The Reason behind the SOCORRO application - Data and AI drive a mindset change in Corrosion Management
The overall challenge set forth in the SOCORRO project is to reduce the costs associated with corrosion. In recent years, the mindset on how to manage corrosion effectively is quickly changing. Corrosion management can be made more efficient using data, allowing to target (parts of) assets having the highest risk of failure due to corrosion. The SOCORRO project explores the possibility to use data about the assets’ environment to predict the overall corrosion risk. Through a combination of collecting sensor data and Artificial Intelligence (AI), the SOCORRO project aims to establish a decision support tool for data driven corrosion management. These tools should not be seen as a way to replace engineers, rational thinking and decision makers, but rather as tools providing data for the decision-making process.
The SOCORRO Approach
The project aims to reduce the costs associated to corrosion, by providing a new tool for corrosion management of submerged steel. This tool will allow companies to independently assess corrosion risks, with as a goal to take earlier preventive actions and evaluate the possibility of live extension of assets. In order to achieve this, the risk and expected severity of corrosion throughout the lifetime of the steel needs to be better understood. The overall goal of SOCORRO is to develop an AI algorithm capable of calculating a corrosion risk, based on environmental parameters, and integrate this AI tool in an easy-to-use and publicly available software platform that can provide decision support to asset owners.
To realise this, three important steps need to be taken: (1) collect real-time data from the exposed steel; (2) building on existing knowledge, further improve our understanding of corrosion in different environments and (3) develop an AI algorithm capable of converting the collected data in a corrosion risk. Key to the approach being developed is that environmental data, i.e. water quality parameters, are used as input for the model, and that a corrosion risk, rather than an exact corrosion rate is calculated. The approach followed in the SOCORRO project is to calculate a corrosion risk as a means to evaluate the environment or medium rather than the structure itself, and use the corrosion risk accumulated over time as a global risk indicator for the corrosion a structure has been exposed to (or part of the structure in case of multiple measurement locations).
Developing AI models to calculate corrosion risks entails two parts: (1) the development of the models themselves and (2) good data with which to train the models. To train the model, data sets are required of high quality. This not only means little noise on the data, but also sufficient variability in the data. Ideally, the data used for training should cover the entire expected parameters space. I.e. all the situations which can be expected to occur should be contained within the training data set. It is impossible to realise this with data from field measurements. Therefore, for the training step, several laboratory experiments have been set up, where the environment can be changed in a controlled manner. These data sets will be used for the initial training of the model and to better understand not only the impact of the environmental parameters, but also the response of the used sensors.
The SOCORRO application
The application developed in the SOCORRO project is described in a number of deliverables and the lessons learned during the analysis of the demonstration data is included in the reports on the demonstrators. The application itself and some videos describing it can be found here.
Integration with existing Management Software
The online SOCORRO Application is only one possible visualization for the model. The trained model itself can be incorporated in existing management platforms that have extensive capabilities to interface with other aspects of the organizational and operational tasks.
Within the framework of the SOCORRO project, this has been illustrated using the Offshare platform, which has been provided by e-BO as an in-kind contribution to the project. Management platforms like Offshare can be used to manage and visualise data, but also for organisational management and to plan operational activities. This showcases the power of having the SOCORRO approach combined with existing management software, and opens the possibility to generate automated alarms, warn the right people, generate work orders, and even generate equipment lists.
The first step in corrosion management using existing management platforms is data storage and visualisation. The Offshare platform for example allows to easily structure the data according to the specific asset being monitored and the parameter being monitored. By doing this, a very clear and easy to use data flow can be established that avoids miscommunications.
Visualisation of data allows users and decision makers to easily compare and see differences between assets. An example of uploaded corrosion risk data is shown in the Figure below, clearly showing an increase in corrosion risk, in this case starting from mid-July 2022. Within Offshare, it is also possible to set thresholds or threshold bands to easily visualize if the corrosion risk is become too high.
The real strength of management platforms like SOCORRO to be used in Corrosion Management then comes in what can be done with these threshold and alarm levels. Automated alarms may be generated when certain parameters cross their thresholds (this may not be limited to the corrosion risk, but may also include certain environmental parameters like pH that can tell you something about how your asset is behaving). Operators, integrity managers, etc. can be warned by e-mail, text message, etc. when immediate action is required.
Taking the example of accumulated corrosion risk, which is a good indicator for the overall corrosion exposure of the asset. When it crosses a certain threshold, that doesn’t necessarily mean that immediate action is required. However, it may be a trigger to plan a visual inspection or routine maintenance. The management platform can then automatically generate a work order for this inspection, and allocate time and resources to this task. Lists of the equipment needed for the task may also be drafted automatically. With all the data available, the tasks and equipment needed may be further detailed based on past events. For example, if there have been certain events of higher than normal pH measured, a task can be added to verify the functionality of the pH-probe and a hand-held pH meter may be added to the equipment list to do so.
Another important aspect of using dedicated management software in corrosion management is that it provides a very powerful means to keep track of data over very long periods of time. When thinking of life extension of offshore wind turbines for example, the life expectancy is 25-30 years. At the end of life, decisions have to be made about life extension and/or ending of life. It is then critical to have useable data from the entire lifetime of the asset. Having a structured framework to keep track of the data, that survives changes of personnel and can be used across multiple windfarms makes sure that (1) this data is available and (2) can more easily be evaluated and compared to other assets.
What has here been demonstrated with the Offshare platform could also be applied to other management platforms.
You can find more information on eBO and their Offshare software here.
If you want to dive deeper into the results, you can find the deliverables here:
Comments