Condition monitoring for the many
Today, equipment can be monitored efficiently and serviced according to business needs, rather than schedules, thanks to the Industrial Internet of Things, says Mohamed Zied Ouertani.
Traditional maintenance work, using time-based schedules, is focused mainly on equipment that already works well. This is time consuming and results in equipment that is working perfectly well being adjusted or even replaced. Most mechanical failures are not related to the age of the equipment, but most maintenance teams act as if the opposite were the case, working their way through a list of time-based actions.
More effective maintenance can be achieved by addressing issues based on level of priority, business needs and actual conditions. This can be done using the monitoring technologies that have become available thanks to the Industrial Internet of Things (IIoT).
According to calculations made by ABB, with more effective monitoring, the cost of maintenance can be reduced by between 15 and 40%. The number of failures during operation can be cut by over 90% and contribute 2 to 3% to plant availability. Even fractional improvements in plant uptime is a big contributor to improved earnings.
Today equipment degradation can be detected before faults occur, reducing downtime, cutting costs and improving safety. With all the relevant data held in cloud repositories, equipment can be analysed with big data technologies, helping to map failure patterns, failure modes and equipment performance.
A structured approach
Failure mode and effects analysis (FMEA) is a structured way of approaching equipment failures and their possible causes. Maintenance experts across a range of industries have used this methodology for many decades, usually with pen and paper. This process has now been packaged into software and the analysis is carried out by computers.
All processes and pieces of equipment are represented by digital twins in the software. Causes of failure, for each piece of equipment are identified - including failure modes that cannot be observed by sensors. All available information, such as data, feedback and expert opinions are used to build these models. They are continuously improved using data and customer feedback.
Once the structured FMEA model is in place, the expected time from the first indication of a fault to actual failure, is calculated. When the first indication appears, for instance bearing vibration, it is possible to work out the remaining time to failure – in this case, bearing failure. With sufficient field data this interval can be calculated with a high degree of accuracy.
This provides a robust basis for the planning of maintenance activities. Knowing, for example, that you have three weeks to replace the bearing enables preparations to be made. Processes can be shifted to redundant equipment or the load can be reduced until maintenance can be carried out.
Handling the data
When expanding monitoring to cover a greater number of machines, the sheer volume of data can become an issue. The challenge, until now, has been that hardwiring is costly and wireless communication is not continuous, as wireless sensors provide readings on a periodic basis to save battery power.
Hardwiring can only be justified for the most critical pieces of equipment, normally about 5% of all machinery. Wireless is not suitable for all types of equipment and conditions – usually, it can only cover about 15% of the equipment. Now, the remaining 80% can be monitored using edge computing, a concept that involves processing the data locally, in peripheral devices, with only the most relevant information being sent to a plant or enterprise asset management system. Several years ago, a world leading Chemical giant started several programs aiming to improve profitability and operations by using leading-edge technologies.
At its main site in Germany, the company has hundreds of different production facilities with thousands of rotating assets. Only a fraction of these used to be monitored. Monitoring thousands of assets is a huge undertaking, bearing in mind the volumes of data that would need to be collected, transferred, processed and analysed. Together with ABB, BASF set out to find ways to address these issues.
Sending raw readings to a high-level system is suitable and efficient when monitoring dozens of assets, but not thousands. The amount of raw data increases as quickly as the number of monitored assets, generating high traffic in the wireless communication infrastructure and requiring a large amount of data storage and processing power.
Every wireless sensor can generate up to 250 megabytes of raw data per day and frequently more than 99% of this data is irrelevant for understating the trends in the machine health. To solve this challenge, a new approach was needed. The solution was found with Edge computing – a method of optimising applications by moving a portion of the application to one of its peripheral parts, such as a field installed sensor.
By processing the data locally, inside the sensor attached to the machine, the volume of data sent to a higher-level monitoring system can be reduced from hundreds of megabytes to a few kilobytes. The work started in Germany has resulted in a range of smart, IIoT-enabled sensors being available.
Having analysed the data, the peripheral asset is not only able to tell the higher-level system what the symptoms are – such vibration or temperature readings – but also what it thinks the problem is. This is a huge benefit to maintenance teams, who no longer need to go out to examine the issue before addressing it. Instead, they can start with a work order that lists the parts needed and the tools required to do the job. The task can then be completed in just one visit, instead of multiple trips. It is also possible to involve artificial intelligence (AI) which can help outline machine condition in a quantitative way.
As well as addressing problems in order of priority and cutting out unnecessary maintenance work, IIoT technologies can help eliminate any unplanned shutdown due to equipment malfunction, failure or maintenance actions gone wrong.
It also becomes easier to share information between daily operations and the maintenance department, as both have access to the same data since they use the same system. For some manufacturers this kind of collaboration between departments is nothing short of revolutionary and can go a long way to improving the bottom-line performance of the facility on a daily basis.
The next step in optimising asset management is to integrate the computerised maintenance management system. This will enable the maintenance department to plan its activities, track spare parts inventory and even order external contractors in the same system that is aggregating condition monitoring data across the enterprise.
Mohamed Zied Ouertani is digital lead & technology manager Chemicals & Refining at ABB.>
Source: Control Engineering Europe - All Articles