Atmospheric Corrosion of Steel on the Australian Pacific Central Coast ^†

Abstract

Comprehensive data are presented for corrosion losses of mild steel exposed for up to 5 years, all obtained from exposing steel coupons at one specific severe marine exposure site on the Pacific Ocean coast. The test programme considered the effects of duration of exposure, inclination, orientation, height, shielding, and coupon variability, using multiple, nominally identical mild steel coupons, all under a single local climatic regime. Such a controlled, consistent, natural environment permits unique, valid comparison of the various influences, both for short-term and longer-term exposures, unlike previous tests of some parameters conducted in the short term at disparate sites. In contrast to coupons exposed only on one side, boldly exposed double-sided coupons corroded severely within 3 years. The effects on corrosion behaviour between individual coupons exposed at different heights and vertical continuous single strips of steel are described. Also reported are corrosion losses for continuous strips and for a series of coupons oriented in different directions. Observations of variability in corrosion losses for nominally identically exposed steel coupons are reported. The effect on corrosion losses with continued exposure to 5 years is reported and compared with information available in the literature.

1. Introduction

Interest in marine corrosion of steel has increased significantly since the late 1900s, growing from a much earlier and longer main focus on the effect of NOx on steel corrosion [1], following the aggressive corrosion observed for defence equipment in the Tropics during the Second World War [2], and early studies, such as those by Ambler and Bain [3], in the harsh tropical atmosphere of Nigeria. The development of interest in the marine corrosion of steel can be seen in major texts [4,5] and major reviews covering basic atmospheric corrosion mechanisms, the effect of metal composition, the effect of external influences, and the development of atmospheric corrosion in short- and long-term exposures [6,7,8].

There is now a substantial body of data and understanding of the effects of steel composition [9], salt contribution [10], and the effects of different exposure sites, including at sea, although often only for shorter-term (1, 2 years) exposures [11,12,13,14,15], as well as the influence of inclination [16,17,18,19,20,21]. Consideration has also been given to the empirical modelling of corrosion loss as a function of exposure time and location [22,23], perhaps with some simplifications [24], and also to corrosion development derived from immersion in seawater [25,26].

Despite this progress, considerable gaps in understanding remain [8], particularly in the basic mechanisms involved, the influence of environmental variables for different orientations of steel surfaces, and the understanding, modelling, and prediction of expected longer-term corrosion losses, such as longer than the common approach of considering only one- or two-year exposures as sufficient for extrapolation to much longer periods [6]. The present paper describes observations and results derived from a 5-year exposure programme employing mild steel at one specific severe marine site. The aspects considered include the effects of duration of exposure, inclination, orientation, height, shielding, and coupon variability, using multiple, nominally identical mild steel coupons.

The location used offers a controlled, consistent, natural environment, thereby permitting valid comparison of the various influences for periods of exposure up to 5 years. There are some other observations of atmospheric corrosion in marine environments, but, as noted, few are for more than one or two years, or in combination with other observations. The programme described below considers four different environmental factors relative to the orientation and placement of corrosion coupons, namely: cardinal direction, inclination, height, and season of exposure, uniquely all at the same site. Furthermore, it considers, for the first time, the variability between coupons exposed to nominally similar environmental conditions. It also updates and extends some results previously reported [27], noting that the results from the programme have so far appeared only in isolated conference proceedings. Throughout, it should be noted that the tests were conducted in the Southern Hemisphere.

2. Experimental Programme

2.1. Test Site

Since the overall aim of the experimental programme was to obtain estimates of the relationships between corrosion loss of mild steel as a function of time and exposure conditions at a severe marine exposure location, the marine atmospheric exposure site at Belmont was chosen. It is known to be highly corrosive and subject to aggressive seasonal marine weather influences [28]. It also is recognized as such in the Australian Standards [29]. The test site is within a secure area on the southeast boundary of the Hunter Water sewage treatment works at Belmont Beach, NSW, approximately 16 km south of Newcastle, NSW (33°02′ S, 152°40′ E, elevation 8 m). The location of the site is such that odours do not normally emanate from the sewage treatment works. Any emissions are rare and originate more than 100 m downwind from the exposure site. Overall, the site is considered free of nitrous or sulphurous oxides or other forms of air pollution. This has been confirmed by periodic testing.

The annual average chloride deposition rate at the site ranges from 250 to 350 mg/(m²·d) as measured in compliance with ISO 9225:2012 [30], but can vary seasonally from 40 to 675 mg/(m²·d). The average annual rainfall is 1150 mm, and the average time-of-wetness is 5650 h/y (66%). Both of these vary with orientation, seasons, and prevailing winds.

The site used for exposure testing was located approximately 200 m inland from the average high-water coastline. The specific site almost always is subject to aerosols from breaking surf. To the east, the site is sheltered to some extent by a line of low sand hills. It is subject predominantly to south-easterly Pacific Ocean spindrift, facilitated by a gap in the sand dunes.

2.2. Test Coupons

The test coupons were all rectangular, 100 × 50 × 3 mm in size. They all were guillotined from the same sheet of mild steel. The steel was analysed using Atomic Emission Spectrometry. The composition is given in

Table 1. Chemical composition of mild steel coupons (wt%).

C	Mn	P	S	Si	Ni	Cr	Mo	V	Al	Cu	Fe
0.12	0.43	0.021	0.005	0.01	0.01	0.03	0.01	0.002	0.045	0.01	Balance

. Coupons were individually identified by a unique numbering system using a small-hole drilling sequence. Immediately prior to exposure, the coupons were cleaned in 16% inhibited HCl, rinsed consecutively in deionised water, alcohol, and acetone, dried, and weighed to the nearest 0.1 mg, all in compliance with the usual atmospheric corrosion testing standards [4,30].

After recovery at various points in time, the recovered coupons were cleaned and weighed using the same cleaning procedure. The mass losses were determined as appropriate for either one- or two-sided exposures. One-sided exposures were achieved by application of a highly adherent polymer–bituminous coating (STOPAQ) to one side of a coupon. In the following, the reported corrosion losses are the average results for the number of coupons recovered at each point in time, in each case derived from mass losses, using conventional procedures [4,5,6,7,8]. Of course, mass loss covers both general corrosion and localized corrosion, such as pitting. Again, for atmospheric corrosion, mass loss and average corrosion loss are the standard measures [4,5,6,7,8].

3. Testing Programme

3.1. General

In all cases, the coupons were attached to a support frame using a nylon bolt, with the coupon held sufficiently far away from the frame by a 15 mm plastic spacer. Figure 1 shows the octagonal drum used for the inclination trial, the mast used for height evaluation and the panel used for the E–W cardinal directions. Figure 2 shows the rack used for general loss determination and the arrangement for N–S exposure. This corresponds to the generally accepted procedure for assessing corrosion loss at a site [4,8,18].

3.2. Effect of Cardinal Direction

The effect of cardinal direction was determined using two methods, namely exposing two sets on both sides and four sets on one side only. For the latter, one side of the coupons was coated as noted above. For the coupons exposed on both sides, one set was deployed facing north–south and the other set facing east–west. The coupons exposed to one direction only were attached to a plastic board (Figure 1 and Figure 2), with the reverse face coated as noted above to ensure that only the exposed face would be able to corrode.

3.3. Effect of Inclination

An octagonal-shaped drum was fabricated and installed east–west to allow the placement of coupons at varying degrees of inclination (Figure 1). In this trial, all of the coupons had their back faces coated with rubber-based mastic so that only the outer face corroded. They were fixed less than 15 mm from the backing frame. The coupons were placed as follows: upper horizontal (0°), 45° N, vertical facing north (90°), 135° N, facing horizontally down (180°), 135° S (225°), vertically facing south (270°), and 45° S (315°). Temperature and time-of-wetness data loggers were installed on this rig at the upper, lower, east-, and west-facing panels (Figure 1). Previous research in tidal environments has shown there to be a difference in corrosion loss between individual isolated coupons and continuous strips [31]. To further investigate this phenomenon, a continuous mild steel strip (50 × 3 mm cross-section, with as-received surface finish) was ‘wrapped’ around the drum and kept from contact with it by means of 100 mm nylon nuts, bolts, and washers. The ends of the strip were welded to create a continuous octagon that sat 100 mm from the outside of the drum.

3.4. Effect of Height

To determine the effect of coupon exposure location with respect to distance above ground level, a 6 m mast was erected with coupons placed at heights above the ground of 0.1 m, 0.5 m, 1.0 m, 2.0 m, 3.0 m, 4.0 m, 5.0 m, and 6.0 m. In addition to the coupons, three 6 m × 50 mm × 3 mm strips of mild steel (with as-received surfaces) were deployed in the vertical direction alongside, but electrically separated from, the mast to determine if there are differences in corrosion losses for similar exposure periods between continuous steel strips and isolated coupons in the same orientation (vertical) and at the same exposure location (Figure 1). The strips were secured only at 1.0 m intervals.

3.5. Effect of Season of First Exposure

The main testing programme commenced at the beginning of summer (December). To evaluate the effect of the season of first exposure, additional sets of coupons were deployed at the start of the following autumn (April), winter (June), and spring (September). This also allowed coupon recoveries to occur at three-month intervals for each of the seasonal trials.

3.6. Variability of Corrosion

Various previous trials have indicated that variability in corrosion loss for coupons increases with time of exposure, although, for atmospheric corrosion, no data appear to be available. To obtain better understanding of this phenomenon, nine nominally identical coupons were exposed with exposed faces facing north–south. These were recovered at six months, twelve months, and annually after that for five years. This test regime was considered likely to yield sufficient information for statistical analysis purposes.

4. Results and Comparisons

4.1. General

In the following, because of the broad scope of the test programme, experimental results are reported together with comments about relevant previous findings in the corrosion literature. Overall aspects are considered in the Discussion.

Although the steel coupons were some 3 mm thick, for a number of exposure conditions with coupons exposed on both sides, it was found that they had corroded through completely within 4 years, leaving only those that were protected on the unexposed face. This was attributed to the aggressiveness of the site. For these cases, it is possible to report comparable results only for the first 3 years of exposure.

Apart from the use of coupons oriented at 45 degrees or at 30 degrees to the horizontal [30], the generally accepted procedure for assessing corrosion loss at a site is with vertically orientated, N–S-facing coupons exposed for one year. This procedure was used also for the present study. For 3 years of exposure, the results for average corrosion loss are shown in Figure 3. For comparison, the rainfall recorded during the exposure periods between coupon recoveries is also shown on Figure 3. Since the coupons were exposed at the start of summer, these intervals correspond to 0–0.2 y, 1.0–1.25 y, and 2.0−2.25 years of exposure. The higher rainfalls in the summer periods correspond approximately to the periods of accelerated corrosion. This is consistent with the literature, since rainfall is not the only factor to influence corrosion, with a major role for layers of rust that build up with increasing corrosion (apart from exfoliation in the longer term) holding some of the moisture in the inner voids, thereby providing the electrolyte necessary for further corrosion [8].

4.2. Effect of Cardinal Direction

Duplicate coupons were recovered and processed from each of the sets facing one direction only (Figure 2). After 11 months, the polymer–bituminous coating on the back surface was still in good condition and showed minimal sign of deterioration, indicating that corrosion had occurred only on the exposed front face. After 18 months, the polymer–bituminous coating had started to break down, allowing some, but slight, corrosion on the rear face. At 30 months, the coating had noticeably deteriorated, resulting in some pitting of the previously protected rear face. Coupons recovered after 3 years had moderate pitting on the rear face, particularly around the bolt hole.

Corrosion losses were calculated for the exposed front face only, without allowances being made for any loss on the rear face. Although the error is not large, it follows that, strictly, the losses reported for this part of the experimental programme are valid for comparative purposes only. The degradation of the backing and consequent increase in pitting on the previously unexposed face also occurred for those coupons exposed in the inclination section of the exposure programme. The corrosion losses for these (one-sided) coupons are shown in Figure 4 by the side of their exposed surface.

For the coupons with both sides exposed from the N–S and E–W racks, the (conventional one-sided) corrosion losses were determined for each of the two sides. They are reported in Figure 4 as N/S and E/W coupons. The results of directional corrosion are shown in Figure 4.

It should be noted that the results shown for the directional coupons are an average of two coupons for the first 3 years, and individual readings thereafter. The N–S readings for the 11- and 24-month exposure periods were obtained as the average losses for nine recovered coupons. Note that after 2.5 years, the coupons exposed on both sides (N–S and E–W) had completely corroded, leaving essentially no remaining steel.

The results for this particular test condition (cardinal direction) show that in the early recoveries (0−20 months), there was relatively little difference in mass loss between pairs of coupons as recovered at any one time from any orientation. However, at 20 months of exposure, the north-facing coupons had corroded 261 μm, which is almost three times as much as that of the west-facing coupons (106 μm) and almost twice as much as that of the south- and east-facing coupons (161 μm and 177 μm, respectively). This differentiation continued, with, at 30 months of exposure, the corrosion loss of the north-facing coupons being up to 50% greater than that for the coupons facing in any of the other directions. After 3 years of exposure, the corrosion losses on the north- and west-facing coupons were similar but greater than the corrosion losses for the south- and east-facing coupons.

These results should be interpreted, noting that the prevailing wind is from the ocean and is predominantly southeast. Thus, it is highly likely that the correspondingly exposed coupons exhibit the greatest loss. On the other hand, the coupons on the lee side of the dominant wind direction were found to show corrosion losses that were more than 30–40% greater than those on the windward side. This may be attributed to the shielding effect retaining a longer ‘time of wetness’.

Overall, however, the most surprising result is that the sum of the corrosion losses for the single-sided coupons, from two opposite directions, was significantly less than that from a coupon exposed on both sides and with the same orientation. For example, after 1 year, the average loss of the east- plus west-facing-only coupons was 100–130 μm. It would be expected that the annual loss for a coupon exposed on both the east and west faces would be about the sum of the losses from the component sides, i.e., about 230 μm. However, the average corrosion loss from the E–W coupons was nearly 50% higher at 340 μm. This apparent disparity was also observed at 20 months, with the typical one-directional loss being 120–250 μm, but for the (two-side-exposed) coupons, facing both in the east and the west directions, the corrosion loss was 860–1000 μm. This is more than double the sum of the combined losses for the individual single-side-exposed coupons.

A similar trend was noted for exposures at 30 months, with the one-sided losses ranging from 350–500 μm. This would have led to an expected combined loss of 800–1000 μm for coupons exposed on both sides. However, at that stage, the coupons, initially 3 mm (3000 μm) thick, had corroded through completely. This indicates that the corrosion loss for these was greater than 1500 μm.

It may be postulated that the explanation for these observations is associated with wind turbulence and marine aerosol size. As well known, aerosol particle size varies from the very small Aitken nuclei (<0.05 μm) to larger coarse or falling particles (>10 μm). The larger particles stay in the atmosphere for only a few hundred metres from their source, whereas the smaller particles may stay airborne for hundreds of kilometres [32]. This differentiation effect also may have influenced the results shown in Figure 4.

Another possibility is that the blank wall shown in Figure 2 used to hold the coupons acted as a shield or boundary condition, forcing most of the air and its accompanying aerosols around the structure. This would create a sudden drop in wind velocity. In turn, this will likely cause the larger components of the saline aerosols to “drop out” of the wind stream before striking the surface of the coupons, thereby leaving the smaller aerosol fractions to deposit on the exposed surface. Because of the aerodynamics on the lee side of the wall, most of the air (including the accompanying larger aerosols) will be swept past the structure and thus are unlikely to have had any further influence on the coupons on the unexposed face.

In contrast, the coupons boldly exposed on both sides would have been subject to direct impingement of all constituent components of marine aerosols, including from all directions as well as from those smaller particles swept around the lee side of the coupon by aerodynamic effects. This suggests that corrosion losses are influenced by the size as well as orientation of any shielding wall or structure and also by the distance of the steel surface from the edge of such a structure or shield. Clearly, this is a matter for further investigation, noting that Ambler and Bain [3] anticipated such effects.

4.3. Effect of Inclination—Coupons

The effect of coupon inclination has been reported previously [16,17,18,19,20]. The experimental investigations considered coupons inclined up to 180° (from horizontal, facing vertically upwards, through vertical and back to horizontal facing down). These were primarily short exposure periods (15 days). In contrast, the programme described below involved complete 360° exposures and continued over five years.

Initially, all the coupons used for the inclination experiment were coated on a single side (as noted above) and thus exposed only on one side. Duplicate coupons were recovered after 1, 1.8, 2.5, 3, 4.5, and 5 years of exposure, cleaned and weighed as described before. It was observed that some coupons had completely corroded at 5 years of exposure. The average mass losses so obtained are shown in Figure 5 and Figure 6. Figure 5 gives a directional perspective looking west. Figure 6 gives a perspective of corrosion loss at each inclination with time.

A number of unexpected trends can be observed in Figure 5 and Figure 6. One is that there is not a smooth circumferential transition from one point to the next, a trend that can be seen to have become more apparent with longer exposure periods.

Figure 5 shows that lower amounts of corrosion consistently occurred on the vertical south-facing coupons. This is perhaps unexpected as these coupons would have received much less solar radiation and hence less drying than those coupons on the opposite side of the exposure drum (see Figure 1, central). These coupons also directly faced the predominant offshore winds. Further, after 2.5 and 3 years of exposure, the points on each side of the lowest amount of corrosion loss show the highest amount of corrosion loss.

As shown in Figure 6, after one-year exposure, the highest corrosion loss (~550 μm) was recorded on the skyward-facing panel. These results are similar to those reported by Vera et al. [20], who observed a higher corrosion loss for steel samples orientated horizontally compared to those orientated vertically. They concluded that corrosion loss was more a function of TOW rather than deposited pollution. Similarly, Sawant and Venugopal [17], in a 15-day exposure experiment, reported a 400% increase for coupons facing skyward compared to those facing downwards. The difference in corrosion loss was attributed directly to chloride deposition, although TOW was not measured. Bullard et al. [19] found that mass loss tended to be greater on the groundward-facing panels (180°) than on the skyward panels (0°), particularly at distances 500 m to 5000 m from the ocean. They also reported that closer to the coast there was little difference between the two coupon-orientation directions.

Again, as shown in Figure 6, after 20 months, the highest corrosion loss occurred both for the skyward-facing coupons (880 μm) and for the 225° coupons (840 μm). The 225°-facing coupons continued to corrode the most from this time onward, and after 3 years had lost 2100 μm, almost four times that of the adjacent vertical south-facing coupons (550 μm).

Overall, the radial trending is as might be expected. After one year of exposure, the (three) coupons exposed to the greatest amount of solar radiation (i.e., those at 0°, 45°, and 90°) had corroded the least, presumably as a result of greater evaporation of electrolyte. The two coupons adjacent to these (i.e., those at 315° and 135°) showed slightly higher corrosion losses, while the coupon least exposed to solar radiation (i.e., at 225°) showed the greatest amount of corrosion loss. The results for these six inclinations are, intuitively, the most likely. However, the corrosion losses for the coupons facing horizontal-down (180°) and vertical-south (270°) are more difficult to explain. It is possible that TOW (Time of Wetness) played a role.

During the experiment, the time of wetness (TOW) was monitored during the first year, but only at the four major positions (top, north, lower, south). The results have been reported previously [27]. This showed that throughout the first 2 years of the experimental programme, the TOW was significantly higher on the horizontal surfaces (75% for upper and 82% for lower) than on the vertical surfaces (42% for N and 51% for S). Observations at various times during the trial showed that almost every day during the summer months, the vertical faces dried out completely but that during the cooler months of April to July, the upper face stayed continually wet compared to the lower surface, which partially dried out to a 40–60% TOW during the day. These observations would suggest that downward-facing coupons would corrode the most; however, this was not observed consistently throughout the experiment. In this context, it is noted that Morals et al. [21] measured the TOW at some 35 atmospheric corrosion sites on the Canary Islands and ranked them by exposure category, from τ₁ (<10 h/y RH > 80%) to τ₅ (>5500 h/y RH > 80%). The τ value was used as parameter to rank the 35 sites. However, no correlation was apparent between TOW and corrosivity. This is consistent with the observations for the Belmont site.

Coupon surface temperature data recorded during the first year of exposure for the four major inclinations indicated an average of less than one degree Celsius (°C) between all four directions. The results showed a slight correlation between the amount of sunshine received on each face and corrosion loss. The upper surface recorded both the maximum and minimum temperatures, whilst the lower sheltered surface had a slightly higher minimum.

4.4. Effect of Inclination—Continuous Strip

Results from the effect of height (see below) and another testing programme relating to marine immersion have shown that isolated coupons and continuous strips tend to show significantly different corrosion profiles along the lengthwise direction of the strip compared to a string of coupons with the same general orientation [31]. To investigate this further, a flat continuous 50 × 3 mm steel strip was bent around the drum such as to have about 400 mm strip length per side. The corrosion of this strip was compared with the corrosion losses of a parallel string of steel coupons, with the coupons located on each side of the octagonal drum. After 9 months continuous exposure, the strip was recovered, cut into 100 mm lengths, 4 for each of the eight inclined sides of the drum (i.e., a total of 32 segments). These were cleaned and weighed as for all other coupons, as described. The mass loss of the continuous strip and those of the accompanying isolated coupons are shown in Figure 7.

4.5. Effect of Height

For each of the heights noted above, coupons were recovered, in duplicate, after 6, 12, and 24 months of exposure. Also, the 6-m-long 50 mm × 3.0 mm strips exposed at the same location were recovered. The strips were cut into precise 100 mm segments, cleaned, and weighed similarly to the coupons, again as described above. Corrosion losses for the sets of data for individual recoveries are shown in Figure 8, Figure 9 and Figure 10. The combined results are shown in Figure 11.

A number of observations can be made for the plots of Figure 8, Figure 9, Figure 10 and Figure 11. For all three exposure periods, the corrosion loss of the coupons increased by a factor of at least three with elevation, from 0.1 m height to 2.0 m height, above which the corrosion losses remained similar or decreased slightly. Similar results have been observed by others in more tropical environments [2,33,34]. Overall, this effect is well-known and documented, and attributed to the fact that wind speed increases with height, with the result that more ‘kilometres of wind’ pass over the higher coupons, thereby assumed to deposit both more moisture and the effect of chlorides. Thus, the increase of corrosion loss with height is probably due to wind turbulence with its subsequent drying effect as well as chloride deposition [6].

For the long steel strips exposed at the same heights, two trends for corrosion loss are of interest, namely (i) the local changes in corrosion loss corresponding to the height of the coupons, and (ii) the corrosion losses for the strips being less than half of those of the coupons. It is noted that the strips were held at the cross-channel brackets of the support frame in an attempt to restrain the strips from excessive flexing under wind conditions. It was anticipated that such flexing could possibly cause delamination of brittle corrosion products, leading to accelerated corrosion. In the experiment, it was noted that when the mast was lowered and the strips removed, the corrosion product consisted of thick layers of rust, similar to those that had formed on the essentially immobile smaller coupons. For the upper 3 m section of the strip, the steel at the hold points was observed to have corroded significantly less than the intermediate sections. In contrast, on the lower 3 m section, the hold points had corroded more than the intermediate sections. It should be noted that at the hold points a 50 mm wide wrapping of plastic tape had been used to ensure insulation from the aluminium brackets. Upon removal of the tape, the area under the tape was only minimally affected by corrosion. In part, this may explain the observations of sections of lower corrosion on the upper 3 m of the strip. A possible explanation for these observations might lie with differential aeration, which is known to cause localised accelerated corrosion on partially covered surfaces [1]. However, if this were the only contributing factor, it would apply to all sections where the strip was restrained. It remains a matter for further investigation.

Overall, the corrosion losses for the coupons were between two to three times those of the continuous strip, even when exposed under nominally identical exposure conditions. Interestingly, similar observations were made for corrosion losses for coupons compared with corrosion losses for continuous strips exposed to tidal sea-waters near Townsville, Australia [35], although with smaller differences.

Many of the coupons from the atmospheric exposures at the Belmont site that showed particularly high corrosion loss were noted, on close examination, to show a tapered profile in the vertical direction with the lower edge being thinner. This is consistent with earlier observations for atmospheric corrosion of coupons in aggressive locations [36]. It may be postulated that this effect is caused by moisture collecting on the lower edge and thus facilitating additional corrosion, but it is a subject open for further investigation.

4.6. Effect of Season of First Exposure

To ascertain the effect of the season of first exposure on the commencement and development of corrosion loss under continued exposure, as noted in Section 3.5, most of the coupons were deployed at the start of summer, with additional sets of coupons deployed at the beginning of each of the following three seasons. Nine coupons were recovered from the main set at the beginning of summer and winter, with duplicate coupons recovered at the start of autumn and spring (Figure 12).

Figure 12 shows clearly that, for all sets of recoveries, corrosion loss is not linear with time but involves a series of steps involving minimal and more rapid corrosion loss. The periods of rapid loss correspond approximately to the summer and autumn seasons. The episodic nature of corrosion has been reported previously [37], where it was noted that there was a general increase in corrosion loss after a rain event. To consider this, Figure 13 shows monthly rainfall figures (extracted from the data held by the Australian Bureau of Meteorology [38]) for the three years corresponding to the period of exposure.

Rainfall in the greater Newcastle region (which includes the area of the Belmont test facility) typically is distributed relatively uniformly throughout the whole year, with the minimum monthly average of 60 mm generally occurring in July, August, and September, and the highest monthly average of 120 mm occurring from February to June. Figure 13 shows that there was some deviation from this overall trend during the period of the exposure trial. Specifically, Figure 13 shows that the rainfall during the two early summer periods was not particularly high, that the winter between was very dry, and that the most recent winter shown had two extremely wet months. This unusual variation has allowed comparison to Figure 12. This shows that corrosion losses are not necessarily associated with rainfall. This is shown more clearly in Figure 14, which shows the increase in corrosion loss between coupon recoveries plotted against the intermediate rainfall. In summary, it is clear from the above results and those in Figure 10 and the TOW analysis for the first year that there is little or no correlation between intermediate corrosion loss and rainfall.

In various standards (e.g., ISO 9223 [30]), it is often the case that the ‘corrosivity’ of a site is taken as the corrosion loss of steel exposed for one year. It is clear from Figure 14 that the season of exposure can influence the annual corrosion loss, both for the first year and for any later time. The data for the present trial show that, in the first year, the annual corrosion loss ranges from 106 μm for the coupons exposed in the autumn to 550 μm for the coupons exposed in spring. In the following 18 months, the corrosion loss for the coupons exposed in autumn was 600 μm, and 1200 μm for those first exposed in the spring. After two years of continuous exposure, most of the coupons had severely corroded and showed some perforation, or had completely corroded away, thereby invalidating any reliable corrosivity reading.

Figure 12 and Figure 15 show that corrosion loss can be quite a nonlinear function. Coupons first exposed during spring produced a corrosion loss trend that is closely linear, with only one “kink” at around one year of exposure (R² = 0.97). The corrosion loss trends for coupons first exposed in winter or in summer show moderate divergence from a linear plot (R² = 0.90 and 0.92, respectively), while those installed during autumn show a trend that diverges more significantly from linearity (R² = 0.85). The non-linear behaviour of atmospheric corrosion for much longer exposure periods has been previously reviewed [39]. Moreover, observations in the tropical atmosphere of India for 14 sites located near the coast and elsewhere [33] have shown corrosion loss patterns remarkably similar to those in Figure 15. This can be also seen in data for the coast of Mauritius [11].

It is clear from these various observations that the corrosion of mild steel under marine atmospheric conditions cannot be considered a linear function of time, nor represented in the longer term by a simple function such as a power–law. This observation has also been expressed in recent papers [6,11,22]. This has implications for the use of terms such as ‘corrosion rate’ and for the meaning to be attached to ‘corrosivity’, which is widely used for comparative purposes but based on one year of exposure [6,12,13,14,40].

4.7. Variability with Identical Conditions

For the sets of nine coupons for variability estimation (Section 3.6), the calculated average losses and the standard deviations are shown in

Table 2. Variability of coupons under identical conditions.

Exposure	6 Months	11 Months	20 Months
Average corrosion loss (μm)	269	327	1003
Standard Deviation	22.8	22.9	28.8

. The variability in appearance of one set of nine coupons, recovered after 20 months’ exposure, is shown in Figure 16. After two years of exposure for similar coupons exposed at the same time, removal of the rust products showed that they had been reduced to consist only of a few grams of steel surrounding the attachment bolt. For longer exposures, the remaining material consisted only of iron oxides. Overall, at each recovery the variability between coupons was generally similar to that shown in Figure 16.

The pattern of corrosion shown in Figure 16 shows that the locations of coupon perforations were similar for all nine coupons. In each case, perforation occurred on the lower, eastern part of the coupon faces. Although not immediately apparent in Figure 16, as for some of the exposures above, all coupons were noted to reduce in thickness in a gentle taper toward the lower edge. The corrosion losses were noted as more severe on the seaward side of all coupons. Further, consistently, all perforations were well away from the edges of the coupons. It may be postulated that these effects are functions of the predominant offshore spindrift; however, it appears that no explanations for these phenomena have appeared in the literature.

4.8. Variability Within Microclimates

As noted, ‘corrosivity’ is sometimes used to characterize the corrosion severity of a site. As also noted, this is the corrosion loss that occurs nominally over a period of one year [29,30]. Typically, this is obtained using coupons exposed vertically at between 1.0 m and 2.0 m above ground level, facing the prevailing wind direction. For the Belmont site, this would use the N–S-facing coupons. From Figure 15, it can be seen that these give a corrosion loss between 300 and 380 μm, based on the corrosion loss in year one, with conventional first exposure in summer–autumn. However, it is evident from the observations noted above that there is a considerable effect of first exposure and of microclimate, as well as of a period of exposure longer than the nominal one year. Figure 17, Figure 18, Figure 19 and Figure 20 show the variability in corrosion losses as affected by period of exposure for each of the following conditions: (a) inclination (inc), (b) height (height), and (c) cardinal direction (N/S, E/W, and N,S,E,W), as noted above. As before, for inclination and cardinal direction, the coupons were coated on one side only and mass loss determined accordingly. The variability of corrosion involved for height, season of exposure, and time-dependence was determined using coupons exposed on both sides.

The above results show that the variability between all coupons located at different heights or at different inclinations may vary by a factor of four, although for coupons under nominally identical conditions, a Coefficient of Variation of 0.2 is typical. No other results for variability appear to have been reported in the literature, apart from immersion corrosion conditions. However, for marine immersion exposure conditions, the Coefficient of Variation of the corrosion of steel coupons under nominally identical conditions is about 0.2 [41].

5. Discussion

Despite considerable differences in corrosion losses depending on season of first exposure, angle of inclination, and exposed surface area, the results given above can be considered both valid corrosion loss results and entirely comparable. All coupons were cut from the same sheet of steel, exposed and recovered on the same days, and processed in an essentially identical manner. They also were all deployed within 10 m of each other. Yet, the corrosion losses between the different orientations and locations within the small test site at Belmont, as used for the present project, can vary by up to 500%. Thus, Figure 17, Figure 18, Figure 19 and Figure 20 emphasise the variability of corrosion losses even within a relatively small geographical location; i.e., for coupons within a few metres of each other. This highlights the importance of defining location, orientation, and other parameters when quantifying the ‘corrosivity’ of a location, as well as for defining the development of corrosion losses. It is clear from the existing corrosion literature that comprehensive values of the relevant parameters have not always been documented as well as, in retrospect, they might have been. Of course, it is recognized that in many locations there are considerable practical difficulties in measuring or even estimating such parameters.

The ‘corrosivity’ notion mentioned above already has practical implications. For example, in the context of the building industry, for an open building with a coated steel roof close to the ocean, as in, for instance, a recreational shelter, the roof sheeting facing north (in the Southern Hemisphere) is more likely to corrode from the lower side than the upper surface. However, as can now clearly be seen in the results presented above, a simplified measure such as ‘corrosivity’ barely is sufficient for assessing the corrosion likely to be experienced over an extended exposure period for major infrastructure. For such assessments, the results given above can be considered as starting points for more detailed, quantitative investigations.

Overall, the result presented herein can be considered to make a unique contribution to the ongoing development of models for prediction of atmospheric corrosion of steel, which is important for much modern infrastructure. In practice, protective coatings and/or cathodic protection systems are applied to attempt to counter marine corrosion. These systems are not without cost and problems themselves, requiring regular servicing and maintenance and, in the case of protective coatings, are problematic for longer-term environmental effects caused by deteriorating coatings. For some infrastructure, such as the interiors of water injection pipelines widely used in the oil and gas industries, the approach is to rely simply on a sacrificial layer of steel, noting that the corrosion products of the FeOOH–Fe₂O₃ type are not environmentally unfriendly; in fact, they are of the same type as iron ore mineral deposits.

6. Conclusions

The data for exposures over 5 years at the marine corrosion test site confirm that already within 12 months or so of first exposure, the rate of corrosion loss declines, and after about 18 months, increases again before declining to what appears to be a longer-term rate. This pattern is delayed by some 6 months for first exposures in autumn–winter periods. There was essentially no correlation between corrosion losses and rainfall.

For the marine climate involved, cardinal orientation had little effect on two-sided corrosion losses of the steel coupons, although corrosion losses were marginally greater for single-sided coupons facing away from the predominant wind direction. The total corrosion losses for coupons exposed simultaneously on both sides were significantly higher than the sum of the one-sided corrosion losses, a result attributed to hydrodynamic effects, but that requires further investigation.

The effect of the angle of inclination of coupons was found to vary with period of exposure. One-sided coupons facing skywards and those facing downwards showed the highest corrosion losses after 12 months of exposure, but after 3 years of exposure, the coupons facing away from solar radiation showed the greatest losses. This effect, attributed for the present to the combined effects of hydrodynamics, time-of-wetness, and solar exposure, warrants further investigation.

The corrosion of coupons increased with height above ground up to about 2 m and steadied with further elevation. Individual coupons were found to corrode at two to three times that of continuous strips exposed under identical conditions.

The coefficient of variation for corrosion loss for coupons exposed to essentially similar conditions of inclination, orientation, and height is around 0.2 and increased slightly with extended exposures, consistent with previous findings. However, the coefficient of variation is up to four times greater when including coupons at different heights or at different inclinations.

Overall, the present results provide a unique set of results for atmospheric corrosion of steel coupons with various orientations and exposure conditions, all at the one marine exposure site.