In this special guest feature, Steve Jennis, Senior Vice President and Head of Global Marketing for ADLINK and PrismTech, discusses the benefits of predictive maintenance over preventative maintenance. A recognized expert in the Industrial Internet of Things marketplace, Steve is regularly invited to comment on IIoT topics and contribute to thought-leadership initiatives. With more than 20 years’ experience in high-technology management, he has achieved success in both large corporate and start-up environments and is widely recognized as a leading business strategist and marketer for infrastructure and platform software. He holds an honors degree in physics from Loughborough University and several postgraduate and professional qualifications. He is based in the United States.
The capability to analyze industrial data very quickly for control loops and instantaneous responses to changes in machine performance has been available for many years, thanks to logic controllers, industrial computers and distributed control systems. What’s emerging in the world of the Industrial Internet of Things (IIoT), however, is also the analysis of longer-term data – big data – whether it’s processed in the cloud or, ideally, closer to the device at the Edge.
We’re not only doing analytics in real-time for machine control, but also longer-term analysis for predictive maintenance and quality management. This way if a machine is showing signs of impending malfunction, that can be detected and remedied before any down time or fatal error occurs that would stop or waste production. The real innovation today is not the very fast analytics that we’ve been doing for years, but the collection of longer-term data for the machine-learning that can predict issues before they happen.
Let’s simplify the concept. We know that preventive maintenance is the scheduled checkup of machines or devices to avoid a sudden breakdown or work stoppage. But that practice is like a human going to a doctor every month or so just for a checkup. It doesn’t make sense and isn’t cost effective. We go to our doctors when we’re not feeling well, they analyze the symptoms and prescribe a remedy based on science (data) collected over many years.
That’s the same result with predictive maintenance: The machines share operational data, the analytics predict any problem and the appropriate corrective action. Then a remedy can be applied, on-demand.
Predictive analytics have been reported to be between 10 percent to 20 percent more cost effective than basic scheduled maintenance. Rather than pre-planned maintenance, we can have the machine report it’s not doing well – whether it’s a sticky bearing or an odd vibration or a strange sound or electrical anomaly – and send an engineer to correct it before a production problem occurs – effectively just-in-time maintenance. This is much more cost effective in terms of resource utilization of service personnel and skills allocation. Instead of sending an expensive engineer to give a machine an unnecessary once-over, the engineer can be allocated to other useful work but dispatched on-demand to the ailing machine.
Furthermore, you can aggregate this new operational intelligence across many, many machines. So, if you’re a machine OEM and you have thousands of machines deployed, you can potentially monitor the performance of all of them, thereby building up a much larger dataset in terms of how the machines are performing, and what is normal and abnormal behavior. That is the essence of machine-learning.
Avoiding production shutdowns is paramount. Something as simple as a conveyor belt roller can cause a work stoppage that could cost millions of dollars in lost production. So, it’s not just the maintenance cost of the machine itself that is lowered, but the much more significant financial risk of lost production is greatly diminished. If you can prevent or reduce downtime, you obviously have a fantastic ROI opportunity.
Persistence Market Research recently reported the value of the predictive maintenance market today is nearly $600 million (U.S.) and is expected to soar to $3.1 billion by 2024. The major players in the market include IBM, General Electric and Bosch, among others.
We’re now not only doing machine control, but also doing longer-term analytics to prevent catastrophic failures and downtime. Oftentimes, the cost of the machine repair is relatively small compared to loss of production. If you can prevent or reduce downtime, you have the opportunity of saving large amounts of money.
And that’s just what the doctor ordered.
Sign up for the free insideBIGDATA newsletter.