Industrial Automation Magazine India

Implementing predictive maintenance using AI can significantly reduce plant downtime

Published on : Monday 03-07-2023

Franziska Rostan, Process Industry Management, Beckhoff Automation.

Can AI be used to optimise performance and safety in the oil and gas industry?
In today's market, users must adopt and implement modern technology appropriately in order to remain competitive in the future. There is still a lot of potential for improvement, particularly in the process automation sector. In order to tap into this potential, plant operators have the opportunity to integrate innovative and new technology, such as machine learning, which is part of artificial intelligence, into the automation system. Operations in the process industry are often complex and time-consuming, and this technology can help to optimise these processes and increase plant safety.  

Implementing predictive maintenance using artificial intelligence can significantly reduce plant downtime. Here, plant data such as temperature, pressure, and vibration are recorded and analysed via AI algorithms. This analysis can identify potential problems in both the process and the plant components. In order be effective, this approach requires long-term data collection to provide sufficient training data for the AI algorithms.

Using predictive maintenance with AI means that deviations from ideal conditions for the plant can be detected. When these deviations are detected at an early stage, maintenance work can be planned and carried out in good time, before major damage occurs to the plant and the operating personnel are faced with dangerous situations. This helps to maximise plant availability, increase productivity, and enhance safety.

In addition, image processing using artificial intelligence offers further optimisation potential for processing plants. By analysing visual data, AI algorithms can identify patterns and correlations that indicate areas for improvement. For example, the algorithms help to monitor the flow of materials, can identify bottlenecks or detect deviations from optimal operating conditions. This ultimately optimises plant performance and increases efficiency.

AI can also help optimise resource consumption. Potential savings can be identified by analysing data and applying special algorithms. For example, AI can monitor energy consumption and suggest ways to reduce energy demand to cut down on raw materials.
Beckhoff offers a seamlessly integrated solution for artificial intelligence especially for machine learning in the TwinCAT 3 automation software. The TwinCAT interfaces to machine learning algorithms enable AI methods to be used in conventional control environments and thus support process optimisation. This allows plants to reap the benefits of machine learning, without having to make extensive changes to their established infrastructure.

The TwinCAT Vision image processing solution from Beckhoff also enables the acquisition, processing, and evaluation of images in real time. Cameras can be integrated into the automation environment to capture and analyse visual information and incorporate it into the control process.

What advantages does edge computing provide for the process industry as compared to the traditional model?
In the main, process technology equipment is still automated centrally. Not only is this solution the most familiar to many users, but it is also believed to be the most straightforward, as it allows all the field devices in the process control system (DCS) to be monitored and controlled from a single location with a clear overview. To this end, all information is sent to a control cabinet and is then processed and analysed via a central industrial PC. The sensors and actuators are connected via remote I/Os.

In edge computing, data processing and analysis takes place directly at the data source. Since the edge device is able to act autonomously, data can be processed continuously even in the event of a network failure, which increases system availability. Instead of sending all data to a central location, the data is pre-selected, sorted and processed to filter the process data so that only the truly relevant data remain. This is particularly important as the data streams are growing continuously, and this process reduces the load on the networks.  

The decentralised architecture consists of several control systems that are each individually assigned to a specific plant section and take over its process control. With the C60xx ultra-compact Industrial PCs, Beckhoff offers maximum computing power that takes up next to no space. Alternatively, the embedded PCs or fieldbus couplers with a modular I/O level allow terminals from the Beckhoff I/O range to be directly connected, so that sensors and actuators can be linked to the edge device with ease.  

However, edge computing is not just a solution for greenfield scenarios as a completely new system concept. Edge devices can also be integrated into existing plants in brownfield scenarios. With the help of a retrofitted edge device, for example, IoT functions can be introduced into the existing application without any software or hardware changes to the existing control unit. The openness of the Beckhoff control architecture also enables communication with all common process control systems by configuring the industrial PC interface or selecting the appropriate fieldbus coupler.

By encapsulating dedicated functions in edge devices, plant operators can scale their control architecture without any extra work. Thus, integrating another edge device means that an additional plant section can be automated or additional functions can be mapped. With this scalability, plant operators can respond to changing industry requirements quickly and adapt their control systems accordingly. Flexibility in production can be increased hugely as a result.

Nevertheless, all relevant and processed process data should also be able to be fed back into a central control system so that plant operators can monitor the process status and intervene if necessary. In addition, the plant can be optimised with this database and predictive maintenance can be implemented to minimise plant downtime. With Beckhoff solutions, data provision is unrestricted because with TwinCAT, a single software package is used for everything from real-time control to data analysis, as well as for all IoT tasks. TwinCAT's modular structure allows the software to be optimally adapted to the application.

(The views expressed in interviews are personal, not necessarily of the organisations represented)