It’s a given that machines will break (often when you least expect it) and is probably the oldest rule in manufacturing. Reliability and predictability are therefore critical elements in asset-intensive industries that use rotating machinery and industrial equipment. Downtime in these industries can result in losses to the tune of millions of rands. Despite this, many manufacturers still make use of supervisory control and data acquisition (SCADA) system and programmable logic controllers. These systems, built following the reactive maintenance model, typically use sensors to detect problems and trigger an alarm when an issue arises. Following a reactive approach to fault detection and resolution does not close the feedback loop to prevent breakdowns or solve problems systematically by following the correct, complying procedures. Furthermore, the supervisor or technician on duty who receives this warning should be equipped with the correct tools to identify the cause of the problem and resolve it correctly. This will avoid applying, temporary or “quick fix” solutions while in a state of alarm. How can this be achieved? Before we delve into the answer, let’s first look at what anomaly detection entails.
What does anomaly detection entail?
Anomaly detection is the process whereby variables that do not belong to an expected pattern in the same dataset are identified. These variables are usually unobservable to the human eye. These anomalies are usually early signs of failure which will ultimately result in equipment breakdown or faults in the working of the equipment. Having an IoT-driven predictive maintenance solution in place means your data is captured and stored on a platform where machine learning can be applied. With this approach, problems that typically lead to breakdowns and downtime can be detected early and addressed before a breakdown occurs. This approach can be followed using the same data that would have been collected using a SCADA system (or similar). This, according to McKinsey & Company “typically reduces machine downtime by 30 to 50 percent and increases machine life by 20 to 40 percent”.
In addition, having a web and mobile app will help to detect the anomaly and alert the right stakeholders to ensure a rapid response and that the correct steps are followed. The app can be built to include all the correct operational safety and maintenance procedures that correlate with certain systems and workflows.
The advantages of predictive maintenance
A predictive maintenance model can yield many benefits such as improved production capability and product quality as well as reduced maintenance cost. With a proper predictive maintenance programme, you can now delay or replace a substantial portion of your preventative maintenance tasks. Additionally, repairs will be limited only to what is broken and maintenance can be scheduled in advance. This further reduces overtime and saves spare parts while improving mechanics’ skills that can now be used for other activities.
Ultimately, anomaly detection in predictive maintenance will detect a problem early, ensure a faster response, inform the right people who can then follow the correct procedures to resolve any problems before a breakdown or downtime can occur. when implemented correctly, it can help you avoid costly repairs and equipment downtime and save your business millions of rands.
DOWNLOAD FREE GUIDE:
How predictive maintenance can reduce manufacturing breakdowns by up to 75%