Abstract
Welded Joints long-term integrity is of particular importance to large vessels, such as frigates or bulk carriers. Downtime due to corrosion concerns can be onerous. Ship classification organisations, such as The American Bureau of Shipping (ABS), standardises processes and product approval programmes for materials manufacturers. High-strength steels are of particular interest due to their high strength to cost ratio for large marine vessel construction. The corrosion performance of such steels has been studied, but data on the specific degradation of such metals when welded and for the welds themselves are scarce. The present paper reports marine immersion corrosion pit depth data for several ABS steel grade weld permutations and exposed to different sites for up to 2.5 years. The implications of the presented results are discussed as well as the possible influence of alloying elements and microstructural differences on the underlying corrosion kinetics.
Introduction
Corrosion of naval vessels can cause significant damage requiring expensive repairs to keep within Classification Society requirements [1 3], and in extreme cases could lead to the loss of vessels [4 6]. Generally, ship hulls are given protective coatings both inside and out and are cathodically protected using either sacrificial anodes or impressed current systems. However, coatings are subject to mechanical damage, wear and deterioration, potentially leaving some areas less well protected or even exposed. On the other hand, cathodic protection systems also deteriorate and require periodic maintenance [6 8]. Further, corrosion may develop under protective coatings (blisters) and perhaps advance rapidly while being difficult to detect, even when protective methods largely are in place [2,7,9]. For these reasons, it is imperative for vessel designers and operators to understand the likely corrosion degradation of ship hulls overall and in particular that at welded zones since these are known to be subject to more severe, preferential, corrosive attack [10 12].
Although there is a considerable amount of practical experience [1 3,13], evidence-based guidance for estimating the likely corrosion of vessel hulls is scarce (e.g. [14]), and there are few quantitative data reflecting field experience for general and pitting corrosion of welded areas. As noted in an explanatory note [8], hulls are prone to aggressive pitting unless cathodically protected. Protective coatings are prone to the possibility of localised mechanical damage in port or in operation and also may deteriorate in various ways [2,7]. As a result, impressed current or sacrificial anode systems are almost always used, with, in some cases, the use of impressed current systems favoured for medium and large ships due to the relative ease of monitoring and adjusting current density [9]. However, other operators tend to favour sacrificial anodes due to the simplicity of the system and long experience, particularly but not exclusively in merchant shipping. In both cases, there is limited guidance to owners and operators regarding the expected residual level of corrosion protection in the case that these galvanic protection measures fail are rendered inoperable or function inefficiently. Further, any corrosion rates that are published are generally described in guidance notes (e.g. [2]) in terms of mm/year, implying that the corrosion rate remains constant over time, which has been shown to be not the case [15].
Corrosion-resistant alloys can reduce asset deterioration but at the expense of high material costs. However, passively protected corrosion-resistant alloys (i.e. stainless steels) are prone to localised or pitting corrosion [16] and this may be severe at or near heat-affected zones (HAZs) of welds [17]. Despite their low corrosion resistance, carbon steel and low-alloy steel are often an economical choice of material with or without cathodic protection. However, under less than adequate maintenance regimes, either for protective coatings or of cathodic protection or both, these steels also may suffer considerable corrosion damage, including at welds and within HAZs, typically in the form of pitting corrosion [18]. Pitting corrosion often is of greater concern and interest to ship owners and operators because of the difficulty of detecting pitting by visual or other simple inspection and because of the potential for sudden failure [19]. Initiation and early development of pitting corrosion have been studied extensively [16,20], but there are less data for longer exposures [21]. As is well known, during fusion welding, the thermal cycles produced by the heat source cause physical state changes, metallurgical phase changes and transient thermal stress and metal movement. Thus, the potential effect on pitting corrosion of differences in microstructure between the zones, caused by either non-standardised filler selection, chemical, mechanical or metallurgical variation has been also considered [22,23]. Overall, there is an agreement that pitting corrosion can advance at rates much greater than the rate of general corrosion, with some researchers reporting pit growth rates in some conditions of up to 10 mm/year for short periods, and 1 mm/year for long periods. Such high rates should be of considerable concern to vessel designers using hull plates of 5-10 mm thickness [1]. The average corrosion rates for welds are less well understood.
It has been demonstrated how severe localised corrosion can be considered in the design to produce a sufficiently safe asset with a more reliable long-term service life [18,24,25]. Also, the underlying causes of severe pitting corrosion of low-carbon mild steels located at or near the weld have been considered many times [11,12,18,24–27] and is known to involve some degree of preferential corrosion but some uncertainty remains regarding the specific mechanism involved. Particular studies could be carried out in order to ascertain the level of effect for each of the aforementioned chemical, electrochemical or variation variables; however, herein the focus is long-term unbiased quantitative empirical data collection and interpretation. Thus, to better understand the dominant factors that drive the preferential corrosion effect, this paper reports an investigation of the effect of microstructure and its variations in mild steel welds on the development of corrosion over some 2.5 years as part of a longer term exposure project. The corrosion developed in the HAZ and the weld zone (WZ) as produced as part of the welding process are considered with respect to that for the parent metal zone (PMZ) for several combinations of steels and welds exposed to sea water immersion conditions. The results from the testing programme are presented and the implications of the microstructure of each zone are discussed, as well as the possible influence of the decarburisation layer commonly present on hot rolled mild steel products.
Materials and methods
ABS, American Bureau of Shipping; DNV, Det Norske Veritas (Norway); AWS, American Welding Society.
Note: Chemical analysis provided by an independent third party. Weld filler chemical composition of all samples complies with ABS specifications [1].
Experimental combinations of welded exposure coupons by steel grade and exposure sites and site conditions.
Note: SAW, submerged arc weld by hand; GMAW, gas metal arc weld by semi-automation.
Each of the individual 200 mm long × 50 mm wide × 10 mm thick centrally welded test coupons (Figure 1) was an offcut from a larger 760 mm × 760 mm × 10 mm thick welded metal sheet supplied by DSTG. Notches cut on the sides of the coupons were used for unique coupon identification, and the centre drilled hole for affixing to the test-racks.
Example coupon before exposure (far left), immediately after 2-year recovery (middle × 2), and after acid clean (far right).
The same combinations of steels and welds were exposed at three different exposure sites. These were chosen so that, apart from annual average dissolved inorganic nitrogen (DIN) concentrations and average sea water temperatures, all other environmental variables could be considered essentially the same, noting that DIN is relevant for microbiologically influenced corrosion [28]. The sites used were Taylors Beach NSW (minimal DIN levels and temperate waters), Port of Newcastle NSW (slightly elevated DIN and temperate waters) and Darwin Marina NT (minimal DIN levels and tropical waters). The water chemistry for each site is presented in Table 3. It is noted that storm events may cause some short-term changes in sea water chemistry; however, these are likely buffered [32]. Annual averages for temperature and DIN concentrations are used throughout. Figure 2 shows an example exposure condition, in this case, Darwin Marina. Retrievals occurred every 6 months from each site, with a total of 8 coupons per site being retrieved for analysis. One complete set of coupons was arranged in a bucket, one for each retrieval, to facilitate easy and unambiguous recovery (one bucket per recovery at each site), thereby ensuring all combinations were considered. Figure 3 shows an example. Following recovery, the coupons were cleaned to remove most marine growth. After reaching the laboratory, they were cleaned using standard practice [22,29,33]. After drying, the coupons were photographed and examined for corrosion morphology and pitting.
Example exposure site conditions (Darwin Marina). Example set of coupons (Taylors Beach) organised in buckets for protection and easy retrieval. Sea water chemistry at each of the different exposure sites. Sources of information: Melchers 2012, 2014, Rosen 2019, NT Gov Marine Water Report (2011).
Corrosion loss was not considered as there were no measurable coupons without the presence of weld material. Only pit depths were considered, for each zone. The pit depths were obtained from repeated and multiple readings based on the microscope focusing on the bottom and at the top of the pit, as is the standard procedure. The use of multiple focusing was to eliminate as much as possible operator error, while multiple pits (typically 20) were measured to produce a reasonable population from which the select the 5 deepest pits as well as ensuring reasonable confidence that the deepest pit had been measured. The latter was also cross-checked by scanning the surface of each zone to locate the deeper regions. As there will inevitably be some overall corrosion loss over each of the weld and PMZs, the pit depths obtained must be considered ‘relative’ pit depths, even though from earlier work, it is known that for extended exposures, there is little difference between absolute and relative pit depths because the general corrosion loss is relatively small compared to pit depth [18,21,24].
Results
Figures 4–12 summarise the average and the maximum pit depths measured in each zone (PMZ, HAZ and WZ) for each combination of steels (e.g. DH36–DH55) and weld types [submerged arc weld (SAW) and gas metal arc (GMAW)] at each exposure period. Also shown are the maximum values of the depth of the 5 deepest pits. Where possible best-fit trends have been drawn through each data set.
Darwin Marina DH36–DH55 pit depth (mm) over time, with average and maximum pit depth (mm) trends also shown. PMZ shown on the top, HAZ centre and WZ shown on the bottom. Darwin Marina DH36–HY80 pit depth (mm) over time, with average and maximum pit depth (mm) trends also shown. PMZ shown on the top, HAZ centre and WZ shown on the bottom. Darwin Marina EH36–EH36 pit depth (mm) over time, with average and maximum pit depth (mm) trends also shown. PMZ shown on the top, HAZ centre and WZ shown on the bottom. Port of Newcastle DH36–DH55 pit depth (mm) over time, with average and maximum pit depth (mm) trends also shown. PMZ shown on the top, HAZ centre and WZ shown on the bottom. Port of Newcastle DH36–HY80 pit depth (mm) over time, with average and maximum pit depth (mm) trends also shown. PMZ shown on the top, HAZ centre and WZ shown on the bottom. Port of Newcastle EH36–EH36 pit depth (mm) over time, with average and maximum pit depth (mm) trends also shown. PMZ shown on the top, HAZ centre and WZ shown on the bottom. Taylors Beach DH36–DH55 pit depth (mm) over time, with average and maximum pit depth (mm) trends also shown. PMZ shown on the top, HAZ centre and WZ shown on the bottom. Taylors Beach DH36–HY80 pit depth (mm) over time, with average and maximum pit depth (mm) trends also shown. PMZ shown on the top, HAZ centre and WZ shown on the bottom. Taylors Beach EH36–EH36 pit depth (mm) over time, with average and maximum pit depth (mm) trends also shown. PMZ shown on the top, HAZ centre and WZ shown on the bottom.
For assigning the pit depths to the various zones, the WZs were defined as the weld itself (∼20 mm wide), the HAZ was defined as the 20 mm band on each side of the weld, and the PMZ was defined as the remaining exposed surface area, excluding all edge effects. In Figures 4–12, the grey scale lines are trends added by hand for the SAW with the solid trend used for the maxima, and the hashed trends used for the average pit depths. The light grey trend lines are similarly for the semi-automated GMAWs.
The pit depths in Figures 4–12 are those as measured, without adjustment for average mass (corrosion) loss as is sometimes done. It is noted that in some cases, the pit depths that were measured for later exposures were less than the pit depths for shorter exposures. As is discussed further below, this obviously cannot occur in continuous exposures as pitting is a monotonic, irreversible process [15,16,21,28]. Almost certainly what is seen is an artifice, resulting from the use of relatively small coupons and relatively short lengths of welds.
Discussion
As can be seen in Figures 4–12, the results for pit depth development are not always as might be expected and it is difficult to obtain clear correlations between pit depths and the variables that may be important. For example, irrespective of weld type, most pitting damage occurs in the HAZ or in the base metal. Also, deeper and broader pits were observed on the lower strength metals, with the higher strength metals showing, subjectively, less corrosion damage overall. Where the welded metals were the same (e.g. EH36 welded to EH36), the overall location of the corrosion damage was judged, subjectively, to be approximately the same on both sides of the weld. Figure 13 shows examples of very deep pits observed in lower strength grade steels, irrespective of weld type and the associated welded steel, and irrespective of the exposure period. Chemical composition, of course, is directly related to strength. For this reason, careful consideration was given to the selection of appropriate weld fillers complying with the standard recommendations [1].
Example of very deep pits observed in lower strength grade steels, irrespective of weld type, associated welded steel. DH36 and EH36 left side of images.
The overall pitting corrosion at the Port of Newcastle site was considered, again subjectively from inspections of cleaned coupon and weld surfaces, to be higher than those observed at Darwin Marina, despite the fact that the pit depths at the Darwin site often were comparatively deeper. Most likely the latter is the result of the higher average sea water temperature at Darwin. Further, the pit depths observed at Taylors Beach are noticeably less deep compared with those observed at the other sites. This was the case for all exposure periods. This appears to be directly correlated with the low DIN concentration at this site, consistent with observations elsewhere [21,28].
The potential effect on pitting corrosion of differences in microstructure between the zones also was considered. As is well known, during fusion welding, the thermal cycles produced by the heat source cause physical state changes, metallurgical phase changes, and transient thermal stress and metal movement [22,23]. The intense heat from the welding process generates a molten pool of weld metal (WM). Some of the heat is conducted into the base metal (BM), resulting in a temperature gradient. Three distinct metallurgical zones are formed in a weldment, comprising the WM, the HAZ and the BM. The peak temperature and the subsequent cooling rates determine the HAZ microstructures [22]. The various phases present in the HAZ of a weld in mild steel typically are a mix of ferrite, pearlite and iron carbides. Figure 14 summarises these [22,23]. The bulk microstructure of mild steel in the normalised condition consists of ferrite and pearlite. Ferrite is relatively pure iron, while pearlite consists of alternating lamellae of ferrite and iron carbide (Fe3C). It is conceivable that these two phases may form a galvanic cell interaction that can drive corrosion effects; however, there is also potential for non-equilibrium phases such as martensite and bainite to form if the cooling rates from above the critical temperature are sufficiently fast. Martensite has a higher energy due to lattice distortion [22,23], and, therefore, may be more susceptible to corrosion than an equilibrium phase such as ferrite or pearlite. Moreover, corrosion also could be exacerbated by the presence of non-metallic inclusions, which in turn could further facilitate the formation of a localised galvanic cell. Residual stresses are also known to be corrosion initiation sites within the weld and HAZ [23,29].
Schematic of HAZ distinct zones formed dependent on the peak temperature reached during the welding process based on [23]. The weight % carbon temperature phase diagram is also correlated for convenience.
Also from Figures 4–12, relative pit depth can be observed from the data plotted in Figures 4–12. Deeper pits can be observed to occur consistently on the relative lower strength steel grades (DH36 and EH36). Close inspection under the microscope confirmed that for the lower grade metals, the pits tend to be broad pits, irrespective of weld technique used. On the other hand, the higher strength steels tended to show pits that were comparatively much narrower, well defined and also shallower. These results are consistent with several previous findings [18,23,25,30,31,34] which showed from detailed microstructure examination that it is plausible that the potential for grain boundary carbide film formation can have a noticeable influence on HAZ corrosion losses. Exactly why the presence of grain boundary carbides may cause increased corrosion rates is, at this stage, not clear, although it is noted that the presence of carbide precipitates is known to lead to preferential attack when metallographic etching is carried out. This is consistent with the notion of ‘etching’ a metallographic sample causing the grain boundaries to be more aggressively attacked by the etchant in order to reveal the microstructure.
Correlations between general corrosion and temperature and DIN concentrations have been demonstrated earlier, based on extensive results from many different marine immersion exposure sites [28]. Similar trends were observed also for the results for pit depth obtained in the present study that included welded metals. The implication of this observation is that the underlying corrosion mechanisms as functions of water temperature and DIN are not affected by the presence of the welds. However, as previously noted, it is likely that microstructural changes due to weld heat gradients do lead to changes in corrosion morphology, although the precise relationships remain to be further investigated.
The fact that for several coupons the pit depths at 1.5 year exposure were greater compared to those at 2 year exposure was noted above, where it was suggested that the relatively small coupon size might be a factor relative to the variability generally noted for pit depths, and often attributed largely to the variability of inclusions and other inhomogeneities in steels [29,31,34]. The pit morphology at the Darwin and Taylors Beach sites was observed to be very similar, suggesting that the corrosion depth trend with time between the sites is similar, but perhaps with rather different overall pit depths, most likely the direct result of the higher average sea water temperatures at the Darwin site. On the other hand, the coupons recovered from the Port of Newcastle site showed somewhat different morphologies, potentially the result of microbiologically influenced corrosion effects under the moderately elevated DIN concentration at the Port of Newcastle site compared with the other sites. The observations of corrosion and pitting given herein potentially are important for the shipbuilding industry as such corrosion potentially is common to weld hulls, internal stiffeners and other structural elements in the construction of steel vessels.
Conclusions
For the exposure site conditions and steel grades tested, the welding technique used (i.e. SAW or GMAW) appears to have little influence on observed pit depths for sea water exposures up to 2.5 years. For the weld zones (WZ, HAZ) elevated temperature and DIN concentrations lead to increased pit depth development with time, generally consistent with that observed previously for unwelded steels. The lower strength steel grades tested all showed more severe and broader pit depth damage compared to their higher strength grade counterparts. Pitting morphology appeared to be influenced by both grain size and microstructure, although this remains to be further investigated. The higher yield steels, known for their relatively finer microstructure, presented roughly narrower and well-defined circular pits, compared to the lower yield steels relative coarse grain samples which presented proportionally broader pits. Overall pit depths were higher in the HAZ, followed by PMZ and WZ.
Footnotes
Acknowledgements
The authors acknowledge the contribution of DST Group in the provision of sponsorship and sample materials. The authors also acknowledge Simon Krismer for his assistance in preparation, cleaning and measurement of the first set of retrieved coupons, and the invaluable support of Bruce Cartwright retrieving the samples from Darwin, and Dr Robert Petersen retrieving from Port of Newcastle and Taylors Beach. Jeremy Rosen acknowledges the support of the University of Newcastle in the provision of a Defence Research Scholarship.
Disclosure statement
No potential conflict of interest was reported by the author(s).