Profiling Corrosion Rates for Offshore Wind Turbines with Depth in the North Sea

Corrosion in the marine environment is a complex and expensive form of damage. It is commonly studied by the deployment of coupons that reflect the marine corrosion a structure will experience, thus allowing design and maintenance prevention strategies to be developed accordingly. This study stems from the lack of information in the literature regarding the profiling of corrosion with respect to marine depth in the North Sea where important wind farm developments have been undertaken. To address such issue a field experiment has been designed and carried out in the vicinity of the Westermost Rough Windfarm in the North Sea. The field experiment consists of deploying steel S355 coupons below the tidal area and capturing the effects of corrosion, the mass loss from which the corrosion rate is derived and the chemical products that makes up the rust with water depth. The study involves proper planning and logistics to ensure that the field experiment survives the rough conditions of the North Sea for a duration of 111 days. A high corrosion rate of 0.83 mm/year has been observed in this experiment. This paper goes into the details of the deployment blueprint employed and the analyses of the coupons to provide a conclusive observation and modelling of corrosion with respect to water depth under free or open sea water corrosion condition.


Introduction
Marine corrosion is a complex process, especially for steel structures. There are various interactions between the metal matrix and the marine environment which alter the corrosion. This interaction is affected by several factors such as the temperature, dissolved oxygen, pH and salinity being flagged as the most important ones.
The temperature of the oceans has changed significantly over the years due to global warming and its effect in such a complex environment requires tracking and comparison. The metals used for structural purposes for ships, offshore wind turbines (OWT), and oil and gas platforms have evolved substantially in terms of the strength of the materials. Their corrosion resistance needs to be assessed accordingly.
The OWT will undoubtedly suffer from corrosion if left unprotected [1]. The designers have borne this in mind and have been careful in the strategy for the corroding piles by including different corrosion protections especially in the immersed sections and splash zones of the monopile where in general it is considered to be highly influenced by corrosion [2][3][4][5][6][7].
If left unprotected, the structures are very likely to suffer from a myriad of corrosion mechanisms. Often, OWT tend to be compared to oil and gas platforms. Despite some similarities and standards being inspired from the oil and gas practices, there are some fundamental differences, particularly in In a sealed structure, the dissolved oxygen is rapidly absorbed and the medium becomes anaerobic causing hydrogen sulphide to be released as a product of the new corrosion mechanism. Within a closed environment it is expected that this form of corrosion would be limited as the internal conditions would not continue to support the chemical reaction [15]. In reality though, partial oxygenation affects corrosion by promoting a more acidic environment and recently pits have been discovered [15]. Corrosion discovered for an OWT was of the following nature [13]: • Mud zone: Differential aeration, microbial-induced corrosion (MIC) and hydrogen induced stress cracking (HISC) • Waterline: pitting corrosion due to differential aeration • Stagnant water causing important environmental discrepancies • Weld defects causing corrosion fatigue and stress corrosion cracking (SCC) • Acidification if sacrificial anodes are installed • Accumulation of gases such as H2S, H2 and CH4 Studies have been done in various parts of the world to capture the effects of marine corrosion on carbon steel. These variations have been observed by several authors but there seems to be absence of this effect with respect to the depth especially in the North Sea and for carbon steel. This study looks at the corrosion loss of carbon steel S355 used in the industry. A novel mooring approach capable of supporting multiple experiments simultaneously was designed and experimented; in this case a corrosion experiment with a marine growth (the colonisation of hard surfaces such as those provided by wind turbine substructures by marine organisms) experiment. The latter of which would not be discussed further beyond the description of shared structural elements. For example, the frames inside which corrosion coupons were mounted also functioned as a sampling structure for marine growth. The frames were made of either plastic (PVC) or S355 steel coated with interzone 954 anti-corrosion coating.
After 111 days of exposure further analyses have been done after cleaning the corrosion coupon. The first analysis being visual inspection and the mass loss and corrosion rates. From the corrosion product further information on corrosion rate with respect to depth can be deduced from the Energy Dissipative Spectra (EDS) analyser connected to the Scanning Electron Microscope (SEM). To characterise the chemical structure of the rust, the X-ray Diffraction method (XRD) has been used. It is to be noted that in the scope of this study biofouling and erosion have not been monitored. In reality though, partial oxygenation affects corrosion by promoting a more acidic environment and recently pits have been discovered [15]. Corrosion discovered for an OWT was of the following nature [13]: • Mud zone: Differential aeration, microbial-induced corrosion (MIC) and hydrogen induced stress cracking (HISC) • Waterline: pitting corrosion due to differential aeration • Stagnant water causing important environmental discrepancies • Weld defects causing corrosion fatigue and stress corrosion cracking (SCC) • Acidification if sacrificial anodes are installed • Accumulation of gases such as H 2 S, H 2 and CH 4 Studies have been done in various parts of the world to capture the effects of marine corrosion on carbon steel. These variations have been observed by several authors but there seems to be absence of this effect with respect to the depth especially in the North Sea and for carbon steel. This study looks at the corrosion loss of carbon steel S355 used in the industry. A novel mooring approach capable of supporting multiple experiments simultaneously was designed and experimented; in this case a corrosion experiment with a marine growth (the colonisation of hard surfaces such as those provided by wind turbine substructures by marine organisms) experiment. The latter of which would not be discussed further beyond the description of shared structural elements. For example, the frames inside which corrosion coupons were mounted also functioned as a sampling structure for marine growth. The frames were made of either plastic (PVC) or S355 steel coated with interzone 954 anti-corrosion coating.
After 111 days of exposure further analyses have been done after cleaning the corrosion coupon. The first analysis being visual inspection and the mass loss and corrosion rates. From the corrosion product further information on corrosion rate with respect to depth can be deduced from the Energy Dissipative Spectra (EDS) analyser connected to the Scanning Electron Microscope (SEM). To characterise the chemical structure of the rust, the X-ray Diffraction method (XRD) has been used. It is to be noted that in the scope of this study biofouling and erosion have not been monitored.

Field Experiment Design and Deployment
Marine corrosion is complex and to understand the corrosion interaction as realistically between the metal and the environment, a field experiment has to be done. Marine field experiments can be considered to exist in three categories namely harbour or port, shallow and deep water. The port or harbour has a different environment with particularly high levels of pollutants. Accelerated low water Energies 2020, 13, 2518 4 of 19 corrosion (ALWC) is usually experienced in those waters which is an aggressive form of microbial influenced corrosion (MIC) that may occur on steel in estuarine and marine structures [19,20]. ALWC has been observed to be enhanced with dissolved inorganic nitrogen (DIN) [21]. In general, coupons are suspended from fixed structure or the piles are removed after a number of years of service to capture the corrosion rate. The piles are able to capture the different regions (atmospheric, tidal, immersed, mudline and below the mudline) corrosion levels. This provides a full profiling which also provides the opportunity to characterise the corrosion mechanisms [22,23].
Deep water corrosion experiments are more challenging to perform. They require robust equipment that will sustain the high pressures [24]. As the offshore oil and gas exploration is moving further offshore, understanding the interaction between the environment and the metal at this depth is important. Coupons have been placed at depth of more than 1000 m for corrosion analysis [25]. Experiments have been done at a number of locations with varying results. Those experiments were done using a submerged buoy which is attached to coupons across different depths. An anchor is used as mooring and it is worth noting that this arrangement is used only in the immersed region [24].
For shallow water, coupons are often suspended from fixed structures like offshore oil and gas platforms, OWTs or offshore meteorological masts. The question for shallow waters is how to test for corrosion in an unexploited site. There have been many experiments done but none in the literature takes a detailed approach in profiling corrosion loss with depth in the past decade from the literature [26,27]. This will require a new design to accommodate the experiment to profile the corrosion rate and loss with depth and is explained below.
The entire experiment in shallow water of maximum water depth of 30 m consisted of five arrays, each comprising four coupons suspended inside a frame. The whole set-up was moored to the seabed by 250 kg gravity anchors made from two 125 kg webbed tubes tied together as shown in Figure 2a.
In adverse weather conditions, the North Sea can be rough, with wave heights (from crests to troughs) of >10 m. The force generated by those waves was immense and would carry or displace large objects on the surface without any difficulty. Whilst completely avoiding large waves would not be achievable given the vertical depth profile of the mooring system, it was decided that a subsurface buoy, submerged below the minimum tidal height, would mitigate the effects in most weather conditions. As an air-filled buoy would suffer from compression effects caused by water pressure as the tide rises and falls, which could cause the buoy to become negatively buoyant and sink to the seabed at high tide, a foam-filled buoy was selected. A buoy which provided 90 kg buoyancy force was selected. It is of note that in using such an arrangement, the data could not be collected from the splash zone.
Energies 2020, 13, x FOR PEER REVIEW  4 of 19 (atmospheric, tidal, immersed, mudline and below the mudline) corrosion levels. This provides a full profiling which also provides the opportunity to characterise the corrosion mechanisms [22,23]. Deep water corrosion experiments are more challenging to perform. They require robust equipment that will sustain the high pressures [24]. As the offshore oil and gas exploration is moving further offshore, understanding the interaction between the environment and the metal at this depth is important. Coupons have been placed at depth of more than 1000 m for corrosion analysis [25]. Experiments have been done at a number of locations with varying results. Those experiments were done using a submerged buoy which is attached to coupons across different depths. An anchor is used as mooring and it is worth noting that this arrangement is used only in the immersed region [24].
For shallow water, coupons are often suspended from fixed structures like offshore oil and gas platforms, OWTs or offshore meteorological masts. The question for shallow waters is how to test for corrosion in an unexploited site. There have been many experiments done but none in the literature takes a detailed approach in profiling corrosion loss with depth in the past decade from the literature [26,27]. This will require a new design to accommodate the experiment to profile the corrosion rate and loss with depth and is explained below.
The entire experiment in shallow water of maximum water depth of 30 m consisted of five arrays, each comprising four coupons suspended inside a frame. The whole set-up was moored to the seabed by 250 kg gravity anchors made from two 125 kg webbed tubes tied together as shown in Figure 2a. In adverse weather conditions, the North Sea can be rough, with wave heights (from crests to troughs) of >10 m. The force generated by those waves was immense and would carry or displace large objects on the surface without any difficulty. Whilst completely avoiding large waves would not be achievable given the vertical depth profile of the mooring system, it was decided that a subsurface buoy, submerged below the minimum tidal height, would mitigate the effects in most weather conditions. As an air-filled buoy would suffer from compression effects caused by water pressure as the tide rises and falls, which could cause the buoy to become negatively buoyant and sink to the seabed at high tide, a foam-filled buoy was selected. A buoy which provided 90 kg buoyancy force was selected. It is of note that in using such an arrangement, the data could not be collected from the splash zone. Between the topmost frame (closest to the sea surface) and the submerged buoy and the bottommost frame and the anchor a device known as a line spacer was added. This device maintained the spread of the ropes, preventing them from twisting or compressing the array under tension.  Between the topmost frame (closest to the sea surface) and the submerged buoy and the bottommost frame and the anchor a device known as a line spacer was added. This device maintained the spread of the ropes, preventing them from twisting or compressing the array under tension.
The attachment of the plates to the frames varied depending on whether the frames were plastic or steel. For the steel frames, at each corner, a triangular bracket with a through hole of 10 mm diameter was drilled. The plates were connected to these attachment points using marine cable ties, which had a breaking strength of >150 kg. For the plastic frames, cable ties tightened around each corner formed the attachment points for the plates. The assembly of the plates and coupons are shown in Figure 1b.

Coupons
Carbon steel of S355 grade was used. Some experiments were carried in the past using coupons and the one most relevant in the literature was done in the North Sea for OWT corrosion monitoring having dimensions of 400 × 90 × 6 mm 3 [1]. The same dimensions were used for this study. Four holes had been drilled into the coupon during manufacturing for attachment to the frame. It was kept to a minimal size as this region would have to be excluded in the study. The exposure time for the experiment covered 111 days.
Coupons were prepared to meet the ASTM G1 standards and were colour marked and numbered accordingly to depth. The depths chosen for deployment are shown in Table 1. The marine tape in the corners was used, as reference in case 3D cloud coordinate analysis would be done as shown in Figure 3. There was a potential problem that could arise from it. The water could flow beneath the layer of the tapes and introduce a limited amount of oxygen resulting in differential oxygen concentration causing crevice corrosion. As such, each of the spots where the tapes were applied were also glued on with extra strong marine glue, commonly used to patch holes in hulls of boats. The plates were colour coded based on which array they were to be deployed in, the plates code named as red, grey, black, blue and green. The front of each plate was denoted one to four strips of tape based on their relative position in their array (as shown in Figure 3): plate 1 (one strip at the top), plate 2 (two strips at the top), plate 3 (two strips at the top and one strip at the bottom), plate 4 (two strips at the top and two strips at the bottom). The corners of each plate were protected with a layer of tape to preserve a reference point for gauging the height of corrosion. The attachment of the plates to the frames varied depending on whether the frames were plastic or steel. For the steel frames, at each corner, a triangular bracket with a through hole of 10 mm diameter was drilled. The plates were connected to these attachment points using marine cable ties, which had a breaking strength of >150 kg. For the plastic frames, cable ties tightened around each corner formed the attachment points for the plates. The assembly of the plates and coupons are shown in Figure 1b.

Coupons
Carbon steel of S355 grade was used. Some experiments were carried in the past using coupons and the one most relevant in the literature was done in the North Sea for OWT corrosion monitoring having dimensions of 400 × 90 × 6 mm 3 [1]. The same dimensions were used for this study. Four holes had been drilled into the coupon during manufacturing for attachment to the frame. It was kept to a minimal size as this region would have to be excluded in the study. The exposure time for the experiment covered 111 days.
Coupons were prepared to meet the ASTM G1 standards and were colour marked and numbered accordingly to depth. The depths chosen for deployment are shown in Table 1. The marine tape in the corners was used, as reference in case 3D cloud coordinate analysis would be done as shown in Figure 3. There was a potential problem that could arise from it. The water could flow beneath the layer of the tapes and introduce a limited amount of oxygen resulting in differential oxygen concentration causing crevice corrosion. As such, each of the spots where the tapes were applied were also glued on with extra strong marine glue, commonly used to patch holes in hulls of boats. The plates were colour coded based on which array they were to be deployed in, the plates code named as red, grey, black, blue and green. The front of each plate was denoted one to four strips of tape based on their relative position in their array (as shown in Figure 3): plate 1 (one strip at the top), plate 2 (two strips at the top), plate 3 (two strips at the top and one strip at the bottom), plate 4 (two strips at the top and two strips at the bottom). The corners of each plate were protected with a layer of tape to preserve a reference point for gauging the height of corrosion.  The mass of each plate was measured by a digital scale to an accuracy of 0.1 g. The results are displayed in the Appendix A.

Assembly
The assembly can be summarised in the charts shown in Figure 4a,b.

Assembly
The assembly can be summarised in the charts shown in Figure 4a,b.

Deployment and Recovery
Five arrays were deployed northwest of Westermost Rough Windfarm as shown in Figure 5. Array 1 was deployed several weeks ahead of arrays 2-4 to test the deployment method. During deployment, a crane was used to lift the anchors over the stern of the vessel (Figure 6a). The frames were each lowered into the water (Figure 6b), starting with that closest to the anchor, followed by the mooring buoy ( Figure 6c). Once the array was in the water the vessel was moved ahead to allow the array to straighten with the mooring buoy at the back. Finally, the surface marker buoy was dropped into the water and the anchors were released.
The marker buoy remained on the surface to warn local vessel traffic of the hazard beneath the surface and for recovery at a later stage. It was planned that during recovery, the recovery line would be pulled aboard and connected to a winch. The anchor would then be winched up to the vessel, allowing the subsurface buoy to float to the surface along with the frames. However, the strength of the mooring ropes proved to be insufficient during the extreme weather and strong winds. Two of the arrays were recovered (Blue and Black) after the ropes failed and the moorings broke away from the anchor. The remaining arrays could not be re-located, but it can be assumed that they suffered a similar fate. For the arrays that were recovered, the steel frames had been damaged, but the corrosion coupons were still in place.

Deployment and Recovery
Five arrays were deployed northwest of Westermost Rough Windfarm as shown in Figure 5. Array 1 was deployed several weeks ahead of arrays 2-4 to test the deployment method.

Deployment and Recovery
Five arrays were deployed northwest of Westermost Rough Windfarm as shown in Figure 5. Array 1 was deployed several weeks ahead of arrays 2-4 to test the deployment method. During deployment, a crane was used to lift the anchors over the stern of the vessel (Figure 6a). The frames were each lowered into the water (Figure 6b), starting with that closest to the anchor, followed by the mooring buoy (Figure 6c). Once the array was in the water the vessel was moved ahead to allow the array to straighten with the mooring buoy at the back. Finally, the surface marker buoy was dropped into the water and the anchors were released.
The marker buoy remained on the surface to warn local vessel traffic of the hazard beneath the surface and for recovery at a later stage. It was planned that during recovery, the recovery line would be pulled aboard and connected to a winch. The anchor would then be winched up to the vessel, allowing the subsurface buoy to float to the surface along with the frames. However, the strength of the mooring ropes proved to be insufficient during the extreme weather and strong winds. Two of During deployment, a crane was used to lift the anchors over the stern of the vessel (Figure 6a). The frames were each lowered into the water (Figure 6b), starting with that closest to the anchor, followed by the mooring buoy (Figure 6c). Once the array was in the water the vessel was moved ahead to allow the array to straighten with the mooring buoy at the back. Finally, the surface marker buoy was dropped into the water and the anchors were released. There is no way to determine when the mooring ropes failed, and the arrays broke away from the anchors. It was therefore assumed that the length of exposure, for the purposes of the analysis, would be from the deployment date to the collection date, which amounted to 111 days. In addition, the results would be displayed with respect to the depth (which is assumed as the exact depth of deployment and orientation cannot be fully traced due to the breaking effect), discussed more thoroughly from the observations, and compared to other experiments or industry standards mainly focussing on the seabed and submerged regions corrosion rate.
Since the profiling of the depth is unknown, the results section has been presented as a methodology to treat the corrosion data. The corrosion data obtained is then tested against other experiments from the literature in the discussion section to observe if there is a trend that can be established. The data reflects the depth profile of the marine corrosion as initially designed before the failure of the mooring ropes.

Results
After collection, the coupons were dried and photographed. Some of the residual material that had fallen off from each coupon was placed in a test tube for further chemical analysis.
The plates were cleaned according to the ASTM G1 standard where 1000 mL HCl, 20 g antimony trioxide (Sb2O3), 50 g stannous chloride (SnCl2) were used and shown in Figure 7a,b [28].

Visual Inspection
The plates cleaned revealed high levels of corrosion and in some cases pitting corrosion was present shown in Figure 8a. In addition, the variation of the pits with respect to pit depth was noticeable and fell mainly in the small and shallow category to no visible pits represented in Figure  8b. An isolated case of crevice corrosion was found under the tape as shown in Figure 8c. At this stage, there were few broad pits, which tended to be formed by the mechanism of pit coalescence The marker buoy remained on the surface to warn local vessel traffic of the hazard beneath the surface and for recovery at a later stage. It was planned that during recovery, the recovery line would be pulled aboard and connected to a winch. The anchor would then be winched up to the vessel, allowing the subsurface buoy to float to the surface along with the frames. However, the strength of the mooring ropes proved to be insufficient during the extreme weather and strong winds. Two of the arrays were recovered (Blue and Black) after the ropes failed and the moorings broke away from the anchor. The remaining arrays could not be re-located, but it can be assumed that they suffered a similar fate. For the arrays that were recovered, the steel frames had been damaged, but the corrosion coupons were still in place.
There is no way to determine when the mooring ropes failed, and the arrays broke away from the anchors. It was therefore assumed that the length of exposure, for the purposes of the analysis, would be from the deployment date to the collection date, which amounted to 111 days. In addition, the results would be displayed with respect to the depth (which is assumed as the exact depth of deployment and orientation cannot be fully traced due to the breaking effect), discussed more thoroughly from the observations, and compared to other experiments or industry standards mainly focussing on the seabed and submerged regions corrosion rate.
Since the profiling of the depth is unknown, the results section has been presented as a methodology to treat the corrosion data. The corrosion data obtained is then tested against other experiments from the literature in the discussion section to observe if there is a trend that can be established. The data reflects the depth profile of the marine corrosion as initially designed before the failure of the mooring ropes.

Results
After collection, the coupons were dried and photographed. Some of the residual material that had fallen off from each coupon was placed in a test tube for further chemical analysis.
The plates were cleaned according to the ASTM G1 standard where 1000 mL HCl, 20 g antimony trioxide (Sb 2 O 3 ), 50 g stannous chloride (SnCl 2 ) were used and shown in Figure 7a,b [28]. There is no way to determine when the mooring ropes failed, and the arrays broke away from the anchors. It was therefore assumed that the length of exposure, for the purposes of the analysis, would be from the deployment date to the collection date, which amounted to 111 days. In addition, the results would be displayed with respect to the depth (which is assumed as the exact depth of deployment and orientation cannot be fully traced due to the breaking effect), discussed more thoroughly from the observations, and compared to other experiments or industry standards mainly focussing on the seabed and submerged regions corrosion rate.
Since the profiling of the depth is unknown, the results section has been presented as a methodology to treat the corrosion data. The corrosion data obtained is then tested against other experiments from the literature in the discussion section to observe if there is a trend that can be established. The data reflects the depth profile of the marine corrosion as initially designed before the failure of the mooring ropes.

Results
After collection, the coupons were dried and photographed. Some of the residual material that had fallen off from each coupon was placed in a test tube for further chemical analysis.
The plates were cleaned according to the ASTM G1 standard where 1000 mL HCl, 20 g antimony trioxide (Sb2O3), 50 g stannous chloride (SnCl2) were used and shown in Figure 7a,b [28].

Visual Inspection
The plates cleaned revealed high levels of corrosion and in some cases pitting corrosion was present shown in Figure 8a. In addition, the variation of the pits with respect to pit depth was noticeable and fell mainly in the small and shallow category to no visible pits represented in Figure   Figure 7. Samples: (a) before cleaning, (b) after cleaning.

Visual Inspection
The plates cleaned revealed high levels of corrosion and in some cases pitting corrosion was present shown in Figure 8a. In addition, the variation of the pits with respect to pit depth was noticeable and fell mainly in the small and shallow category to no visible pits represented in Figure 8b. An isolated case of crevice corrosion was found under the tape as shown in Figure 8c. At this stage, there were few broad pits, which tended to be formed by the mechanism of pit coalescence represented in Figure 9. Smaller pits were also present and when viewed under the magnifying glass provided a clear case that the new pits initiated on the plates were present and not noticeable with the naked eye. There seemed to be more than one could count just signifying that in 111 days, the presence of corrosion cannot be neglected or underestimated and localised corrosion is readily obvious. represented in Figure 9. Smaller pits were also present and when viewed under the magnifying glass provided a clear case that the new pits initiated on the plates were present and not noticeable with the naked eye. There seemed to be more than one could count just signifying that in 111 days, the presence of corrosion cannot be neglected or underestimated and localised corrosion is readily obvious.
(a) (b) (c) Substantial variation was also observed between plates where the topological surfaces varied. More pitted regions accounted for a rougher surface compared to less pitted surface. It was also seen that in some cases on the same plate, some regions had important levels of pits whereas others did not, implying the nature of the pitting corrosion is highly localised on the plate for carbon steel.
It was estimated for instance in the 3-black case that the number of pits on the front side was more than that on its back side. This difference is hard to explain due to a variety of factors that have affected the experiment. This observation has to be re-tested though to come out of the hypothetical shell to a scientific explanation with appropriate instrumentation such as the water velocity, the corrosion current and the biological and chemical influences of the water combined together in the complex sea water without the failure of the ropes and the depth preserved. The surface was more rugged as shown by Figure 9 when pitting corrosion seemed to be the dominating form of corrosion.

Mass Loss and Corrosion Rate
The mass loss and corrosion rate are extremely important parameters those are generally used for design and monitoring of marine structures. They are assumed to be solely under the influence of uniform corrosion. This would give an insight of the harshness of the environment on the metal.
The raw data can be seen in Table A1 in the appendix A section. The mass loss was derived from the following equation: represented in Figure 9. Smaller pits were also present and when viewed under the magnifying glass provided a clear case that the new pits initiated on the plates were present and not noticeable with the naked eye. There seemed to be more than one could count just signifying that in 111 days, the presence of corrosion cannot be neglected or underestimated and localised corrosion is readily obvious.
(a) (b) (c) Substantial variation was also observed between plates where the topological surfaces varied. More pitted regions accounted for a rougher surface compared to less pitted surface. It was also seen that in some cases on the same plate, some regions had important levels of pits whereas others did not, implying the nature of the pitting corrosion is highly localised on the plate for carbon steel.
It was estimated for instance in the 3-black case that the number of pits on the front side was more than that on its back side. This difference is hard to explain due to a variety of factors that have affected the experiment. This observation has to be re-tested though to come out of the hypothetical shell to a scientific explanation with appropriate instrumentation such as the water velocity, the corrosion current and the biological and chemical influences of the water combined together in the complex sea water without the failure of the ropes and the depth preserved. The surface was more rugged as shown by Figure 9 when pitting corrosion seemed to be the dominating form of corrosion.

Mass Loss and Corrosion Rate
The mass loss and corrosion rate are extremely important parameters those are generally used for design and monitoring of marine structures. They are assumed to be solely under the influence of uniform corrosion. This would give an insight of the harshness of the environment on the metal.
The raw data can be seen in Table A1 in the appendix A section. The mass loss was derived from the following equation: mass loss = original mass − final mass after acid cleaning The units measured were in grams. Substantial variation was also observed between plates where the topological surfaces varied. More pitted regions accounted for a rougher surface compared to less pitted surface. It was also seen that in some cases on the same plate, some regions had important levels of pits whereas others did not, implying the nature of the pitting corrosion is highly localised on the plate for carbon steel.
It was estimated for instance in the 3-black case that the number of pits on the front side was more than that on its back side. This difference is hard to explain due to a variety of factors that have affected the experiment. This observation has to be re-tested though to come out of the hypothetical shell to a scientific explanation with appropriate instrumentation such as the water velocity, the corrosion current and the biological and chemical influences of the water combined together in the complex sea water without the failure of the ropes and the depth preserved. The surface was more rugged as shown by Figure 9 when pitting corrosion seemed to be the dominating form of corrosion.

Mass Loss and Corrosion Rate
The mass loss and corrosion rate are extremely important parameters those are generally used for design and monitoring of marine structures. They are assumed to be solely under the influence of uniform corrosion. This would give an insight of the harshness of the environment on the metal.
The raw data can be seen in Table A1 in the Appendix A section. The mass loss was derived from the following equation: mass loss = original mass − final mass after acid cleaning (1) The units measured were in grams.
Corrosion loss was derived from the mass loss and related to the density: The volume was that of a cuboid: The thickness loss could be calculated as: thickness loss = Mass loss Area X Density (4) The mass loss in 1-Black is more than 1-Blue by some margin as shown in Figure 10.
Energies 2020, 13, x FOR PEER REVIEW 9 of 19 Corrosion loss was derived from the mass loss and related to the density: The volume was that of a cuboid: The thickness loss could be calculated as: thickness loss = Mass loss Area X Density (4) The mass loss in 1-Black is more than 1-Blue by some margin as shown in Figure 10. The number of days of exposure represented the thickness loss assumed to be uniform and from that principle the corrosion rate per year could be found: (5) Figure 11 show the results in terms of corrosion rate as a result of the mass loss. The number of days of exposure represented the thickness loss assumed to be uniform and from that principle the corrosion rate per year could be found: corrosion rate mm year = thickness loss Number of exposure days X 365.25 . (5) Figure 11 show the results in terms of corrosion rate as a result of the mass loss. (5) Figure 11 show the results in terms of corrosion rate as a result of the mass loss.

Chemical Analysis and Implications
This section was carried out as it could provide a deduction on the type of corrosion experienced by the metal. This required an analysis of the substrate through scanning electron microscope and

Chemical Analysis and Implications
This section was carried out as it could provide a deduction on the type of corrosion experienced by the metal. This required an analysis of the substrate through scanning electron microscope and the energy dispersive X-ray and the X-ray diffraction technique which will show how the elements combine together to form the rust product.

SEM and EDS
To be able to comprehend the variation, an assessment of the chemical products had been done which could be used to reflect the environmental conditions the plates were subjected to. The results for the EDS and the spectra can be found in Table A2 in Appendix B The scanning electron microscopy (SEM) provides a view of the microstructure of the rust and the Energy Dispersive X-Ray Spectroscopy (EDS) provides the elements present in that specific region. Rust clusters were placed on the adhesive support that was later placed in SEM as demonstrated in Figures 12 and 13. the energy dispersive X-ray and the X-ray diffraction technique which will show how the elements combine together to form the rust product.

SEM and EDS
To be able to comprehend the variation, an assessment of the chemical products had been done which could be used to reflect the environmental conditions the plates were subjected to. The results for the EDS and the spectra can be found in Table A2 in Appendix B The scanning electron microscopy (SEM) provides a view of the microstructure of the rust and the Energy Dispersive X-Ray Spectroscopy (EDS) provides the elements present in that specific region. Rust clusters were placed on the adhesive support that was later placed in SEM as demonstrated in Figures 12 and 13. The results from the remaining clusters showed this similar behaviour and therefore it was concluded that iron and oxygen were the dominating products with important variations in the % composition of other elements. The results from the remaining clusters showed this similar behaviour and therefore it was concluded that iron and oxygen were the dominating products with important variations in the % composition of other elements.

XRD
The chemical products formed due to corrosion were determined using the XRD technique. The rust formed was iron oxide-hydroxide. The equipment is shown in Figure 14.

XRD
The chemical products formed due to corrosion were determined using the XRD technique. The rust formed was iron oxide-hydroxide. The equipment is shown in Figure 14.
This suggested that the corrosion product was actually oxygen rich. No form of anaerobic corrosion was found due to the absence of sulphide-based products. Low sulphide content was observed in the plates after corrosion test from EDS analysis ( Table 2).
XRD analysis of corrosion products can also give substantial information on the environmental condition of the sea compared to Pourbaix diagram of carbon steel. The product obtained falls in the passivation region of the Pourbaix diagram which can also explain the high region of pitting present. composition of other elements.

XRD
The chemical products formed due to corrosion were determined using the XRD technique. The rust formed was iron oxide-hydroxide. The equipment is shown in Figure 14. This suggested that the corrosion product was actually oxygen rich. No form of anaerobic corrosion was found due to the absence of sulphide-based products. Low sulphide content was observed in the plates after corrosion test from EDS analysis ( Table 2).
XRD analysis of corrosion products can also give substantial information on the environmental condition of the sea compared to Pourbaix diagram of carbon steel. The product obtained falls in the passivation region of the Pourbaix diagram which can also explain the high region of pitting present. Table 2. Elemental composition (wt%) of corrosion product collected from colour coded plates.

Discussion
As a result of the trend of the mass loss with respect to depth and checking if the snapping of the rope had a major influence on the experiment with respect to depth, an investigative work was carried out with two objectives to charter out:

•
The trend in corrosion with respect to depth • The corrosion rate magnitude A first point was covered mainly from a literature review. One article was isolated focussing on the immersed and sea bed corrosion and having the same arrangement as this study [29]. Comparing it would be challenging as the location and depth were dissimilar. Therefore, the depth and corrosion rate were normalised using the maximum normalisation method and plotted as shown in Figure 15 Energies 2020, 13

Discussion
As a result of the trend of the mass loss with respect to depth and checking if the snapping of the rope had a major influence on the experiment with respect to depth, an investigative work was carried out with two objectives to charter out:

•
The trend in corrosion with respect to depth • The corrosion rate magnitude A first point was covered mainly from a literature review. One article was isolated focussing on the immersed and sea bed corrosion and having the same arrangement as this study [29]. Comparing it would be challenging as the location and depth were dissimilar. Therefore, the depth and corrosion rate were normalised using the maximum normalisation method and plotted as shown in Figure 15  It can be clearly seen that the general trends for both of them seem to have a trend dipping from the first data point to the third and then from the third to forth going upwards. This trend is synonymous of the quadratic function. The percentage error was found between the two curves is shown in Figure 16. It can be clearly seen that the general trends for both of them seem to have a trend dipping from the first data point to the third and then from the third to forth going upwards. This trend is synonymous of the quadratic function. The percentage error was found between the two curves is shown in Figure 16. Despite the normalisation employed, the values were different. This could be explained because of the experiment in the literature happening at deep water from 500 to 5180 m in the Indian Ocean having a tropical climate and with a very different seawater environment and marine colonisers, whereas the experiment carried out the present study was done in temperate shallow water of the North Sea. It is interesting that similar that this trend does exist and would have to be tested with more results from different sites and depths to simulate and check this observation.
A trend though clear, and it could be suggested from the analysis that the array was in the water long enough to reflect that depth influence with respect to seawater.
To the corrosion rate against the depth, a quadratic equation was fitted to develop a mathematical correlation as shown in Figure 17. Upon interpolation, the splash zone corrosion rate can be found to be more than 1 mm/yr when placed at 20 m. Despite the normalisation employed, the values were different. This could be explained because of the experiment in the literature happening at deep water from 500 to 5180 m in the Indian Ocean having a tropical climate and with a very different seawater environment and marine colonisers, whereas the experiment carried out the present study was done in temperate shallow water of the North Sea. It is interesting that similar that this trend does exist and would have to be tested with more results from different sites and depths to simulate and check this observation.
A trend though clear, and it could be suggested from the analysis that the array was in the water long enough to reflect that depth influence with respect to seawater.
To the corrosion rate against the depth, a quadratic equation was fitted to develop a mathematical correlation as shown in Figure 17. Despite the normalisation employed, the values were different. This could be explained because of the experiment in the literature happening at deep water from 500 to 5180 m in the Indian Ocean having a tropical climate and with a very different seawater environment and marine colonisers, whereas the experiment carried out the present study was done in temperate shallow water of the North Sea. It is interesting that similar that this trend does exist and would have to be tested with more results from different sites and depths to simulate and check this observation.
A trend though clear, and it could be suggested from the analysis that the array was in the water long enough to reflect that depth influence with respect to seawater.
To the corrosion rate against the depth, a quadratic equation was fitted to develop a mathematical correlation as shown in Figure 17. Upon interpolation, the splash zone corrosion rate can be found to be more than 1 mm/yr when placed at 20 m. Upon interpolation, the splash zone corrosion rate can be found to be more than 1 mm/yr when placed at 20 m.
The graph was normalised in terms of the depth where the maximum normalisation method had been employed represented in Figure 18: Corrosion rate[mm/yr] = 1.0754(ND) 2 − 1.2133(ND) + 0.9281 . (6) where CR = corrosion rate, ND = normalised depth: The second analysis is to check the corrosion rate according to the standards based on the corrosion rate for corrosion allowance.
The flowing table based on the DNV J101 standard indicate the level of corrosion rate used for design purposes [30]. The graph was normalised in terms of the depth where the maximum normalisation method had been employed represented in Figure 18: where CR = corrosion rate, ND = normalised depth: The second analysis is to check the corrosion rate according to the standards based on the corrosion rate for corrosion allowance.
The flowing table based on the DNV J101 standard indicate the level of corrosion rate used for design purposes [30].
The values shown in Table 3 were low when compared to the actual experiment results. From the literature, high corrosion rates were reported from 0.9 and 1.4 mm/year in Alaska and the Gulf of Mexico respectively [31]. An observation on corrosion pits showed the multiplying effect of pit growth with regard to dissolved inorganic nitrogen (DIN) [32]. A good observation regarding the 4-Blue plate which is the closest to the seabed indicate the impact of Nitrogen on the corrosion rates shown in Figure 19. The values shown in Table 3 were low when compared to the actual experiment results. From the literature, high corrosion rates were reported from 0.9 and 1.4 mm/year in Alaska and the Gulf of Mexico respectively [31]. An observation on corrosion pits showed the multiplying effect of pit growth with regard to dissolved inorganic nitrogen (DIN) [32]. A good observation regarding the 4-Blue plate which is the closest to the seabed indicate the impact of Nitrogen on the corrosion rates shown in Figure 19. A comparison was carried out based from the data from various researchers from more than 40 sites. The corrosion rate observed in the North Sea from the experiment in the 111 days is the highest observed from Table 4. A comparison was carried out based from the data from various researchers from more than 40 sites. The corrosion rate observed in the North Sea from the experiment in the 111 days is the highest observed from Table 4.  Some pertinent questions on those corrosion rates provided by the standards would have to be asked. The first one would be when and where were those values obtained from? Which depth or relative sea water depth were the coupons deployed? How long has the exposure been keeping in mind that corrosion is a non-linear process? What were the impacts of the (sulphate reducing bacteria) SRB and other corrosion mechanism? Which grade of steel was it subjected to? Which environment zone and how the annual temperature has changed since the experiment? Has there been any changes to the location over the years with respect to fish farms or other sources of pollutants?   [27] Due to complexity of marine corrosion, should the wind farm developers be required to test the environment in corrosion terms to optimise their corrosion allowance and corrosion protection strategies rather than getting the data from the standard? This study denotes such a variation and it is good to keep in mind the dynamic nature of the oceans and the commercial activities like fish farming and the consequences of those parameters corrosion. This can tend to have wider consequences for instance in terms of corrosion fatigue and the SN curve that was derived from artificial seawater. This might be a gross understatement the fatigue life under real conditions especially if the corrosion rate is as high as observed.

Conclusions
The field experiment results can be summarised as: • A new deployment method has been developed to characterise free corrosion (natural corrosion) to profile the corrosion rate with respect to depth for steel S355, a study that has not been carried out as yet according to the literature. • The main advantage of using this system is that it reflects the variabilities in the immersed zones without requiring other structures to suspend it to and that water depth does have an influence on the corrosion rate.

•
High levels of corrosion rate had been observed that is way above prescribed standards with the highest being at the closest point to the seabed being 0.83 mm/yr. This result though is very similar to the lowest astronomical tide being 0.80 mm/yr. This gave a quadratic function of corrosion rate with respect to tidal depth.

•
The highest corrosion rates observed were four times higher than the ones prescribed by the DNV J101-Standards and considerably higher than the field experiments results from 43 other sites.

•
Further investigation on the reason behind the trend in corrosion has been done by carrying out the EDS method on the rust particles. Nitrogen at the seabed might have been the reason for the higher corrosion rates.

•
After 111 days, the main mechanism of corrosion is still aerobic but traces of sulphur indicate that anaerobic corrosion will happen. • There were only two arrays recovered out of five and to boost confidence in the results similar experiments will have to be done.

•
It is important that the array be strengthened especially for the connection of the lower frame and the anchors by using high strength marine chains rather than ropes to survive the harsh environment of the North Sea.