This audio was created using Microsoft Azure Speech Services
One of the many benefits Internet of Things technology brings to companies in the commercial and industrial space is predictive maintenance, where machines and other equipment essentially tell you when they need attention. But it only works if you’ve got a solid industrial edge infrastructure that can process all the data predictive maintenance requires.
The problem with prescriptive maintenance
Predictive maintenance is a major step forward as compared to the prescriptive maintenance that commercial and industrial companies have used for decades. Prescriptive is similar to auto maintenance, where you change the oil, filters, belts, hoses and so on according to a schedule the manufacturer suggests.
The problem with this approach is none of these things fail according to any schedule; it all depends on the conditions they face. A car that routinely travels on dusty country roads will likely need filter changes far more often than one that travels predominantly on highways. Similarly, if you live in an extremely hot or cold climate, or tend to drive fast, that will put additional strain on your engine oil.
The same goes for machines on plant floors and equipment such as refrigerators and ovens. While there’s nothing wrong with maintaining them according to the manufacturers’ instructions, it’s generally more expensive than addressing issues only as they crop up. Prescriptive maintenance is also more risky; there’s no guarantee a machine won’t fail because a part wears out before it’s “scheduled” to, leaving you with costly downtime.
How predictive maintenance works
Predictive maintenance relies on various sensors and software that enable machines and appliances to continually report on their own status.
Consider a machine on a plant floor. When it’s new, you establish a baseline for what normal performance looks like for that machine. In real time, the machine generates status reports covering dozens or hundreds of data points – how much current it’s drawing, pressure readings and so on. Powerful analytics software continually examines all this data, looking for any anomalies that suggest something is veering from the established baseline or expected performance. At that point, maintenance teams can be assigned to look into the issue and address the problem.
The idea is you fix problems only as they crop up, and before they cause a complete failure. That saves significant money on both manpower and parts vs. performing scheduled maintenance. It also reduces the amount of planned and unplanned downtime, which increases productivity.
Industrial edge powers predictive analytics
Predictive maintenance only works if you’ve got the compute power on-site (or nearby) to crunch all the data your machines or appliances are generating. This can be millions of data points per hour, so it’s not practical to ship it all to a faraway data center or the cloud. What’s more, some of this data may be quite sensitive, such as production data; you wouldn’t want it falling into the wrong hands – another reason to keep it local and well-protected.
That’s where edge computing comes in. What’s required is infrastructure such as enclosures that can protect the IT compute and storage equipment required to support the analytics software. You need to supply proper power and cooling, as well as physical security to protect against unauthorized access. Depending on the environment, you may also need protection against the environment, whether that’s excessive dust, noise, cold, heat, moisture or the like.
Shift to predictive maintenance: Access industrial edge resources
Schneider Electric, of course, offers an array of micro data center solutions that are a perfect fit for industrial edge environments, and we have numerous partnerships with IT vendors for compute, storage and networking component. On top of that, our recent merger with AVEVA, which makes powerful analytics software for commercial and industrial environments, positions us to offer everything you need to make predictive analytics a reality.
It’s time to throw the schedule away and look into the benefits that predictive maintenance can deliver in your environment.
