Best Practice in Predictive Maintenance
With increased emphasis on predictive maintenance seen as one of the major trends that will dominate manufacturing in 2021 and beyond, how should manufacturers start to tackle the challenge of assessing and responding to machine health insights?
In this film, Senseye's Alexander Hill and Rob Russell are joined by Make UK's Jim Davison and Dr. Hannah Edmonds from the Manufacturing Technology Centre to explore the best practices that deliver maximum benefit from predictive maintenance, and how the knowledge of condition monitoring specialists and reliability engineers can be democratized to enable predictive maintenance to be done sustainably and at scale.
Alexander Hill, Senseye: Predictive maintenance is going to be a focus of manufacturing and industrial companies this year, and the next few years going forward. It has to be. In order to enjoy these efficiency gains, save money and enable things like remote working and remote maintenance, predictive maintenance is an absolutely essential component.
Rob Russell, Senseye: That main use case for smart factories and digitization has been built in and around predictive maintenance. There is a level of domain knowledge that is required, to first understand how to get the right type of information from the machines, but then also how to use that within the maintenance context.
Jim Davison, Make UK: The traditional plant engineers tend to be like me, with grey hair, getting towards the end of their careers. It’s a challenge for us to recruit the next generation to fulfill those roles. By deploying things like smart technologies, it actually brings a really exciting element to the role that people would not have been familiar with.
Rob Russell, Senseye: When looking at the challenges of how to respond to the outputs, with regard to machine health, you still have to approach that with good engineering policy and practice. The data sources and the identification of those early problems that you’d have within machines that enable predictive maintenance are sometimes detecting quite subtle early stages of failure.
The maintenance team will start to acquire ways that they can substantiate their findings and also understand that early detection is giving them a much larger time window. This enables them to do the planning and investigation when production doesn't need the machines and then also plan the maintenance at a later stage, when again the machines are down and there will be no production impact.
Dr. Hannah Edmonds, Manufacturing Technology Centre: As part of a larger program I’ve been involved with, we had lots of different production elements together, lots of different analytics that were happening - the ability to combine those gave us added value.
We used Senseye’s automated analysis to diagnose issues within the production system and identify the useful remaining life of machinery. We could visualize that through an additional program but we also combined it with a computerized maintenance management system, to schedule maintenance tasks.
Alexander Hill, Senseye: We do not want to replace condition monitoring engineers and reliability specialists. We just want to make sure that their time is used really effectively because they’ll often be looking at tens of thousands of data streams and trying to understand which ones are important. That’s a very taxing, very difficult job to do.
The way that we’d approach this problem is that we automate, as much as possible, the traditional role of a condition monitoring engineer looking at thousands of points of data and trying to understand which ones are important.
We don’t need to be manually building and tweaking models, our patented technology and algorithms allow us to automate that entire process.