conditions.Wika, S.F., 2012. Pitting and Crevice Corrosion of Stainless Steel under Offshore

Conditions. Norwegian University of Science and Technology (NTNU), Norwe-gian University.

Wilde, B., 1972. A critical appraisal of some popular laboratory electrochemical testsfor predicting the localized corrosion resistance of stainless alloys in sea water.Corrosion 28, 283e291.

Williams, P., 1985. Secondary ion mass spectrometry. Annu. Rev. Mater. Sci. 15,517e548.

Wilson, A.G., 1967. A statistical theory of spatial distribution models. Transp. Res. 1,253e269.

Wong, K.P., Alkire, R.C., 1990. Local chemistry and growth of single corrosion pits inaluminum. J. Electrochem. Soc. 137, 3010e3015.

Xiong, S., Zhu, Z., Jing, L., 2012. Influence of Cl-ions on the pitting corrosion of boilerwater-wall tube and its principle. Anti-Corros. Methods Mater. 59, 3e9.

Yari, M., 2014. An intro to pipeline corrosion in seawater. Corrosionpedia 2, 1432.http://www.corrosionpedia.com/2/1432/corrosion-101/an-intro-to-pipeline-corrosion-in-seawater.

Younis, A., El-Sabbah, M., Holze, R., 2012. The effect of chloride concentration andpH on pitting corrosion of AA7075 aluminum alloy coated with phenyl-trimethoxysilane. J. Solid State Electrochem. 16, 1033e1040.

Zakowski, K., Narozny, M., Szocinski, M., Darowicki, K., 2014. Influence of watersalinity on corrosion riskdthe case of the southern Baltic Sea coast. Environ.Monit. Assess. 1e9.

Zamaletdinov, I., 2007. Pitting on passive metals. Prot. Met. 43, 470e475.Zamanzadeh, M., Larkin, E., Gobbin, D., 2004. A Re-examination of Failure Analysis

and Root Cause Determination. Pennsylvania, Pittsburgh.Zatkalíkov�a, V., Bukovina, M., �Skorík, V., Petrekov�a, L., 2010. Pitting corrosion of AISI

316Ti stainless steel with polished surface. Mater. Eng. 17, 15.Zaya, P.G.R., 1984. In: . Evaluation of Theiores for the Initial Stages of Pitting

Corrosion. McMaster University, McMaster.Zhu, W., Rose, J., Barshinger, J., Agarwala, V., 1998. Ultrasonic guided wave NDT for

hidden corrosion detection. J. Res. Nondestruct. Eval. 10, 205e225.

modelling of pitting corrosion in marine and offshore steel

Documents