While metal structures like subsea pipelines, submarines, offshore windfarms, and seawater-cooled power plants stand testament to the resilience of metal, they also showcase the relentless, unforgiving nature of marine environments. Several factors make these environments an ideal breeding ground for marine corrosion:
Many metal structures carry toxic materials that can be harmful to the marine ecosystem through which they pass. Undetected corrosion in these structures can gradually lead to catastrophic failures; hence, timely detection is crucial. Traditional methods are proving to be inadequate to solve these problems, but newer corrosion detection technologies—powered by artificial intelligence (AI)—may be a game changer for the corrosion industry.
Corrosion in Marine Structures
There are four types of marine exposure zones: atmospheric, splash, tidal, and submerged. The type of corrosion in marine structures depends on the zone in which they are installed, the type of corrosive environment that they are subject to, etc. The most common forms of corrosion in marine structures include the following:
From Manual Inspection to Intelligent Detection
Traditional methods of corrosion detection involve physical examination of the structures, such as visual inspection, non-destructive testing (NDT) methods, and direct electrochemical measurements. However, these methods have their limitations due to the difficulty involved in accessing these structures to perform a thorough assessment. They are also dangerous and expensive.
Technical divers inspecting and performing repairs on a ship propeller underwater. (Source: Popartic/iStock)
Recently, advancements in AI have brought improvements to marine corrosion detection. For example, state-of-the-art prediction and detection methods have been developed that enable engineers and maintenance specialists to analyze data and perform preventive maintenance on the structures. These actions decrease the chances of corrosion failure and consequently increase the life span of the structures.
What Technologies Can Detect Marine Corrosion?
Among the various AI technologies used in marine research, pattern recognition (PR), machine learning (ML), and deep learning (DL) are recognized as the most effective and consistent methods in corrosion engineering. PR and DL are subsets of AI that are different. The application areas of PR and ML overlap as they are closely linked with each other; however, ML focuses on learning algorithms, while PR focuses on approaches for classification tasks.
There are mainly two approaches to corrosion detection, which will be discussed in more detail below.
Predictive Maintenance (PdM)
This strategy is based on detecting the early signs of damage by conducting proactive maintenance. PdM has a set of mathematical models that can detect when errors occur and when to perform maintenance work. This method is efficient in predicting the level of corrosion in an affected material and has therefore gained attention in the corrosion industry. It can be categorized into:
- PdM with hybrid models
- PdM with data-based models
- PdM with physics-based models
- PdM with knowledge-based models
Computer Vision and Image Processing
Computer vision imitates human vision using ML; it can see and interpret patterns from optical inputs. This is a type of NDT with high accuracy and quick detection of corrosion boundaries, so it can interpret inputs from various parameters like color, size, and texture.
In image processing, the images are enhanced, improved, and transformed through tuning features and parameters. This is therefore beneficial for corrosion detection because the processing provides a clearer picture of the corroded area. Recently, many algorithms have been established to detect, assess, and predict corrosion based on color, texture, clustering, pixilation, wavelet transformation, etc. There are five main processes:
- Image acquisition
- Image pre-processing
- Image segmentation
- Feature extraction
- Image classification
In real-world applications, AI corrosion detection is applied in marine environments through the following ways:
- Inspecting underwater structures like piers, rigs, piles, bridges, etc.
- Analyzing images of ship hulls for corrosion and breakdown of their protective coating
- Monitoring the integrity of chemical and oil & gas pipelines that pass through harsh marine environments
AI in Marine Corrosion Detection: The Benefits
The use of AI to detect marine corrosion has a wide range of advantages:
Increased accuracy – Corrosion defects are detected with greater accuracy and consistency using AI algorithms when compared with traditional inspection methods.
Improved efficiency – The corrosion detection process can be streamlined to save time and resources by automating inspection tasks and prioritizing maintenance schedules.
Safety – Underwater divers risk their lives to access and inspect remotely located structures. Automating such tasks thus mitigates the risk of injuries and accidents that can be fatal in such hazardous environments.
Cost effective – The use of AI predictive modeling to create proactive maintenance strategies minimizes operational disruptions, downtime, inspection, and repair costs.
A commercial vessel undergoing maintenance in a dry dock, including hull painting and repairs. AI use in the marine industry can reduce the time it takes to both detect and repair issues in underwater assets. (Source: artiemedvedev/iStock)
AI in Marine Corrosion Detection: The Drawbacks
The widespread use of AI in marine corrosion also has its challenges:
Availability of good-quality data – Efficient AI detection and prediction will depend on the quality and diversity of training data.
Dynamic environment – The marine environment is unpredictable and constantly changing (temperature, flow dynamics, water chemistry, etc.), consequently affecting the performance of AI algorithms.
Smooth integration – Integrating new technology with old technology requires changes in workflow, as well as investment in training and equipment. This is not only expensive, but may also be met with resistance by the workforce.
Conclusion
Although corrosion is a mainstream problem in the marine industry, it has limited solutions. Extensive research and testing are required in this field to use AI with machine learning, pattern recognition, and deep learning to come to conclusive methods that can predict and detect marine corrosion effectively. Considering the cost of constructing marine structures, it is worthwhile to invest in the research and adoption of AI technology for corrosion detection.